Portal de Periódicos da CAPES

Resistance to a Bacterial Toxin Is Mediated by Removal of a Conserved Glycosylation Pathway Required for Toxin-Host Interactions

2003; Elsevier BV; Volume: 278; Issue: 46 Linguagem: Inglês

10.1074/jbc.m308142200

1083-351X

Joel S. Griffitts, Danielle L. Huffman, Johanna L. Whitacre, Brad Barrows, Lisa D. Marroquin, Reto Müller, Jillian R. Brown, Thierry Hennet, Jeffrey D. Esko, Raffi V. Aroian,

Insect Pest Control Strategies

Crystal (Cry) proteins made by the bacterium Bacillus thuringiensis are pore-forming toxins that specifically target insects and nematodes and are used around the world to kill insect pests. To better understand how pore-forming toxins interact with their host, we have screened for Caenorhabditis elegans mutants that resist Cry protein intoxication. We find that Cry toxin resistance involves the loss of two glycosyltransferase genes, bre-2 and bre-4. These glycosyltransferases function in the intestine to confer susceptibility to toxin. Furthermore, they are required for the interaction of active toxin with intestinal cells, suggesting they make an oligosaccharide receptor for toxin. Similarly, the bre-3 resistance gene is also required for toxin interaction with intestinal cells. Cloning of the bre-3 gene indicates it is the C. elegans homologue of the Drosophila egghead (egh) gene. This identification is striking given that the previously identified bre-5 has homology to Drosophila brainiac (brn) and that egh-brn likely function as consecutive glycosyltransferases in Drosophila epithelial cells. We find that, like in Drosophila, bre-3 and bre-5 act in a single pathway in C. elegans. bre-2 and bre-4 are also part of this pathway, thereby extending it. Consistent with its homology to brn, we demonstrate that C. elegans bre-5 rescues the Drosophila brn mutant and that BRE-5 encodes the dominant UDP-GlcNAc:Man GlcNAc transferase activity in C. elegans. Resistance to Cry toxins has uncovered a four component glycosylation pathway that is functionally conserved between nematodes and insects and that provides the basis of the dominant mechanism of resistance in C. elegans. Crystal (Cry) proteins made by the bacterium Bacillus thuringiensis are pore-forming toxins that specifically target insects and nematodes and are used around the world to kill insect pests. To better understand how pore-forming toxins interact with their host, we have screened for Caenorhabditis elegans mutants that resist Cry protein intoxication. We find that Cry toxin resistance involves the loss of two glycosyltransferase genes, bre-2 and bre-4. These glycosyltransferases function in the intestine to confer susceptibility to toxin. Furthermore, they are required for the interaction of active toxin with intestinal cells, suggesting they make an oligosaccharide receptor for toxin. Similarly, the bre-3 resistance gene is also required for toxin interaction with intestinal cells. Cloning of the bre-3 gene indicates it is the C. elegans homologue of the Drosophila egghead (egh) gene. This identification is striking given that the previously identified bre-5 has homology to Drosophila brainiac (brn) and that egh-brn likely function as consecutive glycosyltransferases in Drosophila epithelial cells. We find that, like in Drosophila, bre-3 and bre-5 act in a single pathway in C. elegans. bre-2 and bre-4 are also part of this pathway, thereby extending it. Consistent with its homology to brn, we demonstrate that C. elegans bre-5 rescues the Drosophila brn mutant and that BRE-5 encodes the dominant UDP-GlcNAc:Man GlcNAc transferase activity in C. elegans. Resistance to Cry toxins has uncovered a four component glycosylation pathway that is functionally conserved between nematodes and insects and that provides the basis of