PrkC modulates MreB filament density and cellular growth rate by monitoring cell wall precursors

How bacteria link their rate of growth to the external nutrient conditions is not known. To explore how Bacillus subtilis modulates the rate cells expand their encapsulating cell wall, we compared the single-cell growth rate to the density of moving MreB filaments under different conditions. MreB filament density scales with the growth rate, and is modulated by the mur genes that create the cell wall precursor lipid II. Lipid II is sensed by the serine/threonine kinase PrkC, which, among other proteins, phosphorylates RodZ. Phosphorylated RodZ then increases MreB filament density, increasing growth. Strikingly, increasing the activity of this pathway results in cells elongating far faster than wild type in nutrient-poor media, indicating slow-growing bacteria contain spare growth capacity. Overall, this work reveals that PrkC functions as a cellular rheostat, tuning the activities of cellular processes in response to lipid II, allowing cells to grow robustly across a broad range of nutrient conditions. One-sentence summary The serine/threonine kinase PrkC modulates both MreB filament density and cellular growth rate by sensing lipid II in Bacillus subtilis.


Main Text
Since the seminal work of Monod, bacteria have been held to grow as fast as the most limiting nutrient within the media allows (Monod 1949). To accomplish this, bacteria must balance the rates they synthesize their essential components: their RNA/protein ratio Schaechter, Maaloe, and Kjeldgaard 1958), DNA replication frequency, and the 35 rates they synthesize membranes and build external protective structures. For bacteria, a critical essential component is their peptidoglycan (PG) cell wall, a covalently crosslinked meshwork that surrounds cells conferring both shape and protection from lysis. In order to grow and divide, bacteria cells must add new material into this protective structure.

40
The gram-positive Bacillus subtilis inserts new PG into the wall via the action of two different PG synthetic systems: the Class A PBPs and the Rod complex (Cho et al. 2016;Meeske et al. 2016).
The Rod complex contains proteins required for rod shape: the transglycosylase/transpeptidase pair RodA/PBP2a, MreC, MreD, RodZ, and MreB (Fig. 1A). MreB polymerizes into short, curved membrane-associated filaments that orient to point around the rod circumference (Hussain et al. 45 2018). MreB orientation constrains the Rod complex to move around the rod circumference as it synthesizes new PG (Garner et al. 2011;Dominguez-Escobar et al. 2011), resulting in a radial arrangement of glycans that reinforce the cells rod shape against internal turgor (Dion et al. 2019).
It is not known how bacteria regulate the rates PG synthetic enzymes build new cell wall, much less how they scale their activity to match the synthesis of other biomolecules. To explore these 50 questions, we sought to understand how PG synthetic activity scaled with the growth rate in B.
subtilis, subsequently exploring the mechanisms by which this regulation occurs.
Given the directional motions of the Rod Complex are driven by PG synthesis (Garner et al. 2011;Dominguez-Escobar et al. 2011), their activity can be measured by quantifying the number and 55 velocity of directionally moving MreB filaments. As filaments might be too dense to resolve using diffraction-limited microscopy, we used a method that counts the number of directionally moving particles using temporal correlations between adjacent pixels. This approach is more accurate than particle tracking, giving measures equivalent to tracking filaments imaged with TIRF-structured illumination microscopy (TIRF-SIM) (Dion et al. 2019). 60 We started by measuring the density and velocity of MreB filaments in cells growing under agar pads composed of different media, along with their single-cell growth rate . In contrast to previous reports that tracked MreB filaments imaged with TIRF (Billaudeau et al. 2017), we find that filament density increases with growth rate (Fig. 1B-C). We verified this observation 65 using TIRF-SIM (Movie S1); cells growing in rich media contained numerous filaments, many of which were small and sub-diffraction in size, and filament density decreased as cells were grown in progressively less rich media. MreB filament velocity remained mostly constant across growth conditions (Fig. 1C), a finding further verified by tracking single MreB-HaloTag molecules (Fig. 1C, Movie S2). Similar results were obtained in cells expressing msfGFP fusions to both MreB 70 and Mbl (Fig. S1A,B). We believe the differences between this and the Billaudeau study likely Interestingly, during slow growth most MreC-mNeonGreen molecules were diffusive, and as growth rate increased, more MreC molecules moved directionally (Movie S3). Likewise, when cells were shifted from slow growth media to fast growth media, filament number increased while velocity remained constant ( Fig. 1D-E, Movies S4 and S5). Thus, B. subtilis increases the number of directionally moving Rod complexes to increase its growth rate. 80 Next, we explored how cells control the number of active Rod complexes. This regulation is unlikely to occur via changes in expression, as proteomics suggest most Rod Complex components remain relatively constant across different growth media (Fig. S2A). Alternatively, past work suggests a role for lipid II, the immediate precursor for PG synthesis: not only does the relative 85 abundance of Mur proteins scale with growth rate (Fig. S2B) (Dion et al. 2019), perturbations that deplete cells of lipid II cause MreB filaments to depolymerize off of the membrane, inactivating the associated synthetic enzymes (Schirner et al. 2015). To test if lipid II production affected MreB filament density, we titrated the expression of the first enzyme in the Mur pathway, murAA. In nutrient-rich media, low murAA inductions decreased the growth rate, and also MreB and MreC 90 density ( Fig. 2A, B). As induction increased, so did growth rate and MreB and MreC density, with high inductions reaching wild type (WT) values. As before, MreB and MreC velocity remained unchanged (Fig. 2C, Movie S6, and S2C).
When we repeated murAA titrations in cells growing in slow-growth carbon-limited media (S750 95 with glycerol as the carbon source) we observed a surprising effect: while decreased murAA inductions reduced both MreB density and cell growth, high murAA inductions not only increased MreB filament density 25% above that of WT, cells grew 30% faster than WT cells growing in the same media (Fig. 2D). While growth rate could be reduced by low murAA induction in all media tested, the growth acceleration only occurred in slow growth media, suggesting cells in rich media 100 elongate at their maximal rate ( Fig S2D). Thus, not only is both MreB filament density and overall growth rate regulated by the activity of the mur pathway, overexpression of mur genes somehow causes cells to grow faster than normal.
To test if MreB filament formation is regulated by interactions with Mur proteins, we tracked the 105 single-molecule motions of the membrane-associated mur proteins MraY and MurG and the lipid II flippases MurJ and Amj (Meeske et al. 2015). In no case did we see any directional motions (Movie S7), indicating these proteins do not form stable interactions with the Rod complex. These results suggested lipid II signals to MreB via another manner. To determine if this signal arose from cytoplasmic or periplasmic facing lipid II, we observed MreB as we depleted both flippases, 110 which should cause lipid II to accumulate inside the cell in the cytoplasmic orientation. Similar to previous murG/murB and upps depletions (Schirner et al. 2015), these depletions caused MreB filaments to depolymerize (Movie S8), indicating whatever signal controls MreB filaments arose from periplasmic oriented lipid II.

