2. REVIEW OF THE LITERATURE

All the enzymes and their genes in the heme biosynthetic pathway have been characterized (Kappas et al. 1995, Nishimura et al. 1995). The biosynthesis pathway is illustrated in Figure 2.1. The first and the last three of the enzymes in the pathway are localized in mitochondria; the intermediate enzymes are cytosolic. The precursors of heme are glycine and succinyl coenzyme A which are combined by aminolevulinic acid (ALA) synthase to form 5-aminolevulinic acid in the presence of the cofactor pyridoxal 5'-phosphate. Next, two molecules of ALA are condensed to a monopyrrole porphobilinogen (PBG) by ALA dehydratase.

Porphobilinogen deaminase (PBGD) assembles the four rings of PBG in a stepwise fashion in which the pyrrole ring A is first bound to the deaminase followed by rings B, C and finally D (Figure 2.2.). The dipyrromethane cofactor, which arises from the autocatalytic coupling of two molecules of PBG, is covalently linked to the enzyme. The cofactor functions as a primer to which the four substrate molecules are sequentially attached but is not itself incorporated into the product, hydroxymethylbilane.

The fourth enzyme, uroporphyrinogen III synthase, converts the highly unstable hydroxymethylbilane to uroporphyrinogen III and in this reaction the linear tetrapyrrole molecule is closed to form a ring. In the absence of uroporphyrinogen III synthase, hydroxymethylbilane may non-enzymatically close to uroporphyrinogen I. In normal circumstances, this isoform is present only in minute amounts.

The four carboxylic groups of the acetic acid chains in uroporphyrinogen are removed in the reaction catalysed by cytosolic uroporphyrinogen decarboxylase. In the presence of oxygen coproporphyrinogen oxidase catalyses the removal of the carboxyl group and the two hydrogens from the propionic groups of the pyrrole rings and forms vinyl groups at these positions resulting in protoporphyrinogen IX.

Finally, iron (Fe²⁺) is incorporated into protoporphyrin IX by a mitochondrial ferrochelatase, also known as heme synthase.

The heme biosynthesis is regulated in distinctive manners in erythroid and non-erythroid tissues but ALA synthase is the rate-limiting enzyme of heme biosynthesis in all tissues (Kappas et al. 1995). Two different genes code for ALA synthase: the nonspecific ALA synthase is ubiquitous whereas the erythroid ALA synthase is expressed only in erythroid tissues (Kappas et al. 1995). Biosynthesis of heme in the liver and in other non-erythroid tissues is regulated by a feedback mechanism in which the end product heme regulates the synthesis of non-specific isoform of ALA synthase at the transcriptional and translational level (Moore et al. 1987). The activity of nonspecific ALA synthase may be also induced by various chemicals, drugs, lead, and alcohol and reduced by glucose. The erythroid-specific isozyme of ALA synthase is, in contrast to the nonspecific ALA synthase, upregulated by increased heme concentration (Fujita et al. 1991). The ALA synthase activity is also regulated by many other factors, for example by erythropoietin and by the availability of iron (May et al. 1995).

More detailed information about heme biosynthesis is available in the literature (e.g. Moore et al. 1987, Kappas et al. 1995).

The porphyrias are a heterogeneous group of metabolic disorders. They have been divided into two groups, erythropoietic and hepatic porphyrias (Moore et al. 1987) according to the tissue where the excess porphyrins or their precursors are mainly synthesized. In general, the clinical manifestations of porphyrias are either skin photosensitivity due to photoreactivity of porphyrins, neurovisceral dysfunction (acute attacks), or both (Kappas et al. 1995). Most of the porphyrias are inherited dominantly in autosomes. ALA dehydratase deficiency and congenital erythropoietic porphyria have, however, autosomal recessive inheritance. Furthermore, a significant number of patients with porphyria cutanea tarda are sporadic cases (de Verneuil et al. 1978). Hepatoerythropoietic porphyria is a rare homozygous form of the familial type of porphyria cutanea tarda (Toback et al. 1987). A summary of different porphyria types is given in Table 2.1.

