Introduction

Bacterial symbionts of arthropods can cause reproductive abnormalities in their hosts. Wolbachia, which belong to α-Proteobacteria, are well-known sex ratio distorters. Their effects include cytoplasmic incompatibility (CI) in crosses between infected males and uninfected females; feminization, in which genetic males develop as females; parthenogenesis induction by diploidization of unfertilized eggs in haplodiploids; and male killing owing to the death of either male embryos or male larvae (O’Neill et al, 1997; Bourtzis and Miller, 2003). CI is the most common effect of Wolbachia infection and is considered to be related to a delay in chromosome condensation and alignment of the male pronucleus (Tram and Sullivan, 2002; Zabalou et al, 2004). An abnormality in paternal chromosome behavior results in embryonic death in diplodiploids (Hoffmann and Turelli, 1988), and in either increased male production or embryonic death in haplodiploids (Breeuwer and Werren, 1990; Gotoh et al, 2003). Conversely, some Wolbachia strains have no effect on host reproduction in both Drosophila spp (Hoffmann et al, 1994, 1996; Giordano et al, 1995) and spider mites (Gomi et al, 1997; Gotoh et al, 2003). In Drosophila, these Wolbachia strains are referred to as the mod− resc+ strains (Bourtzis et al, 1998; Merçot and Poinsot, 1998; Zabalou et al, 2004), and they are considered to be prevalent in arthropods. The mod− resc+ strains do not modify the sperm but instead rescue the detrimental modification caused by a mod+ resc+ strain.

Other symbiotic bacteria belonging to the Cytophaga–Flavobacterium–Bacteroides (CFB) lineage are found in many arthropods. Flavobacteria are found in termites and cockroaches (Bandi et al, 1994, 1995), and those infecting Adonia variegata and Coleomegilla maculata (Coleoptera) cause male killing in the host offspring (Hurst et al, 1997, 1999). CFB bacteria have been found in false spider mites – Brevipalpus spp (Weeks et al, 2001) – and parasitoid wasps – Encarsia spp (Zchori-Fein et al, 2001). These CFB bacteria are phylogenetically distinct from CFB bacteria known at present, and their closest relative is a bacterium isolated from the tick Ixodes scapularis (Kurtti et al, 1996). They are called Cytophaga-like organisms (CLOs) and a new genus name, Cardinium, has been proposed (Zchori-Fein et al, 2004). Cardinium spp cause parthenogenesis (Zchori-Fein et al, 2001) and CI (Hunter et al, 2003) in parasitoid wasps, Encarsia spp and feminization in false spider mites, Brevipalpus spp (Weeks et al, 2001).

Infections of both Cardinium and Wolbachia in a single host are known in four mite species and three Aphitis species (Hymenoptera; Weeks et al, 2003) and one hymenopteran species and one mite species (Zchori-Fein and Perlman, 2004). As both Cardinium and Wolbachia cause reproductive abnormalities in their host arthropods, it is of interest to know whether they work independently, cooperatively or interferingly. To understand the interaction between the endosymbionts and their host arthropods, we created mite populations that were infected with Cardinium, Wolbachia and both Cardinium and Wolbachia for crossing experiments between infected and uninfected individuals.

In some spider mites, Wolbachia have been reported to induce CI (Breeuwer, 1997; Gotoh et al, 2003) or parthenogenesis (Weeks and Breeuwer, 2001), whereas in other spider mites, no reproductive effects were reported (Gomi et al, 1997; Gotoh et al, 2003). The aim of this study was to clarify whether the Cardinium sp in the spider mites affects reproduction. We determined the phylogenetic relationships of the Cardinium strains found in seven populations of five spider mite species, tested the effect of Cardinium on the reproductive traits of infected spider mites and evaluated the effects of Cardinium infection, Wolbachia infection and infection with both.