115
To determine what pathways sense periplasmic lipid II, we assayed the effects of chemical inhibitors on cell growth. One compound, staurosporine, reduced growth rate in minimal, but not rich media (Fig. 3A), suggesting signaling might be modulated through Hanks type kinases. B. subtilis possesses four serine/threonine kinases, and we assayed the growth of strains lacking each.
The greatest reduction in growth occurred in cells lacking prkC, and a similar reduction occurred 120 if we overexpressed the cognate phosphatase prkC (Fig. 3B).
PrkC contains a cytoplasmic kinase domain and a periplasmic region containing Penicillin-binding And Serine/Threonine kinase Associated (PASTA) domains. The PASTA domains of PrkC and its homologs have been reported in various studies to bind to either lipid II or muropeptides (a 125 moiety within lipid II) (Hardt et al. 2017;Kaur et al. 2019;Mir et al. 2011;Shah et al. 2008;Squeglia et al. 2011;Wamp, Rutter, Rismondo, Jennings, Möller, et al. 2020), thereby increasing the activity of the kinase domain. Mass spectrometry studies have indicated prkC and prkC levels are relatively constant across different growth conditions (Dion et al. 2019) (Fig. S3A).

130
Our next experiments revealed PrkC activity modulates the growth rate of cells in slow growth media: prkC overexpression caused cells to contain not only more MreB filaments, they also grew 53% faster than WT cells in the same media (Fig 3C, S3B). Likewise, ∆prpC cells showed more MreB filaments and grew 28% faster. Conversely, reducing PrkC activity by deleting prkC or overexpressing prpC reduced growth compared to WT and reduced MreB filament density. Similar 135 to murAA, prkC overexpression increased the growth rate above that of WT in minimal, but not rich media (S3C). Growing cells in minimal media with different carbon sources revealed that PrkCs ability to tune growth has an upper limit: While cells growing slowly in poor carbon sources could have their growth reduced or accelerated by the respective removal or overexpression of prkC, cells growing much faster in glucose (an optimal carbon source) could only have their 140 growth rate reduced, and not accelerated by pkrC overexpression beyond the already fast rate that WT cells exhibit in glucose containing media. This growth acceleration required PrkC's activities: cells encoding or overexpressing PrkC containing mutations that block pentapeptide binding (R500A) (Hardt et al. 2017) or abolish kinase 145 activity (K40R) (Madec et al. 2002) caused no change relative to WT cells ( Fig 3D). Furthermore, PrkC's growth enhancing effect is downstream of MurAA, as the growth acceleration caused by murAA overexpression was eliminated in ∆prkC cells (Fig. 3E). Thus, under conditions of slow growth, suggest PrkC modulates both MreB filament density and overall growth rate by sensing the amount of lipid II in the periplasm and relaying this signal to the cytoplasm. 150 The PrkC mediated acceleration of growth posits a dilemma, as increased growth should require more lipid II, and thus require more mur activity. Recent work discovered a connection between PrkC and MurAA that might give this feedback: MurAA degradation is mediated by YrzL and YpiB, each of which increase the rate of MurAA proteolysis (Wamp, Rutter, Rismondo, Jennings, 155 Moller, et al. 2020). Importantly, YrzL mediated degradation of MurAA is inhibited when it is phosphorylated by PrkC. To examine this finding in our context, we measured MurAA levels by fusing the native copy to mNeonGreen and quantitating cell fluorescence with widefield microscopy. Similar to observations by Wamp et al., prkC overexpression increased MurAA, and prpC overexpression reduced MurAA (Fig. 3F). Likewise, ∆yrzL and ∆ypiB cells showed 160 increased MurAA. Importantly, blocking YrzL phosphorylation with a T7A mutation decreased MurAA levels, and the phosphomimetic mutation YrzL(T7E) increased MurAA. Accordingly, our proteomic data indicates that YrzL and YpiB levels decrease as cells are grown in increasingly rich media ( Fig S3D). Thus, PrkC not only senses lipid II to adjust the rate of cell growth, it also feeds back on MurAA to modulate lipid II production. 165 How does PrkC modulate MreB filament number and overall growth rate? This is unlikely to occur via direct interactions with the Rod Complex, as single-molecule tracking of HaloTag-PrkC showed no directional motion, only diffusion within the membrane (Movie S9). Instead, we examined PrkC's substrates. PrkC has been reported to phosphorylate numerous proteins 170 (Ravikumar et al. 2018;Ravikumar et al. 2014), one of which is RodZ, a component of Rod complex, which we verified moves directionally using single-molecule tracking. Like MreC, an increasing number of RodZ molecules move directionally in faster growth media, while the rest diffuse along the membrane (Supplemental Text 2, Movie S10). RodZ is a transmembrane protein containing a periplasmic region that interacts with RodA/Pbp2a, and a cytoplasmic domain that 175 binds to MreB (Bendezu et al. 2009;van den Ent et al. 2010). E. coli lacking RodZ or its cytoplasmic domain contain fewer MreB filaments, suggesting RodZ promotes filament formation via nucleation or stabilization (Hussain et al. 2018;Bratton et al. 2018). Phosphoproteomic studies have shown PrkC phosphorylates RodZ at serine 85, a residue in the linker adjacent to the MreB binding domain (Ravikumar et al. 2014;van den Ent et al. 2010). 