Two Newly Introduced Wolbachia Endosymbionts Induce Cell Host Differences in Competitiveness and Metabolic Responses

ABSTRACT

Wolbachia endosymbionts can induce multiple reproductive manipulations in their hosts, with cytoplasmic incompatibility (CI) being one of the most common manipulations. Two important agricultural pests, the white-backed planthopper (Sogatella furcifera) and the brown planthopper (Nilaparvata lugens), are usually infected with CI-inducing Wolbachia strain wFur and non-CI-inducing Wolbachia strain wLug, respectively. The biological effects of these infections when present in a host cell are unknown. Here, we introduced the two Wolbachia strains into an Aedes albopictus cell line to stably establish a wFur-infected cell line (WFI) and a wLug-infected cell line (WLI). In a mixed culture, WFI cells were completely replaced by WLI cells, pointing to a stronger competitiveness of the WLI cell line. We found that infection by both Wolbachia strains reduced cell growth rates, but WLI had a higher cell growth rate than WFI, and this difference in cell growth rate combined with possible Wolbachia differences in diffusivity may have affected cell competitiveness. By examining gene expression and metabolites in the two lines, we found that some genes and key metabolites responded to differences in cell competitiveness. These results point to potential mechanisms that could contribute to the relative performance of hosts infected by these strains and also highlight the substantial impact of a non-CI Wolbachia on metabolism, which may in turn influence the fitness of its native host.

IMPORTANCE Wolbachia transinfection in insects can be used to suppress pests and block virus transmission. We stably introduced two Wolbachia strains from rice planthoppers into cell lines of an important arbovirus mosquito vector, Aedes albopictus. The levels of competitiveness of host cells from the lines infected by the two Wolbachia strains were different, as were metabolic responses of the cell lines. These results suggest potential metabolic effects of Wolbachia on native hosts that could be exploited when they are transinfected into novel hosts for pest control.

INTRODUCTION

Many microbes form a stable symbiotic relationship with their insect hosts following a long-term evolutionary process (1–3). Wolbachia represents a widely distributed microbe group in arthropods that have a range of positive and negative fitness effects on their hosts (4–8). Because Wolbachia cells live inside the cells of host tissues, they spread through maternal lineages, which can be enhanced through Wolbachia effects on host reproduction to achieve a wide distribution (9–10).

One of the most important reproductive manipulations associated with Wolbachia is cytoplasmic incompatibility (CI), which occurs when a female uninfected with one endosymbiont produces fewer fertile offspring when it mates with a male infected with this endosymbiont (11). Recently, this phenotype has been used for biological pest control when males carrying a newly introduced Wolbachia strain are released into the field. For instance, the release of wPip-infected Aedes albopictus males obtained by artificial transinfection significantly suppressed mosquito populations in the field (12). Some vector populations infected with new Wolbachia strains (e.g., wStri [13], wMel [14], wMelPop [4], and wAlbB [15]) show reduced replication of transmissible viruses in their bodies, resulting in beneficial effects on disease transmission when the CI phenotype induced by these strains is used to invade the strains into field populations. Unlike genetic modification of insect hosts, transinfection of hosts with new Wolbachia strains is seen as a way of generating useful strains that links to the natural processes of transmission (16, 17) and is therefore not subject to the same regulatory controls.

The reproductive effects induced by Wolbachia depend on Wolbachia strains and host species (9, 18, 19). The evolutionary history and genome of a Wolbachia strain often affect its ability to manipulate host reproduction (20–22), which might also include parthenogenesis induction (PI) (e.g., Bryobia praetiosa [23], Leptopilina clavipes [24], and Franklinothrips vespiformis [25]), male killing (MK) (e.g., Drosophila innubila [26] and Ostrinia scapulalis [27]), and feminization (e.g., Eurema hecabe [28] and Zyginidia pullula [29]). A suitable host is also necessary condition for Wolbachia to induce reproductive phenotypes (19). Most Wolbachia strains induce the same phenotypes in hosts before and after transinfection (19), but the phenotypes induced by some Wolbachia strains change after being transferred to new hosts. For example, wMau and wUni lost the ability to induce CI when each of these infection strains was transferred to Drosophila simulans (30, 31), while non-CI-inducible strains (wYak, wTei, and wSan) induced CI when transferred to D. simulans (32), and wKue induced MK and CI, respectively, in donor and recipient hosts (33). In the absence of suitable strains and hosts, Wolbachia strains may not be able to induce CI or other reproductive changes. Strains/hosts such as wLug in Nilaparvata lugens and wAu in D. simulans fall into this category, although these strains have other effects on host fitness, such as improving host fecundity and having antiviral ability (5, 34).

Rice planthoppers are a class of important agricultural pests in rice-growing regions of Asia. They include the brown planthopper (N. lugens), white-backed planthopper (Sogatella furcifera), and small brown planthopper (Laodelphax striatellus). They often cause serious damage by sucking rice seedlings, laying eggs, and transmitting plant viruses. These rice planthoppers can be infected with different endosymbionts (35–37). Recently, the applied potentials of two endosymbionts in rice planthoppers have been explored. Wolbachia wStri of L. striatellus, and Cardinium cSfur of S. furcifera, both with CI inducibility, were stably transferred into N. lugens, and these strains were used to hinder the spread of plant viruses or reduce host fecundity (13, 38). In addition to cSfur, S. furcifera was infected with Wolbachia wFur, which induces CI in its host (36, 39). On the other hand, N. lugens was infected with Wolbachia wLug, which does not induce CI in its host (5). Wolbachia is an obligate intracellular bacterium, so it cannot autonomously be cultured in vitro. Cultured cell lines provide a favorable environment for exploring such differences between Wolbachia strains when placed into the same host.

The Aa23 cell line is naturally infected by Wolbachia strains that infect A. albopictus, and the cell line has been infected with other Wolbachia strains (such as wPip) in previous work, which suggests this cell line might be broadly permissive to supporting different Wolbachia strains. In this study, we introduced the two Wolbachia strains from planthoppers into a Wolbachia-uninfected A. albopictus cell line (WU) and stably established a wFur-infected cell line (WFI) and a wLug-infected cell line (WLI). Then, we explored the biological impact of the two Wolbachia strains and their metabolic effects on the cell hosts. The results showed that competitiveness differed between the two Wolbachia-infected cell lines. This provided an opportunity to compare levels of gene expression and metabolite production in the two lines to understand the potential basis of the competitive difference.

RESULTS

Stable introduction of two Wolbachia strains into cell lines.

In our experiment, a total of two novel cell lines were established by transfecting the Wolbachia strains from planthoppers into an uninfected cell line (WU), namely, the wFur-infected cell line (WFI) and the wLug-infected cell line (WLI) (Fig. 1A). During Wolbachia transinfection and cell passage, the infection status of cells was continuously checked to ensure that cell lines were stably infected with Wolbachia strains (Fig. 1B). The embryos of donor SFCW (Cardinium- and Wolbachia-infected S. furcifera) were coinfected with Wolbachia wFur and Cardinium cSfur. After transinfection, PCR detection of an electrophoretic band from wFur over 10 consecutive generations showed it gradually brightened until it was stable, while the electrophoretic band from cSfur gradually darkened until it disappeared. Thus, only wFur in SFCW embryos could be successfully transinfected into the Aa23 cell line of A. albopictus. The embryos of donor NLW (Wolbachia-infected N. lugens) were singly infected with Wolbachia wLug, which was also successfully transinfected into the Aa23 cell line. These results showed that WFI and WLI were each successfully established.

