Proteins in the periplasmic space and outer membrane vesicles of Rhizobium etli CE3 grown in minimal medium are largely distinct and change with growth phase

Rhizobium etli CE3 grown in succinate-ammonium minimal medium (MM) excreted outer membrane vesicles (OMVs) with diameters of 40 to 100 nm. Proteins from the OMVs and the periplasmic space were isolated from 6 and 24 h cultures and identified by proteome analysis. A total 770 proteins were identified: 73.8 and 21.3 % of these proteins occurred only in the periplasm and OMVs, respectively, and only 4.9 % were found in both locations. The majority of proteins found in either location were present only at 6 or 24 h: in the periplasm and OMVs, only 24 and 9 % of proteins, respectively, were present at both sampling times, indicating a time-dependent differential sorting of proteins into the two compartments. The OMVs contained proteins with physiologically varied roles, including Rhizobium adhering proteins (Rap), polysaccharidases, polysaccharide export proteins, autoaggregation and adherence proteins, glycosyl transferases, peptidoglycan binding and cross-linking enzymes, potential cell wall modifying enzymes, porins, multidrug efflux RND family proteins, ABC transporter proteins, and heat shock proteins. As expected, proteins with known periplasmic localizations (phosphatases, phosphodiesterases, pyrophosphatases) were found only in the periplasm, along with numerous proteins involved in amino acid and carbohydrate metabolism and transport. Nearly one-quarter of the proteins present in the OMVs were also found in our previous analysis of the R. etli total exproteome of MM-grown cells, indicating that these nanoparticles are an important mechanism for protein excretion in this species. IMPORTANCE The reduction of atmospheric nitrogen to ammonia by rhizobia symbiotically associated with legumes is of major importance in sustainable agricultural. Rhizobia excrete a variety of symbiotically important proteins using canonical secretion systems. In this work, we show that Rhizobium etli grown in culture also excretes proteins in membrane-enclosed structures called outer membrane vesicles (OMVs). This study reports OMV production by rhizobia. Proteins identified in the OMVs included Rhizobium adhering (Rap) and autoaggregation proteins, polysaccharidases, RTX toxins, porins and multidrug efflux proteins. Some of these proteins have important roles in the R. etli-common bean symbiosis, and their packaging into OMVs could deliver them to the environment in a concentrated yet diffusible form protected from degradation. The work described here provides a basis for future studies on the function of rhizobial OMVs in free life and symbiosis.


INTRODUCTION 13
Bacterial protein secretion is a vital function involving the transport of proteins 14 from the cytoplasm to other cellular locations, the environment, or to eukaryotic host (18,19). The presence of cytoplasmic proteins in periplasmic protein preparations is MG1655 were predicted to be periplasmic (24). 7 We used the PSLPred program to predict the subcellular localization of R. etli 8 proteins. While the localization predictions made with this program are over 90 % 9 accurate for proteins from gram negative bacteria (25), we noted that several proteins 10 had an unexpected predicted localization. For example, the ribosomal proteins S12 and 11 L31 were predicted to be periplasmic rather than cytoplasmic. Analysis of the R. etli 12 exoproteome using an alternative protein localization prediction program, LocTree3 13 (26) gave significantly different results in comparison to PSLpred. In the case of the 14 ribosomal proteins described above, LocTree3 predicted that S12 is in fact cytoplasmic, 15 but that L31 was secreted. periplasm, all that are classed as phosphatases, phosphodiesterases or pyrophosphatases 25 were found only in the periplasm (Table S1, Supplemental material), consistent with the biochemically determined localization of these enzymes in other rhizobia (18,28). In 2 addition, electron microscopic examination of the cell preparations obtained after hypo-3 osmotic shock showed that the IMs were still intact (results not shown). 4 The R. etli OMV fraction was obtained by differential centrifugation of culture 5 filtrates. Transmission electron microscopic examination of the OMVs purified from 6 6 and 24 h cultures showed that the vesicles were spherical and had diameters of 40 to 7 100 nm, within size range expected for OMVs (16,29). No pili, bacteria, flagella or 8 membrane debris were detected (Fig. 1). SDS-PAGE analysis showed that the OMV 9 protein patterns differed significantly from those of whole cell extracts (data not 10 shown), consistent with the proposal that specific protein sorting mechanisms are 11 important in determining the protein content of bacterial OMVs (3, 10, 30). 12 For proteins present only in the OMVs (Table 1) (Table 1 and Table S3 25 (Supplemental material)) were also found in the previously determined exoproteome (14). Thus, nearly one-quarter of the exoproteins identified in our previous study were 2 apparently excreted in OMVs. It should be noted that the mass spectrometric methods 3 used for protein identification in this and our previous (14, 15) work do not allow the 4 quantitation of proteins, but only reveals their presence or absence in a sample.

