Phosphoglycerol-type wall- and lipoteichoic acids are enantiomeric polymers differentially cleaved by the stereospecific glycerophosphodiesterase GlpQ

The cell envelope of Gram-positive bacteria generally comprises two types of polyanionic polymers, either linked to peptidoglycan, wall teichoic acids (WTA), or to membrane glycolipids, lipoteichoic acids (LTA). In some bacteria, including Bacillus subtilis strain 168, WTA and LTA both are glycerolphosphate polymers, yet are synthesized by different pathways and have distinct, although not entirely understood morphogenetic functions during cell elongation and division. We show here that the exo-lytic sn-glycerol-3-phosphodiesterase GlpQ can discriminate between B. subtilis WTA and LTA polymers. GlpQ completely degrades WTA, lacking modifications at the glycerol residues, by sequentially removing glycerolphosphates from the free end of the polymer up to the peptidoglycan linker. In contrast, GlpQ is unable to cleave unmodified LTA. LTA can only be hydrolyzed by GlpQ when the polymer is partially pre-cleaved, thereby allowing GlpQ to get access to the end of the polymer that is usually protected by a connection to the lipid anchor. This indicates that WTA and LTA are enantiomeric polymers: WTA is made of sn-glycerol-3-phosphate and LTA is made of sn-glycerol-1-phosphate. Differences in stereochemistry between WTA and LTA were assumed based on differences in biosynthesis precursors and chemical degradation products, but so far had not been demonstrated directly by differential, enantioselective cleavage of isolated polymers. The discriminative stereochemistry impacts the dissimilar physiological and immunogenic properties of WTA and LTA and enables independent degradation of the polymers, while appearing in the same location; e.g. under phosphate limitation, B. subtilis 168 specifically hydrolyzes WTA and synthesizes phosphate-free teichuronic acids in exchange.


47
Bacteria are covered by a complex multilayered cell envelope positioned external to the cell membrane,  (11,12) and are responsible for the generally high phosphate content of Gram-positive cell walls (4,13). It 64 was shown that WTA can serve as phosphate storage, allowing B. subtilis to continue growth under

80
In some Gram-positive bacteria, including B. subtilis 168 as well as in S. epidermidis and S. 81 lugdunenis, both LTA and WTA are glycerophosphate polymers (17,18,27,28). Nevertheless, they are 82 synthesized by distinct routes (9,10,29). Figure 1 summarizes the biosynthesis pathways of WTA and LTA 83 of B. subtilis 168. WTA is synthesized from CDP-glycerol in the cytoplasm and the polymer is then flipped 84 outwards (Fig. 1A). In contrast, LTA is synthesized from the precursor phosphatidylglycerol (PG), 85 generated via diacylglycerol(DAG)-CDP and PG-phosphate (PGP). The PG precursor is subsequently 86 translocated and then polymerized on the outside of the cell (Fig. 1B). An important finding has been that 87 the glycerophosphate in the precursors of WTA and LTA have different stereochemistry (17). CDP-88 glycerol has a sn-3-configuration, and in contrast to this, the free glycerol phosphate of PG has a sn-1-89 configuration. The prochirality of glycerol leads to two 3-phosphate products: according to convention 90 (stereochemical numbering: sn-nomenclature), L-glycerol is the configuration that determines the 91 numbering of sn-glycerol phosphates (Fig. 1C). The use of different precursors and the 92 compartmentalization of WTA and LTA synthesis allows to differentially regulate the production of the 93 two polymers, which is important in order to fulfil particular roles in cell envelope integrity and distinct 94 morphogenetic functions during cell cycle and growth phases (18). In contrast, how WTA and LTA execute 95 their distinct physiological functions is not obvious, as they are both present in the same cell envelope 96 compartment.

