© 2001 Nature Publishing Group http://genetics.nature.com
letter
The gene defective in leukocyte adhesion deficiency II encodes a putative GDP-fucose transporter Kerstin Lühn1*, Martin K. Wild1*, Matthias Eckhardt2, Rita Gerardy-Schahn3 & Dietmar Vestweber1
© 2001 Nature Publishing Group http://genetics.nature.com
*These authors contributed equally to this work.
Leukocyte adhesion deficiency II (LAD II) is characterized by the lack of fucosylated glycoconjugates, including selectin ligands, causing immunodeficiency and severe mental and growth retardation1–3. No deficiency in fucosyltransferase activities2,4 or in the activities of enzymes involved in GDP-fucose biosynthesis5 has been found. Instead, the transport of GDP-fucose into isolated Golgi vesicles of LAD II cells appeared to be reduced6. To identify the gene mutated in LAD II, we cloned 12 cDNAs from Caenorhabditis elegans, encoding multi-spanning transmembrane proteins with homology to known nucleotide sugar transporters, and transfected them into fibroblasts from an LAD II patient. One of these clones re-established expression of fucosylated glycoconjugates with high efficiency and allowed us to identify a human homolog with 55% identity, which also directed re-expression of fucosylated glycoconjugates. Both proteins were localized to the Golgi. The corresponding endogenous protein in LAD II cells had an R147C amino acid change in the conserved fourth transmembrane region. Overexpression of this mutant protein in cells from a patient with LAD II
did not rescue fucosylation, demonstrating that the point mutation affected the activity of the protein. Thus, we have identified the first putative GDP-fucose transporter, which has been highly conserved throughout evolution. A point mutation in its gene is responsible for the disease in this patient with LAD II.
Selectins are a family of cell-adhesion molecules that initiate the binding of leukocytes to endothelium and thereby control lymphocyte homing and leukocyte entry into inflamed tissue3. Fucose, α1,3-linked to sialyl-N-acetyllactosamine, is a central binding determinant of all known selectin ligands7. The general defect in fucosylation in patients with LAD II leads to the loss of all known selectin ligands, causing recurrent episodes of infections and persistent leukocytosis1,2. The additional defects in mental and psychomotor development point to the importance of further unknown fucosylated glycoconjugates for the development of the neuronal system. To identify the genetic defect in LAD II, we used a complementation approach in fibroblasts from a patient with LAD II (refs. 4,8). These fibroblasts show strongly reduced expression of d g a fucosylated glycans, which was detected in normal fibroblasts by the Aleuria aurantia lectin (AAL; Fig. 1a–c). AAL staining was completely blocked with 75 mM fucose (data not shown). Our approach was based on the hypothesis that e b h the strong decrease of fucosylation in these cells would be based on the reduced ability of the Golgi to import GDPfucose from the cytosol6. Because no GDP-fucose transporter had been cloned in any species, we used the sequences of the known nucleotide sugar c i f transporters (NSTs) to search the C. elegans genome for sequences encoding all possible homologous, multi-spanning transmembrane proteins9. Based on the fact that C. elegans expresses fucosylated glycoconjugates10,11, we assumed Fig. 1 Rescue of fucosylation in LAD II fibroblasts by complementation with a C. elegans gene. Fucose-specific AAL staining of that one of these sequences wild-type fibroblasts (a) is much stronger than that of LAD II fibroblasts (b); the latter are also shown in phase contrast (c). LAD II cells were co-transfected with a GFP fusion protein as transfection marker and full-length cDNAs for C. elegans ORF-7 (d–f) or might encode a GDP-fucose ORF-8 (g–i). d,g, GFP-label; e,h, AAL staining of the corresponding cells; f,i, phase contrast. Bar, 10 µm. transporter.
1Institut für Zellbiologie, ZMBE, Universität Münster, Münster, and Max-Planck-Institut für Klinische and Physiologische Forschung, Bad Nauheim, Germany. 2Institut für Physiologische Chemie, Universität Bonn, Bonn, Germany. 3Institut für Medizinische Mikrobiologie, Medizinische Hochschule Hannover, Hannover, Germany. Correspondence should be addressed to D.V. (e-mail:
[email protected]).