the dominant mechanism of resistance in C. elegans. Bacterial pore-forming toxins that damage membranes are important virulence factors associated with pathogenic bacteria (1Gilbert R.J. Cell. Mol. Life Sci. 2002; 59: 832-844Crossref PubMed Scopus (159) Google Scholar, 2Schmitt C.K. Meysick K.C. O'Brien A.D. Emerg. Infect. Dis. 1999; 5: 224-234Crossref PubMed Scopus (110) Google Scholar). Examples include aerolysin from Aeromonas hydrophila, α toxin from Staphylococcus aureus, hemolysin from Escherichia coli, and main family crystal (Cry) 1The abbreviations used are: Cry, crystal; ORF, open reading frame; HA, hemagglutinin. proteins from Bacillus thuringiensis. Of these, Cry proteins are unique in that they specifically target invertebrates (insects and nematodes) and are generally considered innocuous to mammals. Mammalian resistance to Cry toxins is thought to be due to toxin insolubility, lack of proper proteolytic processing, and lack of proper receptors in the mammalian gut (3Betz F.S. Hammond B.G. Fuchs R.L. Regul. Toxicol. Pharmacol. 2000; 32: 156-173Crossref PubMed Scopus (352) Google Scholar). As such, Cry proteins demonstrate a degree of safety toward vertebrates that is unmatched by any other pest control product and are used extensively around the world in the control of insect pests that damage crops and carry disease. We previously reported on the isolation of Caenorhabditis elegans bre mutants (for Bt-toxin resistant) that resist Bt Cry protein intoxication (4Marroquin L.D. Elyassnia D. Griffitts J.S. Feitelson J.S. Aroian R.V. Genetics. 2000; 155: 1693-1699Crossref PubMed Google Scholar). When fed the main family crystal protein Cry5B either as protein crystals or as E. coli produced protein, C. elegans becomes rapidly sick and shows degeneration of the intestine, loss of coloration, inhibition of growth, inhibition of progeny production, and eventual death (4Marroquin L.D. Elyassnia D. Griffitts J.S. Feitelson J.S. Aroian R.V. Genetics. 2000; 155: 1693-1699Crossref PubMed Google Scholar, 5Wei J.Z. Hale K. Carta L. Platzer E. Wong C. Fang S.C. Aroian R.V. Proc. Natl. Acad. Sci. U. S. A. 2003; 100: 2760-2765Crossref PubMed Scopus (340) Google Scholar). Cry5B-resistant bre mutants were isolated among the F2 progeny of ethylmethanesulfonate mutagenized larvae. After screening 42,000 mutagenized haploid genomes for animals resistant to Cry5B, we isolated 14 alleles of bre-2, 19 alleles of bre-3, 9 alleles of bre-4, and 2 alleles of bre-5 (4Marroquin L.D. Elyassnia D. Griffitts J.S. Feitelson J.S. Aroian R.V. Genetics. 2000; 155: 1693-1699Crossref PubMed Google Scholar). 2L. D. Marroquin and R. V. Aroian, unpublished results. All are strongly resistant to the pore-forming toxin, and all are healthy in the absence of toxin as well. To date we have also isolated one allele of bre-1 that has significantly weaker resistance to Cry5B (4Marroquin L.D. Elyassnia D. Griffitts J.S. Feitelson J.S. Aroian R.V. Genetics. 2000; 155: 1693-1699Crossref PubMed Google Scholar). No other loci have been identified in these screens. Thus, strong, healthy resistance to the pore-forming toxin inevitably leads to mutation of bre-2, bre-3, bre-4, or bre-5. The identification of bre-5 as a β1,3-glycosyltransferase gene has been reported (6Griffitts J.S. Whitacre J.L. Stevens D.E. Aroian R.V. Science. 