A 49 kDa microtubule cross-linking protein from Artemia franciscana is a coenzyme A-transferase


T. H. MacRae, Department of Biology, Dalhousie University, Halifax, NS, B3H 4J1, Canada.
Fax: + 1 902 4943736; Tel.: + 1 902 4946525;
E-mail: tmacrae@dal.ca


Embryos and larvae of the brine shrimp, Artemia franciscana, were shown previously to possess a protein, now termed p49, which cross-links microtubules in vitro. Molecular characteristics of p49 were described, but the protein's identity and its role in the cell were not determined. Degenerate oligonucleotide primers designed on the basis of peptide sequence obtained by Edman degradation during this study were used to generate p49 cDNAs by RT-PCR and these were cloned and sequenced. Comparison with archived sequences revealed that the deduced amino acid sequence of p49 resembled the Drosophila gene product CG7920, as well as related proteins encoded in the genomes of Anopheles and Caenorhabditis. Similar proteins exist in several bacteria but no evident homologues were found in vertebrates and plants, and only very distant homologues resided in yeast. When evolutionary relationships were compared, p49 and the homologues from Drosophila, Anopheles and Caenorhabditis formed a distinct subcluster within phylogenetic trees. Additionally, the predicted secondary structures of p49, 4-hydroxybutyrate CoA-transferase from Clostridium aminobutyricum and glutaconate CoA-transferase from Acidaminococcus fermentans were similar and the enzymes may possess related catalytic mechanisms. The purified Artemia protein exhibited 4-hydroxybutyrate CoA-transferase activity, thereby establishing p49 as the first crustacean CoA-transferase to be characterized. Probing of Western blots with an antibody against p49 revealed a cross-reactive protein in Drosophila that associated with microtubules, but to a lesser extent than did p49 from Artemia.


microtubule-associated protein

Cell shape and polarity are regulated by microtubules, which serve as key structural elements of mitosis and provide tracks for intracellular transport. Microtubules are polar structures [1] and most are unstable, undergoing assembly and disassembly predominately from the plus end by a process called dynamic instability [2,3]. The formation of microtubules is modulated by tubulin isotypes [4] and microtubule-associated proteins (MAPs), a heterogeneous family defined simply by coassembly with tubulin and adherence to microtubules. Structural MAPs stabilize microtubules and modulate dynamic instability [5–8], whereas dynamic MAPs, or molecular motors, hydrolyse ATP as a prerequisite for vectorial movement of cells and their components [9]. Molecular motors are important during mitosis [10], and the Kin1 kinesin subfamily mediates ATP-dependent microtubule depolymerization [2,11]. Many proteins in addition to those just mentioned associate with microtubules in vivo and in vitro. For example, enzymes involved in tubulin post-translational processing [12,13] and glycolysis [14,15], affiliate with microtubules. Rho family GTPases, their kinases, and Ras GTPases interact with microtubules, seemingly as integral components of cell signaling mechanisms [16,17].

Incubation of cell-free extracts from the brine shrimp Artemia franciscana with paclitaxel (taxol) yielded cross-linked microtubules [18], and in this context, a 49 kDa microtubule interacting protein was isolated [19]. The protein, herein referred to as p49, failed to react with antibodies to structural MAPs such as MAP2 and tau, was moderately heat resistant and consisted of several developmentally invariant isoforms [19–21]. GTP, ATP and their analogues, at final concentrations of 10 mm, disrupted p49 binding to microtubules and weak microtubule-independent nucleotidase activity was detected [20]. In this study, p49 was sequenced and its molecular properties examined, showing that the protein is an acetyl CoA-transferase, the first described for a crustacean. A related protein was observed in Drosophila.

Experimental procedures

Preparation of Artemia and Drosophila cell-free extracts

Sixty grams (dry weight) of Artemia franciscana cysts (Sanders Brine Shrimp Co. or INVE Aquaculture, Inc., Ogden, UT, USA) were hydrated in distilled water at 4 °C for a minimum of 5 h, collected on a Buchner funnel and washed with cold distilled water. The embryos were divided among six 2 L flasks, each containing 1000 mL of Hatch Medium [22] and incubated with shaking at 220 r.p.m. for either 6 or 12 h. The cysts were suction-filtered on a Buchner funnel, washed with cold distilled water followed by Pipes buffer [100 mm 1,4-piperazine-N,N′-bis(2-ethanesulfonic acid) as free acid, 1 mm EGTA, 1 mm MgCl2, pH 6.5], and homogenized with a Retsch motorized mortar and pestle (Brinkman Instruments Canada, Rexdale, ON, Canada) in Pipes buffer for 5 min in three 60 g (wet weight) batches. Cysts developed for 12 h were homogenized as described except that 200 µg of each proteolytic inhibitor, leupeptin, soybean trypsin inhibitor, pepstatin A and phenylmethylsulfonyl fluoride (Sigma Chemical Co., St. Louis, MO, USA), were added to 60 g (wet weight) of cysts during homogenization. The homogenate was centrifuged at 40 000 g for 30 min at 4 °C. The upper two-thirds of each supernatant was removed, placed in a fresh tube, and centrifuged under the same conditions for 20 min. The supernatant was either used immediately or frozen at −70 °C.