FIG 1.

Stable introduction of two Wolbachia strains into cell lines. (A) Schematic diagram representing the establishment process of the two Wolbachia-infected cell lines. The eggs from two species of planthopper were collected, and the cytoplasm of donor embryos was mixed with recipient cells. After continuous culture, two Wolbachia-infected cell lines were obtained. Green, Wolbachia wFur; red, Wolbachia wLug; blue, Cardinium cSfur. (B) PCR detection of Wolbachia (wFur and wLug) and Cardinium (cSfur) during cell passage. The wFur and cSfur infections coexist in SFCW, and wLug infection exists in NLW (n = 3). (C) Quantitative PCR detection of Wolbachia (wFur and wLug) and Cardinium (cSfur) during cell passage. The wFur and cSfur infections coexist in SFCW, wLug infection exists in NLW, and cSfur infection exists in SFC. Data are expressed as mean ± standard error of the mean (SEM [n = 6]). All asterisks indicate the significance of difference between WFI and WLI. **, P < 0.01; ***, P < 0.001; ns, not significant. (D to F) The distribution of Wolbachia in Wolbachia-uninfected (WU) (D), wFur-infected (WFI) (E), and wLug-infected (WLI) (F) cell lines. Nuclear DNA was stained blue by DAPI. White arrows indicate the red regions of Wolbachia staining. SFCW, Cardinium- and Wolbachia-infected S. furcifera; SFC, Cardinium-infected S. furcifera; NLW, Wolbachia-infected N. lugens.

The density dynamics of the two Wolbachia strains were measured to confirm the stability of symbiotic relationships between Wolbachia strains and cell hosts (Fig. 1C). During cell passage, we observed an increasing density up to the seventh generation and a fluctuating density around a high level after the seventh generation, with the two Wolbachia-transinfected cell lines showing a similar pattern. After the eighth cell passage, the relative density of Wolbachia showed a significant difference between the two lines, and wLug in WLI was higher than wFur in WFI (two-way repeated-measures analysis of variance [ANOVA] followed by Bonferroni-adjusted pairwise comparisons: G₈ to G₁₂, P < 0.001). In contrast, there was a gradual decrease of Cardinium density during the cell passages, reflecting the absence of a stable Cardinium transinfection. The distributions of the two Wolbachia strains in their cell hosts could be visualized (Fig. 1D to F). For both cell lines, Wolbachia distributed around all the nuclei, and all cells of the two lines were infected with Wolbachia (100% infection). In addition, a brighter fluorescence was found in WLI than in WFI, consistent with density differences between the infections. These results confirmed the stability of the two new Wolbachia-infected cell lines.

Differences in cell host competitiveness induced by two Wolbachia strains.

Competition experiments between the two cell lines were used to understand the biological properties and regulatory capabilities of the two Wolbachia strains in A. albopictus cells. PCR detection showed that the designed primers could accurately distinguish wFur and wLug (Fig. 2A). When they were cultured independently, the relative densities of the two Wolbachia strains in their own cell hosts fluctuated (Fig. 2B), but after the sixth passage, the density of wLug in WLI was higher than that of wFur in WFI (G_6, G₈, G₉, G₁₁ and G₁₂, P < 0.001; G₇ and G₁₀, P < 0.01). Then the two cell lines were mixed and cultured with the same initial cell number to compare competitiveness (Fig. 2C). The relative densities of the two Wolbachia strains were used to estimate the number of cells infected by the two strains. The relative density of wLug in WLI gradually increased until wFur in WFI was no longer detected. When WFI and WLI were mixed and cultured with an initial cell number of 2:1, the difference in Wolbachia densities between the two cell lines was expected to decrease, but WLI still predominated (Fig. 2D). These results reflected a continued competitive advantage for wLug-infected cells and also indicated that a difference in cell growth rate may contribute to the difference in cell competitiveness.

FIG 2.

Relative densities of two Wolbachia strains and growth rates of two cell lines during cell passage. (A) Accurate detection of wLug- and wFur-specific primers by PCR. (B to D) Relative densities of two Wolbachia strains during cell passage detected when the two cell lines existed independently (B) and the cell number ratios of WFI to WLI were 1:1 (C) and 2:1 (D), respectively. Data are expressed as mean ± SEM (n = 4). (E and F) The cell growth rates of two cell lines during each cell passage were tested when WFI and WLI existed independently (E) and the cell number ratios of WFI to WLI were 1:1 and 2:1 (F). Data are expressed as mean ± SEM (n = 6). (G) Comparison of cell growth rates among the three cell lines (WFI, WLI, and WU). Data are expressed as mean ± standard deviation (SD [n = 18]). All cell growth rates were expressed by cell proliferation multiples within 7 days, which were measured by calculating the ratio of cell number on the seventh day to the that on the first day during each cell passage. All asterisks indicate the significance of difference between WFI and WLI. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant. WU, Wolbachia-uninfected cell line; WFI, wFur-infected cell line; WLI, wLug-infected cell line.

For the two Wolbachia-infected lines, where Wolbachia density differed in many passages, a higher cell growth rate for WLI than WFI was often detected (G₁ and G₁₀, P < 0.001; G₃ and G₅, P < 0.05; G₆ and G₁₁, P < 0.01) when the cell lines were cultured independently, although this was not always the case (Fig. 2E). Cell growth rates fluctuated in the two lines during the cell passages, which may be due to several factors. In particular, cells in the original culture flask were thoroughly mixed during the cell passages and then moved to a new culture flask for adherent growth, and some cells may have died during the mixing process. Cell growth rate may also have been affected by heterogeneity of the Aa23 cells.

During cell passages, the cell growth rate and Wolbachia density of a line did not show consistent passaging curves, suggesting that cell growth rate and Wolbachia density may act independently to influence cell competitiveness. When the initial cell number ratios of WFI to WLI were 1:1 and 2:1, there were no significant differences in cell growth rates between these cultures during most passages (Fig. 2F). Because WLI was still a winner even at the 2:1 ratio (Fig. 2D), this comparison essentially reflects a comparison of WLI cells at later cell passages. We also found a significant difference in cell growth rates between the three lines (one-way ANOVA: F = 89.160, df = 2, 45, P < 0.001), with the uninfected cell line having the higher growth rate (Fig. 2G). Although the two Wolbachia strains both reduced cell growth rates (one-way ANOVA followed by Bonferroni-adjusted test: WFI/WU, P < 0.001; WLI/WU, P < 0.001), WLI had an advantage in cell growth rate over WFI (P < 0.001). These results indicated that WLI cells infected by wLug had a higher growth rate than WFI cells infected by wFur. The different cell growth rates may contribute to the difference in competitiveness of the two infected lines.

Gene expression profile of cell samples.

