Coordination of capsule assembly and cell wall biosynthesis in Staphylococcus aureus

The Gram-positive cell wall consists of peptidoglycan functionalized with anionic glycopolymers, such as wall teichoic acid and capsular polysaccharide (CP). How the different cell wall polymers are assembled in a coordinated fashion is not fully understood. Here, we reconstitute Staphylococcus aureus CP biosynthesis and elucidate its interplay with the cell wall biosynthetic machinery. We show that the CapAB tyrosine kinase complex controls multiple enzymatic checkpoints through reversible phosphorylation to modulate the consumption of essential precursors that are also used in peptidoglycan biosynthesis. In addition, the CapA1 activator protein interacts with and cleaves lipid-linked CP precursors, releasing the essential lipid carrier undecaprenyl-phosphate. We further provide biochemical evidence that the subsequent attachment of CP is achieved by LcpC, a member of the LytR-CpsA-Psr protein family, using the peptidoglycan precursor native lipid II as acceptor substrate. The Ser/Thr kinase PknB, which can sense cellular lipid II levels, negatively controls CP synthesis. Our work sheds light on the integration of CP biosynthesis into the multi-component Gram-positive cell wall.

T he bacterial cell envelope is a complex multilayered structure consisting of peptidoglycan (PG), which in Gram-positive bacteria is densely decorated with glycopolymers such as wall teichoic acid (WTA) and capsular polysaccharide (CP). The coordinated synthesis and assembly of these polymers is pivotal for maintenance of cell wall architecture and function 1 . In contrast to the biosyntheses of PG and WTA, for which the individual enzymatic reactions have already been characterized in vitro 2,3 , the biochemistry underlying capsule formation in Gram-positive bacteria is not well understood. Even more so, it is largely unknown how the different cell wall synthesis pathways, which share building blocks and membrane carriers, function in a coordinated and integrated fashion.
In the case of Staphylococcus aureus, an important opportunistic pathogen 4 , the expression of a polysaccharide capsule contributes substantially to the ability to cause invasive disease [5][6][7] . Serotype 5 and 8 capsular polysaccharide (CP5 and CP8) types are dominant among clinical isolates 6 . S. aureus USA300, which is prevalent in the United States, lacks a capsule due to several conserved mutations within the cap5 locus 8 . However, the majority of USA300-associated infections involved superficial wounds or abscesses 9 , and USA300 isolates are not common outside of North America 10 . Among predominant methicillinresistant S. aureus clones worldwide are the CP8 + lineages ST1, ST30, ST59, ST80, and ST239 and the CP5 + lineages ST5 and ST22.
CP5 and CP8 share similar trisaccharide repeating units, which are identical in monosaccharide composition and sequence, only differing in the glycosidic linkages between the sugars and the sites of O-acetylation 11 .
The CP5 biosynthetic gene cluster comprises 16 genes (cap5A-cap5P; Fig. 1a) 12 encoding for proteins involved in polymer biosynthesis 7,13,14 , acetylation 15 , transport, and the regulation of CP production 16,17 . Database homology searches with amino acid sequences of cap5 operon gene products allowed for the prediction of individual enzymatic functions and the proposal of a pathway for capsule (CP) biosynthesis in S. aureus 18 .
Within this pathway (Fig. 1b), synthesis of the soluble building blocks occurs in the cytoplasm via three distinct reaction cascades, through which the universal cell envelope substrate UDP-D-N-acetylglucosamine (UDP-D-GlcNAc), is converted into the three different nucleotide-coupled sugars UDP-N-acetyl-D-fucosamine (UDP-D-FucNAc), UDP-N-acetyl-L-fucosamine (UDP-L-FucNAc) and UDP-N-acetyl-D-mannosaminuronic acid (UDP-D-ManNAcA). The synthesis of the first soluble precursor UDP-D-FucNAc is allegedly catalyzed in a two-step process by the enzymes CapD and CapN. Only recently, the integral membrane protein CapD was shown to function as a 4,6-dehydratase, which generates the intermediate UDP-2-acetamido-2,6-dideoxy-D-xylo−4-hexulose 19 , proposed to be further converted to UDP-D-FucNAc by the action of the membrane-associated reductase CapN, though experimental evidence is lacking so far. Subsequently, CapM is supposed to transfer the phosphosugar moiety of UDP-D-FucNAc to the membrane-anchored lipid carrier undecaprenyl-phosphate (C 55 P), yielding lipid I cap .
The second cytoplasmic reaction cascade generating the soluble precursor UDP-L-FucNAc involves the enzymes CapE, CapF and CapG, the enzymatic functions of which have already been elucidated biochemically 13 . The transferase CapL is assumed to further attach L-FucNAc to lipid I cap leading to the formation of the second CP lipid intermediate, lipid II cap .
The third nucleotide-activated monosaccharide required for CP5 production, UDP-D-ManNAcA, is generated by the epimerase CapP and the dehydrogenase CapO 7,14 . The transmembrane protein CapI has been proposed to transfer the ManNAcA moiety to lipid II cap , thereby generating the final capsule precursor lipid III cap . The C 55 P-coupled trisaccharide is most likely further modified by the putative acetyltransferase CapH 15 , which catalyzes the O-acetylation of L-FucNAc residues in position C3 in CP5 strains 11 . The complete, modified precursor is then translocated to the outer surface of the cell membrane, where polymerization is assumed to take place. These processes are proposed to be facilitated by the putative flippase CapK and the polymerase CapJ, respectively 12,18 . The attachment of CP precursors to the MurNAc (N-acetylmuramic acid) moiety of peptidoglycan is achieved by a yet unknown mechanism possibly involving a member of the LCP (LytR-CpsA-Psr) family of proteins 20,21 . This process likely releases the lipid carrier C 55 P, which enters new synthesis cycles.
The fact that the undecaprenyl-phosphate carrier is found in limited amounts within the cell 22 and required for the biosyntheses of diverse cell envelope components, like CP, WTA and PG 2,3,18 , makes a well-orchestrated spatial and temporal regulation of these processes crucial for the viability of the cell. The consequences arising from perturbation of this balanced biosynthetic network have been well exemplified for WTA biosynthesis. Interference with late WTA biosynthesis steps has been shown to be lethal, although the polymer per se is not crucial for viability 23 . Similarly, late stage genes are conditionally essential since they are dispensable for viability in an early gene (tarO or tarA) deletion background, a phenomenon referred to as the "essential gene paradox" 24 . Inhibition of late WTA biosynthesis steps causes the accumulation of dead-end lipid-linked intermediates and thus depletes the cellular pool of C 55 P to critical levels impeding peptidoglycan biosynthesis, resulting in cell death 25 .
