The highly conserved protease Lon has important regulatory and protein quality control functions in cells from the three domains of life. Despite many years of research on Lon, only a few specific protein substrates are known in most organisms. Here, we used a quantitative proteomics approach to identify novel substrates of Lon in the dimorphic bacterium Caulobacter crescentus. We focused our study on proteins involved in polar cell differentiation and investigated the developmental regulator StaR and the flagella hook length regulator FliK as specific Lon substrates in detail. We show that Lon recognizes these proteins at their C-termini, and that Lon-dependent degradation ensures their temporally restricted accumulation in the cell cycle phase when their function is needed. Disruption of this precise temporal regulation of StaR and FliK levels in a Δlon mutant contributes to defects in stalk biogenesis and motility, respectively, revealing a critical role of Lon in coordinating developmental processes with cell cycle progression. Our work underscores the importance of Lon in the regulation of complex temporally controlled processes by adjusting the concentrations of critical regulatory proteins. Furthermore, this study includes the first characterization of FliK in C. crescentus and uncovers a dual role of the C-terminal amino acids of FliK in protein function and degradation.
Intracellular proteolysis is a critical process in all cell types that is carried out by dedicated proteases. By removing damaged and non-functional proteins, proteases are necessary for maintaining protein homeostasis, in particular under stress conditions that threaten the proteome. Additionally, proteases have important regulatory roles in precisely adjusting the amounts of specific functional proteins, thus complementing transcriptional and post-transcriptional control mechanisms. Because of their important cellular functions, human proteases are considered as promising therapeutic targets (Bota and Davies, 2016) and their bacterial counterparts as potential antimicrobial drug targets (Culp and Wright, 2016). Hence, extending the knowledge of the substrate pools of specific proteases and the mechanisms underlying substrate selection is vital.
In prokaryotes and in the mitochondria and chloroplasts of eukaryotes, the majority of proteins is degraded by ATP-dependent proteases of the AAA+ (ATPases associated with various cellular activities) protein family (Sauer and Baker, 2011). The protease Lon was the first ATP-dependent protease to be identified and is widely conserved across the three domains of life (Gur, 2013). Lon forms a hexameric protease complex, of which each monomer contains three functional domains: an N-terminal domain, an ATP-dependent unfoldase domain, and a peptidase domain forming the proteolytic chamber (Gur, 2013). As a heat shock protein (Phillips et al., 1984), Lon is upregulated in response to protein unfolding stress, such as thermal stress, and contributes to the degradation of unfolded and misfolded proteins that accumulate under these conditions (Gur, 2013; Gur and Sauer, 2008; Tomoyasu et al., 2001). In addition to its role in protein quality control, Lon exerts important regulatory functions that can be traced to the degradation of specific sets of native substrate proteins involved in stress responses, metabolism, pathogenicity, and cell cycle progression (Tsilibaris et al., 2006).
Despite many years of research on Lon proteases, the number of validated Lon substrates is small in most organisms. Most knowledge about Lon has been obtained by studying bacterial Lon orthologs. In Escherichia coli, Lon specifically degrades several stress-induced regulators (Griffith et al., 2004; Langklotz and Narberhaus, 2011; Mizusawa and Gottesman, 1983), metabolic enzymes (Arends et al., 2018) as well as antitoxins of toxin-antitoxin systems (Muthuramalingam et al., 2016), and in several pathogenic bacteria, Lon was shown to degrade regulators of pathogenicity (Joshi et al., 2020; Puri and Karzai, 2017), thus playing important roles in the regulation of virulence pathways. In the alpha-proteobacterium Caulobacter crescentus, the number of identified Lon substrates to date is small, however, Lon is known to impact cell cycle progression by degrading essential cell cycle regulators.
The C. crescentus cell cycle is characterized by an asymmetric cell division event and morphologically distinct cell cycle phases (Curtis and Brun, 2010). Each division yields a flagellated and piliated swarmer cell and a sessile stalked cell. While the daughter stalked cell initiates DNA replication immediately after cell division, the daughter swarmer cell must differentiate into a stalked cell before entering S-phase. Faithful progression through the C. crescentus cell cycle relies on precise coordination of the polar differentiation events that trigger flagella, pili, and stalk biosynthesis with core cell cycle events, such as DNA replication and cell division (Curtis and Brun, 2010). Previous work established that around one-third of all genes in C. crescentus show cell cycle-dependent fluctuations in their expression levels (Fang et al., 2013; Laub et al., 2000). Many of the corresponding proteins have important developmental functions and peak in abundance in the cell cycle phase in which their function is most needed (Laub et al., 2000). In addition to transcriptional regulatory mechanisms, active proteolysis must occur to rapidly adjust protein concentrations to enforce these transcriptional changes (Grünenfelder et al., 2001). However, only a relatively small subset of cell cycle-regulated factors with developmental functions has so far been found to be subject to proteolysis in C. crescentus and the contributions of distinct proteases in this process remain incompletely defined.
Previous work established that the ClpP protease with its unfoldases ClpA and ClpX has key roles in C. crescentus development by mediating the temporally and spatially controlled degradation of several important cell cycle regulators (Joshi and Chien, 2016; Schroeder and Jonas, 2021). In addition to ClpXP, Lon plays an important role in C. crescentus cell cycle regulation (Joshi and Chien, 2016; Schroeder and Jonas, 2021). It degrades the swarmer cell-specific transcription factor SciP (Gora et al., 2013), the methyltransferase and transcriptional regulator CcrM (Wright et al., 1996), and the conserved replication initiator DnaA (Jonas et al., 2013). Lon-dependent degradation of SciP and CcrM contributes to their cell cycle-dependent regulation (Gora et al., 2013; Wright et al., 1996), while DnaA degradation by Lon ensures rapid clearance of the protein at the onset of nutritional and proteotoxic stress, preventing cell cycle progression under these conditions (Felletti et al., 2019; Jonas et al., 2013; Leslie et al., 2015). Although C. crescentus cells lacking Lon are viable, they grow more slowly and show aberrant chromosome content and division defects (Leslie et al., 2015; Wright et al., 1996), which can in part be attributed to the stabilization of DnaA, SciP, and CcrM. Interestingly, Δlon cells exhibit also characteristic developmental defects, that is, elongated stalks and motility defects (Wright et al., 1996; Yang et al., 2018), suggesting that Lon degrades additional substrates involved in cell differentiation.
Here, using a quantitative proteomics approach, we identified several proteins involved in the dimorphic life cycle of C. crescentus as novel Lon substrates. We studied in detail the transcriptional regulator of stalk biogenesis, StaR, and the flagella hook length regulator FliK as specific Lon substrates. We show that Lon is required to establish cell cycle-dependent fluctuations of these regulatory proteins, thereby contributing to their precise temporal accumulation during the cell cycle phase in which their function is needed. Furthermore, we demonstrate that the increased abundance of these proteins results in aberrant stalk length and motility defects. Taken together, our work revealed a critical role of Lon in coordinating cell differentiation with core cell cycle events in C. crescentus.
Previous work established that Lon degrades the cell cycle regulators DnaA, CcrM, and SciP in C. crescentus (Gora et al., 2013; Jonas et al., 2013; Wright et al., 1996). Absence of Lon, either in a Δlon strain or following Lon depletion, results in increased stability and abundance of these proteins (Figure 1A–B). Conversely, lon overexpression leads to lower protein abundance of DnaA, CcrM, and SciP (Figure 1C). Based on these results, we reasoned that it should be possible to identify novel Lon substrates by monitoring proteome-wide differences in protein stability and protein abundance in strains lacking or overexpressing lon in comparison to wild-type (WT) cells. Thus, we sampled cells from the following strain backgrounds and conditions for quantitative proteomics analysis: (1) 0, 15, and 30 min following protein synthesis shut down in the WT to assess protein stability in the presence of Lon; (2) 0, 15, and 30 min following protein synthesis shut down in Δlon cells to examine protein stability in the absence of Lon; (3) before and after 4.5 hr of Lon depletion in a Pvan-dependent Lon depletion strain; and (4) before and after 1 hr of inducing lon overexpression from a medium-copy plasmid. Tandem mass tag (TMT) labeling and mass spectrometrical (MS) analysis were used to detect proteome-wide differences in protein levels across these samples. We obtained signals for 2270 or 2261 proteins, respectively, in our two biological replicates and sorted the proteins with respect to four criteria (see Materials and methods for details): (1) to be more stable in Δlon cells than in the WT after 30 min of translation inhibition, (2) to be present in higher abundance at t=0 in Δlon cells compared to the WT, (3) to be upregulated after Lon depletion compared to non-depleting conditions, and (4) to be downregulated following xylose-induced lon overexpression compared to non-inducing conditions (Figure 1D). We identified 26 proteins that fulfilled all four criteria, one of them being DnaA. 120 proteins fulfilled three criteria, and because CcrM and SciP were in this group of proteins (Figure 1D), we considered proteins satisfying either three or four criteria as putative Lon substrates. The 146 proteins that fell into this group contained proteins from all major functional categories (Figure 1E). In comparison to all detected proteins, the group of putative Lon substrates contained a lower proportion of metabolic proteins and proteins involved in core genetic information processing, whereas proteins involved in cell cycle and cell differentiation processes, signal transduction and stress responses as well as unclassified proteins showed a higher relative abundance (Figure 1E). Notably, we did not detect FixT and HipB2 in our proteomics experiment, two other recently reported Lon substrates in C. crescentus (Stein et al., 2020; Zhou et al., 2021).
Since Δlon cells have previously been shown to exhibit developmental defects (Wright et al., 1996; Yang et al., 2018), we focused this study on the group of potential Lon substrates annotated to have functions in cell cycle and cell differentiation processes. This group included proteins involved in central cell cycle regulation, chromosome partitioning, stalk morphogenesis as well as motility and chemotaxis (Figure 1F). Interestingly, according to previously published RNA-sequencing data, a large subset of the genes encoding these proteins are subject to cell cycle-dependent regulation and peak in their expression during a specific cell cycle phase (Figure 1F; Lasker et al., 2016; Schrader et al., 2016).