A total of 27 cell samples were sequenced, yielding an average of 6.66 Gb per sample (see Table S1 in the supplemental material). The average matching rate of the samples in the A. albopictus genome was 60.15%, and the average ratio of clean reads to raw reads was 90.25%. A total of 36,114 genes and a total of 51,211 novel transcripts were detected in all the cell samples. The sample correlation heat map for any cell line had high similarity (mean correlations: WU > 0.999, WFI > 0.999, and WLI > 0.996) (Fig. 3A). The levels of similarity between infected and uninfected samples were lower, while the samples compared between the two infected cell lines had high similarities (mean correlation, 0.993). This indicates that infection by the two Wolbachia strains led to many similar effects on the cell hosts. In addition, samples for each cell generation had a high similarity (mean correlations: G₁ > 0.998, G₂ > 0.998, G₃ > 0.994, G₅ > 0.995, G₇ > 0.998, and G₉ > 0.997) (see Fig. S4A in the supplemental material). With cell passage, similarity between the generation samples and the initial G₁ samples decreased. Many differentially expressed genes (DEGs) were observed between each of the Wolbachia-infected cell lines (WLI and WFI) and the noninfected cell line (WU), but far fewer were found between the infection types (Fig. 3B). For the competition experiment, the number of DEGs between successive passage samples and the G₁ samples suddenly increased at G₅ and then decreased slightly (Fig. S4B). The reverse transcription-quantitative PCR (RT-qPCR) validation of arbitrarily screened DEGs confirmed the accuracy of the transcriptome data (see Fig. S1 to S3 in the supplemental material).

FIG 3.

Results for the analysis of transcriptome samples, the number of differentially expressed genes (DEGs), and the Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment of selected DEGs in the three cell lines. (A) Correlation level of gene expression between samples based on Pearson’s correlation coefficient. The color represents the relative size of the correlation coefficient: the darker the color, the higher the correlation, and the lighter the color, the lower the correlation. (B) Numbers of upregulated and downregulated DEGs in each comparison group. (C) Venn diagram comparing the numbers of DEGs in three comparison groups. The bold text indicates mutual DEGs between two Wolbachia-infected cell lines (WFI and WLI) and the Wolbachia-uninfected cell line (WU); these mutual DEGs do not include the genes that are different in the three groups. (D and E) KEGG enrichment analysis of the upregulated (D) and downregulated (E) mutual DEGs from the Venn diagram shown above. (F) Scatter diagram showing the DEGs in the WLI/WFI comparison. (G and H) KEGG enrichment analysis of the upregulated (G) and downregulated (H) DEGs from the scatter diagram shown above. WU, Wolbachia-uninfected cell line; WFI, wFur-infected cell line; WLI, wLug-infected cell line.

Shared and independently expressed genes in the two Wolbachia-infected lines.

To further investigate patterns of gene expression in the two lines, Venn diagrams were constructed to compare mutual and independent changes in DEGs between the two Wolbachia-infected cell lines (WFI and WLI) and the Wolbachia-uninfected cell line (WU). Few genes were shared in all three comparisons, but there was a high degree of overlap when each of the infected lines was compared to the uninfected line (Fig. 3C). A total of 3,197 mutual DEGs, including 1,454 upregulated genes and 1,743 downregulated genes, were found. The Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis found that upregulated DEGs were related to processes, including translation (ribosome) and aging (longevity-regulating pathway), while differentially downregulated genes were mainly related to cell growth and death (cell cycle), amino acid metabolism (valine, leucine, and isoleucine degradation), etc. (Fig. 3D and E). Gene Ontology (GO) enrichment analysis found that DEGs were enriched to many terms (see Fig. S5 in the supplemental material). These results indicated effects of Wolbachia infection at the cell transcription level and decreased expression in genes connected to cell growth and death, consistent with the decreased growth rate of Wolbachia-infected cells. The upregulated cell longevity pathway may create a favorable condition for the successful transinfection and growth of Wolbachia.

Differently expressed genes between two cell lines infected by different Wolbachia strains.

The DEGs between WLI and WFI were compared to understand their connection to competitiveness (Fig. 3F). A total of 235 DEGs, including 94 upregulated genes and 141 downregulated genes, were found. KEGG enrichment analysis found that upregulated DEGs were mainly related to energy metabolism (e.g., carbon fixation pathways in prokaryotes), biosynthesis of other secondary metabolites (e.g., isoquinoline alkaloid biosynthesis), metabolism of other amino acids (e.g., glutathione metabolism), and so on; while downregulated DEGs were mainly related to some metabolism pathways (e.g., glycolysis/gluconeogenesis), the excretory system (e.g., PPAR signaling pathway), and signal transduction (e.g., AMPK signaling pathway). (Fig. 3G to H). GO enrichment analysis found that DEGs were enriched to many terms (see Fig. S6 in the supplemental material). These results revealed differences between WLI and WFI in DEG number, as well as in energy metabolism, metabolite synthesis, and signal transduction, processes likely to be closely related to cell growth and development.

Metabolite profile of cell samples with different infection statuses.

Metabolites of 18 cell samples from three cell lines were determined by nontargeted liquid chromatography-mass spectrometry (LC-MS). The significant aggregation of quality control (QC) samples in the principal-component analysis (PCA) plot indicates that metabolome data were of high quality (Fig. 4A). The distribution of WLI samples was more aggregated than that of the WFI or WU samples, but the cell lines could still be separated in the PCA. After quality control of the data, a total of 1,665 compounds were detected in positive-ion mode, of which 753 compounds were identified; a total of 426 compounds were detected in negative-ion mode, of which 213 compounds were identified (see Data Set S1 in the supplemental material). Therefore, the positive-ion mode with more identified compounds was used for subsequent analyses. Similar numbers of upregulated and downregulated differential metabolites (104 versus 94) were observed in the WFI/WU combination (Fig. 4B). In contrast, a low number of upregulated differential metabolites and a large number of downregulated differential metabolites were observed in the WLI/WU combination (60 versus 283), as well as in the WLI/WFI combination (78 versus 160). This showed that WLI had a greater metabolic response than WFI when they were compared with WU.

FIG 4.

Results from the metabolome samples, the number of differential metabolites, and the Kyoto Encyclopedia of Genes and Genomes (KEGG) terms and enrichment of selected differential metabolites in positive-ion mode. (A) Principal-component analysis (PCA) indicating clustering of samples from different treatments. The coordinate axis represents the percentage of contribution of each principal component (PC) to the total variance. (B) Numbers of upregulated and downregulated differential metabolites in each comparison group. (C) Partial least-squares discriminant analysis (PLS-DA) furthest showing the metabolome sample clustering of WFI and WLI. The coordinate axis represents the percentage of contribution of each principal component (PC) to the total variance. (D) Volcano plot showing the number of differential metabolites in the WLI/WFI comparison. (E) KEGG terms for upregulated and downregulated differential metabolites in the WLI/WFI comparison. (F and G) KEGG enrichment analysis of the upregulated (F) and downregulated (G) metabolites in the WLI/WFI comparison. WU, Wolbachia-uninfected cell line; WFI, wFur-infected cell line; WLI, wLug-infected cell line.

Differential metabolites between two cell lines infected by different Wolbachia strains.

Metabolites that showed different changes in the two cell lines were determined to explore possible reasons for differences in cell competitiveness. The WLI and WFI samples were significantly differentiated in partial least-squares discriminant analysis (PLS-DA) (Fig. 4C). Volcano plots showed the 78 upregulated and 160 downregulated metabolites differing between the infections (Fig. 4D), which showed that the metabolite concentration of WLI was lower than that of WFI in general. KEGG enrichment analyses found that upregulated metabolites were mainly related to nucleotide metabolism (purine metabolism), metabolism of cofactors and vitamins (pantothenate and coenzyme A [CoA] biosynthesis), while downregulated differential metabolites were mainly related to signaling molecules and interaction (neuroactive ligand-receptor interaction), as well as amino acid metabolism (histidine metabolism) (Fig. 4E to G).

Key genes and metabolites relate to competitiveness differences between the two Wolbachia-infected cell lines.