180 We verified RodZ is phosphorylated at S85 by PrkC using PhosTag gel shifts (Fig. S4A). In line with PrkC phosphorylating RodZ in response to lipid II flux, RodZ phosphorylation increased as we titrated murAA expression ( Fig 4A, Fig S4B). Likewise, RodZ phosphorylation scaled with the growth rate of cells growing in different amounts of different carbon sources (Fig. 4B). 185 To test if RodZ phosphorylation itself modulates cell growth, we created strains where S85 was mutated to alanine (S85A, mimicking dephosphorylation) or glutamic acid (S85E, mimicking phosphorylation). Strikingly, cells containing RodZ(S85E) not only showed more MreB filaments (Movie S11), they grew 66% faster than WT cells in the same media, close to the rate of WT cells 190 in glucose. In glycerol, cells containing RodZ(S85A) showed a slightly reduced growth rate (6%) (Fig. 4C). The effects of RodZ phosphorylation are epistatic to MurAA and PrkC,as RodZ(S85A) abolished the faster growth of both murAA (Fig. 4D) and prkC overexpression (Fig. 4C). However, without prkC, RodZ (S85E) is not sufficient to increase growth rate, suggesting that PrkC must also phosphorylate other proteins in order for cells to achieve accelerated growth. in diverse essential processes (Libby, Goss, and Dworkin 2015;Libby, Reuveni, and Dworkin 2019;Pompeo et al. 2015;Shah et al. 2008), PrkC might act to simultaneously tune their overall activities in response to nutrient conditions. If so, decoupling one PrkC regulated process, such as cell wall synthesis, from the others should result in fitness defects, as this would cause one biosynthetic process to proceed too fast or slow relative to the others. To test this, we used the 205 RodZ phospho-mutants to lock PrkC mediated control of PG synthesis into "on" or "off" states while we subjected cells at to extremes of murAA inductions or nutrient conditions. Indeed, cells containing RodZ (S85E) were unable to grow at low murAA inductions (Fig. 4C) and simply lysed, demonstrating that overactivation of PG synthesis is detrimental when the rest of the cell is growing slowly. Likewise, RodZ(S85A) cells grew slower than WT or RodZ(S85E) cells at high 210 murAA induction, indicating cells cannot achieve the maximal possible growth rate when the PrkC regulation of PG synthesis is blocked. (Fig. 4C). Similar effects were observed when these strains were grown in media containing different concentrations of nutrients ( Fig. 4D): In minimal media with saturating carbon (glycerol) or nitrogen (glutamate), "overactivated" RodZ(S85E) cells showed a growth advantage relative to WT and RodZ(S85A) cells. However, when carbon or 215 nitrogen were limiting, cells grew much slower, often lysing. Likewise, "non-activatable" RodZ(S85A) cells were unable to attain the maximal growth rate of RodZ(S85E) in saturating nutrients, However, when nutrients were limiting, RodZ(S85A) cells showed a growth advantage relative to WT and RodZ(S85E) cells. We also observed the PrkC pathway is involved in how fast B. subtilis adapts to nutrient shifts. When shifting from low carbon media to carbon-rich media, 220 ΔprkC cells were slower to adapt than WT cells, and conversely, ΔprpC cells adapt faster than both WT and ΔprkC cells (Fig. 4E).
Together, our data demonstrate that, rather than maximizing growth solely on limiting metabolites under conditions of slow growth, B. subtilis modulates both its amount of active cell wall synthesis 225 enzymes and the overall rate of cell growth by measuring the flux of lipid II through cell wall synthesis. The cell's metabolic state and PG synthetic activity are connected via PrkC, a kinase that phosphorylates multiple proteins when activated by lipid II. This signal is communicated to RodZ via phosphorylation, which then acts to increase the number of MreB filaments, thus increasing the number of directionally moving enzymes inserting material into the wall, thereby 230 causing cells to elongate faster (Fig. 4F). Interestingly, PrkC also controls the rate of lipid II production via YrzL controlling MurAA stability (Wamp, Rutter, Rismondo, Jennings, Moller, et al. 2020, thus modulating the amount of the same molecular signal that controls its activity. This positive feedback might serve to increase the amount of PG precursors needed for increased Rod Complex activity as nutrients increase. 235 While this initial study focused on cell wall synthesis, PrkC and its homologs in other bacteria (Hardt et al. 2017;Cuenot et al. 2019;Wamp, Rutter, Rismondo, Jennings, Möller, et al. 2020) phosphorylate hundreds of proteins involved widely divergent cellular processes (Pereira, Goss, and Dworkin 2011;Ravikumar et al. 2018;Ravikumar et al. 2014), as well as controls cell growth 240 via modulation of (p)ppGpp synthesis (Libby, Reuveni, and Dworkin 2019). This suggests that PrkC might function as a cellular rheostat, tuning the activities of many different cellular processes in response to lipid II so that their activities run at equivalent rates across the range of nutrient conditions.