The modern history of acute porphyrias, reviewed extensively elsewhere (Goldberg and Rimington 1962, Dean 1963, Tschudy et al. 1975), begins at the end of the 19th century, when an increasing number of patients with abdominal pain and neurological symptoms was described. The typical attacks of acute porphyria were associated already in these early days with drug treatment.

Most of the porphyrin nomenclature is based on Hans Fischer's, a Nobel laureate in 1930, studies on the heme biosynthesis in the 1920's. In 1937 Jan Waldenström, who had previously worked in Fischer's laboratory, described over one hundred patients with acute porphyria, most of whom originated from a small village in Northern Sweden (Waldenström 1937). In this study the Mendelian dominant mode of inheritance in AIP was demonstrated for the first time. Later, he proposed that porphyrias could result from enzyme defects in the heme pathway and was also the first to use the term 'acute intermittent porphyria' (Waldenström 1957). A screening test for porphobilinogen was introduced in 1941 by Watson, who also was Fischer's co-worker (Watson and Schwartz 1941). He also introduced hematin for treatment of acute porphyrias (Watson et al. 1978). In the 1960's Granick discovered that ALA synthase, the first step in heme biosynthesis, is induced by chemicals which cause acute porphyria (Granick 1963, Granick 1966). This observation led to the concept that overproduction of ALA may represent the primary genetic defect in all acute porphyrias (Granick 1966), but this hypothesis was soon found to be false because increased activity of this enzyme by itself could not account for the distinct patterns of porphyrin precursors and porphyrin excretion. The biochemical background of AIP was confirmed when it was demonstrated that the disease is due to a defect in PBGD (Miyagi 1970, Meyer et al. 1972).

The characterization of the porphobilinogen deaminase gene (Raich et al. 1986, Grandchamp et al. 1987, Chretien et al. 1988, Lee 1991a, Namba et al. 1991) initiated multiple investigations of the molecular genetic background of AIP, and today several mutations responsible for AIP have been identified (see chapter 2.5.2).

The first Finnish patient with acute porphyria was reported in 1928 (Langenskiöld 1928). The first comprehensive study of the prevalence and clinical characteristics of acute porphyrias in Finland was reported in 1976 by Pertti Mustajoki and Pentti Koskelo (Mustajoki and Koskelo 1976). The prognosis, precipitating factors and associated diseases in Finnish patients with acute porphyrias have been investigated thoroughly by Raili Kauppinen and Pertti Mustajoki (Kauppinen et al. 1992).

AIP is the most common type of acute porphyria in most countries (Kappas et al. 1995), only in the Republic of South Africa is the prevalence of variegate porphyria higher than that of AIP. The prevalence of AIP in Finland has been estimated to be 3:100000 (Mustajoki and Koskelo 1976). In other countries the reported prevalence figures are similar or slightly higher: 5-10:100000 in the United States of America, in Western Australia 3:100000 (Tschudy et al. 1975). The highest prevalence of AIP has been described in northern Sweden where the prevalence is 100:100000 and, furthermore, in two small municipalities the prevalence is as high as 0.5-2%. However, the overall prevalence of AIP in Sweden is only 10:100000 (Andersson 1997). In a study of healthy Finnish blood donors the prevalence of low PBGD activity was 1:500 (Mustajoki et al. 1992) and similar figures were obtained in a study which consisted of healthy French blood donors (Nordmann et al. 1997). This implies that the prevalence of AIP, though asymptomatic, may be much higher than previously estimated.

Approximately 10-20% of the affected individuals experience acute attacks during their lifetime and 50% of patients have milder symptoms typical for AIP (Kauppinen and Mustajoki 1992). Half of the subjects with decreased PBGD activity remain, thus, asymptomatic throughout their lifetimes. Acute attacks are extremely rare before puberty and the first acute attack usually manifests at an age between 15 and 40 years (Mustajoki and Koskelo 1976).

