The complete C. elegans genome 1415 contains only one open reading frame, designated gana-1 (R07B7.11) that has sequence similar to human genes encoding α-GAL and α-NAGA. Similar results were obtained from searching the available C. briggsae genome sequence 15. The gana-1 gene consists of 5 exons and 4 introns and is annotated as an ortholog of human α-NAGA. Several EST clones for this open reading frame (ORF) have been reported and open-reading-frame sequence tag (OST) is present in the Worfdb database 16. Available public database data are in agreement with our findings.

We verified the gene structure by sequencing the PCR products from genomic DNA and cDNA (Figure 1). The analyzed region spanned the entire coding region and the 3' and 5' untranslated regions (UTR). The 5' UTR SL1 element suggests that the gene is either the only gene transcribed from the promoter or is the first gene in an operon including gana-1 and the two predicted downstream genes R07B7.12 and R07B7.13. Although this region has not been reported as an operon, the physical distances between this cluster of three genes are suggestive of an operon 171819. No alternative splicing was found using RT/PCR, a feature similar to both human and mouse orthologs. RNA editing was reported in the 3' UTR of human α-GAL, a finding that another group was unable to confirm 20. We noted no signs of RNA editing in clones derived from the gana-1 cDNA.

We aligned GANA-1 protein sequence with other melibiase family members (Figure 2) and constructed a phylogenetic tree (Figure 3). The alignment showed a striking similarity of GANA-1 to all other included sequences. GANA-1 had the highest sequence similarity with Anopheles gambiae GAL (46%), the lowest similarity was observed with Streptomyces avertimilis GAL (22%). The results of our phylogenetic analysis are in accordance with generally accepted evolutionary concepts. The analysis identified four main clades: animal NAGAs, animal GALs, plant/lower organisms GALs and the clade containing sequences of Drosophila melanogaster, Anopheles gambiae and Caenorhabditis elegans. The branch including C. elegans is positioned between higher animal GALs and NAGAs and plant/lower organisms GALs. This position in the tree infers the evolutionary ancestrality of gana-1 gene to both animal GALs and animal NAGAs. However, this conclusion is not in complete agreement with the presence of pairs of genes in Drosophila and Anopheles genomes annotated as α-GALs and α-NAGAs. The presence of these genes in the Caenorhabditis/Drosophila/Anopheles branch (and not in the GAL and NAGA clades) could be due to low divergence of these sequences from a common ancestral gene or to independent gene duplication in the Drosophila/Anopheles ancestral organism. It is also important to note that the phylogenetic analysis by maximum parsimony algorithm placed the Caenorhabditis/Drosophila/Anopheles branch into the neighborhood of the animal NAGAs branch 8. In this case the computational algorithm probably favored the lower number of necessary sequence changes (parsimony) between GANA-1 and NAGA clade sequences.

In our opinion, the phylogenetic analysis provides evidence that the GANA-1 evolved from a common ancestor of α-GAL and α-NAGA enzymes. However, the topology of the tree could also be explained by a loss of the second gene during the evolution of C. elegans. In this case the enzyme found in C. elegans would probably be the ortholog of human α-NAGA and the lost gene would likely be the ortholog of human α-GAL. The likelihood of these two hypotheses depends on functional divergence of duplicated gene products and their dispensability for organism's metabolic pathways 21.

The best Squared Root of Mean Square Deviations value (RMSD), found between GANA-1 backbone atoms and the chicken α-NAGA template 7, was 0.52 Å. The structural model of the enzyme molecule has a two-domain structure (Figure 4A). Domain I, which contains the predicted active site, adopts a (β/α)₈ barrel structure which represents a common motif in many glycoside hydrolases. Less conserved is domain II that has a β domain with β sandwich structure containing a Greek key motif. The active site pocket of GANA-1 is formed by the same twelve amino acids (W31, D76, D77, Y118, C126, K152, D154, C156, S186, A189, Y190 and R211) (Figure 4B) as in chicken α-NAGA. This finding infers their identical function in catalytic reactions as described for chicken α-NAGA 7. D134 carboxyl starts the initial nucleophilic attack and D215 carboxylic oxygen serves as a subsequent donor and acceptor of the proton during the reaction cycle.

