Outer membrane vesicles catabolize lignin-derived aromatic compounds in Pseudomonas putida KT2440

Results

To explore the interaction of aromatic–catabolic bacteria with lignin, we selected a gram-negative bacterium (P. putida KT2440; hereafter, P. putida) and two gram-positive bacteria (R. jostii RHA1 and Amycolatopsis sp. ATCC 39116; hereafter, R. jostii and Amycolatopsis sp.), which we previously reported to utilize a significant amount of lignin in a soluble, lignin-rich stream compared with other aromatic–catabolic bacteria (7). The lignin-rich substrate utilized in this study is a soluble alkaline extract from corn stover, which contains aromatic monomers and oligomers (SI Appendix, Fig. S1) that retain β-O-4 linkages from native lignin.

We first analyzed bacterial growth and carbon utilization in lignin-rich media (i.e., aromatic compounds, carboxylic acids, and supplemental glucose) and for comparison purposes, also in lignin-free media (containing glucose as the only carbon and energy source) over 120 h. Utilizable aromatic monomers, aliphatic carboxylic acids, and glucose were depleted within 24 h in P. putida cultures and within 72 h in R. jostii and Amycolatopsis sp. (Fig. 1 A–C and SI Appendix, Fig. S2), with oligomeric lignin as the primary remaining carbon source. The initial lignin content (monomeric + oligomeric) in these cultures was 1.7 g/L, and the total concentration of monomeric aromatic compounds was 0.27 g/L (16% of the total lignin) (SI Appendix, Table S1). Since these organisms utilized lignin contents of 23.4, 18.9, and 24.4% (Fig. 1 A–C) as measured with our previous approach (7), it is likely that they utilized a small fraction of lignin oligomers (7.4 ± 2.1% for P. putida, 2.9 ± 2.3% for R. jostii, and 8.4 ± 0.2% for Amycolatopsis sp. by 120 h).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Chemical and proteomics analyses in lignin-free (L−) and lignin-rich (L+) bacterial cultures. (A–C) Bacterial growth (OD₆₀₀) and concentrations of the major carbon sources in (A) P. putida, (B) R. jostii, and (C) Amycolatopsis sp. cultivations. The concentrations of minor carbon sources are shown in SI Appendix, Fig. S2. Error bars show the absolute difference between two biological replicates. (D) Schematic representation of atom signals identified by 2D-HSQC-NMR. (E–H) Structural lignin units analyzed by 2D-HSQC-NMR spectroscopy in 5-d L+ cultures of (E) P. putida, (F) R. jostii, (G) Amycolatopsis sp., and (H) noninoculated 5-d sample (control). Temporal and high-resolution spectra of the side chain and aromatic regions are presented in SI Appendix, Figs. S3–S6. OCH₃, methoxy groups; P, pyridine (internal standard). (I–K) Results from proteomic analyses in (I) P. putida, (J) R. jostii, and (K) Amycolatopsis sp. cultures. The left side of the bar chart shows the number of exclusive (in red) and enriched (in pink) proteins in L+ cultures compared with L− cultures as well as the number of proteins from that pool of proteins that appear only in the exoproteome compared with the intracellular fraction (in blue). The right side of the bar chart shows the number of proteins that appear or increase over time (from 24 to 72 h or from 72 to 120 h) within the pool of exclusive or enriched proteins in L+ cultures. The complete list of proteins is shown in SI Appendix, Tables S3–S8. These analyses originate from two biological replicates and two technical replicates in each organism, each time point (24, 72, 120 h), each media (L+, L−), and each protein location (extracellular vs. intracellular). Proteins in L+ cultures with absolute log ratios value that exceeded 2 SDs of all log ratios within the pairwise comparison L+ vs. L− are denoted as “enriched.” Proteins in 72- or 120-h cultures with absolute log ratios value that exceeded 2 SDs of all log ratios within the pairwise comparison 72 vs. 24 h or 120 vs. 72 h, respectively, are denoted as “increased.”

To understand the chemical changes in lignin, we also conducted two-dimensional (2D) heteronuclear single-quantum correlation (HSQC) NMR spectroscopy as a function of time (Fig. 1 D–H and SI Appendix, Figs. S3–S6). This technique is commonly used to understand lignin chemistry (23, 24), including to elucidate modifications by chemical catalysts (25) and ligninolytic enzymes (26, 27). The major changes occurred from 0 to 72 h. Ferulic acid (FA) and p-coumaric acid (pCA)-derived signals such as FA₂, pCA_β, pCA_2/6, and pCA_α + FA_α disappear at 72 h in P. putida and R. jostii treatments, which corresponds to pCA and FA depletion analyzed by liquid chromatography (LC) mass spectrometry (MS/MS) (SI Appendix, Fig. S2). Conversely, Amycolatopsis sp. still exhibits signals for FA₂ and pCA_α + FA_α at 120 h, although the latter was reduced compared with the noninoculated control. 2D-HSQC-NMR spectra also exhibit a reduction in the A_α and A_β signals from the β-O-4 bond in the three bacteria compared with the control, which suggests that these bacteria are breaking down oligomeric lignin. However, it is worth noting that this cleavage does not necessarily imply increased lignin utilization.

