Whiteflies are small homopterans that feed as nymphs and adults on the phloem sap of plants. The hatching crawler settles near the hatching site, where it goes through four immobile nymphal instars before developing into an adult. The sweet potato whitefly Bemisia tabaci (Gennadius) is a severe agricultural pest in many parts of the world (5). This species consists of several biotypes (3) that have been distinguished largely on the basis of biochemical or molecular diagnostics but whose biological significance is still unclear. Like other phloem-feeding insects, whiteflies require bacteria for supplementing their unbalanced diet. These symbionts are housed in specialized organs called bacteriomes, which are composed of bacteriocytes (2). “Candidatus Portiera aleyrodidarum,” the primary symbiont of whiteflies, is an AT-rich member of the gamma subdivision of the Proteobacteria (2). In B. tabaci, “Ca. Portiera aleyrodidarum” is vertically transmitted through bacteriocyte inclusions into the oocyte at the point which will eventually become the pedicel end of the egg (9, 19).

Secondary symbionts of B. tabaci consist of a diverse array of bacteria which are phylogenetically related to other described symbionts of sap-feeding insects. Using transmission electron microscopy, Costa et al. (8) distinguished three types of secondary symbionts, two of which have been tentatively identified as “Candidatus Cardinium hertigii” (Bacteriodetes) (39) and Fritchea bemisiae (Simkaniaceae) (36), respectively. Sequence-based phylogenetic analysis of secondary symbionts found in various B. tabaci biotypes further revealed the presence of two enteric bacteria, one which resembles the aphid symbiont “Candidatus Hamiltonella defensa” and the other with high sequence similarity to the Arsenophonus-like psyllid symbiont (44) and to Wolbachia spp. (27).

A comprehensive characterization of the bacterial community in different B. tabaci populations is crucial for understanding various aspects of that pest's biology, such as the emergence of more aggressive biotypes and the variation in transmission capabilities of plant viruses (25). The research here aimed to profile the bacterial community found in one B. tabaci population. During the analysis, we identified a Rickettsia bellii-like bacterium, and since this is the first record of that bacterium in whiteflies, we further studied its spatial and temporal localization in various stages of whitefly development.

Whitefly origin and rearing. B. tabaci, B biotype (pesticide-susceptible strain Ssc) was reared on cotton seedlings (Acala) under standard greenhouse conditions, at 26 ± 2°C, 60% relative humidity, and a photoperiod of 14 h of light and 10 h of darkness. This strain was collected from Israeli cotton fields during 1987 and has since been maintained in a closed laboratory culture without exposure to pesticides (Table 1).

TABLE 1. Israeli B. tabaci populations screened for the presence of Rickettsia sp.

PCR amplification and DGGE analysis. To establish the whole range of bacteria associated with the Ssc population, adults of B. tabaci were placed alive in 96% alcohol, and three replicates of one adult female each were ground in lysis buffer as described by Frohlich et al. (15). The 16S rRNA gene fragment (~550 bp) was amplified using PCR from the insect lysate using the primer combination of 341F with a GC clamp (40-nucleotide, GC-rich sequence) and 907R (Table 2), which targets most known Bacteria, with PCR conditions that permit its amplification from most known Bacteria (26). Reactions were performed in a 50-μl volume containing 5 μl of the template DNA lysate, 400 mM concentrations of each primer, 5 μl of 0.2 mM deoxynucleoside triphosphate, 1× ExTaq buffer, and 1 unit of ExTaq (TaKara Bio, Inc.). Five microliters of the PCR mix was tested using agarose gel electrophoresis, and the remaining 45 μl containing the amplified DNA fragments was then subjected to denaturing gradient gel electrophoresis (DGGE) analysis using the following conditions: separation using a 6% (wt/vol) acrylamide gel (acrylamide-N,N′-methylenebisacrylamide, 37.5:1) prepared in 1× Tris-acetate-EDTA buffer with a denaturing gradient ranging from 20% to 60%. Polymerization was carried out with N,N,N′,N′′-tetramethylethylenediamine (0.09% vol/vol) and ammonium persulfate (0.04% wt/vol). Electrophoresis for separation of PCR fragments was performed at 90 V and 60°C for 16 h. After electrophoresis, the gels were incubated in ethidium bromide solution (250 ng/ml) for 10 min, rinsed in distilled water, and photographed under UV illumination. Bands representing bacteria were eluted, cloned into the pGEM T-Easy plasmid vector (Promega), and transformed into Escherichia coli. For each bacterium, two colonies were randomly picked and sequenced (ABI 3700 DNA analyzer; Macrogen, Inc., Korea), and the results obtained were compared to known sequences by using the BLAST algorithm in the NCBI database.