Drosophila melanogaster embryos developed for 12 h were harvested from grape juice agar plates supplemented with yeast paste and washed with Ringer's solution (10 mm Tris/HCl, 182 mm KCl, 46 mm NaCl, 3 mm CaCl2, pH, 7.2). Larvae were collected from Drosophila culture medium plates after 12–36 h of incubation and washed with Ringer's solution. Pupae and adults were collected from culture bottles and rinsed with Ringers solution. All Drosophila samples were homogenized in Dounce homogenizers in Pipes buffer containing leupeptin, soybean trypsin inhibitor and pepstatin A, each at a final concentration of 0.004 µg·mL−1 and phenylmethylsulfonyl fluoride at 0.008 µg·mL−1. The homogenates were centrifuged at 16 000 g for 10 min at 4 °C and the supernatants transferred to fresh tubes before centrifugation at 40 000 g for 20 min at 4 °C. The supernatants were placed in fresh tubes, recentrifuged at 40 000 g for 20 min and these supernatants were either used immediately or stored at −70 °C. Drosophila were obtained from Vett Lloyd, Department of Biology, Dalhousie University, Halifax, NS, Canada.

Purification of p49, gel electrophoresis and protein immunodetection

p49 was prepared from Artemia MAPs as described previously [19] with paclitaxel [23] generously provided by the Drug Synthesis and Chemistry Branch, Developmental Therapeutics Program, Division of Cancer Treatment and Diagnosis, National Cancer Institute, Bethesda, MD, USA. Protein concentrations were determined by the method of Lowry et al. [24] using bovine serum albumin (Sigma) as standard. To assess p49 purity, protein fractions were electrophoresed in 12.5% (w/v) SDS/polyacrylamide gels overlaid with 4% (w/v) stacking gels [25]. Gels were stained with Coomassie blue and protein size was determined by comparison to molecular weight markers (Bio-Rad Laboratories, Mississauga, ON, Canada).

Proteins in SDS/polyacrylamide gels were transferred overnight to nitrocellulose (Bio-Rad) at 100 mA, and membranes were stained with 0.2% (w/v) Ponceau S (Sigma) in 3% (w/v) trichloroacetic acid to verify transfer. Membranes were blocked by incubation with gentle shaking in 5% (w/v) Carnation skimmed milk powder in TBS/Tween [10 mm Tris/HCl, 140 mm NaCl, 0.1% (v/v) Tween 20, pH 7.4] for 45 min, then incubated in primary antibody diluted in TBS/Tween for 15 min. Polyclonal antibodies raised in rabbits included an anti-peptide antibody to the N-terminal 15 residues of p49 [19] and an antibody prepared to native p49 during this study. Rabbits were obtained from Charles River Canada (St. Constant, QC, Canada) and cared for in accordance with guidelines in ‘Guide to the Care and Use of Experimental Animals’ available from the Canadian Council on Animal Care. The blots were washed twice for 3 min each in TBS/Tween and HST (10 mm Tris/HCl, 1 m NaCl, 0.5% Tween 20, pH 7.4) followed by 3 min in TBS/Tween. The membranes were incubated for 15 min with goat anti-(rabbit IgG) IgG horseradish-peroxidase conjugated secondary antibody (Jackson ImmunoResearch Laboratories, Inc., Bio/Can Scientific, Mississauga, Ontario, Canada). The enhanced chemiluminescence technique (PerkinElmer Life Sciences, Boston, MA, USA) was used for detection of antibody-reactive proteins.

