Replication of the Endosymbiotic Bacterium Blochmannia floridanus Is Correlated with the Developmental and Reproductive Stages of Its Ant Host

Abstract

The dynamics of replication of the intracellular endosymbiotic bacterium Blochmannia floridanus was determined during the larval development of its host ant Camponotus floridanus by real-time quantitative PCR. The bacteria were found to proliferate during pupation and immediately after the eclosion of the imagines (adult ants). In older workers the number of bacteria present in the midgut bacteriocytes decreased significantly. In contrast, the bacterial population in the ovaries was dependent on the reproductive state of the animal. An age-dependent degeneration of the midgut bacteriocytes was also investigated by microscopic techniques in males and female castes of the closely related ant species C. herculeanus and C. sericeiventris, respectively, with similar results and supports the concept of age-dependent degeneration of the midgut bacteriocytes in all castes.

Bacteria of the genus Blochmannia are obligate intracellular endosymbionts of ants of the genus Camponotus (17). This ant genus comprises about 1,000 species (5), which are ubiquitously present in most terrestrial habitats. Closely related bacteria are found also in the genus Polyrhachis which, like Camponotus, belongs to the Formicinae. Since Blochmannia could be detected in all Camponotus species investigated so far (>30), it is believed that this endosymbiosis is of great importance for the biology of these insects. The bacteria are localized in specialized cells, the bacteriocytes, which in the case of Camponotus species are intercalated between enterocytes in the midgut tissue of the ants. In queens and workers, the bacteria also occur in the ovaries and oocytes, and transmission of Blochmannia is strictly vertical (6, 12, 19). During embryogenesis the bacteria are concentrated in a ring of endodermal tissue destined to become the midgut. In certain developmental phases, the bacteria appear to spread transiently in various larval tissues, but in the last-instar larvae, they are localized exclusively in the bacteriocytes (16).

Phylogenetically, Blochmannia is closely related to other symbiotic bacteria of insects, such as Wigglesworthia glossinidia, which resides in bacteriocytes of the tsetse fly Glossina brevipalpis, and Buchnera aphidicola, which is located in bacteriocytes of aphids (2, 15, 17). All these endosymbiotic bacteria have extremely reduced genomes of between 450 and 710 kbp in size (1, 7, 10, 20-22). Aphid and tsetse fly hosts are characterized by their strict food specialization to certain diets which are poor in nutrients essential to the animals. In fact, it has been shown previously that these endosymbiotic bacteria supply the diet of their host animals with essential metabolites such as amino acids or vitamins, respectively (9). These physiological data obtained by classical biochemical methods have now been confirmed and extended by the analysis of the genome sequences of several Buchnera species and of W. glossinidia which revealed the strict conservation of the respective biosynthetic pathways in these bacteria, although many other biosynthetic pathways have been subject to genome rationalization processes and are missing or incomplete (1, 20).

In contrast to aphids and tsetse flies, the host animals of Blochmannia endosymbionts are generally considered to be omnivorous, so that a nutritional basis of this symbiosis is not obvious. On the other hand, a recent study of ant species in tropical rain forest canopies indicates that many of these ants, including several Camponotus species, can be considered “secondary herbivores” which mainly feed on plant or insect exudates, indicating that many of these ants are in fact food specialists at least in certain periods (8). It is even possible that food specialization originally was very widespread in these insects, but with the colonization of more temperate habitats, the Camponotus species adapted to the exploitation of a greater diversity of resources. The recently sequenced genome of Blochmannia floridanus suggests that the Blochmannia-Camponotus symbiosis continues to be based on nutrition physiology, because virtually all biosynthetic pathways leading to the biosynthesis of essential amino acids are retained in this microorganism, while those coding for biosynthesis pathways of nonessential amino acids are mostly missing (11).