In S. aureus, the regulation of CP biosynthesis is not only achieved by differential gene expression [26][27][28] , but has additionally been linked to tyrosine phosphorylation 16,17 . Bacterial tyrosine kinases (BY-kinases) are widespread in bacteria and have multifaceted roles in bacterial exopolysaccharide production 29 . BY-kinases belong to the family of P-loop containing kinases 30 , whereby "P-loop" designates a characteristic amino acid sequence resembling the Walker A nucleotide binding motif. BY-kinases of Firmicutes are composed of two interacting polypeptides, a transmembrane activator protein and a cytoplasmic BYkinase 31 . The cytoplasmic kinase carries a C-terminal tyrosine cluster that undergoes autophosphorylation in the presence of ATP. Based on studies in S. aureus and Streptococcus pneumoniae, the cytoplasmic kinase protein alone is not sufficient for phosphotransfer, but has to interact with the C-terminus of the transmembrane adaptor to undergo autophosphorylation 16,32,33 . Concomitantly, phosphate groups can be transferred to tyrosine residues of target proteins, thus modulating their activity. Two triplets of adjacent genes encoding for a transmembrane adaptor, a cytoplasmic BY-kinase, and a cognate phosphotyrosine phosphatase were identified in the genome of S. aureus serotype 5: The capA1/capB1/capC1 triplet (also referred to cap5A/cap5B/cap5C) is located at the 5ʹ-end of the cap5 operon (Fig. 1a), whereas the highly similar capA2/capB2/capC2 triplet is found elsewhere on the bacterial chromosome 12,16 . So far, the cellular roles of the distinct CapAB complexes are not fully understood. Particularly, the nature of the stimulus that triggers CapAB signaling and the exact mode of signal transduction are still elusive.
In this study, we functionally reconstitute the entire CP biosynthetic reaction cascade generating the three membraneanchored CP precursors lipid I cap , lipid II cap and lipid III cap , allowing us to identify crucial enzymatic check points, which are regulated by the tyrosine kinase CapA1B1 complex to control the consumption of essential precursors. Reconstitution of the membrane anchored CapA1 adaptor protein further reveals a yet elusive function. We show that CapA1 is a dual-function kinase activator/phosphodiesterase protein crucial for signaling and processing of the CP polymer. CapA1 interacts with lipid-bound CP precursors to catalyze the cleavage of the pyrophosphate linkage, releasing the essential lipid carrier C 55 P. Moreover, we elucidate the principles of CP attachment to murein precursors in Gram-positive bacteria. We show that the transfer of the capsular phosphosugar moiety is conducted by a member of the LCP protein family and identify the acceptor substrate. We provide biochemical evidence that the attachment of the anionic precursor likely occurs on the level of the lipid-linked peptidoglycan precursor lipid II, and is facilitated in the presence of CapA1, indicating its cooperative functioning with LCP proteins.

Results
In vitro reconstitution of capsule biosynthesis. Homology searches with CapD and CapN suggested that both proteins are involved in the synthesis of the first soluble capsule precursor UDP-D-FucNAc 18 . More recently, CapD was characterized at the molecular level, and the enzymatic product was shown to be UDP-2-acetamido-2,6-dideoxy-D-xylo−4-hexulose 19 . CapN is proposed to further convert the CapD reaction product to UDP-D-FucNAc by stereospecific reduction of the C-4 keto group 18 .
As confirmed by capillary electrophoresis (CE) and mass spectrometry (MS), purified CapN (Supplementary Figure 1) catalyzed the NADPH-dependent conversion of the intermediate UDP-2-acetamido-2,6-dideoxy-D-xylo−4-hexulose to a sugar nucleotide species having a molecular mass of m/z 590.4 for the negatively charged molecule, consistent with the formation of UDP-D-FucNAc (Supplementary Figure 2). The subsequent transfer of the phosphosugar moiety of UDP-D-FucNAc to the lipid anchor C 55 P is thought to be catalyzed by the polyprenylphosphoglycosyltransferase CapM, to initiate the assembly of the lipid-anchored trisaccharide repeating units 18 . Incubation of purified CapM protein with C 55 P and UDP-D-[ 14 C]FucNAc revealed the formation of a new radiolabeled lipid species, not present in the negative control (Fig. 2a). In comparison to C 55 P (Fig. 2a (Fig. 2b), which is consistent with the calculated neutral mass for undecaprenyl-pyrophosphoryl-D-FucNAc (lipid I cap ) of 1113.7162. Moreover, the GT-B type 35 glycosyltransferase CapL was able to use lipid I cap as acceptor substrate and to catalyze the addition of the second 14  Conversion to the ultimate lipid-linked CP precursor lipid III cap was achieved by the attachment of a 14 C-D-ManNAcA residue to lipid II cap , catalyzed by the glycosyltransferase CapI (Fig. 2a, lane 5; Rf = 0.53). Mass spectrometric analysis of lipid III cap did not yield any signal, presumably due to reduced ionization with addition of further sugar moieties. However, in line with the altered migration behavior of the different CP intermediates on TLC (Fig. 2a, lanes 3-5) resulting from the consecutive sugar residue addition, the lantibiotic nisin formed extraction-stable complexes with all C 55 P-containing CP precursors further validating their identity (Fig. 2a, lanes 6-8). Taken together, these findings confirm the proposed functions of the different glycosyltransferases (GT) and demonstrate that CapM functions as priming GT, initiating the synthesis of membranebound CP precursors by coupling the first sugar moiety to the lipid carrier C 55 P.
CapM is homologous to a wide range of GTs from various pathogens ( Supplementary Figure 3a), which catalyze the transfer of UDP-activated sugars to C 55 P 36,37 . Prediction of transmembrane topology (http://www.cbs.dtu.dk/services/TMHMM/) and comparison with hydrophobicity plots (ProtScale 38 ) of homologous GTs predicts that CapM is anchored to the cytoplasmic membrane by a 28-amino acid long α-helical transmembrane domain that is linked to a large C-terminal catalytic domain located in the cytoplasm (Supplementary Figure 3b). In contrast to the in silico predicted single-pass transmembrane (TM) helix geometry, structural analysis of the homologous PglC of Campylobacter concisus suggests that the TM segment adopts an unusual architecture 37  Interestingly, CapM was inhibited in the presence of tunicamycin (Supplementary Figure 4a), a nucleoside antibiotic shown to inhibit members of the UDP-HexNAc-1-phosphate transferase family (i.e., WecA, MraY or TarO) through competitive binding to the DDxxD Mg 2+ -binding site 39 . Tunicamycin had an IC 50 of 129 µM for CapM, which is in the range determined for PglC of C. jejuni (IC 50 = of 100 μM) 40   architectures and catalytic sites. Since we observed a major impact of anionic phospholipids (dioleoylphosphatidylglycerol and cardiolipin) on the enzymatic activity of CapM, we further investigated the effect of different detergents. However, CapM showed lower activity in the presence of detergents or DMSO indicating that the enzyme preferred the native phospholipid environment (Supplementary Figure 4b).