Co-assembly of Artemia p49 and tubulin

Purified p49 at 0.5–1.0 µg·mL−1 and Artemia tubulin at 1.0 µg·mL−1[26], were incubated for 30 min at 37 °C with 1.8 mm GTP and 10 µm paclitaxel in final volumes of either 50 or 100 µL. Assembly conditions were the same for Artemia cell-free extract which was used at a final concentration of approximately 35 mg·mL−1. Assembly mixtures were centrifuged at 40 000 g for 30 min at 22 °C after overlaying on either 500 or 1000 µL 15% sucrose cushions in Pipes buffer. Pellets were rinsed gently with Pipes buffer at 37 °C, resuspended in 18 µL of the same buffer, and examined for tubulin and p49 by SDS/polyacrylamide gel electrophoresis followed by Western blotting. Microtubule cross-linking was detected by transmission electron microscopy with 5 µL samples of assembly mixtures fixed in 4% (v/v) glutaraldehyde applied to formvar-covered, carbon-coated, 200-mesh copper grids for 1 min. Excess liquid was removed by blotting with filter paper and grids were negatively stained with 1% (w/v) uranyl acetate for 30 s. Specimens were examined in a Philips Tecnai transmission electron microscope and images were captured with analysis, version 2.1.

Detection of a Drosophila p49 analogue

Drosophila cell-free extract was examined for p49 analogues by electrophoresis in SDS polyacrylamide gels and immunoprobing of Western blots using procedures described for Artemia. Drosophila tubulin was induced to assemble by paclitaxel addition to cell-free extracts and microtubules were collected by centrifugation through sucrose cushions. Pellets were rinsed, resuspended in Pipes buffer and processed for SDS/PAGE, immunoprobing of Western blots and electron microscopy as described earlier.

4-Hydroxybutyrate CoA-transferase assay

The presence of 4-hydroxybutyrate CoA-transferase activity was detected by formation of thiophenolate anion [27]. Reaction mixtures of 1.0 mL contained 100 mm KH2PO4, pH 7.0, 200 mm sodium acetate, 1 mm oxaloacetic acid, 1 mm 5,5′-dithiobis(2-nitrobenzoate), 0.1 mm butyryl CoA, 0.5 U citrate synthase (Sigma) and p49. Absorbance increase at 412 nm was measured at 20 °C and enzyme activity is reported in arbitrary units as ΔA412 min−1·mg protein−1.

Sequencing of p49 peptides

For sequencing by Edman degradation purified p49 was electrophoresed in 12.5% (w/v) SDS/polyacrylamide gels overlain with 4% (w/v) stacking gels and transferred to Immobilon poly(vinylidene difluoride) membrane (Millipore, Mississauga, ON, Canada) at 100 V for 1 h in 10 mm 3-(cyclohexylamino)-1-propane-sulfonic acid (CAPS) buffer, pH 10.5 containing 20% (v/v) methanol. The membranes were stained with Coomassie blue for 2 min, destained with 90% methanol/10% acetic acid (v/v), and rinsed with deionized water before drying. Edman sequencing was performed in a Hewlett Packard Model G1005A protein sequencer using the routine 3.1 PVDF program and analysis of PTH amino acids on line with a Hewlett Packard Model 1100 HPLC. When the sequence became difficult to read the sequencing cartridge contents were treated in situ with acetic anhydride to block partially degraded proteins at the amino terminus [28]. The acetylated proteins were cleaved at methionine residues with BrCN, excess reagent and reaction products were removed, and sequencing resumed.

Cloning and sequencing of p49 cDNA

Approximately 1.5 g (wet weight) of Artemia nauplii were homogenized in 0.5 mL TRIzol® reagent (Life Technologies, Boston, MA, USA) for 1 min in a glass homogenizer and RNA was recovered. Polyadenylated mRNA was purified with an mRNA Purification Kit (Amersham Biosciences, Baie d'Urfe, QC, Canada) and used with Ready-To-Go™ RT-PCR beads (Amersham Biosciences) to synthesize cDNA and amplify p49 cDNA. Fifty-microliter reaction mixtures contained 900 ng of poly(A)+ mRNA, 0.5 µg of oligo d(T)18 primer, 2 µm of each gene specific primer and 2.0 mm MgCl2. The mixtures were topped with 50 µL of mineral oil and reverse transcription was carried out at 42 °C for 30 min followed by 5 min at 95 °C. PCR amplification was performed immediately as follows: 95 °C for 45 s, 48 °C for 60 s, 72 °C for 90 s, sequentially for 35 cycles, and 10 min at 72 °C. Samples of reaction mixtures were diluted 1 : 100 in RNase-free water and amplified by nested PCR. Clone p49-1 was amplified by using degenerate primers 1a and 2 designed from peptide data, followed by seminested PCR using primers 1b and 2 (Fig. 1). The remaining 3′-p49 cDNA sequence, represented by Clone p49-2, was amplified with gene-specific primer 3a and an adaptor-coupled oligo d(T) primer, 5a, followed by seminested PCR using primer 3b and adapter primer 5b (Fig. 1). Clone p49-3 was amplified with primers 4a and 5a, followed by seminested PCR with primers 4b and adapter primer 5b (Fig. 1). PCR products were sized in 1% (w/v) agarose gels in TAE buffer (40 mm Tris/HCl, 20 mm acetic acid, 1 mm EDTA, pH 8.5) with a 1000 bp marker set (MBI Fermentas, Burlington, Ontario, Canada), then cloned with the pGEM®-T Easy Vector Systems Kit (Promega, Madison, WI, USA) and Escherichia coli JM109. Plasmid DNA was isolated using Wizard®Plus SV Minipreps DNA Purification System (Promega), followed by DNA digestion with EcoR1 and agarose gel electrophoresis to confirm insert size. Cloned DNA was sequenced at least twice in both directions at the DNA Sequencing Facility, Centre for Applied Genomics, Hospital for Sick Children, Toronto, ON, Canada.