Recently, we were able to remove the endosymbionts from C. floridanus workers by treating the host with antibiotics. These laboratory-reared animals did not show any adverse effects to this treatment and although they were virtually devoid of their bacterial endosymbionts they could not be distinguished in any way from the control group which was not subjected to antimicrobial treatment (16). This may indicate that the symbiosis is not essential for the adult insects; instead it may be mainly relevant for early developmental stages. In agreement with this assumption, we recently observed a degeneration of the symbiosis in C. floridanus queens several years of age which no longer harbored bacteria in their midgut bacteriocytes, although the ovaries continued to contain bacteria (18). In aphids or tsetse flies treated with antibiotics the development ceased or was delayed and the mortality increased. This, too, suggests an important role of the endosymbionts during host development. In the case of Buchnera it has been reported that the bacterial load of the animals increases during development, reaching a peak in the first-instar nymphs (14). On the other hand, real-time quantitative PCR experiments demonstrated an equal increase of aphid weight and Buchnera population, indicating that there are no population peaks of Buchnera in its host (3). Since multiplication of the bacteria might be indicative for those periods of the host life cycle in which the symbiosis is most relevant, we investigated the changes in the bacterial replication activity during the development of C. floridanus from the egg to the imago using a quantitative real-time PCR approach. Moreover, we provide further evidence for degenerative processes in the Blochmannia-Camponotus endosymbiosis which occur with increasing age of adult workers and males of Camponotus species.

MATERIALS AND METHODS

Ants.

The colonies of Camponotus floridanus, Camponotus sericeiventris, and Camponotus herculeanus were cultivated at a constant temperature of 30°C, 50% humidity, and a 12-h day-night rhythm. A fertile queen was present in each colony. They were fed with honey water and cockroaches (Nauphoeta cinera) twice a week. For male production young workers were isolated from their colony harboring a reproductive queen. After 4 to 6 weeks oviposition started in these orphaned workers.

Isolation of total DNA from Camponotus floridanus.

Ants were divided into different developmental stages: eggs; larval stages L1 (2 to 3 mm), L2 (3 to 4 mm), and L3 (>4 mm); pupal stages P1 (shortly after pupation), P2 (after metamorphosis), and P3 (short time before hatching); and workers W1 (immediately after hatching), W2 (age estimated only a few days), and W3 (age estimated a couple of weeks).

To isolate DNA, the animals were crushed in liquid nitrogen and homogenized in buffer DNA-A (10 mM Tris-Cl [pH 7.5], 60 mM NaCl, 10 mM EDTA). An equal volume of buffer DNA-B (200 mM Tris/Cl, pH 9.0, 30 mM EDTA, 2% sodium dodecyl sulfate) supplemented with RNase (1 mg/ml) and lysozyme (2 mg/ml). The sample was mixed carefully and incubated at 37°C for 3 to 4 h. Subsequently, proteinase K (0.25 mg/ml) was added and incubation was continued overnight. Finally, the DNA was purified by phenol-chloroform extraction and ethanol precipitated. The DNA pellets obtained from six individuals or approximately 40 eggs, respectively, were resuspended in 60 μl of distilled H₂O.

Real-time quantitative PCR.

A 130-bp fragment of the dnaX gene of B. floridanus was amplified by PCR using primers dnaXfor (CTC GTA AAT GGC GTC CAA AG) and dnaXrev (TCC AGA TCC TCT TGT ACC TGT G). RT-PCRs were performed using a DNA Engine Opticon System (MJ Research, Waltham, Mass.) and the qPCR Core Kit for Sybr Green I (Eurogentec, Cologne, Germany). All reactions were performed according to the manufacturers' recommendations. Genomic DNA of Blochmannia floridanus was isolated as described earlier (11), and 45 ng to 4.5 pg of genomic DNA was used as a standard in RT-PCR. The molecular weight of the B. floridanus genome was deduced from the genome size to be 4.356 × 10⁶ g/mol. In each RT-PCR, 20 μg of total ant DNA, isolated from different developmental stages, was used. Six independent DNA preparations of each developmental stage were measured six times each in RT-PCR, resulting in an overall replicate number of 36 for each developmental stage. The threshold cycle number was set to a fluorescence of 0.01. The same procedure was conducted using primers gyrBfw (GAG GAA GAG ATG TAT CAA TAC C) and gyrBrev (GAA ACT GCC GTT CCA TAC G), specifically amplifying a 160-bp fragment of the gyrB gene of B. floridanus, which is located far apart from the dnaX gene and close to the replication termination point.