Residues forming the "N-acetyl recognition loop" in the chicken α-NAGA 8 (S172, A175, Y176) have the closest contact with the N-acetyl moiety of the ligand. These residues are completely conserved between human and chicken NAGAs, but in human α-GAL serin 172 is replaced by glutamine and alanine 175 is replaced by leucine. The replacements with bulkier residues apparently discriminate between α-GAL and α-NAGA substrates. While NAGAs can accommodate α-galactose and can have some α-GAL activity, GALs do not have α-NAGA activity and are not inhibited by N-acetylgalactosamine. The corresponding residues of GANA-1 in the NAGA recognition loop are S186 and A189 and are characteristic for NAGAs.

According to Garman 8 the key residue in the dimer interface in human α-GAL is F273. Residues corresponding to this position in other orthologs can serve as predictive markers of the protein quartenary structure. Phenylalanine or tyrosine is present in enzymes that act as homodimers while glycin indicates a monomeric structure 8. The equivalent residue to human α-GAL F273 in GANA-1 is lysine at position 257 which is suggestive of homodimeric structure due to its sterical properties. The homology modeling showed that a groove opposing K257 is formed by residues T260, L261, D262, M263, I389, V390 and V391 of the other monomer unit of GANA-1. In the case of chicken α-NAGA these residues are equivalent to S246, Y247, E248, Q249, N375, P376 and S377 (for details see Additional file 1).

Standard bacteria/nematode separation protocol previously used by other authors 2223 while evaluating lysosomal enzyme activites is based on sucrose flotation approach. We avoided standard sucrose flotation of worms because we could not exclude unpredictable artifacts caused by this compound, which is known to induce artificial lysosomal storage in eukaryotic cells and to alter lysosomal gene expression at concentrations significantly lower 24 than those used in flotation protocols.

We found both α-GAL and α-NAGA enzymatic activities in the homogenates from C. elegans N2 strain using 4-methylumbelliferyl (MU) substrates. The α-NAGA activity was dominant over the α-GAL activity. The activity of α-NAGA measured at 37°C was 430 nmol.mg^-1.h^-1 with MU-α-N-acetylgalactosaminide compared to the activity of α-GAL with MU-α-galactopyranoside of 43 nmol.mg^-1.h^-1 (about 10% of that of α-NAGA).

In the assay of α-GAL, the degradation of the MU-α-galactopyranoside was inhibited up to 95% in the presence of N-acetyl-D-galactosamine (D-GalNAc), whereas in the presence of D-galactose (D-Gal) the degradation of the same substrate was inhibited up to 75%. In the assay of α-NAGA, the degradation of the MU-α-N-acetylgalactosaminide was inhibited up to 97% by D-GalNAc and up to 90% by D-Gal. No inhibition of α-NAGA and α-GAL activity by D-glucose was observed.

According to published observations in human enzymes, D-GalNAc has no inhibitory effect on α-GAL activity. On the other hand, human α-NAGA activity is inhibited by both, D-GalNAc and D-Gal 25. The model of GANA-1 predicts only one active site per monomer of the enzyme. If the enzyme had activity toward both substrates, MU-α-D-galactopyranoside and MU-α-N-acetylgalactosaminide, it is to be expected that D-GalNAc and D-Gal would inhibit both activities. The strong inhibitory effect of D-GalNAc on the α-GAL activity, which is not present in human α-GAL, supports the hypothesis that C. elegans has only one enzyme with both α-GAL and α-NAGA activities. Nevertheless, these experiments were not conducted with the pure enzyme and thus do not provide absolute proof of this hypothesis.

RNA interference assays directed against the whole coding region of gana-1, employing combination of microinjection and feeding approaches, did not reveal any abnormal morphological phenotypes. Nevertheless, measurement of α-GAL and α-NAGA activities in four different experiments showed a simultaneous decrease of both enzymatic activities in RNAi-treated worms (Table 1) as compared with control animals. In all RNAi experiments, both α-GAL and α-NAGA activities decreased proportionally, usually by tens of percent of activity of appropriate controls. The activity of the control enzyme (β-hexosaminidase) did not differ between the RNAi-treated nematodes and controls (data not shown). This finding supports the specificity of gana-1 RNAi. The differences between individual experiments are not surprising due to the well-known variability in the efficiency of RNAi 26. The results of RNAi experiments further support the hypothesis that GANA-1 has both enzymatic activities.