Figure 1.

Cloning of p49 cDNA. p49 cDNA was cloned in three sections called clones p49-1, p49-2 and p49-3. Primer locations and names are indicated on the schematic and arrows indicate the 5′ to 3′ direction of each primer. Primer sequences are listed.

Sequence analysis of p49

Sequences obtained for p49 were compared to archived sequences at the National Center of Biological Information (NCBI) using Basic Local Alignment Search Tool (BLAST) [29], including blastx for DNA and blastp for proteins. Motif searches were performed with PROSITE database [30] using predict protein at the EMBL website, Heidelberg, Germany. Secondary structure predictions were made with Prof_ s, accessible via predict protein. Multiple alignments were performed with clustalx[31] with output files formatted by boxshade (http://www.ch.embnet.org/software/BOX_form.html). To examine evolutionary relationships, all 46 sequences in fasta format from the NCBI nonredundant protein database showing a high similarity to p49 were collected using the arbitrary cutoffs of E-value = 1e−50, and greater that 35% identity based on the observed distinct classes of similarity among all matches. Twenty-nine sequences remained after eliminating redundant entries representing partial sequences and splicing variants of the same gene. Sequence alignments and neighbor-joining trees were generated with clustalx using the Gonnet protein comparison matrix and 1000 bootstrap trials. The tree was viewed and printed with treeview[32].


Purification of p49

Purification of p49 to apparent homogeneity was obtained from Artemia cysts developed for either 6 or 12 h. Briefly, Artemia tubulin and MAPs, the latter defined by their ability to coassemble in vitro with tubulin [19], were induced to form microtubules by addition of paclitaxel and GTP to cyst cell-free extracts. After centrifugation of assembly mixtures through sucrose cushions, MAPs were recovered by incubating microtubule pellets in Pipes buffer containing 0.5 m NaCl. Enrichment of p49 was by heating MAPs to 50 °C for 5 min followed by centrifugation to remove precipitated proteins, chromatography on phosphocellulose P11 and (NH4)2SO4 fractionation. Only one weakly staining band of 49 kDa was observed in Coomassie blue stained gels after (NH4)2SO4 fractionation (Fig. 2A) and it interacted strongly with an antibody to native p49 even though there was almost no reaction in equivalent positions in lanes containing cell-free extract (Fig. 2B). A higher molecular mass protein of unknown identity that reacted with anti-p49 was observed routinely on Western blots containing cell-free extract, MAPs and heated MAPs, but only occasionally in more purified fractions. Approximately 0.2 mg of pure p49 was obtained from 2970 mg of starting protein. When the 49 kDa protein was incubated with Artemia tubulin at 37 °C in the presence of GTP and taxol, the resulting microtubules were cross-linked by irregularly shaped, randomly distributed particles (Fig. 2C).

Figure 2.

Purification of p49. Protein fractions obtained during p49 purification were electrophoresed in 12.5% (w/v) SDS/polyacrylamide gels and either stained with Coomassie blue (A) or blotted to nitrocellulose and immunostained with anti-(native p49) by enhanced chemiluminescence (B). (A) Lane 1, 60 µg of Artemia cell-free extract; lane 2, 35 µg of MAPs; lane 3, 35 µg of heated MAPs; lane 4, 2.5 µg of 0.2 m NaCl fraction from P11; lane 5, 1 µg of purified p49. (B) Lanes 1–3 each received 10 µg of protein; lane 4, 2.5 µg of the 0.2 m NaCl fraction from P11; lane 5, 1 µg of purified p49. Because low yields for the final two purification steps precluded accurate determination, protein amounts reported for lanes 4 and 5 were estimates based on staining intensity of bands. Lane M, molecular mass markers × 10−3; arrows, p49; arrowhead, cross-reactive protein. (C) Purified tubulin was assembled in the presence of p49, microtubules were collected by centrifugation, resuspended in Pipes buffer, applied to grids and negatively stained with 1% (w/v) uranyl acetate. Arrows, p49 microtubule cross-linking particles. The bar represents 200 nm.