5
The 770 proteins identified in the R. etli periplasmic and OMV fractions at 6 6 and/or 24 h (Table S1 in Supplemental Material) represent 12.8 % of the 6022 predicted 7 ORFs encoded in its genome. Only 14.2 % of these proteins are plasmid-encoded, 8 representing less than half of the 32 % of the R. etli proteome that is 9 extrachromosomally encoded. There was no significant difference of the relative 10 proportion of plasmid-encoded proteins in the periplasm-only, OMV-only, and in both 11 the periplasm and OMV categories.

12
Of the 770 proteins identified, 568 and 164 (74 and 21 %) occurred exclusively 13 in the periplasm (Table S1, Supplemental material) and OMV (Table 1) fractions,   14 respectively. Remarkably, only 4.9 % of the total proteins were found in both fractions 15 (Table S3, Supplemental material), which argues against the random inclusion of 16 periplasmic proteins in the OMVs during their formation and supports the idea that 17 specific protein sorting mechanisms are at least partly responsible for determining OMV 18 protein content (30).

19
The number and identity of periplasmic and OMV proteins produced by bacteria 20 changes with culture age and growth conditions (4,30,33). In R. etli, we found 21 significant differences in the identities of the proteins present in the periplasm and 22 OMVs at 6 versus 24 h (Table S1, Supplemental material). In the periplasm-only 23 fraction (Table S2, Supplemental material), 31 and 45 % of the proteins were present 24 only at 6 and 24 h, respectively, and 24 % were present at both 6 and 24 h. For the proteins found exclusively in the OMVs (Table 1), 49 and 42 % were present only at 6 and 24 h, respectively, and 9 % were present at both times. The largely distinct protein 2 profiles for OMVs from log and stationary phase cultures, with relatively few proteins 3 present at both sampling times, indicates a time-dependent differential packaging of 4 proteins into the OMVs. For example, 9 of the OMV-exclusive protein COG groups are 5 comprised of a majority of proteins that are present at 6 but not 24 h. What accounts for 6 the disappearance of these proteins between 6 and 24 h? Possibly, these proteins are 7 selectively degraded within the OMVs, or are released from them, as the culture ages. It 8 has been proposed that different subpopulations of OMVs with a distinctive protein 9 content could exist in the same bacterial culture, but this has hardly been addressed 10 experimentally (35).

11
Proteins without a dedicated transport mechanism might enter the exoproteome 12 by interacting with one or more proteins that are specifically excreted. We determined 13 potential protein-protein interactions in the R. etli proteome using the ProLinks server 14 (http://prl.mbi.ucla.edu/prlbeta/ (36). While highly probable (P = 1.0) interactions were 15 predicted to occur between some members of the total proteome, none were found 16 among the proteins identified in the periplasm and/or OMVs, even at the lowest 17 probability setting (P = 0.4).    Proteins involved in energy production and conversion represent more than 11 % 5 of the OMV-exclusive proteins, over 2-fold more than among the periplasm-only 6 proteins. For oxidative phosphorylation, numerous components of NADH 7 dehydrogenase, cytochrome c, NADH-quinone oxidoreductase and ATP synthase were 8 found principally in the OMVs, although not all of the proteins required to completely 9 assemble these complexes was present.

10
Among the cytoplasmic proteins present only in OMVs, we found Tme, the 11 NADP + -specific malic enzyme that in S. meliloti appears to serve as a secondary               Table 1.