nature genetics • volume 28 • may 2001
69
letter
© 2001 Nature Publishing Group http://genetics.nature.com
© 2001 Nature Publishing Group http://genetics.nature.com
a
Fig. 2 Sequence comparison of ORF-7 homologs in different species, dendrogram of ORF-7, ORF-11 and other NST family members, and predicted membrane topology of ORF-7. a, Alignment of the protein sequences of ORF-7 from C. elegans with its nearest homologs in Homo sapiens and Drosophila melanogaster. The human protein exists as two splicing variants. The shorter version corresponds to the protein FLJ11320, starting after the arrow; the longer version is described in the accompanying report13. b, Dendrogram showing the predicted relationship of known NSTs with the ORF-7 homologs from C. elegans, H. sapiens and Drosophila, and ORF-11 from C. elegans. c, Predicted membrane topology for ORF-7 in C. elegans and H. sapiens as concluded from structure predictions and comparison with known NSTs.
extending the amino terminus of our sequence for 13 amino acids, is described in the accompanying report13 and represents a splicing variant (Web Figs. A and B). b Also highly homologous is a sequence from Drosophila melanogaster (gene product CG9620), which shows 44% amino acid sequence identity (Fig. 2a,b). In accordance with its putative function as a GDPfucose transporter, the polypeptide of ORF-7 is predicted to span the membrane several times. We propose 10 transmembrane domains for ORF-7 and its human homolog based on two different structure-prediction algorithms in combination with sequence comparison of ORF-7 with the two known GDP-mannose transc porters (Fig. 2c). Ten transmembrane domains were also predicted for the CMP-sialic acid transporter14. Because fucosylation occurs in the Golgi, a function of ORF-7 as a GDP-fucose transporter required that this protein be specifically targeted to the Golgi. We found that a fusion protein containing C. elegans ORF-7 fused at its carboxy terminus to GFP was targeted to the Golgi We found 16 sequences with amino acid sequence identity above in transfected COS-7 cells, as determined by its co-localization 20%, of which 12 were isolated as full-length cDNAs using a C. ele- with the Golgi marker golgin-97 (Fig. 3a–c). We conclude that gans cDNA library. Each clone was transiently transfected into LAD II fibroblasts and the transfected cells were analyzed by immunoflud a orescence for staining with AAL. To detect transfected cells, we cotransfected the C. elegans cDNAs with a GFP fusion protein. One of the cDNAs, ORF-7, induced bright AAL staining in approximately 40–50% of the GFP-positive cells (Fig. 1d–f). A second clone, ORF11, gave rise to approximately 10% AAL-positive cells in the GFPpositive population (data not shown). All other clones were negative; ORF-8 is shown as an example (Fig. 1g–i). The difference observed with ORF-7 and -11 was not due to differences in the efficiency of transfection, as it was reproducible in independent transfection experiments using identical amounts of DNA. e b Analysis of the sequence of both C. elegans cDNAs revealed a protein of 363 amino acids (calculated molecular weight (MW) of 40.4 kD) for ORF-7 (Fig. 2a) and a protein of 324 amino acids (calculated MW of 35.8 kD) for ORF-11. The dendrogram illustrates the relationships between the newly identified genes (ORF7, ORF-11) and known NSTs (Fig. 2b). Searching for related sequences in the gene database revealed that ORF-11 is most closely related (27% amino acid sequence identity) to the UDPN-acetylglucosamine (UDP-GlcNAc) transporter from Kluyveromyces lactis12. In contrast to ORF-11, none of the known c f NSTs in any species was as closely related to ORF-7 as the human protein FLJ11320 (of unknown function), showing 55% amino acid sequence identity (Fig. 2a,b). An isoform of this protein, Fig. 3 Subcellular localization of C. elegans and human ORF-7. Fusion proteins consisting of full-length ORF-7 from either C. elegans (a–c) or H. sapiens (d–f) fused at their C termini to GFP were transfected into COS-7 cells. Cells were analysed for GFP staining (a,d) and labeled with antibodies specific for golgin97 (b,e). c,f, Merged pictures. Bar, 10 µm.
70
nature genetics • volume 28 • may 2001
© 2001 Nature Publishing Group http://genetics.nature.com
© 2001 Nature Publishing Group http://genetics.nature.com
Fig. 4 Rescue of fucosylation in LAD II fibroblasts by complementation with the human ortholog of ORF-7. LAD II cells were co-transfected with a GFP fusion protein as transfection marker and full-length cDNA for either human ORF-7 (a–c) or mutated human ORF-7 cloned from LAD II cells (d–f). GFP label is shown in (a) and (d), AAL staining of the corresponding cells, in (b) and (e), and phase contrast, in (c) and (f). Bar, 10 µm.
