A Molecular Vision Research Article

Received 19 Jun 1996 | Accepted 28 Oct | Published 4 Nov 1996
Mol. Vis. 2:11, 1996 <http://www.emory.edu/molvis/v2/tink>

Organization, Evolutionary Conservation, Expression and Unusual Alu Density of the Human Gene for Pigment Epithelium-Derived Factor, a Unique Neurotrophic Serpin

Joyce Tombran-Tink^*1, Krzysztof Mazuruk^*, Ignacio R. Rodriguez^*, Daniel Chung^*, Timothy Linker^*, Ella Englander⁺ and Gerald J. Chader^*

^*Laboratory of Retinal Cell and Molecular Biology, National Eye Institute
and the
⁺Laboratory of Molecular Growth Regulation, National Institute of Child Health and Human Development,
National Institutes of Health, Bethesda, MD 20892

¹Corresponding author: jtink@helix.nih.gov

PEDF is a neurotrophic serpin that promotes a neuronal phenotype and augments neuronal cell survival. The isolation, sequence and structural analysis of the human PEDF gene and its promoter along with its evolutionary conservation and expression in human tissues are now described. The gene spans approximately 16 kb and is divided among 8 exons and 7 introns, the junctions of which conform to the AG/GT consensus rule. PEDF appears to fall into the ovalbumin/PAI-2 subgrouping of serpins and is structurally far different from GDN/PN-1, the only other neurotrophic serpin reported to date. The immediate 5'-flanking region is dominated by a dense cluster of Alu repeats in which are embedded several promoter consensus sequences. A CAAT box is present at -43. The putative promoter region is also far different >from that reported for GDN/PN-1. Comparable hybridization signals of 23 kb EcoRI fragments containing the PEDF gene are observed by Southern blot analysis in all primate, mammal and avian species examined; conservation is particularly evident among the primates. Northern blot analysis confirms the presence of the PEDF transcript in a broad range of human fetal and adult tissues including almost all brain areas examined, underscoring differences with GDN/PN-1 which, in the adult brain, is only expressed in glia and a subset of neurons.

Serpins are an important family of proteins that all share a similar, basic globular structure but demonstrate wide functional diversity. Although the primary role of most serpins is in the regulation of proteolytic events (eg. a-l-antitrypsin, a-1-antichymotrypsin), others have evolved alternate functions which include hormone transport (thyroxine-binding globulin, corticosteroid-binding globulin) and blood pressure regulation (angiotensinogen) (1-4). Another well known member of the serpin family, glia-derived nexin/protease nexin-1 (GDN/PN-1) exhibits neuronal differentiative properties (5). It appears to function as a classical serpin antiproteinase by inhibition of the action of thrombin (6).

Pigment epithelium-derived factor (PEDF) was first identified as a biological activity in conditioned medium of cultured human fetal retinal pigment epithelial (RPE) cells that induces neuronal differentiation of cultured human Y-79 retinoblastoma cells (7). Specifically, morphologically undifferentiated Y-79 cells extend long, neurite-like processes in response to addition of the conditioned medium. Expression of neuronal marker molecules occurs coincident with the morphological changes. Subsequently, the factor was isolated and its cDNA cloned (8,9), demonstrating it to be a 50 kDa protein exhibiting sequence homology with members of the serpin gene family. Functionally, PEDF has also been implicated in the cell cycle and aging since it is only expressed at G₀ in cultured human fibroblasts (9) and disappears at the onset of senescence in both fibroblasts (10) and pigment epithelial cells of the neuroretina (11). Moreover, we have recently shown that PEDF markedly enhances the survival of neurons in culture (12).

PEDF behaves as a noninhibitory serpin in that its neurotrophic activity does not require the serpin reactive loop (13). Since GDN/PN-1 acts by inhibiting the action of thrombin (14), it is clear that, although both GDN/PN-1 and PEDF are both neurotrophic serpins, they have markedly different mechanisms of action. PEDF is also of interest since its gene is localized to human chromosome 17p (14), a region to which numerous hereditaty diseases have been mapped. We thus felt it of importance to characterize the human PEDF gene and now report its structural organization. Moreover, we have investigated the evolutionary conservation of PEDF in a number of phylogenetically-related species and determined the expression of the gene in human tissues including major regions of the brain.

