Proteomic profiling of a robust Wolbachia infection in an Aedes albopictus mosquito cell line

Abstract

Wolbachia pipientis a widespread vertically transmitted intracellular bacterium, provides a tool for insect control through manipulation of host-microbe interactions. We report proteomic characterization of wStr, a Wolbachia strain associated with a strong cytoplasmic incompatibility phenotype in its native host, Laodelphax striatellus. In the Aedes albopictus C/wStr1 mosquito cell line, wStr maintains a robust, persistent infection. MS/MS analyses of gel bands revealed a protein “footprint” dominated by Wolbachia-encoded chaperones, stress response and cell membrane proteins, including the surface antigen WspA, a peptidoglycan-associated lipoprotein and a 73 kDa outer membrane protein. Functional classifications and estimated abundance levels of 790 identified proteins suggested that expression, stabilization and secretion of proteins predominate over bacterial genome replication and cell division. High relative abundances of cysteine desulfurase, serine/glycine hydroxymethyl transferase, and components of the α-ketoglutarate dehydrogenase complex in conjunction with above average abundances of glutamate dehydrogenase and proline utilization protein A support Wolbachia genome-based predictions for amino acid metabolism as a primary energy source. wStr expresses 15 Vir proteins of a Type IV secretion system and its transcriptional regulator. Proteomic characterization of a robust insect-associated Wolbachia strain provides baseline information that will inform further development of in vitro protocols for Wolbachia manipulation.

Introduction

Wolbachia pipientis (Rickettsiales; Anaplasmataceae), an obligate intracellular alphaproteobacterium originally described as a rickettsia-like microbe in reproductive tissues of mosquitoes (Hertig, 1936), is now known to infect a wide range of insects and other arthropods as well as filarial nematodes (Haegman et al., 2009; Ferri et al., 2011; Zug and Hammerstein, 2012). In contrast to its sister genera Anaplasma and Ehrlichia, which include vector-borne pathogens of vertebrates, Wolbachia does not infect vertebrates. Wolbachia is classified into phylogenetic clades referred to as supergroups (Lo et al., 2007). The best-studied supergroups are A (WOL-A) associated with insects, B (WOL-B) primarily associated with insects and isopods, and C and D (WOL-C and WOL-D) associated with filarial nematodes. Based on the abundance and diversity of its insect hosts, Wolbachia may be the most widespread intracellular bacterium on earth (Zug and Hammerstein, 2014).

WOL-C and WOL-D strains are obligate mutualists, with phylogenies concordant with those of their hosts, and are not known to undergo horizontal gene transfer. In contrast, discordant phylogenies between Wolbachia and insect hosts, and evidence for lateral gene transfer and recombination among WOL-A strains, have confounded designation of Wolbachia strains as unique species (Klasson et al., 2008; Ellegard, 2013). Although prophage, repetitive and mobile DNA sequences and protein genes encoding ankyrin repeats are more abundant in genomes of Wolbachia that occur in arthropods, relative to those that infect filarial worms (Wu et al., 2004; Newton and Bordenstein 2011), these elements also occur in genomes of other members of the Rickettsiales that undergo both vertical and horizontal transmission (Cho et al., 2007; Darby et al., 2007; Felsheim et al., 2009; Newton and Bordenstein, 2011).

Wolbachia infections in insects, typified by wMel, a WOL-A strain from Drosophila melanogaster, and wPip, a WOL-B strain associated with Culex pipiens mosquitoes (Klasson et al., 2008; Salzberg et al., 2009) are considered reproductive parasites, rather than mutualists. Unlike Wolbachia in filarial worms (Foster et al., 2005; Darby et al., 2012), insect-associated Wolbachia can be eliminated by antibiotic treatment without deleterious effects on the host and obligate mutualism is rare (Zug and Hammerstein, 2014). In insects, phenotypes of Wolbachia infection include parthenogenesis, altered sex ratios, and reduced egg hatch through mechanisms such as cytoplasmic incompatibility (CI). These effects on reproduction facilitate Wolbachia's invasion of uninfected host populations (Yen and Barr, 1971, Werren, 1997), and can be harnessed as a gene drive agent for insect control (Sinkins and Gould, 2006). Although underlying mechanisms remain to be discovered, Wolbachia's interference with arthropod immunity and replication of arboviruses, bacteria and plasmodia (Molina-Crux et al., 2008; Moreira et al., 2009; Pan et al; 2012) increases enthusiasm for developing Wolbachia-based control of arthropod vectors that transmit pathogens of humans, livestock and plants.

An important advantage of Wolbachia strains that infect arthropods is their adaptability to cell culture. For example, wMel has been used as a surrogate to evaluate antibiotic-induced changes in the less tractable, but more medically relevant filarial Wolbachia (Darby et al., 2013). Although the genome of wMel has been sequenced, wMel does not induce a strong CI phenotype in its Drosophila host. Because CI and other reproductive distortions induced by Wolbachia are particularly relevant to implementation of Wolbachia infections for vector control (Laven, 1967b; Yen and Barr, 1971), we have focused on strain wStr, which expresses strong CI in its planthopper host (Noda et al., 2001, 2002; Nakamura et al., 2012), and, unlike the mosquito strain, wAlbB (Fallon, 2008), maintains a robust, persistent infection in the well-characterized C7-10 mosquito cell line (Fallon et al., 2013a). Here we report a wStr proteome of 790 proteins, of which 68% are based on identification of at least three unique peptides, the most in-depth proteomic analysis of a Wolbachia strain to date. Our results provide a significant advance in the characterization of a strong and persistent Wolbachia infection in the clonal C7-10 mosquito cell line, which exhibits a robust immune response (Fallon and Sun, 2001). We anticipate that our results will facilitate investigation of Wolbachia-host interactions and development of cell-based in vitro tools, such as transformation, to support genetic manipulation of Wolbachia and its development for control of arthropod-borne disease. Successes with other genera in the Rickettsiales (reviewed by Beare et al., 2011) support the likelihood that genetic transformation of Wolbachia will eventually be achieved.

Results

Cell Growth

To facilitate eventual genetic manipulation of Wolbachia, we developed a persistently infected Aedes albopictus mosquito cell line (C/wStr) that supports robust growth of Wolbachia strain wStr from the planthopper, Laodelphax striatellus in a chemically defined medium in which nutritional components and/or selective drugs can be systematically evaluated (Fallon et al., 2013a). C/wStr1 cells grow as a monolayer culture with a doubling time of approximately 35 h, and have been maintained without supplementation with uninfected cells for more than 70 passages. Cytoplasm of infected cells contains abundant Wolbachia that stain green with BacLight^TM (Fig. 1, compare panel A to B) and associate with membranous structures (C, arrow), while host cell nuclei begin to fluoresce red, indicating compromised membrane integrity. Visual examination shows that heavily infected cells contain a minimum of 50 – 100 Wolbachia. Heavily infected clusters have vesicle-like structures (D) that escape into the culture medium (E). In late stage infections, cells detach from the substrate as aggregated masses, vesicle-like structures are largely absent, while most nuclei fluoresce red with strong contrast to fluorescent greenWolbachia (F).

Figure 1.

Clusters of uninfected A. albopictus C7-10 (A) and wStr-infected C/wStr1 cells (B - F) stained with BacLight^TM Live/Dead stain. Arrows indicate green-stained wStr associated with apparent membranous and vesicle-like structures.

Wolbachia enrichment

We adapted the sucrose-phosphate buffer (SPG) and density gradient centrifugation methods developed for enrichment of rickettsiae and mitochondria (Bovarnick et al., 1950; Weiss et al., 1975; Dubin and HsuChen, 1983) to fractionate C/wStr1 host cells. Following lysis of host cells by sonication, Wolbachia associate with particulate material in the lipid-rich fraction at the top of the gradient (GF-30; Fig. 2). Treatment with lipid-disruptive organic reagents, with phophoplipases A and D, or with lipase Type VII releases Wolbachia from the particulate material but their usual rod-like morphology is lost, a predominance of smaller rounded forms distribute throughout the gradients, and the Wolbachia stain red, suggesting loss of membrane integrity. Addition of EGTA to manipulate lipase activity results in nearly complete dispersal of Wolbachia, regardless of the presence of exogenous lipases. Furthermore, most Wolbachia stain green in the presence of EGTA, retain normal morphology, and sediment to the 50-60% sucrose interface (GF-50/60), while most Mitotracker Red^TM -stained mitochondria sediment to GF-30/40. The intermediate GF-40/50 fraction contains a mixture of both bacteria and mitochondria. The pattern of Wolbachia distribution within the sucrose gradient resembles that of wBol1 cultivated in A. albopictus RML-12 cells and enriched on a Percoll gradient (Duplouy et al., 2013). Our final buffer formulation differed from SPG (Bovarnick et al., 1950) by removal of glutamate and inclusion of 10 mM EGTA (SPE buffer). Wolbachia were recovered from gradient fractions by dilution to 15% sucrose, centrifugation and resuspension in SPE.

