Expansion of genes encoding a novel type of dynamin in the genome of the pea aphid, Acyrthosiphon pisum


Atsushi Nakabachi, Advanced Science Institute, RIKEN, Wako, Saitama 351-0198, Japan. Tel.: +81 48 4679332; fax: +81 48 462 9329; e-mail: bachi@riken.jp


Screening of the entire genome of the pea aphid, Acyrthosiphon pisum, detected 15 genes for putative dynamin superfamily proteins – self-assembling large GTPases that are involved in the fission and fusion of membranes. In addition to a single gene each for Dyn, Drp1, and Opa1, orthologues that are common in Metazoa, 12 genes encoding a novel type of dynamin were found. Phylogenetic analyses showed that these novel-class genes are monophyletic. Quantitative reverse transcription-PCR demonstrated that expressions of four novel-class dynamin genes are highly up-regulated in the midgut, through which aphids take in phloem-sap diets and plant viruses. As this type of dynamin is absent from all other fully sequenced organisms, they may function in processes unique to aphids.


Dynamins are self-assembling large GTPases that are ubiquitously distributed in eukaryotic organisms. They form a large superfamily and are involved in various processes that are associated with membrane dynamics, including division of organelles such as mitochondria and plastids, endocytosis, cytokinesis, and budding of transport vesicles (Praefcke & McMahon, 2004). Their basic function appears to be to constrain the membrane. Animals, including insects, typically have three types of dynamins – the classical dynamin (Dyn), dynamin-related protein 1 (Drp1), and optic atrophy 1 (Opa1) – which are involved in endocytosis and cytokinesis, mitochondrial and peroxisomal division, and mitochondrial fusion, respectively (Miyagishima et al., 2008). In many cases, a single gene corresponds to each type.

Aphids are the primary vectors of plant viruses. Of the 697 plant virus species recognized by the International Committee on Taxonomy of Viruses, aphids transmit 197 species (28%) (Hogenhout et al., 2008), which makes them notorious agricultural pests. Whereas some viruses are transmitted from plant to plant simply by binding to the cuticular lining of the aphid mouthparts or the foregut, certain plant viruses need to enter the aphid haemocoel before being transmitted to uninfected host plants (Gray & Gildow, 2003; Hogenhout et al., 2008). Since dynamins are essential for the infection of various viruses in animal hosts (Cherry & Perrimon, 2004; Marsh & Helenius, 2006), it is plausible that dynamins are involved in the infection and transfer of plant viruses in aphids.

Furthermore, aphids harbour an obligate mutualist, Buchnera aphidicola (γ-Proteobacteria), within the cytoplasm of specialized cells called bacteriocytes (Buchner, 1965; Moran et al., 2008; Nakabachi, 2008). Since the initial infection more than 100 million years ago (Moran et al., 1993), as in the case of organelles such as mitochondria and chloroplasts, Buchnera have been subjected to strict vertical transmission through generations of hosts, and the mutualism between Buchnera and their host has evolved to the point that neither can reproduce in the absence of the other (Febvay et al., 1995; Sasaki & Ishikawa, 1995; Nakabachi & Ishikawa, 1997; Douglas, 1998; Nakabachi & Ishikawa, 1999). Since each Buchnera cell is encased in a membrane of host origin (Hinde, 1971; Griffiths & Beck, 1973), and dynamins are essential for the maintenance of endosymbiont-derived organelles such as mitochondria and plastids (Praefcke & McMahon, 2004; Miyagishima et al., 2008), it is also plausible that aphids make use of dynamins to facilitate symbiosis with Buchnera.

The genome of the pea aphid, Acyrthosiphon pisum, was screened for genes encoding dynamin superfamily proteins. We further analysed the expression profiles of all of the identified A. pisum dynamin genes in the whole body, the bacteriocyte, the embryo, and the midgut, using the real-time quantitative RT-PCR technique.