245
Our findings also have implications for our understanding of the limitations of bacterial growth.
While it has been long held that the limiting nutrient determines their maximal growth rate, these experiments reveal that, under limiting nutrient conditions, B. subtilis still contains the potential to grow ~1.5 faster when given regulatory pathways are activated (Towbin et al. 2017;Cheng et al. 2014). Along with recent work demonstrating carbon limited bacteria contain an excess of 250 inactive ribosomes (Bosdriesz et al. 2015), our findings suggest that, rather than growing at the maximal possible rate, bacteria actively hold back some growth capacity in reserve, activating or deactivating this capacity to tune their growth in response to the nutrient state or presence of antibiotics.     Fig. S1A and Movies S1 and S3. 360 (C) MreB velocity remains the same in different media -Cells expressing MreB-mNeonGreen (bYS09) were grown in different media, and MreB velocity determined by particle tracking and MSD analysis. See also S1B and Movie S1-2.

(D to E) MreB filament density, but not velocity increases during nutrient upshifts -Cells 365
were grown in S750 maltose for 6 hours in liquid, washed in CH, then placed under a pad made of CH media immediately before imaging with TIRF-SIM. Plotted is the growth rate and either the density (D) or velocity (E) of directionally moving MreB filaments. See also Movies S4 and S5.   (B) Effects of kinases and phosphatases on cell growth -The growth rate of cells in S750 glucose and glycerol was assayed in strains containing deletions of different Ser/Thr kinases 400 (bYS542, bYS969, bYS970 and bYS971) and PrkC or PrpC overexpression (bYS730 and bYS545 with 30mM xylose).

(D) PrkC's lipid II binding and kinase activity are essential for growth acceleration
The growth rate of overexpressed PrkC mutants was assayed in S750 glycerol with 30mM 410 xylose. PrkC(R500A) (bYS727) has a mutation in the PASTA domain. PrkC(K40R) (bYS729) has a mutation in the kinase catalytic domain.  (E) RodZ phosphorylation gives a growth advantage in excess nutrients, but a detriment when nutrients are limiting -Growth rates of WT (PY79), RodZ(S85A)(bYS125), and RodZ(S85E)(bYS127) cells assayed in either S750-glycerol or the same media but where carbon (glycerol), nitrogen (glutamate) or phosphate (KH2PO4) were reduced by 50-fold. 465 (F) PrkC modulates the rate cells adapt to increased nutrients -The growth rate of WT, ΔprkC (bYS542), and ΔprpC (bYS543) cells was assayed as they were shifted from growth in S750 glycerol (0.02%) to growth in S750 glycerol (1%).