Figure 2.

Schematic flow chart of sample preparation and MS/MS methods used to produce MS data sets D – G. As depicted at left, MS data sets D and E were derived from total cellular and subcellular protein extracts that were fractionated by SDS PAGE and trypsin-digested for protein detection by LTQ MS/MS. As depicted at right, MS data sets F and G were derived from gradient fraction GF-50/60 (enriched in Wolbachia from C/wStr cells) digested in-solution with trypsin for HPLC separation of peptides and protein detection by Orbitrap MS/MS.

We monitored Wolbachia enrichment based on a Coomassie blue-stained 24 kDa protein band (labeled “C” in Fig. 3) that contains the major surface antigen Wsp, a widely utilized marker of infection (Fallon et al., 2013a). Band C is absent in uninfected C7-10 cell extracts (Fig. 3, lanes 2 - 4) but is present in C/wStr1 extracts (lanes 5 – 8). Band C staining intensity is lowest in lane 5, containing total protein from infected cells, higher in lanes 6 and 7, containing cytoplasmic and pooled GF-30/40 and GF-40/50 protein, and highest in lane 8, containing C/wStr1 GF-50/60 protein. Additional prominent bands unique to infected cells migrate at approximately 9, 16, 56, and 70 kDa (lane 8, bands A, B, D, and E). Bands enriched in uninfected extracts (marked by > symbols, compare lanes 2-4 with lanes 5 – 8) represent a potential host cell response to wStr infection, and will not be further described here.

Figure 3.

SDS PAGE of Wolbachia proteins enriched by cell fractionation and sucrose step gradient ultracentrifugation. Lane 1: Markers, range 6 – 200 kDa, indicated at left. Lanes 2 and 3: Uninfected C7-10 cell total proteins and cytoplasmic protein, respectively. Lane 4: Pooled C7-10 gradient fractions GF-30/40, -40/50 and -50/60. Lanes 5, 6 and 7: Wolbachia infected C/wStr1 (from cells at passage16) total cell, cytoplasmic, and pooled GF-30/40 and -40/50 protein extracts, respectively. Lane 8: Extract of C/wStr1 GF-50/60 with Coomassie blue-stained bands indicated by A – E at right. The relative order of lanes in the original gel image has been shifted to facilitate comparisons. Coomassie blue-stained bands in C7-10 lanes 2 - 4 that appear depleted in C/wStr1 lanes are indicated by > symbols at left of lane 2.

Proteomic definition of the wStr footprint

MS/MS analysis of bands A-E excised from four replicate gels identified 53 proteins with a confidence level of ≥ 99%, based on a match to at least three unique peptides with ≥ 95% confidence. Of those, 41 (77%) are from wStr (Wolbachia) versus 12 from A. albopictus host cells (Table S1). The most abundant proteins, representing ≥ 0.05% of total peptide spectra, include 14 Wolbachia proteins and a single A. albopictus host protein (Table 1). Among them are GroES (Hsp10) homologs from both wStr and host cells (band A), peptidoglycan-associated lipoprotein (PAL or OmpA; band B), surface antigen Wsp (band C), chaperones GroEL and DnaK (Hsp60 and Hsp70) and a 73 kDa outer membrane protein (bands D and E). The highest peptide counts represented GroEL and DnaK. The stress response proteins Hsp20 and the ATPase subunit of protease HslVU are the second most abundant proteins in bands B and D, respectively. A TypA membrane GTPase stress response protein (BipA) is the fourth most abundant protein in band E. The reproducible profile represented by bands A – E (Table 1) constitutes an SDS PAGE protein footprint of wStr dominated by chaperone, stress response, and membrane surface proteins. Analysis of gels loaded with ³⁵S[methionine/cysteine]-labeled extracts confirmed the wStr protein footprint and revealed an additional radiolabeled band in the Wolbachia-enriched GF-50/60 lane that migrated at approximately 66 kDa (Fig. 4, lane 2, asterisk). It may contain a 66 kDa flavoprotein subunit of succinate dehydrogenase that has a Met/Cys codon frequency of 6% versus an average of 3.6% among all wBm proteins (http://www.kazusa.or.jp/codon/).

Table 1.

Most abundant proteins in the wStr SDS PAGE protein footprint.

Proteins1	Band	kDa	# Pep.	%Cov.	%Spectra
GroESWolbachia	A	10	8	71	1.34
GroES Aedes	A	11	7	66	1.15
PAL2	B	18	13	40	0.25
Hsp20	B	17	6	26	0.07
Surface antigen Wsp	C	24	8	15	0.70
ribosomal protein L4	C	23	3	18	0.20
ribosomal protein S4	C	24	9	37	0.06
chaperonin GroEL	D	59	38	53	0.59
HslVU ATPase subunit	D	54	12	27	0.13
MBL3 family protein	D	60	8	17	0.08
ribosomal protein S1	D	62	12	22	0.08
ATP synth. F1 β subunit	D	52	9	22	0.07
Chaperonin dnak	D	69	9	17	0.06
outer membrane protein	D	73	7	12	0.06
chaperonin dnak	E	69	26	36	0.37
outer membrane protein	E	73	16	20	0.31
chaperonin GroEL	E	59	10	20	0.09
TypA membrane GTPase	E	68	9	17	0.08

¹Only proteins represented by ≥ 0.05% of total peptide spectra are shown here with gel band of origin, mass in kilodaltons, maximum number detected peptides among four replicate gel slices, % protein sequence coverage and % of total spectra. See Table S1 for complete list of 53 proteins and genbank accession numbers of most similar homologs among Wolbachia strains in the MS data set searchbase.

²peptidoglycan-associated lipoprotein.

³metallo-beta-lactamase.

Figure 4.

Autoradiographic analysis of [³⁵S]methionine/cysteine-labeled proteins separated by SDS PAGE confirms enrichment of Wolbachia proteins. Equal radioactivity from uninfected C7-10 GF-50/60 and infected C/wStr (p20) GF-50/60 protein extracts was electrophoresed in lanes 1 and 2, respectively. Bands corresponding to Coomassie-stained bands A – E of Fig. 3 are indicated at right. Asterisk indicates additional prominent band associated with wStr that migrated at approximately 66 kDa. Relative migration positions of unlabeled mass markers, range 6 – 200 kDa, are indicated at left.

Derivation of a composite wStr proteome from four MS data sets

The wStr genome has not been sequenced, but comparisons among available Wolbachia genomes indicate a “core genome” of approximately 650-700 orthologs that is well conserved between even distantly related strains (Foster et al., 2005; Dunning-Hotopp, 2006; Klasson et al., 2009; Ishmael et al., 2009). In each genome only a small proportion of genes are unique as in the example of wBol1b with only 44 genes that have no homologs in other Wolbachia strains (Duplouy et al., 2013). In silico analysis of genomes from two cytoplasmically incompatible Wolbachia strains associated with Culex pipiens mosquitoes identified only 29 open reading frames with <75% sequence identity (J. Beckmann, unpublished). Because most wStr proteins would thus be expected to have homologs in Wolbachia strains with annotated genomes, we used an MS/MS searchbase of WOL-A, -B and -D genomes to identify proteins from wStr.

We began by establishing a profile of proteins for which we could identify at least two unique, 95% confidence peptides derived from a single homolog within at least one of four MS data sets (Fig. 2). For final inclusion in the wStr proteome, we reduced protein prediction confidence to 50% and manually examined the Scaffold output files using NCBI BLASTp and Protein Cluster tools to resolve annotation ambiguities and reveal un-annotated homologs. Proteins with a mass > 14 kDa were accepted based on at least two 95% confidence unique peptides from a single homolog or from different regions of two or more homologs within at least one MS data set. Because smaller proteins contain fewer trypsin cleavage sites, we accepted those with a mass ≤ 14 kDa that matched a single peptide.