One of the proteins that satisfied the four criteria in our proteomics experiment most clearly was the transcriptional regulator StaR, a protein previously reported to be involved in the regulation of stalk biogenesis and holdfast production (Figure 1D and F; Biondi et al., 2006; Fiebig et al., 2014). Like the three known substrates DnaA, CcrM, and SciP, StaR was more stable in the Δlon strain compared to the WT and showed increased steady-state levels in Δlon and Lon-depleted cells, but reduced protein levels in cells overexpressing lon (Figure 2A). To validate these proteomics data, we monitored the stability of StaR in WT and Δlon cells using a StaR-specific antibody (Fiebig et al., 2014). In the WT, the levels of StaR were below the limit of detection, however, we observed a strong and stable band corresponding to StaR in the Δlon strain (Figure 2B), indicating that StaR is upregulated in Δlon cells, a result that is consistent with the proteomics data (Figure 2A). To analyze the rate of StaR degradation in the presence of Lon, we expressed staR from a medium-copy vector to elevate its levels. In this strain, StaR was degraded with a half-life of approximately 12 min, when Lon was present (Figure 2C). Absence of Lon resulted again in complete stabilization of StaR and increased steady-state levels, which is in line with our proteomics data, and confirms that StaR degradation depends on Lon. To directly test if StaR is a Lon substrate, we performed an in vitro degradation assay with purified StaR and Lon. This assay showed that Lon readily degrades StaR in an ATP-dependent manner (Figure 2D). Hence, StaR is a Lon substrate and no additional factors are required for recognizing and degrading StaR, at least in vitro.
We also investigated if Lon recognizes StaR via a degron sequence at one of the termini and monitored the degradation of FLAG-tagged StaR variants. Addition of the 3xFLAG tag to the C-terminus of StaR (StaR-F) completely abolished degradation (Figure 2E), indicating that a freely accessible C-terminus of StaR is required for degradation by Lon. Addition of the tag to the N-terminus of StaR (F-StaR) still enabled notable degradation within 60 min after shutting down protein synthesis (Figure 2E). However, degradation of this N-terminally tagged StaR was also observed in the Δlon strain, suggesting that another protease degrades this StaR variant, probably because of changes in StaR folding that result from the addition of the tag. These data show that native N-termini and C-termini of the protein are required for Lon-dependent degradation.
As a transcriptional regulator of stalk biogenesis and holdfast production, StaR function is expected to be particularly needed at the beginning of the cell cycle, when the swarmer cell differentiates into a stalked cell (Figure 3A). Indeed, previously published RNA sequencing data show that staR mRNA levels fluctuate during the cell cycle, peaking in the swarmer and early stalked cell before declining during S-phase and remaining low until cell division (Figure 3B; Lasker et al., 2016; Schrader et al., 2016). Furthermore, existing ribosome profiling data show that StaR is translated predominantly in the swarmer and early stalked cells, but only at low levels during the remaining cell cycle (Lasker et al., 2016; Schrader et al., 2016). Quantification of StaR protein levels by Western blot analysis in synchronized cultures showed that protein levels follow this expression pattern in WT cultures (Figure 3C). StaR was detectable within 30 min after synchronization before it was strongly downregulated and remained below the limit of detection for the rest of the cell cycle. Strikingly, the cell cycle-dependent changes in protein abundance were absent in the Δlon strain, in which StaR levels remained high until 75 min after synchronization (Figure 3C). Thus, Lon-dependent degradation of StaR is required for establishing oscillations of StaR levels during the cell cycle.
Next, we assessed the importance of Lon-mediated StaR proteolysis for correct stalk biogenesis. Consistent with a previous study (Wright et al., 1996), we observed that the stalks of Δlon cells are significantly elongated compared to the WT (Figure 3D–E). Because StaR overexpression was shown to lead to an increase of stalk length (Biondi et al., 2006), we reasoned that the abnormal stalk length of Δlon cells might be caused by the higher abundance of StaR in these cells. To address this hypothesis, we introduced the ΔstaR deletion into the Δlon strain background and assessed stalk length of this ΔstaR Δlon double mutant (Figure 3D–E, Figure 3—figure supplement 1). The ΔstaR Δlon mutant phenocopied the ΔstaR single mutant, in which stalks are shortened compared to Δlon cells, demonstrating that Lon affects stalk length through StaR (Figure 3D–E). To further investigate the relationship between Lon, StaR, and stalk lengths, we also analyzed stalk morphology under phosphate-limiting conditions, in which stalks are drastically elongated in C. crescentus (Schmidt and Stanier, 1966). Stalk length in the different strain backgrounds followed the same trend as under optimal conditions (Figure 3—figure supplement 1). However, all four strains were able to strongly elongate their stalks under phosphate starvation, suggesting that StaR and Lon are not required for starvation-dependent stalk elongation. Taken together, our data demonstrate that Lon-mediated degradation of StaR is required for proper stalk biogenesis during the cell cycle.
In addition to StaR, our proteomics experiments identified several proteins involved in flagella-based motility and chemotaxis as potential Lon substrates (Figure 1F). Particularly promising hits in this group of proteins were CCNA_00944 and MotD (CCNA_00945) that are encoded by partly overlapping open reading frames and showed similar changes in protein abundance and stability in the Lon deficient and Lon overproducing strains in our proteomics experiments (Figure 4A–B). While CCNA_00944 is annotated as a flagella hook length determination protein, MotD is annotated as a chemotaxis protein. According to a signature-based annotation of MotD in UniProtKB (entry A0A0H3C6R3), MotD contains a conserved Flg hook domain that is commonly present in the C-terminal portion of FliK proteins that control flagella hook length in many bacteria (Waters et al., 2007). When we attempted to clone CCNA_00944, we repeatedly observed one additional guanosine in the cloned gene sequence that was not present in the reference genome sequence of C. crescentus NA1000, the strain that we use. This insertion, which is also present in the sequence reads of previously published RNA-sequencing data (Schrader et al., 2014), generates a frameshift that merges the CCNA_00944 gene with the downstream located motD gene, thus forming one single open reading frame, of which the 3′ portion corresponds to motD (Figure 4B). This resulting gene corresponds to a single open reading frame (CC_0900) in C. crescentus CB15, the isolate from which NA1000 is derived. These observations indicate that CCNA_00944 and motD are incorrectly annotated in the reference genome of NA1000 and instead form one continuous open reading frame. Consistently, when we performed Western blot analysis with antiserum raised against the protein portion corresponding to MotD, we detected one single protein band that runs at high molecular weight (Figure 4—figure supplement 1), confirming that CCNA_00944 and motD form together one single open reading frame. Because the C-terminal portion of this new gene encodes a Flg hook domain that is a characteristic of FliK proteins in other bacteria and because no other gene has so far been annotated as fliK in C. crescentus, we named the new gene fliK (Figure 4B). This annotation of fliK is also in line with a recent study in Sinorhizobium meliloti, which suggested that the motD gene from alpha-proteobacteria should be renamed fliK (Eggenhofer et al., 2006).
Having established that CCNA_00944 and MotD are part of the same FliK protein, we next investigated its regulation by Lon. Consistent with our proteomics data, we found that FliK abundance and stability were notably increased in Δlon cells, consistent with FliK being a Lon substrate (Figure 4C). To investigate the sequence determinants required for Lon to interact with FliK, we expressed an N-terminally FLAG-tagged version of FliK and monitored its stability in vivo. Similar to the non-tagged FliK, 3xFLAG-FliK was efficiently degraded in WT cells but stable in cells lacking Lon (Figure 4D). This result led us to hypothesize that the recognition by Lon occurs via the C-terminal domain of FliK. Therefore, we monitored the stability of the C-terminal portion of FliK containing the Flg hook domain, which corresponds to the formerly annotated MotD protein. Like the full-length protein, degradation of this truncated FliK protein (FliK-C), with a FLAG-tag at the N-terminus, depended strongly on Lon (Figure 4E). Furthermore, in vitro degradation assays showed that Lon degrades non-tagged FliK-C in an ATP-dependent manner (Figure 4F, Figure 4—figure supplement 2). Based on these results, we conclude that FliK is a Lon substrate, and that its C-terminal portion containing the Flg hook domain is sufficient for Lon-dependent degradation.
To further pinpoint the regions within FliK-C that are required for Lon-dependent turnover, we analyzed the degradation of a set of additional truncation mutants. According to the MobiDB database (Piovesan et al., 2021), the C-terminal portion of FliK lists two unordered regions. We engineered FliK-C variants that lack either of these unordered regions (FliK-C∆N44 and FliK-C∆59–91) or the C-terminal part (FliK-C∆C44). In vitro degradation assays showed that deletion of the unordered regions did not influence degradation, whereas removal of the 44 C-terminal amino acids completely stabilized the protein (Figure 4F). This result, and the fact that some of the degrons recognized by Lon are located at the very C-terminus of Lon substrates (Burgos et al., 2020; Ishii et al., 2000; Puri and Karzai, 2017; Zhou et al., 2019), prompted us to determine the in vivo stability of a FliK-C variant lacking only the C-terminal five amino acids with the sequence LDIRI (3xFLAG-FliK-C∆5) (Figure 4G). This deletion abolished FliK-C degradation (Figure 4G), demonstrating that the interaction between Lon and FliK depends on these C-terminal amino acids.
Like many other proteins involved in flagella biogenesis in C. crescentus, the transcription of the two annotated genes CCNA_00944 and motD that together form the fliK gene is cell cycle regulated (Lasker et al., 2016; Schrader et al., 2016), with mRNA levels and ribosome occupancy peaking in late S-phase when a new flagellum at the pole opposite the stalk is being assembled (Figure 5A). Consistently, Western blot analysis with synchronized WT C. crescentus cultures showed that FliK protein was not detectable in the beginning of the cell cycle, but began to accumulate in late stalked cells before reaching a maximum in abundance in predivisional cells shortly before cell division (Figure 5B). This cell cycle-dependent pattern of FliK abundance was completely absent in the Δlon strain, in which FliK was already detectable in swarmer cells and remained at high levels throughout the cell cycle (Figure 5B). This result shows that, as in the case of StaR, Lon is absolutely necessary to ensure that the protein is eliminated in the cell cycle phase when its function is no longer needed.