The association analyses between the DEGs and differential metabolites in the WLI/WFI comparison were used to understand possible reasons for the competitiveness difference between WFI and WLI. First, a regularized canonical correlation analysis (rCCA) showed a general correlation between DEGs and differential metabolites in the WLI/WFI comparison (Fig. 5A), which indicates that genes and metabolites are related. Then, strong correlations between DEGs and differential metabolites were established through Spearman correlation analyses at a conservative threshold. A total of 6 key metabolites and 25 key genes were significantly correlated at the conservative threshold, and these genes and metabolites were significantly differentially expressed in the WLI/WFI comparison (Fig. 5B and Table 1). Among these metabolites, most were upregulated in the WLI/WFI comparison, including choline, guanine, hypoxanthine, inosine, and 7-aminomethyl-7-deazaguanine, reflecting higher concentrations of these key metabolites in WLI than in WFI. These upregulated metabolites may promote some important functions in WLI cells, such as membrane transport, metabolism of cofactors and vitamins, global and overview maps, amino acid metabolism, nucleotide metabolism, and lipid metabolism. In contrast, only anandamide was downregulated in the WLI/WFI comparison, which could hinder signaling molecules and interactions in cells. In general, the changes in these key metabolites are expected to be conducive to the growth of WLI cells relative to WFI cells.

FIG 5.

Correlation analysis between differentially expressed genes (DEGs) and differential metabolites in the WLI/WFI comparison. (A) Clustering heat map of transcriptome and metabolome correlations based on regularized canonical correlation analysis (rCCA). Each row represents a differential metabolite, and each column represents a DEG. Red represents a positive correlation, and blue represents a negative correlation: the darker the color, the stronger the correlation. (B) Network diagram of the correlation between DEGs and differential metabolites based on Spearman’s rank correlation coefficients. Red and blue lines indicate positive and negative correlations between DEGs and differential metabolites, respectively. Red and blue circles represent upregulated and downregulated DEGs or differential metabolites, respectively.

TABLE 1.

Correlations between significantly enriched differential metabolites and differentially expressed genes in the WLI/WFI comparison

Compound ID	Metabolite name	Metabolite up/downregulation	Gene symbol	Gene product description	Gene up/downregulation	P value	Relativitya
0.726_103.0998	Choline	Up	LOC109421749	E3 ubiquitin-protein ligase MYLIP	Up	0.003	+
0.726_103.0998	Choline	Up	LOC109423716	Venom allergen 5	Up	0.003	+
0.726_103.0998	Choline	Up	LOC109430941	Uncharacterized LOC109430941	Up	0.003	+
0.726_103.0998	Choline	Up	LOC109430960	Protein artichoke-like	Down	0.003	−
2.404_151.0495	Guanine	Up	LOC109405454	Alpha-mannosidase 2	Down	0.003	−
2.404_151.0495	Guanine	Up	LOC109408847	Protein catecholamines up-like	Down	0.003	−
2.404_151.0495	Guanine	Up	LOC109411864	Innexin inx3-like	Down	0.003	−
2.404_151.0495	Guanine	Up	LOC109424324	Prostatic acid phosphatase-like	Down	0.003	−
2.404_151.0495	Guanine	Up	LOC109426331	Arrestin domain-containing protein 2-like	Down	0.003	−
2.404_151.0495	Guanine	Up	LOC109427327	Neprilysin-like	Down	0.003	−
2.404_151.0495	Guanine	Up	LOC109430062	Integrin alpha-PS2-like	Down	0.003	−
2.412_136.0386	Hypoxanthine	Up	LOC109411406	CDP-diacylglycerol–glycerol-3-phosphate 3-phosphatidyltransferase, mitochondrial	Down	0.003	−
2.412_136.0386	Hypoxanthine	Up	LOC109424328	Putative mediator of RNA polymerase II transcription subunit 26	Down	0.003	−
2.412_268.0808	Inosine	Up	LOC109400873	Phosphoenolpyruvate carboxykinase [GTP]-like	Down	0.003	−
2.412_268.0808	Inosine	Up	LOC109403795	Probable cytochrome P450 9f2	Down	0.003	−
2.412_268.0808	Inosine	Up	LOC109412576	Endoplasmin-like	Down	0.003	−
2.412_268.0808	Inosine	Up	LOC109415499	Phosphoenolpyruvate carboxykinase [GTP]-like	Down	0.003	−
2.412_268.0808	Inosine	Up	LOC109416951	Uncharacterized LOC109416951	Down	0.003	−
2.412_268.0808	Inosine	Up	LOC109423295	Activin receptor type-1	Down	0.003	−
2.412_268.0808	Inosine	Up	LOC109433487	Fatty acid amide hydrolase 2-B-like	Down	0.003	−
2.609_179.0809	7-Aminomethyl-7-deazaguanine	Up	LOC109400455	Phenoloxidase 2	Up	0.003	+
2.609_179.0809	7-Aminomethyl-7-deazaguanine	Up	LOC109403781	Probable cytochrome P450 9f2	Down	0.003	−
2.609_179.0809	7-Aminomethyl-7-deazaguanine	Up	LOC109414751	Cystinosin homolog	Down	0.003	−
2.609_179.0809	7-Aminomethyl-7-deazaguanine	Up	LOC109416785	Uncharacterized LOC109416785	Down	0.003	−
2.609_179.0809	7-Aminomethyl-7-deazaguanine	Up	LOC109430294	RWD domain-containing protein 1	Down	0.003	−
9.952_347.2824	Anandamide	Down	LOC109411406	CDP-diacylglycerol–glycerol-3-phosphate 3-phosphatidyltransferase, mitochondrial	Down	0.003	+
9.952_347.2824	Anandamide	Down	LOC109424328	Putative mediator of RNA polymerase II transcription subunit 26	Down	0.003	+

^aRelativity refers to the correlation between differential metabolites and DEGs. +, positive correlation; −, negative correlation.

Moreover, most genes that were related to upregulated key metabolites were downregulated in the WLI/WFI comparison, such as those encoding alpha-mannosidase 2, prostatic acid phosphatase-like protein, and probable cytochrome P450 9f2. Low expression of these genes leading to increases of the key metabolites may be linked to an increase in cell growth and competitiveness, while the effect of high expression of genes is less clear. The expression of some genes was positively correlated with upregulated metabolites, such as E3 ubiquitin-protein ligase MYLIP, venom allergen 5, and phenoloxidase 2, leading to increases in the concentrations of key metabolites. The greater numbers of upregulated differential metabolites and downregulated DEGs in WLI cells may work together to make WLI cells more competitive than WFI cells in a mixed culture.

DISCUSSION

In the present study, we introduced two Wolbachia strains with different CI inducibilities from rice planthoppers into an A. albopictus cell line and investigated their biological properties and regulatory capabilities in A. albopictus cell hosts. The two Wolbachia strains were stably infected and passaged in cell lines, and their infections caused changes in host gene expression. The competitiveness of WLI cells was stronger than that of WFI cells in mixed culture. Although the infection of both Wolbachia strains reduced cell growth rates, the WLI cell line we generated had a higher cell growth rate than the WFI cell line. The relative density of wLug was also higher than that of wStri in their respective lines in most passages. In addition, transcriptome and metabolome analyses further found that cells from the WLI line had more downregulated DEGs and key upregulated metabolites than WFI cells.