97
Modifications of the polyols by alanylation and glycosylation are important means to alter the 98 physiological properties of WTA and LTA and also affect recognition by the innate immune system 99 (6,30,31). D-alanylation adds positive charge (free amino groups) to the polyol-phosphate polymers, 100 thereby rendering the anionic character and as a consequence the binding properties (10,32). The 101 multienzyme complex DltABCD is responsible for adding D-Ala modifications onto LTA on the outer 102 4 leaflet of the cell membrane, and indirectly also onto WTA ( Fig. 1) (32)(33)(34)(35). LTA and WTA can also be 103 α-, or β-glycosylated, which severely increases the stability of these polymers against alkaline hydrolysis 104 (9,10). In B. subtilis the enzyme TagE transfers α-glucosyl residues from UDP-glucose onto preformed 105 WTA within the cytoplasm and constitutes the only WTA glycosylating enzyme in this bacterium (Fig.

106
1A) (36,37). Although WTA glycosylation usually occurs prior to the translocation of the polymer across 107 the cell membrane, it was recently proposed that it may also occur after translocation in Listeria 108 monocytogenes (37,38). Alike alanylation, glycosylation of LTA generally occurs, along with synthesis,

112
Besides synthesis, also the turnover of WTA and LTA needs to be differentially regulated, which 113 so far has not been explored in much detail. Recently, the exo-acting sn-glycero-3-phosphate 114 phosphodiesterase GlpQ, along with an endo-acting phosphodiesterase PhoD, has been implicated in the 115 degradation of WTA during phosphate starvation (41). However, apart from WTA degradation during 116 adaptation to phosphate starvation, turnover of WTA likely occurs also along with the turnover of PGN of 117 the cell wall in B. subtilis and other Gram-positive bacteria (42-44). Since strains of B. subtilis lacking 118 both WTA and LTA are not viable, the simultaneous degradation of both polymers would be detrimental 119 (18). We thus wondered how differential degradation of WTA and LTA by hydrolases ("teichoicases") is 120 possible. Previous studies with the glycerophosphodiesterase GlpQ of B. subtilis as well as orthologous 121 enzymes from Escherichia coli and S. aureus (amino acid sequence identities of 29 and 54 %, respectively) 122 have revealed strict stereospecificity for glycerophosphodiesters harbouring sn-glycerol-3-phosphoryl 123 groups, e.g. produced by phospholipases from membrane phospholipids (41,(45)(46)(47). Accordingly, 124 phosphatidylglycerol or lysophosphatidylglycerol, which harbour only free sn-glycerol-1-phosphoryl ends 125 are not hydrolyzed by GlpQ and also bis(p-nitrophenyl)-phosphate, a chromogenic substrate for other 126 phosphodiesterases, is not cleaved by GlpQ (45,46). Intriguingly, LTA of S. aureus was found to be not a 127 substrate of GlpQ, which however could be due to phosphoglycerol backbone modifications (47). In 128 contrast, the enzyme shows broad substrate specificity with respect to the alcohol moiety and can hydrolyse 129 a variety of different phospholipid head groups, such as glycerophosphocholine, 130 glycerophosphoethanolamine, glycerophosphoglycerol, and bis(glycerophospho)glycerol (41,45,47).

131
So far, differential cleavage of WTA and LTA polymers by GlpQ has not been examined in detail.

132
In this work, we show that the stereospecific sn-glycerol-3P phosphodiesterase GlpQ acts as an exo-lytic 133 hydrolase that sequentially cleaves off sn-glycerol-3-phosphate (Gro3P) entities from the exposed end of 134 WTA, however, it is unable to hydrolyse intact LTA. Thereby we provide biochemical evidence that these and 1517 min -1 , were determined for B. subtilis GlpQ (41,45,47). We confirmed the stereospecificity of 146 recombinantly expressed, B. subtilis GlpQ for sn-glycero-3-phosphoryl substrates and determined the 147 enzyme's stability and catalytic optima, using sn-glycero-3-phosphocholine (GPC) as substrate 148 (Supporting Information, Fig. S1). Our analysis revealed that GlpQ is rather temperature sensitive. It 149 readily loses stability at temperatures above 30°C, e.g. within 30 min at 37°C more than 50% of its activity 150 was lost. At the same time however, the enzymatic turnover steadily increases with temperature up to an 151 optimum at 55°C and about half maximum activity at 30°C (Supporting Information, Fig. S1B).

