Alanylated lipoteichoic acid primer in Bacillus subtilis

Lipoteichoic acid is a major lipid-anchored polymer in Gram-positive bacteria such as Bacillus subtilis. This polymer typically consists of repeating phosphate-containing units and therefore has a predominant negative charge. The repeating units are attached to a glycolipid anchor which has a diacylglycerol (DAG) moiety attached to a dihexopyranose head group. D-alanylation is known as the major modification of type I and type IV lipoteichoic acids, which partially neutralizes the polymer and plays important roles in bacterial survival and resistance to the host immune system. The biosynthesis pathways of the glycolipid anchor and lipoteichoic acid have been fully characterized. However, the exact mechanism of D-alanyl transfer from the cytosol to cell surface lipoteichoic acid remains unclear. Here I report the use of mass spectrometry in the identification of possible intermediate species in the biosynthesis and D-alanylation of lipoteichoic acid: the glycolipid anchor, nascent lipoteichoic acid primer with one phosphoglycerol unit, as well as mono- and di-alanylated forms of the lipoteichoic acid primer. Monitoring these species as well as the recently reported D-alanyl-phosphatidyl glycerol should aid in shedding light on the mechanism of the D-alanylation pathway of lipoteichoic acid.

This feature makes bacterial membrane the easy target of host immune molecules such as cationic antibiotic peptides [2][3][4] . Bacteria have been known to constantly modulate membrane components 1,5 . There are at least three pathways which may contribute to surface charge modulation: biosynthesis of phosphatidylethanolamine (PE), L-lysyl-phosphatidylglycerol (lysyl-PG), and D-alanylation of lipo-and wall-teichoic acids. In comparison with Gram-negative bacteria, Gram-positive bacteria typically have noticeably less PE 1 , but have an abundance of lysyl-PG or other aminoacylated PGs which most Gram-negative bacteria lack 5,6 . Besides, lipo-and wall-teichoic acids are only found in Gram-positive bacteria.
Gram-positive bacteria lack the outer membrane as well as phosphate-rich lipid A and lipopolysaccharide found in Gram-negative bacteria. Instead they have a diverse category of polymeric teichoic acids made of phosphate-containing repeating units. Peptidoglycanattached wall-teichoic acids and glycolipid-anchored lipoteichoic acids were discovered six decades ago 7 . There are five types of lipoteichoic acids 10 and four types of wall teichoic acids 11 . The biosynthesis pathways of the two types of teichoic acids have been characterized [8][9][10][11] . Glycerol or ribitol residues in the repeating units of type I and type IV lipoteichoic acids 10 as well as type I wall teichoic acids 11 are known to undergo D-alanine esterification 7,9,12 , which is known to be carried out by the four dlt operon-coded proteins DltABCD 13 . This surface charge modulation has been observed to significantly affect the antigenicity of the bacteria 2 . In the cytosol, DltA (~500 amino acid residues) catalyzes with the consumption of ATP first the adenylylation of D-alanine and then the thioester formation with D-alanyl-carrier protein DltC (~80 amino acids) [13][14][15] . Crystal structures of DltA 16,17 have proven that DltA is homologous to adenylation domains (also called AMP-forming domains) found in modular nonribosomal peptide synthetases 18 as well as fatty acylcoenzyme A synthetases 19 and firefly luciferases 20 . The functionally uncharacterized DltB (~400 amino acid residues) is predicted to be an integral membrane protein with multiple putative transmembrane helices with a low level of similarity to a putative group of membrane-bound O-acyltransferases 21 . DltD (~400 amino acid residues), with a single putative N-terminal transmembrane helix and a large globular domain, has been reported to bind DltC and possibly catalyzes the final D-alanyl transfer from DltC to teichoic acid 22 .
We have recently characterized the presence of D-but not L-alanine in lipid lysate from Bacillus subtilis, implying the presence of D-alanyl-PG in the bacterial membrane 23 . Observation of other D-alanylated species in the bacterial membrane would help sketching a transfer route for the D-alanyl group from inside the cytosol to teichoic acids on the outer surface. Here I report profiling of B. subtilis lipids and identification of mono-and di-alanylated derivatives of nascent lipoteichoic acid primer with a single phosphoglycerol unit attached to the glycolipid anchor (chemical structures shown in Figure 1).