The proteome derived from MS data sets D - G included 790 proteins with predicted molecular weights (MW) that ranged from < 10 to > 300 kDa. The exhaustive gel-based data set D (Fig. 2) provided the greatest contribution to the proteome, particularly for ≤ 14 kDa proteins identified by a single unique peptide. We noted an 18% increase in protein identification from peptides separated by 2D- versus 1D-HPLC following in-solution digestion of Wolbachia-enriched GF-50/60 extracts (data sets G versus F). Due to a non-normal distribution, we report the protein median MW (33 kDa) and measure of dispersion through Interquartile Range (19-50 kDa) for each MS data set (Table 2). The non-parametric Kruskal-Wallis test showed that median MW was not different between datasets (p-value = 0.21) but number of detected peptides per protein differed (p-value <0.0001), with higher numbers returned in datasets F and G. Nine percent (73) of the proteins were matched to a single peptide, 22% (177) to two peptides, and 68% (540) to three or more peptides. Based on matched peptides, 66% of wStr proteins most closely resembled proteins from wPip (WOL-B), while 23% were most similar to WOL-A strain proteins and 11% to those from the WOL-D strain, wBm (see Table S2 for number of unique peptide matches to all homologs and protein sequence coverage).

Table 2.

Comparison of aggregate wStr protein numbers to numbers recovered from individual MS Data sets D – G.

MS Dataset	N	Molecular Weight1		Peptides
		Median (IQR)	p-value2	Mean	p-value2
All	790	33 (19-50)	-	-	-
D	719	34 (20–51)	0.21	3 (2-5)	<0.0001
E	404	33 (21-47)		3 (2-5)
F	395	36 (20-51)		4 (2-6)
G	456	36 (22-52)		4 (3-7)

¹Molecular weight in kDa with median and interquartile range (IQR) and mean number peptides (SD) returned.

²Kruskal-Wallis Test.

Over 20% of the wStr proteome consists of hypothetical proteins, of which we assigned 27 to one of 20 functional classes, where conserved domains and BLASTp analyses gave firm indication of cellular function (annotated in Table S2). The distribution of proteins by functional class among the four MS data sets (Fig. 5A) was statistically significant only if the largest class, Function unknown was included in the comparison (χ² = 89.6, p-value = 0.0037 versus χ² = 42.2, p-value = 0.8776 after exclusion of Function unknown class; Pearson's Chi-Squared test).

Figure 5.

Distribution of identified wStr proteins by protein functional class and MS data set, and by peptide count and mass. (A) Relationship between proteins (n=1974) and functional class in data sets D, E, F and G. (B) Relationship between protein mass and MS-detected peptide count expressed as mean of Univariable model studentized residuals, a measure of the difference between expected and observed log peptide count adjusted by estimated standard error. An SR of 0 corresponds to average abundance.

Estimation of protein abundance by peptide spectral counting

Because the complete absence of Wolbachia proteins from uninfected cells precludes absolute quantitation of proteins using labels such as iTRAQ, we based estimates of relative abundance levels (RALs) on counts of unique peptides matched to each protein (mean of all matched peptides Table S2, column H). This method is a well-established alternative (reviewed in Lundgren et al., 2010) applied in earlier analyses of wBm and wOo from nematodes (Bennuru et al., 2011; Darby et al., 2012).

We used Spearman correlation coefficients to test associations between peptide count, predicted protein mass (MW) and MS data set. Correlation between peptide count and MW was 0.43057 (p<0.0001), between peptides and data set was 0.13096 (p<0.0001), and between MW and data set was 0.03861 (p=0.086). These results indicated an association between peptide count and MW, and between peptide count and MS data set, which confirmed our previous nonparametric test of those associations (Table 2). We further tested strength of association between peptide count and MW through linear regression models, after log transformations to meet linearity assumptions. In a Univariable model, we found that log MW was a relatively weak (r² = 0.2263) but statistically significant (p<0.0001) predictor of peptide count: log(peptides) = −0.41236 + 0.49670 * log(mw). We then used a forward selection procedure to determine the most parsimonious Multivariable model including run (MS dataset D, E, F, or G) and protein functional class as categorical variables and found that they increased the coefficient of correlation (r²=0.3458), but did not significantly alter the association (Beta: 0.47470, p-value: <0.0001) of log MW with peptide count. Complete results of the Univariable and Multivariable models are presented in Table S3 including modeled expected values of peptide counts for all proteins in all runs (n=1974) and studentized residuals (SR), a measure of deviance from expected values adjusted for estimated standard deviation from the mean. Association between protein MW and peptide count (as mean SR from all MS data sets) in the Univariable model is shown in Fig. 5B.

The most abundant wStr proteins

Based on the Univariable model (see above), protein mass explains only 23% of variation in number of detected peptides for the 790 proteins in the wStr proteome. We defined abundant proteins as those present in at least three MS data sets with a mean SR > 1 (average abundance = 0) and high abundance proteins as those with a mean SR > 2 (Table S2, column I). Of the 67 proteins with SR values > 1 (Table 3), 24 are among the 46 Wolbachia proteins enriched in wStr GF-50/60 that constitute our SDS PAGE protein footprint (Fig. 3 and Table S1). Nine members of the top67 are high abundance proteins, including 5 chaperonins (mean SR 2.66). The most abundantly represented protein functional classes in the top64 are Ribosome/Translation, Protein modification/Chaperones, Transcription and Energy production with mean SRs of 1.52, 1.83, 1.65, and 1.31, respectively.

Table 3.

The 67 most abundant proteins in wStr.