The DNA fragments to be amplified in quantitative real-time PCR were amplified by PCR from 50 ng of genomic DNA of B. floridanus using primer pairs dnaXfor-dnaXrev and gyrBfw-gyrBrev, yielding products of 130 and 160 bp, respectively. A control PCR using genomic DNA of E. coli gave no results, demonstrating the specificity of the reaction. RT-PCR products were sequenced using primer dnaXfor for the dnaX gene and primer gyrBfw for the gyrB gene, and the sequences were shown to originate from B. floridanus.

Microscopic techniques.

Midgut preparations of C. sericeiventris workers were fixed in a solution of 4% cacodylate buffered paraformaldehyde containing 1% glutaraldehyde for 5 h. After being washed with 50 mM cacodylate buffer, the tissue samples were fixed for 90 to 120 min with 2% buffered OsO₄ solution and washed with water. The samples were stained overnight in 0.5% uranyl acetate (Sigma Aldrich), dehydrated in an ethanol series and embedded in Epon. For electron microscopic studies, ultrathin sections were stained with 2% methanolic uranyl acetate and lead citrate (Sigma Aldrich), and for light microscopic studies, 1- to 5-μm-thick histological sections were used. If not stated otherwise, all chemicals were supplied by Roth (Karlsruhe, Germany).

RESULTS

Bacterial replication during host development.

The bacterial genome copy number from different developmental stages of the ant C. floridanus was determined using a quantitative RT-PCR approach, amplifying a fragment of the dnaX gene. The bacterial genomic copy number per host individual was found to decrease by a factor of 10 from about 10⁷ to 10⁶ copies during the development of the egg to the second-instar larva (Fig. 1a). In contrast, a 10-fold increase in the bacterial genome complement was observed immediately after pupation, but prior to the onset of metamorphosis and a second boost in replication could be noted in older pupae immediately before eclosion. Bacterial replication then continued in young callow workers for a short period directly after eclosion, during which a further 10-fold increase in genome copy number occurred. After this peak, however, the bacterial genome copy number declined abruptly from stage W2 to W3 and reached a minimum of 10⁶ genomic copies in older workers. Overall, a 1,000-fold multiplication of the bacterial genome copy number was detected when comparing the minimum of genome copies per individual in larvae L2 and the maximum in young workers W1. These results were verified by amplifying a fragment of the gyrB gene of B. floridanus (Fig. 1b). In contrast to dnaX, gyrB is located far from the putative origin of replication, but nevertheless the results obtained from the amplification of gyrB were very similar to the dnaX data.

FIG. 1.

Number of genome dnaX copies per C. floridanus individual in different developmental stages of the ant determined from the number of dnaX fragments (a) and gyrB fragments (b). The average body weights of the sampled individuals were (in grams) 0.00013 (E), 0.00098 (L1), 0.00318 (L2), 0.01362 (L3), 0.01033 (P1), 0.00803 (P2), 0.00731 (P3), 0.00675 (W1), and 0.01001 (W3). Abbreviations: E, egg; L1 to L3, larval stages; P1 to P3, pupal stages; W1 to W3, adult stages.

Age-dependent degeneration of midgut bacteriocytes.

The decrease of the bacterial genome copy numbers in older imagines (adult ants) indicates that the symbiosis may degenerate with increasing age of the animals. In agreement with this result, earlier electron microscopic studies of C. floridanus queens revealed an age-dependent degeneration of the midgut bacteriocytes (18). In the present study we had the unique chance to investigate bacteriocyte degeneration in workers of the closely related ant species C. sericeiventris, which had been isolated from their colony and maintained in the laboratory for more than three years. The comparison of the bacteriocytes of several young workers of C. sericeiventris with these unusually old animals derived from the same nest revealed striking differences. In the young C. sericeiventris workers the bacteriocytes looked virtually identical to those described previously for several other Camponotus species (19). These bacteriocytes were completely filled with bacteria (Fig. 2). In contrast, in old C. sericeiventris workers only very few, if any, bacteriocytes could be detected and bacteria could only rarely be found in the remaining bacteriocytes. Occasionally, in young C. sericeiventris workers bacteria were found to be enclosed in vacuoles (Fig. 2). This is in contrast to all previous observations in other Camponotus species, including C. floridanus, C. rufipes, C. herculeanus, and C. ligniperdus, in which bacteria were found exclusively in the cytosol of the bacteriocyte (16, 19). It is not clear if these very rare events reflect a transient vacuole location of the bacteria during certain developmental stages or the onset of degeneration.