2001; 293: 860-864Crossref PubMed Scopus (181) Google Scholar). To our knowledge this approach is unique in its objective to genetically characterize how whole animals interact with and resist a bacterial pore-forming toxin. In the case of Cry proteins, identifying the genes that mutate to confer resistance is important since the major threat to the long term utility of insecticidal Cry proteins is the emergence of resistance. More generally, understanding the pathways that can mediate resistance might suggest therapeutic strategies for coping with bacterial toxins that target mammals. Here, we identify the Cry5B toxin resistance genes bre-2 and bre-4 as glycosyltransferase genes that previously have not been genetically characterized. We find that bre-2 and bre-4 function in the intestine to confer susceptibility to Cry5B and are also required for the interaction of toxin with intestinal cells in vivo. We also identify the bre-3 resistance gene as sharing significant homology to Drosophila egghead (egh). Because bre-5 shares sequence similarity to the Drosophila brainiac (brn) gene and since egh and brn function in a single pathway in Drosophila, we hypothesized that the Cry protein intoxication pathway in C. elegans is equivalent to the Drosophila pathway. We indeed show that bre-3 and bre-5 function in a common pathway in C. elegans, that bre-2 and bre-4 are also part of this pathway, and that the C. elegans bre-5 gene has the same in vivo and in vitro activities as Drosophila brn. C. elegans Culture and Microscopy—C. elegans was propagated using standard techniques with strain Bristol N2 as the wild type (7Brenner S. Genetics. 1974; 77: 71-94Crossref PubMed Google Scholar). All nematode assays were carried out at 20 °C. For microscopy of live animals, nematodes were mounted on pads consisting of 2% agarose and 0.1% sodium azide. Low magnification microscopy was carried out on an Olympus BX-60 microscope using a 10× objective coupled to a 0.5× camera mount and a DVC camera. Endocytosis assays were imaged and deconvolved using a DeltaVision system (Applied Precision) on an Olympus IX-70 microscope using a 40× objective (NA1.35) and differential interference contrast or fluorescence optics. Immunofluorescence images were captured on the DeltaVision system using a 40× or 100× objective (NA1.35) and differential interference contrast or fluorescence optics. Cloning of bre Genes—bre-2(ye31) was placed in trans to dpy-18(e364) unc-64(e246). 25/25 Dpy nonUncs segregated bre-2(ye31), and 23/23 Unc nonDpys segregated bre(+). The deficiency eDf2 also failed to complement bre-2(ye31), suggesting that bre-2 resided between unc-64 and the right breakpoint of eDf2. The bre-2(ye71) allele was generated by mutagenesis in the Hawaiian strain background (CB4856) and then placed in trans to unc-25(e156) bli-5(e518). Unc nonDpy and Dpy non-Unc recombinants were used to follow the segregation of single nucleotide polymorphisms, which indicated that bre-2 mapped between nucleotides 3,645 and 153,209 of Y39E4B. This region contains 14 predicted open reading frames (ORFs). Sequencing of cDNAs isolated from two bre-2 alleles indicated that both alleles contained point mutations in Y39E4B.9. Identification of the putative glycosyltransferase Y39E4B.9 as bre-2 was confirmed by reproducing the phenotype using RNA-mediated interference (data not shown) and by cDNA rescue from a bre-2 mutant. Analogous to bre-3 (below, including spliced leader 1 primer at the 5′ end), a complete bre-2 cDNA was constructed using reverse-transcribed RNA from wild-type nematodes. The predicted amino acid sequence is exactly as predicted for Y39E4B.9. bre-4(ye13) was placed in trans to unc-35(e259) dpy-5(e61). 