Materials and methods

Spider mites

Seven populations representing five species of spider mite were used in this study (Table 1). The species were Eotetranychus suginamensis, Oligonychus ilicis, Amphitetranychus quercivorus, Tetranychus urticae (three populations) and Tetranychus pueraricola. A PCR survey of 94 populations comprising 27 spider mite species showed that some of these populations were infected with Cardinium (H Noda et al, unpublished data). Mites were reared on leaf discs of either of the original host plants or of kidney bean – for T. urticae and T. pueraricola – in a climate-controlled room (25°C, L:D=16:8, RH 60%). The Taiwanese population of E. suginamensis was imported to Japan with the authorization of the Ministry of Agriculture, Forestry and Fisheries of Japan (no. 14-Y-583) on 25 October 2002.

Table 1 Mite populations collected for this study
Full size table

Polymerase chain reaction

The DNA template was prepared by homogenizing a single female adult in a 25 μl mixture of STE buffer (100 mM NaCl, 10 mM Tris-HCl, 1 mM EDTA, pH 8.0) and 2 μl proteinase K (10 mg/ml). The mixture was incubated at 37°C for 30 min, and proteinase K was inactivated at 95°C for 5 min. The samples were centrifuged briefly in a microfuge tube and used immediately for the PCR reaction, or stored at −30°C for later use. All PCR reactions were performed in 20 μl of buffer: 14 μl H2O, 2 μl 10 × buffer, 1 μl dNTP (2.5 mM each), 5 U/μl Taq polymerase (TAKARA rTaq, Tokyo), 1 μl sample and 1 μl of primers (10 pmol each). The primers used for detection of Cardinium were CLO-f1 (5′-GGA ACC TTA CCT GGG CTA GAA TGT ATT) and CLO-r1 (5′-GCC ACT GTC TTC AAG CTC TAC CAA C), which amplified 468 bp of 16S ribosomal DNA (rDNA). The primers used for obtaining 16S rDNA for sequencing were fD1 (5′-AGA GTT TGA TCC TGG CTC AG) and rP2 (5′-ACG GCT ACC TTG TTA CGA CTT) (Weisburg et al, 1991) and CLO-specific primers (CLO-f2, 5′-GGT GCG TGG GCG GCT TAT T; CLO-r2, 5′-AAA GGG TTT CGC TCG TTA TAG). The 16S rDNA of A. quercivorus was amplified using fD1 and rP2. Two parts of 16S rDNA were separately amplified using primer combinations of fD1/CLO-r2 and CLO-f2/rP2 in other spider mite species. These Cardinium-specific primers were designed on the basis of the 16S rDNA sequences of Cardinium from I. scapularis (Kurtti et al, 1996), Brevipalpus phenicis (Weeks et al, 2001), Encarsia spp (Zchori-Fein et al, 2001) and A. quercivorus (this study). Reactions were cycled 35 times for 30 s at 95°C, 30 s at 54°C and 90 s at 72°C. The PCR products were electrophoresed in a 1.0% agarose gel in TAE.

Wolbachia infection was examined using Wolbachia-specific 16S rDNA (99F-992R; O’Neill et al, 1992), ftsZ (ftsZ f1-r1; Holden et al, 1993) and wsp (wspf-wspr; Zhou et al, 1998).

Sequencing and phylogenetic analysis

The phylogenetic relationship of Cardinium strains in spider mites was based on the nucleotide sequences of the 16S rDNA genes, which were amplified by primers fD1/rP2, fD1/CLO-r2 and CLO-f2/rP2. The PCR product was cloned into a pGEM-T Vector (Promega). The template DNA was amplified by PCR using M13-20 and reverse primers. The sequence was determined by the dye terminator sequencing method by using a DNA sequencer (models 377 and 3700, PE Applied Biosystems).

The 16S rDNA sequences of 32 Cardinium strains were aligned by using the CLUSTAL X program (Thompson et al, 1997). The aligned data set consisted of 1108 residues. A neighbor-joining analysis was performed with CLUSTAL X, and a bootstrap analysis was performed with 1000 replications.