152
Furthermore, the enzyme was shown to be stable over a remarkably wide pH range between 2 to 10, but 153 has a very narrow optimum at pH 8.0 (Supporting Information, Fig. S1B). We thus conducted all 154 experiments with the enzyme GlpQ in this study at 30°C and pH 8.0.

155
Although the detailed mechanism of phosphodiester-cleavage by GlpQ is currently unknown, Ca 2+ 156 ions were recognized as crucial for catalytic activity (yet they can be substituted with Cd 2+ and partially 157 with Mn 2+ and Cu 2+ ) (45,48). Accordingly, the catalytic reaction was inhibited with 158 ethylenediaminetetraacetic acid (EDTA). Nevertheless the addition of Ca 2+ ions was not required when 159 using the recombinant GlpQ that was purified from the cytosolic extracts of E. coli. The recently solved

167
The substrate binding cleft can be divided into a hydrophilic side including the active site Ca 2+ ion and a 168 hydrophobic side consisting of hydrophobic amino acids including phenylalanine and tyrosine (Phe190,

169
Tyr259, Phe279) (Fig. 2). The active site Ca 2+ ion adopts a pentagonal bipyramidal coordination. It is held 6 in place by glutamic and aspartic acid residues (Glu70, Glu152, Asp72) and is also coordinated by the two 171 hydroxyl groups of Gro3P (Fig. 2). The phosphate as well as the C2 and C3 hydroxyl-group of Gro-3P are 172 drawn towards the active site Ca 2+ ion, and are moved away from the hydrophobic side of the binding cleft.

173
Coordination of the Ca 2+ ion by amino acids with charged side chains and the hydroxyl and phosphate 174 groups of the substrate as well as the orientation of the hydrophobic C-H groups of the substrate towards 175 the hydrophobic side of the binding cleft restricts the productive binding to the unsubstituted sn-glycero-176 3-phosphoryl stereoisomer, thus allowing productive binding only of sn-glycerol-3-phosphoryl groups.

177
Instead, the C2 hydroxyl group of sn-glycerol-1-phosphoryl would face towards the hydrophobic side,  The glycerophosphodiesterase GlpQ of B. subtilis has been identified recently as a teichoicase that 187 preferentially digests polyGroP-type WTA lacking modifications on the glycerol subunits (41). However, 188 in this study, product formation with GlpQ had not been followed using polymeric teichoic acids as 189 substrates and thus, neither the strict specificity for unmodified WTA nor the exo-lytic mechanism have 190 been unequivocally shown. We thus aimed at directly monitoring product release by GlpQ from cell wall 191 (PGN-WTA complex) preparations using high performance liquid chromatography-mass spectrometry 192 (HPLC-MS). We first applied cell wall preparations containing glycosylated WTA, which were extracted 193 from B. subtilis 168 wild-type cells, and cell wall preparations containing non-glycosylated WTA, which 194 were extracted from ΔtagE::erm cells lacking the WTA alpha-glucosyl transferase TagE (cf. Fig. 1). These 195 samples were digested with GlpQ and product formation was followed by HPLC-MS; Figure

217
D-Ala substitutions were removed in the teichoic acid samples by pretreatment as well as applying 218 the GlpQ reaction at pH 8. It has been reported earlier that alanyl esters are rather labile at pH >7, with a 219 half time of hydrolysis at pH 8 and 37°C of 3.9 h (32,50). Accordingly, no difference in the release of GroP 220 was observed, when non-treated and pH 8-pretreated the WTA samples were compared (data not shown).

