Structure of the Human Liver / Bone / Kidney Alkaline Phosphatase Gene

In man, there are multiple forms of alkaline phosphatase encoded by at least three homologous genes: placental, intestinal, and liver/bone/kidney. This report describes the characterization of the human liver/ bonekidney alkaline phosphatase locus. The gene appears to exist as a single copy in the haploid genome and is comprised of 12 exons distributed over more than 50 kilobases. In liver, kidney, SAOS-2 human osteosarcoma cells, and cultured fibroblasts, there is a single major start for transcription situated about 25 nucleotides downstream of an A/T-rich motif. The promoter region is extremely G/C-rich, is relatively abundant in the dinucleotide CpG, and contains four copies of the consensus sequence for SP1 binding (GGGCGG). The liverponekidney alkaline phosphatase gene is at least five times larger than the intestinal and placental alkaline phosphatase genes, mainly due to intron size differences. Intron-exon junctions occur at analogous positions in all three genes, but there is an extra noncoding exon at the 5‘ end of the liver/bone/kidney alkaline phosphatase gene. The relevance of our findings with respect to the evolution of the human alkaline phosphatase multigene family is discussed.

1986; Henthorn et al., 1987;Berger et al., 1987) has established that they are encoded by separate homologous gene loci. A placental-like ALP found in trace amounts in testis and thymus is probably the product of a fourth locus (Millan and Stigbrand, 1983;Goldstein et al., 1982;Knoll et al., 1987). It is believed that all of the human ALP genes evolved from a single ancestral gene (Harris, 1982). Placental and intestinal ALPs are closely related, exhibiting about 87% identity at the amino acid level (Henthorn et al., 1987). L/B/K ALP is more evolutionarily distant from the other human ALPs, 52 and 57% amino acid identity to placental and intestinal ALP, respectively (Harris, 1982;Weiss et al., 1986;Henthorn et al., 1987). The genes encoding intestinal ALP, placental ALP, and probably the placental-like ALP, all map to bands q34-q37 of human chromosome 2 (Martin et al., 1987;Griffin et al., 1987), while the L/B/K ALP gene maps to the distal short arm of human chromosome 1, bands p34-p36.1 (Smith et al., 1988).
Each form of human ALP exhibits a characteristic pattern of tissue distribution (McComb et al., 1979). Placental and intestinal ALPs are found predominantly in placenta and small intestine, respectively. In contrast, L/B/K ALP is expressed in numerous tissues, including those for which it is named. Slight variations in thermostability and electrophoretic mobility between the liver/bone/kidney-type ALPs found in different tissues have been attributed to differences in post-translational modification (Harris, 1982;Moss and Whitaker, 1985).
The physiological role of ALP is unknown except that the bone isoenzyme is believed to play a role in normal skeletal mineralization (Robison, 1923;McComb et al., 1979;Wuthier and Register, 1985). Evidence for this role is provided by the rare genetic disease hypophosphatasia (Rathbun, 1948;Fraser, 1957;Rasmussen, 1983). Affected patients suffer from rickets or osteomalacia and a deficiency of L/B/K ALP in all tissues (placental and intestinal ALP levels are unaffected). The most severe form of the disease, usually lethal in infancy, is inherited as an autosomal recessive trait. The genetic defects that result in hypophosphatasia may involve mutations at the L/B/K ALP locus, although this remains to be proven.
The physiological regulation of L/B/K ALP expression has been most closely examined in bone cells. ALP is a differentiation marker for the osteoblastic phenotype and an indicator of bone formation and turnover (Rodan and Rodan, 1984). In uiuo, levels of osteoblast and chondrocyte ALP increase markedly at sites where mineralization is actively occurring (McComb et al., 1979). In cultured osteosarcoma cells, ALP expression is modulated by growth conditions and various biological agents including glucocorticoids, vitamin D, and parathyroid hormone (Rodan and Rodan, 1984).