CapAB phosphorylation positively modulates CapM and CapE activity. Full-length integral membrane proteins CapA1 and CapA2 and cytoplasmic kinases CapB1 and CapB2 were overexpressed as His 6 -tag fusion proteins, purified by Ni-NTA chromatography (Supplementary Figure 1), and utilized to reconstitute CapAB tyrosine kinase activity in vitro in the presence of γ-labeled [ 33 P]ATP. CapB2 autophosphorylation was effectively activated by full-length CapA1 or CapA2 (Fig. 3a) producing a single radiolabeled band with an apparent molecular mass of 25.3 kDa, corresponding to phosphorylated CapB2 (lanes 3 & 4). In contrast, the negative control in which a CapA activator was absent (Fig. 3a, lane 1 and 2, Supplementary  Figure 5a) showed no autophosphorylation. In line with previous findings 16 Fig. 3a, lane 7). A full-length CapA2B2 fusion protein, which was constructed for reasons of comparison, showed comparable autokinase activity (lane 8). This finding contradicts recent suggestions that CapB1 might be a pseudokinase devoid of catalytic activity 41 . CapA1B1-mediated phosphotransfer on putative target proteins involved in CP biosynthesis revealed tyrosine phosphorylation of the glycosyltransferase CapM (21 kDa) and the dehydratase CapE (38.6 kDa) (Fig. 3b, Supplementary Figure 5b). In contrast, recombinant CapD, CapN, CapF, CapG or CapL were not phosphorylated (Fig. 3b, Supplementary Figure 5c). The opposing PHP class phosphatases CapC1 and CapC2 were able to antagonistically dephosphorylate CapB kinase and target proteins (Supplementary Figure 5d).
Phosphorylation of CapM by the CapA1B1 fusion increased lipid I cap synthesis 4-fold (Fig. 3c), showing that CapAB-mediated signaling stimulates the priming step of CP biosynthesis in S. aureus by enhancing the rate of CapM-catalyzed glycosyl transfer and thus likely controlling the consumption of the shared carrier C 55 P.
In silico phosphorylation site prediction (NetPhos 3.1) 42 identified tyrosines 75 and 157 as putative phosphosites in CapM (Fig. 3d). The highly conserved Tyr157 appears to be the primary phosphorylation site in CapM, since tyrosine phosphorylation and the concomitant stimulatory effect on the catalytic activity was completely abolished in the CapM_Y157F mutant (Supplementary Figure 6a,b). In contrast, changing Tyr75 to Phe had only a minor effect on the protein's ability to be catalytically activated by CapA1B1-mediated phosphotransfer. Both mutant proteins retained catalytic activity comparable to wild-type CapM in the absence of CapA1B1 (Supplementary Figure 6b) showing that Tyr157 represents the crucial regulatory CapAB phosphorylation site on CapM.
The CapA1B1-mediated stimulation of CapM activity is also reflected by enzyme kinetics, revealing a Km value of 2,309 ± 311.8 µM for UDP-D-FucNAc and a V max value of 0.5559 ± 0.02645 pmol min −1 μg −1 in the absence of CapA1B1 and a lowered K m value of 996.2 ± 136.2 µM and an increased V max of 1.0688 ± 0.0358 pmol min −1 μg −1 in the presence of CapA1B1 (Supplementary Figure 6c). Moreover, a CapM phosphomimetic in which Tyr157 was exchanged to Glu displayed a K m value of 1383 ± 267.3 µM and a V max value of 0.8625 ± 0.04708 pmol min −1 μg −1 in the absence of CapA1B1, substantiating the stimulatory impact of CapM phosphorylation on Tyr157 (Supplementary Figure 6c).
In vitro kinase assays further identified CapE as a CapA1B1 target (Fig. 3e). Tryptic fragments of in vitro phosphorylated CapE were analyzed by nanoscale liquid chromatography coupled to tandem mass spectrometry (nanoLC-MS/MS), and Tyr76 of the CapE protein was found to be phosphorylated (Supplementary Figure 7a). This residue is located in proximity to the conserved TGxxGxxG motif required for cofactor binding and to the MxxxK catalytic site 43 (Fig. 3f). Three other phosphotyrosine residues (Tyr290, Tyr293 and Tyr329), were mapped to a C-terminal loop of CapE ( Fig. 3f; Supplementary Figure 7a). In CapE in vitro assays with active CapAB, the conversion of UDP-D-GlcNAc was increased by 40%, demonstrating that the CapE catalyzed reaction is positively modulated through phosphorylation (Fig. 3e).
The role of the potential regulatory CapAB phosphorylation sites was probed in a site-directed mutagenesis study. As determined by CE quantification, replacing either Tyr76 of CapE or the three C-terminal tyrosine residues Tyr290, Tyr293, and Tyr329 with Phe diminished the activating effect of CapA1B1 on CapE. The stimulatory effect of CapA1B1 on the conversion of UDP-GlcNAc was completely abolished when all four tyrosine residues were mutated to Phe, confirming their role as regulatory phosphorylation sites on CapE (Supplementary Figure 7b).
CapA1 is a dual-function phosphodiesterase/kinase activator protein. When purified lipid I cap was incubated in the presence of CapA1, an additional lipid band was detected by TLC analysis (Fig. 4a), which was not visible in the negative control with heatinactivated CapA1, indicating enzymatic conversion of the first CP lipid precursor. Intriguingly, the reaction product displayed an identical R f value as the lipid carrier C 55 P. MALDI-TOF MS analysis determined a mass of m/z 845.610 (negative mode; [M-H] − ) for the purified lipid product, confirming that CapA1 is able to catalyze the cleavage of the pyrophosphate linkage within the lipid I cap intermediate to release C 55 P (Fig. 4b). Of note, CapA1 was unable to cleave the peptidoglycan precursor lipid II PG , but was able to hydrolyze the WTA precursor lipid III (Supplementary Figure 8). The facts that CapA1 interacts with lipid-linked CP precursors, and exhibits phosphodiesterase activity towards lipid I cap , implies additional functions for this protein in CP biosynthesis, polymerization and attachment processes that go beyond being a mere "transmembrane activator". Importantly, no cleavage of lipid I cap or lipid III WTA was observed with the CapA1 paralogue CapA2 (Fig. 4a), revealing functional differences between the two proteins.