Similarity searches detected 15 dynamin genes in the pea aphid genome

BLASTP searches of the A. pisum Gnomon-predicted protein database detected 15 gene models for putative dynamin superfamily proteins, whereas BLASTP searches of the A. pisum RefSeq protein database detected only 14 (Table 1). This was because the pipeline of the National Center for Biotechnology Information (NCBI) to generate the RefSeq data set detected a suppressed stop codon near the end of the CDS that corresponds to the locus LOC100168088 (Gnomon prediction ID: hmm664634) and annotated this locus as a pseudogene. However, the manual checking of trace sequences of the A. pisum genome and the orthology-based protein evidence of closely related organisms strongly suggest that a frameshift that causes the suppressed stop codon was erroneously introduced through sequencing and assembly errors, and that this locus encodes true protein(s) (T. Murphy, pers. comm.). Moreover, LOC100168088 is the only candidate in A. pisum that can encode an orthologue for classical dynamins, which are found in all of the Metazoa sequenced to date (see below), further supporting the functionality of this locus. TBLASTN searches of A. pisum genome scaffolds (Acry_1.0) detected no further candidates.

Table 1.  Genes for putative dynamins detected in the Acyrthosiphon pisum genome
Gene symbolGene nameGene IDsRefSeq mRNARefSeq proteinProtein length (aa)Note
DynDynaminhmm664634, LOC100168088XR_045901.1(747->) 874frameshift corrected
Drp1Dynamin-related protein 1ACYPI007635, LOC100166793XM_001949334.1XP_001949369.1705 
AcypiDlp1Dynamin-like protein 1ACYPI010125, LOC100169513XM_001945309.1XP_001945344.1(658->) 6575′-end corrected
AcypiDlp2Dynamin-like protein 2ACYPI005508, LOC100164495XM_001943916.1XP_001943951.1666 
AcypiDlp3Dynamin-like protein 3ACYPI005629, LOC100164635XM_001950654.1XP_001950689.1663 
AcypiDlp4Dynamin-like protein 4ACYPI007980, LOC100167166XM_001946739.1XP_001946774.1665 
AcypiDlp5Dynamin-like protein 5ACYPI003894, LOC100162764XM_001952144.1XP_001952179.1(653->) 6713′-end corrected
AcypiDlp6Dynamin-like protein 6ACYPI007504, LOC100166649XM_001944952.1XP_001944987.1691 
AcypiDlp7Dynamin-like protein 7ACYPI009835, LOC100169195XM_001949539.1XP_001949574.1681 
AcypiDlp8Dynamin-like protein 8ACYPI006414, LOC100165463XM_001942959.1XP_001942994.1641 
AcypiDlp9Dynamin-like protein 9ACYPI007252, LOC100166374XM_001946251.1XP_001946286.1633 
AcypiDlp10Dynamin-like protein 10ACYPI002209, LOC100160943XM_001943516.1XP_001943551.1496 
AcypiDlp11Dynamin-like protein 11ACYPI009671, LOC100169013XM_001945961.1XP_001945996.1643 
AcypiDlp12Dynamin-like protein 12ACYPI004494, LOC100163403XM_001943013.1XP_001943048.1384 
Opa1Optic atrophy1-like proteinACYPI000326, LOC100158908XM_001950135.1XP_001950170.1946 

Acyrthosiphon pisum possesses 12 genes encoding a novel type of dynamin

Molecular phylogenetic analyses were performed to further identify the 15 dynamins of A. pisum. Amino acid sequences of highly conserved GTPase domains and moderately conserved middle domains (Praefcke & McMahon, 2004; see below) of dynamin superfamily proteins were aligned and used for the analyses. Phylogenetic trees were inferred by the maximum likelihood (ML) and Bayesian (BI) methods. For the analyses, data of essentially all fully-sequenced Metazoa, together with various other lineages of organisms, were used as references. Subsequently, only non-redundant and fully informative data were selected (e.g. only Homo sapiens was used as a representative of mammalians. The taxa used are listed in Supporting Information Table S1) and were used to construct the final tree (Fig. 1). The analyses demonstrated with robust statistical support that three (LOC100168088, ACYPI007635, and ACYPI000326) (Table 1) out of the 15 A. pisum dynamins are orthologues of Dyn, Drp1 and Opa1, respectively, which are ubiquitously distributed in Metazoa (Fig. 1, Supporting Information Table S1). Statistical values (boot strap percentage for ML/posterior probability percentage for BI) to support the monophyly of insect Dyn and Drp1, including those for A. pisum, were (97/100) and (84/99), respectively. Statistical values in support of the monophyly of metazoan Dyn, Drp1, and Opa1 were (100/100) (92/99), and (100/100), respectively. Overall, the phylogenetic relationships appeared to be congruent with the species tree of Metazoa, which places the pea aphid with Pediculus humanus, another member of the Paraneoptera clade, basal to the Holometabola (International Aphid Genomics Consortium, 2010), although the Opa1 of A. pisum was not nested within the clade of arthropods (Fig. 1). Dyn (LOC100168088, 874 aa), Drp1 (ACYPI007635, 705 aa), and Opa1 (ACYPI000326, 946 aa) of the pea aphid are 76.2, 70.2, and 53.2% identical to their orthologues in Drosophila melanogaster, i.e. Shibire (CG18102, isoform I, NP_001036279.1, 830 aa), Drp1 (CG3210, NP_608694.2, 735 aa), and Opa1 (CG8479, NP_725369.1, 972 aa), respectively (for alignments, see Supporting Information Figs S1–S3).