(G) Model for regulation of cell growth by PrkC 470
Top -Schematic of PrkC mediated control of cell growth by lipid II. PrkC, by binding to lipid II, senses the amount, or flux, of cell wall precursors. Activated PrkC phosphorylates RodZ, YrzL, and other cellular targets. Phosphorylated RodZ then increases the amount of MreB filaments (either by nucleation or stabilization), which then activate the enzymes within the rod complex. PrkC also phosphorylates Yrzl, which increases the amount of MurAA by preventing 475 its degradation by ClpC, thus increasing lipid II levels.
Bottom Left -When nutrients are limiting, there is low rate of lipid II, and thus PrkC is relatively inactive. This causes RodZ to be mostly unphosphorylated, resulting in only a few MreB filaments per cell. 480 Bottom right -As external nutrients increase, lipid II abundance increases, which leads to increased PrkC activity. This results in an increased phosphorylation of RodZ, which increases the number of MreB filaments in the cell, causing it to elongate faster.

Supplementary Movie Captions
HaloTag-PrkC (bYS765) was incubated with 10uM of HaloTag-JF549 ligand for 30 minutes, washed, then prepared as detailed in methods. Cells were imaged using TIRFM at 1 second 590 intervals with 0.5s exposures.

Movie S10: TIRF Imaging of HaloTag-RodZ
Cells were imaged using TIRFM at 1 second intervals using an exposure time of 0.5 seconds.

Notethis section will be updated in a updated preprint as the cross data analysis between labs evolves. 605
Previously, Billaudeau and colleagues conducted a study detialing how the number of directionally moving MreB filaments (measured by particle tracking) changed with growth rate (Billaudeau et al. 2017). They came to the opposite conclusion that we arrive to here, that as B. subtilis grows faster, the filament velocity increases, not the number of MreB filaments.

610
We are confident in our data: we see same results not only with different fluorescent fusions to both MreB and Mbl; we gain the same results 1) conducting single-molecule tracking of sparsely labeled HaloTag-MreB, and, importantly, when 2) calculating the density and velocity MreC-mNeonGreen fusions. However, we trust in the good faith and scientific ability of colleagues, and thus, are working to understand the differences between our two studies. We are exchanging both 615 data, strains, and code so that we can determine if the discrepancies in the two studies arise from 1) the strains and fusions used, or 2) the different analysis pipelines.
During the interim, we can only speculate why these differences occur. We think they may arise from either source: 1) The difference may arise from the MreB fusions used, as we and others have noted differences 620 in the structures different MreB fusions make in the cell. For example, some fusions show numerous small filaments, while other fusions show fewer, but much longer filaments.
2) The difference may also come from the method of analysis, as while Billaudeau used particle tracking, which is subject to the diffraction limit. In contrast, our approach uses temporal correlations that between adjacent pixels and thus is able to count moving particles even when 625 their density exceeds the diffraction limit (Dion et al. 2019). However, to control for this, we conducted single-particle tracking of sparsely labeled MreB Halo-Tag fusions and again saw no change in MreB velocity in different growth media. Thus, we currently disfavor the idea that these differences come from analysis methods, and likely think they arise from the fusion used.
We will update this preprint as we conduct our cross-lab comparison. 630

Supplementary Text 2
Upon initiating our study of RodZ, we noted the correct open reading frame (ORF) of Bacillus subtilis rodZ(ymfM) was previously overlooked (https://www.uniprot.org/uniprot/O31771). We verified this using SDS PAGE gels, where HaloTag-RodZ (bYS163) (which contains an N-635 terminal HaloTag fusion directly before the overlooked upstream start codon) resulted in a single band at 70KDa (Fig. S4A, bottom panel). This indicated the full length of B. subtilis RodZ should be 304 aa instead of 288aa in the original annotation, where the first helix H1 of RodZ was truncated. Therefore, the phosphorylation site on RodZ should be S85 instead of S69 (Ravikumar et al. 2014). Single-molecule imaging of RodZ gave further verification this was the functional 640 protein, showing that only HaloTag-RodZ (bYS163) with the correct ORF moves directionally around the cell width (Movie S10), while the original annotation, which truncated the protein (bYS151) moves freely, diffusing within the membrane, likely due to the lost interaction between MreB and RodZ (van den Ent et al. 2010).

Culture growth.
Unless otherwise noted, B. subtilis was grown in casein hydrolysate (CH) medium or S750 minimal media with different carbon sources. Where indicated, xylose or isopropyl thiogalactoside (IPTG) was added. Cells were grown at 37°C to an OD600 of ~0.4 to 0.6. The default concentration 650 of carbon source was 1%.

General imaging conditions.
All components were prewarmed to 37 °C including Matek dishes, coverslips, agarose pads, and media. Exposure times varied between 300ms and 1 second, as indicated. And the interval between 655 frames was 1 sec. A 488nm laser line was used for the imaging of mNeonGreen, and a 561nm laser line was used for the imaging of JF549-Halo dye.