29/32 Dpy nonUncs-segregated bre-4(ye13) indicated bre-4 was near and to the right of unc-35. Bre nonDpy recombinants from bre-4(ye13) dpy-5(e61)/CB4856 heterozygotes placed bre-4 to the right of the single nucleotide polymorphism Y71G12B.60148. Dpy recombinants from heterozygotes unc-35(e259) dpy-5(e61),bre-4(ye43) (bre-4 allele in the Hawaiian background) placed bre-4 to the left of single nucleotide polymorphism Y92H12A.28829. The 150-kilobase region between these boundary single nucleotide polymorphisms contains 20 predicted ORFs. Sequencing of cDNA isolated from the bre-4 alleles revealed changes in the enzymatically characterized glycosyltransferase Y73E7A.7 (8Kawar Z.S. Van Die I. Cummings R.D. J. Biol. Chem. 2002; 277: 34924-34932Abstract Full Text Full Text PDF PubMed Scopus (70) Google Scholar). cDNA rescue of a bre-4 mutant confirmed the identity of bre-4. A complete cDNA was assembled as for the other bre genes (including using spliced leader 1-specific primers). The predicted protein sequence corresponds exactly as predicted for ORF Y73E7A.7. bre-3(ye28) was placed in trans to sma-2(e502) ced-7(n1892) unc-69(e587). Of 25 Sma non-Unc recombinants, 12 were sma-2(e502) bre-3(ye28) ced-7(+), 12 were sma-2(e502) bre-3(+) ced-7(n1892), and 1 was sma-2(e502) bre-3(+) ced-7(+). These data indicated bre-3 was close and to the left of ced-7. We co-injected bre-3(ye28) animals with cosmids in this region and plasmid pRF4, which encodes the dominant rol-6 marker, and found that 2/2 stably transmitting lines containing cosmid B0464 were rescued to toxin susceptibility. Subcloning and injection of different regions within B0464 indicated that the rescuing activity (8/8 lines) was conferred by a 4.3-kilobase FspI-NheI fragment containing nucleotides 13,437-17,729. This fragment contains a single predicted ORF B0464.4 and an additional ∼800 bases at both the 5′ and 3′ ends of the ORF. Sequencing of genomic DNA and cDNA from three bre-3 mutant alleles indicated these alleles contained alterations in this ORF, confirming the identity of the gene. A complete wild-type cDNA was assembled and sequenced from two overlapping fragments generated 1) at the 5′ end by using random-primed total cDNA, a spliced leader 1 primer, an internal primer, and PCR and 2) at the 3′ end by using a λ phage cDNA library (generously provided by Robert Barstead), a 3′ end-specific vector primer, an internal primer, and PCR. The predicted protein sequence from this assemble cDNA corresponds exactly to that of the predicted B0464.4 ORF. Transmembrane domains in BRE-3 and other related proteins were predicted using TMHMM2.0. Intestine-specific Expression and Rescue—pBluescript (KS) (Stratagene) was used as the vector backbone for test constructs. The cpr-1 promoter (9Britton C. McKerrow J.H. Johnstone I.L. J. Mol. Biol. 1998; 283: 15-27Crossref PubMed Scopus (47) Google Scholar), corresponding to nucleotides 9,541-11,581 of cosmid C52E4, was amplified from C. elegans genomic DNA and cloned upstream of bre gene coding regions (for bre-2, bre-3, and bre-5, respectively, nucleotides 66,617-69,835 of Y39E4B, 14,285-16,925 of B0464, and 22,694-23,958 of cosmid T12G3; the bre-4 coding region was amplified from C. elegans cDNA). Coding regions were fused in-frame to a double HA tag at their 3′ ends, with a downstream transcription termination region derived from the bre-5 locus (T12G3 nucleotides 21,578-22,699), except in the bre-3 construct, which contained the bre-3 transcription termination region downstream of the HA tag (B0464 nucleotides 13,437-14,287). Constructs were co-injected with the dominant rol-6 marker (pRF4) into mutant animals to obtain stably transmitting lines carrying extrachromosomal arrays. Eggs laid by Roller (or N2 control) hermaphrodites were placed on modified ENG agar plates containing 1 mm isopropyl β-d-thiogalactoside and 50 μg/ml carbenicillin and spread with E. coli cells expressing Cry5B (5Wei J.Z. Hale K. Carta L. Platzer E. Wong C. Fang S.C. Aroian R.V. Proc. Natl. Acad. Sci. U. S. A. 2003; 100: 2760-2765Crossref PubMed Scopus (340) Google Scholar) or with vector-only cells. Eggs were allowed to hatch and grow for 72 h, and resistance was scored based on the ability of larvae to progress to adulthood during the course of the experiment. For each construct tested at least