FIG. 2.

Electron micrograph of a cross section of a bacteriocyte of a young (a) and old (b) C. sericeiventris worker demonstrating the degeneration in older workers. Abbreviations: B, bacterium; V, vacuole.

Bacteriocyte degeneration not only occurs in female castes, it was also found in aging ant males (Fig. 3). These findings are based on light microscopic studies of midgut bacteriocytes in C. herculeanus males. Males of this species hibernate in the nest before they take off for the mating flight, after which the ant males die. In such males the bacteriocytes appeared to be ejected into the lumen of the midgut (Fig. 3c), and contrary to workers, not only the bacteriocytes but also a portion of epithelial cells were discharged into the lumen, resulting in a complete degeneration of the midgut epithelium. This phenomenon was even more pronounced upon starvation of the animals (Fig. 3d). Moreover, as already reported previously for C. floridanus males, in C. herculeanus males, bacteria could also be detected in epithelial cells flanking the bacteriocytes, a phenomenon which we did not observe so far in workers or queens (data not shown).

FIG. 3.

Sagittal sections through the midgut of C. herculeanus males of different age groups demonstrating the secretion of bacteriocytes and epithelial tissue. (a) One day after eclosion. Abbreviations: MG, midgut; C, crop; T, testes. (b) Midgut epithelium of a 3-month-old male. BC, bacteriocyte. (c) Midgut epithelium of an 8-month-old male. (d) Midgut epithelium of an old male that has ejected all its bacteriocytes after starvation.

Conservation of replication-competent bacteria in the ovaries of imagines.

While it seemed that the midgut population of Blochmannia degenerates with increasing age of the hosts, no quantitative data were available about the bacterial population in the ovaries. The characterization of a few C. floridanus queens, however, indicated that their ovaries retain the bacteria to guarantee their transmission to the progeny throughout the life span of the reproductive females (18). Since ant queens are difficult to obtain in a sufficient number, the following experiments were carried out with workers. C. floridanus workers, isolated from their colony, start to deposit haploid eggs which develop into males, while workers remaining in the colony do not produce offspring. In order to investigate the dynamics of the bacterial population in the ovaries, the number of genome copies was again determined by quantitative RT-PCR in sterile workers taken directly from the colony, and compared to the bacterial genome content of workers of a similar age that had been isolated from the colony for several months and subsequently produced male offspring. Since the midgut population should not be affected by the reproductive state of the animal, whole animals were used in this study. In nonreproducing workers, a bacterial population of 10⁶ genomic copies per host individual could be detected (Fig. 4), while in reproducing workers 10⁸ to 10⁹ bacterial genome copies were found. The bacterial population of the males produced by workers was similar to that of nonreproducing workers. Since the bacterial population in the midgut of workers significantly decreased with aging, as we know from microscopic studies, the high number of bacterial genomes detected in reproducing females should be assigned to bacteria located in the ovaries. This suggests that the animals maintain replication-competent bacteria at least in their ovaries throughout their life span and that replication of the bacteria is in some way linked to the reproductive activity of the host.

FIG. 4.

Number of dnaX copies per C. floridanus individual. Abbreviations: W3, old worker, not reproducing; Wm, reproducing worker; M, male.