Single-cell growth rate time-lapse experiments.
2ul of cell culture was spotted on No. 1.5 glass-bottomed dishes (MatTek, MA) under 3% agarose 660 pads made with growth media. Phase-contrast images were collected on a Nikon Ti microscope equipped with a Hamamatsu ORCA Flash4 CMOS camera with a Nikon Plan Apo λ 100×/1.4NA objective with exposure times of 200ms, and each pixel was 64 nm. Images were acquired from 30-60 fields of view every 2 minutes for a total of 2-4 hours using Nikon NI Elements. Analysis of phase-contrast time-lapse movies was performed using a custom-built package in MATLAB. 665 For nutrient upshift analyses, first cells were grown in slow growth media to mid-log (OD600 of ~0.4 to 0.6) and then imaged under 3% agarose pads with the fast growth media. Phase-contrast time-lapse movies were then segmented, and single cell growth rates were calculated based on the piece-wise linear fitting of the area of cells. The instantaneous rate of growth during shift was calculated by rolling window regression using a window of 15 min. 670 Bulk growth rate measurements.
Cultures were grown at 37 °C to an OD600 of ~0.6. The cultures were diluted back to a calculated OD 600 of 0.05 (replicated in 2-5 wells for each culture) and their growth measured using an Epoch microplate spectrophotometer (BioTek, VT) at 37 °C with continuous shaking. The growth rates were calculated from OD600 measurements that were recorded every 5 min for 6 h~10h. 675

TIRF-SIM Imaging of MreB.
Cells were prepared as described in "Culture growth." Cells were placed under an agarose pad in a No. 1.5 glass-bottomed dish (MatTek, MA) for imaging. Images were collected on a GE OMX in TIRF-SIM mode, using an Edge 5.5 sCMOS camera (PCO AG, Germany) and a 60x objective. 680 100 msec exposures from a 488 nm diode laser were used for each rotation. Raw images were reconstructed using SoftWoRx (GE Healthcare, MA) software. The pixel size after reconstruction was 40 nm.

MreB-mNeonGreen and MreC-mNeonGreen dynamics assayed by TIRFM. 685
Cells were prepared as described in "Culture growth." Images were collected on a Nikon Ti microscope equipped with a Hamamatsu ORCA Flash4 CMOS camera with a Nikon 100X NA 1.45 objective. The pixel size was 65 nm. Cells were imaged with TIRFM for 120 seconds, followed by a single phase-contrast image.

690
For the nutrient upshifts conducted on strains expressing MreB-mNeonGreen, cultures grown in slow growth media to mid-log (OD600 of ~0.4 to 0.6) were concentrated by centrifugation at 4500g for 60 seconds. The cell pellet was resuspended in nutrient-rich media (CH). 2ul of cell culture was then spotted on an ethanol-cleaned Matek disk with a No. 1.5 glass coverslip, under a 3% agarose pad containing CH, placed into a 37 °C chamber immediate imaged. Time-lapse imaging 695 of MreB-mNeonGreen was performed every minutes 15 minutes with 300ms exposures, and the time interval is 1 sec.

Imaging MreB-HaloTag , RodZ-HaloTag and PrkC-HaloTag by TIRFM.
MreB-HaloTag cells were incubated with 50 pM of HaloTag-JF549 ligand for 30 minutes during 700 growth at 37 °C with rotation and then washed twice with 2 volumes of growth media for sparse labeling. Cells were imaged at 1 second intervals using TIRF microscopy. The exposure time was 0.5 seconds. RodZ-HaloTag and PrkC-HaloTag were incubated with 1uM of HaloTag-JF549 ligand for 30 minutes for complete labeling. Following this, cells were washed twice with 2 volumes of growth media before imaging. 705

Analysis of the density of MreB filaments and MreC enzymes.
The method used to quantitate the density of MreB filaments and MreC enzymes has been published and extensively characterized previously (Dion et al. 2019). Briefly, phase images were segmented, and the width and midline of each cell were calculated. Next, the fluorescence time-710 lapses were analyzed based on the segmentation mask of the phase image. Cell contours and dimensions were calculated using the Morphometrics software package. Then kymographs were generated for each row of pixels along the midline of the cell, and the time-lapse movie for each cell was converted into a single 2D image. To identify filaments in the kymograph, closed image contours were generated in the 2D image. For these contours, we calculated the total intensity, the 715 centroid and orientation. Next, to identify cases where the same MreB filament appears in multiple sequential kymographs, each object in a given kymograph was linked to a corresponding object in the previous and following kymographs based on the above properties of the object, and these objects counted as one event. All of the image analyses were performed using custom MATLAB code. 720

Analysis of the velocity of directional moving MreB and MreC
Particle tracking of MreB and MreC was performed using the software package FIJI (Schindelin et al. 2012) and the TrackMate plugin (Tinevez et al. 2017). First, the phase images of cells were segmented, and then a segmentation mask was used to reject the tracks outside of the cells. And 725 only tracks that are longer than seven frames were used for calculations of particle velocity.
Particle velocity for each track was calculated based on nonlinear least-squares fitting using MSD(t)=4Dt+(vt)2, where MSD is the mean squared displacement, t is the time interval, D is the diffusion coefficient, and v is velocity. The time interval used for the velocity analysis was 80% of the track length. Tracks were excluded if the R-squared for the fitting of log[MSD] versus log[t] 730 was less than 0.95. Single-molecule trajectories were discarded if displacement was < 250 nm.