Sequencing of p49

Fifty eight cycles of Edman degradation were performed on p49 blotted to poly(vinylidene difluoride) membrane. The first 50 cycles yielded the sequence FYSYSQEPFHPIQGRSPKWTSLEDSVKAVRSGDTVFVHsaaxtpxxxlxa with some residues either not determined (x), or assigned tentatively (lower case). Residues 51–58 could not be assigned. The protein sample, after sequential treatment with acetic anhydride and BrCN, gave a readable sequence from the seventh Edman cycle onward, although the identity of some residues was uncertain QVDFLRGAAIxPEAGXPILALPATTxRGES.

cDNA for p49 was cloned in sections with PCR amplification of first strand cDNA achieved by the use of degenerate oligonucleotide primers 1a and 2 (Fig. 1) designed on the basis of peptide sequence. Seminested PCR was then performed with primers 1b and 2, giving Clone p49-1, a 1107 bp DNA fragment encoding 369 amino acid residues (Fig. 3). Peptides identified by Edman degradation were encoded by Clone p49-1. Clone p49-2, containing the rest of the p49 cDNA, was obtained by PCR amplification. The sequences of these clones were partially confirmed by analysis of Clone p49-3. The assembled p49 cDNA, deposited in GenBank under the accession number AY304544, was 1411 bp and it contained an ORF of 1320 bp encoding 440 amino acid residues. The ORF was flanked by a stop codon (TAA) composed of nucleotides 1321–1323, followed by a 3′ noncoding region of 70 nucleotides containing the poly adenylation signal AAGTAAA and a poly(A) tail of 18 nucleotides. The combined cDNAs represent the complete p49 sequence, except for six N-terminal residues determined only by Edman degradation (Fig. 3). The initiator methionine was not observed, indicating removal of the residue during protein maturation. The calculated molecular mass of the protein was 48.3 kDa, in agreement with SDS/PAGE. Motif searches with PredictProtein E-mail Server and gene runner displayed several putative phosphorylation motifs recognized by different classes of kinases, but no typical microtubule binding regions (Fig. 3, Table 1).

Figure 3.

Nucleotide and amino acid sequences of p49. The nucleotide sequence of p49 was obtained as described in Experimental procedures, and from this the amino acid sequence was deduced. Amino acid residues determined by Edman degradation are in bold. The termination codon (TAA) is underlined and the polyadenylation signal is boxed. Putative phosphorylation motifs are underlined in the deduced amino acid sequence. The first six amino acid residues, revealed only by peptide sequencing, are in brackets.

Table 1. Kinase recognition motifs in p49. Protein motif searches were performed with gene runner (Hastings, Inc) and the PROSITE database (40) using the predict protein E-mail server at EMBL, Heidelberg, Germany.
Kinase classMotif sequence and position
cAMP/cGMP-dependent protein kinase56 KKSS 59
Protein kinase C16 SPK 18
25 SVK 27
59 SLK 61
198 TTK 200
284 SKK 286
371 TTK 373
391 TTR 393
412 SLR 414
Casein kinase II20 TSLE 23
134 SPPD 137
172 TFGD 175
181 SHFD 184
202 TDVE 205
207 TIGE 210
322 SCIE 325

Identification of p49 as a CoA-transferase by sequence analysis

p49 has significant sequence similarity to CoA-transferases encoded in the genomes of D. melanogaster, Anopheles gambiae, and Caenorhabditis elegans(Fig. 4). C. elegans has two members in this gene family with 52% identity to one another. Analysis with the Conserved Domain Database disclosed a region of p49 beginning at residue 84 with 64.3% similarity to the acetyl CoA hydrolase/transferase domain, indicating that p49 belongs to this family. Phylogenetic analysis exposed evolutionary relationships between p49 and other proteins and revealed distinctive phylogenetic protein groups (Fig. 5). p49 and the homologous invertebrate sequences formed a subcluster within a main branch with p49 positioned between C. elegans and insect proteins, this in line with established lineage relationships. Highly similar p49 homologs exist in many prokaryotic species, including the archaebacteria and eubacteria (Fig. 5). Homologues of lower similarity levels are present in many of the species represented in Fig. 5, in several other bacteria species and in the Saccharomyces cerevisiae and Schizosaccharomyces pombe genomes, suggesting multiple subfamilies within the large hydrolase/CoA-transferase family. All purification fractions displayed 4-hydroxybutyrate CoA-transferase activity ranging from 0.22 units for the cell-free extract, to 0.88 units for heated MAPs and 0.51 units for purified p49.