Figure 1.

Phylogenetic tree of dynamins. A Bayesian (BI) tree is shown, while the maximum likelihood (ML) tree exhibited substantially the same topology. Bootstrap support values over 50% (left) and Bayesian posterior probability percentages over 95% (right) are shown at the nodes. Dashes indicate statistical values of less than 50 (ML) or 95 (BI). Branch lengths are proportional to the number of amino acid substitutions, which are indicated by the scale bar below the tree. The names of proteins or GenBank accession numbers are shown with the organism names. The Acyrthosiphon pisum dynamins are shown in red.

All of the remaining 12 dynamins that were found in A. pisum (ACYPI010125, ACYPI005508, ACYPI005629, ACYPI007980, ACYPI003894, ACYPI007504, ACYPI009835, ACYPI006414, ACYPI007252, ACYPI002209, ACYPI009671, and ACYPI004494) (Table 1) appeared to form a novel class (Fig. 1). The phylogenetic tree provided robust statistical support (100% both in ML and BI) that they are monophyletic and no other proteins from any sequenced organisms fell within this clade, implying that the novel class genes relatively recently emerged from a single aphid gene. It appears that these genes basally diverged from the metazoan Drp1 genes, although statistical support for the grouping was weak (64% in ML, whereas 100% in BI). The novel class genes appeared to form a cluster with the Dyn, Drp1, and Vps1 genes that are involved in the endocytosis and cytokinesis, mitochondrial and peroxisomal division, and vesicle trafficking from the Golgi complex, respectively. This topology was supported by moderate statistical values (78% in ML, 100% in BI), but their phylogenetic relationships within this clade were not clearly resolved (Fig. 1).

Structure of aphid dynamins

Structural analyses were performed to further characterize the A. pisum dynamins. The architectural features common to dynamin superfamily proteins which are distinct from other GTPases are the structure of the large GTPase domain (∼300 amino acids) and the presence of two additional domains: the middle domain and the GTPase effector domain (GED), which are involved in oligomerization and regulation of the GTPase activity (Praefcke & McMahon, 2004). The GTPase domain contains the GTP-binding motifs (G1-G4) needed for guanine-nucleotide binding and hydrolysis. In addition to these, the classical dynamins have two more domains that are involved in targeting. These are the Pleckstrin-homology (PH) domain, which interacts with various phospholipids, and the proline-rich domain (PRD), which binds to SRC-homology (SH3) domains of other proteins. Dynamin superfamily proteins other than the classical dynamins may lack one or more domains, or have additional domains that are not present in the classical dynamins (Praefcke & McMahon, 2004).