two independent transgenic lines were assayed, and all lines possessed the reported phenotypes with >90% penetrance, with ∼20 rollers and 20 non-rollers tested per line. Fixation and Staining for Immunofluorescence Microscopy—The fixation procedure was a modification of that of Finney and Ruvkun (10Finney M. Ruvkun G. Cell. 1990; 63: 895-905Abstract Full Text PDF PubMed Scopus (495) Google Scholar). A mixed population of a rescuing line expressing cpr-1::bre-5::2XHA or cpr-1::bre-3::2XHA was suspended in phosphate-buffered saline, 25% methanol, 3% formaldehyde and flash-frozen. After thawing under tap water worms were incubated on ice for 30 min followed by 2 washes in 50 mm sodium borate buffer, 0.01% Triton X-100, pH 9.5 (BO3T). Worms were then incubated for 20 min in BO3T plus 10 mm dithiothreitol followed by another wash in BO3T and a 20-min incubation in BO3T plus 0.3% hydrogen peroxide. Fixed and permeabilized worms were then washed in phosphate-buffered saline, 0.5% bovine serum albumin, 0.05% Triton X-100, 0.02% sodium azide (antibody buffer). For staining, all steps were carried out in antibody buffer using monoclonal anti-HA antibodies (16B12, Covance), and a final wash included 4′,6-diamidino-2-phenylindole for DNA staining. Assays for Endocytosis of Rhodamine-labeled Cry5B—Cry5B protoxin was purified as described previously (6Griffitts J.S. Whitacre J.L. Stevens D.E. Aroian R.V. Science. 2001; 293: 860-864Crossref PubMed Scopus (181) Google Scholar) and labeled using a 5-fold molar excess of NHS-rhodamine. Trypsinization was carried out at a 1:200 (trypsin:toxin) mass ratio for 2 h at 25 °C. Trypsinized toxin was still toxic but 5-10-fold less based on growth assays. We used Edman degradation sequencing and determined that trypsin removes the first 170 amino acids, which includes predicted α helix 3 but not helix 4. Consistent with our results studies with insecticidal Cry toxins show that α helices 4 and 5 are critical for toxin function and that helices 1-3 may contribute some activity (11Uawithya P. Tuntitippawan T. Katzenmeier G. Panyim S. Angsuthanasombat C. Biochem. Mol. Biol. Int. 1998; 44: 825-832PubMed Google Scholar, 12Vachon V. Prefontaine G. Coux F. Rang C. Marceau L. Masson L. Brousseau R. Frutos R. Schwartz J.L. Laprade R. Biochemistry. 2002; 41: 6178-6184Crossref PubMed Scopus (40) Google Scholar). Based on the apparent molecular weight of trypsinized toxin, trypsin is predicted to remove most of the carboxyl-terminal protoxin domain. Animals (L4 stage) were incubated in M9 medium containing 50 μg/ml protoxin or 120 μg/ml trypsinized toxin for 3 h and washed 3 times in M9 before imaging. For wash-out experiments, bre-4(ye13) and bre-5(ye17) animals fed rhodamine toxin for a few hours in wells were pipetted onto standard NG plates and examined every few minutes on an Olympus green fluorescent protein dissecting scope fitted with a rhodamine filter. Single Mutant Versus Double Mutant Resistance Assays—All double mutants were constructed and confirmed using a similar strategy. For example, the bre-4(ye13),bre-5(ye17) double mutant was created by mating bre-4(ye13) males into the strain dpy-5(e61),bre-5(ye17) (bre-4 and dpy-5 are both on chromosome I). NonDpy F1 progeny were cloned, and in the F2 nonDpy Cry5B-resistant animals were picked that segregated F3 Dpy (bre-4(y13),dpy-5(e61),bre-5(ye17)). NonDpy F3 were cloned out, and those that did not segregate Dpy were selected as bre-4(ye13),bre-5(ye17) putative doubles. The presence of both mutations was verified by sequencing DNA. To measure resistance of single and double mutants we used the Cry5B-related crystal protein Cry14A (5Wei J.Z. Hale K. Carta L. Platzer E. Wong C. Fang S.C. Aroian R.V. Proc. Natl. Acad. Sci. U. S. A. 2003; 100: 2760-2765Crossref PubMed Scopus (340) Google Scholar, 6Griffitts J.S. Whitacre J.L. Stevens D.E. Aroian R.V. Science. 