Knockout of ykcB, a putative glycosyltransferase, leads to vancomycin resistance in Bacillus subtilis

Vancomycin resistance of gram-positive bacteria poses a serious health concern around the world. In this study, we searched for vancomycin-resistant mutants from a gene deletion library of a model gram-positive bacterium, Bacillus subtilis, to elucidate the mechanism of vancomycin resistance. We found that knockout of ykcB, a glycosyltransferase that is expected to utilize C55-P-glucose to glycosylate cell surface components, caused vancomycin resistance in B. subtilis. Knockout of ykcB altered the susceptibility to multiple antibiotics, including sensitization to β-lactams, and increased the pathogenicity to silkworms. Furthermore, the ykcB-knockout mutant had: i) an increased content of diglucosyl diacylglycerol, a glycolipid that shares a precursor with C55-P-glucose, ii) a decreased amount of lipoteichoic acid, and iii) decreased biofilm formation ability. These phenotypes and vancomycin resistance were abolished by knockout of ykcC, a ykcB-operon partner involved in C55-P-glucose synthesis. Overexpression of ykcC enhanced vancomycin resistance in both wild-type B. subtilis and the ykcB-knockout mutant. These findings suggest that ykcB deficiency induces structural changes of cell surface molecules depending on the ykcC function, leading to resistance to vancomycin, decreased biofilm formation ability, and increased pathogenicity to silkworms. IMPORTANCE Although vancomycin is effective against gram-positive bacteria, vancomycin-resistant bacteria is a major public health concern. While the vancomycin resistance mechanisms of clinically important bacteria such as Staphylococcus aureus, Enterococcus faecium, and Streptococcus pneumoniae are well-studied, they remain unclear in other gram-positive bacteria. In the present study, we searched for vancomycin-resistant mutants from a gene deletion library of a model gram-positive bacterium, Bacillus subtilis, and found that knockout of a putative glycosyltransferase, ykcB, caused vancomycin resistance in B. subtilis. Notably, unlike the previously reported vancomycin-resistant bacterial strains, ykcB-deficient B. subtilis exhibited increased virulence while maintaining its growth rate. Our results broaden the fundamental understanding of vancomycin-resistance mechanisms in gram-positive bacteria.

Although vancomycin is effective against gram-positive bacteria, vancomycin-resistant 36 bacteria is a major public health concern. While the vancomycin resistance mechanisms 37 of clinically important bacteria such as Staphylococcus aureus, Enterococcus faecium,38 and Streptococcus pneumoniae are well-studied, they remain unclear in other gram-39 positive bacteria. In the present study, we searched for vancomycin-resistant mutants 40 from a gene deletion library of a model gram-positive bacterium, Bacillus subtilis, and 41 found that knockout of a putative glycosyltransferase, ykcB, caused vancomycin 42 resistance in B. subtilis. Notably, unlike the previously reported vancomycin-resistant 43 bacterial strains, ykcB-deficient B. subtilis exhibited increased virulence while 44 maintaining its growth rate. Our results broaden the fundamental understanding of 45 vancomycin-resistance mechanisms in gram-positive bacteria. 46

INTRODUCTION
suggest that loss of ykcB function leads to vancomycin resistance in B. subtilis. 96 We examined the sensitivity of ΔykcB to antibiotics other than vancomycin. ΔykcB 97 became sensitive to the cell wall synthesis inhibitors ampicillin, oxacillin, and 98 ceftazidime, and the DNA synthesis inhibitor levofloxacin (Fig. 1C). On the other hand, 99 ΔykcB became resistant to the protein synthesis inhibitor chloramphenicol and showed 100 no change in sensitivity to tetracycline (Fig. 1C). These results indicate that ykcB 101 deficiency alters the susceptibility to various antibiotics. 102 As vancomycin-resistant S. aureus strains are known to have attenuated pathogenicity 103 (12-14), we investigated the pathogenicity of ΔykcB using the silkworm infection model. 104 Contrary to our expectation, silkworms injected with ΔykcB died earlier than the parent 105 strain, indicating increased virulence of ΔykcB (Fig. 1D). ΔykcB showed the same growth 106 rate as the parent strain in nutrient medium (Fig. 1E), ruling out the possibility that the 107 change in pathogenicity against silkworms depends on the bacterial growth rate. 108 Introduction of a stop codon mutation in the ykcB gene in the delA background 119 (delA/ykcBstop) caused a vancomycin-resistant phenotype to the same extent as ΔykcB, 120 but introduction of a stop codon mutation into both the ykcB and ykcC genes in the delA 121 background (delA/ykcBstop/ykcCstop) did not cause a vancomycin-resistant phenotype, 122 indicating that ykcC knockout cancelled the effect of ykcB knockout ( Fig. 2A, 2C). These 123 findings suggest that ykcB deficiency confers vancomycin resistance to B. subtilis in the 124 presence of ykcC. 125 vancomycin sensitivity indistinguishable from that of the parent strain (Fig. 3D). 144 Therefore, diglucosyl diacylglycerol does not contribute to vancomycin resistance. 145