DISCUSSION

In this work the replication activity of B. floridanus, the obligate intracellular endosymbiont of C. floridanus, was investigated at various developmental stages of the host animal by means of quantitative RT-PCR. This method allows the quantification of genomic copy numbers, albeit without generating any direct information about the actual population size of the bacteria, since so far the genome copy number per Blochmannia cell is unknown. The genome copy number may play an important role for endosymbiotic bacteria of insects. Buchnera, a close relative of Blochmannia, contains several hundred genome complements per bacterial cell (13). Because these bacteria have lost most of their specific transcription factors, it is likely that genome amplification is an effective way of global regulation of bacterial gene expression in Buchnera. Such a polyploid state cannot be ruled out for Blochmannia, since a striking feature of these bacteria located in the midgut bacteriocytes is their extraordinary length of up to 12 μm (4). The present work therefore focused on the determination of the number of genome copies in host specimens rather than the absolute number of bacteria.

A stepwise increase of bacterial genome copy numbers could be observed in prepupae, old pupae prior to eclosion, and in callow workers immediately after eclosion, while the bacterial genome number declined significantly in older adults. The multiplication of the symbionts at definite points of the larval development of the host may be indicative of those phases in which the symbiosis is of greatest importance. In fact, the steps of multiplication coincide with developmental phases of the host when high metabolic activities can be expected, suggesting that the symbiosis is important during pupation, and apparently also in very young adults. However, the latter phenomenon could in part be due to some “inertia” of the system, preventing an immediate stop of bacterial propagation as soon as the imagines eclosed, e.g., due to the half-life of the degradation of the proteins involved in DNA replication.

Insects need large amounts of aromatic amino acids during their development to produce catecholamines which are needed for the sclerotization of the cuticle. Aromatic amino acids are poorly soluble in water, and usually insects store large amounts of soluble tyrosine or phenylalanine compounds in their hemolymph to keep sclerotization running. The biosynthetic pathways for nonessential amino acids are largely reduced in Blochmannia. However, Blochmannia retained the ability to synthesize tyrosine (11). Therefore, in Camponotus, the problem of tyrosine supply might have been solved elegantly by harboring endosymbiotic tyrosine synthesizing bacteria. This is in agreement with the observed age-dependent degeneration of the midgut bacteriocytes in adults, when no more ecdyses take place.

The multiplication of the bacteria in young pupae also coincides with previous results demonstrating an emigration of bacteria from the bacteriocytes to the ovaries of the host individuals (6). Therefore, the observed bacterial replication at this stage may also be correlated with these events. It is, however, interesting that the onset of the first efficient bacterial replication bout takes place immediately after the bacteria reach the bacteriocyte in the midgut of the last-instar larvae (16).

In adult animals, RT-PCR indicated an age-dependent degeneration of the bacterial population in the midgut bacteriocytes. These data are in agreement with the observations made with old C. sericeiventris workers and C. herculeanus males reported here, but also with previously published data about bacteriocyte degeneration in older queens and reorganization of the midgut tissue with increasing age in C. floridanus workers (16, 19). This indicates that the symbiosis is of little or no importance in older adult animals. Accordingly, the removal of the bacteria from adult workers by treatment with antibiotics did not result in detectable consequences for laboratory-reared ants (16). In striking contrast, preliminary results indicate that antibiotic treatment of C. floridanus queens results in dramatic consequences for fertility since the development of the offspring is arrested at the stage of the first-instar larvae (F. Wolschin, unpublished data). In this developmental stage, RT-PCR revealed a decrease in the bacterial population and the development seems to stop immediately before the first increase of the bacterial population takes place. Therefore, Blochmannia may not only play a role during metamorphosis, but may also be essential for the very early developmental stages of the host.

The incremental character of the increase in the bacterial population demonstrates that bacterial replication is not a constitutive but a highly regulated process. Moreover, the reproductive activity of the host is in some way involved in the control of bacterial replication, since the bacteria in the ovaries even of old animals do not degenerate irreversibly, while the endosymbionts located in the bacteriocytes of adult animals may have lost their ability to replicate (16). This may indicate that either the bacteria have developed in different ways in the midgut and in the ovaries, respectively, or that putative host signals that induce bacterial replication in the ovaries of adult animals may no longer be present in the midgut bacteriocytes.