Direct examination and comparison of the genes that encode the various human ALPS should further elucidate the structure and evolutionary history of this multigene enzyme family. Examination of the promoter regions of the ALP genes may provide insight into mechanisms that mediate their characteristic patterns of tissue expression. To address these issues and to facilitate the study of hypophosphatasia, we report here an investigation of the human L/B/K ALP locus. The structure of the intestinal and placental ALP genes are presented in accompanying papers Knoll et al., 1988); comparison with the L/B/K ALP gene supports the hypothesis that the human ALP multigene family arose by a series of duplications from a common ancestor.

RESULTS
The Human L/B/K ALP Gene Locus-A map of the L/B/ K ALP locus was constructed from overlapping genomic DNA fragments (Fig. 1A). All genomic fragments which are detected by the full-length cDNA in Southern blots ( Fig. 1B) are accounted for in this map, suggesting that the locus exists as a single copy. The gene consists of 12 exons distributed over more than 50 kb. Exon sequences were first localized by hybridization of radiolabeled L/B/K ALP cDNA to restriction digests of cloned genomic DNAs. Intron-exon boundaries were precisely defined by DNA sequence analysis of appropriate genomic fragments (Fig. 2 (in the Miniprint) and Table I). The sequences at the 5' and 3' ends of each intron are in agreement with the consensus sequence for intron-exon boundaries of other eukaryotic genes (Green, 1986;Mount, 1982). All introns begin with the dinucleotide GT and end with AG. Intron number 1, at least 25 kb in length, interrupts the 5"untranslated sequence 105 bp upstream of the initiation methionine codon. All other introns interrupt the gene within protein coding regions. Exon 12, about 1025 bp, contains 263 nucleotides of coding sequence, the termination codon, and the entire 3"untranslated region.
At the end of exon 12, there are putative 3'-mRNA processing signals (shown in Table I) that are commonly found in other eukaryotic genes (Birnstiel et al., 1985); the mRNA cleavage/polyadenylation site is flanked by the sequence AA-TAAA about 12 bp upstream, and a G/T-rich region about 12 bp downstream. L/B/K ALP Genomic Sequences-The nucleotide sequence of each exon was determined in full. There are 11 positions where the L/B/K ALP gene exon sequences differ from our published sequence of the cDNA (Weiss et al., 1986). These discrepancies are summarized in Table I1 (numbered according to the convention used to describe the cDNA).
Four of the eleven differences, at cDNA positions 1712-1713, 1944, 2373, and 2397, identify errors in the reported cDNA sequence. One of these errors occurs within the protein coding region, replacing a leucine for a valine at amino acid position 496. The correct sequence at nucleotides 2204-2207 in the 3"untranslated region is unresolved since the two strands of the cDNA yield ambiguous results at a single base position after resequencing in the presence of deoxy-7-deazaguanosine to alleviate compression artifacts (Mizusawa et al., 1986).
There are six authentic differences between the cDNA and * Portions of this paper (including "Materials and Methods" and genomic exon sequences. Two of these differences occur within protein coding regions: a T to C transition at position 506 represents a silent mutation, and a G to A transition at nucleotide position 880 changes a glycine codon in the cDNA to a glutamic acid codon in the genomic DNA sequence. This difference at base 880 may represent a reverse transcriptase error introduced during cDNA cloning, or a null allele in SAOS-2 cells (see below). Differences at positions 1718,1816,1983, and 2104 occur within the 3'-untranslated region and may represent either cloning artifacts or DNA sequence polymorphisms.
The nucleotide sequence at the 5' end of the L/B/K ALP gene is shown in Fig. 3. Nucleotide positions are numbered with the first base of the methionine initiation codon designated as +1; nucleotides 5' to this position are shown as negative numbers. Over 90% of this region was sequenced on both strands. The sequence of a stretch of G/C-rich DNA, base pairs -280 to -260, could not be determined on one strand. This region was read unambiguously on the complementary strand.