Figure 4.

Sequence alignment of p49 and related invertebrate proteins. The deduced amino acid sequence of p49 was aligned by clustalw with the following proteins: D_mel (NP_651762.1 Drosophila melanogaster), A_gam (EAA09276.1 Anopheles gambiae str. PEST with three residues, ‘SEK’, at the N-terminal removed based on the alignment and annotation practice by Celera of not defining the start codon), C_ele1 (AAN63431.1 representing partial sequence of NP_495409.2 which has 261 additional residues at the C-terminus, Caenorhabditis elegans), C_ele2 (CAA87047.1 Caenorhabditis elegans). Black, identical residues; grey, similar residues; no shading, different residues.

Figure 5.

Phylogenetic tree of p49-related CoA-transferases. The following bacterial proteins were used, in addition to the proteins in Fig. 5: M_sp. (ZP_00042556.1, Magnetococcus sp. MC-1), C_tep (AAM71277.1, Chlorobium tepidum TLS), L_int (AAN51814.1, Leptospira interrogans serovar), SS_one (ANN54762.1, Shewanella oneidensis), C_klu (AAA92344.1, Clostridium kluyveri), C_ami (CAB60036.1, Clostridium aminobutyricum), C_tet (AAO35111.1, Clostridium tetani), F_nuca (NP_603518.1, Fusobacterium nucleatum ssp. nucleatum ATCC 25586), F_nucb (EAA24344.1, Fusobacterium nucleatum ssp. vincentii ATCC 49256), S_sp. (BAA17706.1, Synechocystis sp.), G_met1 (ZP_00079959.1, Geobacter metallireducens), G_met2 (ZP_00082143.1, Geobacter metallireducens), G_met3 (ZP_00082133.1, Geobacter metallireducens), G_met4 (ZP_00080028.1, Geobacter metallireducens), T_10 (AAM23830.1, Thermoanaerobacter tengcongensis), D_haf1 (ZP_00099788.1, Desulfitobacterium hafniense), D_haf2 (ZP_00099512.1, Desulfitobacterium hafniense), D_hal3 (ZP_00098805.1, Desulfitobacterium hafniense), A_ful1 (AAB90101.1Archaeoglobus fulgidus), A_ful2 (AAB89400.1 Archaeoglobus fulgidus), N_aro (ZP_00095224.1, Novosphingobium aromaticivorans), Y_pes (CAA21375.1, Yersinia pestis), S_ent (AAK97550.1, Salmonella enteritidis), R_pal (ZP_00009324.1, Rhodopseudomonas palustris), B_jap (BAC52055.1, Bradyrhizobium japonicum). Bootstrap values above 700 (70%) out of 1000 trees are indicated at the nodes. The branch length is proportional to distance. The subbranch for sequences from invertebrate species is shaded.

Higher order structure of p49

The secondary structures of p49, 4-hydroxybutyrate CoA-transferase from Clostridium aminobutyricum and glutaconate CoA-transferase from Acidaminococcus fermentans were predicted with Prof_s (Fig. 6). Notwithstanding limited sequence similarity, p49 and the C. aminobutyricum 4-hydrodybutyrate CoA-transferase have related secondary structure predictions, an observation which correlates with their structural and functional similarities. A striking congruence between the two proteins is alternation of relatively short stretches of α-helical and β-sheets throughout much of their lengths. Additionally, the glutaconate CoA-transferase, for which the crystal structure has been determined, is only weakly related in sequence to p49 and C. aminobutyricumm 4-hydroxybutyrate CoA-transferase, but predicted secondary structures are similar. The tertiary structure of p49 is uncertain, although secondary structure predictions suggest similarities between the CoA-transferases. The size of microtubule cross-linking particles in concert with molecular mass measurements indicate p49 forms a homomultimeric complex of 10–20 subunits, but how monomers self-associate is not apparent.

Figure 6.

Predicted secondary structures of p49 and bacterial CoA-transferases. The secondary structures of p49 from A. franciscana (p49), hydroxybutyryl CoA-transferase from C. aminobutyricum (HBCoA) predicted according to Profile network prediction HeiDelberg (Prof_ s) accessible via PredictProtein, and glutaconate CoA-transferase A chain (GlutCoA) from A. fermentans derived from its crystal structure (NCBI Structure entry POIA) were compared. The number of residues for each sequence is indicated on the right side of the figure. Single underline, α-helix; double underline, β-sheet.