CapA1 is anchored to the cytoplasmic membrane via two transmembrane domains flanking an outside loop comprising 130 amino acids (Fig. 4c), which likely represents a dual function sensory/catalytic domain involved in recognition and processing of membrane-bound CP precursors.
As evidenced in previous research, CapA1 is crucial for efficient capsule formation in S. aureus 44 . Corroborating these studies, complementation with pCapA1 in trans enhanced CP production in the serotype 8 strain MW2 (Fig. 4d). Strain MW2 carries a frameshift mutation in capA1 (cap5A) that results in expression of a truncated version (171 aa) of the full-length gene product (222 aa) 45 . In contrast, deletion of capA2 in S. aureus Newman did not affect in vivo CP production (Fig. 4d). Similarly, the deletion of cognate CapB kinases revealed, that capB1 but not capB2 is required for CP production (Supplementary Figure 7c). In spite of the functional redundancy and overlapping protein target specificities of CapA1B1 and CapA2B2 observed in our in vitro phosphorylation assays, the two kinase complexes clearly have distinct roles in cell physiology.
In contrast to CapA1, CapA2 is not encoded within the cap5 operon, and does not exhibit phosphodiesterase activity towards lipid I cap , which led us to conclude that CapA2 has a distinct function in the biosynthesis of the bacterial cell envelope. More recently, the S. aureus exopolysaccharide poly-N-acetyl-β-(1,6)glucosamine (PNAG; also referred to as polysaccharide intercellular adhesin) has been reported to bind to the CapA2 receptor loop 46 , suggesting that PNAG might represent a molecular signal detected by CapA2. We therefore investigated the influence of purified PNAG and CP5 on the autophosphorylation rate of the CapAB complexes. As revealed by in vitro kinase assays, CapA2-induced autophosphorylation of CapB2 was inhibited in the presence of PNAG, as well as in the presence of CP5, in a concentration-dependent manner (Supplementary Figure 9). In contrast, CapA1-induced phosphorylation of CapB2 was not diminished, indicating that the observed modulatory effect of exopolysaccharide molecules is mediated via an interaction with the transmembrane activator protein CapA2, and not via a direct inhibitory interaction with the protein kinase CapB2. This finding further substantiates differential roles for CapA1 and CapA2 in cell envelope biosynthesis.
LcpC catalyzes the ligation of CP to lipid II. Structural analysis and phenotypic studies suggest that members of the LCP protein family catalyze the transfer of undecaprenyl-linked intermediates onto the C6-hydroxyl function of N-acetylmuramic acid in PG, thereby promoting attachment of WTA and CP in Gram-positive bacteria 20,47 . S. aureus encodes three LCP enzymes with semiredundant functions 21 . As deduced from knockout mutant studies, CP attachment is preferentially catalyzed by LcpC, though LcpA and LcpB may partially compensate for the loss of LcpC 21 .
To investigate the proposed role of LcpC in vitro purified [ 14 C] lipid I cap was incubated with purified LcpC protein and lipid II PG as a potential acceptor substrate. In this setup, LcpC was able to catalyze cleavage of the donor substrate lipid I cap and catalyze attachment of the phosphoryl-sugar moiety to the ultimate PG precursor lipid II (Fig. 5a), resulting in a reaction product (LII PG − 14C CP) that migrates slower on TLC. Identical results were obtained when [ 14 C]-labeled lipid II PG and non-labeled lipid I cap were used as the reaction substrates (Fig. 5b), further verifying the identity of the ligation product. Quantitative analysis revealed that~50% of the CP lipid precursor was attached to lipid II (Fig. 5c). Strikingly, the LcpC-mediated transferase reaction was significantly enhanced when CapA1 was included in the reaction mixture, indicating cooperative action of the transmembrane activator/phosphodiesterase protein and LcpC (Fig. 5c).
CapA1 alone was unable to catalyze the attachment of the CP precursor (Supplementary Figure 10a). Of note, LcpC was able to catalyze hydrolytic cleavage of lipid I cap , but not of lipid II PG , a reaction that would likely be deleterious in vivo ( Supplementary  Figure 10b, c).
Interestingly, PBP2 added posterior to the LcpC reaction efficiently catalyzed the transglycosylation of lipid II PG -CP (Fig. 5a, b). The resulting polymerized PBP reaction product is not extracted from the reaction mixture, thus lipid bands vanish from the TLC. The antibiotic moenomycin, known to inhibit the PBP2 catalyzed transglycosylation, fully blocked the polymerization reaction (Fig. 5b). Importantly, LcpC was further able to efficiently transfer the disaccharide of lipid II cap and the trisaccharide of the ultimate CP lipid intermediate lipid III cap to lipid II PG , as evidenced by the altered migration of the respective ligation products (Rf = 0.26 and Rf = 0.22, respectively). Efficient processing of ligation products by PBP2 shows that CP attachment could occur in parallel to PG assembly (Fig. 5d).
Interestingly, antisense-RNA mediated depletion of CapA1 in a triple Δlcp mutant was lethal, indicating that CapA1 hydrolysis of C 55 P-coupled CP and WTA precursors contributes to rescue S. aureus from the accumulation of toxic intermediates (Supplementary Figure 10d) and further supports a functional link between CapA1 and LcpC.  Capsule biosynthesis is negatively controlled by PknB. As evidenced in previous research, the eukaryotic serine/threonine kinase (ESTK) PknB of S. aureus is involved in the regulation of cell envelope biogenesis processes [48][49][50] and was more recently shown to sense lipid II PG 51 . Testing CapM in an in vitro kinase assay showed PknB-mediated phosphorylation of the priming GT (Fig. 6a). Moreover, quantitative analysis of the in vitro CapMcatalyzed reaction revealed an up to 30% decrease in lipid I cap synthesis when purified PknB kinase and ATP were included in the reaction mixture, compared to a control reaction containing heat-inactivated PknB, suggesting that ESTK signaling negatively controls the activity of CapM (Fig. 6b). LC/MS-analysis revealed PknB-mediated phosphotransfer onto Thr67, Thr128 and Thr134 of CapM (Supplementary Figure 11). Thr134 is highly conserved among homologous bacterial glycosyltransferases, located in proximity of the membrane interface and the amphipathic helices D and I (Supplementary Figure 3c) and may thus be the most promising candidate for a regulatory PknB-phosphorylation site on CapM. Intriguingly, the kinase protein CapB1 was also identified as potential phosphorylation target of PknB (Fig. 6c), indicating the possibility of kinase cross-talk. To further explore this finding, the autophosphorylation activity of the CapA1B1 complex was assessed in the presence of PknB. The overall phosphorylation intensity of CapA1B1 was decreased by 34% in the presence of active PknB (Fig. 6d), compared to the heat-inactivated negative control, showing that PknB-mediated serine/threonine phosphorylation inhibits CapB1 autophosphorylation on tyrosine. Thr8 was identified as the specific phosphorylation site on CapB1 by LC/MS analysis (Supplementary Figure 12a) and a CapA1B1 fus phosphomimetic in which this Thr was exchanged to Glu resulted in a significantly reduced CapB1 autophosphorylation (Supplementary Figure 12b, c), suggesting that PknB-mediated phosphorylation interferes with the CapB1 activation by CapA1 41,52 . In contrast, an exchange of Thr8 to Ala did not affect CapA1B1 fus autophosphorylation (Supplementary Figure 12b,c).