Both enzymatic activities were lower in RNAi-treated and control worms cultured on the bacterial strain HT115 26 compared to a N2 strain cultured on the OP50 strain.

RNAi previously provided sufficient level of inhibition of structural lysosomal proteins for development of abnormal phenotypes in the worm 2728; however, it is apparently not efficient enough for lysosomal catalytic proteins.

To study the expression of gana-1 in C. elegans, we created transgenic worms with a reporter gene containing the entire coding region of gana-1 C-terminally tagged with green fluorescent protein (GFP) under the control of a 3 kb region of the gana-1 hypothetical promoter. The presence of the gana-1::GFP transgene in the worms was confirmed on the level of genomic DNA, cDNA and protein. However, no GFP signal was observed by fluorescence microscopy under the standard laboratory conditions. As Western blotting showed the presence of fusion protein of the expected size (data not shown), we assumed that the absence of the GFP signal was caused by a pH-dependent quenching of GFP fluorescence, which has neutral to alkaline optimum (pH 5.5–12) 29.

To study the tissue and intracellular distribution of the fusion protein, we resorted to immunofluorescence detection of the transgene product. Immunofluorescence detection of GFP fusion protein showed a specific and coarsely granular cytoplasmic pattern of fusion protein expression. This transgene product was limited to body wall muscle cells (30% of population) (Figure 5A, Additional file 2) or found in a broader tissue distribution that included body wall muscle cells, intestinal cells and coelomocytes (3–5% of population) (Figure 5B, Additional file 3).

This latter staining pattern is consistent with the GFP detection in NH₄Cl and concanamycin A (CON A) experiments (Figure 6) discussed below. The expression of the transgene was observed in about 30% of the population which corresponded to usual expression efficiency of extrachromosomal array transgenes 3031. The immunofluorescence staining protocol resulted in a significant decrease of inherent intestinal granular autofluorescence previously assigned to secondary lysosomes 32. The decrease of autofluorescence intensity together with its poorly defined emission spectra hampered co-localization study.

To confirm that the absence of the GFP signal was due to the quenching of fluorescence by low pH in the acidic cellular compartment, we used two agents specifically alkalizing acidic cellular compartment 3334 to enhance the GFP emission. Soaking of gana-1::GFP transgenic animals in NH₄Cl or CON A resulted in a distinct GFP signal in a vesicular compartment of endocytically active coelomocytes in a small proportion of worms (3–5% of population). The GFP signal intensity was dependent on the time of incubation and the concentration of the alkalizing agent used. The first visible GFP signal was observed after 8 hour incubation in 100 mM NH₄Cl and within 2 hours of incubation in 50 nM CON A. Lower concentrations of both NH₄Cl and CON A did not result in visible GFP signal even after prolonged incubation of up to 24 hours. The reappearance of the GFP signal after treatment of the worms with compounds increasing the acidic compartment pH indirectly confirms lysosomal localization of the fusion protein. The GFP signal in coelomocytes had the same coarsely granular pattern as that observed after immunostaining.

Limited access of alkalizing agents to the tissues can explain the differences between the results of immunofluorescence and alkalization studies.

The wild type Bristol N2 strain was used for all experiments and was handled under standard laboratory conditions as described previously 35. Standard methods were used for DNA microinjection 36 and dsRNA synthesis and microinjection 37. Nomenclature is in agreement with available Genetic Nomenclature for Caenorhabditis elegans 15 and has been approved prior to manuscript submission.

Wormbase (2002–2004 versions and freeze versions 153839) databases were repeatedly searched for human α-GAL and α-NAGA orthologs using the BLASTP 40 program set at default values. Amino acid sequences of human lysosomal α-GAL and α-NAGA (acc. no. NP_000160 and acc. no. NP_000253 41) were used as query sequences.