ORF-7 encodes a putative GDP-fucose transporter located in the Golgi and that expression of this transporter in LAD II cells rescues the fucosylation defect. We found that a fusion protein containing human ORF-7 and GFP was targeted to the Golgi of COS-7 cells (Fig. 3d–f). Transient co-transfection of LAD II cells with full-length human ORF-7 and a GFP marker re-established fucosylation in all of the GFP-positive cells (Fig. 4a–c), providing strong evidence that the human ortholog is a GDP-fucose transporter. To detect the ORF-7 mutation of the LAD II patient, we isolated the cDNA by RT–PCR from fibroblasts of the patient (identical with patient A.C. in the accompanying report13). From seven independent RT–PCR reactions (based on four RT reactions), we obtained identical sequences that all displayed a point mutation (C439T) causing replacement of arginine 147 by a cysteine (R147C). According to the predicted topology, the mutation is located in the fourth transmembrane domain, a region highly conserved between C. elegans ORF-7 and its human and Drosophila homologs (Fig. 5). An arginine is conserved at an analogous position in the fourth transmembrane region of ORF11, as well as in the K. lactis UDP-GlcNAc transporter and the C. elegans SQV-7 protein (Fig. 5), which was reported to translocate UDP-glucuronic acid, UDP-N-acetylgalactosamine (UDPGalNAc) and UDP-galactose in vitro15. Overexpression of the mutated human ORF-7 gene in LAD II cells did not result in the rescue of the LAD II phenotype (Fig. 4d–f), indicating that the identified mutation inactivates the protein and is responsible for the fucosylation defect in LAD II cells. We and others have found that the lack of fucosylation in LAD II fibroblasts can be corrected by adding fucose to the culture medium4,16. Likewise, treatment of a LAD II patient with oral fucose can restore selectin ligands and correct the immunodeficiency4,17. These results suggest that the LAD II Golgi contains a low GDP-fucose import activity and that increased cytosolic levels of GDP-fucose synthesized from external fucose drive amounts of GDP-fucose into the Golgi sufficient to restore fucosylation. This suggests that either the mutant transporter is not completely inactive or there is yet another, low-efficient mechanism available. Whether ORF-11, the second gene we detected in C. elegans, is related to such an activity needs further investigation.
a
d
b
e
c
f
letter
Based on the available information on the complete C. elegans genome, we searched for a C. elegans gene that would rescue the human LAD II fucosylation defect (also called congenital disorder of glycosylation (CDG)-IIc; ref. 13). We have found such a gene that encodes a protein with the membrane topology and subcellular localization of a Golgi NST. Its highly conserved sequence allowed the identification of the human homolog, which rescued the fucosylation defect even more efficiently. The corresponding endogenous gene in LAD II cells is mutated in a highly conserved region, rendering the protein inactive. We conclude that a mutation in this gene represents the defect in this LAD II patient and provide evidence that the corresponding protein is a GDP-fucose transporter, the first identified so far. It is still unknown why the lack of fucosylated glycans affects neurodevelopment. Understanding the evolutionarily conserved mechanism of GDP-fucose import into the Golgi will facilitate the generation of gene-deficient animals devoid of any fucosylation, and may allow us to gain insights into why fucose is needed for higher neurological functions. Note: supplementary information is available on the Nature Genetics web site (http://genetics.nature.com/supplementary_info/).
Methods
Fig. 5 Comparison of a conserved region in the fourth transmembrane domain of ORF-7, which bears the point mutation of LAD II. The mutation replaces a highly conserved arginine (R147 in human ORF-7) by a cysteine. The arginine is conserved in the predicted fourth transmembrane domain of homologs of ORF-7, ORF-11 as well as in the UDP-GlcNAc transporter of K. lactis and in the C. elegans gene sqv-7 and its human homolog. The product of C. elegans sqv-7 is able to transport UDP-glucuronic acid, UDP-GalNAc and UDP-galactose in vitro when expressed in S. cerevisiae15.