The conserved domain (CD-)search at the NCBI website and the Pfam domain search detected a GTPase domain, a middle domain, a PH domain, and a GED in the A. pisum Dyn protein (Supporting Information Fig. S1). The GED was followed by a proline-rich region (32/116 aa = 28%; proline residues are denoted with asterisks in Supporting Information Fig. S1). Among 32 proline residues, 16 residues (50%) are conserved among the A. pisum Dyn, D. melanogaster Shibire, and H. sapiens Dyn1, Dyn2, and Dyn3. This domain structure (the full set of five distinct domains) of the A. pisum Dyn typifies the classical dynamins, further supporting the results of the phylogenetic analyses. The GTPase domains, middle domains, and GEDs were detected in the Drp1 protein and the novel type dynamins of A. pisum, although Dlp10 and Dlp12 appeared to lack GED, as their carboxyl termini are somewhat shorter than the others (Fig. 2, Supporting Information Fig. S2). This similarity in domain structures of the A. pisum Drp1 and novel class dynamins is consistent with their relatively close relationship implied by the phylogenetic analyses. The domain searches detected only a GTPase domain and a middle domain in the A. pisum Opa1 (Supporting Information Fig. S3). GED, one of three components common in dynamin superfamily proteins, was not detected. This was also true for Opa1 proteins of D. melanogaster and H. sapiens, which might be due to the low sensitivity of the searches and/or the highly divergent GED structure in Opa1 proteins.

Figure 2.

Structure of the novel class dynamins of Acyrthosiphon pisum. Amino acid sequences of novel class dynamins of A. pisum are aligned with the Drp1 sequence of Drosophila melanogaster using ClustalX. Residues conserved in all lineages, 80% of lineages, and 60% of lineages are shaded in black, dark grey, and light grey, respectively. Residues contributing to the GTPase domain, middle domain, GTPase effector domain (GED), and GTP-binding motifs (G1-G4) within the GTPase domain, are boxed. Amino acid residues that are required for the GTPase activity are indicated with arrowheads.

All the 15 dynamins of A. pisum have four conserved GTP-binding motifs (G1-G4) within their GTPase domains. The G1 motif (in the so-called P-loop) coordinates the phosphates, whereas the threonine in the G2 motifs is involved in catalysis. The glycine in the G3 motif forms a hydrogen bond with the γ-phosphate of GTP. The G4 motif is involved in base and ribose coordination (Praefcke & McMahon, 2004). The key residues (indicated with arrowheads in Fig. 2, and the Supporting Information Figs S1–S3) that are required for GTPase activity are conserved in all cases, suggesting that all the A. pisum dynamins have GTPase activity.

SignalP detected no significant evidence for the presence of signal peptides in the 15 dynamins of A. pisum. TargetP detected a transit peptide at the N-terminus of the A. pisum Opa1 (1–35 aa), predicting that this protein localizes in mitochondria (the highest neural network score of TargetP = mTP; 0.733) (Supporting Information Fig. S3). TMHMM Server ver. 2.0 and coiled-coil predictions at the European Molecular Biology Laboratory (EMBL) further detected a single transmembrane region and two coiled-coil structures, respectively, in the A. pisum Opa1. These features are common to Opa1 orthologues of various animal lineages (Praefcke & McMahon, 2004). TargetP predicted that all the other A. pisum dynamins are localized in the cytoplasm.

Expression profiles of dynamin genes in Acyrthosiphon pisum

To examine the expression profiles of the 15 dynamin genes identified in the genome of the pea aphid, we quantified their transcripts in the whole bodies, bacteriocytes, embryos, and midguts of A. pisum by using the real-time quantitative RT-PCR technique. Preceding RT-PCR amplified all of the target cDNAs other than that for Dlp12. As the genomic PCR successfully amplified the Dlp12 genomic sequence, thereby verifying the effectiveness of the primer set, the failure in the RT-PCR amplification implied that this locus is not transcribed at a significant level. The expression levels of other 14 genes were further analysed.

The analyses demonstrated that the expression levels of two canonical dynamin genes, Drp1 and Opa1, and four novel class dynamin genes, Dlp5, Dlp6, Dlp7, and Dlp10, were significantly higher in the midgut than in the other organs (Fig. 3). The transcripts for Drp1, Dlp5, Dlp6, Dlp7, Dlp10, and Opa1 were 2.38, 29.7, 2.61, 94.6, 2.80, 2.70-fold more abundant in the midgut than in the whole body, respectively (P < 0.001, one-way anova followed by Tukey-Kramer test). The analyses also showed that the overall expression level of the A. pisum Dyn was relatively high among the 15 A. pisum dynamins, and that this gene is more highly expressed in the embryo and in the midgut (Fig. 3). The transcript for the A. pisum Dyn was 1.464 and 1.461-fold more abundant in the embryo and in the midgut, respectively, than in the whole body (P < 0.001, Tukey-Kramer test). Because endocytosis and cytokinesis are expected to be active in the midgut and in the embryo, respectively, this observation is consistent with the previous report that Dyn proteins of various organisms are involved in endocytosis and cytokinesis (Praefcke & McMahon, 2004; Miyagishima et al., 2008).