RNA Analysis-Northern analysis was performed on cellular RNAs from SAOS-2 cells, human fibroblasts, human liver, and human kidney using the L/B/K ALP cDNA as hybridization probe. As shown in Fig. 4A, the L/B/K ALP mRNA present in these cells and tissues is of similar size (about 2.5 kb), although the steady-state levels vary, being highest in SAOS-2 cells and lowest in liver. The L/B/K ALP mRNA levels from these sources roughly correlate with ALP enzyme activity levels (data not shown).
To ascertain the start(s) of mRNA transcription, S l nuclease protection and primer extension analyses were performed (Fig. 4B).
For S1 protection analysis, a 5' end-labeled DNA probe, corresponding to nucleotides -642 to -129 in Fig. 3, was denatured, hybridized to a variety of cellular RNAs, and digested with S1 nuclease. One major fragment (69-74 bp) and one minor one (122-124 bp), labeled a and c, respectively, are specifically protected by RNA from SAOS-2 cells, normal human fibroblasts, and human kidney. Longer exposure of the autoradiogram revealed that these same fragments are also protected by human liver RNA (data not shown). Thus, L/B/K ALP mRNA transcription appears to start at the same sites in the four cell types examined, although the level of L/ B/K ALP message varies among cell types.
The SI-protected fragment labeled b in Fig. 4B is present in all samples, including the tRNA control. This fragment, as well as primer extension product 2 in Fig. 4B, may represent artifacts caused by a stem-and-loop structure in the L/B/K ALP mRNA (see "Discussion").
The 5' ends of the L/B/K ALP mRNA, as mapped by primer extension and S1 nuclease protection, are summarized in Fig. 3. The two assays are in agreement. The 5' end of the majority of L/B/K ALP mRNA maps to within a few nucleotides centered around position -195. A minor transcription start site maps to within a few nucleotides of position -145. The start sites mapped by S1 protection are situated a few nucleotides downstream of those mapped by primer extension,  Ray et al., 1988). The asterisk indicates a BclI site which is present in the cDNA. B, Southern hybridization analysis of genomic DNA. Human genomic DNA was digested with the indicated restriction enzymes, size fractionated on a 0.8% agarose gel, transferred to a nylon membrane, and hybridized to 3ZP-radiolabeled L/B/ K ALP cDNA. Molecular weight size markers, determined from BstEII-digested wild type X DNA, are shown at left.

TABLE I Intron-exon organization of the L/B/K ALP gene
Nucleotide sequences of intron-exon junctions were determined according to the strategy in Fig. 2 (in the Miniprint). Exon sequences are shown in upper case letters; intron sequences are shown in lower case. The position of coding nucleotides that border the 5' and 3' ends of the introns are numbered above the DNA sequence with the first base of the ATG initiation codon designated as +1 (nucleotides within introns are not numbered). Amino acid codons bordering the splice junctions are shown numbered with the first amino acid in mature L/B/K ALP designated as +l. Exon sizes and approximate intron sizes are indicated. The region of mRNA cleavage/ polyadenylation is at the 3' end of exon 12 is shown at the bottom of the figure. Nucleotides not present in the cDNA, which are 3' to the polyadenylation site, are shown in lower case. Putative 3'-mRNA processing signals, common to many other eukaryotic genes, are underlined.
- probably due to "nibbling" of the S1 nuclease at the ends of the protected fragments. The L / B / K ALP Gene Promoter-To examine promoter activity of DNA within the 5' region of the L/B/K ALP gene, a 672-bp AuaII fragment, corresponding to bases -801 to -129 in Fig. 3, was inserted upstream of the promoterless bacterial chloramphenicol acetyltransferase gene in derivatives of the plasmid pSVOcat, as described under "Methods" in the Miniprint. The ability of these constructs to express chloramphenicol acetyltransferase activity was measured 48 h after transfection into SAOS-2 human osteosarcoma cells.