nature genetics • volume 28 • may 2001
Sequence analysis. We performed BLAST searches for putative NSTs in C. elegans based on the protein sequences of all known NSTs (BLAST version 2.1, BLOSUM62 matrix). Sequences with identities of at least 20% were selected and newly identified sequences were used in a second round of BLAST searches. We identified 16 putative NST sequences in the C. elegans genome, numbered each ORF from 1 to 16 and cloned 12 of them (except ORF-2, -4, -12 and -13). In several cases, the coding sequences of these genes as they had been predicted by the GENEFINDER program18 differed from the ORF products that we cloned from the cDNA library. The corrected sequences of ORF-1, ORF-7, ORF-8 and ORF-14 were deposited in GenBank. To predict the membrane topology of C. elegans ORF-7, we used the PredictProtein algorithm19 (http://www.embl-heidelberg.de/predictprotein). The PredictProtein algorithm predicted nine transmembrane domains. The second hydrophobic peak from the C terminus that was visible in a Kyte-Doolittle hydrophobicity plot20 was not classified as a transmembrane segment; however, the same program predicted 10 transmembrane
71
letter
© 2001 Nature Publishing Group http://genetics.nature.com
© 2001 Nature Publishing Group http://genetics.nature.com
domains for the related Leishmania and yeast GDP-mannose transporters. Moreover, an even number of transmembrane domains was suggested for all Golgi NSTs studied so far, as N and C termini of these proteins face the cytosolic side14,15,21,22. Finally, an additional algorithm (TMHMM, available from http://www.cbs.dtu.dk/services/TMHMM-1.0) predicted 10 transmembrane domains for C. elegans ORF-7. These considerations make a 10transmembrane topology for C. elegans ORF-7, as well as for the highly conserved human homolog, the most likely topology. We performed sequence alignments using the ClustalX method and created Dendrograms using MegAlign from the DNASTAR software. DNA cloning. We cloned full-length cDNAs of 12 of 16 putative C. elegans NSTs from a commercially available cDNA library (ProQuest Two-Hybrid C. elegans cDNA library 11288-016, Life Technologies) by PCR, using standard protocols and the Pfx-enzyme (Life Technologies). We used the following primers (sense primer followed by antisense primer): ORF-1, 5´–ATC AACATGAGGGTTGTGAA–3´, 5´–TTATGCTTTGCTCATCCATG–3´; ORF2, 5´–CAAAAATGCTTTGCGCTTCAC–3´, 5´–TTACCATAAATATGACTT CACTTC–3´; ORF-3, 5´–CCACAATGGGGCTCACAAAA–3´, 5´–TTATGATTTGCTGCTCTCGAC–3´; ORF-4, 5´–GGAAATATGAAGACGGC AATT–3´, 5´–TCAAACTATAAAACTTTTTAATCTC–3´; ORF-5, 5´–TTG CAATGTTCTACAAAGAATGT–3´, 5´–CTACTCGTTAACAGAGTTA TAC–3´; ORF-6, 5´–TGCCAAATGGCTGCTGCAGT–3´, 5´–TTAAAAGCG GGCAAATGACAC–3´; ORF-7, 5´–CGTCGTTATGTATCAGTCAGC–3´, 5´–TTAAACTGATTCCTCGGCCG–3´; ORF-8, 5´–AAGCCACAGAGTTACAAGCG–3´, 5´–GTGGCAACCGAACAATTTCG–3´; ORF-9, 5´–CTTTCA AATGACTGCAGCTC–3´, 5´–CTAAACCGTCATTGGATCCT–3´; ORF-10, 5´–ACAAGATGGTGGCGTTCGCG–3´, 5´–TTATGCCTCCATAAGGTGTT CT–3´; ORF-11, 5´–TAAGAATGGCATCGGCAGTT–3´, 5´–CTACTGTTT CTTCTTCTCATCT–3´; ORF-12, 5´–GCAAAATGATCAGAAGTGAGG–3´, 5´–TTATTCATCCGTATCTTCTGTG–3´; ORF-13, 5´–GAAAAATGCCGAT TTTCAAGC–3´, 5´–TCAGGCATTATGAGCTTCGG–3´; ORF-14, 5´–GCTGTATTTGTAACTCTCTGA–3´, 5´–CCGTGGAAAACCAATAGTAC–3´; ORF15, 5´–TCAAAATGAGTTCAAGTGAAAATG–3´, 5´–CTACTTTTCCGGTTTAATCATCA–3´; ORF-16, 5´–TTCAGATGACGTCAACAGTAC–3´, 5´–TCA GTTCCTTGGCTTGTGCA–3´; human ORF-7, 5´–CCACCATGGCGCTGACCGGGGCCTCAGACCCCTC–3´, 5´–TCACACCCCCATGGCGCT–3´. For cloning of GFP-chimeras the following antisense