MATERIALS AND METHODS

Materials. Restriction enzymes, SuperScript RT and Kanamycin were purchased from GIBCO-BRL (Gaithersburg, MD). Dynabeads, Oligo dT(25) and MPC-E magnetic separator were from Dynal Inc.(Lake Success, NY). Retrotherm^TM RT was obtained from Epicentre Technologies, (Madison,WI). RNAsin and the Wizard PCR Preps DNA purification kit were from Promega (Madison, WI). Midi Qiagen kits for plasmid purification were from Qiagen Inc. (Chatsworth, CA.). Perfect Match DNA Polymerase enhancer, the plasmid vector pBlueScript used for subcloning, l DASH II library, Random Prime It kits and Nuctrap push columns for DNA labeling and purification were from Stratagene (La Jolla, CA). The cosmid library in pWE15 and 5'-AmpliFINDER RACE kit were from Clontech (Palo Alto, CA). Select-D G-50 columns were from 5 Prime - 3 Prime (Boulder, CO).

Total RNA from neural retina and retinal pigment epithelium (RPE) was purified from human tissues obtained from the National Disease Research Interchange (NDRI, Philadelphia, PA) according to the method of Chomczynki and Sacchi (15). [³²P]-a-dATP and [³²P]-a-l-ATP (3000 Ci/mmol) used for labeling and sequencing respectively were from Amersham (Arlington Hts. IL). Superbroth (Bacto-Tryptone) at 12g/l, yeast extract at 24 g/l, K₂HP0₄ at 12.5 g/l, KH₂PO₄ at 3.8 g/l and glycerol 5 ml/l, denaturing solution (0.2 N NaOH, 1.5 M NaCl), neutralizing solution (1 M Tris-Cl, pH 7.0, 1.5 M NaCl), 2OX SSC (3.0 M NaCl, 0.3 mM sodium citrate), 1OX TBE (1 M Tris-borate, 2 mM EDTA, pH 8.3), 5OX TAE (2 M Tris-acetate, 50 mM EDTA, pH 8.0) and 2OX SSPE (3 M NaCl, 0.2 M NaH₂PO₄, 20 mM EDTA, pH 7.4) were from Digene Diagnostics Inc. (Silver Spring, MD). Ampicillin was from Sigma Chemical Co. (St. Louis, >MO). Human multiple tissue RNA blots were purchased from Clonetech.

Oligonucleotide synthesis and polymerase chain reaction (PCR). Oligonucleotide primers were synthesized using a DNA synthesizer, model 392, >from Applied Biosystems Inc. (Foster City, CA). The oligonucleotides were deprotected and used without further purification. 2X PCR mixes of dNTPs, PCR buffer and Taq polymerase were prepared according to specifications of Perkin Elmer (Norwalk, CT). 1 ul of Stratagene's Perfect Match DNA polymerase enhancer was added to samples containing human genomic DNA; 100 ng of each oligo was used in a 100 ul PCR reaction. Template DNA (20 ng for P1 clones, 500 ng for genomic DNA) was denatured followed by a 30 cycle program with 1 min at 95⁰ C, 30 sec annealing (temperature between 55⁰ and 65⁰ C depending on the primers) and an extension at 72⁰ for 1-5 min depending on the length of the product amplified.

Screening of genomic libraries and Southern blot analysis. The human genomic cosmid library was plated at a density of 10,000 colonies per plate on LB plates containing 150 ug/ml ampicillin and 20 ug/ml kanamycin. Nitrocellulose filter lifts of the colonies were denatured, neutralized and then hybridized with a a-[³²P]-labeled PEDF cDNA insert (10⁶ cpm/ml). This resulted in the isolation of a 7 kb cosmid clone (JT1). A lDASH II library was also screened similarly and a 7 kb NotI-NotI (JT6A) fragment isolated. A P1 library was screened by Genome Systems (St. Louis, MO) and two identical P1 clones, p410 and p147, containing the entire PEDF gene and flanking regions were isolated. Oligos used for P1 isolation were jt10UP01/jt10DP01:

5'GGGAGCTGCTTTACCTGTGGATAC3'/5'GGTGTGCAAATGTGTGCGC- CTTAG3' and 1590/1591: 5'GGACGCTGGATTAGAAGGCAGCAA3'/ 5'CCACACCCAGCCTAGTCC3' designed from the 5'-flanking PEDF sequence.