To determine whether the observed inhibitory effect of PknB in vitro would translate into an in vivo effect, we compared CP5 production in S. aureus Newman and in an isogenic pknB deletion mutant. Consistent with the proposed role of PknB as negative regulator of CP biosynthesis, the amount of cellassociated CP5 was 5-fold higher in the pknB mutant compared to the parental strain and complementation of the ΔpknB mutant in trans restored CP production to wild-type level (Fig. 6e).
The in vitro and in vivo data reveal that PknB signaling serves to reduce CapM GT activity, as well as CapA1B1 autokinase activity, allowing for a shutdown of CP production likely ensuring a sufficient supply of precursors for PG formation, and thus maintenance of cell wall architecture and function.

Discussion
Biosynthesis of the S. aureus CP shows similarity to the syntheses of PG and WTA, in that all pathways share a pool of essential precursors, i.e. the lipid carrier undecaprenyl-phosphate and UDP-D-GlcNAc 1,53 . Since the availability of C 55 P within the bacterial cell is limited 22 , distribution and prioritization for the alternate metabolic pathways need to be tightly controlled in time and space to ensure bacterial viability. Reversible protein phosphorylation appears an elegant mechanism to ensure a coordinated and temporally controlled flux of intimately shared cell envelope metabolites. Several biosynthetic enzymes involved in polysaccharide production have been identified as endogenous substrates of bacterial tyrosine kinases, for instance the UDP-glucose 6-dehydrogenases TuaA and YwqF of Bacillus subtilis 54 , the homologous E. coli UDPglucose dehydrogenase Ugd 55 and the UDP-ManNAc dehydrogenase CapO 17 involved in S. aureus CP biosynthesis.
The fact that CapO, involved in the third CP reaction cascade (Fig. 1), is a regulatory target of the CapAB kinase complex 17 , led us to assume that additional checkpoints within the biosynthetic pathway are also controlled by CapAB, which turned out to be the case for CapE and CapM. Four tyrosine residues in the dehydrogenase CapE were identified as the specific phosphorylation sites, with Tyr76, Tyr290 and Tyr293 being widely conserved among homologous proteins from different pathogens. Interestingly, these residues are located in strategic functional regions of CapE: next to the cofactor binding site, in close proximity to the active site, and within a recently described mobile loop (Fig. 3f). Crystal structure analysis identified this mobile latch to connect two CapE protomers within the hexameric complex (trimer of dimers) and showed that the latch of one dimerization partner is associated with the substrate-binding domain of the contiguous CapE monomer, and vice versa, suggesting that this mobile loop is involved in regulating the access of the UDP-D-GlcNAc substrate to the active site 56 . Since contacts to Tyr290 and Tyr293 appear to be involved in the interaction, phosphorylation might induce conformational changes that facilitate access of UDP-D-GlcNAc to the active site, thereby increasing CapE enzymatic activity.
For the monotopic GT CapM, we identified the conserved tyrosine residue at position 157 as the CapA1B1 regulatory phosphorylation site which is in good agreement with the work of Minic et al. (2007) and supports the hypothesis that phosphorylation of the equivalent site in the S. thermophilus The error bars represent the ± SD from three biological replicates. Statistical significance was analyzed by an unpaired t-test (**p < 0.05). c CapB1 is phosphorylated by PknB. Putative target proteins (3 µg) were incubated with PknB in the presence of [ 33 P]ATP, and reactions were analyzed by SDS-PAGE and phosphoimaging. d Quantification of CapB autophosphorylation in the absence and presence of PknB. Experiments were performed in triplicate. The error bars represent the ± SD from three biological replicates. Statistical significance was analyzed by an unpaired t-test (***p < 0.005). e ELISA-based quantification of CP production in a ΔpknB deletion strain, the corresponding parental strain S. aureus Newman (WT) and complementation of the ΔpknB mutant in trans. Experiments were performed in triplicate. The error bars represent the ± standard error of the mean (SEM) from three biological replicates. Statistical significance was analyzed by an unpaired t-test (***p < 0.005) phosphogalactosyl-transferase EpsE results in activation in vivo 57 . A recent structural analysis showed, that the equivalent tyrosine residue in PglC is involved in a hydrogen-bonding network that establishes intramolecular interactions between helices A, F and G 37 (Supplementary Figure 3c), suggesting that the phosphorylated amino acid side chain may affect relative strengths of hydrogen bonds and critical interactions in CapM. Current hypotheses suggest a multifaceted role for BY-kinases in bacterial exopolysaccharide production. BY-kinase signalling may not only regulate the catalytic activity of polysaccharide biosynthetic enzymes, but also ensure the correct cellular localization of protein targets 58 . Moreover, it has been suggested that BY-kinases may interact with the export and polymerization machinery to control the level of CP production and/or CP chain length 52,59 . However, the exact mechanism through which this control is exerted, as well as the molecular signal sensed, remains enigmatic so far.
The phosphodiesterase activity of CapA1 described here is in good agreement with previous studies demonstrating the release of CP and WTA into the culture supernatant in LCP deletion mutants of different species [60][61][62][63] . An lcpC deletion mutant of S. aureus accumulated CP in the culture supernatant, indicating that LcpC is the key LCP enzyme for attachment of CP to the peptidoglycan in S. aureus 21 .
This study provides biochemical proof that LcpC catalyzes the attachment of CP to PG. Importantly, CP lipid precursor cleavage and transfer of the phosphosugar moiety were found to be enhanced in the presence of CapA1, suggesting that the transmembrane activator cooperates with LcpC by forming an interaction complex, thereby modulating the attachment of CP to the cell wall precursor. In line with this interpretation, Toniolo et al. (2015) reported that the extracellular domain of the CapA1 homolog CpsC of Streptococcus agalactiae may modulate LCPmediated CP attachment in response to the phosphorylation state of the BY kinase CpsD 64 .