To provide insights into the extracellular enzymes potentially involved in lignin catabolism, we conducted differential (lignin-free and lignin-rich media), spatial (intracellular and extracellular), and temporal (24, 72, and 120 h of incubation) proteomic analyses with the three bacteria (Dataset S1). P. putida reached stationary phase at 24 h, and R. jostii and Amycolatopsis sp. reached stationary phase between 48 and 72 h, when most of the readily available carbon sources were depleted (Fig. 1 A–C). Despite reaching stationary phase—a growth phase that leads to cell lysis events—the percentages of differential proteins between the extracellular and intracellular fractions were 54, 33, and 60% in P. putida, R. jostii, and Amycolatopsis sp., respectively (SI Appendix, Fig. S7), indicating that the intracellular and extracellular fractions are different. The extracellular fraction contained a less diverse set of proteins compared with the intracellular fraction in all cases, and P. putida exhibited the highest number of lignin-induced extracellular proteins compared with both the intracellular fraction and the other organisms (Fig. 1 I–K and SI Appendix, Fig. S8). It is notable that the other microbes exhibit a higher number of predicted proteins from the genome than P. putida (SI Appendix, Table S2). Furthermore, the P. putida exoproteome showed that a higher number of proteins were enriched when oligomeric lignin became the major carbon source (from 24 to 72 h) compared with the other two organisms (from 72 to 120 h) (Fig. 1 I–K).

To identify putative enzymes that could be involved in lignin modification and/or depolymerization from these large pools of proteins, here we focused on lignin-induced extracellular proteins that were not detected in the intracellular compartment (SI Appendix, Fig. S9). The most represented catalytic functions, according to gene ontology annotation, were hydrolases and oxidoreductases in the three organisms (∼30 to 40%) (SI Appendix, Tables S3–S5). Notably, these functions were further enriched in the subset of proteins that appeared or increased at the time at which oligomeric lignin was the major carbon source—up to 56% in P. putida and 80% in Amycolatopsis sp. (the percentages remained similar in R. jostii, ∼40%) (SI Appendix, Tables S6–S8). Most of the lignin-depolymerizing enzymes described to date, including some potentially involved in β-O-4 cleavage [i.e., dye decolorizing peroxidases (DyPs) (28⇓–30), multicopper oxidases (31, 32)], belong to the broad group of oxidoreductases. In the current analysis, we have only detected a DyP in P. putida. However, we also found other enzymes that could be potentially involved in the modification of both oligomeric and monomeric aromatic or phenolic compounds (SI Appendix, Tables S3–S8). For instance, azoreductases (33, 34), a xenobiotic reductase (35), and a 2,3-quercetin dioxygenase (36) were detected in P. putida. All but DyP appeared or were enriched when oligomeric lignin was the major carbon source. In the case of R. jostii, a cholesterol oxidase (37) and a cytochrome P450 (38) were also observed. In Amycolatopsis sp., another cholesterol oxidase, cytochrome P450, and two oxidoreductases (one potentially belonging to the glucose–methanol–choline superfamily) (39) were detected. Among these enzymes, only the cytochrome P450 in R. jostii and cholesterol oxidase in Amycolatopsis sp. appeared from 72 and 120 h, when oligomeric lignin is the major remaining carbon source (protein access numbers are in SI Appendix, Tables S3–S8). Additional information on these oxidoreductases as well as other enzymes that were not part of this protein subset and that are potentially involved in lignin modification is presented in SI Appendix, Text S1.