DNA from a P1 clone was subjected to digestion with a number of restriction enzymes. The digested DNA was separated on 1% agarose TBE gel using a CHEF DRII pulse field apparatus (Bio-Rad Labs., Hercules, CA) set for 10 hrs at 200 volts and 2 second pulses. A sizing ladder purchased >from Life Technologies was used for size determination.

DNA purification and sequencing. PCR products were purified using the Wizard PCR Preps DNA purification kit. P1 clones and plasmid subclones were purified using the Qiagen Midi plasmid purification kit. Sequencing of most of the intronic and flanking regions was performed by Lark Sequencing Technologies Inc. using standard sequencing techniques. Clones used were JT1, JT6A, JT8A, JT9, JT14 and JT15. Sequencing of DNA from PCR products of the P1 clones that encompassed intron-exon junctions was performed in our laboratory using an automated DNA fluorescent sequencer (ABI, Foster City, CA) to independently confirm all intron-exon boundaries and junctions across clones and to fill in gaps between subclones. Intron F was amplified from a P1 subclone and sequenced directly without subcloning. Other regions of interest or question were also independently confirmed. The ABI PRISM Dye Deoxy Terminator Cycle Sequence Kit was used following the manufacturer's protocol. Typically, 0.5 pmoles of template and 3 pmoles of primer were used per sequencing reaction. Sequencing products were purified using Select-D G-50 columns, dried and dissolved in 5 ul formamide and 1 ul of 50 mM EDTA prior to heating and sequencing. cDNA synthesis on Dynabeads oligo(dT)₂₅. cDNA was synthesized on Dynabeads as previously described by Rodriguez and Chader (16). Dynabeads (0.5 mg) were washed with 100 ul of 10 mM Tris-Cl buffer, pH 7.0 containing 1 mM EDTA and 1 M KC1. 30 ul of total RNA (1 ug/ul in H₂0) was mixed with 30 ul of the above buffer and the equilibrated Dynabeads heated to 55⁰ C for 2 min. Poly A⁺ RNA was annealed to the beads for 15 min at room temperature and excess RNA removed by binding the beads to an MPC-E magnetic separator. The beads were subsequently suspended in 2.5 ml buffer A (200 mM Tris-Cl buffer, pH 8.3 containing 1.0 M KC1), 2.5 ul buffer B (30 mM MgCl₂ and 15 mM MnCl₂), 20 ul 10 mM dNTPs (2.5 mM each), 1 ul RNAsin, 2 ul SuperScript RT, 5 ul Retrotherm RT (1 Unit/ul) and 16 ul H₂0. The 50 ul reaction mixture was incubated at 40⁰ C for 10 min, then at 65⁰ C for 1 hr. The beads were bound to the MPC-E magnetic separator, washed once with 100 ul 0.2 N NaOH, once with 1OX SSPE and twice in 1X TE. The cDNA-bound beads were then suspended in 100 ul of 1X TE.

5' Rapid Amplification of cDNA Ends (RACE). 5'- RACE was performed using a modified method of the 5'- AmpliFINDER RACE kit (17). The AmpliFINDER anchor primer was ligated to the 3' end of Dynabeads-immobilized RPE cDNA. The anchor primer was used in conjunction with PEDF primer #2744 (5'CCTCCTCCACCAGCGCCCCT3') and 2 ul of anchor-ligated human RPE-Dynabeads cDNA for PCR amplification of the 5' end. Amplification was performed for 30 cycles as described above. The PCR product was cloned into pGEM-T (Promega). Several clones were selected for sequencing.

Gene expression: Northern blots containing 2 ug poly A⁺ RNA >from a number of human tissues, including brain, endocrine and immune-related tissues, were hybridized similiarly as for Southern analysis and blots exposed overnight at -70^o C prior to autoradiography.

RESULTS

5'-RACE of PEDF: After 5'-RACE amplification of the PEDFcDNA, one main product was observed at approximately 230 bp (Fig. not shown). This fragment was cloned and found to extend the 5' end of PEDF by 20 bp (see Fig. 2) from the previously published cDNA sequence (9).