Next, we wanted to investigate if the Lon-dependent regulation of FliK abundance is required for proper motility in C. crescentus. Consistent with a previous study (Yang et al., 2018), we observed that Δlon cells show reduced motility in soft agar compared to the WT (Figure 5C), which might be caused by reduced flagella function in addition to growth and cell division defects. Additionally, our proteomics data revealed that the levels of α-flagellins FljJ, FljK, and FljL and β-flagellins FljM and FljN, which compose the structural components of flagella, are strongly downregulated in the Δlon mutant (Figure 5D). Although a Lon-dependent effect on flagellin levels was not apparent after 4.5 hr of Lon depletion and only to a lesser extent upon lon overexpression (Figure 5D), these data point to an indirect involvement of Lon in the regulation of flagella biosynthesis. One explanation for the motility defect and the reduced flagellin levels in the Δlon mutant might be the stabilization and higher levels of SciP in this strain (Figure 1A; Gora et al., 2013), which is known to negatively affect flagellin gene expression through CtrA (Gora et al., 2010). Since correct regulation of FliK levels was shown to be critical for proper flagella biosynthesis in other species (Muramoto et al., 1998; Waters et al., 2007), we thought that the stabilization and thus increased abundance of FliK in the absence of Lon might contribute to the motility defect of Δlon cells as well. To specifically study the consequences of increased FliK abundance, we overexpressed FLAG-tagged FliK, FliK-C, and FliK-CΔ5 from a medium copy vector in otherwise WT cells and assessed soft agar motility and flagellin levels. Overexpression of FliK led indeed to a reproducible reduction in swim diameter to 85% compared to the vector control strain (Figure 5E), indicating that elevated levels of FliK impair motility. Interestingly, while this effect was clearly exacerbated in the strain overexpressing FliK-C (Figure 5E–F), it was absent in the strain overexpressing FliK-CΔ5, demonstrating that the C-terminus of FliK is critical for the FliK-dependent effect on motility. When analyzing flagellin protein levels in the different overexpression strains, we found that the motility defects of the FliK and FliK-C overexpression strains correlated with a significant downregulation of flagellin levels to 55% and 30%, respectively (Figure 5G). Conversely, overexpression of the FliK-CΔ5 affected flagellin levels only mildly (Figure 5G).
Taken together, our data indicate that an oversupply of FliK, which can either be caused by overexpression or absence of Lon-dependent degradation, leads to reduced motility and flagellin levels. This suggests that FliK likely contributes to the motility defects of Δlon cells along with SciP and potentially other Lon substrates affecting flagella function that are stabilized and upregulated in the absence of Lon. Furthermore, our data revealed that the C-terminal portion of FliK is critical not only for FliK degradation, but also for its effects on motility and flagellin protein levels.
In all cells, the concentrations of specific proteins must be precisely regulated to maintain cellular functions and to orchestrate complex cellular behaviors in response to external and internal cues. This study uncovered a novel role of the highly conserved protease Lon in coordinating cell differentiation with cell cycle processes in the dimorphic bacterium C. crescentus. In this bacterium, each cell cycle phase is coupled to a distinct morphological state (Curtis and Brun, 2010). This coupling of cell differentiation with core cell cycle events requires sophisticated mechanisms that coordinate these processes in space and time. Previous work established that in C. crescentus large sets of genes are transcriptionally regulated in a cell cycle-dependent manner (Laub et al., 2000; McGrath et al., 2007; Schrader et al., 2016). Using a proteomics approach, we found that several proteins encoded by cell cycle-regulated genes are Lon substrates. The identified Lon substrates include important regulators and structural components required for C. crescentus development and cell cycle progression. Our results show that active proteolysis of at least some of these proteins by Lon is required to rapidly clear these proteins following a cell cycle-dependent decrease in their transcription, thus restricting their accumulation to the cell cycle phase when their function is needed (Figure 6). The abundance of Lon does not change during the cell cycle (Wright et al., 1996) and a recent study suggested that the catalytic activity of Lon is cell cycle independent (Zhou et al., 2019). However, it is possible that the degradation of Lon-dependent degradation is affected by the accessibility or functional state of Lon substrates. For example, previous work showed that degradation of both SciP and CcrM is modulated by DNA binding (Gora et al., 2013; Zhou et al., 2019), and in the case of the AAA+ ATPase DnaA, ATP binding seems to increase protein stability (Liu et al., 2016; Wargachuk and Marczynski, 2015). Future work will show if similar mechanisms modulate the degradation of the herein identified Lon substrates.
We focused our studies on the stalk regulator StaR and the flagella regulator FliK, which were among the proteins whose abundance and stability were most strongly affected by Lon. Like DnaA, CcrM, and SciP, the transcriptional regulator StaR is a DNA-binding protein. It was initially identified as a positive regulator of stalk biogenesis (Biondi et al., 2006) and was later shown to regulate holdfast development by directly inhibiting the expression of the holdfast inhibitor HfiA (Fiebig et al., 2014); holdfast is a polysaccharide-rich adhesin that is produced at the nascent stalked cell pole in late swarmer cells and allows C. crescentus to attach to surfaces (Curtis and Brun, 2010). Stalk biogenesis and surface attachment must be tightly regulated during the cell cycle, in particular, under environmental conditions (Fiebig et al., 2014), and our work revealed that Lon-mediated proteolysis contributes to this by regulating the stability of StaR and likely other proteins involved in this process, such as the histidine phosphotransferase ShpA and the polysaccharide biosynthesis protein CCNA_02361 that we identified as putative Lon substrates in our proteomics screen (Figure 1F). The gene encoding CCNA_02361 (CC_2278), which was previously shown to contribute to surface attachment (Sprecher et al., 2017), shows a similar cell cycle-dependent pattern in mRNA levels and ribosome occupancy as staR (Lasker et al., 2016; Schrader et al., 2016), with the highest mRNA abundance and translation rate in the swarmer state (Figure 1F). Thus, Lon may also contribute to temporally regulating the abundance of this protein during the cell cycle.
In addition to proteins involved in stalk biogenesis and surface attachment, we identified several proteins required for flagella-mediated motility and chemotaxis as potential Lon substrates (Figure 1F), and investigated FliK in detail. FliK proteins are thought to function as molecular rulers that, while being exported themselves, precisely measure flagellar hook length via the N-terminal domain and mediate via the C-terminal domain a switch from export of hook protein to filament protein, once the hook has reached a certain length (Erhardt et al., 2011; Minamino, 2018; Minamino et al., 1999; Moriya et al., 2006; Shibata et al., 2007). Although previous work made significant progress in understanding the molecular function of this important protein in enterobacteria, proteolytic control mechanisms contributing to FliK regulation have so far not been described. Moreover, in bacteria outside the gamma-proteobacteria FliK remains poorly studied. In C. crescentus, a fliK gene has previously not been annotated, and our new data revealed that fliK corresponds to a gene that was previously annotated as two separate but overlapping genes. We demonstrate that Lon-mediated proteolysis ensures cell cycle-dependent accumulation of FliK during late S-phase, when the cell prepares for cell division by building a new flagellum and chemotaxis apparatus at the swarmer pole (Figure 6). The results that overexpression of FliK results in reduced flagellin levels and impaired motility (Figure 5E and G) indicate that this precise regulation of FliK abundance is critical for correct flagella biosynthesis. Based on studies in other bacteria (Muramoto et al., 1998), we consider it possible that an excess of FliK protein causes a premature termination of hook synthesis, leading to shorter hooks, which may cause downstream effects, namely reduced flagellin levels and motility. Our observation that overexpression of FliK-C, a truncated version of FliK lacking the N-terminal domain, which is required for export in other bacteria (Hirano et al., 2005), exhibits an even stronger effect on flagellin levels and motility may be explained by a dominant-negative effect that the truncated version of FliK has over the WT. In this scenario, overexpressed FliK-C would block the normal function of native FliK, possibly by occupying binding sites on interacting proteins, for example, the export apparatus protein FlhB (Kinoshita et al., 2017; Minamino et al., 2009). Our result that deletion of the C-terminal five amino acids of Caulobacter FliK-C abolishes the negative overexpression effects (Figure 5F and G) is consistent with the finding that FliK’s export switch-inducing function depends on its C-terminal five amino acids in other bacteria (Kinoshita et al., 2017; Minamino et al., 2006; Williams et al., 1996). Importantly, we also uncovered a critical role of the C-terminal five amino acids of FliK for proteolytic control (Figure 4G), thus, our data indicate a tight coupling between FliK function and degradation.
Interestingly, in addition to the flagella hook length regulator FliK, our data indicate that the flagella hook protein FlgE itself is regulated by Lon in C. crescentus. FlgE was among the proteins that satisfied two criteria in our proteomics approach, and we verified by Western blot analysis that FlgE degradation partly depends on Lon (Figure 6—figure supplement 1). This finding is also consistent with previous work showing that Lon degrades FlgE in E. coli (Arends et al., 2018), and reinforces the previously made notion that precise regulation of the cellular concentrations of FlgE as well as FliK is an important requirement for the correct temporal order of flagella assembly (Inoue et al., 2018).
In several other bacteria, including important human pathogens, Lon has been linked to flagella-mediated motility (Ching et al., 2019; Clemmer and Rather, 2008; Fuchs et al., 2001; Rogers et al., 2016). In many of these cases the nature of substrates mediating the observed Lon-dependent effects on motility remains unknown. However, in Bacillus subtilis, it was shown that Lon specifically degrades the master regulator of flagellar biosynthesis SwrA by a mechanism requiring the adaptor SmiA (Mukherjee et al., 2015). In contrast to SwrA, both StaR and FliK are robustly degraded by Lon in vitro (Figures 2D and 4F, Figure 4—figure supplement 2). While these data suggest that no adaptors or other accessory proteins are required to mediate the interaction between these proteins and Lon, it is possible that additional factors exist that modulate the rate of Lon-mediated proteolysis of these proteins in response to specific conditions.
In conclusion, our work provides new insights into the cellular roles of Lon and emphasizes the importance of proteolysis in adjusting the amounts of regulatory proteins involved in critical cellular processes, including cell differentiation and cell cycle progression. Importantly, in addition to StaR and FliK our work identified many other proteins as putative Lon substrates and the precise role of Lon in the regulation of these proteins will be worthwhile to investigate in detail in future studies. Furthermore, our work highlights quantitative proteomics using isobaric mass tags as a powerful approach for the identification of novel protease substrates that could be exploited to identify candidate substrates under diverse growth conditions, in different species or of other proteases.
All bacterial strains, plasmids, and primers used in this study are listed in Supplementary file 1.