2001; 293: 860-864Crossref PubMed Scopus (181) Google Scholar). Cry14A is required for these studies since bre-2, -3, -4, and -5 are very highly resistant to Cry5B (e.g. they are not killed by even 1 mg/ml Cry5B) but are only moderately resistant to Cry14A. Cry14A was produced in E. coli and incorporated into the assay diet as described (6Griffitts J.S. Whitacre J.L. Stevens D.E. Aroian R.V. Science. 2001; 293: 860-864Crossref PubMed Scopus (181) Google Scholar). Assays were carried out in 24-well plates, with each well containing 400 μl of S medium with 30 μg/ml each of tetracycline and chloramphenicol and ∼60 newly hatched L1 animals, with a final bacterial density of A 600 = 0.25. After a 60-h exposure at 20 °C, animals were mounted and imaged, and area measurements were made outlining the nematodes by hand and processing with NIH Image. bre-5 Rescue of the Drosophila brainiac Mutant—The bre-5 cDNA was cloned into pUAST (13Brand A.H. Perrimon N. Development. 1993; 118: 401-415Crossref PubMed Google Scholar) and used to transform yellow, white embryos using standard techniques. Two independent lines containing the transgene on chromosome II were used for brn complementation assays, which were carried out as described in Müller et al. (14Müller R. Altmann F. Zhou D. Hennet T. J. Biol. Chem. 2002; 277: 32417-32420Abstract Full Text Full Text PDF PubMed Scopus (48) Google Scholar). Briefly, male UAS-bre-5 transformants were crossed to females of the genotype forked, brn 1.6P6/FM6,w 1, actin-, heat shock- or armadillo-GAL4,w +. Rescue was scored based on the ability of forked male progeny to hatch. The GAL4 drivers used for this experiment carry the Bloomington stock numbers 1560 (armadillo-GAL4), 2077 (heat-shock GAL4), and 4414 (actin-GAL4). These lines were generously provided by Markus Noll and Erich Frei (Institute of Molecular Biology, University of Zürich). Flies carrying UAS-brainiac and UAS-β3GnT-IV (a human N-acetylglucosaminyltransferase) were used as positive and negative controls, respectively. bre-5-dependent N-Acetylglucosaminyltransferase Assays—N2 and bre-5(ye17) mixed larval-staged animals were reared on ENG medium (standard NG with an additional 2.5 g of peptone and 1 g of yeast extract per liter), washed, flash-frozen in 50 mm Tris, pH 7.5, as a 0.2-ml pellet, and stored at -70 °C. Samples were sonicated on ice in 0.05% Triton-X-100, 50 mm Tris and centrifuged at 15,000 × g for 10 min to obtain a clarified supernatant. The assay mixture (25 μl) contained 50 mm cacodylate, pH 6.5, 0.1% taurodeoxycholate, 10 mm MnCl2, 1 μCi of UDP-[6-3H]GlcNAc (39.7 Ci/mmol, PerkinElmer Life Sciences),10 mm Manβ1,4Glc-pNP (custom synthesis of Toronto Research, ON, Canada), and 8 μg of extracted protein (N2 or bre-5) as the enzyme source. After incubation at 25 °C for 7 h, the reaction products were diluted with 0.5 ml of 0.5 m NaCl and applied to a Sep-Pak C18 cartridge (100 mg; Waters). After washing the cartridge with 25 ml of water, the products were eluted with 50% methanol, dried, and counted by liquid scintillation. To better understand how resistance to pore-forming Cry toxins develops, we used single-nucleotide polymorphisms to map the bre-2 and bre-4 genes to small molecular intervals of ≤20 genes, each of which contained a glycosyltransferase gene. Given that the previously identified bre-5 was shown to encode a glycosyltransferase gene (6Griffitts J.S. Whitacre J.L. Stevens D.E. Aroian R.V. Science. 