Sequence and structure of the human PEDF gene: The full sequence is given in Genbank, accession # U29953. Comparison of the restriction pattern between a P1 clone and genomic DNA demonstrates that the entire PEDF gene is contained in the clone (Fig. 1 Top). A scale map of the human PEDF gene is given in Fig. 1 (Bottom). The gene spans approximately 16 kb and contains 8 exons; the positions of several of the restriction sites are indicated. All splice-site junctions and junctions across clones were confirmed by two independent sequencings. The intron-exon junctions and their flanking sequences are shown in Table 1. All of the junctions obey the AG/GT rule.

Table 1. Intron-exon junctions and flanking sequences

No.	bp.	EXON	Donor splice-site	.....	Acceptor splice-site	Exon	No.
			PROMOTER	.....	..aaaggagta	GCTGTAATC	1
1	128	TATCCACAG	gtaaagtag..	4793 bp	..ttcttgcag	GCCCCAGGA	2
2	92	CCGGAGGAG	gtcagtagg..	2863 bp	..tcctgccag	GGCTCCCCA	3
3	199	TCTCGCTGG	gtgagtcgc..	980 bp	..ctctggcag	GAGCGGACG	4
4	156	TTGAGAAGA	gtgagtcga..	688 bp	..tcttctcag	AGCTGCGCA	5
5	204	ACTTCAAAGG	gtgagcgcg..	2982 bp	..tctttccag	GGCAGTGGG	6
6	143	AGCTGCAAG	gtctgtggg..	1339 bp	..ttgtctcag	ATTGCCCAG	7
7	211	AGGAGATGA	gtatgtctg..	444 bp	..tctctacag	AGCTGCAAT	8
8	377	TTTATCCCT	aacttctgt

5'-Flanking region: Few of the conventional promoter elements (eg. TATA box, sequences, etc.) are found within the putative promoter region although a putative CAAT box is present at -43 (Fig. 2). Within the immediate adjacent 150 bps, the sequence GTTAAAGTTAAC at position -62 resembles the binding consensus GTTAATNATTAAC for HNF-1, the hepatocyte nuclear factor (18). At -113, the sequence GTGCAAAT deviates by one base from the consensus sequence ATGCAAAT for the octomer (Oct) family of transcription factors (19). The sequence AGGAAG/A for PEA3, the polyomavirus enhancer activator-3, (19), is exact and present in tandem at -122, -129 and again at -141. Upstream at -654, the sequence GTGGTTATG is within the consensus sequence GTGGT/AT/AT/AG recognized by the C/EBP (CAAT-enhancer binding protein) family of transcription factors (19). Other than these, the 5 kb 5'-flanking region is dominated by a dense cluster of Alu repeats consisting of 8 complete and 3 partial Alu elements. A partial inverted Alu element (left monomer) is located relatively near to the transcriptional start site (-154 to -284) of the gene; this is closely followed by a complete inverted Alu repeat (-286 to -591) and a third Alu in the sense orientation at position -790 to about -1010. Interestingly, embedded within these Alus are sequences similar or identical to known promoter elements. The sequence TCAGGTGATGCACCTGC at -202 is very similar to the artificial palindromic sequence (TREp) TCAGGTCATGACCTGA (19). The sequence AGGTGATGCACCT at -204 contained within TREp is also similar to the developmentally-regulated retinoic acid receptor (RAR) motif whose consensus sequence is AGGTCATGACCT (19). Three possible CACCC boxes are present, one at -845 and two in the reverse orientation at -826 and -905, within the third Alu repeat along with a consensus SP-1 site (GGCGGG) at its distal border (-1030). Finally, two possible estrogen receptor-dependent transcriptional enhancers are present within the Alu repeats (see Discussion).

Evolutionary conservation. The evolutionary conservation of PEDF throughout the animal kingdom as well as among a number of phylogenetically-related higher species was examined by Southern blotting (Fig. 3). Strong homology is observed in human, mouse and chicken (left panel). No hybridization signals were detected in lower species. All the mammalian species examined show strong conservation of PEDF but with varied restriction patterns (middle panel). A large EcoRI fragment of approximately 23 kb is present in all primate species examined (right panel). The common chimpanzee, however, has an additional, strongly-hybridizing fragment at approximately 9 kb.