Figure 3.

Expression of the pea aphid dynamin genes. Ivory, blue, green, and orange columns represent expression levels in the whole body, the bacteriocyte, the embryo, and the midgut, respectively; bars, standard errors (n= 6). The expression levels are shown in terms of mRNA copies of the target genes per copy of mRNA for RpL7. Asterisks indicate statistically significant differences (Tukey-Kramer test; *P < 0.05; **P < 0.01; ***P < 0.001).

None of the transcripts for the A. pisum dynamins was more abundant in the bacteriocyte than in the other organs.


Screening of the genome of the pea aphid, Acyrthosiphon pisum, revealed it to have 15 genes for putative dynamin superfamily proteins. This is the largest number of dynamin genes detected in any single organism reported to date. Metazoa, including insects, typically have only three dynamin genes, namely Dyn, Drp1, and Opa1, although Dyn is triplicated in mammalians, and several copies of Mx, another class of dynamin superfamily gene, are present in vertebrates (Praefcke & McMahon, 2004; Miyagishima et al., 2008). Among the 15 dynamin genes of A. pisum, 12 appeared to form a novel class, whereas three genes were orthologues for canonical dynamins, Dyn, Drp1, and Opa1. Phylogenetic analyses suggested a close relationship of the novel class of dynamins with the Dyn, Drp1, and Vps1 proteins, which are involved in endocytosis and cytokinesis, mitochondrial and peroxisomal division, and vesicle trafficking from the Golgi complex, respectively, implying that the novel dynamins are also involved in similar processes.

Quantitative RT-PCR demonstrated that four of the 12 novel class dynamin genes as well as the canonical dynamin genes, Dyn, Drp1 and Opa1, are highly expressed in the midgut, through which aphids take in a phloem-sap diet and plant viruses. This may suggest that these dynamins are involved in the ingestion of phloem sap and/or plant viruses in aphids.

Aphids are responsible for transmission of 28% of all plant viruses (Hogenhout et al., 2008), which makes them notorious agricultural pests worldwide. Some viruses are transmitted in the ‘stylet-borne’ or ‘foregut-borne’ manner, and do not enter the aphid haemocoel, but in the ‘circulative’ and ‘propagative’ modes, viruses are transmitted in the following manner: (1) During phloem sap feeding with stylets, viruses are ingested from the infected host plant into the lumen of the aphid alimentary canal; (2) Viruses enter the aphid haemocoel through the gut; (3) Viruses propagate (propagative mode) or are simply retained (circulative mode) in the aphid tissues and haemolymph; (4) Viruses are transported into the salivary gland; (5) While feeding, viruses are injected into the phloem tissue of an uninfected host plant. In the 2nd step, viruses enter the epithelial cells of the aphid gut by endocytosis and exit these cells to enter the haemocoel by exocytosis. Transport of viruses through the salivary gland cells (4th step) also involves endocytosis and exocytosis (Gray & Gildow, 2003). Thus, the proteins involved in endocytosis, vesicle transport and exocytosis (dynamins, together with NSF, alpha-SNAP, SNAREs, Munc18, Rab, annexins, etc.) are expected to play important roles in virus transmission (Tamborindeguy et al., 2009). Indeed, Dyn proteins have been demonstrated to be essential for the infection of various viruses in animal hosts (Cherry & Perrimon, 2004; Marsh & Helenius, 2006). As the expansion of dynamin genes and the rampant transmission of plant viruses are among unique features of aphids, high expression levels of novel as well as canonical dynamin genes in the midgut suggest they may be involved in certain steps of viral transmission.

We were unable to find any evidence that aphid dynamins play important roles in the symbiotic relationship with Buchnera, at least in the adult maternal bacteriocyte. However, we cannot exclude the possibility that dynamins are more important in the bacteriocyte of nymphs and embryos, where Buchnera cells divide more frequently than in the adult maternal bacteriocyte (Whitehead & Douglas, 1993).