The natural PG acceptor substrate of LcpC is elusive; possible acceptor structures include the ultimate PG precursor lipid II, as well as "nascent" and crosslinked PG. More recently, LcpA and LcpB were shown to attach a shortened, soluble (C 20 P) WTA precursor to a preformed "nascent PG" oligomer in vitro 65 , but not to a lipid II PG mimetic lacking the natural undecaprenyl tail. In contrast, the biochemical analyses presented here clearly demonstrate that CP is efficiently attached to lipid II PG by the LcpC enzyme, and that PBPs are able to polymerize the resulting reaction product. With regard to the membrane localization of both, substrates and enzymes, and the interdependence and intimate connection of the enzymatic machineries, lipid II PG seems the most plausible acceptor.
CP lipid intermediates were further hydrolyzed by LcpC in the absence of the acceptor substrate as predicted by structural analysis of Cps2A of S. pneumoniae 20 , which was not observed for LcpA and LcpB proteins 65 . In vitro, all three CP lipid intermediates were efficiently processed by LcpC, although the proximal full-length undecaprenyl-pyrophosphoryl-linked sugar moiety appears sufficient for CP precursor recognition. Likewise, the first C 55 P-linked sugar unit of the O-antigen repeat unit contains the recognition information necessary for catalysis by the O-antigen ligase WaaL 66 .
The biosyntheses of the cell envelope components PG, WTA and CP have to be coordinated in time and space, since the enzymatic machineries and their individual components are functionally related to each other or intimately connected 1,53 . The complex interplay between different cell envelope pathways becomes evident upon inhibition of individual biosynthetic steps. In analogy to the WTA biosynthesis in S. aureus, where late stage biosynthetic genes have been shown to be conditionally essential ("essential gene paradox"; 67) , Yother and co-workers reported that deletion of the late stage CP biosynthesis genes, responsible for side chain assembly, polymerization or transport (cps2K, cps2J, and cps2H) in Streptococcus pneumoniae, is lethal 68 . Since the capsule is not required for cell viability per se, the damage to the cell envelope is most likely due to an inhibitory effect on PG biosynthesis, resulting from sequestration of CP lipid intermediates probably reducing the undecaprenyl-phosphate level to a critical point where PG synthesis is affected. Since we found CapA1 to hydrolyze CP lipid intermediates even in the absence of LCP proteins and that depletion of CapA1 in a S. aureus Δlcp triple mutant was lethal, we conclude that CapA1-mediated cleavage of dead-end lipid-linked products might serve as a rescue mechanism counteracting depletion of C 55 P to critical levels.
The ESTK PknB is another important player in the orchestration of cell wall polymer biosynthesis and has been implicated in coordinating PG cross-wall formation, autolysis and cell division in S. aureus [48][49][50] . Our findings that PknB downregulates CapM and with the BY-kinase CapA1B1 operating in an antagonal manner underpins the importance of CapM as crucial enzymatic checkpoint. Moreover, in vitro kinase assays indicate that PknB-mediated phosphotransfer onto CapB1 modulates the activity of the CapA1B1 BY-kinase complex itself, thus identifying another activity by which PknB may exert control on CP biosynthesis. Cross-phosphorylation of ESTKs and BYkinases has also been reported in B. subtilis 69 . The CapB homolog PtkA was identified as in vitro phosphorylation target of the ESTK PrkC 69 . Moreover, PtkA autophosphorylation in vivo was strongly enhanced in a ΔprkC strain, suggesting that PtkA BY-kinase activity is negatively regulated by PrkC-mediated phosphotransfer 70 .
Corroborating the in vitro results, CP production was substantially elevated in a pknB deletion mutant of S. aureus. Besides decreasing the activity of CP biosynthesis proteins by direct phosphorylation, PknB kinase activity may serve to increase the cellular concentration of UDP-MurNAc-pentapeptide and lipid II PG 71 , cell wall metabolites that were identified as inhibitors of CapD enzymatic activity 19 . Moreover, PknB signaling was reported to influence the transcription of the cap5 gene cluster 72 , suggesting that this sensor kinase is part of a multilayered signaltransduction network that coordinates CP formation in space and time.
The finding that CP and PG production are inversely controlled by PknB further substantiates a key role for PknB in the coordinate regulation of Gram-positive cell surface glycopolymers. In exponentially growing cells, PknB-mediated Ser/Thr phosphorylation positively controls PG biosynthesis, while exerting negative control on CP production (Fig. 7).
With respect to biosynthesis of the S. aureus CP, the CapA1B1 complex may be particularly important for short-term regulation. More recently, Elsholz and co-workers showed that exopolysaccharide (EPS) production in Bacillus subtilis is subject to a positive feedback loop that ties the synthesis of the polymer to its own concentration 46 . Regulation of EPS synthesis is mediated by the EpsAB tyrosine kinase complex, whereby direct recognition of EPS by the extracellular domain of the membrane component EpsA seems to result in stimulation of kinase activity 46 .
The activator/phosphodiesterase CapA1 directly and specifically interacts with lipid-bound CP precursors, and CapA1B1 signaling stimulates the activity of CP biosynthetic enzymes, triggering enhanced CP production in late-exponential and stationary phase (Fig. 7).
S. aureus PknB specifically interacts with lipid II PG , likely responding to cellular pool levels of the ultimate peptidoglycan precursor 51 . It is thus conceivable that in order to balance the pool levels of the lipid II PG acceptor and the CP donor substrate, which are both built on the lipid carrier C 55 P, PknB and CapA1B1 antagonistically control biosyntheses to ensure vital cycling of the essential C 55 P lipid carrier. In addition, CapA1 might contribute to the correct localization of the CP assembly machinery, as proposed for S. agalactiae 64 , which has to integrate precursor transport and CP polymerization. This interaction protein network thus appears to contribute to the spatiotemporal coordination of CP biosynthesis with PG synthesis (Fig. 7), rendering proteins such as CapA1 and LcpC, exposed at the cell surface and devoid of eukaryotic homologs, potentially attractive antibacterial or antivirulence targets.