Gene Expression of PEDF: The 1.5 kb PEDF transcript was seen in almost all 44 human fetal and adult tissues examined (Fig. 4). In particular, intense hybridization signal is seen in both fetal and adult liver, adult testis, ovaries, placenta and pancreas. Of the 16 brain regions examined, all except the putamen and cerebellum exhibited hybridization signals (Fig. 5). An actin probe was used as a positive control to determine RNA loading efficiency in each lane (data not shown).

DISCUSSION

Serpins are an ancient family of proteins that are widely distributed in plants, viruses and animals. They exhibit a broad range of functions and are involved in numerous disease processes (20,21). Phylogenetically, Marshall (20) has divided serpin proteins into 3 main groupings; I: ovalbumin/PAI-2 type; II: a-1-antitrypsin antichymotrypsin (AAC) type and III: angiotensinogen/GDN type. The gene structures of several of these proteins have been described and range from 5 exons in AAC (22) to 10 exons in a₂-plasmin inhibitor (23) and expand over areas from about 9 kb to 26 kb. The human PEDF gene is composed of 8 exons interrupted by 7 introns and spans approximately 16 kb. The simplicity of its restriction pattern (Fig. 1 Top) indicates that there is probably only one gene for PEDF. The genomic structure suggests that PEDF belongs to the ovalbumin family of serpins in which ovalbumin, plasminogen activator inhibitor (PAI-2) and other "ov-serpins" have similar gene organizations consisting of 8 exons and 7 introns (24). Interestingly, this organization is distinct from that reported for the only other known neurotrophic serpin, GDN/PN-1; its gene has been reported to be separated by 8 intronic sequences, most closely resembling PAI-1 (25). Characteristics of the GDN/PN-1 protein also place it in a branch of the serpin family tree separate >from the ov-serpins (20).

Although, the positions and sizes of the introns between serpins vary widely, the sizes of the introns in PEDF as well as the entire gene size appear to be relatively similar to those of PAI-2. Both genes are approximately 16 kb in length and have introns of about the same size. Both genes comply with the AG/GT consensus splice-site rule as do those of other serpins. Consistent with gene organization features of most serpins, the first exon of the PEDF gene is non-coding and the 3' UTR is not interrupted by a splice junction (26).

The 5'-flanking region of the PEDF gene is interesting since it exhibits very few of the sequence motifs recognized by known transcriptional factors but rather is exceptionally rich in Alu elements. PEA3, is one of the few complete consensus motifs present outside the Alu sequences; it occurs three times in close juxtaposition within residues -120 to -150. It is recognized by members of the ETS family of transcription factors and is thought to be involved in oncogene interactions and cellular growth responses (27). Besides a sequence recognized by the C/EBP family of transcription factors at -654, most other sequences such as those that might recognize HNF-1 and Oct proteins are not complete consensus motifs. Functionality of these elements and of the PEA3 sites remains to be determined in future experiments. Certainly, the most striking and unique feature of the first 5 kb of 5'-flanking region is the dense cluster of Alu elements which comprise 70% of the most proximal 1 kb upstream region. Although single (28) or multiple (29) Alu repeats have been detected in intronic regions of other serpin genes, this is a feature that, to our knowledge, has not been observed in other serpin 5'-flanking regions to date. These repeats belong to one of the two major Alu subfamilies, AluS (30), a family thought to have transposed ~45 million years ago. Embedded within the most proximal Alu repeat in the putative promoter region is a close consensus sequence for TREp/RAR (-202) and, in a more distal Alu, an exact consensus sequence for SP-1 (-1030). Although such sequences are generally thought to be nonfunctional, recent studies have demonstrated the functionality of a retinoic acid response element provided by an Alu sequence (31). This is a situation similar to the presence of the TREp/RAR sequence in the 5'flanking region of the PEDF gene. Moreover, Norris et al. (32) have recently identified a new Alu subclass (consensus sequence GGTCA(n)₃ TGGTC(n)₉TGACC) which can function as an estrogen receptor-dependent transcriptional enhancer. A sequence identical to this consensus is present in the PEDF upstream Alu repeat and a sequence different by only one nucleotide is present in the proximal Alu repeat, i.e., within 200 bps of the translational start site. As with the more classical elements discussed above, it will be interesting to determine if any of the elements embedded in the Alu repeats are functional. Finally, with recent evidence that Alu repeats may affect nucleosome formation over neighboring DNA sequences (33) and, in this way, influence gene expression, it will be important to analyze the corresponding chromatin structure with respect to normal/abnormal PEDF expression.