primers were used: C. elegans ORF-7, 5´–GAACGTATTCCTCGGCCGCA–3´; human ORF-7, 5´–GCACCCCCATGGCGCTCTT–3´. We cloned PCR products into the TOPO-TA vector (pcDNA3.1/CT-GFP-TOPO, Invitrogen) either as fulllength sequences ending at its C terminus with a stop codon (for complementation experiments) or as fusion proteins with a C-terminal GFP (for targeting experiments). All PCR products were analysed by sequencing. The cDNAs for human ORF-7 and mutated human ORF-7 from LAD II cells were cloned by RT–PCR using MMLV-reverse transcriptase (Stratagene) and total RNA isolated with Trizol (Life Technologies) either from normal human blood leukocytes or from LAD II fibroblasts. We performed PCR reactions by standard protocols. Sequencing of human ORF-7 cDNAs isolated from 4 different RNA sources revealed a T at nt 772 instead of a C at the corresponding position. The resulting codon codes for a phenylalanine at position 258 in the polypeptide. Complementation and targeting experiments. For complementation, we co-transfected LAD II fibroblasts8 (from a patient named A.C. in the accompanying report13) with vector DNA (2.5 µg), containing a cDNA encoding a putative NST, and pcDNA3 vector DNA (1.5 µg), containing a cDNA encoding a JAM-GFP fusion protein (consisting of the truncated junctional adhesion molecule (JAM) fused at its C terminus to GFP). For targeting experiments, we transfected COS-7 cells with vector DNA (4 µg) encoding chimeric fusion proteins consisting of either human or C. elegans ORF-7 fused at its C terminus to GFP. Transfections were performed using the GeneJammer transfection reagent (Stratagene) according to the manufacturer’s instructions. After 48 h, we fixed and permeabilized the samples as described23. Fibroblasts were stained with the biotinylated AAL, which recognizes fucose specifically in α1,3- and in α1,6-linkage (Vector Laboratories). We incubated COS-7 cells with anti-golgin-97 monoclonal antibody (mouse IgG, Molecular Probes) and anti-mouse-IgG-Biotin conjugate (Dianova). Subsequently, we stained cells with Streptavidin-Cy3 (Dianova) and analysed them using a ×40 objective on a Zeiss Axioskop 50 microscope (Zeiss) and a SPOT imaging system (Diagnostic Instruments). 72
GenBank accession numbers. Human ORF-7, AF323970; human protein FLJ11320, NM_018389; Drosophila gene product CG9620, AAF54215. C. elegans ORFs 1 to 16 (accession numbers of corresponding cosmids and gene names are given, genes that were cloned as full-length cDNAs are underlined): ORF-1, Z82274, JC8.12 (corrected sequence now given under AF324487); ORF-2, Z68011, Z68004, T21B6.5; ORF-3, Z82288, ZK896.9; ORF-4, Z81102, Z82288, M02B1.1; ORF-5, M98552, ZK370.7; ORF-6, Z68215, C53B4.6; ORF-7, Z70750, C50F4.14 (corrected sequence now given under AF323969); ORF-8, AF045639, B0212.4 (corrected sequence now given under AF324488); ORF-9, U00067, F54E7.1; ORF-10, U23169, C29H12.2; ORF-11, AF036696, F15B10.1; ORF-12, AF024500, K06H6.3; ORF-13, AF016674, C03H5.2; ORF14, AF016438, F44C8.7 (corrected sequence now given under AF324489); ORF-15, AF003383, ZC250.3; ORF-16, U50135, C52E12.3 Acknowledgments
We thank T. Marquardt for the LAD II fibroblasts; M. Aurrand-Lion and B. Imhof for the GFP-JAM cDNA construct; and J. Brosius and S. Stamm for discussions. This work was supported by a grant of the Deutsche Forschungsgemeinschaft, SFB 293 to D.V. Received 7 December 2000; accepted 28 March 2001. 1. 2. 3. 4. 5.
6.
7.
8. 9.
10.
11.
12.
13.
14. 15.
16.
17.
18. 19. 20. 21.
22.
23.