Antibiotic treatment

Small discs (ca 1 cm2) of the original host or kidney bean leaves were placed on a cotton bed soaked in either tetracycline hydrochloride (0.1%, w/v; TC, Wako) (to eliminate both Cardinium and Wolbachia) or in penicillin G (0.1%, w/v; PCG, Sigma) (to eliminate Cardinium only; Morimoto et al, 2006). The procedure was carried out in plastic dishes (9 cm in diameter), 1 day before the start of rearing. Newly hatched larvae were placed on the leaf discs, and distilled water was added daily to keep the cotton bed wet. The cotton and the leaf discs were replaced every 4 days, and the mites were reared in the antibiotic environment for one generation (Gotoh et al, 1995). Mites were checked for infection with Cardinium and/or Wolbachia by PCR after three generations, by using the specific primers mentioned above. Offspring from adults that were found to be Cardinium-free and/or Wolbachia-free were used in the crossing experiments.

Heat treatment

Wolbachia was eliminated by heat treatment (van Opijnen and Breeuwer, 1999). To determine the most effective heat treatment, T. pueraricola eggs laid during a 24 h period were kept at two different temperatures (35 and 40°C) for two different times (3 and 7 days) and then transferred to 25°C. Emerging adults were checked for infection with Cardinium and Wolbachia by using PCR. Offspring from adults that were found to be Cardinium-infected and Wolbachia-free were allowed to mate.

Crossing experiments

To determine the effects of Cardinium infection, we carried out crossing experiments between infected and tetracycline-treated cured mites in all Cardinium-infected species. A crossing pair was denoted as the female/male strain and antibiotic-cured colonies were designated as ‘Es−’. For example, ‘Es+/Es−’ denotes a cross between a Cardinium-infected female and an antibiotic-cured male in E. suginamensis. Single females in the teleiochrysalis stage (the final immature stage) obtained from each stock culture were transferred onto a small leaf disc (ca 4 cm2) of the appropriate plant together with an adult male (1- to 5-day-old) from either the same or a different culture. Males were removed 2 days after emergence of the adult females. Each female was allowed to lay eggs for 5 days after oviposition started and was then removed. This ensured that only young mites were used. Eggs on leaf discs were checked daily to determine eclosion rates from eggs, survival rates at the immature stages and sex ratio (percentage of females). PCR was carried out for all females and males used in the crossing experiments to confirm that the expected combinations were achieved in the pairs. Data from the pairs that did not achieve the expected combination were discarded. All experiments were carried out at 25°C and L:D=16:8.

Results

Sequence and phylogenetic relationships

Nucleotide sequences of the 16S rDNA genes amplified from the seven populations showed little variation, with 97.7–100% similarity. The phylogenetic tree based on the 16S rDNA genes clearly indicated that the microorganisms in the spider mites were Cardinium, forming a monophyletic group with other Cardinium. The Cardinium from the spider mite populations seemed to be closely related to the symbionts from other mite and tick species in the phylogenetic tree (Figure 1). Three populations of T. urticae had identical sequences, whereas the other four populations had unique sequences, indicating that they harbored different strains of Cardinium.

Figure 1
figure 1

Phylogenetic tree based on 16S rDNA sequences of Cardinium, constructed by using a neighbor-joining procedure in CLUSTAL X. Each Cardinium is shown by its host name. Numbers on the nodes indicate bootstrap values (%). T. urticae A, T. urticae B and T. urticae C are populations collected at Nanae (Hokkaido), Matsukawa (Nagano) and Iida (Nagano), respectively. Asterisks designate symbiont sequences obtained in this study.

Full size image

The effects of Cardinium infection on spider mites

The only population to show any reproductive abnormality in the crossing experiments was the E. suginamensis population, which showed CI.

Eotetranychus suginamensis

The population of E. suginamensis showed unidirectional CI when penicillin-cured females were crossed with infected males (Es−/Es+) (Table 2). This cross was carried out twice with similar results. The hatchability of eggs and the number of F1 females in the Es−/Es+ cross were significantly lower than those of the other crosses (Es+/Es+, Es+/Es− and Es−/Es−). However, the overall lower number of eggs hatching and the overall lower number of F1 females produced per pair are underestimated owing to extensive variation between pairs. The hatchability (Figure 2) and proportion of F1 females (Figure 3) varied greatly in the Es−/Es+ cross, but not so much in the other three crosses. Five pairs of the Es−/Es+ cross produced no F2 females, indicating a strong CI effect by Cardinium. Conversely, Es+/Es+, Es+/Es− and Es−/Es− never showed much lower values in hatchability and proportion of females.