Total RNA was isolated from mixed stages of N2 cultures 42 and reverse transcribed with an oligodT-T7 (5'-AATACGACTCACTATAG) primer and Superscript reverse transcriptase (Invitrogen). The entire coding region of R07B7.11 was PCR amplified in two overlapping PCR products, with intragenic primers designed according to available Wormbase 15 and Worfdb 16 data. SL1 primer (5'GGTTTAATTACCCAAGTTTGAG) and SL2 primer (5'GGTTTTAACCCAGTTACTCAAG) 17 together with gene specific primer (5'ATCCTGATTAATTTTAATTGC) were used to amplify 942 bp of the 5' part of the cDNA and to evaluate the mode of trans splicing; the 1142 bp fragment of the 3' end of cDNA was amplified with T7 primer and a gene specific primer (5'CTTAAGTTTGGAATTTATGAA). The dominant PCR products were cloned with TOPO TA cloning kit (Invitrogen) into the pCR 2.1 vector. Positive clones were sequenced using the Li-Cor automated fluorescent sequencer and sequences were aligned with R07B7 reference cosmid sequence in the AlignIR software (Li-Cor) to evaluate splicing boundaries and overall gene organization.

Confirmed or predicted amino acid sequences of melibiase family members 43 representing plant, unicellular, and animal kingdoms were aligned using ClustalW algorithm 44 and Blosum62 matrix. The SwissProt/TrEMBL 45 accesion code and source organism of the sequences are depicted in Figure 2. The sequence alignment was used for phylogenetic analysis with the software package PHYLIP 46. The phylogenetic tree is based on 100 bootstraped input alignments and was constructed by maximum likelihood method with Jones-Taylor-Thornton matrix model 47. Sequence identities between species were calculated without signal sequence in EMBOSS by Needleman-Wunsch global alignment algorithm with Blosum62 matrix, gap penalty – 10 and gap extension penalty – 0.5 84849. Signal peptides were predicted at the SignalP server 50 both by algorithms using neural networks and Hidden Markov Models. The results were compared to known signal sequences. The differences between signal peptides predicted by the algorithms are depicted in Figure 2.

The 3D model of GANA-1 is based on the X-ray structure of chicken α-NAGA, rice α-GAL and human α-GAL (PDB codes 1ktcA, 1uas and 1r47, respectively) 78951. The model was created using the automated homology modeling server SwissModel with structure refinement and model evaluation in the DeepView program 52. The print quality figures (Figure 4) and animations (Additional file 1) were ray traced using PovRay software package 53.

The entire coding region of the gana-1 gene, including 3 kb of its 5'upstream sequence, was amplified from N2 genomic DNA through a nested PCR reaction using DyNAzyme EXT™ PCR system (Finnzymes) and two pairs of primers: the external pair (5'GTGAGAGTGGGGAGATAGAA and 5'TCAATTTGCTTGAGGTACATA) and the internal primers, with overhangs containing SphI and SalI restriction sites respectively (5'ACATGCATGCAACTTTCACAGGAACACAAC and 5'CGACGTCGACAATTGAACTCTATTGGTTCTCAA). The amplified DNA fragment (4709 bp) was cloned using TOPO-XL cloning kit (Invitrogen) into the pCR-XL-TOPO vector. The SphI and SalI gana-1 restriction fragment was recloned into the GFP reporter vector pPD95.67 (supplied by A. Fire, Stanford University). The in-frame nature of the insert was confirmed by sequencing. The green fluorescent protein (GFP) fusion construct pJH3 (50 ng/μl) and pRF4 plasmid (50 ng/μl) used as the dominant marker were co-injected into the gonads of young adult N2 worms. Transgenic animals were screened for GFP signal. Nikon Eclipse E800 with C1 confocal module and 488 nm and 543 nm lasers and differential interference contrast (DIC) optics was used for specimen examination. EZ-C1 software (Nikon) was used for picture analysis and 3D rendering (Additional Files 2, 3).

Mixed stage pJH3 and N2 (control) cultures were harvested from NGM OP50 plates and washed with water. Worms were pelleted by centrifugation (max. 1000 RPM, 2 min.) between the washes. Worms were treated with either one of two agents (NH₄Cl, concanamycin A – CON A) 3334, that are known to specifically increase pH in the cellular acidic compartment. For the NH₄Cl method, animals were suspended in 0, 10, 25, 50, 75 and 100 mM aqueous solutions of NH₄Cl. Small aliquots of worms were examined after 30 min, 2, 4, 6, 8 and 24 hours. For CON A, animals were suspended in 0, 10, 20, 50, 100, 150, 200 nM solutions of CON A in aqueous media. Small aliquots of worms were examined after 1, 3, 6 and 24 hours.

Microscopical examination was performed as described above.