Our results revealed that the two Wolbachia strains from rice planthoppers could be stably established in A. albopictus cell lines, which raises the issue of whether endosymbionts from agricultural insects can be used in a broader context. Recently, the value of transfer of endosymbionts from agricultural insects into other vectors has become apparent from the phenotypes they induce. For instance, the establishment of wStri from L. striatellus transferred to A. albopictus cells could block Zika virus growth at two independent stages of viral replication (40), and its transinfection into N. lugens could induce CI and hinder plant virus replication (13); Tetranychus truncatus wTtru was also stably transinfected into A. albopictus cells (41), while a Cardinium endosymbiont transinfected into N. lugens reduced host fecundity (38). While wLug lacks CI inducibility in N. lugens, it could still have useful applications in other hosts. For instance, transinfection of wAu from Drosophila, where it lacks CI inducibility, into Aedes aegypti blocked virus replication (34). Moreover, some Wolbachia strains without CI inducibility, such as wYak, wTei, and wSan, were able to induce CI when transferred to new hosts (32). Both wFur and wLug may have antiviral activity, because they are related to wStri and both belong to the Wolbachia B-group, but this needs further work.

Although host transfers of endosymbionts are possible across substantial evolutionary distances, they will depend on host background as well. In the current work, Wolbachia and Cardinium of S. furcifera were cotransferred into the cells, but Cardinium was gradually lost. In contrast, after these two endosymbionts were transferred into N. lugens by embryo microinjection, Cardinium was stable and Wolbachia was lost (38). Perhaps A. albopictus is an incompatible host for Cardinium, unlike Wolbachia. Perhaps Cardinium does become established but has low competitiveness compared to Wolbachia in A. albopictus cells. Transcriptome analysis indicates that infection with the two Wolbachia strains could upregulate longevity-regulating pathways, potentially resulting in a longer life span and providing an advantage in competition with Cardinium.

For Wolbachia-transinfected lines, Wolbachia was aggregated or sparse in many cells, although all cells were infected, indicating that these cell lines also retained heterogeneous properties, as reported by O'Neill et al. (42). Most insect cell lines consist of a heterogeneous population of cell types, and Aa23 cells contain at least two cell types (42, 43). In our experiments, cell heterogeneity may contribute to the difference in Wolbachia distribution. Khoo et al. found that the number of Wolbachia bacteria varies in cells and that Aa23 cells can be classified into cell types with high, medium, and low numbers of Wolbachia cells (44). When researchers explored the cytological properties of the Wolbachia-infected cell line and the adaptability of Wolbachia to long-term passage in cell hosts, heterogeneity of Aa23 cells was noted (45, 46). In our study, several measures were used to reduce the impact of cell heterogeneity on stable cell culture and competition experiments, including thorough sampling of cells in the competition experiments and in the transcriptome and metabolite assessments, as well as the use of the double-standard-curve method for accurately determining the relative density of Wolbachia. We are therefore confident of the overall difference in competitiveness of the cell lines and transcriptional as well as metabolic differences.

When hosts are coinfected by two endosymbionts, only one endosymbiont may impact host phenotype, while the other plays an auxiliary role. For example, Wolbachia induced CI in Encarsia inaron, but Cardinium did not (47); in contrast, Cardinium induced CI and Wolbachia only played an auxiliary role in Tetranychus piercei (48). Likewise, Wolbachia dominated Spiroplasma and induced CI in T. truncatus (49). In other cases, two endosymbionts can also work together to impact a host’s phenotype. For instance, Wolbachia and Cardinium coinduced CI in S. furcifera (36, 39), while wAu and wAlbB cotransinfected into A. aegypti both blocked virus transmission and induced CI (34).

In our current work, the two infections did not coexist because WFI with a high initial cell number was gradually replaced by cells from the WLI line, indicating that WLI was more competitive than WFI in this cell host. In contrast, the two Wolbachia strains cultured separately could be stably infected and passaged in the cells. We suspect that the difference in cell growth rates may directly affect competitiveness of the two Wolbachia-infected cell lines. The faster division of WLI cells should lead to the replacement of WFI cells in mixed culture. Nevertheless, cells infected by both Wolbachia strains had a reduced cell growth rate compared to the uninfected cell line. Perhaps Wolbachia consumes host nutrients for survival, decreasing growth rate. However, growth might the be expected to depend on density, which was not the case given the higher density of wLug in WLI compared to wFur density in WFI. Endosymbiont density is thought to affect several phenotypes: for instance, a high density of endosymbionts leads to stronger CI in insect and spider mite hosts (50–52), and a high Wolbachia density confers stronger virus protection (53, 54). Differences in intracellular density may also be related to the regulation of Wolbachia strains through host autophagy (55).

Many DEGs and differential metabolites between the two Wolbachia-infected cell lines may underlie the difference in competitiveness. WLI had fewer upregulated DEGs and more downregulated DEGs compared with WFI, indicating that WLI generally had lower gene expression levels than WFI. The upregulated DEGs were mainly related to energy metabolism, the synthesis of secondary metabolites, and the metabolism of other amino acids, which suggested that WLI had a partially upregulated metabolic capability relative to WFI. For example, glutathione plays an important role in antioxidant defense, nutrient metabolism, and regulation of cellular events (56), and upregulation of it may help to delay senescence and enhance immunity to the host. Apart from metabolic functions, many downregulated DEGs were related to signal transduction, which may increase the time taken for molecular signals to pass through the cell membrane, perhaps promoting the rapid growth of the cells. For example, AMPK (AMP-activated protein kinase) is an important kinase in regulating energy homeostasis and a key protein in signal transduction pathways, and it can maintain regular physiological activities and prevent cellular rapid proliferation (57). Its downregulation may decrease the inhibition of rapid cellular metabolism. Although there were more downregulated metabolites than upregulated metabolites in the WLI/WFI comparison, the association analysis indicated that particular genes and key metabolites may affect differences in cell competitiveness. The key metabolites play an important role in cell metabolism functions. For example, choline is an important component of biofilms, which can promote fat and transmethylation metabolism and regulate cell apoptosis (58); inosine is a component of ATP, coenzyme A, RNA, and DNA, which participates in energy metabolism and increases the level of ATP (59). The key genes and metabolites that caused differential competitiveness could be artificially changed to further verify their functions.

Combined with the above considerations, we suggest that the replacement of WFI cells by WLI cells involved two processes (Fig. 6). First, our results showed that the cell growth rate of WLI was higher than that of WFI, indicating that infections by the two Wolbachia strains caused significant differences in cell growth rates. This difference in cell growth rates was likely to have directly contributed to cell replacement in the competition experiment, given that there would have been more WLI cells than WFI cells. Second, when the two cell lines were mixed and cultured, the higher growth rate of wLug over wFur may have contributed to the replacement of WFI by WLI. Regardless of whether the WFI/WLI initial cell number ratio was 1:1 or 2:1, the relative density of wLug increased. A difference in density could have resulted in wLug having stronger diffusivity than wFur in cells. Wolbachia strains may survive in the extracellular medium for a short time across the cell membrane. The wLug infection may have survived better in the culture medium, penetrating the membrane of the WFI cell and then attaching to cell microtubules (60). This attachment would have promoted the spread of newly infecting wLug into more cells with cell mitosis (61). In contrast, wFur may be less stably established in WLI cells due to its low density and show low diffusivity. Therefore, differences in cell competitiveness may be affected both by differential cell growth rates and by Wolbachia diffusivity.

FIG 6.

Model for a mechanistic understanding of the competitiveness difference between WFI and WLI. WLI had a greater Wolbachia density and more metabolites, such as choline, guanine, hypoxanthine, and inosine, than WFI. The same cell numbers of WLI and WFI were mixed, then WFI cells were completely replaced by WLI cells. The stronger competitiveness of WLI over WFI may depend on the high cell growth rate and the strong diffusion ability of wLug.