Drosophila contain a p49 analogue

A protein of 49 kDa was detected on Western blots containing cell-free extract from Drosophila adults but not from embryos, larvae and pupae (Fig. 7). The Drosophila 49 kDa protein often appeared as a doublet and reaction with anti-p49 was stronger than for Artemia p49 in cell-free extract. As demonstrated by immunoprobing of blots, the Drosophila p49 analogue coassembled with taxol-induced microtubules, albeit in reduced quantity as compared to Artemia microtubules (Fig. 8). Small amounts of Drosophila tubulin and the p49 analogue were detected under control conditions, even when reaction mixtures were centrifuged prior to assembly and incubation time was shortened, perhaps due to limited tubulin polymerization. The mAb DM1A, directed against tubulin, detected tubulin and a polypeptide lower in molecular mass than tubulin thought to be a proteolytic degradation product. Microtubules assembled in Drosophila cell-free extract were distributed sparsely on grids with no evidence of cross-linking.

Figure 7.

Detection of a Drosophila p49 analogue. Cell-free extracts from Drosophila and Artemia were electrophoresed in 12.5% SDS polyacrylamide gels and either stained with Coomassie blue (A) or blotted to nitrocellulose and probed with anti-native p49 antibody (B). Panels A and B, lane 1, 60 µg of Artemia cell-free extract protein; lane 2, 1.0 µg of purified p49; lanes 3–6, 60 µg of Drosophila cell-free extract from embryos, larvae, pupae and adults, respectively. Molecular mass markers are shown in lane M and represent 97, 66, 43, 31 and 22 kDa. Arrow, p49; arrowhead, cross-reactive high molecular mass protein.

Figure 8.

Coassembly of Drosophila tubulin and the p49 analogue. Tubulin in Artemia and Drosophila cell-free extracts was assembled by the addition of taxol and GTP. Microtubules were collected by centrifugation through sucrose, resuspended in Pipes buffer, electrophoresed in 12.5% SDS polyacrylamide gels and either stained with Coomassie blue (A) or blotted to nitrocellulose and stained with anti-tubulin mAb, DM1A (B), or anti-p49 (C). In all panels, lane 1, complete assembly reaction with Artemia cell-free extract; lane 2, assembly reaction lacking GTP and taxol with Artemia cell-free extract; lane 3, complete assembly reaction with Drosophila cell-free extract; lane 4, assembly reaction lacking GTP and taxol with Drosophila cell-free extract. All assembly mixtures contained 630 µg of protein in a final volume of 50 µL. Molecular mass markers represent 97, 66, 43, 31, 22 and 14 kDa. Arrow, p49; tub, tubulin; arrowhead, cross reactive high molecular mass protein.


A 49 kDa protein, termed p49, was purified to apparent homogeneity from Artemia developed either 6 or 12 h, and cross-linked microtubules were produced when the protein was incubated with Artemia tubulin. This observation, coupled with the finding that an antibody raised previously to a 49 kDa microtubule cross-linking protein recognized p49 (not shown), and the N-terminal 15 amino acid residues of both proteins were identical, demonstrated p49 is the protein described by Zhang and MacRae [19–21]. Sequencing by Edman degradation yielded amino terminal and internal peptides essential to primer design for PCR amplification of p49 cDNA. Clone p49-1, a cDNA fragment of 1107 bp beginning near the N-terminus and representing about 84% of p49, was obtained initially. Clone p49-2, which overlapped with the 3′ end of clone p49-1 and contained the remaining p49 sequence, included the polyadenylation signal, AAGTAAA. Clone p49-3 overlapped partially with clones p49-1 and p49-2, confirming a portion of each sequence. According to motif analysis, p49 has several phosphorylation sites, in line with the presence of two phosphorylated p49 isoforms [21], and suggesting how protein function is regulated. p49 lacks microtubule binding domains typical of MAP2, MAP4 and tau.