Materials and methods
Bacterial strain and cell culture. The BL21 (DE3) strain of E. coli (Novagen) and B. subtilis strain 168 (Bacillus Genetic Stock Center) were first plated from freezer stock onto LB-agar media. A single colony was transferred into 100 ml of LB media. After incubation overnight at 37°C and 220 rpm in an environmental shaker, it was transferred to 1 liter of LB media. When the cell density reached ~1.0 at 600 nm, 200 ml cell culture supplemented with 2.0 ml of 1.0 M NaAc buffer at pH 4.6 was centrifuged at 5,500 rpm in a Beckman JLA-8.1 rotor for 16 minutes at 4°C. The wet cell pellet was used for lipid extraction.
Lipid extraction. HPLC-grade organic solvents (Fisher Scientific) and distilled and deionized water were used throughout the experiment. The lipid extraction procedure was following that of Bligh and Dyer 24 . Briefly, the wet cell pellet was re-suspended in a glass tube in 0.5 ml ice-chilled water and 2.0 ml of ice-chilled methanol. Then 1.0 ml of cold chloroform was added. The suspension was vortexed for 3 seconds every 5 minutes and incubated on ice for a duration of 10 minutes. After that, 2.0 ml cold chloroform was added followed by 1.5 ml of cold water. The tube was vortexed for 3 seconds and placed on a rocking platform at a room temperature of 21°C for 3 minutes. Phase separation was assisted by centrifugation at 1,300 rpm for 5 minutes with a Beckman Allegro X-22R centrifuge. The heavier chloroform-rich phase was transferred by a glass syringe to a second glass centrifuge tube. Another 2.0 ml cold chloroform was added to the first tube and vortexed for 3 seconds. Then the first tube was put back on the rocking platform at room temperature for 10 minutes. Centrifugation at 1,300 rpm for 5 minutes and transfer of the heavier chloroform-rich phase to the second glass tube followed. The combined chloroform-rich phase was

Amendments from Version 1
All abbreviation "Pho" for phosphate are replaced with "P".
All abbreviation "LTA" for lipoteichoic acid primer are replaced with "LTAP".
The chemical structures in Figure 1 and Figure 5 are redrawn to show double bond in the same thickness and glucosyl residues in their chair conformation.
Type I teichoic acids in B. subtilis as well as other four types of teichoic acids are mentioned. Type I and type IV teichoic acids undergo D-alanylation.
In the abstract, it becomes "in the identification of possible intermediate species".
In the abstract, "would aid in" becomes "should aid in".
In the introduction, lipid A and lipopolysaccharide as well as the outer membrane of Gram-negative bacteria has been mentioned.
In the discussion, abundance of D-alanine over L-alanine in the lipid lysate is discussed.
In the introduction, "DltB is predicted to be an integral membrane protein".
In figure legends, "aicd" has been corrected as "acid".
"FA -fatty acid" has been added to the notes of Table 2, Table 3 and Table 4. mixed with 0.5 ml 0.5 M NaCl, vortexed for 3 seconds and gentle shaking by hand for 1 minute. After centrifugation at 1,300 rpm for 5 minutes, the chloroform-rich phase, 4.0-4.5 ml in volume, was collected in a third glass tube for storage at -80°C. Typically, the total lipid concentration was estimated as 0.5 mg/ml.
Lipid profiling by mass spectroscopy. The lipid samples were diluted by adding 2-fold volume of methanol to a concentration of ~0.15 mg/ml (or 150 ppm) for direct infusion at a rate of 0.6 ml/hour to a SCIEX 4000 QTRAP mass spectrometer. Electrospray ionization was achieved at a temperature of 500°C and a pressure of 20 psi for curtain gas as well as ion source gas 1 and 2. The collision energy in the ion trap was tested between 30 and 100 electronvolts for most efficient detection of target substructures in the lipids. The SCIEX Analyst 1.6 software was used to acquire and export averaged mass spectra with the 4000 QTRAP system. Agilent MassHunter B.06.00 was used to process mass spectra with an Agilent Q-TOF 6500 system. MS spectra in the figures were also analyzed with Mass++ 2.7.4 software 25 and presented with Microsoft Excel.
Tandem mass spectroscopy. The targeted MS/MS spectra were first acquired using the SCIEX 4000 QTRAP system with multiple collision energy settings between 50 and 90 electronvolts. Highaccuracy MS/MS spectra were acquired using the Agilent Q-TOF 6550 system with collision energy ranging from 30 to 80 electronvolts. Direct infusion was also employed but at a faster flow rate of 2.0 ml/hour for the Q-TOF 6550 system. Polar lipid extraction on ice produced more species in the sample -Ice-chilled solvents instead of room-temperature ones were used during the well-established polar lipid extraction procedure devised by Bligh and Dyer 24 . The new lipid preparations did not show marked differences on thin-layer chromatograms. However, their mass spectra showed noticeable difference with the cold extraction producing more species than room-temperature extraction. The alanylated derivatives of lipoteichoic acid primer were not observed in lipids extracted at room temperature.