The symptoms during the acute attacks are mainly related to neurological dysfunctions. The incidence of symptoms and physical signs are listed in Table 2.2. Abdominal pain is almost invariably present and in severe cases it may mimic 'acute surgical abdomen'. Dark or red urine is also frequently observed but the presence of other symptoms varies considerably which sometimes causes difficulties in making the right diagnosis. Unless treated, the symptoms of an acute attack may proceed to lethal respiratory paralysis. Nowadays, however, life-threatening symptoms during acute attacks seldom occur because of increased knowledge of precipitating factors and improved therapy (Kauppinen and Mustajoki 1992). The course of an acute attack is variable and porphyric symptoms may last from a few days to several weeks (Kappas et al. 1995). Some patients may have symptoms almost continuously, but the majority of symptomatic patients have a few well-delineated attacks with relatively asymptomatic intervening remissions (Kappas et al. 1995).

The porphyric attacks are often precipitated by environmental or endogenous factors, such as changes in hormonal state (menstrual cycle and pregnancy), several therapeutic drugs (e.g. barbiturates and sulphonamides), alcohol, infections, stress, and inadequate caloric intake (Kauppinen and Mustajoki 1992, Kappas et al. 1995, Andersson 1997). According to a Finnish survey, at present the most common cause for an acute attack is the menstrual cycle in females and excessive alcohol consumption in males (Kauppinen and Mustajoki 1992). Avoiding the precipitating factors usually decreases the incidence of attacks. Women at fertile age may need hormonal therapy which suppresses their endogenous sex hormone production.

Some diseases are more common in patients with acute porphyria than in the normal population. The risk for hepatocellular carcinoma is increased 61 to 114-fold (Kauppinen and Mustajoki 1992, Andersson et al. 1996, Linet et al. 1999) and is the cause of death in about one-quarter of AIP patients (Andersson et al. 1996). Chronic hypertension has been suggested to be more common in patients with manifest AIP than in those with latent AIP or in healthy controls in Swedish material of one large AIP family (Andersson and Lithner 1994): 56% of the patients with manifest AIP had hypertension. However, the prevalence of chronic hypertension was much lower (~24%) in the Finnish material which consisted of both AIP and variegate porphyria patients from 53 families and, furthermore, the occurrence of hypertension did not correlate with clinical manifestations (Kauppinen and Mustajoki 1992). When Finnish AIP patients were compared to the general Finnish population, the prevalence rates of hypertension with medication were higher only in two age-specific subgroups (Kauppinen and Mustajoki 1992). Cardiovascular mortality was not higher among AIP patients (Kauppinen and Mustajoki 1992, Andersson et al. 1996). The prevalence of chronic renal failure has been suggested to be higher among AIP patients (Kauppinen and Mustajoki 1992, Andersson and Lithner 1994).

Under normal conditions, the heme biosynthesis pathway is able to fulfill its task inspite of decreased PBGD activity. It is widely presumed that increased metabolic demand for heme in the liver leads to an induction of ALA synthase expression (Kappas et al. 1995, Elder et al. 1997). This results in accumulation of intermediates of the heme biosynthesis pathway, namely delta-aminolevulinic acid and porphobilinogen, since the defective enzyme is not able to convert all the porphobilinogen to hydroxymethylbilane. The resulting deficiency of the final product, heme, leads to further induction of ALA synthase provoking circulus vitiosus, an escalating metabolic chain reaction.

About 65 percent of the aminolevulinic acid produced in the rat liver is utilized for the formation of CYP isoenzymes (Sassa and Kappas 1981). Some drugs, alcoholic beverages and steroid hormones, i.e. substances which are known to precipitate porphyric symptoms, are known to induce or increase the inducibility of hepatic ALA synthase and may also increase the intrahepatic requirement for heme by inducing the synthesis of CYP isoenzymes (Elder et al. 1997). This leads to the accumulation of delta- aminolevulinic acid and porphobilinogen because the deficient PBGD cannot fulfill the requirements of the accelerated heme biosynthesis.

The symptoms of AIP result from neurologic dysfunction and acute attacks are known to be caused by the induction of ALA synthase. However, the details of molecular pathogenesis are still obscure. Three major hypothesis has been proposed to explain the symptomatology of an acute attack (Moore et al. 1987, Kappas et al. 1995):

1. depletion of heme in the cells of the central nervous system (Lindberg et al. 1996, Lindberg et al. 1999),

2. neurotoxicity of accumulated intermediates, especially aminolevulinic acid (Bonkovsky 1993), and

3. neurotransmitter disturbance secondary to the deficiency of heme and tryptophan pyrrolase (a heme dependent enzyme) in the liver (Litman and Correia 1983).