Plasmids used for protein expression are based on the pSUMO-YHRC backbone and were constructed as follows: the coding sequences of the staR (pMF56-c88) and fliK-C (formerly motD; pMF61) genes were amplified from C. crescentus NA1000 genomic DNA with the primers listed in Supplementary file 1 (see sheets listing primers and vector fragments for sequences and primer combinations, respectively). The backbone vector pSUMO-YHRC was amplified in two parts disrupting the kanamycin resistance gene in order to reduce background (using primer pairs oMJF34/oMJF36 and oMJF37/oMJF38, Supplementary file 1). Following the PCR, the template was digested with DpnI (10 U) and the remaining PCR fragments were subsequently purified by gel extraction. Fragments were then assembled using Gibson assembly (Gibson et al., 2009). Vectors containing deletions of an annotated gene (FliK-C truncations: pMF66, pMF67-A, and pMF68-A) were derived from vectors harboring the full-length coding sequence in a similar manner using the primer pairs specified in Supplementary file 1.
pDJO145 (pBX-MCS-4 containing NdeI-3xFLAG-KpnI): plasmid pBX-MCS-4 (Thanbichler et al., 2007) was amplified using primers oDJO13 and oDJO41. The sequence encoding the triple FLAG tag (3xFLAG; GAC TAC AAA GAC CAT GAC GGT GAT TAT AAA GAT CAT GAC ATC GAC TAC AAG GAC GAC GAC GAC AAG) was amplified from a plasmid using primers oDJO42 and oDJO43 adding a KpnI-site followed by a stop codon to the 3′ end of 3xFLAG. The two amplified fragments were then joined by Gibson assembly (Gibson et al., 2009).
pDJO151 (pBX-MCS-4 containing Pxyl-staR-3xFLAG): staR was amplified with primers oDJO44 and oDJO45 using chromosomal C. crescentus NA1000 DNA as template and cloned into NdeI-cut pDJO145 using Gibson assembly.
pDJO157 (pBX-MCS-4 containing Pxyl-3xFLAG-staR): staR was amplified with primers oDJO46 and oDJO47 using chromosomal C. crescentus NA1000 DNA as template and cloned into KpnI-cut pDJO145 using Gibson assembly.
pDJO173 (pBX-MCS-4 containing Pxyl-flgE-3xFLAG): flgE was amplified with primers oDJO65 and oDJO66 using chromosomal C. crescentus NA1000 DNA as template and cloned into NdeI-cut pDJO145 using Gibson assembly.
pDJO200 (pBX-MCS-4 containing Pxyl-3xFLAG-fliK-C): fliK-C (CCNA_00945) was amplified with primers oDJO87 and oDJO88 using chromosomal C. crescentus NA1000 DNA as template and cloned into KpnI-cut pDJO145 using Gibson assembly.
pDJO410 (pBX-MCS-4 containing Pxyl-3xFLAG-fliK-CΔ5): fliK-CΔ5 was amplified with primers oDJO87 and oDJO179 using C. crescentus NA1000 DNA as template and cloned into KpnI-cut pDJO145 using Gibson assembly.
pDJO487 (pBX-MCS-4 containing Pxyl-3xFLAG-fliK): fliK was amplified with primers oDJO75 and oDJO88 using chromosomal C. crescentus NA1000 DNA as template and cloned into KpnI-cut pDJO145 using Gibson assembly.
To generate the ΔstaR Δlon strain (KJ1037), the ΔstaR deletion was introduced into the Δlon strain (KJ546) by two-step recombination (Skerker et al., 2005) after transformation with plasmid pNTPS138-ΔstaR (pAF491; Fiebig et al., 2014). Briefly, transformants were selected on kanamycin plates, single colonies were grown overnight in PYE and plated on PYE containing sucrose. Single sucrose-resistant colonies were subsequently screened for kanamycin sensitivity and the staR knockout was confirmed by colony PCR using primers oDJO40 and oDJO38.
C. crescentus strains carrying replicating plasmids were created by transforming the plasmids into the respective strain backgrounds by electroporation.
C. crescentus strains were routinely grown at 30°C in PYE medium while shaking at 200 rpm and, if necessary, regularly diluted to assure growth in the exponential phase. If required, the medium was supplemented with xylose (0.3% final), glucose (0.2% final), or vanillate (500 µM final). Antibiotics were used at following concentration (liquid/solid media): gentamycin 0.625/5 µg/ml, chloramphenicol 1/1 µg/ml, and oxytetracycline 1/2 µg/ml. Experiments were generally performed in the absence of antibiotics when using strains in which the resistance cassette was integrated into the chromosome. For phosphate starvation experiments, log-phase cells grown in M2G (minimal medium with 0.2% glucose) were washed and transferred to M5G medium lacking phosphate.
E. coli strains for cloning purposes were grown in LB medium at 37°C, supplemented with antibiotics at following concentrations (liquid/solid media): chloramphenicol 20/40 µg/ml, gentamycin 15/20 µg/ml, kanamycin 30/50µg/ml, and oxytetracyclin 12/12µg/ml.
To synchronize C. crescentus cultures, cells were pelleted by centrifugation at 8000 rpm for 4 min at 4°C. The supernatant was aspirated, and tubes were kept on ice. Pellets were resuspended in 1 ml of 1× M2 salts on ice. 1 ml of cold Percoll was added and samples were mixed well. The mixture was aliquoted into two Eppendorf tubes and centrifuged at 10,000×g for 20 min at 4°C. The top layer of the cells was aspirated, and the swarmer cells were moved into a new tube. Swarmer cells were washed twice with 1 ml of cold 1x M2 salts and finally resuspended in 20 ml prewarmed PYE medium containing antibiotic if required. Samples were taken immediately after resuspending the cell pellet and subsequently at the indicated time points for immunoblot analysis.
For whole-cell extract analysis, 1 ml culture samples were collected after the indicated treatments and time points, and cell pellets were obtained by centrifugation. Cell pellets were resuspended in appropriate amounts of 1× SDS sample buffer, to ensure normalization of the samples by units OD600 of the cultures. Samples were boiled at 98°C for 10 min and frozen at –20°C until further use. The thawed samples were separated by SDS-PAGE using TGX Stain-free gels (Bio-Rad), and subsequently transferred to nitrocellulose membranes by a semi-dry blotting procedure as per the manufacturer’s guidelines. The protein gels and membranes were imaged using a Gel Doc Imager before and after the transfer, respectively, to assess equal loading of total protein as well as the quality of the transfer.
Membranes were blocked in 10% skim milk powder in TBS-Tween (TBST) and protein levels were detected using the following primary antibodies and dilutions in 3% skim milk powder in TBST: anti-CcrM 1:5000 (Stephens et al., 1996), anti-DnaA 1:5000 (Jonas et al., 2011), anti-Lon 1:10,000 (kindly provided by R.T. Sauer), anti-FLAG M2 antibody 1:5000 (Sigma-Aldrich), anti-SciP 1:2000 (Gora et al., 2010), anti-flagellin 1:2000 (kindly provided by Y. Brun) (Brun and Shapiro, 1992), anti-FliK-C 1:500, anti-StaR 1:500 (Fiebig et al., 2014), and anti-TipF 1:5,000 (kindly provided by P. Viollier) (Davis et al., 2013). Secondary antibodies, 1:5000 dilutions of anti‐rabbit or anti-mouse HRP‐conjugated antibodies (Thermo Fisher Scientific), and SuperSignal Femto West (Thermo Fisher Scientific) were used to detect primary antibody binding. Immunoblots were scanned using a Chemidoc (Bio‐Rad) system or an LI-COR Odyssey Fc system. Relative signal intensities were quantified using the Image Lab software package (Bio‐Rad) or ImageJ software.
To assess protein degradation in vivo, cells were grown under the appropriate conditions (e.g., for 1–2 hr in the presence of xylose to induce expression of FLAG-tagged proteins), and subsequently protein synthesis was shut off by addition of chloramphenicol (100 μg/ml) or tetracycline (10 μg/ml). Samples were taken at the indicated time points and snap frozen in liquid nitrogen before preparation for analysis by Western blot analysis.
Sample preparation was performed as previously described (Schramm et al., 2017). In brief, two independent cultures for each analyzed condition were harvested by centrifugation. Cell pellets were washed using cold ddH2O and stored at –80°C. Protein digestion, TMT10 plex isobaric labeling, and the mass spectrometrical analysis were performed by the Clinical Proteomics Mass Spectrometry Facility, Karolinska Institute/Karolinska University Hospital/Science for Life Laboratory.
To identify putative Lon substrates, first, the protein abundances for each condition within one biological replicate were considered to calculate the following ratios (see Figure 1—source data 1): protein levels in Δlon at t=0 min/protein levels in WT at t=0 min (Δlon 0 /WT 0) to identify proteins with a higher steady-state level in the absence of Lon; protein levels after Lon depletion/protein levels before Lon depletion (Lon depletion (no van)/lon expression (+van)) to identify differences in protein levels after Lon depletion; protein levels before lon overexpression in the presence of glucose/protein levels after 1 hr xylose-induced lon overexpression (gluc/+xyl 1 hr lon overexpression), to identify proteins that are downregulated by lon overexpression; protein levels in WT at t=0 min/protein levels in WT at t=30 min (WT 0 /WT 30), to identify proteins that are degraded in WT cells after protein synthesis shut off. Additionally, we calculated the ratios of protein levels in Δlon cells at t=0 min/protein levels in Δlon cells at t=30 min after protein synthesis shut off (Δlon 0/Δlon 30). Subsequently, the average of the ratios obtained from the two biological replicates was calculated and used for further analysis. Some proteins, as indicated in Figure 1—source data 1, were detected in only one of the replicate data sets. In these cases, only the ratio from the replicate, in which they were detected, was considered.
Next, we selected all proteins with a ratio of (Δlon 0 /WT 0) ≥1.1 for the group ‘higher levels in Δlon’. Similarly, we selected all proteins with a ratio of (Lon depletion (no van)/lon expression (+van)) ≥1.1 for the group ‘higher levels after Lon depletion’. For the group ‘lower levels after lon overexpression’, we selected all proteins with a ratio of (gluc/+xyl 1 hr lon overexpression) ≥1.05. For the group ‘stabilized in Δlon’, we first selected all proteins with a ratio of (WT 0 /WT 30) ≥1. For those, we calculated the ratio of (WT 0/WT 30)/(Δlon 0/Δlon 30) and chose the proteins with a ratio ≥1.05. The low threshold ratios were chosen due to a low dynamic range of the data and to ensure that any promising Lon substrate candidates were not missed. To eliminate false positives that may have passed these individual thresholds, an additional filter was applied that at least three of the threshold ratios needed to be met in order to consider a protein a putative Lon substrate. For this, overlaps between the groups of proteins passing the individual thresholds were determined and graphically displayed using jvenn (Bardou et al., 2014; Figure 1D). Functional categories were assigned to the group of putative Lon substrates as well as all detected proteins as previously described (Schramm et al., 2017).
Protein purification was adapted from Holmberg et al., 2014. In brief, BL21-SI/pCodonPlus cells were transformed using a pSUMO-YHRC derived vector by electroporation. Transformants were selected on LB agar plates lacking NaCl (LBON) supplemented with kanamycin and chloramphenicol and pre-cultures inoculated with 20 colonies before being cultivated at 30°C overnight. Pre-cultures were diluted 1:100 in 1 L 2xYTON and grown to OD600 1.0–1.5. Protein expression was induced with 0.5 mM IPTG and 0.2 M NaCl either for 4.5 hr at either 30°C (StaR and FliK-C) or overnight at 20°C (in case of FliK-C truncations). After centrifugation, cell pellets were stored dry at −80°C.