The results of these transfections are shown in Fig. 5. When the chloramphenicol acetyltransferase gene is transcribed from the SV40 early promoter (pSV2Acat), chloramphenicol acetyltransferase activity is produced as indicated by the faster migrating acetylated forms of chloramphenicol. When the L/B/K ALP promoter is placed in front of the chloram-phenicol acetyltransferase gene (pSVALBKcat), activity corresponding to roughly 30% that observed with the SV40 promoter is obtained. This activity is visibly increased by the addition of an SV40 enhancer (pSVALBKcat-LR). Relatively lower levels of chloramphenicol acetyltransferase activity are observed when the L/B/K ALP promoter is situated in the reverse orientation, in the absence or presence of the SV40 enhancer (pSVALBK(r)cat and pSVALBK(r)cat-LR, respectively). When no promoter is present, no chloramphenicol acetyltransferase activity is observed either in the absence or presence of the SV40 enhancer (pSVOAcat(x) andpSVOAcat-LR).
Expression of the L / B / K ALP cDNA-To authenticate the identity of our L/B/K ALP cDNA, we attempted to demonstrate its ability to express active ALP. The original cDNA insert (from the plasmid pS3-1, Weiss et al., 1986) was placed immediately downstream of the SV40 early gene promoter to Differences between genomic and cDNA sequences Nucleotide positions are numbered according to the convention used to describe the cDNA as previously reported (Weiss et al., 1986). The location of the discrepancies and codon changes, if any, are noted. Several errors in the reported cDNA sequence are indicated. In some cases this was confirmed by resequencing portions of the cDNA in the presence of deoxy-7-deazaguanosine to alleviate compression artifacts. Cases where all sequences are unambiguous represent either cloning artifacts or DNA sequence polymorphisms (see text).

t t t~r t t t c t~g~T l i . G A 1 C A l c A G i l~A~C~~C C A C l -74 G C C A~C C A C C C C C~C C C A C C C A C G~C~~~G C A~C~C~G~~~C~G~l~C f f i G l C~l G G G G l G C~C~l G .
. , , . , . form the plasmid pSV2Aalp, and ALP activity was determined after transfection into COS cells. Initial attempts to detect ALP activity above background failed. We thought it possible that the inability to express ALP was due to the mutation discussed above at position 880 of the cDNA. We therefore replaced a segment of the coding sequence in pSV2Aalp containing a guanosine residue at position 880 with a homol-ogous segment from a different cDNA isolate which contained an adenosine residue at this position (as found in the genomic sequence). This modified plasmid (pSV2Aalp') is capable of expressing ALP enzymatic activity: in two separate experiments, COS cells transfected with pSV2Aalp' expressed ALP activity levels over 20-fold above background (data not shown). This result indicates that active L/B/K ALP contains a glutamic acid at amino acid position 218, and not glycine as previously reported (Weiss et al., 1986). Hence, this glutamic acid residue is conserved in human placental, intestinal, and L/B/K ALP and Escherichia coli ALP (see Henthorn et al. (1987) and Weiss et al. (1986)), as well as the liverbone/ kidney-type ALPs found in rat (Thiede et al., 1987), mouse (Terao and Mintz, 1987), and cow . NIH 3T3 cells and CV-1 cells transfected with pSV2Aalp' stain positively for histochemical ALP (not shown), indicating that the transiently expressed enzyme resides on the cell surface. Because of the ease and economy with which ALP can be detected, we are exploring the use of ALP genes as reporters for promoter studies.