Methods
Strains. S. aureus Newman NCTC8178 was maintained on blood or lysogeny broth (LB; Oxoid) agar. Escherichia coli strains BL21 and C43 used for overexpression of recombinant His 6 -tag fusion proteins were maintained on LB-agar plates containing 50 µg ml −1 ampicillin or 25 µg ml −1 kanamycin. CP5 production was quantified from S. aureus cultivated on Columbia medium supplemented with 2% NaCl and 1.5% agar. The Newman ΔpknB mutant was kindly provided by Dr. Knut Olsen (Wuerzburg, Germany), and RN4220 carrying pCapA1 was provided by Dr. Gabriele Bierbaum (Bonn, Germany) 44 . S. aureus strain MW2 (NRS123) and strains from the Nebraska transposon mutant library 73 were obtained from the Network on Antimicrobial Resistance in Staphylococcus aureus program, which was supported by the National Institute of Allergy and Infectious Diseases of the National Institutes of Health (contract HHSN272200700055C). The Nebraska library is comprised of derivatives of the serotype 5 strain USA300 LAC (cured of three plasmids) in which individual nonessential genes were disrupted by the insertion of the mariner transposon (Tn) bursa aurealis and includes mutants in capA2 (NE1286), capB2 (NE75), capA1 (NE302a), and capB1 (NE135). Construction of cap deletion mutants is described in the Supplementary Methods. The pknB and Tn mutants were maintained on tryptic soy agar (TSA) plates with 5 µg ml −1 erythromycin, and pCapA1 was maintained in S. aureus cultivated on TSA with 10 µg ml −1 chloramphenicol. S. aureus Newman ΔlcpABC mutants were kindly provided by Dr. Olaf Schneewind (Chicago, USA) 21 maintained on LB-agar plates. S. aureus strains carrying the plasmid pEPSA5 or the antisense plasmid pEPSA5-capA1AS were maintained on TSA-agar plates supplemented with 34 µg ml −1 chloramphenicol.
The supernatant was incubated with Ni-NTA-agarose slurry (Qiagen, Germany) for 2 h at 4°C under gentle stirring. The mixture was then loaded onto a column support. After washing with lysis buffer, weakly bound material was removed with 10 and 20 mM imidazole. Recombinant proteins were eluted with buffer containing 300 mM imidazole. Five 500-µl fractions were collected each and stored in 30% (v/v) glycerol at −20°C; preparations of CapE were dialyzed against 10 mM potassium phosphate buffer (KPi), pH 7.5 before storage. For PknB purification the pellet was resuspended in lysis buffer containing 1% (v/v) Triton X-100 and the suspension was incubated on ice for 1 h. Cell debris was spun down (20 min, 4°C, 34,000 x g) and the clear supernatant was loaded onto a Strep-tactin agarose column (IBA, Germany). The column was washed twice with lysis buffer and PknB was eluted with Strep-tactin elution buffer (IBA, Germany). Purity of elution fractions was assessed by SDS-Page (NuPAGE; Invitrogen); protein concentrations were measured using Bradford reagent (Bio-Rad, Germany). Expression and purification of CapA1B1 fus , CapE and CapM protein variants were carried out as described for the wild-type proteins.
In vitro syntheses of soluble CP precursors. UDP-D-GlcNAc and cofactors for enzyme assays were obtained from Sigma-Aldrich. Synthesis of the precursor UDP-2-acetamido-2,6-dideoxy-D-xylo−4-hexulose for mass spectrometric analysis was performed with a recombinant truncated version of the enzyme PglF from C. jejuni, as previously described 19 . The PglF reaction product was used as substrate for reconstitution of CapN catalytic activity. For this purpose, PglF reactions were carried out overnight and quenched by heating (5 min, 95°C). CapN catalyzed synthesis of UDP-D-FucNAc was carried out in a total volume of 40 µl. CapN (12 µg) was incubated in the presence of~3 mM UDP-2-acetamido-2,6-dideoxy-D-xylo−4-hexulose, 1.875 mM NADPH, 8% (v/v) Triton X-100 and 10 mM KPi, pH 7.5, for 2 h at 30°C. Synthesis of larger quantities of soluble capsule precursors for MALDI-TOF MS analysis was achieved by a 10-fold upscale of the synthesis reaction.
In vitro modulation of CapE catalytic activity by CapAB. The influence of CapAB-mediated tyrosine phosphorylation on CapE catalytic activity was examined in vitro using purified recombinant proteins. Assays were performed in a total volume of 100 µl containing 17 µg of CapE (or of a CapE protein variants) and 8.5 µg of purified CapAB kinase complex. The enzymes were incubated in the presence of 3 mM UDP-D-GlcNAc and 1 mM ATP, in 10 mM KPi, 10 mM MgCl 2 , pH 7.5 supplemented with 1 mM DTT, for 30 min at 30°C. Reactions were quenched by heating (5 min, 90°C), then subjected to capillary electrophoresis (CE) analysis 19 . Control reactions were performed with heat-inactivated (10 min, 100°C) CapAB protein.
MALDI-TOF mass spectrometric analysis. Soluble capsule precursors were purified by RP18-HPLC 19 , and cleaved C 55 P intermediates were purified by preparative TLC. Lipid spots were visualized using iodine vapour and extracted off the silica plates with methanol. Samples were spotted onto a ground steel MALDI-TOF target plate and allowed to dry at room temperature. Subsequently, each sample was overlaid with 1 µl of matrix (saturated solution of 6-Aza-2-thiothymine in 50% (v/v) ethanol, 10 mM diammonium hydrogen citrate) and air dried at room temperature again. Spectra were recorded either in the reflector negative mode within a mass range from 300 to 3000 Da (soluble capsule precursors) or in the linear negative mode within a mass range from 400 to 3000 Da (lipid intermediates), at a laser frequency of 9 Hz on a Biflex III mass spectrometer (Bruker Daltonics). Data analysis was performed using flexAnalysis software (Bruker Daltonics).
In vitro kinase assays. In vitro BY-kinase assays were carried out in a total volume of 10 µl containing either 0.5 µg of CapA and CapB, or 2 µg of a CapAB protein fusion construct. For identification of protein substrates, 2 µg of a recombinant target protein were added. The proteins were incubated in the presence of 10 μCi γ-labeled [ 33 P]ATP (~300 nM; Hartmann Analytic) in 50 mM Tris-HCl, 10 mM MgCl 2 , pH 7.5 supplemented with 0.5 mM DTT, 0.5 mM EDTA and 10 µM ATP. Assays with CapAB protein variants were carried out as described for the wild-type proteins.
For identification of PknB protein substrates, purified PknB (0.5 µg) was incubated in presence of 3 mM MnCl 2 in the reaction mixture described above. Cross-phosphorylation of CapAB by PknB was assessed analogously.
After 30 min of incubation at 30°C, reactions were stopped by addition of 4x LDS sample buffer (Invitrogen), and analyzed by SDS-PAGE (NuPAGE, Invitrogen). Radioactive protein bands were visualized using a storage phosphor screen in a Storm imaging system (GE Healthcare). A detailed description of the identification of phosphorylation sites by nanoLC-MS/MS is given in the Supplementary Methods.