Results
Profiling and tandem mass spectroscopy of polar lipids with dihexose head group -The sodiated form of the lipid anchor of lipoteichoic acid in B. subtilis has been identified by mass spectrometry previously 26 . Several mass spectrometric scans with the 4000 QTRAP system in search for the lipid anchor were experimented. The anchor has a common structure of DAG-dihexose, with the hexose being either glucose or galactose depending on the identity of the microbial organism 27 . The unbranched and typically glycerolphosphate polymer is attached to C-6 of the non-reducing hexopyranosyl end of the glycolipid anchor by a phosphodiester bond 27 . In B. subtilis, the head group is diglucose 27 ( Figure 1A). The sodiated dehydrated diglucose (342 -18 + 23 = 347 amu) at a collision energy of +80 electronvolts revealed the two most intense peaks (887 and 915 amu) matching expected sizes of the lipid anchor with the two dominant fatty acyl compositions of (30:0) and (32:0), respectively 26 ( Figure 2A). Tandem mass spectra of the most abundant 915 amu species was then acquired with the SCIEX 4000 QTRAP system ( Figure 2B) and the Agilent Q-TOF 6550 system ( Table 1). The QTRAP system produced less noisy spectra which are shown in Figure 2- Figure 4. The m/z values obtained from the Q-TOF system were more accurate and are listed in Table 1-Table 4 Table 1). The twin peaks at 365 and 347 amu corresponded to the sodiated diglucose head group and its dehydrated form, respectively. It is worth noting that the signature [DAG -OH] + ion (551 amu) for glycerolphospholipids was missing. However, the two [MAG -OH] + ions at 299 and 327 amu were observed at lower intensity. Even though the 405 amu ion was more intense than the 347 ion, lipid profiling by searching for precursors of the 405 amu cation was inferior to the precursor scan for the 347 amu cation.
Profiling and tandem mass spectrometry of lipids with phosphoglycerol terminus -The phosphoglycerol head group has a molecular mass of 172 and produced a cyclic, equivalent to dehydrated, residual anion at 153 amu. The 153 amu fragment peak is most intense for phospholipids with a terminal phosphoglycerol, and weak for phospholipids -such as cardiolipin (CL) and aminoacylated PGswith such an embedded group. This scan between 400 and 1700 amu at a collision energy of -95 electronvolts was most effective in hitting larger precursor ions (part of the mass range is shown in Figure 3A). The spectrum revealed a cluster of cardiolipin (CL) double anions in the 650-680 amu range and a more intense cluster centered around two major anions at 693 and 721 amu, corresponding to the dominant lipids of (30:0) and (32:0) PGs, respectively. There was an 887 amu unknown species as well as mostly dehydrated lysocardiolipins (lyso-CL) close to 1100 amu and cardiolipins close to 1300 amu. There were no noticeable hits below 600 amu or between 1400 and 1700 amu. Besides, the 1017 amu and 1045 amu anions matched expected masses of lipoteichoic acid primer ( Figure 1B) with dominant fatty acyl compositions of (30:0) and (32:0), respectively.
The 1017 amu anion had two identical (15:0) fatty acyl chains and therefore made assignment of fragments less difficult. The MS/ MS spectra of the 1017 ion acquired with the QTRAP system at a collision energy of -90 electronvolts is shown in Figure 3B, and m/z values of fragments are listed in Table 2. In addition to the 79 amu phosphate residue, the pair of 153 amu and 171 amu ions which corresponded to glycerolphosphate residue, the dominant fatty acid ion at 241 amu matched the expected (15:0) composition. Fragmentation at the two glycosyl bonds likely produced the 315 amu and 477 amu ions. At the other end of the spectrum, the 943 amu ion was likely due to the neutral loss of cycloglycerol (74 amu). A further loss of (15:0) fatty acid (242 amu) or ketene (224 amu) likely produced the pair of 701 and 719 amu ions, respectively. Another pair at 775 and 793 amu were produced similarly but from the molecular ion. The 1017 amu molecular ion matched structural characteristics of a lipoteichoic acid primer with a single glycerolphosphate unit attached to the lipid anchor of diglucosyldiacyglycerol.
Profiling of lipids with ester-linked alanine -In negative mode, ester-linked fatty acids are known to form intense fragment [FA-H]ions. This is also true for ester-linked amino acids 23 . A precursor scan between 400 and 1700 amu at an optimized collision energy of -95 electronvolts for 88 amu [Alanine-H] -(part of the mass range is shown in Figure 4A) revealed as expected a cluster of alanyl-PGs with two dominant peaks at 764 and 792 amu corresponding to (30:0) and (32:0) alanyl-PG, respectively. The precursor scan also revealed two adjacent clusters of alanylated lipids separated by 71 amu which corresponded to the molecular mass of a dehydrated alanine. The first cluster with dominant 1088 and 1116 amu anions matched expected m/z values of mono-alanylated lipoteichoic acid  Figure 1D. P -phosphate; Gro -glycerol; MAG -monoacylglycerol; DAG -diacylglycerol; Glc -glucose. There are equivalent choices such as between a 1 and a 2 , as well as between b 1 and b 2 .   Figure 1D. FA -fatty acid; P -phosphate; Gro -glycerol; MAG -monoacylglycerol; DAG -diacylglycerol; Glcglucose; LTAP -lipoteichoic acid primer DAG-Glc 2 -P-Gro. Cleavage at a 1 and a 2 , as well as at b 1 and b 2 produces fragments of identical sizes. primers ( Figure 1C), while the second cluster centered around the 1159 and 1187 amu anions matched those of di-alanylated lipoteichoic acid primers ( Figure 1D). There were no noticeable hits below 700 amu or between 1200 and 1700 amu.
The overall mass of the 1088 amu anion matched that of (30:0) alanyl-lipoteichoic acid primer. The MS/MS spectra of the 1088 ion acquired with the QTRAP system at a collision energy of -90 electronvolts is shown in Figure 4B, and m/z values of fragments are listed in Table 3 The m/z value of the 1159 amu anion matched that of (30:0) di-alanyl-lipoteichoic acid primer. The MS/MS spectra of the 1159 ion acquired with the QTRAP system at a collision energy of -80 electronvolts is shown in Figure 4C, and m/z values of fragments are listed in Table 4. Due to the lower abundance of the 1159 anion, its MS/MS spectrum was noisier than those of the 1017 and 1088 anions. The spectrum shared 79, 88, 153, 171 and 241 amu ions with that of the 1088 amu anion. Unexpectedly, a 159 amu species corresponding precisely to deprotonated alanylalanine dipeptide was also observed. The mid-section of the spectrum did not reveal reoccurring ions in spectra collected at collision energies 10 electronvolts apart and therefore were likely noise due to the low abundance of this molecular ion. The whole head group of dialanylated glycerolphosphate was not observed. At the high end of the spectrum, an 846 amu fragment was probably generated by neutral losses of both (15:0) fatty acid (242 amu) and a cyclo-alanine (71 amu). The 925, 943 and 999 amu ions were in common with the fragments from the 1088 amu anion of mono-alanylated lipoteichoic acid primer. A further dehydrated 981 amu ion was observed, which corresponded to neutral losses of two alanine molecules (2 × 89 = 178 amu). A larger 1017 amu ion precisely matched that of (30:0) lipoteichoic acid primer. Another even larger 1070 amu ion corresponded to the 89 amu neutral loss of one alanine    from the 1159 amu molecular anion. Surprisingly, a 1115 amu ion corresponding to a neutral loss of 44 amu was observed. As shown in Figure 5A, a two-step reaction may account for the formation of alanylalanyl-LTA primer and subsequent fragmentation into the 159 amu alanylalanine anion. An alternate reaction shown in Figure 5B may rearrange the putative bis-alanyl-LTA primer to expose a terminal carboxyl group in one of the two alanine residues, which could subsequently release the 44 amu CO 2 and produce the 1115 amu fragment anion. Due to the lack of any fragment ion corresponding to linear (313 amu) or cyclic (295 amu) bis-alanylglycerolphosphate, the result was not definitive on the location of two ester-linked alanine residues. Based on its similar fragmentation pattern to that of mono-alanylated LTA primer, this 1159 amu species is tentatively assigned as bis-alanyl-LTA primer ( Figure 1D).