Comparison of the deduced amino acid sequence to archived sequences demonstrated p49 is a CoA-transferase family member. Representatives of this family in the anaerobic bacteria C. aminobutyricum and Clostridium kluyveri[27,33], catalyze the formation of 4-hydroxybutyryl CoA from 4-hydroxybutyrate, using either butyryl-CoA or acetyl-CoA as coenzyme A donors in a fully reversible process thought to be important in meeting the energy needs of these anaerobic organisms. The signature motif EXG, located near the C-terminus of CoA-transferases, and encompassed by residues 402–404 in p49, may play a critical role in the catalytic formation of a thiol ester between glutamate and the substrate CoAS-moiety [34]. Propionate CoA-transferase from Clostridium propionicum was rapidly inactivated by borohydride mediated modification of Glu324 in the presence of propionyl CoA [35]. Glu324 corresponds to p49 Glu402, suggesting residues 402–404 of p49 are important catalytically and both proteins have similar reaction mechanisms. Purified p49 exhibited 4-hydroxybutyrate CoA-transferase activity, reinforcing the conclusion that the protein belongs to the CoA-transferase family. Because the intent was to demonstrate enzyme activity and low yields precluded extensive analysis, assays were not optimized nor were other potential substrates determined. Of interest, however, purified p49 was less active than heated MAPs, suggesting the loss during purification of a cofactor required for maximal enzyme activity. No other descriptions of CoA esters and their hydrolysis products are, to our knowledge, available for Artemia.

Secondary structure predictions for p49, 4-hydroxybutyrate CoA-transferase from A. aminobutyricum and glutaconate CoA-transferase from A. fermentans resemble one another and CoA-transferases are generally thought to have similar tertiary structures even though sequence identity is limited. As one example, the crystal structure of glutaconate CoA-transferase indicates a globular protein accommodating many secondary structural elements, in which β-strands form a barrel-like structure [36]. The quaternary structure of p49 probably includes 10–20 subunits, an estimate based on microtubule cross-linking particle size and monomer molecular mass. In comparison, C. aminobutyricum 4-hydroxybutyrate CoA-transferase is a homodimer [27], and other bacterial CoA-transferases organize into heterooctomers [37].

Most species displayed in the phylogenetic tree (Fig. 5) have a single gene, but the bacterium Geobacter metallireducens has four CoA-transferase genes, the largest number known. Three family members arose by recent gene duplications, as indicated by identities geqslant R: gt-or-equal, slanted 77% and membership in the same phylogenetic tree subbranch. The large gene family may relate to the ability of G. metallireducens to live in extraordinarily high iron concentrations, suggesting a role for CoA-transferases in metal metabolism or detoxification. Desulfitobacterium hafniense, capable of reductive dechlorination of hydrocarbons and use of sulfite and thiosulfate as terminal electron acceptors, has three gene family members, with two probably from a recent duplication.

The Drosophila genome encodes 4-hydroxybutyrate CoA-transferase that is analogous to p49 but the protein has not been characterized. Drosophila cell-free extract from adult flies contains a protein that reacts strongly with anti-p49, indicating it is 4-hydroxybutyrate CoA-transferase, but it is lacking from embryos, larvae and pupae. The results contrast with the situation in Artemia where p49 is expressed in encysted embryos and early larvae. Microtubules assembled in Drosophila cell-free extract associate with a 49 kDa protein, but to a lesser degree than for Artemia and they are distributed sparsely on grids with no evident cross-linking. These differences, perhaps reflecting protein sequence variation, indicate the Drosophila analogue is less dependent than p49 on microtubules for spatial organization. p49 displays a weak nucleotide-independent nucleotidase [20] and there are no intact nucleotide binding sites in p49. The Drosophila gene product has a putative nucleotide recognition site encompassing residues 79–86, and it may bind GTP efficiently, thus causing protein dissociation from microtubules. This is the first time Drosophila 4-hydroxybutyrate CoA-transferase has been shown to associate with microtubules.

Microtubules organize many proteins in the cytoplasm, and one example is the nucleotide–dependent association of enolase with these polymers [14]. Glyceraldehyde-3-phosphate dehydrogenase, involved in transporting vesicular tubular clusters between the endoplasmic reticulum and Golgi [38,39], binds to microtubules in a phosphorylation-dependent mechanism [15]. Hexokinase, a key enzyme in glucose metabolism, associates with brain microtubules [40]. Signaling molecules such as the Rho family of kinases engage microtubules [16,17], as does the tumor suppressor protein p53 [41] and the transcriptional coordinator P/CIP [42]. Clearly, microtubules recognize many cytoplasmic proteins in addition to those classically designated as MAPs, and some associations have functional implications, as may be reflected in the relationship between p49 and microtubules.


The authors thank Dr Robert Schultz, National Cancer Institute, Bethesda, MD, USA, for the generous gift of paclitaxel and Dr Vett Lloyd, Dalhousie University, for supplying Drosophila. The work was supported by a Natural Sciences and Engineering Research Council of Canada Discovery Grant to THM.