The fixation and immunofluorescence staining procedures were based on the approaches of Nonet et al. 54. Mixed stage N2 cultures were harvested from NGM OP50 plates and washed thoroughly in M9 buffer to remove intestinal bacteria. Worms were pelleted by centrifugation (1000 RPM, 2 min.) between the washes. Worms were fixed overnight in 4% paraformaldehyde in 100 mM sodium/potassium phosphate buffer. Afterwards the pellets were washed three times in 1 × PBS, and incubated in 1% Triton X-100, 100 mM Tris (pH 7.0), 1% β-Mercaptoethanol overnight at 37°C to reduce the cuticle. After 5 washes in 1 × PBS, the worms were incubated for 5 hours in 900 U/ml collagenase type IV (Sigma) diluted in Krebs-Ringer solution (119 mM NaCl, 25 mM NaHCO₃, 11.1 mM glucose, 1.6 mM CaCl₂·H₂O, 4.7 mM KCl, 1.2 mM KH₂PO₄, 1.2 mM MgSO₄·7H₂O, pH 7.4). The reduction/digestion step was performed twice. Pellets were washed 3 times with 1 × PBS and stored for further processing in AbA buffer (1 × PBS, 0.1% Triton X-100, 1% BSA, 0.05% NaN₃). AbA buffer was used for antibody dilution. Primary antibody (polyclonal rabbit anti-GFP IgG (Abcam)) was diluted 1:500. Secondary antibody (goat anti-rabbit IgG Alexa Fluor 488 labeled (Molecular Probes)) was diluted 1:1000. Both incubations were performed overnight at room temperature, with AbB buffer (1 × PBS, 0.1% Triton X-100, 0.1% BSA, 0.05% NaN₃) washes in between.

Nuclei were counterstained with SYTOX orange (Molecular Probes) and the microscopic evaluation was performed as described above.

Mixed stage pJH3 and N2 cultures were harvested from NGM OP50 plates. Worms were homogenized by sonication and the concentration of protein was measured by the Hartree method 55. The proteins (equivalent of 25–50 μg of protein per lane) were separated by SDS-PAGE gradient gel (4% to 20% polyacrylamide) and transferred onto nitrocellulose membrane by semi-dry blotting. The membrane was treated according to a common Western blotting protocol with chemiluminiscence detection (SuperSignal, West Pico) 56. Rabbit polyclonal anti-GFP IgG (Abcam, dilution 1:5000) was used as the primary antibody, the secondary antibody was goat anti-rabbit IgG/Px (Pierce, diluted 1:20 000).

The PCR product containing the entire gana-1 cDNA was cloned into pCRII-TOPO vector (Invitrogen) and L4440 double promoter vector for microinjection and feeding RNAi respectively. In-vitro transcription employing T7 and SP6 RNA polymerases (Promega) was used to generate antisense single stranded RNA molecules, which were annealed to generate double stranded RNA (dsRNA). dsRNA was microinjected into N2 worms which were fed on HT115 26E. coli strain carrying L4440 plasmid with gana-1 insert. The F₁ and early F₂ progeny was screened for morphological phenotypes. N2 worms microinjected with water and fed on HT115 E. coli transformed with L4440 vector without insert were used as a control. 5–7 worms were microinjected both with dsRNA and water in each of 4 separate experiments, single worm progeny reaching 110–150 individuals.

Prior to all activity measurements worms were washed from culture plates and repeatedly (6 times) washed and centrifuged in M9 buffer and finally pelleted by centrifugation. 4-methylumbelliferyl (MU)-α-D-N-acetylgalactosaminide (1 mM), 4-MU-α-D-galactopyranoside and 4-MU-β-D-glucopyranoside in theMcIlvaine buffer (0.1 M citrate/0.2 M phosphate buffer at acidic pH) were used as enzyme substrates. Reaction mixtures (sample and enzyme substrate) were incubated at 37°C and reactions were stopped by 600 μl of 0.2 M glycine/ NaOH buffer (pH 10.6) 1357. Fluorescence signal of the 4-methylumbelliferone was measured on the luminiscence spectrofotometer LS 50B (Perkin Elmer) (emission 365 nm and excitation 448 nm). Inhibitors (N-acetyl-D-galactosamine, D-galactose and D-glucose) were used in 0.1 M final concentration. All measurements were performed in doublets.