Protein	Entry	kDa	SR	RAL	Functional class	Ft	wOo	wBm	wMel
Cysteine desulfurase	1	46	1.07	9.5	Amino acid metab.
Serine/glycine hydroxymethyltransferase	2	47	1.04	8.8	Amino acid metab.		X
Putative transaldolase	3	23	1.11	6.5	Carbohydrate metab.
Peptidoglycan-associated lipoprotein	4	18	2.28	11.5	Cell envelope/membr.	X	X	X
Putative outer membrane protein	5	73	1.94	18.8	Cell envelope/membr.	X	X	X	X
Surface antigen (aka YaeT)	6	89	1.61	17.0	Cell envelope/membr.
Wsp major surface antigen	7	24	2.20	13.0	Cell envelope/membr	X	X	X	X
Antioxidant Ahp/Tsa family protein	8	22	1.97	10.5	Cellular defense mech.
Cell division protein FtsZ	9	42	1.11	8.5	DNA replic./Cell division
Exinuclease ABC, A subunit	10	104	1.30	15.5	DNA replic./Cell division
RecA protein	11	39	1.06	8.0	DNA replic./Cell division
ATP synthase F0F1, alpha subunit	12	56	1.16	10.8	Energy metab.	X	X	X
ATP synthase F0F1, beta subunit	13	52	1.81	14.8	Energy metab.	X			X
α-KGDH, E2 component	14	43	1.55	11.2	Energy metab.		X
Isocitrate dehydrogenase, NAD-depend.	15	53	1.10	7.2	Energy metab.
Malate dehydrogenase	16	34	1.21	8.2	Energy metab.				X
Pyruvate/α-KGDH dehydrogenase E3 comp.	17	49	1.22	10.5	Energy metab.		X
Succinyl-CoA synthetase α subunit	18	30	1.01	6.8	Energy metab.
Succinyl-CoA synthetase β subunit	19	42	1.40	10.2	Energy metab.		X
Conserved hypothetical protein	20	29	1.92	9.2	Function unknown
Hypothetical protein WP0198	21	21	1.18	6.5	Function unknown
Hypothetical protein WP0641	22	14	1.11	5.8	Function unknown
Class II aldolase/adducin domain protein	23	27	1.09	6.8	General function
3-oxoacyl-(acyl-carrier-protein) reductase	24	16	1.06	6.8	Lipid metab.
Adenylate kinase	25	24	1.02	6.2	Nucleotide metab.
Nucleoside diphosphate kinase	26	16	1.35	6.2	Nucleotide metab.
Putative phage-related protein	27	9	1.50	5.0	Phage/Viral proteins
ATP-binding subunit ClpB	28	95	1.41	17.0	Protein modify/Chaperones	X
ATP-binding subunit ClpX	29	47	1.30	10.2	Protein modify/Chaperones
Chaperonin dnaJ	30	41	1.04	8.0	Protein modify/Chaperones
Chaperonin dnaK	31	69	2.82	31.2	Protein modify/Chaperones	X	X	X	X
Chaperonin GroEL	32	59	3.69	52.5	Protein modify/Chaperones	X	X	X	X
Chaperonin GroES	33	10	2.33	9.0	Protein modify/Chaperones	X	X		X
Heat shock protein HslVu-ATP-bind. Sub.	34	54	1.36	11.5	Protein modify/Chaperones	X
Hsp20 heat shock protein	35	17	2.27	11.8	Protein modify/Chaperones	X
HtpG heat shock protein	36	73	2.19	22.0	Protein modify/Chaperones	X			X
hflC protein	37	33	1.89	12.2	Protein modify/Chaperones		X	X
hflK protein	38	39	1.11	8.2	Protein modify/Chaperones
Peptidyl-prolyl cis-trans isomerase D	39	70	1.28	12.2	Protein modify/Chaperones	X	X	X
Protease DO	40	53	1.56	10.2	Protein modify/Chaperones
Trigger factor	41	52	1.43	12.8	Protein modify/Chaperones
Arginyl-tRNA synthetase	42	64	1.46	13.2	Ribosome/Translation	X
Polyribonucleotide nucleotidyltransferase	43	85	1.95	21.2	Ribosome/Translation	X			X
Ribosomal protein L3	44	26	1.11	7.2	Ribosome/Translation
Ribosomal protein L7/L12	45	14	1.04	5.2	Ribosome/Translation				X
Ribosomal protein L14	46	13	1.19	5.0	Ribosome/Translation
Ribsomal protein L25/Stress protein Ctc	47	22	1.25	7.0	Ribosome/Translation
Ribosomal protein S1	48	62	1.58	14.2	Ribosome/Translation	X
Ribosomal protein S2	49	30	1.24	8.2	Ribosome/Translation				X
Ribosomal protein S4	50	24	1.68	9.2	Ribosome/Translation	X
Ribosomal protein S9	51	16	1.02	5.0	Ribosome/Translation	X
Ribosome recycling factor	52	21	1.55	8.5	Ribosome/Translation
Translation initiation factor IF2	53	86	1.54	16.2	Ribosome/Translation
Translation elongation factor G	54	76	1.90	19.2	Ribosome/Translation	X			X
Translation elongation factor Ts	55	31	1.46	10.0	Ribosome/Translation
Translation elongation factor Tu	56	43	2.80	25.8	Ribosome/Translation		X	X
Isoprenoid biosynthesis protein	57	26	1.31	7.8	Secondary metabolism
GTP-binding protein TypA (aka BipA)	58	67	1.45	13.2	Signal transduction	X
CtrA two-component SHK regulator	59	29	1.76	10.8	Signal transduction
DNA-directed RNA poly. β/β′ subunits	60	319	3.03	58.2	Transcription/RNA modify				X
DNA-directed RNA poly. omega subunit	61	15	1.07	5.0	Transcription/RNA modify	X
Metallo-beta-lactamase family protein	62	60	1.95	17.5	Transcription/RNA modify	X
Ribonuclease Rne/Rng	63	66	1.08	11.8	Transcription/RNA modify
Transcription elongation factor GreA	64	18	1.48	7.0	Transcription/RNA modify
Transcription elongation factor NusA	65	57	1.08	9.8	Transcription/RNA modify	X
Transcription termination factor Rho	66	52	1.88	16.5	Transcription/RNA modify	X
Iron compound ABC transporter periplasmic	67	38	1.84	12.8	Transporter

Only proteins with a mean SR > 1 (column K of Table S3) are included. See Table S2 for Accession #s and RAL (mean all peptides in MS data sets D –G.). Ft = present in wStr protein footprint (Fig. 3 and Table S1). wOo, wBm, wMel = presence of homolog in high abundance proteins (Darby et al. 2012, 2013; Tables S11 and S3, respectively.

Protein functional classes and partitioning of metabolic resources

In the Multivariable analysis using functional class as a variable and the largest class, Function Unknown as the referent (Table 4), beta coefficients (range: 0.16859 – 0.90069) correlated positively with peptide counts and class RAL scores (range: 1 – 7.7), which for individual proteins range from 2 – 94 and 0.2 – 58.2, respectively (Table S2). The Protein modification/Chaperones, Transcription/RNA modification and Ribosome/Translation functional classes have the highest beta coefficients, indicating high level expression as functional classes, while that of the DNA Replication/Cell Division class is among the lowest. Only three top67 proteins (Table 3, entries 9 – 11) participate in DNA metabolism: cell division protein FtsZ, the A subunit of excinuclease ABC, and RecA. The Nucleotide metabolism class, which supports both DNA replication and transcription, has above average beta coefficient and RAL values (Table 4), and includes two abundant proteins, adenylate kinase (Table 3, entry 25), involved in AMP metabolic signalling and cellular energy homeostasis, and nucleoside diphosphate kinase (entry 26), which maintains equilibrium between NTPs. Proteins involved in RNA transcription are prominent members of the top67 proteins (entries 60 – 66), including elongation and termination factors for mRNA transcription, the RNA polymerase omega subunit and the β/β′-subunit, which is second only to GroEL in abundance. Fifteen proteins with direct or indirect roles in translation are among the top67 (entries 43 – 56), including polyribonucleotide nucleotidyltransferase, which is involved in quality control of rRNA precursors, mRNA modification and RNA degradation (Mohanty and Kushner, 2000) as a degradosome component with enolase and the abundant ribonuclease E (entry 63). In aggregate, the results suggest that wStr in mosquito cells devotes a greater proportion of metabolic resources to transcription, translation and protein modification/stabilization than to DNA replication/Cell division.

Table 4.

Protein functional classes in the wStr proteome.

Functional Group		% Total	RAL1	Beta2
Protein modification/Chaperones	39	4.9	7.7	0.90069
Transcription/RNA modification	29	3.7	6.0	0.62210
Ribosome/Translation	103	13.0	4.3	0.52874
Signal transduction	7	0.9	4.8	0.50104
Cellular defense	11	1.4	3.2	0.49942
Energy metabolism	42	5.3	4.6	0.47887
Nucleotide metabolism	37	4.7	3.6	0.39646
Amino acid metabolism	21	2.7	4.5	0.38148
Motility/Secretion	30	3.8	4.0	0.33995
Secondary metabolism	9	1.1	3.0	0.33957
General function	36	4.6	2.8	0.33708
Inorganic ion metabolism	3	0.4	2.6	0.30392
Carbohydrate metabolism	15	1.9	3.2	0.30292
Cell envelope/biogenesis	45	5.7	3.1	0.28183
Lipid metabolism	18	2.3	2.2	0.27313
Coenzyme metabolism	29	3.5	2.3	0.20685
DNA replication/Cell division	82	10.4	2.5	0.19242
Transporter	21	2.7	2.0	0.19198
Phage/Virus-related	39	4.9	1.5	0.16859
Function unknown	175	22.2	1.0	0 (referent)3

¹Mean relative abundance level of all proteins in functional class; mean of all classes is 3.4.

²Beta coefficient of functional class in Multivariable model; mean of all classes is 0.362.

³Function unknown was modeled as the referent category in Multivariable linear regression model.

Chaperonin and stress response proteins

Eight of 14 abundant and highly abundant proteins of the Protein modification/chaperone class (Table 3, entries 26 – 41) are consistent with identities of prominent gel bands in the wStr protein footprint (Fig. 3 and Tables 1 and S1). High abundance chaperonins that are involved in stress responses and prevention of protein denaturation such as GroEL, GroES, DnaK, Hsp20 and HtpG are among the abundant proteins of wBm, wOo, and wMelPop (Table 3). Other abundant stress-related proteins in wStr include protease Do (entry 40) involved in periplasmic protein degradation and heat stress responses, and hflC and hflK (entries 37, 38) with roles in suppressing accumulation of abnormal membrane proteins. Trigger factor (entry 41) is a ribosome-associated chaperone that maintains open conformation of nascent polypeptides and has peptidyl-prolyl cis-trans isomerase activity, while peptidyl-prolyl cis-trans isomerase (entry 39), is a chaperone whose activity at proline residues can be rate limiting in protein folding.

In aggregate, these results suggest that Wolbachia survival in diverse hosts may be associated with significant investment of resources in prevention of protein denaturation and aggregation accompanied by increased levels of other stress response proteins involved in signal transduction, protein translation and oxidative defenses. Examples include an Ahp/TCA family antioxidant protein (Table 3, entry 8) and an isoprenoid biosynthesis protein (entry 57) that is elevated 1.6-fold in wMelPop stressed by exposure to doxycycIine (Darby et al., 2013, Table S6). Isoprenoids are associated with resistance to oxidative stress and are essential for survival of intracellular bacteria (reviewed in Heuston et al., 2012).