Discussion
Aminoacylated lipids play an apparent role in surface charge modulation of Gram-positive bacteria 5 . The least known part of charge modulation is arguably the D-alanylation pathway of lipoteichoic acids. The Bligh and Dyer method 24 carried out at an icy temperature appeared to be essential for successful extraction of species that are almost certainly lipoteichoic acid primer and its monoand di-alanylated derivatives. My lab has recently observed that lysate of B. subtilis lipids contained predominantly D-alanine. Lysate of lipids in this study showed the expected predominance of D-alanine over L-alanine. The lipoteichoic acid primers were possibly esterized with D-alanine. The possible existence of the putative bis-alanylated lipoteichoic acid primer indicates that these species are unlikely to be hydrolyzed fragments of lipoteichoic acids since hydrolysis could only produce mono-alanylated derivative. Hydrolysis of lipoteichoic acid should also produce detectable amount of residue with more than one phosphoglycerol units attached to the lipid anchor, which was apparently lacking in the lipid extract. It also implies that lipoteichoic acid is unlikely to be transferred as D-alanyl-glycerolphosphate unit directly from D-alanyl-PG to the growing lipoteichoic acid chain by the LtaS polymerase as that would only produce mono-alanylated derivative. The observable abundance of lipoteichoic acid primer also appear to suggest that one of the four LtaS paralogs in B. subtilis 28 may indeed act like LtaP primase in Listeria monocytogenes for the biosynthesis of lipoteichoic acid primer 29 .
My lab has recently hypothesized that D-alanyl-PG may serve as the lipid intermediate for subsequent D-alanylation of teichoic acids. Taken together, a putative pathway is shown in Figure 6. Author contributions YL conceived and carried out this study.