221
Furthermore, in a time course experiment we observed that already after a few seconds the majority of the

241
The linker disaccharide was obtained from both wild-type and unglycosylated PGN-WTA complexes by 242 chemical digestion in equal amounts as shown in Figure 4A and B. The amount of linker disaccharide 243 released by chemical digestion was set as to 100% of linker disaccharide in the substrate. As another 244 control, the PGN-WTA complex was treated with TCA alone to determine the amounts of linker 245 disaccharide that TCA can release without requiring NaOH pretreatment. Very small amounts of linker 246 disaccharide (ca. 3.6% of the total) were released from both PGN-WTA variants ( Fig. 4C and D). The 247 difference, however, becomes significant once the substrate was pre-digested with GlpQ. While from the 248 wild-type PGN-WTA no more linker was released with GlpQ than with TCA treatment alone, GlpQ was  ΔcsbB::kan cells using established protocols (51,52). Since LTA has been reported to be extensively 256 modified by D-alanyl esters, we aimed to remove also these modifications prior to GlpQ treatment. In a 257 previous study, it has been shown that incubation of LTA at pH 8.5 for 24 h at room temperature leads to 258 an almost complete removal of D-alanyl esters (22). However, according to these data, also partial 259 degradation occurs during this treatment, albeit the degradation apparently is very limited (the degree of 260 polymerisation dropped from 48 to 43). Since we wanted to avoid absolutely any degradation of the LTA 261 samples, we decided to apply slightly milder conditions, and thus preincubated the LTA preparations in

285
Subsequent digestion of these fragments with GlpQ released significant amounts of GroP. As seen for precursors could be co-purified. Since these WTA precursors provide free sn-glycero-3-phosphoryl ends 295 the low amount of GroP product released from LTA might come from its degradation.

296
In summary, GlpQ releases GroP in significant amounts from WTA (see Fig. 3), but only very 297 small amounts from LTA preparations (see Fig. 5C and 6C). These findings confirm differences in the

342
Oligonucleotide primers are listed in supplemental Table S1. The PCR products were purified (Gene JET 343 purification kit and Gene Ruler, 1-kb marker, Thermo Fisher Scientific) and then digested with appropriate

365
were harvested and resuspended in 30 ml piperazine-acetate buffer (50 mM, pH 6) with 12 U proteinase K 366 and boiled for 1 h. The cytosolic fractions were removed by centrifugation (3000 x g, 15 min, 4°C). The 367 pellet was resuspended in 6 ml buffer (10 mM Tris, 10 mM NaCl, 320 mM imidazole, adjusted to pH 7.0 368 with HCl) and 600 µg α-amylase, 250 U RNase A, 120 U DNase I and 50 mM MgSO4 were added. The 369 sample was incubated at 37°C for 2 h while shaking, 12 U Proteinase K was added the incubation continued 12 for 1 h. 4% SDS solution was added 1:1 and the mixture was boiled for 1 h. The SDS was removed by 371 repeated ultracentrifugation steps (20 times at 140 000 x g, 30 min, 40°C) and suspension in H2Obidestilled as 372 well as dialysis against H2Obidestilled. The SDS content was controlled with the methylene blue assay 373 described earlier (54). The cell wall preparation was dried in a vacuum concentrator.

645
The other side of the binding cleft (right of Gro3P) consists of hydrophobic amino acids like phenylalanine, 646 tyrosine and leucine (Phe190, Tyr259, Phe279). The Ca 2+ ion adopts a pentagonal bipyramidal coordination 647 and is coordinated by Glu70, Glu152, and Asp72) as well as by the two hydroxyl groups of Gro3P. The

726
GlpQ is able to cleave off Gro-3P from the terminal ends of WTA. Conversely, GlpQ is not able to chip 727 off Gro-1P from the terminal ends of LTA. The orientation of the hydroxyl group on the C2 is distinguished 728 by the stereospecific enzyme GlpQ. Treatment with NaOH allows to pre-cleave phosphodiester bonds 729 within the LTA chain polymer, resulting in fragments that contain Gro-3P terminal ends. From these ends 730 GlpQ is able to cleave off Gro-3P moieties. The differential cleavage of WTA and LTA by GlpQ reveals 731 different stereochemistry of the polymers.