Symbiotic Bacterium Modifies Aphid Body Color

See allHide authors and affiliations

Science  19 Nov 2010:
Vol. 330, Issue 6007, pp. 1102-1104
DOI: 10.1126/science.1195463


Color variation within populations of the pea aphid influences relative susceptibility to predators and parasites. We have discovered that infection with a facultative endosymbiont of the genus Rickettsiella changes the insects’ body color from red to green in natural populations. Approximately 8% of pea aphids collected in Western Europe carried the Rickettsiella infection. The infection increased amounts of blue-green polycyclic quinones, whereas it had less of an effect on yellow-red carotenoid pigments. The effect of the endosymbiont on body color is expected to influence prey-predator interactions, as well as interactions with other endosymbionts.

The world is full of colors, and many animals have color vision, recognizing their environment, habitat, food, enemies, rivals, and mates by visual cues. Body color is thus an ecologically important trait, often involved in species recognition, sexual selection, mimicry, aposematism, and crypsis (1, 2). In the pea aphid Acyrthosiphon pisum, red and green color morphs are found in the same populations. Early work has shown that the aphid body color is genetically determined, with red being dominant over green (3). Ecological studies show that ladybird beetles tend to consume red aphids on green plants (4), and parasitoid wasps preferentially attack green aphids (5). The predation and parasitism pressures appear to maintain the color variation in natural aphid populations (1, 4). An unexpected recent discovery showed that the aphid genome contains several genes for carotenoid synthesis not found in animal genomes. The genes are of fungal origin and appear to have been acquired in the evolutionary history of aphids via ancient lateral transfer. One of the genes is involved in synthesis of red color pigments, and the presence or absence of the gene is responsible for the red or green coloration of the aphids (6). Here, we report another factor affecting aphid color polymorphism: a previously unrecognized endosymbiont that modifies insect body color in natural populations.

While screening pea aphid strains from natural populations collected in France, we found several strains of green aphids producing red nymphs. As the nymphs grew, their body color changed from reddish to greenish, and the adults became green (Fig. 1A and table S1). A survey of endosymbiotic microbiota in these aphid strains from distinct geographic origins and bearing different genotypes (7) identified γ-proteobacterial facultative endosymbionts (one of which is a Hamiltonella- or Serratia-like organism) that confer protection to their host aphids against parasitoid wasps (8). In addition, we found a previously unrecognized aphid endosymbiont of the genus Rickettsiella, whose members are insect pathogens that are phylogenetically related to the human pathogens Coxiella and Legionella (Fig. 2A) (9).

Fig. 1

(A to H) Rickettsiella-induced body-color change in pea aphids of different genetic backgrounds. Scale bars, 1 mm. For details of the aphid strains, see table S1.

Fig. 2

(A) Phylogenetic analysis of Rickettsiella endosymbionts from European pea aphids on the basis of 16S ribosomal RNA gene sequences. A maximum likelihood phylogeny inferred from 1384 aligned nucleotide sites is shown with bootstrap values. (B to G) In situ hybridization of Rickettsiella. (B) A mature embryo (blue) containing many primary bacteriocytes harboring Buchnera (green) and a secondary bacteriocyte harboring Rickettsiella (red) that together constitute a huge bacteriome. (C) Enlarged image of the secondary bacteriocyte. (D) Sheath cells harboring Rickettsiella adhering to the periphery of primary bacteriocytes (white arrowheads). (E) Oenocytes (oe) infected with Rickettsiella. (F) Posterior part of an ovary, where ovariole pedicels are heavily infected with Rickettsiella (yellow arrowheads). em, embryo. (G) Enlarged image of an ovariole pedicel. (H to J) Electron microscopy images of Rickettsiella. (H) Image of a secondary bacteriocyte harboring Rickettsiella and a primary bacteriocyte harboring Buchnera. vac, vacuole; b, Buchnera; asterisks, Rickettsiella. (I) Enlarged image of a Rickettsiella cell. m, mitochondrion. (J) Image of the wall of an ovariole pedicel infected with Rickettsiella. hem, hemocoel; lu, lumen of ovariole pedicel; nu, nucleus.

By antibiotic treatments (7), we successfully eliminated the Hamiltonella/Serratia infection from the aphids without affecting Rickettsiella and Buchnera infections. Their body-coloring patterns did not change after the treatments (Fig. 1B). When we injected diverse aphid strains, which harbored only Buchnera, with hemolymph from the Rickettsiella-infected strains (7), the aphids produced both Rickettsiella-infected and -uninfected offspring. Notably, Rickettsiella-infected red aphids of distinct genotypes consistently changed body color to green as they developed, whereas neither uninfected red aphids nor originally green aphids were affected (Fig. 1, fig. S1, and table S1). The body-coloring patterns in the experimentally infected aphid strains were similar to those in the naturally Rickettsiella-infected strains (Fig. 1). Quantitative polymerase chain reaction (PCR) analyses revealed that the intensity of green color was positively correlated with the infection density of Rickettsiella for different host and endosymbiont genotypes (fig. S1). These results indicate that the Rickettsiella infection is responsible for green body color in at least some green pea aphids in natural populations.