Etzioni, A. et al. Recurrent severe infections caused by a novel leukocyte adhesion deficiency. N. Engl. J. Med. 327, 1789–1792 (1992). Becker, D.J. & Lowe, J.B. Leukocyte adhesion deficiency type II. Biochim. Biophys. Acta 1455, 193–204 (1999). Vestweber, D. & Blanks, J.E. Mechanisms that regulate the function of the selectins and their ligands. Physiol. Rev. 79, 181–213 (1999). Marquardt, T. et al. Correction of leukocyte adhesion deficiency type II with oral fucose. Blood 94, 3976–3985 (1999). Körner, C. et al. Decreased availability of GDP-L-fucose in a patient with LAD II with normal GDP-D-mannose dehydratase and FX protein activities. J. Leukoc. Biol. 66, 95–98 (1999). Lübke, T., Marquardt, T., von Figura, K. & Körner, C. A new type of carbohydratedeficient glycoprotein syndrome due to a decreased import of GDP-fucose into the golgi. J. Biol. Chem. 274, 25986–25989 (1999). Maly, P. et al. The α-(1,3)fucosyltransferase Fuc-TVII controls leukocyte trafficking through an essential role in L-, E-, and P-selectin ligand biosynthesis. Cell 86, 643–653 (1996). Marquardt, T. et al. Leukocyte adhesion deficiency II syndrome, a generalized defect in fucose metabolism. J. Pediatr. 134, 681–688 (1999). Gerardy-Schahn, R. & Eckhardt, M. Nucleotide sugar transporters in carbohydrates. in Oligosaccharides in Chemistry and Biology: A Comprehensive Handbook, Vol. II. Biology of Saccharides (eds. Ernst, B., Hart, G. & Sinay, P.) 19–36 (Wiley-VCH, Weinheim, 2000). van Die, I. et al. Core α1,3)-fucose is a common modification of N-glycans in parasitic helminths and constitutes an important epitope for IgE from Haemonchus contortus infected sheep. FEBS Lett. 463, 189–193 (1999). DeBose-Boyd, R.A., Nyame, A.K. & Cummings, R.D. Molecular cloning and characterization of an α1,3-fucosyltransferase, CEFT-1, from Caenorhabditis elegans. Glycobiology 8, 905–917 (1998). Abeijon, C., Robbins, P. & Hirschberg, C.B. Molecular cloning of the Golgi apparatus uridine diphosphate-N-acetylglucosamine transporter from Kluyveromyces lactis. Proc. Natl. Acad. Sci. USA 93, 5963–5968 (1996). Lübke, T. et al. Complementation cloning identifies CDG-IIc, a new type of congenital disorders of glycosylation, as a GDP-fucose transporter deficiency. Nature Genet. 28, 73–76 (2001). Eckhardt, M., Gotza, B. & Gerardy-Schahn, R. Membrane topology of the mammalian CMP-sialic acid transporter. J. Biol. Chem. 274, 8779–8787 (1999). Berninsone, P., Hwang, H.-Y., Zemtseva, I., Horvitz, H.R. & Hirschberg, C.B. SQV-7, a protein involved in Caenorhabditis elegans epithelial invagination and early embryogenesis, transports UDP-glucuronic acid, UDP-N-acetylgalactosamine, and UDP-galactose. Proc. Natl. Acad. Sci. USA 98, 3738–3743 (2001). Karsan, A. et al. Leukocyte adhesion deficiency type II is a generalized defect of de novo GDP-fucose biosynthesis. Endothelial cell fucosylation is not required for neutrophil rolling on human nonlymphoid endothelium. J. Clin. Invest. 101, 2438–2445 (1998). Lühn, K., Marquardt, T., Harms, E. & Vestweber, D. Discontinuation of fucose therapy in LADII causes rapid loss of selectin ligands and rise of leukocyte counts. Blood 97, 330–332 (2001). The C. elegans Sequencing Consortium. Genome sequence of the nematode C. elegans: a platform for investigating biology. Science 282, 2012–2018 (1998). Rost, B. PHD: predicting 1D protein structure byprofile based neural networks. Methods Enzymol. 266, 525–539 (1996). Kyte, J. & Doolittle, R.F. A simple method for displaying the hydropathic character of a protein. J. Mol. Biol. 157, 105–132 (1982). Ishida, N. et al. Indispensability of transmembrane domains of Golgi UDPgalactose transporter as revealed by analysis of genetic defects in UDP-galactose transporter-deficient murine Had-1 mutant cell lines and construction of deletion mutants. J. Biochem. (Tokyo) 126, 1107–1117 (1999). Gao, X.-D. & Dean, N. Distinct protein domains of the yeast Golgi GDP-mannose transporter mediate oligomer assembly and export from the endoplasmic reticulum. J. Biol. Chem. 275, 17718–17727 (2000). Ebnet, K., Schulz, C.U., Meyer-zu-Brickwedde, M.-K., Pendl, G.G. & Vestweber, D. Junctional adhesion molecule (JAM) interacts with the PDZ domain containing proteins AF-6 and ZO-1. J. Biol. Chem. 275, 27979–27988 (2000).
nature genetics • volume 28 • may 2001