Table 2 Compatibility of crosses between Cardinium-infected (Es+) and antibiotic-treated (Es−) colonies of the Taiwanese population of Eotetranychus suginamensis
Full size table
Figure 2
figure 2

Egg hatchability of the F1 offspring produced by crosses between Cardinium-eliminated (Es−) and Cardinium-infected (Es+) individuals of E. suginamensis. See also Table 2.

Full size image
Figure 3
figure 3

Sex ratio (proportion of females) in F1 offspring produced by the crosses between Cardinium-eliminated (Es−) and Cardinium-infected (Es+) individuals of E. suginamensis. See also Table 2.

Full size image

Oligonychus ilicis

The population of O. ilicis showed no sex distortion (data not shown). No reduction in egg hatchability (91.8±2.33 to 96.5±1.62, mean±s.e.m.) was observed among the four combinations (F3,52=0.938, P>0.05, ANOVA). The Oi−/Oi+ cross resulted in a slightly reduced proportion of females (66.2±2.15; F3,52=6.403, P<0.001, ANOVA) among the four combinations, but it was not significantly different from that in the Oi−/Oi− cross (68.3±1.40; P>0.05, Tukey HSD test), indicating that Cardinium apparently did not affect host reproduction.

Amphitetranychus quercivorus

A Cardinium-free colony was established by treatment with penicillin, and four combinations of crosses were observed between Cardinium-infected and Cardinium-free colonies. No differences (P>0.05, ANOVA) were observed in hatchability (F3,52=2.427; 96.8±1.07 to 99.4±0.61) and survival rate at immature stages (F3,52=1.161; 89.0±1.61 to 93.8±1.55) among the four crosses (data not shown). The sex ratios between the Aq+/Aq− cross (80.7±1.94) and the Aq−/Aq+ cross (71.7±2.50) were significantly different (F3,52=3.159, P<0.05), but values in these two crosses did not differ from those in the Aq+/Aq+ (78.2±1.88) and Aq−/Aq− (74.2±2.43) crosses (P>0.05, Tukey HSD test). These results suggest that Cardinium in this population did not have the perceptible ability to manipulate the host sex ratio.

Tetranychus urticae (red form)

Members from the original three infected populations were crossed with antibiotic-cured colonies (data not given). Four combinations of crosses in the A (Nanae) and B (Matsukawa) populations showed no significant differences (P>0.05) in egg hatchability (98.9±0.64 to 100±0.00, F3,45=1.545, for A; and 99.3±0.45 to 99.7±0.35, F3,44=0.192, for B), survival rate at the immature stages (96.7±1.02 to 99.8±0.24, F3,45=2.430, for A; and 99.1±0.59 to 99.7±0.31, F3,44=0.384, for B) and sex ratio (72.3±2.03 to 75.9±2.26, F3,45=0.153, for A; and 76.2±1.46 to 82.1±1.66, F3,44=2.524, for B). In the C (Iida) population, hatchability (F3,48=4.674, P<0.001; 96.9±0.96 to 100±0.00) was significantly different among the four combinations, because all eggs in the C+/C− cross hatched. However, no male-biased sex ratio was observed among the four combinations (F3,48=1.104, P>0.05). Thus, we observed no reproductive abnormalities in the three populations of T. urticae.

Tetranychus pueraricola

The population of T. pueraricola harbored infection with both Cardinium and Wolbachia. Bacteria-free colonies, in which both bacteria species were eliminated from the host, were established by tetracycline treatment. Four combinations of crosses were observed between members from the original population and those from bacteria-free colonies. Egg hatchability and survival rate at immature stages were not significantly different (P>0.05, Tukey HSD test) among the four combinations. The sex ratio was different between the Tp+/Tp− and Tp−/Tp− crosses, but the values were similar (P>0.05) to those of the Tp+/Tp+ and Tp−/Tp+ crosses (first group in Table 3). These results suggest that no reproductive abnormalities occurred when T. pueraricola was infected with the two bacteria simultaneously.