A question that arises from these data is how the various proteins are transported from the cell to the extracellular milieu. We used several algorithms (i.e., SignalP, TatP, SecP, PSORTb) to predict protein location or export from the cytoplasm (40⇓⇓–43). Only 40% of the lignin-induced proteins (that were exclusively detected in the extracellular fraction) were predicted to be extracellular and/or exported in P. putida by at least one of the algorithms, whereas 70% in R. jostii and Amycolatopsis sp. are predicted to be secreted (SI Appendix, Tables S9–S11). Moreover, a larger number of lignin-induced proteins in the exoproteome of P. putida lignin cultures is predicted to participate in catabolic (32 to 37 proteins) and biosynthetic processes (74 to 89 proteins) compared with R. jostii (10 to 16 and 25 to 33 proteins, respectively) and Amycolatopsis sp. (5 to 13 and 20 to 40 proteins, respectively) (SI Appendix, Figs. S10–S12). Interestingly, some enzymes from the βKA pathway (considered to be cytoplasmic) were detected in the extracellular fraction of P. putida and R. jostii in lignin cultures and to a lesser extent, in Amycolatopsis sp. (SI Appendix, Text S2 and Figs. S13–S18 show detailed trends in both intracellular and extracellular fractions for the βKA pathway and alternative routes for the catabolism of aromatics). Taken together, these results highlight the importance of considering other enzyme secretion mechanisms that cannot be predicted by these algorithms or explained due to cell lysis.

Based on these results, we performed electron microscopy to investigate cell integrity and the response to the lignin-rich substrate. Interestingly, we observed that P. putida secretes a striking amount of OMVs in lignin-rich media (Fig. 2 A and B), especially when compared with P. putida in lignin-free media or R. jostii and Amycolatopsis sp. in either media (SI Appendix, Fig. S19). In view of these observations, we hypothesized that some of the proteins not predicted to be secreted could be trafficked to the extracellular compartment via OMVs. Thus, we next focused on P. putida OMVs. While the production of OMVs by this bacterium has been previously described (44), their interaction with lignin has not yet been reported. Dynamic light scattering (DLS) measurements and transmission electron microscopy (TEM) (Fig. 2 E and F) identified two distinct sizes of OMVs in both media formulations with no clear trend in vesicle diameter over time (Fig. 2 C and D). A 7% increase in cell death is observed in lignin-rich media compared with lignin-free media (SI Appendix, Fig. S20), which may contribute to production of vesicles via explosive cell lysis (17). Regardless, both scanning electron microscopy (SEM) and TEM data confirm that OMVs are also secreted via blebbing events from the outer membrane of P. putida cells with undamaged membranes even in late stationary phase (Fig. 2 G and H).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

P. putida secretes OMVs of two distinct sizes in lignin-rich media. Representative SEM images of P. putida cultivations at 72 h in (A) lignin-free and (B) lignin-rich media. Diameter of vesicles from P. putida cultivations in (C) lignin-free and (D) lignin-rich media as measured by DLS (intensity). Diameter is reported as an average ± SD of 3 biological replicates and 10 technical replicates for two distributions (D1 and D2), which represent two vesicle subpopulations. (E and F) TEM images of the two size populations of OMVs from P. putida cultivations in lignin-rich media at 72 h. Vesicles were drop cast and stained prior to imaging. (G) SEM image of a single P. putida with vesicles blebbing (white arrows). (H) TEM image of a P. putida cell with vesicles blebbing (white arrows) from the outer membrane (black asterisk). Cells were cryofixed, embedded, sectioned, and stained prior to imaging.

To gain insight into extracellular protein sorting in P. putida, the exoproteome was separated into an OMV fraction and the vesicle-free secretome (VFS) (Fig. 3A) at 24, 48, and 72 h of cultivation in lignin-free and lignin-rich media and analyzed by proteomics. This timeframe was selected because 1) most changes in the exoproteome were observed (Fig. 1I) and 2) oligomeric lignin was the primary carbon source. A rich dataset was generated exhibiting excellent reproducibility between biological triplicates with protein abundance differences driven by fraction, media composition, and time of harvest (SI Appendix, Fig. S21 and Dataset S2). In lignin-free media, 50% of the proteins (382 proteins) are differentially abundant between OMV and VFS fractions, from which 58% are enriched in OMVs and 42% are enriched in VFS (Fig. 3B). However, during cultivation in lignin-rich media, the number of proteins with differential abundance between fractions increased up to 70% (486 proteins), from which 51% are enriched in OMVs and 49% are enriched in VFS (Fig. 3B). Between lignin-rich and lignin-free OMVs, only 32% of the protein cargo are shared (Fig. 3B). Of the proteins enriched in OMVs in lignin-free media, only 1% (four proteins) change in abundance across 72 h of cultivation (Fig. 3B). Remarkably, growth in lignin-rich media triggers a highly dynamic OMV proteome, with 54% of OMV cargo (263 proteins) changing in abundance across the 72-h cultivation (Fig. 3B). Together, OMVs from lignin-rich cultivations exhibit 125 proteins with both lignin-dependent spatial enrichment and temporal changes (SI Appendix, Table S12).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