The promoter region of the only other neurotrophic serpin reported to date, GDN/PN-1, has been cloned (34) and is markedly different from that of PEDF. The GDN/PN-1 promoter region contains a typical TATA box, 5 SP-1 consensus sequences, 4 MyoD1 sites and a number of other putative binding sites for other transcription factors none of which are found in PEDF. In addition, the Alu cluster present in PEDF is not present in GDN/PN-1 (31), highlighting the potentially different control routes of expression of the two genes. We have previously reported that PEDF is secreted by human fetal RPE cells (7,8,11) and Pignolo et al. (10) have reported PEDF (EPC-1) to be synthesized by WI-38 fibroblast cells. Since then, Pignolo et al. (35) have reported PEDF (EPC-1) mRNA to be present in a number of cultured human cell lines. We now give evidence for the presence of PEDF transcripts in most human tissues. Our Southern blot analyses not only indicate that PEDF expression is more widespread than previously suspected but that the gene is well conserved throughout the vertebrates. For example, a 23 kb fragment is found in almost all primates; only chimpanzee exhibits an additional 9 kb band. Functionally, loss of PEDF expression has been linked to the onset of senescence in two human cell types (10,11). The phylogenetic conservation of PEDF may indicate that, at least among the vertebrates, PEDF does perform a generally important cell survival function in vivo. This notion is supported by our recent findings that PEDF is a potent neuron-survival factor for cerebellar granule cells in culture (12). The presence of PEDF transcripts in almost all regions in the adult brain might indicate that PEDF could be of particular importance in brain as an autocrine/paracrine neurotrophic factor. The fact that PEDF transcripts are very poorly expressed in cerebellum in light of our finding that PEDF exerts such a strong neuronal survival effect on cultured cerebellar granule cells although, as a secreted product, PEDF could very well be pruduced in another brain area and be transported to the cerebellum for action. Our mRNA studies also demonstrate that PEDF has a markedly different pattern of expression than does GDN/PN-1 which has a much more limited distribution of its mRNA in the adult animal (14,36). In the developing brain, for example, GDN/PN-1 is only transiently expressed in most neurons.

Availability of the genomic clones and sequence will make it possible to better study PEDF function in human tissues as well as PEDF's possible involvement as a candidate gene in human diseases. The PEDF gene maps to human chromosome 17p13 (37), a region containing many disease genes such as medulloblastoma (38) ovarian cancer (39) and inherited retinal degeneration (40). Of particular interest is the tight linkage shown between the PEDF gene and an autosomal dominant retinitis pigmentosa (RP13) that maps to 17p13.3 (41). Finally, the unusual cluster of Alu repeats in the 5'-flanking region of the gene and their embedded putative promoter/enhancer elements may mediate novel functional control of normal or defective gene expression.

ACKNOWLEDGEMENTS

Dr. Mazuruk was supported as the John W. Kluge Research Fellow of the Foundation Fighting Blindness.

REFERENCES

1. Carell, R.W. and Travis, J. (1985) Trends Biochem. Sci. 10, 2023-2026.

2. Monard, D. (1988) Cell-derived proteases and protease inhibitors as regulators of neurite outgrowth. Trends Neurosci. 11, 541-544.

3. Stein, P.E., Tewkesbury, D.A. and Carre, R.W. (1989) Ovalbumin and angiotensinogen lack serpin S-R conformational change. Biochem. J.262, 103-107.

4. Patston, P.A., Rodi, N., Schifferli, J.A., Bischoff, R., Courtney, M. and Schapira, M. (1990) Reactivity of alpha 1-antitrypsin mutants against proteolytic enzymes of the kallikrein-kinin, complement, and fibrinolytic systems. J. Biol. Chem. 265, 10786-10791.