In summary, we introduced two Wolbachia strains with different CI inducibilities from rice planthoppers into A. albopictus cell lines. In mixed culture, WLI cells were more competitive than WFI cells. We found that the infection of both Wolbachia strains reduced cell growth rates, but faster cell growth rate and potentially stronger Wolbachia diffusivity of WLI over WFI could account for the difference in levels of competitiveness. WLI had fewer upregulated metabolites and more downregulated metabolites than WFI, and the association analysis found that specific genes and key metabolites may be the key to cell competitiveness. The study therefore highlights mechanisms potentially underlying interactions between Wolbachia and hosts at the cellular level. However, additional cell lines need to be developed with these and other Wolbachia strains from planthoppers to test the generality of the findings.

MATERIALS AND METHODS

Wolbachia strain and cell line.

The wFur and wLug strains of Wolbachia were from S. furcifera (white-backed planthopper) and N. lugens (brown planthopper), respectively. S. furcifera was collected from Yunnan Province, China, in 2017. Consistent with previous research, the field S. furcifera population (SFCW) was coinfected with Wolbachia wFur and Cardinium cSfur (36). Only the cSfur-infected S. furcifera population (SFC) was selected by treating SFCW with tetracycline for three consecutive generations. The Wolbachia wLug-infected N. lugens population (NLW) was collected and selected from Guangdong Province, China, in 2015. The Aedes albopictus cell line Aa23 was a gift from Xiaoying Zheng of Sun Yat-Sen University. The Wolbachia wAlbB strain, which was naturally present in Aa23 cells, was eliminated by treatment with 10 μg/ml rifampin. Cells of the Wolbachia-uninfected cell line (WU) were reared for 12 months before being used in experiments. All cells were cultured in 5 ml Schneider’s Drosophila medium (Gibco, United Kingdom) supplemented with 10% fetal bovine serum (FBS) (Gibco, Australia) in 25-cm³ cell culture flasks at 27°C. To avoid the influence of Aa23 cell heterogeneity on the experiments, the cells in the culture flask were thoroughly mixed during cell passage, and then the line was passaged with 4.5 × 10⁵ living cells into new cell culture flasks every 7 days, which counted as a passage.

Wolbachia transinfection.

The transinfection of Wolbachia in cell lines was performed with a modified shell vial technique (62). Briefly, mature fertilized females of SFCW and NLW were placed on fresh rice seedlings, where they laid eggs for 8 h. The newly laid eggs (which served as donors) of each line were carefully picked with a pinch and immersed in 1× phosphate-buffered saline (PBS), immediately washed three times for 5 min in 1× PBS, washed three times for 30 s in 2% liquor natrii hypochloritis, rinsed with 75% ethanol three times for a minute, and washed three times for 5 min in 1× PBS again (Fig. 1A). The sterilized eggs were immersed and ground in culture mediums. The grinding solutions were poured into culture mediums with WU cells (which served as recipients) in six-well plates. To make Wolbachia particles and cells fully fuse, six-well plates were centrifuged at 350 × g for 3 h at 4°C and then cultured at 27°C in a dark incubator. After 5 days, the transinfected cells were transferred into new cell culture flasks and then passaged to new culture flasks every 7 days thereafter. DNA was extracted from the remaining cells at each passage, and PCR was performed to detect their infection status by using the specific primers of the Wolbachia wsp gene and specific primers of the Cardinium 16S rRNA gene (Table 2) (63, 64). Once one cell line was infected with Wolbachia, PCR was performed in subsequent generations to ensure that Wolbachia could stably infect the cell line. The recipient cell line derived from the donor SFCW was tested for both Wolbachia and Cardinium, and the cell line derived from the donor NLW was only tested for Wolbachia. The experiments described above were continued until two Wolbachia-infected cell lines were stably obtained. To confirm that Cardinium derived from SFCW would not be transinfected into the cell line, the transinfection of donor SFC embryos in recipient WU cells was also tried according to the method described above.

TABLE 2.

Primer and probe sequences used for PCR, qPCR, and FISH

Organism	Target gene	Purpose	Primer or probe sequence (5′→3′)	Reference
Wolbachia	wsp	PCR for transinfection expt	Forward: TGGTCCAATAAGTGATGAAGAAAC	63
			Reverse: AAAAATTAAACGCTACTCCA
Cardinium	16S rRNA	PCR for transinfection expt	Forward: CGGCTTATTAAGTCAGTTGTGAAATCCTAG	64
			Reverse: TCCTTCCTCCCGCTTACACG
Wolbachia	ftsZ	qPCR for cell passage detection	Forward: TTATCACAGCAGGGATGGGT	36
			Reverse: TTTTTTCTTTTGCTCCTTTATCTTTAACTA
Cardinium	ftsZ	qPCR for cell passage detection	Forward: AGCATGTGCAAGCCCAAGAAGG	36
			Reverse: TGCTTTTGGCGGCAGTGGTT
A. albopictus	Actin	qPCR for housekeeping gene	Forward: GCAAACGTGGTATCCTGAC	61
			Reverse: GTCAGGAGAACTGGGTGCT
Wolbachia	16S rRNA	FISH for distribution detection	W1: AATCCGGCCGARCCGACCC	38
			W2: CTTCTGTGAGTACCGTCATTATC
Wolbachia wFur	wsp	qPCR for competition expt	Forward: CAGCAATCCTTCAAAAGCTGGT
			Reverse: AGCATCATCCTTAGCTGCCTT
Wolbachia wLug	wsp	qPCR for competition expt	Forward: AAGACGGAGATGTGGCTGATAA
			Reverse: TTGTCGCTAAAGGGTTGCTTAC

Density and distribution of Wolbachia during cell passage.

In order to understand the interaction between Wolbachia and Cardinium during cell passage, the relative densities of the two endosymbionts were characterized. After transinfection, genomic DNA of cell samples in 12 consecutive passages was extracted and used for quantitative PCR (qPCR) detection. The recipient cell line derived from the donor SFCW was detected for both Wolbachia and Cardinium, and the other two recipient cell lines, respectively, derived from the donors NLW and SFC, were only tested for Wolbachia or Cardinium. The specific primers of the Wolbachia single-copy ftsZ gene and the specific primers of the Cardinium single-copy ftsZ gene were used to determine gene copy numbers (Table 2) (36). The gene copy numbers of two endosymbionts were normalized against the actin gene copy numbers of A. albopictus to calculate the relative densities of two endosymbionts (Table 2) (61). The double-standard-curve method expressed the copy level of Wolbachia per unit cell gene copy, which should accurately reflect the relative densities of the two Wolbachia strains. Relative densities between the two Wolbachia strains were compared with two-way repeated measures ANOVA with post hoc Bonferroni-adjusted pairwise comparisons, and P values of <0.05 were considered significant.