DISCUSSION
In higher organisms there are multiple forms of ALP, each exhibiting characteristic patterns of tissue distribution at different stages of development. Complementary DNAs encoding each of the major forms of human ALP have recently been isolated and sequenced (Kam et al., 1985;Millan, 1986;Henthorn et al., 1986;Weiss et al., 1986;Henthorn et al., 1987;Berger et al., 1987). These studies have confirmed the existence of at least three separate ALP genes. Direct examination of these genes provides the next step toward understanding the structure, evolution, and regulation of the various ALPs. This report describes the gene encoding the liver/ bone/kidney form of human ALP. Characterization of the human intestinal and placental ALP genes are presented in Prior to S1 nuclease digestion, radiolabeled probe was incubated with various amounts of human cellular RNAs indicated in the autoradiogram as follows: S, SAOS-2,5 pg; L, liver, 100 pg; K, kidney, 50 pg; F, fibroblast, 50 pg; C, control, yeast tRNA, 100 pg. The RNA concentrations of all samples were normalized to 100 pg with yeast tRNA. The DNA fragments which were protected from S1 nuclease digestion (labeled a, b, and c), and the undigested probe are indicated in the autoradiogram and shown in schematic form. 32P-labeled ends are marked by an asterisk. The primer extension experiment is shown at the bottom of the figure and is described in the text. Radiolabeled primer was annealed to 10 pg of SAOS-2 cellular RNA and incubated with and without AMV reverse transcriptase (RV. Extended products 1, 2, and 3 are labeled in the autoradiogram and indicated as thick lines in the schematic diagram (products 2 and 3 are visible after longer exposure of the autoradiogram (not shown)). The primer is symbolized as an open box. Molecular sizes of extended products were determined by a DNA sequencing ladder (not shown). accompanying articles , Knoll et al., 1988. All Liver/Bone/Kidney-wpe ALPs Are the Product of a Single Gene-Various liver/bone/kidney-type ALPs from different tissues display characteristic differences in thermostability and electrophoretic mobility. Based on a comparison of Southern blots and cloned genomic DNAs, it appears that the L/B/K ALP locus exists as a single copy in the haploid human genome. L/B/K ALP mRNA from SAOS-2 osteosarcoma 12007 s1 PE cells, fibroblasts, liver and kidney has the same major 5' end and is processed to the same size, consistent with the hypothesis that liver/bone/kidney-type ALPs found in different tissues are all products of the same gene and contain the same polypeptide moiety (Harris, 1982;Moss and Whitaker, 1985). The 5' End of the L/B/K ALP Gene-The start of transcription of L/B/K ALP mRNA has been mapped by S1 protection and primer extension. There appears to be a single major start site and at least one minor one (Fig. 3) which are the same in SAOS-2 cells, fibroblasts, kidney, and liver. One S1-protected fragment (b, Fig. 4) is present in all samples including the yeast tRNA control and, therefore, may not represent a true start of transcription. The 3' end of this protected fragment maps to an A/T-rich region centered around nucleotide -225. There is a potential stem-and-loop structure (AG = -29.6 kcal/mol) in this region a t positions -241 to -212. Formation of a cruciform structure between L/ B/K ALP mRNA, and the end-labeled probe would render the loop within the probe (positions -224 to -226) sensitive to S1 nuclease digestion. It is also possible that primer extension product 2, whose 5' end maps to the same region, may be due to premature termination of reverse transcriptase at the stem-and-loop structure within the mRNA.

The LiverlBonelKidnev Alkaline Phswhatase Gene
We have shown that a 672-bp DNA segment situated 5' to the L/B/K ALP gene can direct chloramphenicol acetyltransferase gene transcription in mammalian cells. It is reasonable to assume that this activity is due to the L/B/K ALP gene promotor. However, verification will depend on demonstrating that the sites of transcription initiation in these transfections are in agreement with the 5' ends of authentic L/B/K ALP mRNA. These studies, along with detailed functional mapping of this region, are in progress.