2001; 293: 860-864Crossref PubMed Scopus (181) Google Scholar), we sequenced cDNAs isolated from bre-2 and bre-4 mutant alleles and found that multiple alleles of each were associated with point mutations in the glycosyltransferase genes Y39E4B.9 and Y73E7A.7, respectively (Fig. 1, A and B). At least one allele for each gene (bre-2(ye68) and bre-4(ye13)) is predicted to eliminate protein function by truncation of the putative glycosyltransferase catalytic domain, indicating that Cry5B resistance is the null phenotype for each. That RNA-mediated interference (RNAi) of bre-2 recapitulates Cry5B resistance supports this (data not shown). Rescue data (see below) confirm that these glycosyltransferase genes encode bre-2 and bre-4. Cloning and characterization of bre-3, which encodes an unusual glycosyltransferase of striking resemblance to a known glycosyltransferase in Drosophila, is dealt with below. bre-2 encodes a putative family 31 β1,3-glycosyltransferase with a single amino-terminal membrane-spanning domain and an appropriately positioned DXD motif that is conserved in this enzyme family and is thought to be critical for catalysis (Fig. 1A). In general, BRE-2 shares ∼25% amino acid identity in its putative enzymatic domain with other β1,3-glycosyltransferases such as with BRE-5 (a β1,3-GlcNAc transferase; see Griffitts et al. (6Griffitts J.S. Whitacre J.L. Stevens D.E. Aroian R.V. Science. 2001; 293: 860-864Crossref PubMed Scopus (181) Google Scholar) and below) and insect and mammalian β1,3-Gal and GlcNAc-transferases. bre-2 is also a member of a more closely related and expanded subfamily of C. elegans β1,3-glycosyltransferase genes that includes T09F5.1, F14B6.6, T09E11.10, E03H4.11, C54C8.3, F14B6.4, C47F8.3, C47F8.5, C47F8.6, and T15D6.5 and to which it shares ∼40% identity in its glycosyltransferase domain. To our knowledge, bre-2 is the first one of these family members to be characterized genetically and to be demonstrated to have a loss of function phenotype. bre-4 is predicted to encode a member of the family 7 glycosyltransferases, of which β1,4-galactosyltransferases are the best characterized (15Ramakrishnan B. Boeggeman E. Qasba P.K. Biochem. Biophys. Res. Commun. 2002; 291: 1113-1118Crossref PubMed Scopus (39) Google Scholar). BRE-4 is predicted to contain a single membrane-spanning region near its amino terminus as well as conserved residues involved in catalysis (Fig. 1B). BRE-4 has been previously characterized biochemically as a UDP-GalNAc:GlcNAc β1,4-N-acetylgalactosaminyltransferase (Ceβ4 GalNAcT) (8Kawar Z.S. Van Die I. Cummings R.D. J. Biol. Chem. 2002; 277: 34924-34932Abstract Full Text Full Text PDF PubMed Scopus (70) Google Scholar). BRE-4/Ceβ4GalNAcT specifically synthesizes the GalNAcβ1,4GlcNAc sequence commonly found on glycoproteins and glycolipids. No mutational analysis of bre-4 has been previously reported. There are two other related glycosyltransferase genes in the C. elegans genome. Blast searches using both BRE-4 full-length protein and its enzymatic domain indicate a best hit and likely homologue in Drosophila melanogaster (predicted gene CG8536) and Anopheles gambiae (ORF from nucleotides 565443 to 566142, chromosome II). The predicted catalytic domains of these proteins, respectively, share 45 and 51% identity to the BRE-4 catalytic domain. Expression of bre-5, bre-2, and bre-4 in the Intestine Is Required for Cry Protein Intoxication—Based on the fact that Cry toxins appear to specifically attack the gut cells of insects and nematodes, we predicted that bre-2 and bre-4 glycosyltransferase genes should be required in intestinal cells for Cry protein intoxication. To demonstrate this intestinal requirement while simultaneously confirming the correct identities of bre-2 and bre-4, we placed wild-type bre-2 genomic coding region and bre-4 cDNA downstream of the intestinal-specific C. elegans promoter from the cpr-1 gene (9Britton C. McKerrow J.H. Johnstone I.L. J. Mol. Biol. 1998; 283: 15-27Crossref PubMed Scopus (47) Google Scholar). As