Competing interests
No competing interests were disclosed. The present manuscript describes a mass spectrometric study of Bacillus subtilis lipids in which possible 12.

16.
The present manuscript describes a mass spectrometric study of Bacillus subtilis lipids in which possible intermediates in the biosynthesis and D-alanylation of type I lipoteichoic acid are detected. In general, the title is appropriate, and the abstract provides an adequate summary of the article. The conclusions are based on the raw data presented. My major critique is that the author actually does not show the presence of the pathway, he only detects molecular ions my MS that might be pathway intermediates. The other comments should be addressed to improve the manuscript. Comments: Abstract: I think the word "possible" should be added, "Here I report the use of mass spectrometry in the identification of possible intermediate species…", because although that it is very probable that this is case, strictly speaking, it has not been shown.
Abstract: In the same sense, I think it is preferable to write "…D-alanyl-phosphatidyl glycerol should aid in…" Introduction: It should be specified that the author refers to type I LTA.
Introduction: The first part of the introduction is very simplified, for example no lipid A is mentioned, nor the modification it can suffer to make it less anionic. I suggest rewriting and improving this paragraph. Strictly speaking the authors don't show that it is D-alanine in the detected structures.
Results/Discussion: What does the author think is the explanation that the new species only can be detected after a cold extraction?
Methods: Why do the authors add sodium acetate to the growth medium before centrifugation of the culture?
Page 2, left column, middle paragraph: I think it should say "adenylylation" not "adenylation", because it is an AMP-transfer.
Page 2, left column, middle paragraph: Instead of "The functionally uncharacterized DltB appears to be an integral membrane protein…", it would be better to write "is predicted to be…".
Page 2, left column, middle paragraph: It should say "O-acyltransferase", not "o-acyltransferase" In the legends to figures 3 and 4, it should say "acid" not "aicd".
Methods/Results: Why are the electronvolts sometimes "+" and sometimes "-" Tables 2 and 4: The abbreviation FA for fatty acid should be explained.
Page 8, left column, middle paragraph: The word "smaller" should be deleted.
I agree with reviewer one that P is probably a better abbreviation of the phosphate group.