Signal transduction

We identified seven wStr proteins involved in signal transduction, a functional class with relatively high RAL and beta coefficient values (Table 4). Bacteria detect and respond to environmental change through sensor histidine kinase (SHK) two-component signal transduction systems that include CtrA, which is abundant in wStr (Table 3, entry 59), and a potential SHK partner (Table S2, entry 740). An abundant conserved hypothetical protein (Table 3, entry 20) has sequence identity to an SHK associated with stress response signalling in a Botryotinia sp. CtrA is associated with stress-dependent differential gene expression in a close relative of Wolbachia (see Discussion), and with differential expression of genes in response to nutrient fluctuations in Caulobacter crescentus (England et al., 2010).

The abundant TypA membrane stress GTPase (aka BipA; Table 3, entry 58) is a ribosome-binding GTPase (TRAFAC class) and functions as a highly conserved signal transduction protein in bacteria and plants. In Salmonella enterica, normal association of BipA with the 70S ribosome in a GTP-bound state is altered under stress conditions (deLivron and Robinson, 2008). Co-ordination of translation in Wolbachia with a potential stress response is further suggested by the abundant ribosomal protein L25 (CTC general stress protein; entry 47), required for accurate translation under stress conditions but not for ribosome assembly, and arginyl-tRNA synthetase (entry 42). It is elevated 1.6-fold in wMelPop exposed to doxycycIine (Darby et al., 2013) and an arginyl/seryl-tRNA synthetase complex enhances aminoacylation during the early phase of bacterial translation, particularly in response to stress (Godinic-Mikulic et al., 2011).

Energy and amino acid metabolism

Energy production is of particular interest because genomic evidence indicates that amino acid metabolism predominates over carbohydrate metabolism in Wolbachia. Absence of glycolytic and gluconeogenic enzymes, reduced pathways for amino acid biosynthesis and presence of transporters for proline, aspartate/glutamate and alanine suggest reliance on host amino acids as a major energy source (Wu et al., 2004; Foster et al., 2005; Dunning-Hotopp et al., 2006).

Proteins with roles in energy and amino acid metabolism constitute 8.0% of the wStr proteome and as functional classes they have RALs and beta coefficients that are higher than the proteome mean values, while those of the carbohydrate and lipid metabolism classes are lower (Table 4).

Active metabolism is suggested by the abundant F0F1 ATP synthase α- and β-subunits (Table 3, entries 12, 13) and the well-represented NADH dehydrogenase respiratory chain components (Table S2, entries 223-228). The wStr top67 include abundant proteins of TCA cycle energy-producing steps (Fig. 6) such as the E2 component of the α-ketoglutarate dehydrogenase (α-KGDH) complex (Table 3, entry 14) and its lipoamide dehydrogenase E3 component (entry 17), which is shared with the the pyruvate dehydrogenase complex in most Gram-negative bacteria. The α-KGDH complex converts α-ketoglutarate to succinyl-CoA and in conjunction with above average abundance proline utilization protein A (SR 0.5) and glutamate dehydrogenase (SR 0.5), which convert proline to glutamate to α-ketoglutarate, is the critical link between amino acid metabolism and the TCA cycle (Fig. 6). Both succinyl-CoA synthetase subunits are abundant (Table 3, entries 18, 19), while the succinate dehydrogenase subunits, including the candidate component of the 66 kDa band from Wolbachia-enriched GF-50/60 (Fig. 4) are of above average abundance (Suppl. Table S2, entries 239, 240; SR 0.5, 0.8). Other abundant TCA cycle enzymes include the isocitrate and malate dehydrogenases (entries 15, 16). The importance of amino acids as an energy source for Wolbachia is supported by the low abundance of the E2 component of the pyruvate dehydrogenase complex, which links glycolysis to the TCA cycle, relative to that of the α-KGDH complex (SR -0.2 versus 1.6, respectively).

Figure 6.

Schematic model of central metabolism in Wolbachia Gray shading with black text represents proteins of the TCA cycle, carbohydrate metabolism and inorganic ion transporters with SR values in wStr as indicated. White text on black background represents amino acids and proteins with roles in amino acid metabolism. Black text on white background represents biosynthetic pathways, metabolism of iron and sulfur, and glutathione (GSH). Stars indicate ROS generation and/or neutralization. DH indicates dehydrogenase. HMT indicates hydroxmethyltransferase. SOD indicates superoxide dismutase.

Interconversion of serine and glycine

Wolbachia retain enzymes for interconversion of glycine with serine (Dunning Hotopp et al., 2006), suggesting that these amino acids are important in Wolbachia metabolism, and/or limited in host amino acid pools. In both wStr and wOo, serine/glycine hydroxymethyltransferase (SG-HMT; Table 3, entry 2) is an abundant enzyme that catalyzes reversible conversion of serine and tetrahydrofolate to glycine and 5,10-methylenetetrahydrofolate (Fig. 6). In many cell types, SG-HMT is the primary source of folate pathway single carbon units that are consumed in biosynthesis of metabolites such as purines and thymidylate. Folate pathway proteins in the wStr proteome (Table S2, entries 100, 104, 106, 107) include folylpolyglutamate synthase. Glutamate and glycine are precursors in the folate and glutathione synthesis pathways and glycine is a precursor of riboflavin and porphyrin/heme as well. Wolbachia can also derive glycine from activity of two threonine aldolases that share only 31% sequence identity (Suppl. Table S2, entries 16, 17).

Coversion of cysteine to alanine

The abundant cysteine desulfurase (Table 3, entry 1) converts cysteine to alanine by removal of the sulfur atom as a persulfide intermediate that is transferred to the sulfur relay pathway (Fig. 6). An anabolic demand for sulfur is exerted by Wolbachia proteins in the Coenzyme/Heme metabolism class with roles in biosynthesis of biotin, lipoic acid, thiamine, and iron-sulfur clusters that are dependent on cysteine desulfurase for sulfur (Hidese et al., 2011). Such cofactors serve as critical redox-sensitive components of proteins involved in a wide range of activities, including the TCA cycle and oxidative respiration. Free cysteine causes oxidative damage in bacterial cells (Park and Imlay, 2003) and is regulated by additional mechanisms including its condensation with glutamate and glycine to form the antioxidant, glutathione, catalyzed by below average abundance glutamate cysteine ligase and glutathione synthetase (Table S2, entries 13 and 86).

Amino acid, phosphate and iron transporters

Sodium symporters for alanine and glutamate and an ABC transporter involved in amino acid import (Fig. 6) are low-abundance proteins in wStr (Table S2, entries 790, 791, 772). A phosphate ABC transporter permease and its periplasmic phosphate- and ATP-binding partners are of below and above average abundance, respectively (entries 785-787). An abundant periplasmic iron-compound binding protein is the critical component of an ABC transporter for uptake of iron (Table 3, entry 67).

Production of reducing equivalents and pyruvate phosphate dikinase

The TCA cycle produces NADH, and conversion of succinyl CoA to succinate and then to fumarate is accompanied by formation of FADH₂ and GTP. Wolbachia lack an NADP-dependent isocitrate dehydrogenase and the oxidative branch of the pentose phosphate pathway, which produce NADPH that can support glutathione-dependent oxidative defenses and activities of lipid and nucleotide biosynthetic pathways. However, they can produce NADPH through conversion of malate to pyruvate by malic enzyme and aspartate to L-4-aspartyl phosphate by aspartate semialdehyde transferase (Fig. 6). The enzymes are of above average abundance (Table S2, entries 7, 229).

Pyruvate kinase converts phosphoenolpyruvate and ADP to pyruvate and ATP in the final step of glycolysis in host cells. In Wolbachia, an alternative enzyme of below average abundance in wStr, pyruvate phosphate dikinase (PDK; Table S2, entry 35), catalyzes conversion of ATP and pyruvate to phosphoenolpyruvate, AMP and pyrophosphate. PDK is a Wolbachia-specific target for design of anti-filarial drug therapies (Palayam et al., 2012), and if coupled to activity of the abundant adenylate kinase (Table 3, entry 25), it might function in energy conservation as well (Raverdy et al., 2008). Activity of PDK and its role in central metabolism are likely influenced by two below average abundance proteins: a pyrophosphatase and a conserved hypothetical protein (Table S2, entries 459 and 23) with a serine kinase domain and 58% BLASTp identity to a PDK regulatory protein (gi:85687555) from an Ehrlichia sp.