a control using the previously cloned bre-5 gene we also made a cpr-1::bre-5 construct in which a 2× HA tag was fused in-frame at the carboxyl terminus of bre-5. Intestine-specific expression of the cpr-1 promoter was confirmed by HA antibody staining of a stably transmitting line transformed with the cpr-1::bre-5:HA construct (Fig. 2, A and B). This construct was fully capable of restoring Cry5B susceptibility to a bre-5(ye17) mutant (Fig. 2, C and D), demonstrating that intestinal expression of wild-type bre-5 in an animal otherwise lacking bre-5 is sufficient for Cry protein intoxication and providing a complementary result to our mosaic experiments (6Griffitts J.S. Whitacre J.L. Stevens D.E. Aroian R.V. Science. 2001; 293: 860-864Crossref PubMed Scopus (181) Google Scholar). We similarly found that intestinal-specific expression of wild-type bre-2 in a bre-2(ye31) mutant background and intestinal-specific expression of wild-type bre-4 in a bre-4(ye13) mutant background fully restored Cry toxin susceptibility to each (Fig. 2, E and F). Thus, like bre-5, both glycosyltransferase genes act in the intestine in their mediation of Cry protein intoxication. The fact that the bre-2 gene complements a bre-2 mutant and that the bre-4 gene complements a bre-4 mutant also confirms the identities of these genes. To demonstrate specificity of rescue, we transformed a bre-5(ye17) mutant with the wild-type cpr-1::bre-2 construct that fully rescues a bre-2 mutant. No rescue of the bre-5 mutant by the bre-2 gene was seen (Fig. 2G) despite the fact that both are members of the extended β1,3-glycosyltransferase family. Toxin Uptake in C. elegans Is Associated with Active Toxin and Requires bre-2, bre-4, and bre-3—We previously demonstrated that bre-5 is required for Cry5B toxin to interact with intestinal cells since bre-5(ye17) mutant animals fail to endocytose Cry5B protoxin labeled with rhodamine (6Griffitts J.S. Whitacre J.L. Stevens D.E. Aroian R.V. Science. 2001; 293: 860-864Crossref PubMed Scopus (181) Google Scholar). However, because Cry proteins are produced as large (protoxin) precursors that are proteolytically activated, it was not demonstrated that toxin molecule taken up was the active toxin fragment or that the toxin had to be functional to be taken up. To extend these results, we fed C. elegans rhodamine-labeled Cry5B that was full-length (protoxin), proteolytically activated with trypsin (Fig. 3A), or boiled to inactive it. As reported (6Griffitts J.S. Whitacre J.L. Stevens D.E. Aroian R.V. Science. 2001; 293: 860-864Crossref PubMed Scopus (181) Google Scholar), full-length protoxin is readily taken up by intestinal cells into autofluorescent gut granules, which are reported to be the secondary lysosome (Fig. 3, B and C). We found that trypsin-activated toxin is also endocytosed by intestinal cells (Fig. 3D). In contrast, inactivated toxin is not taken up (Fig. 3E). Thus, toxin uptake into intestinal cells is associated with activated, functional toxin. In contrast to wild-type animals, bre-2(ye31) bre-4(ye13), and bre-3(ye28) (bre-3 is discussed further below), mutant animals do not endocytose either functional protoxin or trypsin-activated toxin into their intestinal cells (Fig. 3, F-K). Rhodamine toxin seen in the intestinal lumen is not stably associated with the intestinal membrane since bre-4(ye13) and bre-5(ye15) mutant animals fed a pulse of rhodamine toxin flush the labeled toxin from their lumen within 5-10 min; no residual binding was detected (Fig. 3L, shown for bre-4(ye13)). These results suggest that the glycosyltransferase genes make a component required for toxin to interact with the plasma membrane of the intestine. An alternative explanation, that these mutants are all defective in general endocytosis by the intestine