Although all these theories are supported by some experimental findings, none of them can explain all the various symptoms during an acute attack, and the pathogenetic mechanisms of an acute attack remain, thus, unsolved.

The development of a transgenic porphobilinogen deaminase deficient mouse (Lindberg et al. 1996, Lindberg et al. 1999) is the first experimental model for acute porphyrias. Studies with PBGD (-/-) mice suggest that heme deficiency and consequent dysfunction of hemeproteins cause chronic and progressive neuropathy leading to impairment of motor coordination and muscle weakness (Lindberg et al. 1999). However, the mouse model more resembles the chronic neuropathy seen in some patients and it has not yet been able to elucidate the pathogenetic mechanisms of acute attacks.

During an acute attack AIP patients always excrete increased amounts of the porphyrin precursors, delta-aminolevulinic acid and porphobilinogen, in urine (Bissel 1982). Qualitative tests, e.g. Watson-Schwartz and Hoesch tests (Bissel 1982), have been developed for rapid confirmation of diagnosis, but sensitivity or specificity of these tests limit their usefulness (Lamon et al. 1977). A quantitative assay (Mauzerall and Granick 1956) has been used for more precise diagnosis. During the asymptomatic phase more than one half of AIP patients do not excrete PBG in urine and these patients are less susceptible to acute attacks (Kauppinen and Mustajoki 1992).

Biochemical diagnosis of acute intermittent porphyria is based on measurement of erythrocyte porphobilinogen deaminase activity (Magnussen et al. 1974, Ford et al. 1980). In this assay, porphobilinogen is used as a substrate and the uroporphyrin I formed is measured by spectrofluorometry. In general, the erythrocyte PBGD activity in AIP patients is ~50% of that found in normal subjects. Due to alternative splicing of the erythroid and wild type isoform (Grandchamp et al. 1987), however, the erythrocyte PBGD activity is normal in the variant form of AIP (Mustajoki 1981). Furthermore, there are wide individual variations of PBGD activity among healthy controls (Mustajoki et al. 1992) and the normal and pathological values have an overlapping zone. In addition, many diseases - for example renal insufficiency, iron deficiency anemia and various malignancies - may affect PBGD activity (Moore et al. 1987). These confounding factors limit the usefulness of biochemical methods in the diagnosis of AIP.

Before the DNA era, AIP was divided into two major subtypes according to the ratio of enzyme activity and the amount of immunoreactive protein in erythrocytes (Anderson et al. 1981, Desnick et al. 1985, Lannfelt et al. 1989a). When the amount of immunoreactive protein reflects the amount of enzymatically active protein, the patient is classified as cross-reactive immunologic material (CRIM) negative. CRIM negative patients have been further divided into two groups according to the PBGD activity in their erythrocytes: in type 1 the enzyme activity is 50% of the normal activity whereas in type 2 the erythrocyte activity is normal. Several explanations have been proposed for the absence of detectable mutant protein, i.e. CRIM negativity (Mustajoki and Desnick 1985): instability of mutant mRNA, insufficient translation, rapid intracellular decay of the mutant polypeptide, or due to an altered structure, the antibody fails to detect the mutant polypeptide.

Enzymatically inactive but immunologically detectable protein is detected in erythrocytes of CRIM positive patients. Two CRIM positive subtypes have been identified. In type 1 the ratio of the protein and enzyme activity is ~1.6, i.e. the amount of inactive enzyme is 60-80% of that of the active enzyme. In type 2 the CRIM/activity ratio is ~5.7, i.e. the amount of inactive enzyme is 5.7-fold that of the active enzyme. In both CRIM positive subtypes the amount of enzyme intermediates is increased (Anderson et al. 1981, Desnick et al. 1985), suggesting that the enzyme protein is synthesized normally but it cannot catalyze deamination or elongate the pyrrole chain normally.