TABLE 2. PCR primer sets used in this study

Screening for the presence of Rickettsia sp. Of the three bands detected and analyzed, one showed 98% identity to various Rickettsia spp. by BLAST searches. Based on that rickettsial 16S rRNA gene and on similar sequences found in the databases, primers which are capable of specifically amplifying that gene from the whitefly Rickettsia sp. were designed. Samples of B. tabaci from various agricultural crops in Israel were collected in different years and seasons and kept in the laboratory as separate populations as described above or placed in ethanol. To detect the presence of Rickettsia in these populations, one adult was ground individually in lysis buffer, as described above, in at least nine replicates. Each sample was subjected to a PCR using the Rickettsia-specific primers Rb-F and Rb-R (Table 2), a combination expected to yield a product of about 900 bp. PCR parameters were as follows: denaturation for 2 min at 95°C; 30 cycles of 30 s at 92°C, 30 s at 58°C, and 30 s at 72°C; and a 5-min final extension at 72°C. The whitefly populations screened are summarized in Table 1. All reactions included a negative control of sterile water and a positive control of Rickettsia-infected B. tabaci. As an internal control, primers known to amplify the mitochondrial cytochrome oxidase I (COI) gene (15) were used on all samples.

Characterization of Rickettsia sp. in B. tabaci. To determine the phylogenetic affiliation of the newly discovered Rickettsia sp. with rickettsial groups previously classified (spotted fever, typhus, and ancestral), we have followed the genotypic scheme suggested by Fournier et al. (12). All PCR analyses were performed using specific primers and PCR amplification conditions as specified in the literature (Table 2). A nearly full-length segment of the 16S rRNA gene was obtained using the primer combinations 27F/Rb-R and Rb-F/1494R with the parameters described above. The two 16S rRNA gene contigs were assembled using DNAMAN (Lynnon Biosoft Vaudreuil, Quebec, Canada). The citric acid cycle (citrate synthase) gene gltA was amplified and assembled in the same way, and the presence of the rickettsial outer membrane protein (rOmp) encoding genes ompA and ompB was tested. The sequences of the Rickettsia genes obtained were deposited in GenBank.

Establishment of a clean B. tabaci line. In order to characterize the Rickettsia sp. distribution in B. tabaci, a Rickettsia-free line that can serve as a negative control was required. Most of the B. tabaci populations tested exhibited variation in their infection status (Table 1); therefore, an attempt was made to establish a Rickettsia-free line out of the same whitefly colony by isolating 30 mated whitefly females from a line collected in a sweet pepper greenhouse during June 2004 in order to form separate reproductive lines. Each female was allowed to oviposit individually on a sweet pepper (Capsicum annum) leaf disk (55-mm diameter) placed on 1% agar in a transparent plastic cup, maintained at 25°C and 60% ± 10% relative humidity and a photoperiod of 14 h of light and 10 h of darkness, until she died. Upon emergence, at least five progeny of each female were placed alive in 96% alcohol, and the infection status of the various lines was tested by subjecting samples from three individuals to PCR with Rickettsia-specific primers.

In situ hybridization. Adults, eggs, and the various instars were collected with a needle, while ovaries were dissected in a drop of saline buffer under a stereoscopic microscope. The fluorescence in situ hybridization (FISH) procedure generally followed the method of Sakurai et al. (32), with slight modifications. Specimens were collected directly into Carnoy's fixative (chloroform:ethanol:glacial acetic acid, 6:3:1) and fixed overnight. After fixation, the samples were decolorized in 6% H₂O₂ in ethanol for 2 h and then hybridized overnight in hybridization buffer (20 mM Tris-HCl [pH 8.0], 0.9 M NaCl, 0.01% sodium dodecyl sulfate, 30% formamide) containing 10 pmol of fluorescent probes/ml. Based on the Rickettsia sp. and “Ca. Portiera aleyrodidarum” 16S rRNA sequences, two DNA probes were designed using Primer3 software (30) (source code available at http://fokker.wi.mit.edu/primer3/) and were checked for specificity using the Ribosomal Database Project II “probe match” analysis tool (http://rdp.cme.msu.edu/); the probe BTP1-Cy3 (5′-Cy3-TGTCAGTGTCAGCCCAGAAG-3′) was designed to specifically target “Ca. Portiera aleyrodidarum,” and the probe Rb1-Cy5 (5′-Cy5-TCCACGTCGCCGTCTTGC-3′) was designed to target Rickettsia. Stained samples were whole mounted and viewed under an IX81Olympus FluoView500 confocal microscope. Specificity of the detection was confirmed using the following controls: no-probe control, RNase-digested control, and Rickettsia-free whiteflies.