The exoproteome is selectively compartmentalized into OMVs, and the protein content of OMVs from lignin-rich cultivations is highly dynamic over time. (A) The extracellular milieu (= exoproteome) was fractionated into OMVs and VFS prior to proteomics analysis of each fraction from biological triplicates grown in lignin-rich and lignin-free media at 24, 48, and 72 h of cultivation. (B) Proteins with differential abundance between the OMV and VFS or OMV over time. Tests were conducted with a pairwise ANOVA on data from biological triplicates, and a Benjamini–Hochberg correction was performed on the resulting P values; proteins that had a P value < 0.05 are considered to have differential abundance, and those that did not meet this significance cutoff are considered to have similar abundance. Proteins with differential abundance between each time point in lignin-rich OMVs are presented in SI Appendix, Table S14. (C) Temporal trends of abundance in the OMV and VFS for the 263 proteins (p.) that change over time in lignin-rich OMVs. Clusters were identified by k-means clustering, and protein abundance is presented as a z score. Trend lines represent the 95% CI for VFS proteins (red) and OMV proteins (blue) in each cluster. The numbers of proteins in each cluster are presented in the upper left-hand corner and listed individually in SI Appendix, Table S13. (D and E) Heat-mapped log₂ protein abundance for select proteins for each biological triplicate at 24, 48, and 72 h of cultivation in the VFS and OMV in lignin-rich and lignin-free media. National Center for Biotechnology Information (NCBI) RefSeq identifiers are provided for each protein in parentheses following the description. (D) Select proteins with enrichment into lignin-rich OMVs and function (known or putative) in oxidation–reduction processes. (E) Select proteins with differential abundance between lignin-rich and lignin-free OMVs and function (known or putative) as a lipoprotein, small molecular binding protein, transport related, or OMP.

Of the OMV proteins that change over time in lignin-rich media, clusters of temporal trends highlight distinct patterns of protein sorting between the OMV fraction and the VFS (Fig. 3C and SI Appendix, Fig. S22 and Table S13). A pairwise comparison of OMVs in lignin-rich media between 24 and 48 h, between 48 and 72 h, and between 24 and 72 h identified 58, 2, and 236 proteins, respectively, with changing abundance (adjusted P value ≤ 0.05) (SI Appendix, Fig. S22 and Table S14). Oxidoreductases increase in abundance in the OMVs from 24 to 48 h (Fig. 3D and SI Appendix, Table S14), corresponding to the shift from catabolism of monomers to interaction with oligomeric lignin in stationary phase. These results corroborate and refine the earlier findings. For example, the first proteomics dataset identified an flavin mononucleotide (FMN)-dependent nicotinamide adenine dinucleotide (NADH)-azoreductase and a quercetin 2,3-dioxygenase to be dynamically enriched in the exoproteome (Fig. 1I); here, we refine spatial localization to the OMV and temporal sorting to between 24 and 48 h (Fig. 3D). The low (or below the limit of detection) abundance of these proteins in the VFS, even at 72 h of cultivation, suggests that enrichment into vesicles is not an artifact of cell lysis and/or that proteolysis in the VFS occurs rapidly (Fig. 3D and SI Appendix, Fig. S23).

Surprisingly, many lipoproteins and outer membrane porins (OMPs)—which were originally hypothesized to be constitutive elements of the vesicles—exhibit differential abundance between OMVs in lignin-rich relative to lignin-free media (Fig. 3E and SI Appendix, Fig. S24). Presence in the VFS under both media conditions precludes the possibility that differences in gene expression underlie this observation (Fig. 3E) and thus suggests that these membrane-bound or membrane-associated proteins are selectively excluded from OMV blebbing sites when lignin is present. Selective exclusion of canonical outer membrane proteins from the OMV has been previously reported (45) but is not widely described. Lipoproteins, OMPs, and lipopolysaccharides are involved in models for selective packaging of cargo and subsequent vesicle biogenesis (46), making these attractive candidates for studying lignin-specific vesiculation and cargo sorting into OMVs.