Transcription mapping studies, combined with the demonstration of functional promoter activity, verify that the DNA sequence shown in Fig. 3 contains the L/B/K ALP promoter. This region contains three types of DNA sequence elements worth noting (Fig. 3). First, an A/T-rich sequence is present between nucleotides -228 and -222, about 25 bp upstream of the major transcription start site. Similar sequences, referred to as TATA boxes, are found at analogous positions in many eukaryotic genes and are thought to be important in positioning the site of transcription initiation (Serfling et al., 1985;Dynan and Tjian, 1985). Second, four copies of the consensus sequence for binding of transcription factor SP1, GGGCGG (Kadonaga et al., 1986), are present at nucleotide positions -437, -320, -280, and -268. This sequence exists within the promoters of a diverse group of cellular and viral genes and its functional significance has been established in several cases (Dynan and Tjian, 1985;Dynan et al., 1986). Further studies are necessary to establish any role for SP1 in the transcription of the L/B/K ALP gene.
Third, between nucleotides -740 and -460, there is a cluster of short, purine-rich, imperfect direct repeats of the consensus AGAGACG; two of these repeats are within a larger 15-bp perfect repeat. The significance of these repeats is unknown.
The 5' end of the L/B/K ALP gene contains a region that is extremely G/C-rich and relatively abundant in the dinucleotide sequence CpG: a 650-bp segment of DNA beginning about 250 bp upstream of the major start of transcription (nucleotide position -450 in Fig. 3) and extending into intron l3 has a G/C content of 76.3%, and a CpG to GpC dinucleotide ratio of 79/93 = 0.85. CpG-rich regions, also called CpG' islands, are associated with many cellular genes, especially those that exhibit a widespread tissue distribution, and may function in the regulation of gene expression (Bird, 1986). All other regions of the L/B/K ALP gene which have been sequenced (mostly exons and adjacent intron sequences) contain relatively few CpG dinucleotides (CpG to GpC ratio equal to approximately 0.2).
There is some evidence to suggest that divergent transcription may occur frequently at CpG islands (Lavia et al., 1987). Consistent with this observation, we show that the L/B/K ALP promotor initiates bidirectional transcription of a bacterial chloramphenicol acetyltransferase gene. This phenomenon has been observed to be a property of numerous other promoters (Efrat and Hanahan, 1987;Farnham et al., 1985;Gidoni et al., 1985), but is poorly understood. We have not searched for the existence of cellular RNAs that are transcribed opposite to the coding strand of L/B/K ALP mRNA.
Comparison of the Intestinal, Placental, and L/B/K ALP Genes-The L/B/K ALP gene is about five times larger than the intestinal and placental ALP genes (see the accompanying articles, Henthorn et al. (1988) and Knoll et al. (1988)). However, the protein coding regions of the three ALP genes are interrupted at analogous positions by introns with splice junctions at identical positions within the triplet codons (data not shown), further supporting the hypothesis that members of the human ALP multigene family are the product of a series of duplications of a single ancestral gene (Harris, 1982).
The first exon of the L/B/K ALP gene, occurring in the 5"untranslated region, does not have an analogous counterpart in the intestinal and placental ALP genes. This noncoding exon and the adjacent L/B/K ALP gene promoter may have been acquired as a distinct genetic unit, analogous to functional domains of certain proteins that are encoded by separate exons (Gilbert, 1978). If so, the acquisition of this genetic unit during ALP evolution may have conferred the widespread pattern of tissue expression that is characteristic of L/B/K ALP and in marked contrast to the much more tissue specific expression of intestinal and placental ALP. The existence of a noncoding "regulatory exon" has been proposed in the growth hormone/prolactin gene family (Cooke and Baxter, 1982). It has been suggested that 10-14-bp direct repeats flanking the promoter and first exon of each of these genes may be remnants of a DNA transposition event. Similarly, there is an 11-bp direct repeat flanking the promoter and first exon of the L/B/K ALP gene (shown in Fig. 3).