I have read this submission. I believe that I have an appropriate level of expertise to confirm that
F1000Research YL: I will revise the statements concerning D-alanylation. I will briefly report that the lipid lysate has much more D-alanine than L-alanine.
7. Results/Discussion: What does the author think is the explanation that the new species only can be detected after a cold extraction? YL: It is possible that it is not stable at room temperature. I think the readers can explain this.
8. Methods: Why do the authors add sodium acetate to the growth medium before centrifugation of the culture? YL: It is out of respect to previous publications that teichoic acid-linked D-alanine tend to be most stable under mildly acidic condition. It is also part of my effort to standardize protocols. As the grow dense, pH typically gradually raises from neutrality to B. subtilis 8.5. Unlike lipid extraction under low temperature, adding sodium acetate before centrifugation was not a required procedure as unbuffered cells also produced similar amount of the reported lipid species.
9. Page 2, left column, middle paragraph: I think it should say "adenylylation" not "adenylation", because it is an AMP-transfer. YL: Strictly speaking "adenylylation" is correct. It will be corrected.
10. Page 2, left column, middle paragraph: Instead of "The functionally uncharacterized DltB appears to be an integral membrane protein…", it would be better to write "is predicted to be…". YL: I agree.
12. In the legends to figures 3 and 4, it should say "acid" not "aicd". YL: I was typing too fast.
13. Methods/Results: Why are the electronvolts sometimes "+" and sometimes "-". YL: The sign of voltage values appears to follow the positive/negative mode of ionization. I had to specify the sign during the experiment.
14. Tables 2 and 4: The abbreviation FA for fatty acid should be explained. YL: I will add this abbreviation.
15. Page 8, left column, middle paragraph: The word "smaller" should be deleted. YL: I was emphasizing the relative ease of analyzing the smaller 1017 amu species than the larger 1045 amu species. As the sentence starts a paragraph, I agree it is better to remove "smaller".
16. I agree with reviewer one that P is probably a better abbreviation of the phosphate group. YL: I was trying to use 3-letter codes consistently. I will simplify it as "P". The work of Luo Y. describes an identification of the biosynthetic intermediates of the lipoteichoic acid utilizing mass spectrometrical approach.

None
Generally, the conclusions are based on the solid raw data, the title is appropriate, and the abstract provides subsequent summary of the article.
I have a few comments, which addressing can help to improve the manuscript: In the abstract and introduction given description of LTA refers mainly to Type 1 of LTA. I would specify it, and in the introduction mention other types of LTA, that are structurally different The use of the sentence with the characterization of D-but not L-Alanine, is confusing, as to date all bacterial Ala is D-configured. The given formula of glycerol is wrong -no double bond is present, better to use standard abreviation Gro, e.g. Gro-Glc-Glc (concerns also Table 1).
Phosphate groups are commonly abbreviated as P, not Pho. Table 1 -I would write LTA primer, not only LTA, otherwise confusing.
Peak at 887 corresponds to 2x Hex, Gro, 2x15:0, Na, so corresponds to sodiated (but not dehydrated -this word should be removed) linker (Figure 2A). The confusion with dehydrated or not was present a few times in the text (e.g. 753 amu represents loss of glucose and not dehydrated glucose).
On page 4 -the QTRAP spectra are shown in Fig 3 -Fig 4, not as in text Fig 2 -Fig 4   I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.
No competing interests were disclosed.

Competing Interests:
Author Response ( ) 03 Mar 2016 Member of the F1000 Faculty