Table 3 Compatibility of crosses between Cardinium–Wolbachia double-infected (TpCW+) and antibiotic-treated (TpCW−) colonies, crosses between Cardinium-infected (TpC+) and antibiotic-treated (TpC−) colonies and crosses between Wolbachia-infected (TpW+) and antibiotic-treated (TpW−) colonies in the Hitachiohta population of Tetranychus pueraricola
Full size table

Culturing the T. pueraricola strain at 35°C for 3 or 7 days completely eliminated Wolbachia, but not Cardinium (Figure 4). Cardinium populations tolerated heat up to 40°C for 7 days, with several exceptions. Cardinium infection rates tested by PCR after heat treatment in females were 27/30 (90%) at 35°C for 7 days, 28/30 (93.3%) at 40°C for 3 days and 24/30 (80%) at 40°C for 7 days. We used a colony treated at 35°C for the 7 days in the experiments. In crosses between mites of Cardinium-infected and Cardinium-free colonies, no significant differences (P>0.05, ANOVA) were found in egg hatchability, survival rate at immature stages and sex ratio of the F1 generation (second group in Table 3), indicating that Cardinium did not independently cause any sex distortion.

Figure 4
figure 4

PCR assay showing the effects of 3- and 7-day heat treatment on Wolbachia and Cardinium infection in T. pueraricola. Wolbachia was sensitive and Cardinium was resistant to heat. Lanes for Wolbachia: M, marker (λ-EcoT14I digest); 1, 16S rDNA; 2, ftsZ; 3, wsp. Lanes for Cardinium: M, marker; 1, 16S rDNA amplified by CLO-f1/CLO-r1; 2, 16S rDNA amplified by fD1/CLO-r2.

Full size image

Wolbachia-infected and Cardinium-free colonies were established by penicillin treatment, and four combinations of crosses were observed between Wolbachia-infected and Wolbachia-free colonies. No significant differences (P>0.05, ANOVA) were found in egg hatchability, survival rate at immature stages or sex ratio among the four crosses (third group in Table 3), showing that Wolbachia by itself did not affect the sex ratio of T. pueraricola.

Discussion

Phylogenetic analysis of the 16S rDNA sequences clearly revealed that the bacteria in the seven populations of spider mites belonged to Cardinium and that similar Cardinium strains were harbored by all known acarian species. The 16S rDNA sequence of Cardinium infecting E. suginamensis had 98.1% similarity to the symbionts in the tick I. scapularis (Kurtti et al, 1996), in which Cardinium was first described. The sequence also had 96.2% similarity to the symbiont in the parasitic wasp Encarsia pergandiella (Hunter et al, 2003), in which CI was first observed.

Cardinium induced CI in one out of the seven populations of spider mites examined in this study. This is the second instance of a bacterial symbiont other than Wolbachia that can induce CI. The other known instance is Cardinium in the parasitoid wasp E. pergandiella (Hunter et al, 2003). Cardinium, as well as Wolbachia in a number of mite species (Breeuwer, 1997; Gotoh et al, 2003, 2005), did not have a pronounced effect on crosses in E. suginamensis. However, five pairs of E. suginamensis produced no F1 females, showing a severe CI effect. Females in the incompatible crosses produced female offspring (daughters) but the number of daughters produced was 25–29% less than the number of daughters produced by females in the compatible crosses. The reduced number of daughters was mainly the result of the death of female eggs; that is, the hatchability of eggs produced by females in these crosses was 20–26% less than the hatchability of eggs produced by females in the compatible crosses. The egg hatchability observed in the incompatible crosses was slightly higher in Cardinium-infected E. suginamensis (71–76%) than in spider mites infected with Wolbachia (56–99%) (Gotoh et al, 2003, 2005). However, a reduction of hatchability was not always observed in the incompatible crosses caused by Wolbachia infections, even in the same combination. In some crosses, a reduction of hatchability occurred in the first trial, but not in the second trial (Gotoh et al, 2005). Therefore, it is unclear whether Cardinium infection always results in reduced hatchability, because this study dealt with only one example. In the present study, we used 1- to 5-day-old males for crossing experiments. We may need to examine further the effects of male age on incompatibility level by comparing the 1- and 5-day-old males, because older males usually express less CI than younger ones (Hoffmann et al, 1986). In the six populations of the four species, no reproductive abnormalities, such as a female ratio less than 50%, were observed in the F1 generation in our experimental conditions. Vala et al (2000) reported that Wolbachia infection causes a more severe hybrid breakdown phenotype in the F2 generation of a cross between uninfected females and infected males than in the F1 generation. The possible effects of Cardinium in the F2 generation are untested and could not be ruled out in our study.