Nucleotide sequence accession numbers. The sequences of the Rickettsia genes obtained in this study were deposited in GenBank under the accession numbers DQ077707 (16S rRNA gene) and DQ077708 (gltA).

Identification of bacterial assembly. A banding pattern composed of three distinct bands was evident when the PCR products were analyzed by DGGE (Fig. 1). When these bands were eluted, cloned, sequenced, and compared to other sequences found in GenBank, they showed high similarities to known bacteria; two bands were most similar to the primary whitefly symbiont “Ca. Portiera aleyrodidarum” (100%) and the secondary symbiont “Ca. Hamiltonella defensa” (100%), respectively. The third band gave a 98% similarity value to the tick symbiont Rickettsia bellii. Because the first two bands revealed the presence of previously described bacteria (44), we have concentrated our efforts on characterizing various aspects of the interactions between B. tabaci and the newly discovered Rickettsia sp.

FIG. 1. DGGE analysis of PCR-amplified, 16S rRNA gene fragments of bacteria found in three individual Bemisia tabaci females. Portiera, “Ca. Portiera aleyrodidarum;” Hamiltonella, “Ca. Hamiltonella defensa.”

Screening for the presence of Rickettsia sp. A total of 20 B. tabaci populations were screened using the Rickettsia-specific primers (Table 1). These populations were collected between the years 1987 and 2003 from Haifa (north) to Hazeva (south), Israel, from eight different host plants (Table 1). Rickettsia sp. could be detected in all of the B. tabaci populations tested, with infection rates ranging from 22% to 100%.

Establishment of Rickettsia sp. identity. The combination of most known Bacteria primers with the Rickettsia-specific primers Rb-F and Rb-R yielded a 1,445-bp sequence of the 16S rRNA gene which exhibited highest sequence similarity to the proteobacterium R. bellii (99%). The use of specific primers for the gltA gene resulted in sequences of 1,210 bp showing 97% similarity to the tick symbiont R. bellii citrate synthase gene. Presence of the ompA and ompB genes could not be detected in PCR using specific primers.

Establishment of a clean B. tabaci line. Out of 30 isofemale lines, the progeny (F₁) of five females tested negative for Rickettsia sp. Consequently, the other 25 lines were discarded, and the five Rickettsia-free lines were further reared. The F₂ and F₃ generations of these five lines were also found to be free of Rickettsia sp. and were mixed into one population.

In situ hybridization. (i) Primary symbionts. Throughout the life cycle of B. tabaci, the probe designed to specifically target the primary symbiont “Ca. Portiera aleyrodidarum” consistently produced signal exclusively inside the bacteriomes (Fig. 2 to 4).

FIG. 2. FISH of B. tabaci adults. (a) Bacteriocytes (arrow) and Rickettsia sp. (arrowhead) in a female abdomen (combined Z sections). (b) Bacteriocytes (arrow), and Rickettsia sp. (arrowhead) in female legs and abdomen (one section). (c) Rickettsia sp. concentrated (more ...)

FIG. 4. FISH of B. tabaci nymphs. (a) Crawlers; (b) third instar; (c) fourth instar. Rickettsia sp. (blue), “Ca. Portiera aleyrodidarum” (red). Note the “Y”-shaped distribution. Right panels, combined Z sections of bright field (more ...)

(ii) Distribution of Rickettsia sp. in adults. Spheres of bacteriocytes are seen around and between the ovaries in the female abdomen (Fig. 2a and b). The signal specific to “Ca. Portiera aleyrodidarum” indicates that these bacteria are located on the surface and inward of the spheres (Fig. 2b). This organization is in agreement with transmission electron microscopy data that show a large nucleus in the middle of each bacteriocyte (35; M. Ghanim, unpublished data). In females, Rickettsia is located around the oocytes, around the follicle cells, and among (but not inside) the bacteriocytes (Fig. 2a to c). In some individuals (males or females), Rickettsia is located in specific polygon-like structures (Fig. 2a) and along the gut (Fig. 2c). Rickettsia can be detected in the hemolymph of both sexes, occurring in all body parts, including head and legs (Fig. 2b and d). Bacteriocytes are hardly ever detected in males and, overall, Rickettsia is more abundant in females (data not shown). Each mature oocyte usually incorporates one bacteriocyte (9, 19) (Fig. 3a), where Rickettsia can be hardly detected or is detected in the pedicel only (Fig. 3a).