5. Bock, S.C. (1991) in Hemostasis and Thrombosis (Hoyer, L.W. and Drohan, W.N. Eds.) pp. 25-45, Plenum Press, New York.

6. Evans, D.L., McGrogan, M., Scott, R.W. and Carrell, R.W. (1991) Protease specificity and heparin binding and activation of recombinant protease nexin I. J. BiolChem. 266, 22307-22312.

7. Tombran-Tink, J. and Johnson, L.V. (1989) Neuronal differentiation of retinoblastoma cells induced by medium conditioned by human RPE cells. Invest. Ophthalmol. Vis. Sci. 30, 1700-1707.

8. Tombran-Tink, J., Chader, G.J. and Johnson, L.V. (1991) PEDF: a pigment epithelium-derived factor with potent neuronal differentiative activity Exp.Eye.Res.53, 411-414.

9. Steele, F.R., Chader, G.J., Johnson, L.V. and Tombran-Tink, J.(1993) Pigment epithelium-derived factor: neurotrophic activity and identification as a member of the serine protease inhibitor gene family. Proc. Natl. Acad. Sci. U.S.A. 90, 1526-1530.

10. Pignolo, R.J., Rotenberg, M.O. and Cristofalo, V.J. (1995) Analysis of EPC-1 growth state-dependent expression, specificity, and conservation of related sequences. J.Cell Physiol. 162, 110-118.

11. Tombran-Tink, J., Shivaram, S.M., Chader, G.J., Johnson, L.V. and Bok, D. (1995) Expression, secretion, and age-related downregulation of pigment epithelium-derived factor, a serpin with neurotrophic activity. J. Neurosci. 15, 4992-5003.

12. Taniwaki, T., Becerra, S.P., Chader, G.J. and Schwartz, J.P (1995) Pigment epithelium-derived factor is a survival factor for cerebellar granule cells in culture. J. Neurochem. 64, 2509-2517.

13. Becerra, S.P., Sagasti, A., Spinella, P. and Notario, V. (1995) Pigment epithelium-derived factor behaves like a noninhibitory serpin. Neurotrophic activity does not require the serpin reactive loop. J. Biol. Chem. 270, 25992-25999.

14. Reinhard, E., Suidan, H.S., Pavlik, A. and Monard, D. (1994) Glia-derived nexin/protease nexin-1 is expressed by a subset of neurons in the rat brain. J. Neurosci. Res. 37, 256-270.

15. Chomczynski, P. and Sacchi N. (1987) Nucleic Acid Res 17, 2919-2932.

16. Rodriguez, I.R. and Chader G.J. (1992) A novel method for the isolation of tissue-specific genes.Nucl. Acid Res. 20,3528.

17. Rodriguez, I.R., Mazuruk, K., Schoen, T.J. and Chader, G.J. (1994) Structural analysis of the human hydroxyindole-O-methyltransferase gene. Presence of two distinct promoters. J. Biol. Chem. 269, 31969-31977.

18. Frain, M., Swart, G., Monaci, P., Nicosia, A., Stampfli, S. Frank, R. and Cortese, R. (1989) The liver-specific transcription factor LF-B1 contains a highly diverged homeobox DNA binding domain.Cell 59, 145-157.

19. Faisst, S. and Meyer, S. (1992) Compilation of vertebrate-encoded transcription factors. Nucleic Acid Res. 20, 3-26.

20. Marshall, C.J. (1993) Evolutionary relationships among the serpins.Phil. Trans. Soc. Lond. B 342, 101-119.

21. Potempa J., Korzus, E. and Travis, J. (1994) The serpin superfamily of proteinase inhibitors: structure, function, and regulation. J. Biol. Chem. 269, 15957-15960.

22. Bao, J.J., Sifers, R., Kidd, V., Ledley, F and Woo, S. (1987) Molecular evolution of serpins: homologous structure of the human alpha 1-antichymotrypsin and alpha 1-antitrypsin genes Biochemistry 26, 7755-7759.

23. Hirosawa, S., Nakamura, Y., Miura, O., Sum, Y. amd Aoki, N. (1988) Human genes for complement components C1r and C1s in a close tail-to-tail arrangement.Proc. Natl. Acad. Sci. U.S.A. 85, 6836-6840.