Secretion systems - Type IV

In bacteria, the Type IV secretion system (T4SS) is a major mediator of intercellular transfer of macromolecules (reviewed in Alvarez-Martinez and Christie, 2009). We recovered high confidence peptides from all proteins encoded by the virB8-D4 and virB3-B6 operons, with the exception of the 11 kDa VirB3 matched to a single 90% confidence peptide, as well as the “orphan” gene-encoded VirB4-2, VirB8-2 and VirB9-2, and putative VirB2 and VirB7 proteins (Table 5). Expression of T4SS proteins is stringently co-regulated and a system-wide transcription factor that is conserved in the Anaplasmataceae (Cheng et al., 2008; Gillespie et al., 2009, 2010) has been characterized as wBmxR1 in wBm (Li and Carlow, 2012). In wStr, the transcription factor, the VirB8, B9, B9-2, B10, and B11 membrane channel components and the VirD4 substrate docking protein are of above average abundance (mean SR 0.69), while the other proteins, including the two VirB4 and four VirB6 paralogs, are below average. Collectively, the 16 proteins have a mean RAL of 4.1 and a mean SR of -0.27. As a benchmark, we note that RALs of a nearly complete set of 50 ribosomal proteins and associated housekeeping proteins likely to be expressed at above average abundance range have a mean RAL of 3.4 and a mean SR 0.285.

Table 5. The wStr Type IV secretion system proteins.

Protein1	kDa	Pep.	Cov.	RAL2	SR3
Transcription factor	11	4	33	2.8	0.30
VirB2 (putative)	12	1	17	0.2	-1.32
VirB7 (putative)	9	1	15	0.2	-1.09
VirB4	91	8	13	6.0	-0.12
VirB6-1	95	7	11	4.0	-0.87
VirB6-2	89	2	3	0.5	-1.81
VirB6-3	109	8	10	4.2	-0.56
VirB6-4	123	5	15	3.8	-1.14
VirB8	26	7	44	5.0	0.61
VirB9	31	8	35	6.0	0.69
VirB10	54	10	26	8.2	0.77
VirB11	37	10	29	6.8	0.75
VirD4	77	10	18	5.8	0.35
VirB4-2 orphan	90	5	7	4.0	-0.80
VirB8-2 orphan	25	2	10	1.5	-0.79
VirB9-2 orphan	30	110	39	6.0	0.65

¹Predicted wStr T4SS proteins with maximum number MS-matched unique peptides in a single MS data set and % protein sequence coverage.

²RAL - mean number peptides in all data sets.

³Mean studentized residuals based on the Univariable model (Table S3).

Type II

Secretion of proteins to the bacterial periplasm and outer membrane is meditated by a multimeric Sec preprotein translocase energized by the SecA ATPase and the SecB chaperone that directs proteins to the translocase (Table S2, entries 479, 480). Both are of above average abundance, while the Sec D, E, G, Y and interacting proteins YidC and YajC of the translocase (Nouwen and Driessen, 2002) are below average with the exception of YajC (Table S2, entries 481-486). SecF, represented by single peptides in separate MS data sets, did not meet criteria for inclusion in the proteome. Nascent membrane proteins are also directed to the translocase by signal recognition particle GTPase and its FtsY docking partner (below average abundances), while the above average signal peptidase I cleaves preproteins after membrane transit (Table S2, entries 487-489). The abundant YaeT surface antigen (Table 3, entry 6) facilitates protein folding and membrane insertion (Werner and Misra, 2005).

Type I

The wStr proteome contains above average abundance HlyD membrane fusion and PrtD-like ATPase proteins and a low abundance putative membrane-spanning periplasmic partner of a trimeric Type I secretion system (Table S2, entries 478, 491, 490, respectively). Probable substrates include proteases and lipases involved in host cell interactions or nutrient acquisition, which are translocated without an amino terminal cleavage (Kanonenberg et al., 2013).

Discussion

Transcriptomic and proteomic-based analyses have been used to explore host-microbe interactions. We focused directly on a proteomic analysis because microbial mRNAs, including those of slow growing Rickettsiales bacteria (Winkler, 1987) have short half-lives (Rauhut and Lug, 1999; Bernstein et al., 2002), and in studies with Mycoplasma pneumoniae (Maier et al., 2011) and in Escherichia coli cells under varying growth conditions (Taniguchi et al., 2010), protein abundances and cognate mRNA levels are poorly correlated. The reduced genomes of Wolbachia encode a limited repertoire of genes involved in transcriptional control (Wu et al., 2004; Foster et al., 2005), and in the context of our long term goal of manipulating Wolbachia in vitro, proteins are more likely to mediate interactions between Wolbachia and its host cell that may be influenced by selective conditions based on modification of culture media. Relative to earlier studies with the filarial-associated Wolbachia strains wBm and wOo (Bennuru et al., 2011; Darby et al; 2012), the present analysis of a robust and persistent wStr infection in a mosquito cell line demonstrates greater depth of peptide coverage, with a total of 790 identified proteins. Analyses of relative protein abundance based on peptide spectral counting (reviewed in Lundgren et al., 2010) revealed that wStr devotes more metabolic resources to protein expression and stabilization, relative to DNA replication and cell division (Table 4), as manifested by the dominant proteins in the SDS PAGE protein footprint (Figs. 3 and 4; Tables 1 and S1) and the top67 abundant proteins (Table 3).

Potential stress response

Mechanisms by which the host cell tolerates Wolbachia and Wolbachia in turn maintains a persistent infection are largely unknown. High abundance wStr proteins include a wide array of stress response proteins in addition to classic heat-shock chaperonins such as GroEL and DnaK that have established roles in protein folding and prevention of aggregation and are also abundant in wBm, wOo and wMelPop. Although we cannot exclude the possibility that Wolbachia enrichment induces expression of stress response proteins, we note that elevated levels of GroEL have been detected in other members of the Rickettsiales, including tick-borne pathogens of vertebrates (Renesto et al. 2005; Hajem et al., 2009). More distantly-related intracellular bacteria, such as Wigglesworthia from tsetse flies and Buchnera from aphids, constitutively express levels of GroEL that in free-living E. coli are approached only under rigorous heat-shock conditions (Aksoy 1995; Baumann et al., 1996). Overproduction of GroEL and a proteomic buffering role have been proposed to play a role in bacterial endosymbiosis, particularly in obligate intracellular bacteria that undergo strict vertical transmission, experience population bottlenecks, and accumulate detrimental genetic variation during genome reduction (Moran, 1996; Fares et al., 2004). In aggregate, our data suggest that the extent to which a potential stress response contributes to survival of Wolbachia in arthropod hosts merits further investigation. A similar stress response in the nematode-associated wOo (Darby et al., 2012) has been attributed to loss of DNA repair activities and accompanying accumulation of mutations. That scenario is less likely for wStr, which as an insect-associated strain would be expected to have greater access to mobile DNA (Newton and Bordenstein, 2001), and has genes encoding DNA recombination and repair proteins that do not occur in the more reduced genome of wOo.

Other abundant proteins in wStr have E. coli homologs with diverse stress response functions, such as hflC and hflK, protease DO, ribosomal protein L25 and Hsp20. The closest homologs of Wolbachia Hsp20 are found on rickettsial plasmids (Baldridge et al., 2010) and among bacteria associated with soil or other environments that are subject to variations in temperature and pH and/or osmotic and oxidative conditions. We note that mosquito host cell antioxidant proteins are upregulated upon Wolbachia infection (Brennan et al., 2008), that exposure of infected cells to the oxidizing agent paraquat reduces Wolbachia levels (Fallon et al., 2013b), and that wStr expresses an abundant Ahp/TCA family antioxidant protein and an isoprenoid biosynthesis protein (Table 3, entries 8 and 57; Fig. 6) that is associated with oxidative resistance in intracellular bacterial pathogens (Heuston et al., 2012).