FIG. 3. FISH of B. tabaci eggs. (a) Mature oocyte surrounded by external bacteriocytes (red) and a Rickettsia sp. (blue). Note the concentration of the Rickettsia sp. in the pedicel area (arrow). (b) Less than 24 h old, showing bacteriocyte (red) and Rickettsia (more ...)

(iii) Distribution of Rickettsia sp. in eggs. More than 50 eggs were collected and viewed under the confocal microscope. In embryos 0 to 24 h old, a single bacteriocyte is located near the pedicel. “Ca. Portiera aleyrodidarum” is first detected in the bacteriocyte around the perimeter of the cell, with invasion toward the center (Fig. 3b); later on, it occupies most of the bacteriome sphere (Fig. 3c). In older eggs (4 to 5 days of embryogenesis), the bacteriome divides to form two separate organs (data not shown). At early stages of embryogenesis, signal from Rickettsia sp. can be seen only among the “Ca. Portiera aleyrodidarum” within the bacteriocyte (Fig. 3b). However, at later stages, signal from Rickettsia sp. can be detected at a random distribution throughout the embryo (Fig. 3c and d).

(iv) Distribution of Rickettsia sp. in nymphs. The distribution of the Rickettsia sp. in B. tabaci nymphs seems to be random in terms of both quantity and space; the signal in some larvae suggests the presence of large numbers of symbionts, while the signal in others is lower (Fig. 4a). Rickettsia seems to be located throughout the nymphal body, excluding the bacteriomes, as in the eggs. Although the signal can be detected throughout the nymph, there is higher intensity in a “Y”-shaped structure, following the contour of the whitefly's gut (19) (Fig. 4b and c). The random spatial distribution can be seen in all nymphal stages (Fig. 4). The absolute confinement of “Ca. Portiera aleyrodidarum” to the bacteriomes, as opposed to the characteristic random signal for Rickettsia is noticeable when both probes are hybridized on the same crawler (Fig. 4c).

(v) Controls. The no-probe and RNase-digested controls showed fine autofluorescence of the 543 and 633 laser lines. In whiteflies from the Rickettsia-free line, bacteriomes are stained with the “Ca. Portiera aleyrodidarum”-specific probe, while there is no signal from the Rickettsia-specific probe (data not shown).

Insects harbor a diverse assemblage of bacteria, and here we applied a method used for assessing microbial compositions in environmental samples for characterizing the microbial assembly in Israeli populations of B. tabaci. This community was found to be composed of three different bacteria, namely, “Ca. Portiera aleyrodidarum,” “Ca. Hamiltonella defensa,” and Rickettsia sp. (Fig. 1). It should be noted, however, that it has been suggested that only bacterial populations that make up 1% of a complex bacterial community can be detected by PCR and DGGE (26) and, in general, PCR amplification with species- or genus-specific primers is more sensitive than broad-range 16S rRNA gene amplifications. The bacteria identified in this study may therefore not represent the entire diversity of the microbial community within that whitefly population, and the presence of other, less numerous bacteria should not be excluded.

The order Rickettsiales is composed of a coherent group of obligate intracellular symbionts of eukaryotic cells within the alpha subdivision of the Proteobacteria (42). The genus Rickettsia is usually described from blood-feeding arthropods, renowned for the ability of some of its members to cause rickettsioses, and was rarely reported from phytophagous insects (1). To establish the identity of the whitefly bacterium and further characterize its phylogenetic affiliation, we followed the guidelines of Fournier et al. (12). These authors suggested that a bacterium can be ascribed the genus Rickettsia if it shares >98.1% similarity of the 16S rRNA gene and >86.5% similarity of the gltA gene of any known Rickettsia sp. The 16S rRNA and gltA genes of the symbiont described from B. tabaci exhibit 99 and 97% similarity (respectively) with the previously described Rickettsia bellii, and that bacterium could therefore be considered a member of the genus Rickettsia. According to Fournier et al. (12), a Rickettsia sp. in which the ompA and ompB genes are absent belongs to the ancestral group. Because our PCR analyses failed to detect the presence of these genes, it was concluded that the B. tabaci Rickettsia sp. is a member of the ancestral group, together with R. bellii. Other than the pea aphid Rickettsia sp., the first record of a Rickettsia sp. in a phytophagous insect, that bacterium has been reported from the orders Psocoptera, Coleoptera, and Hymenoptera, which are not known to have interaction with vertebrates (20, 22, 41, 43). Outside the Insecta, a Rickettsia sp. was also found in phytophagous organisms, such as the spider mite Tetranychus urticae (21).