24. Samia, J., Alexander, S., Horton, K., Auron, P., Byers, M., Shows, T.and Webb, A. (1990). Chromosomal organization and localization of the human urokinase inhibitor gene: perfect structural conservation with ovalbumin. Genomics 6, 159-167

25. McGrogan, M., Kennedy, J., Golini, F., Ashton, N., Dunn, F., Bell, K., Tata, E., Scott, R and Simonsen, C. (1990) in Serine Proteases and their Serpin Inhibitors in the Nervous System. (B. Festoff, Ed.) Plenum Press, New York, NY, pp. 147-1161.

26. Prochownik, E.V., Bock, S.C.., Orkin, S.H. (1985) Intron structure of the human antithrombin III gene differs from that of other members of the serine protease inhibitor superfamily. J. Biol. Chem. 260, 9608-9612.

27. Macleod, K., Leprince, D. and Stehelin, D. (1992) The ets gene family. TIBS 17, 251-256.

28. Long, G., Chandra, T., Woo., S., Davie, E. and Kurachi, K. (1984) Complete sequence of the cDNA for human alpha 1-antitrypsin and the gene for the S variant. Biochemistry 23, 4828-4837.

29. Carter, P., Duponchel, C., Tosi, M. and Fothergill, J. (1991) Complete nucleotide sequence of the gene for human C1 inhibitor with an unusually high density of Alu elements. Europ. J. Biochem. 197, 301-308.

30. Jurka, J. and Miloslavljevic, A. (1991) Reconstruction and analysis of human Alu genes. J. Mol. Evol. 32,105-121.

31. Vansant, G. and Reynolds, W. (1995) The consensus sequence of a major Alu subfamily contains a functional retinoic acid response element. Proc. Natl. Acad. Sci.92, 8229-8233.

32. Norris, J., Fan, D., Aleman, C., Marks, J., Futrea, P., Wiseman, R., Iglehard, J., Deininger, P. and McDonnell, D. (1995) Identification of a new subclass of Alu DNA repeats which can function as estrogen receptor-dependent transcriptional enhancers. J. Biol. Chem.270, 22777-22782.

33. Englander, E. and Howard, B.H. (1995) Nucleosome positioning by human Alu elements in chromatin. J. Biol. Chem. 270, 10091-10096.

34. Erno, H. and Monard, D. (1993) Molecular organization of the rat glia-derived nexin/protease nexin-1 promoter. Gene Expression 3, 163-174.

35. Pignolo, R., Rotenberg, M. and Cristofalo, V. 1995. Analysis of EPC-1 growth state-dependent expression, specificity, and conservation of related sequences. J. Cell. Physiol. 162, 110-118.

36. Mansuy, I.M., van der Putten, H., Schmid, P., Meins, M., Botter, F. and Monard, D. (1993) Variable and multiple expression of Protease Nexin-1 during mouse organogenesis and nervous system development. Development 119, 1119-1134.

37. Tombran-Tink, J., Pawar, H., Swaroop, A., Rodriguez, I. and Chader, G.J. (1994). Localization of the gene for pigment epithelium-derived factor (PEDF) to chromosome 17p13.1 and expression in cultured human retinoblastoma cells. Genomics 19, 266-272.

38. Biegel, J., Burk, C., Barr, F. and Emanuel, B. (1992) Evidence for a 17p tumor related locus distinct from p53 in pediatric primitive neuroectodermal tumors. Cancer Res.52, 3391-3395.

39. Phillips, N.J., Ziegler, M.R., Fair, K.L., Steinbrueck, T, Xynos, F.P. and Donis-Keller, H. (1996) Allelic deletion on chromosome 17p13.3 in early ovarian cancer. Cancer Res. 56, 606-611.

40. Greenberg, J., Goliath, R., Beighton, P. and Ramesar, R. (1994) A new locus for autosomal dominant retinitis pigmentosa on the short arm of chromosome 17.Hum. Mol. Genet. 3, 915-918.

41. Goliath, R., Tombran-Tink, J., Rodriguez, I., Chader, G.J., Ramesar, R. and Greenberg, J. (1996) The gene for PEDF, a retinal growth factor, is a prime candidate for retinitis pigmentosa and is tightly linked to the RP13 locus on chromosome 17p13.3. Mol. Vision 2: 5, 1996.