Two abundant signal transduction proteins in wStr, CtrA and BipA, are associated with co-ordination of stress responses in other bacteria. In Wolbachia's close relative, Ehrlichia chaffeensis, CtrA regulates differential gene expression in replicative reticulate cells and stress resistant dense-cored cells that are more infectious (Cheng et al., 2011). Regulated genes include ompA (PAL) and surE, encoding a stationary phase survival protein, whose wStr homolog (Table S2, entry 422) is of above average abundance. In invasive pathogens such as Salmonella, Bordetella and Pseudomonas spp. and enteroinvasive E. coli, the BipA (aka TypA) membrane-associated GTPase plays critical roles in environmental sensing and co-ordinate regulation of gene expression, and is involved in capsular protein secretion and host cell adhesion, re-organization of the host cytoskeleton, and resistance to defensive peptides (Stockbauer et al., 2001; Grant et al., 2003; Neidig et al, 2013). In nitrogen-fixing Sinorhizobium meliloti, BipA is required for symbiotic association with plant hosts and stress adaptation (Kiss et al., 2004), while a chloroplast-specific BipA from a salt tolerant plant is believed to regulate oxidative stress tolerance (Wang et al., 2008).

Amino acids as an energy source

How Wolbachia coordinates its minimal capacity for biosynthesis and interconversion of amino acids (Dunning-Hotopp et al, 2006) with uptake from host cells to generate an amino acid pool sufficient for both energy generation and protein synthesis is of considerable interest. Wolbachia presumably competes for amino acid resources that are also needed for host survival and reproduction. For example, in mosquitoes and tsetse flies, proline is an energy source for flight, a nitrogen sink during metabolism of the blood meal, and depending on developmental state, is partitioned between somatic and reproductive tissues and the hemolymph, in which it is the most abundant amino acid (Goldstrohm et al., 2003; Scaraffia et al., 2010). In some insects, proline concentrations are up 100-fold higher than that of any other amino acid (Sowa et al., 1996). Although it is not clear how the extraordinarily high levels of free amino acids in insect hemolymph (Wyatt, 1964) translate into intracellular amino acid pools, it seems plausible that Wolbachia sequentially converts host-derived proline to glutamate and then to α-ketoglutarate for TCA cycle energy metabolism (see Results), and retains the proline biosynthetic pathway to supplement host proline under limiting conditions.

Reduced pathways for amino acid biosynthesis in Wolbachia suggest that transporters provide access to degraded host cell components and organic nutrients (Wu et al., 2004), while increased protein ubiquitination in newly infected host cells (Fallon and Witthuhn, 2009) is consistent with the possibility that Wolbachia derives amino acids from degraded host proteins. Abundant proteins in wStr that may participate in scavenging host peptides include periplasmic protease DO, bacterial proteasome components HslVU, ClpB and ClpX (Table 3) and an above average cytosol aminopeptidase (Table S2 entry, 598) also noted in wMelPop (Darby et al., 2013).

Outer membrane proteins

Bacterial OMPs are rapidly evolving and highly variable proteins that mediate host interactions through adhesion, invasion, molecular transport, and interactions with signal, defense and immunity pathways (Lin et al., 2002). A major class of OMPs is characterized by antiparallel sheets that form a transmembrane β-barrel porin domain embedded in the outer membrane, while variable loop domains are exposed to the extracellular surface (Koebnik et al., 2000). The most abundant β-barrel OMPs in wStr, wOo, wBm and wMelPop are Wsp and the 73 kDa major components of the wStr protein footprint (Fig. 3 and Tables 1 and 3). WspA is strongly expressed by Wolbachia in Drosophila eggs and is thought to play a role in establishment and maintenance of symbiosis (Braig et al. 1998). Analysis of the Brugia malayi filarial secretome indicates that WspA is excreted or secreted (Bennuru et al., 2009). It co-localizes with host cytoskeleton proteins and peptidoglycan-associated lipoprotein (PAL) in the Wolbachia-containing vacuole and interacts with glycolytic enzymes (Melnikow et al., 2013). PAL is a wStr protein footprint component and high abundance protein in wStr, wOo and wBm (Table 3, entry 4).

The Wolbachia T4SS

Phylogenetic analyses suggest that the unique surface properties of Rickettsiales bacteria may reflect co-evolutionary and/or functional interactions between the T4SS and OMPs such as WspB, which is associated with the virB8-D4 operon in Wolbachia (Gillespie et al., 2010) and may be a secreted effector (Melnikow et al., 2011). In Orientia tsutsugamushi (Rickettsiales), the T4SS genes have undergone massive expansion and are associated with genes encoding proteins with possible signaling and host cell interaction functions (Cho et al., 2007). The Wolbachia T4SS operons are highly conserved and transcriptionally active in insect host ovaries and may be involved in secretion of effectors that influence host reproduction (Wu et al., 2004; Felix et al., 2008; Rances et al., 2008, et al., 2009). Known Rickettsiales T4SS effectors include an Anaplasma ankyrin repeat-containing (Ank) protein that regulates transcription and suppresses host innate immunity, and Anaplasma and Ehrlichia proteins that target mitochondria, inhibit host cell apoptosis and reduce reactive oxygen species (Rikihisa and Lin, 2010; Liu et al., 2012). Ank proteins are established T4SS effectors of intracellular bacterial pathogens (Pan et al., 2008; Ge and Shao; 2011), and ank genes associated with a prophage in wPip are linked to polymorphisms in cytoplasmic incompatibility phenotypes in C. pipiens mosquitoes (Sinkins et al., 2005). The wStr proteome contains over 20 Ank proteins, including one of above average abundance (Table S2, entry 249), and a homolog of prophage protein Gp15 (entry 555), which contains a conserved motif found in toxin-translocating proteins secreted by bacteria and has been suggested to be a T4SS effector (Fujii et al., 2004). Gp15 is associated with variation in cytoplasmic incompatibility phenotypes in C. pipiens (Duron et al., 2006) and has homology to a probable sialidase, VrlC, encoded by a pathogenicity island in Dichelobacter nodosus, a pathogen of sheep (Billington et al., 1999). If Gp15 is secreted, it might influence host cell adhesion interactions or nutrient acquisition functions analogous to the Legionella pneumophila AnkB effector that assembles polyubiquinated host proteins on the Legionella-containing vacuole as an amino acid source for growth (Price et al., 2011). Our own work with C. pipiens mosquitoes reveals a possible secreted effector, WPIP0282 that is enriched in mature sperm (Beckmann and Fallon, 2013).

The Rickettsiales tandemly encoded VirB6 paralogs are unique among bacteria and may be involved in DNA transfer (Gillespie et al., 2009; 2010). Immunofluorescence assays indicate expression of VirB6 proteins by wAtab3 and wRi in insect oocytes and by wMelPop and wAlbB in mosquito cell lines (Rances et al., 2008) and we detected all four paralogs at below average abundance in wStr. In Agrobacterium tumefaciens, VirB6 domains participate in DNA transfer (Jakubowski et al., 2004) and have similarities to Helicobacter T4SS ComEC DNA channels (Christie and Cascales, 2005; Draskovic and Dubnau, 2005). VirB6 might compensate for absence of VirB5 in the Anaplasmataceae, which is involved in pilus formation and mating events (Bao et al., 2009). wStr expresses a putative pilus formation protein (Table S2, entry 492). A multifunctional T4SS with both secretory and DNA transfer capacities could facilitate access to the phage/bacterial mobile gene pool associated with horizontal transfer among strains of insect-associated Wolbachia (Ishmael et al., 2008; Kent et al., 2011).

Our investigation of the wStr proteome provides an important baseline for further study of Wolbachia interactions with its host cell in an in vitro context that will eventually facilitate genetic manipulation of Wolbachia. Cell lines provide a tractable approach for identifying interactions between Wolbachia and arthropod hosts that can advance practical applications of Wolbachia to pest control. It has been nearly 50 years since cytoplasmic incompatibility in C. pipiens was causally linked to Wolbachia and proof-of principle in its exploitation for pest control was established by population replacement of a filariasis vector in Burma (Laven, 1967a and b). Much remains to be done.

Materials and Methods

Cultivation of Cells and Microscopy

Aedes albopictus C7-10 (control) and C/wStr1 (Wolbachia-infected) cells were maintained in Eagle's minimal medium supplemented with 5% fetal bovine serum at 28°C in a 5% CO₂ atmosphere as described previously (Shih et al., 1998;Fallon et al., 2013a). Wolbachia infection was monitored by Giemsa-staining and with the Live/Dead Baclight™ Bacterial Viability Stain Kit (Life Technologies Corp., Grand Island, NY), using an Olympus IX70 microscope equipped with epifluorescent illumination and fluorescein isothiocyanate and Texas Red filter sets. Mitochondria were visualized by staining with Mitotracker Red^TM (Life Technologies Corp.). Images were collected with the SPOT photo-imaging system (Diagnostic Instruments Inc., Sterling Heights, MI).