The distribution of two Wolbachia strains in cells was visualized to determine the differences of their positions by using fluorescence in situ hybridization (FISH) with a modification (38, 65). Briefly, two Wolbachia-infected cell lines were continuously reared for more than 30 generations. These cells were cultured on sterile culture dishes with round slides, incubated at 27°C for 48 h, washed once for 5 min in sterile 1× PBS, immediately fixed for 30 min in fresh 4% paraformaldehyde, gently washed two times for 5 min in 1× PBS, immersed in hybridization solution overnight in 46°C, washed once for 5 min in each of four solutions (2× sodium chloride-sodium citrate buffer [SSC] with 0.015% [wt/vol] dithiothreitol [DTT], 1× SSC with 0.015% [wt/vol] DTT, 0.5× SSC with 0.015% [wt/vol] DTT, and 1× PBS alone), stained for 30 min with Vectashield antifade mounting medium with DAPI (4′,6-diamidino-2-phenylindole) (Vector Laboratories, Burlingame, CA, USA) at room temperature, and washed three times in 1× PBS. The slides were taken out from the culture dish and mounted with an antifluorescence quenching mounting medium, and then scanned immediately with a Leica TCS SP8 laser confocal microscope. Two rhodamine 5′-end-labeled Wolbachia 16S rRNA probes were used (Table 2) (38). WU as a negative control was also stained by the method described above.

Competitiveness of cell lines with different Wolbachia strains.

Competition experiments were performed to investigate differences in competitiveness between the two cell lines infected by different Wolbachia strains. First, the relative densities of wFur and wLug were measured when WFI and WLI were passaged independently with the same initial cell number. The 4.5 × 10⁵ living cells of WFI or WLI were used as the initial bases for cell expansion. Afterward, when the initial cell number ratio of WFI/WLI was 1:1, the relative densities of wFur and wLug were determined during cell passage. Total numbers of 4.5 × 10⁵ mixed cells, consisting of 2.25 × 10⁵ WFI cells and 2.25 × 10⁵ WLI cells, were used as the initial base for cell expansion. Finally, based on the above results, when the initial cell number ratio of WFI to WLI was 2:1, the relative densities of wFur and wLug were detected during cell passage. A total of 4.5 × 10⁵ mixed cells, consisting of 3 × 10⁵ WFI cells and 1.5 × 10⁵ WLI cells, served as the initial base for cell expansion. All cells were passaged every 7 days for 12 consecutive generations, and 4.5 × 10⁵ cells were passaged each generation. In order to reduce any influence of cell heterogeneity on competition experiments, the three cell lines were cultured by an identical passaging method. The cells in a culture flask were thoroughly mixed during each cell passage. The relative densities of two Wolbachia strains during the cell passage were determined by qPCR, which was used to distinguish two cell lines and their relative cell numbers. Specific quantitative primers of the wFur wsp gene and the wLug wsp gene were designed to determine gene copy number (Table 2). Copy numbers of two Wolbachia strains were normalized against actin from A. albopictus to calculate their relative densities. Three biological replicates were performed during cell passage, and each biological sample was performed with four technical replicates.

To understand the difference in cell growth rates between two Wolbachia-infected cell lines, cell proliferation multiples within 7 days were measured by calculating the ratio of cell number on the seventh day to the that on the first day during each cell passage. For the first day (i.e., at the beginning of each cell passage), the total cell number in each shell flask was 4.5 × 10⁵. For the seventh day (i.e., at the end of each cell passage), the total cell number was counted by measuring cell density using an Invitrogen Countess 3 automatic cell counter. Cell proliferation multiples within 7 days at each passaging were tested for all treatments. Three biological replicates were performed, and each biological sample was tested twice (two technical replicates). During cell passages, cell growth rates and relative densities between the two Wolbachia strains were tested with two-way repeated-measures ANOVAs with post hoc Bonferroni-adjusted pairwise comparisons. In addition, to compare the growth rates of Wolbachia-infected and uninfected cells, cell proliferation multiples within 7 days were determined as described above. Eight biological replicates were performed, and each biological sample was tested twice. A one-way ANOVA with post hoc Bonferroni-adjusted pairwise comparisons was used to compare the differences between the groups. All P values of <0.05 were considered significant.

Transcriptome sequencing and analyses.

Samples of the three cell lines were collected for transcriptome sequencing to determine changes in gene expression induced by the Wolbachia strains. A total of 4.5 × 10⁵ living cells of WFI, WLI, and WU were used as initial bases for cell expansion, and cells were collected on the seventh day. Cells of G₁, G₂, G₃, G₅, and G₇ in the competition experiment with a WFI/WLI ratio (2:1) were also collected. Each treatment involved three biological replicates from different shell flasks. Total RNA of cell samples was extracted with TRIzol (Invitrogen, Carlsbad, CA, USA), and the samples’ quality and integrity were determined by using a fragment analyzer. The mRNA with a poly(A) tail was enriched from total RNA by using magnetic beads with oligo(dT). Subsequently, a cDNA library was prepared and sequenced with a BGISEQ-500 platform (BGI, Shenzhen, Guangdong, China). Raw reads obtained by sequencing were used for quality control by internal BGI SOAPnuke software (v1.5.2) to filter out low-quality reads and generate clean reads. Then, clean reads were mapped to the reference genome of A. albopictus by using HISAT (Hierarchical Indexing for Spliced Alignment of Transcripts, v2.0.4). The second quality control was performed according to the mapping rate of the reference genome and the distribution of reads on the reference sequence. After quality control, subsequent analyses were performed.

Clean reads were mapped to reference gene sequence using Bowtie2 (v2.2.5), and then gene expression levels were calculated by RSEM (v1.2.12). The values of FPKM (fragments per kilobase per million) indicate gene expression levels. The fold change in gene expression between different groups was determined by calculating the ratio of average FPKM values of comparable groups. The significance of differentially expressed genes (DEGs) was tested with DESEQ2 (v1.4.5). DEGs between the comparison groups were identified using the following parameters: |log₂ fold change| of >1 and Benjamini-Hochberg-adjusted P value of <0.05. To reflect the correlation of gene expression between samples, the Pearson correlation coefficients of all gene expression between each two samples were calculated, and these coefficients were reflected in the form of the heat maps. Venn diagrams and scatterplots of DEGs in different comparison groups were performed based on the above parameters. DEGs with average FPKM values of >1 were selected to describe gene function more accurately. KEGG and GO annotations were obtained from the KEGG Ortholog (KO) database (https://www.genome.jp/kegg/ko.html) and Gene Ontology (GO) database (http://www.geneontology.org/). KEGG (https://www.kegg.jp/) and GO (http://www.geneontology.org/) enrichment analyses of annotated DEGs were performed by Phyper based on a hypergeometric test. The statistical significance values of terms and pathways were calculated with a Bonferroni-adjusted P value (P < 0.05). To verify the accuracy of transcriptome sequencing and analysis, arbitrarily selected DEGs were subjected to reverse transcription-quantitative PCR (RT-qPCR) using designed primers (Table 3).

TABLE 3.