Diagnostic PCR surveys detected 7.9% [28 of 353 insects (28/353)] Rickettsiella infection in Western European populations of A. pisum (fig. S2A). Fitness measurements revealed that infection status did not affect growth rate and body size for two aphid genotypes, although we observed significantly larger body size and faster growth with Rickettsiella infection for one aphid genotype (fig. S3). Similarly to Hamiltonella and Serratia (10, 11), Rickettsiella resided in secondary bacteriocytes and sheath cells in vivo and was also found intracellularly and extracellularly in various tissues and the hemolymph (Fig. 2, B to J). In natural populations, not all green aphids were infected with Rickettsiella, and some strains of red aphids were found with Rickettsiella infection (fig. S2, B and C). It appears that the combination of aphid genotype and the endosymbiont contribute to body color (3, 6): Some endosymbiont genotypes may fail to induce green coloration, whereas some host genotypes may attenuate or inhibit the activity of Rickettsiella. Similar interactions have been documented for other facultative endosymbionts in the pea aphid (12, 13).

Aphid body color mainly consists of two major groups of pigment molecules: (i) yellow-red colors from carotenoid pigments such as β-carotene, lycopene, and torulene (6, 14) and (ii) blue-green and other pigments from structurally complex polycyclic quinones and their glycosides, called aphins or aphinins (14, 15). When the naturally red aphid strain was infected with Rickettsiella and became green, we observed some changes in carotenoid compositions, but the differences were not sufficient to account for the degree of green pigmentation (fig. S4). Moran and Jarvik (6) have found that a fungus-derived carotenoid desaturase gene, tor, is present in red aphid clones but is absent in green aphid clones and is responsible for production of the red carotenoids. Our quantitative reverse transcription–PCR assay showed that expression levels of the tor gene were not significantly affected by the Rickettsiella infection (fig. S5).

We recovered almost all of the green pigments into the butanol fraction by water-butanol extraction of the aphids (7). Thin-layer chromatography of the extracts separated one major and several minor green bands, which presumably represented polycyclic quinone glycosides like aphinins (14), although their exact structures were not determined (7). The intensity of the green bands was greater in the Rickettsiella-infected green-aphid strains than in the original red-aphid strains (fig. S6A). Densitometric quantification of the major band revealed a 2.4- to 4.6-fold increase of the green pigment in the Rickettsiella-infected aphid strains compared with the uninfected strains (fig. S6B). We presume that Rickettsiella does not synthesize new green pigments for itself but stimulates aphid metabolism to increase green-pigment production (fig. S7). The murky hue of the Rickettsiella-induced green aphids (Fig. 1 and fig. S1) is probably a result of the combination of the green pigments and the reddish carotenoid pigments. Genome sequencing of the Rickettsiella endosymbiont and transcriptomic analysis of the infected and uninfected aphid hosts should provide insights into the molecular and metabolic interactions between host and endosymbiont that lead to the body-color change.

Previous studies have identified a variety of biological roles for facultative endosymbionts in the pea aphid, including tolerance to high temperature, resistance against natural enemies, and plant adaptation (8, 1619). We have added another endosymbiont relationship that potentially affects a host trait of ecological importance in ways that have yet to be verified. For example, the induced green color may reduce the predation risk by ladybird beetles. Notably, Rickettsiella is frequently found in co-infections with either Hamiltonella [55.6% (35/63)] or Serratia [20.6% (13/63)] endosymbionts (fig. S2D), both of which are protective against parasitoid wasps (8) and may act to offset the risk of green aphids attracting parasitoids.

Supporting Online Material

Materials and Methods

Figs. S1 to S7

Table S1


References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. We thank J. Peccoud, Y. Outreman, S. Stoeckel, L. Mieuzet, J. Jaquiéry, and W. Weisser for aphid sampling; J. Bonhomme, L. Mieuzet, and J. Makino for aphid rearing and genotyping; and X.-Y. Meng and S. Hanada for help with electron microscopy. This work was supported by the Japan-France Integrated Action Program SAKURA of the Japan Society for the Promotion of Science (to T.F. and J.-C.S.); the Grant-in-Aid for Scientific Research on Innovative Areas (22128007) of the Ministry of Education, Culture, Sports, Science and Technology, Japan (to T.F.); and funds from INRA AIP Bioresources (to J.-C.S.). T.T. was supported by a RIKEN special postdoctoral research fellowship.
View Abstract

Navigate This Article