Preparation of Subcellular Fractions and Enrichment of Wolbachia

C/wStr1 cells from six 25-cm² tissue culture flasks were resuspended by gentle pipetting and recovered by centrifugation at 800 × g, 10 min, room temperature. Cell pellets were washed in 30 ml serum free medium, centrifuged, resuspended in 6 ml ice-cold SPE buffer (250 mM sucrose, 3.8 mM KH₂PO₄, 7.1 mM K₂HPO₄, 10 mM EGTA, pH 7.2) and divided into 1 ml aliquots on ice. One aliquot was used to prepare a total cellular protein extract (see Fig. 2). To prepare subcellular fractions for reduction of sample complexity and maximized peptide detection, the remaining aliquots were lysed with a Kontes GE 70.1 ultrasonicator (Kontes Co., Vineland, NJ) in 1.5 ml tubes at the amplitude 40 setting for 5 seconds. The lysates were pooled and passed sequentially through 25 mm diameter glass microfiber syringe filters with 5.0 μm and 1.5 μm pores (Whatman, Florham Park, NJ) to remove nuclei and large particulates. The filtrate was centrifuged at 16,100 × g, 10 min, 4°C in 2 ml tubes and the supernatant was retained as the soluble cytoplasmic subcellular fraction. The pellets, containing bacteria, mitochondria, and other membranous particles, were re-suspended in 3 ml of ice-cold SPE buffer and 1 ml aliquots were loaded onto sucrose step gradients consisting of 2.5 ml each of 60, 50, 40 and 30% w/v sucrose in SPE buffer. The gradients were centrifuged in an SW-41 rotor at 210,000 × g for 90 min at 4°C. One ml fractions were collected as diffusely banded material within the upper portion of the 30% sucrose layer (GF-30) and as more distinct bands at the 30 – 40% (GF-30/40), 40 – 50% (GF-40/50), and 50 – 60% (GF-50/60, enriched in Wolbachia) sucrose boundaries. Bacteria were recovered from GF-50/60 by dilution to 15% sucrose, centrifugation and resuspension in SPE, and stored at -80°C until needed for protein precipitation.

SDS-polyacrylamide Gel Electrophoresis

Proteins were precipitated from cellular extracts and gradient fractions by addition of trichloroacetic acid (10% final concentration). Pellets were recovered by centrifugation, washed with 90% acetone, re-suspended in 2X SDS sample buffer and loaded onto 10 - 15% polyacrylamide gradient separating gels (SDS PAGE) with 4% stack layers (Laemmli, 1970). After electrophoresis at 40 mAmps for 5 hr, gels were stained with Coomassie blue.

Mass Spectrometry

Proteins were identified by mass spectrometry (MS) using two complementary approaches. In the first approach, proteins from total cellular and cytoplasmic extracts, and from sucrose density fractions were subjected to SDS PAGE. Individual gel bands or entire lanes (in up to 22 sections of 3-4 mm each) were subjected to in gel digestion with trypsin and peptides were analyzed by LC-MS/MS on an LTQ mass spectrometer as described previously (Beckmann and Fallon, 2013). In a second approach, enriched wStr were recovered from gradient fraction GF-50/60 (see Fig. 2) by dilution to 15% sucrose and centrifugation in a 2 ml tube at 16,100 × g for 15 min at 4°C. The pellet was resuspended in SPE buffer without EGTA and re-pelleted twice to deplete EGTA before tryptic digestion. Peptides were subjected to on-line reverse phase high pressure liquid chromatography (HPLC) for protein identification using an Orbitrap mass spectrometer (1D LC MS/MS, data set F), or were first subjected to off-line high pH HPLC before protein identification (2D LC MS/MS, data set G), as described previously (Tran et al., 2013).

Database Searching and Protein Identification

Tandem mass spectra were extracted by Sequest (Thermo Fisher Scientific, San Jose, CA, USA; version SRF v.3 or version 27, rev. 12) and searched against an rs_wolbachia_aedes_v200808_cRAP_flavivirusREV database that contained 74,570 protein entries from sequenced Wolbachia genomes, the Aedes aegypti genome, and flavivirus genomes available as of July 2011. Assembled Wolbachia genomes included those of the wPip WOL-B strain associated with Culex pipiens quinquefasciatus Pel mosquitoes from Sri Lanka (Klasson et al., 2008), the Drosophila-associated WOL-A strains, wMel (Wu et al., 2004) and wRi (Klasson et al., 2009), and the nematode-associated WOL-D strain, wBm (Foster et al., 2005). Incomplete genomes included the Drosophila-associated wAna and wWill WOL-A strains. Sequest parameters and protein sequence database information are reported in Table S7.

Scaffold (version 4.2.1, Proteome Software Inc., Portland, OR) was used to validate MS/MS based peptide and protein identifications. Peptide identifications were accepted if they could be established at greater than 95.0% probability by the Peptide Prophet algorithm (Keller et al., 2002). Protein identifications were accepted at probability settings based on at least 1, 2 or 3 unique identified peptides, as described in the Results section. Protein probabilities were assigned by the Protein Prophet algorithm (Nesvizhskii et al., 2003). Proteins that contained similar peptides and could not be differentiated based on MS/MS analysis alone were grouped to satisfy the principles of parsimony. False discovery rates for each collective dataset at different minimum unique peptide number (1, 2 and 3) are reported in Table S4. Identified proteins were classified into functional groups, including some hypothetical proteins without annotated functional identities, based on presence of conserved domains and use of the Protein Cluster and BLASTp tools (http://BLAST.ncbi.nim.nih.gov).

Statistical Analysis

All tests of association were performed with SAS v9.3 (Cary, NC; http://www.sas.com/en_us/home.html/).

Supplementary Material

Supp TableS1

Supplemental Table S1. Proteins in the wStr SDS PAGE protein footprint. (A) Wolbachia and Aedes (indicated by * symbol) proteins represented by at least three 95% confidence peptides in tryptic digests of bands A - E from four replicate gels. (B) Genbank Accession number. (C) Predicted protein mass (kDa). (D) Maximum number of detected peptides and percent sequence coverage of the protein. (E) Mean of the percentage of total spectra representing the protein averaged from four replicate gels.

Supp TableS2

Supplemental Table S2. The wStr proteome. (A) The 790 identified wStr proteins, as based on detection of at least two 95% confidence peptides in any one of four mass spectrophotometry data sets D - G. (B) Genbank Accession number. Underlined proteins were identified based on detection of peptides from multiple homologs in the MS search database and the Acc. # represents the homolog with the highest number of detected peptides, or highest protein coverage when peptide numbers from each homolog are equal. (C) Predicted protein mass in kDa. (D - G) Total number of peptides and percent coverage of the protein sequence in MS data sets D - G are shown in Columns D - G, respectively. Within columns D - G, underlined, boldface, underlined italic, and italic type fonts indicate, respectively, that one or more peptides were derived from an homolog different from that represented by the Acc. # in column B. (H) The relative abundance level (RAL) of the protein calculated as the mean of all peptides representing that protein in columns D - G. (I) Mean of all studentized residuals with standard deviation from Table S3, column K (Univariable model). (J) The functional class of the protein. (K) Annotation notes and homolog distribution of peptides from identified proteins. The number of peptides and percent sequence coverage from homologs different from that represented by the Acc. # in column B are indicated in type fonts corresponding to those in columns D - G. Note that the fonts representing the corresponding peptides in Column D - G are separated by semicolons.

Supp TableS3

Supplemental Table S3. Results of Univariable and Multivariable Analyses after log transformation of the outcome, Peptide Count, and predictor, Molecular Weight. See tabs at bottom for All Residuals, Univariable Model and Multivariable Model (adjusted for functional class and MS Dataset; see Table 6 for beta coefficients of each functional class). Runs 1, 2, 3 and 4 correspond to MS data sets D, E, F and G, respectively. Residuals (the difference between expected and observed log peptide count) are calculated from respective models. Studentized residuals (SR), which are further adjusted by estimated standard error, are used for all comparisons between individual proteins (column AG), and between functional classes (column AL). An SR of 0 corresponds to average abundance. We define proteins of above average abundance as those present in at least three runs with a mean SR that is > 0 to 1, abundant as > 1 but < 2, and high abundance as > 2.

Supp TableS4

Supplemental Table S4: Sequest and Scaffold program Publication Reports for MS data sets D – G and the wStr SDS PAGE protein footprint data sets 1 – 4. The file contains seven individual data set reports and a summary of protein false discovery rates across all data sets as tabs at bottom.