The sequence of the last 200 bp of this region, not shown in Fig.  3, has been determined on only one strand. The sequence of this region is available upon request. The signal peptide at the amino terminus and the hydrophobic stretch of amino acids at the carboxyl terminus, in exons 2 and 12, respectively, are shown in black. Regions which comprise the active pocket that are conserved in intestinal ALP, placental ALP, and E. coli ALP are shown as follows: small rectangles above the exons indicate conserved units of amino acid sequence which exist as discrete units of secondary structure in E. coli ALP (black for &sheets, white for a-helices); the open circles indicate metal ligands, and the closed circles indidate residues that directly interact with incoming substrate.
The promoter region of the L/B/K ALP gene is quite distinct from the intestinal and placental ALP gene promoters. The L/B/K ALP gene contains a G/C-rich promoter and a CpG island at its 5' end. These features have been observed in many other genes that exhibit a widespread tissue distribution, most notably the "housekeeping" genes (Serfling et al., 1985;Bird, 1986;Dynan, 1986). In contrast, the intestinal and placental ALP genes, which are highly tissue-specific, contain promoters that are much less G/C-rich and are not associated with CpG islands  and Knoll et al. (1988), accompanying articles).
Although the L/B/K ALP gene resembles the housekeeping genes in its promoter structure and widespread tissue distribution, it is not completely accurate to apply this term to the L/B/K ALP gene. Housekeeping implies that the gene encodes an enzyme which performs an essential metabolic function (Dynan, 1986); the function for L/B/K ALP in most tissues is unknown. The only biological role which has been established for ALP is one in bone mineralization, a tissuespecific function. Furthermore, there is a marked variation in L/B/K ALP expression in different cell types. Most notably, L/B/K ALP is expressed at high levels in mineralizing chondrocytes and osteoblasts (McComb et al., 1979). Hence, the L/B/K ALP gene displays characteristics of both housekeeping and tissue-specific genes.
Intron-Exon Organization of the ALP Genes-A diagram of the relationship between L/B/K ALP gene and protein primary structure is shown in Fig. 6. Except for the first exon (discussed above), this diagram also applies to the intestinal and placental ALP genes. The signal peptide along with the first three amino acids of human liver/bone/kidney, intestinal, and placental ALP are encoded by a separate exon. Contrary to the prediction of Kam et al. (1985), the carboxylterminal stretch of hydrophobic amino acids encoded by the ALP cDNAs, which presumably participate in membrane localization, are not represented by a separate exon.
Further analysis of the relationship between ALP protein structure and intron-exon organization is impeded by a lack of structural information on the mammalian ALPs. However, some observations can be made by considering the structure of E. coli ALP, which has been determined to a high degree of resolution by crystallographic studies (Sowadski et al., 1985;Wyckoff et al., 1983). Although the E. coli and human ALP polypeptides demonstrate only about 25% positional identity in an optimal alignment, the amino acid sequences that comprise the active pocket are highly conserved (Kam et al., 1985;Millan, 1986;Weiss et al., 1986). This active pocket consists of several components distributed throughout the polypeptide chain: three functional metal binding sites located near the carboxyl ends of four parallel @-sheet strands and the amino end of an antiparallel P-sheet, a reactive serine linked to the amino end of an a-helix, and an arginine residue thought to stabilize the transition state during catalysis. Localization of these regions within the human ALP genes (Fig. 6) reveals that the enzyme active pocket is formed by structural elements from six separate exons.
Localization of introns within the three-dimensional structure of E. coli ALP, based upon alignment of homologous regions of the bacterial and human polypeptide sequences (Sowadski et al., 1985;Weiss et al., 1986), suggests that the splice junctions coincide with regions at the surface of the enzyme and tend to fall in between stretches of a-helix or psheet units of secondary structure. These trends have been noted in other genes which encode globular proteins (Lonberg and Gilbert, 1985;Craik et al., 1982;Blake, 1985). Confirmation of these observations awaits detailed structural analysis of mammalian ALPs.

The Liver/Bone/Kidney
Alkaline Phosphatase Gene