Next, we examined enzymes from the βKA pathway (Fig. 4A) as they were detected in previous exoproteome analysis (SI Appendix, Fig. S14). Of the detected proteins in the βKA pathway, all are strongly enriched in the OMVs from lignin-rich media compared with the VFS and the lignin-free exoproteome (Fig. 4B). Intra-OMV abundance of βKA enzymes increases temporally after 24 h of cultivation (Fig. 4B). Since enrichment is not observed in the VFS, we posit that enzymes in the βKA pathway are secreted via OMVs for the extracellular turnover of lignin-derived aromatics.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Catabolism of lignin-derived aromatics via the βKA pathway is enhanced by OMVs. (A) The βKA pathway for catabolism of lignin monomers p-coumarate and ferulate into succinyl-Coenzyme A (succinyl-CoA) and acetyl-Coenzyme A (acetyl-CoA), which support cellular growth. Metabolites that were used for in vivo and in vitro assays are shown in blue and orange. Proteins in bold were detected in the proteomics analysis. Due to the nontargeted proteomics methodology used throughout this study, we note that lack of detection of certain enzymes from the βKA pathway does not necessarily mean biological absence. (B) Proteins detected within the βKA pathway in the VFS or OMV are plotted as a heat map of log₂ abundance. Values for biological triplicate at each time point are plotted separately. *P < 0.05 from an ANOVA with Tukey’s honest significant difference (HSD) of the VFS relative to OMV; **P < 0.005 from an ANOVA with Tukey’s HSD of the VFS relative to OMV. (C) Improvement of final culture growth (OD₆₀₀) from complementation of P. putida ΔpobAR or P. putida ΔpcaHG with wild-type OMVs (boiled or fresh) for growth on M9 minimal media with 4-HBA or PCA, respectively, as the sole carbon and energy source. Growth enhancement was calculated as the percentage increase in OD₆₀₀ from the mutant strain + substrate to the mutant strain + substrate + OMV. *P < 0.05 from a one-tailed paired t test. (D) In vitro turnover of 4-HBA or PCA by boiled or fresh OMVs. Values are displayed as the percentage decrease in substrate from t = 0 h to t = 24 h of incubation. n.s., P > 0.05 from a one-tailed paired t test. *P < 0.05 from a one-tailed paired t test. Fcs, feruloyl-CoA-synthase; Ech, feruloyl-CoA hydratase-lyase; Vdh, vanillin dehydrogenase; PobA, 4-hydroxybenzoate 3-monooxygenase; VanAB, vanillate O-demethylase; PcaHG, protocatechuate 3,4-dioxygenase; PcaC, 4-carboxymuconolactone decarboxylase; PcaD, 3-oxoadipate enol-lactonase; PcaIJ, 3-oxoadipate CoA-transferase; PcaF, beta-ketoadipyl-CoA thiolase.

To test this hypothesis, we engineered two P. putida deletion mutants that cannot grow on two catabolic intermediates: P. putida ΔpobAR on 4-HBA and P. putida ΔpcaHG on PCA. These mutants were inoculated to the same cell density and separately complemented with wild-type OMVs for growth analyses (SI Appendix, Fig. S25 and Table S15). A boiled OMV control was also included to test if incorporated or encapsulated metabolites (e.g., lipids, amino acids) complement growth. Boiled OMVs enhance growth (measured as optical density at 600 nm [OD₆₀₀] after 72 h) of P. putida ΔpobAR by 175% and P. putida ΔpcaHG by 153% (Fig. 4C). Active OMVs enhance growth of P. putida ΔpobAR cultivations in 4-HBA by 336% and P. putida ΔpcaHG cultivations in PCA by 319% (Fig. 4C). While these growth enhancements are roughly double to what is observed in the boiled control, active OMV complementation does not fully restore the growth observed in wild-type cultivations (SI Appendix, Fig. S26 and Table S15), suggesting a limited enzyme content, lifetime in the exoproteome, and/or lack of cofactors.

To further test the hypothesis that OMVs mediate aromatic catabolism, we tested in vitro turnover of 4-HBA and PCA by OMVs. PCA is known to be unstable, which was reflected in the 24 and 27% turnover in abiotic and boiled controls, respectively (Fig. 4D and SI Appendix, Fig. S27). Interestingly, active OMVs increase the turnover of PCA by 59%, demonstrating that the protocatechuate 3,4-dioxygenase PcaHG remains catalytically active into the OMVs (Fig. 4D). Conversely, 4-HBA turnover is not improved by active OMVs (Fig. 4D). It is noteworthy that the 4-hydroxybenzoate 3-monooxygenase PobA requires nicotinamide adenine dinucleotide phosphate (NADPH) as a cosubstrate, whereas PcaHG does not (Fig. 4A). Considering NAD(P)H instability, its content in isolated OMVs remains unknown. In fact, when OMVs were provided exogenous reducing equivalents, NAD(P)H only decreased in the presence of 4-HBA (SI Appendix, Fig. S28). While this does not provide information on the in vivo presence of NAD(P)H or the generation thereof, these results indicate that PobA activity is retained inside the OMV. Together, these data demonstrate that OMV-encapsulated enzymes actively contribute to the turnover of model lignin-derived aromatic compounds.