In this study, the Sapporo population of A. quercivorus was infected with the non-CI strains of Cardinium. Our previous study shows that the Sapporo (43° N) females were incompatible with the Tsukuba (36° N) males, which resulted in low egg hatchability and a male-biased sex ratio, whereas the reciprocal crosses were compatible and produced normal progeny with a female-biased sex ratio (Gotoh et al, 1995; A. quercivorus was referred to as Tetranychus quercivorus). In that study, 12 types of antibiotic, including tetracycline hydrochloride, were used to treat the Sapporo females, but they were not effective in restoring the compatibility of the two populations (Gotoh et al, 1995). Wolbachia also did not infect the Sapporo and Tsukuba populations (Gotoh et al, 2003). We first found Cardinium in A. quercivorus and no other symbiotic bacteria were detected. These results clearly show that unidirectional reproductive incompatibility between the Sapporo and Tsukuba populations of A. quercivorus is not due to intracellular bacteria such as Cardinium and Wolbachia.

An interesting aspect in the present study is that the Cardinium strain responsible for CI in E. suginamensis is closely related to the non-CI strains found in T. pueraricola and A. quercivorus, based on the 16S rDNA tree. A similar phenomenon was found in Wolbachia strains infecting spider mites. On the basis of wsp (Wolbachia surface protein) gene sequences, Wolbachia in Panonychus mori Yokoyama had an identical sequence to those in T. urticae (green form), but CI occurred only in the former (Gotoh et al, 2003, 2005). Phylogenetic similarity of symbionts is not related to their phenotypes due to one, or a combination of, the following three factors. First, symbiont genotype is different, and a gene involved in reproductive alteration might have a mutation in its sequence. Second, the host species have different genetic backgrounds. Third, symbiont density in the host animals is different, which seems to be caused by host–symbiont interaction and results in different phenotypes. The second and third factors seem to be under the control of some other factors. Future studies in which a CI-causing Cardinium strain is transferred to a Cardinium-uninfected host or to a host harboring non-CI Cardinium might further clarify this interesting biological difference.

Infections of both Cardinium and Wolbachia in the same host are known in four Aphytis species (Hymenoptera) and four mite species (Acari) (Weeks et al, 2003; Zchori-Fein and Perlman, 2004), although it is unknown whether the symbionts infecting these arthropod species induce reproductive abnormalities. Infections of both Cardinium and Wolbachia have not so far been found in spider mites but were found in T. pueraricola. Cardinium were sensitive to tetracycline, rifampicin, penicillin G, ampicillin and chloramphenicol antibiotics (Morimoto et al, 2006). Tetracycline and rifampicin were effective against Wolbachia, but penicillin G was less effective (Fenollar et al, 2003). Therefore, we used penicillin G for elimination of Cardinium alone and tetracycline for elimination of both Cardinium and Wolbachia. Heat treatment is sometimes used for elimination of symbionts from host arthropods. In this study, it successfully eliminated only Wolbachia from T. pueraricola, creating a population infected with only Cardinium (Figure 4). No abnormalities were observed in the double-infected population or in populations infected with either one of the symbionts. However, selective elimination of symbionts will help in the analysis of reproductive abnormalities in double-infected populations or species.