Pellets were resuspended in HK500MG (40 mM HEPES-KOH pH 7.5, 500 mM KCl, 5 mM MgCl, and 5% Glycerol) supplemented with 1 mM PMSF, 1 mg/ml Lysozyme, and 3 µl Benzonase/10 ml suspension and topped up to 20 ml. Cells were then lysed by 2–4 passes through an EmulsiFlex-C3 high-pressure homogenizer. Lysate was cleared by centrifugation at 32,500×g at 4°C for 0.5–1 hr. The protein of interest was bound to 1 g Protino Ni-IDA beads at 4°C/on ice for 30 min. After washing five times with approx. 45 ml HK500MG protein was eluted using HK500MG + 250 mM Imidazole and fractions with protein concentrations ≥1 mg/ml were pooled. For 6xHis-SUMO tag removal, 4 µg/ml Ulp1-6xHis was added and imidazole was removed in parallel by dialysis against HK500MG. Tag depletion (except for StaR) was achieved by binding to 1 g Protino Ni-IDA beads as before and flow through containing purified protein was collected. Protein concentration was then checked via SDS-PAGE (Bio-Rad 4–20% Mini-PROTEAN TGX Stain-Free Protein Gel) and InstantBlue Protein Stain (Expedeon) and quantified using Bio-Rad ImageLab 6.0.1. Protein samples were aliquoted and stored at −80°C.
Because StaR was not soluble in neither of the tested buffer conditions and formed precipitates, the protein was refolded (adapted from De Bernardez Clark et al., 1999; Thomson et al., 2012). The precipitate was collected (centrifugation at 7197×g, 4°C) and solubilized in buffer S (50 mM HEPES pH 8.0, 6 M guanidinium-HCl, 1mM EDTA, and 10 mM DTT) and the protein concentration was adjusted to 0.2 mg/ml. The protein solution was then diluted with an equal volume dialysis buffer D1 (50 mM HEPES pH 8.0, 2 M guanidinium-HCl, and 2 mM EDTA) and dialyzed against 125 volumes dialysis buffer D1 followed by dialysis against 125 volumes dialysis buffer D2 (50 mM HEPES pH 8.0, 1 M guanidinium-HCl, 2mM EDTA, 0.4 M Sucrose, 500 mM KCl, and 2 mM DTT). The dialysis buffer was then diluted with one buffer volume of dialysis buffer D3 (50 mM HEPES pH 8.0, 2 mM EDTA, 0.4 M Sucrose, 500 mM KCl, and 2 mM DTT) and dialyzed. This was followed by a final dialysis against 125 volumes of dialysis buffer D3 to remove the remaining guanidinium-HCl. Each dialysis step was carried out at 4°C for approx. 24 hr.
Afterward, insoluble StaR molecules were removed by centrifugation (20,000×g, 4°C, 10 min) and cleared refolded StaR was supplemented with additional 2 mM DTT and concentrated using a centrifugal filter with a MWCO of 3 kDa (VWR #516-0227P) and stored at −80°C. The final concentration was determined by SDS-PAGE using a BSA standard and visualized by InstantBlue Protein Stain (Expedeon).
Purified FliK-C was used as antigen to generate rabbit polyclonal antisera (Davids Biotechnologie GmbH).
In vitro degradation assays were performed as published previously (Jonas et al., 2013). The reaction was carried out in Lon reaction buffer (25 mM Tris pH 8.0, 100 mM KCl, 10 mM MgCl2, and 1 mM DTT) employing 0.75 µM Lon (0.125 µM Lon6), 4 µM substrate (if not stated otherwise), and an ATP regeneration system (4 mM ATP, 15 mM creatine phosphate, and 75 µg/ml creatine kinase). The reaction and the ATP regeneration system were prepared separately pre-warmed to 30°C (approx. 4 min). The reaction was started by adding the ATP regeneration system. Samples were taken at indicated time points and quenched by 1 vol. 2× SDS sample buffer (120 mM Tris-Cl pH 6.8, 4% SDS, 20% glycerol, and 0.02% bromophenol blue) and snap frozen in liquid nitrogen. Samples were boiled at 65°C for 10 min and separated by SDS-Page (Bio-Rad 4–20% Mini-PROTEAN TGX Stain-Free Protein Gel), visualized by InstantBlue Protein Stain (Expedeon) and quantified using Bio-Rad ImageLab 6.0.1. Substrate levels were normalized to Lon and/or creatine kinase levels (in case of ‘–Lon’ samples).
Cells were fixed by addition of formaldehyde (1% final) to culture samples and stored at 4°C. For visualization, fixed cells were transferred onto 1% agarose pads attached to glass slides, covered with a coverslip, and transferred to the microscope. A Ti eclipse inverted research microscope (Nikon) with 100×/1.45 numerical aperture (NA) objective (Nikon) was used to collect phase-contrast images. The images were processed with Fiji (ImageJ).
To quantify the stalk length of cells, microscopy images were analyzed using ImageJ. Briefly, the scale was set to one pixel representing an equivalent of 0.0646 μm. The stalk was then manually selected and length was determined (Figure 3—figure supplement 1). For half-automated analysis (Figure 3E), the software BacStalk was used (Hartmann et al., 2020). The data files containing microscopic images were added to the program, the channel ‘phase contrast’ was selected, and the scale was set to 1 px≙0.0646 μm. Stalk detection was activated by the setting ‘My cells have stalks’. For cell detection, the default settings were used.
To assess motility, strains were grown in PYE media, supplemented with gentamycin to maintain replicating plasmids when necessary, and cultures were diluted to an OD600 of 0.1. Subsequently, 1 µl of each sample was injected about 2 mm vertically into PYE soft agar plates (0.35%), supplemented with gentamycin or gentamycin and xylose when indicated, using a pipette. The plates were incubated at 30°C and pictures were taken with the setting ‘Blots: Colorimetric’ using a ChemiDoc (Bio-Rad).
All data generated or analysed during this study are included in the manuscript and supporting files.
NCBI Gene Expression OmnibusID GSE68200. Ribosome Profiling Reveals Translational Control During the Caulobacter crescentus cell cycle.
Lon protease has multifaceted biological functions in Acinetobacter baumanniiJournal of Bacteriology 201:e00536-18.https://doi.org/10.1128/JB.00536-18
The Lon protease regulates swarming motility and virulence gene expression in Proteus mirabilisJournal of Medical Microbiology 57:931–937.https://doi.org/10.1099/jmm.0.47778-0
Bacterial proteases, untapped antimicrobial drug targetsThe Journal of Antibiotics 70:366–377.https://doi.org/10.1038/ja.2016.138
Getting in the loop: Regulation of development in Caulobacter crescentusMicrobiology and Molecular Biology Reviews 74:13–41.https://doi.org/10.1128/MMBR.00040-09
De- and repolarization mechanism of flagellar morphogenesis during a bacterial cell cycleGenes & Development 27:2049–2062.https://doi.org/10.1101/gad.222679.113
Inhibition of aggregation side reactions during in vitro protein foldingMethods in Enzymology 309:217–236.https://doi.org/10.1016/s0076-6879(99)09017-5
Regulation of the replication initiator DnaA in Caulobacter crescentusBiochimica et Biophysica Acta. Gene Regulatory Mechanisms 1862:697–705.https://doi.org/10.1016/j.bbagrm.2018.01.004
Proteolytic degradation of Escherichia coli transcription activators SoxS and MarA as the mechanism for reversing the induction of the superoxide (SoxRS) and multiple antibiotic resistance (Mar) regulonsMolecular Microbiology 51:1801–1816.https://doi.org/10.1046/j.1365-2958.2003.03952.x
Recognition of misfolded proteins by Lon, a AAA(+) proteaseGenes & Development 22:2267–2277.https://doi.org/10.1101/gad.1670908
Bacstalk: A comprehensive and interactive image analysis software tool for bacterial cell biologyMolecular Microbiology 114:140–150.https://doi.org/10.1111/mmi.14501
A versatile bacterial expression vector designed for single-step cloning of multiple DNA fragments using homologous recombinationProtein Expression and Purification 98:38–45.https://doi.org/10.1016/j.pep.2014.03.002
Regulated proteolysis in bacteria: CaulobacterAnnual Review of Genetics 50:423–445.https://doi.org/10.1146/annurev-genet-120215-035235
Caulobrowser: A systems biology resource for Caulobacter crescentusNucleic Acids Research 44:D640–D645.https://doi.org/10.1093/nar/gkv1050
ClpAP is an auxiliary protease for DnaA degradation in Caulobacter crescentusMolecular Microbiology 102:1075–1085.https://doi.org/10.1111/mmi.13537
The type III flagellar export specificity switch is dependent on FliK ruler and a molecular clockJournal of Molecular Biology 359:466–477.https://doi.org/10.1016/j.jmb.2006.03.025
Effect of cellular level of FliK on flagellar hook and filament assembly in Salmonella typhimuriumJournal of Molecular Biology 277:871–882.https://doi.org/10.1006/jmbi.1998.1659
Lon gene product of Escherichia coli is a heat-shock proteinJournal of Bacteriology 159:283–287.https://doi.org/10.1128/JB.159.1.283-287.1984
AAA+ proteases: ATP-fueled machines of protein destructionAnnual Review of Biochemistry 80:587–612.https://doi.org/10.1146/annurev-biochem-060408-172623
Fiji: an open-source platform for biological-image analysisNature Methods 9:676–682.https://doi.org/10.1038/nmeth.2019
The protein quality control network in Caulobacter crescentusFrontiers in Molecular Biosciences 8:682967.https://doi.org/10.3389/fmolb.2021.682967
FliK regulates flagellar hook length as an internal rulerMolecular Microbiology 64:1404–1415.https://doi.org/10.1111/j.1365-2958.2007.05750.x
Biological roles of the Lon ATP-dependent proteaseResearch in Microbiology 157:701–713.https://doi.org/10.1016/j.resmic.2006.05.004
The FliK protein and flagellar hook-length controlProtein Science 16:769–780.https://doi.org/10.1110/ps.072785407
Mutations in FliK and FlhB affecting flagellar hook and filament assembly in Salmonella typhimuriumJournal of Bacteriology 178:2960–2970.https://doi.org/10.1128/jb.178.10.2960-2970.1996
Caulobacter Lon protease has a critical role in cell-cycle control of DNA methylationGenes & Development 10:1532–1542.https://doi.org/10.1101/gad.10.12.1532
Sonja V AlbersReviewing Editor; University of Freiburg, Germany
David RonSenior Editor; University of Cambridge, United Kingdom
In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.
[Editors' note: this paper was reviewed by Review Commons.]