Epifluorescence measurements of MurAA
Cells were prepared as described in "Culture growth." Epifluorescence and phase images were collected using a Nikon Eclipse Ti equipped with a Nikon Plan Apo λ ×100/1.4NA objective and 745 an Andor Clara camera. The pixel size was 64.5 nm. Exposure time is 500ms.

Spot Plating Assay for the determination of survival of B. Subtilis
Cells were prepared as described in "Culture growth." Overnight culture in LB or S750 glycerol was grown to OD600~0.6. 8x108 CFU/ml was used as the conversion ratio of OD600 to CFU/ml. 750 Then the culture was serially diluted to obtain 1x105, 1x104, and 1x103 CFU/ml culture. Then 10 μl of 1x105, 1x104, and 1x103 CFU/ml culture were plated onto the LB plate respectively. The fosfomycin concentrations on LB plates were 25ug/ml and 50ug/ml. Plates were incubated at 37oC overnight.

Strain Construction
765 bYS170 harboring mreC-mNeonGreen was generated by transforming bMD88 (Schirner et al. 2015) with a Gibson assembly consisting of three fragments: (1) PCR with primers oMD117 and oYS230 and PY79 template genomic DNA (containing the upstream of mreC); (2) PCR with primers oYS232 and oYS233 and gBlocks gene fragment containing mNeonGreen; (3)  bYS125 harboring rodZ(S85A) was generated by transforming bMK169 with a Gibson assembly consisting of two fragments: 1) PCR with primers oZB36 and oYS286 and PY79 template genomic 785 DNA (containing the upstream of rodZ locus); (2) PCR with primers oYS287 and oZB44 and PY79 template genomic DNA (containing the downstream of rodZ locus). Selection on LB plates for growth in the absence of xylose and kan antibiotics resulted in strain bYS125. Counterselection was done in the presence of 30 mM xylose. The genotype was confirmed by PCR and sequencing. 790 bYS127 harboring rodZ(S85E) was generated by transforming bMK169 with a Gibson assembly consisting of two fragments: 1) PCR with primers oZB36 and oYS288 and PY79 template genomic DNA (containing the upstream of rodZ locus); (2) PCR with primers oYS289 and oZB44 and PY79 template genomic DNA (containing the downstream of rodZ locus). Selection on LB plates for growth in the absence of xylose and kan antibiotics resulted in strain bYS127. Counterselection 795 was done in the presence of 30 mM xylose. The genotype was confirmed by PCR and sequencing.
bYS151 harboring NTD truncated HaloTag-rodZ was generated by transforming bMK169 with a Gibson assembly consisting of three fragments: 1) PCR with primers oZB36 and oYS290 and PY79 template genomic DNA (containing the upstream of rodZ locus); (2) PCR with primers 800 oYS616 and oYS601 and Halo-tag template genomic DNA (3) PCR with primers oYS291 and oZB44 and PY79 template genomic DNA (containing the downstream of rodZ locus). Selection on LB plates for growth in the absence of xylose and kan antibiotics resulted in strain bYS151. Counterselection was done in the presence of 30 mM xylose. The genotype was confirmed by PCR and sequencing. 805 bYS163 harboring HaloTag-rodZ was generated by transforming bMK169 with a Gibson assembly consisting of three fragments: 1) PCR with primers oZB36 and oYS290 and PY79 template genomic DNA (containing the upstream of rodZ locus); (2) PCR with primers oYS616 and oYS601 and Halo-tag template genomic DNA (3) PCR with primers oYS291 and oZB44 and 810 PY79 template genomic DNA (containing the downstream of rodZ locus). Selection on LB plates for growth in the absence of xylose and kan antibiotics resulted in strain bYS163. Counterselection was done in the presence of 30 mM xylose. The genotype was confirmed by PCR and sequencing.
bYS165 harboring HaloTag-rodZ(S85A) was generated by transforming bMK169 with a Gibson 815 assembly consisting of two fragments: 1) PCR with primers oZB36 and oYS286 and bYS163 template genomic DNA(containing the upstream of rodZ locus and HaloTag); (2) PCR with primers oYS287 and oZB44 and PY79 template genomic DNA. Selection on LB plates for growth in the absence of xylose and kan antibiotics resulted in strain bYS165. Counterselection was done in the presence of 30 mM xylose. The genotype was confirmed by PCR and sequencing. 820 bYS365 harboring pxyl-murAA::spec was generated upon transformation of PY79 with a fourpiece Gibson assembly reaction, that contained the following PCR products. (1) PCR with primers oMD388 and oMD389 and PY79 template genomic DNA.; (2) PCR with primers oJM028 and oJM029 and the spectinomycin-resistance cassette loxP-spec-loxP(amplified from pWX467 [gift 825 of D. Rudner]); (3) a fragment containing the xylR gene, and the PxylA promoter with an optimized ribosomal binding sequence (amplified from pDR150 using primers oMD73 and oMD226); and (4) a 1271 bp fragment containing the murAA coding region (amplified from PY79 genomic DNA using primers oMD394 and oMD395). 