Primer sequences used for RT-qPCR detection

Gene symbol	Gene product description	Primer sequence (5′→3′)
		Forward	Reverse
LOC109398643	PIN2/TERF1-interacting telomerase inhibitor 1-like	GAGCCTTCACCAGAGGAACC	TCTTCCGCGGGAGCTTTATC
LOC109400264	Heat shock protein 70 B2	GTTCTGGTGAAGATGCGGGA	TGTCGCTGGGAATCGTTGAA
LOC109404421	PIN2/TERF1-interacting telomerase inhibitor 1-like	CTCCTTCGGATGAAGCTCCC	TGCTGTTCTTCGGTGACCTC
LOC109417533	Lipopolysaccharide-induced tumor necrosis factor-alpha factor homolog	AAACTCACATCTTCGCCGGT	AGTTCGGGCAGTAGTGGTTG
LOC109417676	60S ribosomal protein L37a	GTACCCGTTATGGTGCCTCC	TTGCAGGACCAGATTCCGAC
LOC109403945	Endoribonuclease Dicer	GGTGGATCCGAAGACTAGCG	GTCTTGTCCCCCATGGTCTG
LOC109414904	Histone PARylation factor 1-like	GTGCAAACGTCCACCAACTC	TAGATGTCACTCATCGCGCC
LOC109417111	Deoxynucleoside kinase	GCCATTCACCGTGTTCATCG	TCGCCACTTTTCCACAGGTT
LOC109420799	Ribonucleoside-diphosphate reductase large subunit	AACGGCATCTCCAACGGATT	CGCGCCTGTGTTCTTTTTCA
LOC109421080	Innexin inx2-like	ACATAGCTCAGCCGGGAGTA	TCTTGATCCTGCCTGCTTCC
LOC109403076	Protein SHQ1 homolog	AACTGGTGTTCACGCTTCCA	GGCGTTTCGTTTTTCACCGT
LOC109407766	Isocitrate dehydrogenase [NADP] cytoplasmic	GCCACGGACTTTGTTGTTCC	ACTCGTGAACGACGTAGCTG
LOC109418926	Isocitrate dehydrogenase [NADP] cytoplasmic-like	CGATCGTGATCGGTCGTCAT	ACTCGTGAACGACGTAGCTG
LOC109421750	Uncharacterized	TCATGATAAGCGGAGGTCGC	TCTGAAGGTCTGGCGGAAAC
LOC109426551	NADH dehydrogenase [ubiquinone] flavoprotein 2, mitochondrial-like	GGCGAAACAACGAAGGATGG	TGCTGCAAGTCCGTCAAGAT
LOC109397824	tRNA dimethylallyltransferase, mitochondrial-like	GCAGTAGGCCGGAAGAAAGT	TCCACTTGATCAGGCGCAAT
LOC109398140	Sphingomyelin phosphodiesterase-like	TTCTTCGGCCATACGCACTT	GTCACCGAAGGCCCAATGTA
LOC109405369	tRNA dimethylallyltransferase, mitochondrial-like	GCAGTAGGCCGGAAGAAAGT	TCCACTTGATCAGGCGCAAT
LOC109414445	Uncharacterized	TCCAAAGCCTACCCCCAGTA	ATGGCTGATTCGTCGGAGTC
LOC109415499	Phosphoenolpyruvate carboxykinase [GTP]-like	GGATCAGTATGGAAGCGCGA	AGCATTCGTGTTCGTCCACT
LOC109404421	PIN2/TERF1-interacting telomerase inhibitor 1-like	CTCCTTCGGATGAAGCTCCC	TGCTGTTCTTCGGTGACCTC
LOC109407505	Polyadenylate-binding protein 4-like	GTTCCGTGCGAACGCTAATC	AGCCGTTTGACGGTATTGGT

Metabolomics analysis.

To compare changes in intracellular metabolites induced by Wolbachia strains, nontargeted metabolomics of three cell lines were determined by liquid chromatography-mass spectrometry (LC–MS). A total of 4.5 × 10⁵ living cells of WFI, WLI, and WU were used as the initial bases for cell expansion. A total of 1 × 10⁷ cells for each of three cell lines were collected on the seventh day, and six biological replicates for each cell strain were obtained from different shell flasks. The collected cell samples were frozen in liquid nitrogen for 15 min, and then metabolites were extracted. Metabolite profiles were analyzed using a high-performance liquid chromatograph (Waters 2D UPLC, Waters, USA) and high-resolution mass spectrometer (Q Exactive HF; Thermo Fisher Scientific, USA), and the data from positive-ion and negative-ion modes were collected separately to improve the coverage of metabolites. To endure data accuracy, metabolites with the same volume as for the 6 biological replicate samples of each cell line were mixed into a sample used for quality control (QC) during metabolite determination. Six needles of blank solvent were performed before determining metabolites in the samples, and the first 100 ions of the response intensity in the sixth needle of blank solvent were excluded. Five needles of QC samples were continuously used to monitor the status of the instrument and balance the LC-MS system. After confirming the integrity of the instrument, cell samples were tested. In the testing process of cell samples, QC samples were randomly tested. The repeatability data of QC samples were used to evaluate the quality of the experimental data. To decrease the influence of instrument status on sample detection, the test sequence of experimental samples was randomized.

Principal-component analyses of all measured cell samples indicated that QC samples were highly clustered (see Results), reflecting high-quality data. At the same time, the number of ions with small fluctuations (i.e., coefficient of variation of ≤30%) for ionic strength in all QC samples accounted for more than 90% of the variation, which also indicates that the collected data are qualified. Compounds with a coefficient of variation of relative peak area greater than 30% in all QC samples were deleted. Metabolites were identified by processing raw data collected from LC-MS in Compound Discoverer 3.0 (Thermo Fisher Scientific, USA), in combination with BGI Library, mzCloud, and ChemSpider (HMDB, KEGG, and LipidMaps). All identified metabolites were listed and compared between different groups. Partial least-squares discriminant analysis (PLS-DA) was used to reflect the difference between two comparison groups by using metaX software (BGI, Shenzhen, Guangdong, China). Differential metabolites between comparison groups were screened according to the following parameters: VIP values in first two principal components of PLS-DA model of ≥1, fold change of ≥1.2 or ≤0.83, and P value of <0.05. Venn diagrams and volcano plots of differential metabolites for the comparison groups were constructed based on the parameters described above. The KEGG functional annotation of differential metabolites was performed to understand KEGG terms and pathways. The statistical significance values of terms and pathways were calculated with a Bonferroni-adjusted P value (P < 0.05).

Association analyses of differentially expressed genes and differential metabolites.

An association analysis between the cell transcriptome and metabolome can help establish connections between changes in gene expression and biological phenotypes. To understand the reasons for the competitiveness difference between WFI and WLI, association analyses between DEGs and differential metabolites in WLI/WFI comparison were performed. Each treatment of transcriptome had three replicates, and each treatment of metabolome had six biological replicates. Thus, all replicates of the transcriptome and a random three replicates of the metabolome in the association analyses were used. The association analyses were carried out in two steps. For the first step, the purpose was to test the correlation between levels of gene expression of the transcriptome and metabolite concentrations of the metabolome in the WLI/WFI comparison. The DEGs (Benjamini-Hochberg-adjusted P value of <0.05) and differential metabolites (fold change of ≥1.2 or ≤0.83 and P value of <0.05) in the WLI/WFI comparison were screened according to a general threshold. A regularized canonical correlation analysis (rCCA) (66) was performed between these DEGs and differential metabolites to measure their correlation. For the second step, the purpose was to indicate links between specific genes and metabolites that might affect the competitiveness of the two Wolbachia-infected cell lines. The DEGs (|log₂ fold change| of ≥1, average FPKM values of ≥1, and Benjamini-Hochberg-adjusted P value of <0.05) and differential metabolites (fold change of ≥1.2 or ≤0.83, VIP values of ≥1, and P value of <0.05) in the WLI/WFI comparison are screened according to the original rigorous threshold. Spearman correlation analysis between substantial changes in the expression of the genes and the concentration of metabolites was undertaken. The correlation network was plotted with the following parameters: correlation coefficient of greater than 0.98 or less than −0.98 and P value of <0.001. The original input data used in both analyses were intensity values, which were gene expression levels from the transcriptome and the metabolite concentration levels from the metabolome.

Data availability.

The data from transcriptome sequencing have been deposited in the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) under BioProject accession no. PRJNA706551. The profiles of all metabolites are shown in Data Set S1.