PBGD is a monomeric enzyme and in humans two isoforms are present: the 44 kD house-keeping enzyme and the 42 kD erythroid-specific enzyme. The ubiquitous polypeptide, which consists of 361 amino acid residues, is encoded by a single gene on chromosome 11q, and the two isoforms differ only in their N-terminal amino acid sequences; the first 17 amino acids of the wild type enzyme are not present in the erythroid isoform (Grandchamp et al. 1989c). Since, the deaminases show exceptional heat stability, their characterization has been very convenient (Shoolingin-Jordan 1995).

PBGD catalyzes the tetramerization of porphobilinogen. The reaction produces a highly unstable 1-hydroxymethylbilane (preuroporphyrinogen) intermediate which is followed by the formation of cyclic tetramere uroporphyrinogen by uroporphyrinogen III synthase. PBGD assembles the four rings of PBG in a step-wise fashion, in which the pyrrole ring A is first bound to the deaminase followed by rings B, C, D (Figure 2.2.) The dipyrromethane cofactor anchors the substrate molecules at the catalytic center and directs the construction of the tetrapyrrole (Jordan and Warren 1987). The cofactor is formed by autocatalytic coupling of two molecules of PBG, the same molecule which also acts as the substrate for the reaction.

PBGD has, thus, the ability to perform multiple chemical reactions despite its small size. The enzyme catalyzes the repetitive condensation of PBG units with an acceptor pyrrole chain and is able to 'count' precisely and terminate the reaction when the tetrapyrrole chain has been assembled. In addition, the apoenzyme installs its own dipyrromethane cofactor to form the holoenzyme.

The structure of human PBGD has been predicted using the three-dimensional structure of PBGD purified and crystallized from E. coli (Louie et al. 1992, Brownlie et al. 1994). The molecule is composed of three domains of similar sizes linked together by flexible hinge regions (Figure 2.3.). The enzyme possesses a single active site that is used for each porphobilinogen condensation (Warren and Jordan 1988, Louie et al. 1992). In E. coli (the corresponding human residues are in parenthesis), the dipyrromethane cofactor is covalently bound to cysteine-242 (Cys261). In addition, residues aspartate-84 (Asp99), arginines 131, 132, 149, 155 (Arg149, 150, 167, 173), lysine-83 (Lys98), as well as some other residues form salt-bridge and hydrogen bond interactions with the cofactor (Brownlie et al. 1994). The positively charged side chains of Arg11, 132, and 155 (Arg26, 150, 173) are thought to interact with the substrate (Lambert et al. 1994).

Fifty-eight residues in the PBGD polypeptide are invariant according to amino acid sequence deduced from the nucleotide sequence determination from a wide range of organisms, and the conserved amino acids are clustered in hydrophobic core of the molecule, in other conformationally important locations, or at the active site. This suggests a similar structure and, furthermore, similar mechanism of catalytic activity in organisms with different phylogenetic age (Brownlie et al. 1994).

The gene coding for PBGD, which has been identified and thoroughly characterized (Raich et al. 1986, Grandchamp et al. 1987, Chretien et al. 1988, Lee 1991a, Namba et al. 1991, Yoo et al. 1993), is assigned to chromosome 11q24 (Namba et al. 1991). The size of the gene is 10 kb of which 1.3 kb represents coding sequence. The genomic sequence is divided into 15 exons ranging from 39 and 438 bp and 14 introns ranging from 87 to 2913 bp (Figure 2.4.).

Two isoforms, arising from different promoters, are transcribed (Chretien et al. 1988). The mRNA of the housekeeping (non-erythropoietic) isoform contains exons 1 and 3 to 15 coding for an enzyme of 361 amino acids, whereas the erythroid isoform is encoded by exons 2 to 15 (Figure 2.4). The translation initiation codon for the housekeeping isoform is located in exon 1 and for the erythroid-specific isoform in exon 3. The erythroid isoform, thus, lacks the first 17 amino acids of the amino terminus.