In vitro phosphatase assays. CapC phosphatase activity was examined in vitro by the addition of CapC1 or CapC2 (2 µg each) with 2 mM MnCl 2 to the CapAB in vitro kinase assays (see above) and subsequent heat-inactivation. After incubation for 1 h at 30°C, phosphatase reaction was stopped by the by addition of 4x LDS sample buffer (Invitrogen), and analyzed by SDS-PAGE (NuPAGE, Invitrogen). Radioactive protein bands were visualized using a storage phosphor screen in a Storm imaging system (GE Healthcare).
In vitro syntheses of lipid I cap , lipid II cap , and lipid III cap . Lipid I cap synthesis was carried out in a total volume of 50 µl containing 5 nmol C 55 P (Larodan, Sweden), 2 mM UDP-D-GlcNAc, 10 mM NADPH, 0.5 mM NADP and 0.6% (v/v) DMSO in 50 mM Tris-HCl, 10 mM MgCl 2 , pH 7.5. The reaction was initiated by the addition of 4 µg of biosynthetic enzymes CapD, CapN and CapM, and incubated for 16 h at 30°C.
To assess the influence of different detergents and phospholipids on CapM activity, newly synthesized UDP-D-FucNAc (described above) was incubated with 4 µg CapM, 5 nmol C 55 P in a reaction mixture containing 50 mM Tris-HCl, 10 mM MgCl 2 , pH 7.5 supplemented with an appropriate amount of triton X-100, n-dodecyl-β-D-maltoside (DDM), n-lauroyl sarcosine (LS), dimethylsulfoxide (DMSO), dioleoylphosphatidylglycerol (DOPG) or cardiolipin (CL). To investigate the modulatory effect of CapAB on CapM activity 4 µg CapAB and 10 mM ATP were added to the CapM reaction mixture. To assess the influence of PknB on CapM activity in vitro, 1 µg PknB was added to the CapM reaction mixture supplemented with 2 mM MnCl 2 and 10 mM ATP.
The second precursor lipid II cap was synthesized in a total volume of 50 µl by incubating the capsule biosynthesis proteins CapE, CapF, CapG and CapL (4 µg each) in the presence of 2 nmol purified lipid I cap and 2 mM UDP-D-GlcNAc, 2 mM NADPH, 0.5 mM NADP and 0.6% (v/v) DMSO in 50 mM Tris-HCl, 10 mM MgCl 2 , pH 7.5.
Lipid III cap was synthesized by incubating 2 nmol purified lipid II cap with CapO, CapP and CapI (4 µg each) and 2 mM UDP-D-GlcNAc, 6 mM NAD + , 0.5 mM dithiothreitol (DTT; Sigma-Aldrich) in 50 mM Tris-HCl, 10 mM MgCl 2 , pH 7.5. Synthesized CP precursors were extracted from the reaction mixture with an equal volume of n-butanol/pyridine acetate, pH 4.2 (2:1, v/v) and analyzed by TLC on silica plates (Merck) according to Rick (chloroform, methanol, water, ammonium hydroxide, 88:48:10:1) 75 . Nisin was added at a molar ratio of 2:1 with respect to the lipid precursor prior to the extraction procedure. Reaction mixtures were extracted and unbound lipids were analyzed by TLC. For mass spectrometric analysis of lipid intermediates see Supplementary Methods.
For synthesis of [ 14 C]-labeled lipid-bound precursors, the assays were further supplemented with 0.333 nmol UDP-D-[ 14 C]GlcNAc (Hartmann Analytic, Germany). Radiolabeled spots were visualized using a storage phosphor screen in a Storm imaging system (GE Healthcare). Non-radiolabeled lipid intermediates were analyzed using PMA staining reagent (2.5% (w/v) phosphomolybdate, 1% (w/v) ceric sulfate, 6% (v/v) sulfuric acid). Isolation of small quantities of CP precursors was achieved by synthesis and subsequent purification via preparative TLC. To this end, lipid spots were visualized using iodine vapour and material was scratched off the silica plates. Lipids were extracted by incubation in 100 µl of methanol for 60 min. Larger quantities of CP precursor were purified using high-performance liquid chromatography (HPLC) over a DEAE-FF (5 ml; GE Healthcare) and eluted in a linear gradient from chloroform/ methanol/ water (2:3:1) to chloroform, methanol, 300 mM ammonium bicarbonate (2:3:1).
CapA1 catalyzed cleavage of lipid-bound CP precursor. For reconstitution of the CapA1 mediated cleavage of the pyrophosphate-linkage of lipid-linked cell wall intermediates, purified CapA1 (8 µg) was incubated with 2 nmol lipid I cap , lipid III WTA, lipid II PG in 10 mM MgCl 2 and 50 mM Tris-HCl, pH 7.5. After incubation for 16 h at 30°C, cleavage products were extracted from the reaction mixture with an equal volume of n-butanol/pyridine acetate, pH 4.2 (2:1, v/v) and analyzed by TLC on silica plates (Merck) according to Rick (chloroform, methanol, water, ammonium hydroxide, 88:48:10:1) and visualized by PMA staining 75 .
In vitro LCP assays. For reconstitution of the CP ligation reaction purified lipid I cap or [ 14 C]lipid I cap was incubated in the presence of LcpC (4 µg), using 2 nmol of purified lipid II PG or [ 14 C]lipid II PG as acceptor substrate in a 50 µl reaction mixture with 0.6% DMSO, 18 µg DOPG, 10 mM MgCl 2 in 50 mM MES buffer, pH 5.5. Synthesis and purification of the acceptor substrate lipid II is described in the supplementary methods. If indicated, assays additionally contained 4 µg of CapA1. After incubation for 16 h at 30°C, samples were analyzed by TLC, followed by phosphoimaging. Polymerization of LcpC reaction products were catalyzed by the subsequent addition of PBP2 with 2 mM CaCl 2 to the LCP reaction mixture, followed by incubation at 30°C for 1 h. Moenomycin (MOE) was used to block the PBP2 catalyzed reaction and added at a final concentration of 10 µM. For evaluation of LcpC mediated hydrolysis of CP precursor in the absence of the acceptor substrate, 2 nmol of the individual lipid-linked precursor was incubated with 4 µg LcpC, 18 µg DOPG, 10 mM MgCl 2 and 0.6% DMSO in 50 mM MES buffer, pH 5.5. After incubation for 16 h at 30°C, samples were analyzed by TLC and visualized by PMA staining.
Reporting summary. Further information on experimental design is available in the Nature Research Reporting Summary linked to this article.