Whole-mount fluorescence in situ hybridization of various developmental stages established, for the first time, the long-assumed specific localization of the B. tabaci primary symbiont “Ca. Portiera aleyrodidarum” in the bacteriocytes. This technique also shows a random and uneven localization of the Rickettsia sp. in most of the B. tabaci eggs, nymphal stages, and adults tested. In various adult and nymph specimens, the bacteria are seen aggregating along the gut structure (Fig. 2 and 4). Although the function of the polygonal structures is currently unknown, this specific Rickettsia distribution may serve as a clue in future studies. The detection of Rickettsia around the gut structure and in between the follicle cells is unique compared with other localization data: secondary symbionts of plant-feeding insects have been reported from bacteriomes, within the secondary bacteriocytes, and in the hemolymph (aphid) (16, 17) in the syncytium (part of psyllid bacteriome) (34) and within bacteria themselves, forming a secondary symbiosis (in mealybugs) (38). Whole-mount FISH of one Acyrthosiphon pisum strain revealed the presence of Rickettsia in two types of cells, secondary bacteriocytes and sheath cells. These cells, together with the primary bacteriocytes containing a Buchnera sp., form the bacteriome in the body cavity of aphids (32).

The fact that the Rickettsia sp. can be detected inside the eggs of B. tabaci, and that it is detected throughout the insect development, implies vertical transmission of that bacterium. Our data suggest that the Rickettsia enters the oocyte in the ovaries either by penetrating the oocyte via the pedicel (as suggested from Fig. 3) or by penetrating the bacteriocytes. After entering the egg, the bacteria start multiplying and spreading during embryogenesis. The hatching nymph carries Rickettsia throughout its body, with seemingly higher concentrations around the gut. Adult females apparently carry a higher load of Rickettsia than males, possibly to enable transmission of the bacteria to the next generation.

Studies exploring the influence of secondary symbionts in aphids revealed quite a number of roles these tenants play in their host's biology, including conferring resistance to parasitoids (11, 28), influencing host plant preferences (11, 23, 37), and conferring heat resistance (24). On the other hand, Sakurai et al. (32) showed that a Rickettsia-infected A. pisum strain exhibits a smaller fresh body weight and a lower total number of offspring than a Rickettsia-free strain. These authors also found that the presence of Rickettsia significantly suppressed the population of Buchnera and postulated that these phenomena may be correlated (32). Other studies investigating the fitness effects of the Rickettsia sp. indicate that the presence of that bacterium generally induces negative effects on the aphid hosts, but the intensity of these effects and their consequences depend on environmental factors (6, 7, 24).

The effect of Rickettsia sp. on B. tabaci is yet to be resolved; however, since it is highly prevalent in all tested Israeli B. tabaci populations (Table 1), it can be speculated that its phenotype is advantageous under certain conditions but may be deleterious enough under others to prevent fixation. Moreover, the concentration of the bacteria around the gut tube may indicate a nutritional dependence. Another possible hint for Rickettsia influence may come from the work of Gerling and Fried (18) who found a unique phenomenon of density-dependent sterility in the B. tabaci parasitoid Eretmocerus mundus (Hymenoptera: Aphelinidae) and suggested the involvement of microorganisms. A fitness comparison between Rickettsia-infected and noninfected whitefly individuals would provide the information required for understanding the cost and the benefit of this association for the whitefly host.

We thank Amir Sherman for invaluable technical help and gratefully acknowledge three anonymous reviewers for their very useful comments.

This work was supported by Research Grant no. IS-3633-04 R from BARD, the United States-Israel Binational Agricultural Research and Development Fund, to E. Zchori-Fein and by a financial contribution made by Koppert Biological Systems, The Netherlands.