Protein turnover is a basic feature of life. The Lon protease, a component of the protein degradation machinery, is conserved in all three domains of life. Yet its protein substrates remain poorly characterised. Here the authors present a proteomic-based approach to the discovery of such substrates in the model bacterial organism Caulobacter crescentus. Two substrates have been characterised in considerable detail, highlighting Lon's role in bacterial physiology.https://doi.org/10.7554/eLife.73875.sa1
Reviewer #1 (Evidence, reproducibility and clarity (Required)):
The Lon protease is highly conserved across 3 domains of life, yet in most organisms very few substrates are known. In this study, the authors build on observations of Lon-dependent protein stability modifications to set up a proteomics-based screen for new Lon substrates in the model bacterium Caulobacter crescentus (in which only 3 substrates had been identified so far). They focus the rest of their work on validating and characterising further 2 Lon substrates, StaR (a transcriptional regulator important for stalk biogenesis) and FliK (a regulator of flagellar hook assembly). They were able to confirm that these two proteins are Lon substrates, both in vivo and in vitro (the latter showing that the effect of Lon on the levels of these proteins is not an indirect effect). They could also identify portions of these substrates that are required for Lon-mediated degradation. In the case of FliK, they found that the last 5 residues at the C-terminus of the protein are needed for Lon-dependent degradation and are also important for the effect of FliK on flagellins amounts in the cell. Along the way, they corrected a misannotation of the ORF encoding that protein. This suggest a hypothesis for how the Lon-dependent degradation of FliK might lead to flagellin misregulation, therefore affecting bacterial motility. In the case of StaR, the authors show that the stalk length defects previously observed in the ∆lon mutants are attributed to elevated levels of StaR during the cell cycle. Overall, this is a carefully conducted study and a clear manuscript. The text and figures are easy to understand and the story flows. The abstract and introduction are engaging for a broad audience.
The authors suggest the possibility that cell-cycle-dependent Lon-mediated degradation of Lon substrates might depend on their functional state, such as their DNA-bound state. This is a very attractive hypothesis considering that the 3 previously identified Lon substrates in Caulobacter are DNA-binding proteins, as well as StaR that they identified here as another Lon substrate. In the discussion, the authors also mention the possibility of accessory proteins to regulate Lon activity. However, I miss the information about Lon protein levels during a (synchronized) cell cycle (either a ref to published data or a new experiment), as this might be constitute another factor for cell cycle regulation and contribute to understanding how Lon can impact protein levels of its substrates in a cell-cycle dependent manner.
According to a previous study by Wright et al. 1996, Lon levels do not change in a cell cycle-dependent manner. As the reviewer suggested, we have added this information to the discussion in l. 329-335.
Related to this, the authors refer to a paper showing that Lon catalytic activity does not seem to change during the cell cycle, in the discussion part, but this could be developed a bit and presented earlier in the introduction.
In this mentioned study by Zhou et al. 2019, it was shown that Lon-dependent degradation of a constitutively expressed model substrate was not affected by the cell cycle phase, which led to the conclusion that Lon activity is cell cycle independent. We have modified the section about this previous work and also added a sentence to mention that future work will be dedicated to mechanisms regulating the degradation of the newly identified substrates. Given that our study was primarily focused on the identification of new Lon substrates and their contributions to Lon-dependent phenotypes, we think this information is better placed in the discussion than in the introduction.
Finally, in their legend of the model figure (Figure 6), the authors suggest that the Lon protease acts on its substrates when their expression level drops. Do the authors imply some sort of stoichiometry at play between Lon and its substrates that would determine when Lon will have an overall impact on their protein levels, or do they consider a more complex mechanism? It would be interesting to discuss this a bit more (I understand that going in this direction experimentally would be another story, and not required to support the data shown here).
Based on the reviewer’s comment, we realized that the figure legend might have been misleading. We do not think that Lon only acts on these substrates when their expression levels drop, but rather that Lon-mediated proteolysis in combination with regulated transcription ensures rapid clearance of the proteins. We have modified the text of this figure legend to enhance clarify (l. 1092).
- Figure 1E: Optional: It might be informative to add the level of enrichment of these categories compared to the whole proteome content.
We thank the reviewer for this suggestion. In the new version of our manuscript, we included, for comparison, a graph to Figure 1E that shows how all detected proteins sort to the different functional categories. To fit both graphs into the main figure, we replaced the pie chart with stacked bar graphs. We have also modified the text (l. 125-130) to emphasize the comparison of protein factions between whole proteome content and the pool of Lon substrates.
– Line 193: the ∆staR∆lon mutant phenocopies the ∆staR single mutant regarding stalk length. I would suggest to add the result of the ANOVA for this comparison on Figure 3 (as done in the related Figure 3—figure supplement 1 showing non-significant difference between these mutants).
As suggested by the reviewer, we have added the results of the statistical tests for the remaining comparisons to the figure and figure legend.
– Figure 4F: I could not find the explanation for the asterisk next to the band below FliK-C. Please add it in the legend.
Thanks for pointing this out. We have added the explanation for the asterisk to the figure legend.
– Figure 4F: did the authors test Lon-dependent degradation in vitro for the FliK-C∆5 mutant as well? They show degradation in vivo only (Figure 4G).
We did not test the degradation of the FliK-C∆5 mutant in vitro. However, based on our in vivo stability data of this mutant and our in vitro data of the FliK-CΔC44 mutant (which shows complete stabilization), we expect this mutant to be stable.
– Line 286: at this point, I would consider that the authors have shown misregulation of flagellin protein levels, not flagellin expression (transcriptional).
We agree with the reviewer that it is better to refer to “flagellin protein levels” rather than “flagellin expression” and changed the wording accordingly.
– Figure 5D: results from the three different lon mutant strains seem contradicting, as Lon deletion and Lon depletion have opposite effects on flagellins protein levels, and Lon overproduction gives results similar to Lon depletion. Could it be that -van and +van labels were inadvertently swapped?
If yes, please correct.
While the authors appropriately use the term "misregulated" to describe the flagellins in the various lon mutants, the absence of comment on this discrepancy at that place in the text might bring confusion about their interpretation of these results. Adding a clarification here might also help introducing more clearly the experiments that come next.
The Lon-dependent effect on flagellin levels is indeed not apparent in the Lon depletion strain and only to a lesser extent in the Lon overexpression strain. Although we cannot fully explain this difference between the ∆lon strain and the conditional Lon strains at this point (see also our response to reviewer 4), we still think that the flagellin data, along with the motility data, support an indirect involvement of Lon in flagella regulation. As the reviewer suggested, we have described the flagellin data more explicitly in the text. Based on the comments by reviewers 3 and 4, we have also restructured this section to enhance clarity.
Related to this, at lines 303-304 (Figure 5F-G), the authors conclude that the reduced flagellin levels and motility defects of ∆lon can at least in part be attributed to the stabilisation of FliK in these cells. Here as well, it might be worth giving a few words on why Lon depletion data (Figure 5D) do not fit with this idea.
In our new version we have improved the section describing the link between FliK, Lon and motility (see comments below). With these changes we hope that we satisfy this point by reviewer 1.
– Figure 5E: please clarify whether a specific a-flagellin was monitored here or all of them.
The antiserum that we used detects all flagellins. We have modified the text and added a reference to indicate the source of this antibody.
– Figure 5E: how do the authors explain that FliK-C overproduction has a stronger effect on a-flagellin levels than the full-length FliK (as both are supposed to be able to interact with FlhB – cfr their hypothesis for the importance of the last 5 aa at lines 292-294).
It is indeed striking that overexpression of FliK-C results in a stronger phenotype than overexpression of the full length FliK protein. We consider it possible that the truncated version of FliK has a dominant negative effect over the wild type. In this scenario, overexpressed FliK-C would block the normal function of native FliK, possibly by occupying binding sites on interacting proteins, such as the export apparatus protein FlhB (Kinoshita et al. 2017). We have added a paragraph to the discussion in which we discuss this possibility (l. 371-383)
– Methods: could the authors explain how they chose the 1.1 or 1.05 ratios (lines 526-534), which may seem quite low?
The main reason for these low thresholds was the low dynamic range of our proteomics data that is due to technical aspects of the TMT-labeling approach as well as our experimental conditions. Even the well-studied substrates, DnaA, SciP and CcrM, that we used as positive controls, showed only small fold changes (1.1-1.7 fold) in the different strain comparisons. Because we did not want to miss promising Lon substrates when setting our filters for the individual conditions, we decided to choose the low thresholds of 1.1 in the lon deletion and depletion experiments and 1.05 in the stability assay and overexpression experiment (in these latter experiments the dynamic range was even smaller due to the early time points that we used).
Importantly, to filter out the false positives that may have passed these low thresholds in the individual conditions, we applied the additional requirement that at least three of the thresholds needed to be met in order to consider a protein a putative Lon substrate. With this multi-layered filtering approach, we were able to identify 146 proteins that included the known substrates DnaA, SciP and CcrM as well as the validated new substrates StaR and MotD/CCNA_00944. Based on these results we are confident that our group of putative Lon substrates contains a high rate of true Lon substrates. Nevertheless, we think it is important to perform follow-up experiments with the candidates generated in our study. For this reason, we were careful to call unvalidated hits “putative Lon substrates”. Furthermore, we have presented our proteomics data in a transparent and comprehensive way in dataset 1 that will allow other researchers to carefully evaluate Lon-dependent effects on other proteins of interest.
Based on this comment and the comments by reviewer 2 we have added some more explanation and details to the description of our proteomics data analysis in the Methods section.
Reviewer #1 (Significance (Required)):
The Lon protease is highly conserved across 3 domains of life, yet in most organisms very few substrates are known. This work advances the fields of protein homeostasis / bacterial cell biology by identifying new substrates of Lon, using the bacterium Caulobacter crescentus (a model for bacterial cell cycle control, development and differentiation) using an interesting proteomic approach. Two of these substrates are characterized further, StaR and FliK. This work certainly highlights yet another layer of sophistication in the complex cell cycle control mechanisms in Caulobacter crescentus. While here the authors present Lon as an important factor for regulating/coordinating cell cycle events, how exactly Lon-mediated degradation can be cell-cycle regulated in Caulobacter crescentus (at least for the substrates characterized in this study) is an important question that future studies will need to address.
In my opinion, the audience will include researchers involved in protein homeostasis (whether working directly on Lon or on other protein degradation/stability systems), bacteriologists (including the broad field of cell cycle regulation), and of course more specifically all researchers working on Caulobacter crescentus.
My field of expertise is bacterial cell biology in general, including bacterial cell cycle control, intracellular processes, live microscopy imaging and protein localisation in Caulobacter crescentus and other species. I do not have sufficient expertise to evaluate the methodological details of the identification and quantification of substrates by mass-spectrometry.