830 bYS395 harboring murAA-mNeonGreen::cat was generated upon transformation of PY79 with a Gibson assembly consisting of three fragments: (1) PCR with primers oYS171 and oYS172 and PY79 template genomic DNA (containing the upstream of murAA locus); (2) PCR with primers oYS147 and oYS865 and gBlock gene fragment containing mNeonGreen template DNA (3) PCR with primers oYS175 and oYS176 and chloramphenicol-resistance cassette loxP-cat-loxP 835 template (amplified from pWX465[a gift from D. Rudner]); and (4) PCR with primers oYS177 and oYS178 and PY79 template genomic DNA.
bYS542 harboring ∆prkC is a markerless prkC deletion strain. The loxP-cat-loxP antibiotics cassette of bYS537 was looped out using a cre-expressing plasmid pDR244. 850 bYS539 harboring amyE::Pxyl-prkC::kan was generated upon transformation of PY79 with a four-piece Gibson assembly reaction, that contained the following PCR products. (1) PCR with primers oMD191 and oMD108 and PY79 template genomic DNA; (2) PCR with primers oJM028 and oJM029 and loxP-kan-loxP template (amplified from pWX470); (3) PCR with primers oMD73 and oMD226 and pDR150 [gift of D. Rudner] template DNA (containing the xylR gene, 855 and the PxylA promoter with an optimized ribosomal binding); (4) PCR with primers oYS1014 and oYS1015 and PY79 template genomic DNA; and (5) PCR with primers oMD196 and oMD197 and PY79 template genomic DNA. bYS543 harboring ∆prpC::erm is a prpC deletion strain. It was generated upon transformation of 860 PY79 with a three-piece Gibson assembly reaction that contained the following PCR products. 1) PCR with primers oYS934 and oYS935 and PY79 template genomic DNA (containing the upstream of prpC locus); (2) PCR with primers oJM028 and oJM029 and loxP-erm-loxP template (amplified from pWX467 [gift of D. Rudner]); (3) PCR with primers oYS936 and oYS937 and PY79 template genomic DNA. The loxP-erm-loxP antibiotics cassette was subsequently looped 865 out using a cre-expressing plasmid pDR244.
bYS545 harboring amyE::Pxyl-prpC::erm was generated upon transformation of PY79 with a four-piece Gibson assembly reaction, that contained the following PCR products: (1) A 1228 bp fragment containing sequence upstream of the amyE locus (amplified from PY79 genomic DNA 870 using primers oMD191 and oMD108); (2)  with primers oYS864 and oYS981 and PY79 template genomic DNA; and (5) PCR with primers oMD196 and oMD197 and PY79 template genomic DNA.
bYS551 harboring amyE::Phyper-PrpC::erm was generated upon transformation of PY79 with a five-piece Gibson assembly reaction, that contained the following PCR products.
bYS747 harboring rodZ(S85E), Pxyl-mreC::erm was generated upon transformation of bYS127 with genomic DNA from bMD88. 950 bYS736 harboring ∆prkC, mreB-mNeonGreen was generated upon transformation of bYS734 with genomic DNA from bYS09. Counterselection was done in the presence of 30 mM xylose. Selection on LB plates for growth in the absence of xylose and erm antibiotics resulted in strain bYS736. 955 bYS738 harboring ∆prpC, mreB-mNeonGreen was generated upon transformation of bYS735 with genomic DNA from bYS09. Counterselection was done in the presence of 30 mM xylose. Selection on LB plates for growth in the absence of xylose and erm antibiotics resulted in strain bYS738. 960 bYS748 harboring rodZ(S85A), mreB-mNeonGreen was generated upon transformation of bYS746 with genomic DNA from bYS09. Counterselection was done in the presence of 30 mM xylose. Selection on LB plates for growth in the absence of xylose and erm antibiotics resulted in strain bYS748. 965 bYS749 harboring rodZ(S85E), mreB-mNeonGreen was generated upon transformation of bYS747 with genomic DNA from bYS09. Counterselection was done in the presence of 30 mM xylose. Selection on LB plates for growth in the absence of xylose and erm antibiotics resulted in strain bYS749. 970 bYS763 harboring HaloTag-prkC::cat was generated by transforming bYS542 with a Gibson assembly consisting of five fragments: 1) PCR with primers oMD464 and oYS1031 and PY79 template genomic DNA (containing the upstream of prkC locus); (2) PCR with primers oYS616 and oAB60 and Halo-tag template genomic DNA; (3) PCR with primers oYS1086 and oYS984 and PY79 template genomic DNA (4) PCR with primers oJM028 and oJM029 and loxP-cat-loxP 975 template (amplified from pWX465); (5) PCR with primers oYS861 and oYS862 and PY79 template genomic DNA (containing the downstream of prkC locus).