Experimental procedures

Screening of the Acyrthosiphon pisum genome for dynamin genes

Basic Local Alignment Search Tool (BLAST) searches (Altschul et al., 1997) were performed at the website of the National Center for Biotechnology Information (NCBI) using amino acid sequences of the classical Dynamin (Shibire, NP_001036279.1), Drp1 (NP_608694.2), and Opa1 (NP_725369.1) of D. melanogaster as queries. The RefSeq protein database and the ab initio Gnomon-predicted protein database of A. pisum were screened by using BLASTP. The A. pisum genome scaffolds (Acry_1.0) were screened by using TBLASTN. Default parameters were used for the analyses (E-value cutoff = 0.001).

Molecular phylogenetic analysis

Fully sequenced genomes of representative eukaryotes including all of the sequenced insects were screened for genes that encode dynamin superfamily proteins in the same manner as described above. Retrieved gene models (Supporting Information Table S1) together with A. pisum dynamin gene models were used for the phylogenetic analyses. Deduced amino acid sequences were aligned using ClustalW (Thompson et al., 1994) embedded in MEGA 4.0 (Tamura et al., 2007), followed by manual refinement. Amino acid sites corresponding to alignment gap(s) were omitted from the data set. Only unambiguously aligned sequences were used for the phylogenetic analysis. There were a total of 653 amino acid positions in the final dataset. Phylogenetic trees were inferred by the ML (Felsenstein, 1981) and the BI methods (Ronquist & Huelsenbeck, 2003). ML trees were constructed using RAxML 7.0.4 (Stamatakis, 2006) with 100 replicates, using the WAG matrix of amino acid replacements, assuming a proportion of invariant positions and four gamma-distributed rates (WAG+I+gamma model). Bayesian inference was performed with the program MrBayes version 3.1.2 (Ronquist & Huelsenbeck, 2003) using the WAG+I+gamma model. For the MrBayes consensus trees, 1 000 000 generations were completed with trees collected every 100 generations.

Structural analysis

Domain structures of gene models were analysed using the CD-search at the NCBI website (Marchler-Bauer et al., 2007) and the Pfam domain search (Finn et al., 2008). The presence and location of signal peptides were predicted by using the program SignalP 3.0 (Bendtsen et al., 2004). Subcellular localization of proteins was predicted with TargetP (Emanuelsson et al., 2000). Transmembrane regions and coiled-coil structures were analysed using TMHMM Server ver. 2.0 (Krogh et al., 2001) and coiled-coil predictions at the European Molecular Biology Laboratory (EMBL) (Lupas et al., 1991), respectively.

Real-time quantitative reverse transcription-PCR

Strain ISO, a parthenogenetic clone of the pea aphid that is free from secondary symbionts was used for the analysis. The insects were reared on Vicia faba at 15 °C in a long-day regime of 16 h light and 8 h dark. RNA was isolated from the whole bodies, bacteriocytes, embryos, and midguts of 12–15 day-old parthenogenetic apterous adults using TRIzol reagent, followed by RNase-free DNase I treatment. First-strand cDNAs were synthesized using pd(N)6 primer and PrimeScript reverse transcriptase (Takara). Quantification was performed with the LightCycler instrument and FastStart DNA MasterPLUS SYBR Green I kit (Roche), as described previously (Nakabachi et al., 2005). The primers used are listed in Table 2. The running parameters were: 95 °C for 10 min, followed by 45 cycles of 95 °C for 10 s, 55 °C for 5 s, and 72 °C for the time shown in Table 2. Results were analysed using LightCycler software version 3.5 (Roche), and the relative expression levels were normalized to the mRNA for the ribosomal protein RpL7. Statistical analyses were performed using one-way anova and the Tukey-Kramer test.

Table 2.  Primer sets used for quantitative reverse transcription-PCR
TargetForward primer 5′–3′Reverse primer 5′–3′Amplicon size (bp)Extension time (s)


We thank Terence Murphy at the National Center for Biotechnology Information (NCBI) for his helpful advice. This study was supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan (AN).