Reviewer #2 (Evidence, reproducibility and clarity (Required)):
Omnus et al. utilized a quantitative proteomics approach to globally identify novel substrates of Lon protease in the bacterium Caulobacter crescentus. Among candidate protease substrates revealed by large proteomic datasets, the authors focused on the developmental regulator StaR and the flagella hook length regulator FliK as direct Lon substrates. With a number of elegant follow-up experiments, their findings established a critical role of Lon in coordinating developmental processes with cell cycle progression. Overall, this work is carefully designed and well performed. The reviewer would recommend its publication upon addressing a few minor issues, which are mostly associated with the quantitative analysis of proteomics data.
1. Typically three biological replicates would be needed for proteomic experiments, thus permitting statistical analyses of differentially regulated proteins.
We are aware that two biological replicates are not sufficient to do sophisticated statistical analyses. Therefore, we did not perform a more in-depth analysis that would provide statistical significances on the observed changes. Instead, we treated the proteomics results as a list of potential substrate candidates that provide the basis for rigorous further investigation and validation.
2. Line 524, some proteins are only detected in one replicate and should not be used for quantitative analyses. Given that only two replicates were considered, it is important to stay on the safe side and control the rate of false positives.
We agree that it is important to exclude false positives, and our approach to consider only proteins as putative Lon substrates that satisfied a combination of different criteria helped us certainly to do so. Our decision to include also proteins that were detected in only one of the replicates was taken not to miss any promising Lon substrate candidates. Note that for example CCNA_00945, i.e., MotD, which turned out to correspond to FliK-C, one of the validated new Lon substrates, was detected only in one of the replicates. We also want to stress that one replicate in the proteomics analysis included in fact data points for three independent strain comparisons (1. WT/∆lon stability and steady state levels, 2. Lon depletion, depleted vs. non-depleted, 3. Lon overexpression, induced vs. uninduced). Comparing the results from these different experiments against each other gave us good indications regarding the validity of our data.
3. Line 526, the cutoff of the protein ratio is 1.1 (or 1.05 in some cases), which is way too low to tease out the altered proteins……The author should apply the threshold of 1.5 at least.
Please see our response to the last comment by Reviewer #1, in which we have explained why we use these low threshold ratios in our data analysis. Applying a threshold ratio of 1.5 would exclude essentially all proteins from our list of putative Lon substrates including the known substrates DnaA, SciP, and CcrM and the validated substrates StaR, MotD, CCNA_00944. Hence, a threshold of 1.5 was not appropriate for our dataset.
4. When reporting protein fold changes (in the supplemental dataset), two decimals are sufficient. The authors used eight decimals, which do not make any sense.
We agree and have changed the column type to “Number” and set the amount of shown decimals to two decimal places, in order to preserve the full numbers while increasing readability.
Reviewer #2 (Significance (Required)):
The findings should add significantly to our growing knowledge about the Lon protease, as well as the regulation of critical biological processes by intracellular proteolysis.
Reviewer #3 (Evidence, reproducibility and clarity (Required)):
In this work, the authors use a combination of proteomics to identify novel Lon substrates in the bacterium Caulobacter crescentus. They focus their search on substrates involved in cell differentiation and identified two new Lon substrates, StaR and FliK, which they show are degraded in a Lon-dependent manner in vivo and are directly proteolyzed by Lon in vitro as well. The authors go on to show that these substrates seem to be recognized by Lon from their C-termini and that Lon-mediated degradation of these substrates allows them to oscillate during the cell cycle. Finally, the authors link misregulation of these substrates in cells lacking Lon with defects in stalk biogenesis and motility. The authors also take the extra effort to clarify the annotation of the fliK gene, which is appreciated. Overall, this is a very strong manuscript, with some concerns as raised below.
1. The authors should explicitly state on line 188-191 that overexpression of StaR was shown to increase stalk length (Biondi, et al. 2006) as this is a key premise for their model that stabilization of StaR explains in the increased stalks of ∆lon. As written, it is unclear that StaR has been previously shown to be a positive inducer of stalk length.
We completely agree with this point and have modified the text to mention more explicitly that staR overexpression causes stalk elongation and that our experiment is based on this finding (l. 191).
On a related note, the implication of the model is that o/exp of StaR in a ∆lon would not increase stalk length. Is that true?
According to our model, Lon regulates stalk length via StaR, which makes it unlikely that lon deletion would diminish the effect of StaR overexpression. Rather, we consider it possible that staR overexpression in ∆lon cells would lead to even stronger StaR accumulation, which in turn might be reflected in even longer stalks. Because we think that this experiment is not really needed to support the main conclusions of our paper, we have not included it in our study.
2. The authors show convincing data that FliK is a Lon substrate. However, the data tying FliK stabilization to motility defects in a Δlon strain is not as compelling, especially as motility in this media is a combination of cell growth, division, and actual motility. Ideally, the authors could show that flagellar function itself is affected by using chemotaxis assays or something similar. If this cannot be experimentally addressed, then there should be a clear discussion as to the reservations with interpreting these swimming assays in light of the combination effects that change growth.
We agree with the reviewer that the swim diameters in low percentage agar are likely to be affected by changes in growth rate and cell division rate. For this reason, we decided to also include the quantification of flagellin levels in the different strain backgrounds, as we thought that the observed misregulation of flagellin levels further indicates an indirect involvement of Lon in flagella regulation. However, we agree that the limitations of the motility assays should be mentioned in our manuscript and have added a sentence in l. 283 to indicate that growth and division defects may contribute to reduced motility.
3. Why did the authors not determine if deletion of fliK could reverse ∆lon motility defects similar to that shown for StaR?
Studies in other bacteria have shown that FliK is an essential regulator of motility that is critical for proper hook synthesis and filament protein secretion. The deletion of fliK or absence of functional FliK leads to elongated hooks without attached filament structure, i.e., to an immotile polyhook phenotype (e.g., Muramoto et al. 1998, Eggenhofer et al. 2006). We consider it likely that deletion of fliK in Caulobacter will have a similar effect. Therefore, we do not expect to see a reversion of the motility defects when knocking out flik in a ∆lon strain.
To studying the link between Lon and FliK, we decided to analyze the consequences of FliK overexpression in otherwise wild type cells, thus mimicking the stabilization and higher abundance of FliK that occurs in ∆lon cells, while eliminating the effect of Lon on other substrates. Our data indicate that increased FliK abundance has a negative effect on motility and highlights the importance of precisely regulating FliK levels.
The case of StaR is different: though it is a positive regulator too, it is not essential for stalk biogenesis. The stalks of ∆staR cells are shorter than WT stalks but they are not completely absent.
4. We found it surprising that there was not more description of how stabilization of SciP (a known Lon substrate) could be contributing to the flagella phenotypes, given that it seems to directly repress expression of many flagella genes. It seems likely that overabundance of SciP underlies why FliK stabilization might only partially explain the ∆lon motility defects.
We agree with the reviewer that stabilization of SciP in ∆lon cells likely contributes to the motility phenotype. As the reviewer suggested, we discuss this idea more explicitly in the new version of our manuscript (l. 289-292).
5. In Figure 5C, the quantification of the ∆lon swim diameter should also be shown, so that it can be compared with those diameters shown in F and G, especially as the ∆lon swim diameter seems much smaller.
We have added the quantification including statistical analysis of the relative swim diameters to Figure 5C and modified the figure legend accordingly.
There also should be some statistical treatment of the measurements in F and G to show they are different.
We have modified the graphs of Figure 5E and F (formerly Figure F and G) to include the results of the statistical analysis (see also figure legends). For consistency, we have also added statistical analysis of the data displayed in Figure 5G (formerly Figure 5E).
Finally, why were the panels in F and G taken at different days of growth?
The experiments shown in panels 5E and F (formerly Figure 5F and G) were not conducted side by side, hence we happened to photograph the plates after slightly different incubation times (5 or 6 days of incubation at 30°C). Both experiments include a vector control used for normalization, and the relevant experimental samples to allow comparisons and thus can stand for themselves.
6. We found the connection to α and β flagellin levels in Figure 5 confusing. The authors reason that FliK overabundance reduces flagellin levels but in Figure 5D, the authors show that while flagellin levels decrease in a Δlon strain (which is likely also due to SciP stabilization as described above) this is not the case for the Lon depletion strain which would presumably have elevated FliK levels as well. At a minimum, this should be pointed out and discussed in the text.
As explained in our responses to reviewer 1, we have improved and restructured the text describing the data shown in Figure 5D to better explain the link between FliK, flagellins, Lon and motility.
7. It is not clear from the methods how loading of the westerns was controlled. This is particularly important given the reliance on quantification for the in vivo degradation measurements.
We verified equal loading and the quality of transfer by visualizing total protein using TGX Stain-free protein gels (Bio-Rad) prior to and after blotting. We have added this information to the description of immunoblotting in the Materials and methods part of the manuscript.
Reviewer #3 (Significance (Required)):
This work reveals the substrate profile of an important protease. It demonstrates both the caveats and advantages of a multi-proteomic strategy. One major addition is the identification of the StaR degradation, which explains some phenotypes of the ∆lon mutant. Overall, this work will be highly valuable to those interested in post translational control in bacteria.
Reviewer #4 (Evidence, reproducibility and clarity (Required)):
Using a proteomics approach the authors find that the Lon protease regulates the stability of hundred proteins including the StaR stalk transcriptional regulator and the FliK flagellar protein. They show that the degradation of the two proteins by Lon is direct using in vitro assays.
The manuscript is clearly written and the data is solid and the experiments are well executed. The data is displayed clearly, sometimes in more than one (redundant) panel. There might be problem with Fig5D, middle panel (see below).
The major shortcoming is that this work lacks conceptual novelty: it does not advance the field substantially and it does not establish (convincing) causality between Lon, motility, FliK and the flagellins.
We want to emphasize that our study includes a comprehensive search for Lon substrates in one of the primary bacterial model organisms. This search yielded a list of more than 100 Lon substrate candidates, of which we have characterized StaR and FliK as Lon substrates in detail. We were able to demonstrate that Lon-dependent degradation is critical for the cell cycle-dependent regulation of these proteins and for proper control of stalk length and motility. We think our findings will appeal to a wide readership, including scientists studying proteolysis, cell cycle regulation as well as bacterial motility, and reviewers 1-3 seem to agree with this.
We admit that the part regarding FliK, Lon, flagellins and motility might have been somewhat unclear in the first version of our manuscript, which may have caused confusion. To improve our paper, we have made some larger text changes to this section and have also added a new paragraph to the discussion, in which we explain the FliK and FliK-C overexpression effects. Finally, we believe that the discovery and initial characterization of a FliK protein potentially functioning as a hook length regulator in Caulobacter is of high relevance. In particular our finding that FliK is proteolytically regulated will likely be appreciated by all groups working on bacterial motility/flagella assembly and function.