The erythroid promoter region is located in intron 1 and its structure is very similar to that of the -globin gene (Chretien et al. 1988). The erythroid-specific transcript, thus, appears to be regulated by similar set of transcription factors, suggesting that both genes are similarly regulated during erythroid differentiation (Mignotte et al. 1989). The promoter region for the housekeeping transcript is located upstream of exon 1 (Chretien et al. 1988, Yoo et al. 1993). The minimal promoter sequence has been identified by deletion mapping (Lundin and Anvret 1997) and it is located between -243 and -115 nucleotides from the translation initiation codon. This region contains two Sp1 consensus recognition sequences, a 13 bp repeat sequence, and an AP1 consensus recognition sequence. In addition, sequence elements that have negative regulatory function are located closer to the translational initiation site (Lundin and Anvret 1997).

To date, thirteen intragenic polymorphisms have been identified in the PBGD gene (Lee and Anvret 1987, Llewellyn et al. 1987, Lee et al. 1988, Gu et al. 1991, Lee 1991b, Picat et al. 1991, Daimon et al. 1993a, Yoo et al. 1993, Law et al. 1999, Whatley et al. 1999). One of the polymorphisms is exonic, located in exon 10, but it does not alter the amino acid sequence. Nine polymorphisms are located in intron regions of the gene; four of them are found in intron 1, the longest (2.9 kb) intron in the PBGD gene. In addition to the intragenic polymorphisms, two polymorphisms have been identified in the non-erythroid promoter region (Picat et al. 1991, Schreiber et al. 1992, Lundin and Anvret 1997).

Six Alu elements have been found in the intronic sequences of the PBGD gene. Alu elements are derived from 7SL RNA by internal deletions or point mutations of the 7SL sequence followed by dimerization (Makalowski et al. 1994). They integrate throughout the primate genomes via a process called retroposition (Rogers 1985) which involves generation of an RNA polymerase III transcript, reverse transcription, and integration at staggered nicks within AT-rich regions of the genome (Jagadeeswaran et al. 1981). A typical Alu element is 282 nucleotides long and it is composed of two homologous but distinct subunits. Alu repeats can be divided into various subfamilies based on the presence of commonly shared diagnostic mutations (Jurka 1993). Two of the Alu sequences in the PBGD gene are located in the 5' flanking region, three in intron 1. The sixth Alu element is in intron 9 and is the only Alu in sense orientation. Five of the Alu sequences belong to the relatively modern Sa subfamily whereas one is homologous to the older J subfamily (Yoo et al. 1993).

Up to July 1999, 149 different mutations resulting in AIP had been identified in the PBGD gene worldwide (Table 2.3.).

Of the mutations, 56 (38%) are substitutions of one amino acid and 37 (25%) mutations affect splicing. Fifteen mutations result in an immediate termination codon and, furthermore, 36 mutations cause frame shifts which have been predicted to lead to a premature termination codon after a variable number missense amino acids. One mutation hits the translation initiation codon and the remaining three mutations are deletions or insertions, which do not cause frame shift.

Figure 2.5. shows the distribution of the mutations by location and mutation type. No mutations causing AIP have been identified in exon 2 and its flanking sequences. This was somewhat expected, because exon 2, transcribed only to the erythroid mRNA, does not code for polypeptide since the initiation codon of this isoform is located in exon 3. Mutations responsible for AIP are dispersed quite evenly in the PBGD gene and there is no hot spot for the mutations. Twenty-nine (19%) of the mutations are located in exon 12, which is clearly the highest number of all exons, but this exon the second longest exon in the PBGD gene and it is almost twice as long as the majority of other exons (Figure 2.4.). Notably, six of eight residues known to be crucial for the enzyme activity (R26, K93, D94, R149, R150, R167, R173, and C261) are substituted resulting in AIP, only at positions K93 and R150 there are no reports of mutations so far.

In the light of the mutations published in the PBGD gene so far, the molecular genetic background is highly heterogeneous despite the rather uniform clinical manifestations of the disease. Although studies of geno/phenotype comparisons have not yet been published, nothing implies that the type or the location of the mutation affects the clinical manifestations of the disease. Furthermore, it seems that most of the mutations in the PBGD gene dramatically decrease the enzymatic activity of the polypeptide encoded by the mutant allele.