Reviewer #4 (Significance (Required)):
StaR has been studied before, its role in stalk length control and its transcriptional profile has been determined. Here the contribution is that StaR is regulated at the level of stability by Lon and that the StaR accumulation profile during the cell cycle (in G1 phase) matches that of its transcriptional profile (reported previously), simply because StaR is unstable protein and that this instability is mediated by Lon. If StaR levels are elevated by preventing its degradation, then cells have longer stalks. However, despite this causality,there is no new insight on StaR function in this study. StaR overexpression defects have been described before, and this work only shows that such overexpression can also arise by interfering with degradation.
The main goal of our study was to identify novel Lon substrates and to study the role of Lon in regulating these proteins in order to gain molecular insights into the general mechanisms by which Lon impacts critical cellular processes such as cell differentiation and development. The intention was not to in detail characterize the specific functions of individual Lon substrates, such as StaR, as this would have exceeded the scope of this paper.
The part of FliK is enigmatic. The two first panels of figure 5D appear inconsistent regarding the abundance of flagellins (in panel 2: why are the flagellin levels higher in Δ-lon cells harboring Pvan-lon in the absence of inducer).
We agree that the discrepancy between the ∆lon and Lon depletion data is unexpected and at this point we can only speculate about the underlying reasons. Given that Lon affects flagellin protein levels most likely indirectly, it is possible that the time point of 4.5 hours that we chose in our depletion experiment was too short to detect an effect on flagellin levels. Despite this apparent inconsistency, we still think that the altered flagellin levels in the ∆lon and lon overexpression strains support an indirect role of Lon in regulating flagella-based motility. In the new version of our manuscript, we describe the flagellin data in more detail.
There is no null phenotype of FliK reported and the observed effect on flagellins could simply arise from a block of secretion caused by the absence of Lon which feeds back on flagellin expression. These effects are well known.
We completely agree that the changes in flagellin levels are likely caused by indirect effects that arise from problems in secretion. In the new version of our manuscript, we discuss this possibility in the discussion (l. 371-373).
It is also not clear that a 50% reduction in flagellins will block motility to the extent seen here or if this is due to other effects of Lon. Causality is missing. Can flagellin overexpression restore motility and what are the effects of deleting fliK also in the Lon mutant background?
The stabilization of FliK in ∆lon cells is unlikely to be the only cause of the motility defect of these cells. As explained in our responses to reviewer 3, we think that stabilization of SciP and potentially other Lon substrates with functions in flagella assembly / regulation likely contribute to this phenotype. In our new version of this manuscript, we mention this possibility more explicitly in l. 289-292 and l- 309-310. We have also restructured the text dealing with this aspect to improve clarity.
Because we think that the altered flagellin levels in the FliK overexpression strains are an indirect consequence of defects during the flagella assembly process (i.e., premature termination of hook length and secretion problems), we consider it unlikely that flagellin overexpression would restore motility in a ∆lon mutant. Although our FliK overexpression data suggest that FliK contributes to the motility phenotype of ∆lon cells, we agree that our data do not completely prove causality and we have carefully chosen our wording to not imply this.
Regarding the effects of deleting fliK in the Lon mutant background, please see answer to reviewer 3, question 3.
As it stands here, the work only shows that FliK (whose function is unknown in Caulobacter) is degraded by Lon. FliK is cell cycle regulated, but this may again arise from temporal control at the level of synthesis of an unstable protein that is constitutively degraded by Lon.
We disagree that our work “only shows that FliK is degraded by Lon”. Our work is the first study describing a FliK protein in Caulobacter and provides an important first characterization of this protein: we show that it accumulates in a cell cycle phase-specific manner and that Lon is required for this. Furthermore, our detailed mutational analysis in combination with the study of FliK-dependent phenotypes uncovered a critical dual role of the C-terminus of FliK in Lon-dependent degradation and FliK function. We agree that our study raised many new questions regarding this interesting protein. However, answering all these questions would certainly go beyond the scope of this manuscript.
Why FliK needs to be degraded seeing that it is likely a protein that is exported by the flagellar system and whether this export and its function is important remain unclear.
These are indeed interesting and very relevant points. It is noteworthy that the cytoplasmic abundance of many other secreted flagella components is precisely regulated by transcriptional, translational and in some cases by proteolytic control mechanisms. It has been proposed by others (e.g., Bonifield et al. 2000, J Bact) that precise regulation of secreted flagella components ensures correct order of secretion and may help to avoid competition between secreted flagella subunits during the assembly process. It will be interesting to address these hypotheses in future studies.
It is also unclear how FliK would regulate flagellins and whether this can impact motility.
As explained above, we think that the effect of FliK on flagellin levels is a consequence of improper secretion that feedbacks to flagellin expression levels. Our data showing that FliK and FliK-C overexpression results in reduced swimming, demonstrate that FliK impacts motility. Based on FliK studies in Salmonella (e.g., Muramoto et al. 1998, JMB), we think that the main reason for this result is the premature termination of hook synthesis in the presence of excess FliK, leading to shorter hooks, which may cause downstream effects, i.e., reduced flagellin levels and motility. We have added a new paragraph to our discussion (l. 371-385) in which we discuss this possibility.
In sum, Lon constantly degrades proteins, also those that are cell cycle regulated at the level of transcription/synthesis. Lon may simply fulfill a passive role in promoting the instability of many cytoplasmic proteins, including those that are cell cycle regulated. Thus, it may serve as a factor that renders a selection of cellular proteins unstable, perhaps to ensure that their presence in the cell cycle is determined when they are transcribed, but I think this global role is not sufficiently developed here.
This is the essence of the model we describe in this manuscript and that is illustrated in Figure 6, however we want to emphasize that, according to our data, Lon does not “simply fulfill a passive role” but has a crucial active role in establishing the fluctuation in levels of important proteins encoded by cell cycle regulated genes. We show that in the absence of Lon-dependent proteolysis, the cell cycle-dependent transcriptional control of important cell differentiation proteins becomes completely ineffective (Figure 3C, Figure 5B), thus highlighting the importance proteolysis in complementing gene regulatory mechanisms. We have demonstrated this concept by focusing on the newly identified Lon substrates StaR and FliK. Investigating additional putative Lon substrates to establish a more “global role” for Lon will be an interesting task for the future, but would in our eyes go beyond the scope of this manuscript.https://doi.org/10.7554/eLife.73875.sa2
- Kristina Jonas
- Kristina Jonas
- Kristina Jonas
- Kristina Jonas
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
The authors thank Peter Chien and his lab for sharing aliquots of purified Lon and for their help with Lon purifications, members of the Jonas lab for discussions and specifically Roya Akar for technical assistance, Claes Andréasson and his lab for providing the BL21-SI/pCodonPlus strain and the pSUMO-YHRC vector and for their help with the His-SUMO protein purification procedures, Yves Brun and Patrick Viollier for sharing aliquots of antibodies and Sean Crosson for providing plasmids and the ∆staR strain. The authors also thank the Clinical proteomics facility at KI/KS for support and advice as well as the Protein Expression and Characterization facility at SciLifeLab for sharing their equipment and providing help with the protein purifications. The study was financially supported by grants from the Swedish Research Council (Dnr. 2016-03300 and 2020-03545), the future leaders grant from the Swedish Foundation for Strategic Research (FFL15-0005), and funding from the Strategic Research Area (SFO) program distributed through Stockholm University.
- David Ron, University of Cambridge, United Kingdom
- Sonja V Albers, University of Freiburg, Germany
- Received: September 14, 2021
- Accepted: September 22, 2021
- Version of Record published: October 25, 2021 (version 1)
© 2021, Omnus et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
In the adult Drosophila midgut, basal intestinal stem cells give rise to enteroblasts that integrate into the epithelium as they differentiate into enterocytes. Integrating enteroblasts must generate a new apical domain and break through the septate junctions between neighbouring enterocytes, while maintaining barrier function. We observe that enteroblasts form an apical membrane initiation site (AMIS) when they reach the septate junction between the enterocytes. Cadherin clears from the apical surface and an apical space appears between above the enteroblast. New septate junctions then form laterally with the enterocytes and the AMIS develops into an apical domain below the enterocyte septate junction. The enteroblast therefore forms a pre-assembled apical compartment before it has a free apical surface in contact with the gut lumen. Finally, the enterocyte septate junction disassembles and the enteroblast/pre-enterocyte reaches the gut lumen with a fully-formed brush border. The process of enteroblast integration resembles lumen formation in mammalian epithelial cysts, highlighting the similarities between the fly midgut and mammalian epithelia.
Major genomic deletions in independent eukaryotic lineages have led to repeated ancestral loss of biosynthesis pathways for nine of the twenty canonical amino acids1. While the evolutionary forces driving these polyphyletic deletion events are not well understood, the consequence is that extant metazoans are unable to produce nine essential amino acids (EAAs). Previous studies have highlighted that EAA biosynthesis tends to be more energetically costly2,3, raising the possibility that these pathways were lost from organisms with access to abundant EAAs in the environment4,5. It is unclear whether present-day metazoans can reaccept these pathways to resurrect biosynthetic capabilities that were lost long ago or whether evolution has rendered EAA pathways incompatible with metazoan metabolism. Here, we report progress on a large-scale synthetic genomics effort to reestablish EAA biosynthetic functionality in mammalian cells. We designed codon-optimized biosynthesis pathways based on genes mined from Escherichia coli. These pathways were de novo synthesized in 3 kilobase chunks, assembled in yeasto and genomically integrated into a Chinese Hamster Ovary (CHO) cell line. One synthetic pathway produced valine at a sufficient level for cell viability and proliferation, and thus represents a successful example of metazoan EAA biosynthesis restoration. This prototrophic CHO line grows in valine-free medium, and metabolomics using labeled precursors verified de novo biosynthesis of valine. RNA-seq profiling of the valine prototrophic CHO line showed that the synthetic pathway minimally disrupted the cellular transcriptome. Furthermore, valine prototrophic cells exhibited transcriptional signatures associated with rescue from nutritional starvation. 13C-tracing revealed build-up of pathway intermediate 2,3-dihydroxy-3-isovalerate in these cells. Increasing the dosage of downstream ilvD boosted pathway performance and allowed for long-term propagation of second-generation cells in valine-free medium at a consistent doubling time of 3.2 days. This work demonstrates that mammalian metabolism is amenable to restoration of ancient core pathways, paving a path for genome-scale efforts to synthetically restore metabolic functions to the metazoan lineage.