Knockout of ykcB decreases the amount of lipoteichoic acid and attenuates biofilm-147
forming ability 148 Based on the observation that DykcB exhibits altered sensitivity to antibiotics, increased 149 killing activity against silkworms, and an increased amount of diglucosyl diacylglycerol, 150 we hypothesized that knockout of ykcB alters the amount of lipoteichoic acid or changes 151 biofilm formation, both of which have important roles in antibiotic resistance and 152 virulence. DykcB and the delA/ykcBstop mutants had decreased amounts of lipoteichoic 153 acid (Fig. 4A, 4B). DykcB and the delA/ykcBstop mutants formed less biofilm than the 154 parent strain and the delA mutant, respectively (Fig. 5A, 5B). These findings suggest that 155 the ykcB knockout decreases the amount of lipoteichoic acid, and decreases biofilm 156 forming ability. The delA/ykcBstop/ykcCstop mutant did not have a decreased amount of 157 lipoteichoic acid or decreased biofilm formation (Fig. 4A, 4B, 5A, 5B), indicating that 158 ykcC knockout cancelled the phenotypic changes caused by ykcB knockout. 159 160

Overexpression of ykcC increases vancomycin resistance 161
Because ykcC knockout cancelled the vancomycin resistance caused by the ykcB 162 knockout, expression of ykcC is hypothesized to have a positive role in vancomycin 163 resistance. To evaluate this hypothesis, we transformed the parent and DykcB strains with 164 a multicopy plasmid encoding FLAG-tagged ykcC under the ykcBC native promoter. 165 Western blot analysis revealed that the expression of FLAG-tagged ykcC was higher in ykcB knockout upregulates the ykcBC promoter. The parent and DykcB strains 168 transformed with FLAG-tagged ykcC exhibited higher vancomycin resistance than those 169 transformed with an empty vector (Fig. 6B). In addition, the DykcB strain transformed 170 with FLAG-tagged ykcC exhibited slightly higher vancomycin resistance than the parent 171 strain transformed with FLAG-tagged ykcC (Fig. 6B). These findings suggest that ykcC 172 confers vancomycin resistance to B. subtilis in an expression-dependent manner. 173

175
The findings of the present study revealed that knockout of ykcB, a putative 176 glycosyltransferase gene, confers B. subtilis resistance against vancomycin in a ykcC-177 dependent manner. Knockout of ykcB also leads to bacterial sensitivity to beta-lactams, 178 decreases the amount of lipoteichoic acids, attenuates biofilm formation, and increases 179 silkworm-killing activity. This study is the first to reveal that knockout of a specific gene 180 leads to vancomycin resistance in B. subtilis. 181 The ykcC knockout mutant did not exhibit the same phenotypes as the ykcB knockout 182 mutant. In addition, in the ykcC-stop codon mutant background, the stop codon mutation 183 of ykcB led to no phenotypic changes. Therefore, the ykcC gene is required for the 184 phenotypic changes triggered by ykcB knockout. Furthermore, overexpression of ykcC 185 increases vancomycin resistance in the B. subtilis parent strain and the ykcB knockout 186 strain. These findings suggest that expression of ykcC as well as knockout of ykcB 187 increases the amount of some biologic molecule that leads to vancomycin resistance (Fig.  188   7). A previous study predicted that YkcC catalyzes UDP-glucose to C55-P-glucose and accumulate in the ykcB-knockout mutant and lead to various phenotypic changes, 192 including vancomycin resistance (Fig. 7). C55-P-glucose is utilized as a sugar donor by 193 YfhO to glycosylate lipoteichoic acid (17,20). C55-P acts as an anchor to synthesize wall 194 teichoic acid and peptidoglycan (21). Therefore, accumulation of C55-P-glucose might 195 alter the structures of peptidoglycan, lipoteichoic acid, and wall teichoic acids, which may 196 underlie the phenotypic changes observed in the ykcB-knockout mutant. 197 In the ykcB-knockout mutant, the amount of diglucosyl diacylglycerol was increased. 198 We speculate that the accumulation of C55-P-glucose in the ykcB-knockout mutant 199 increases UDP-glucose, a precursor of C55-P-glucose, and the increase in UDP-glucose 200 leads to an increase in diglucosyl diacylglycerol (Fig. 7). Because knockout of ugtP, a 201 synthetase gene of diglucosyl diacylglycerol, did not alter vancomycin resistance ( Fig.  202 3D), the increased amount of diglucosyl diacylglycerol in the ykcB-knockout mutant does 203 not contribute to the vancomycin resistance. In addition, in the ykcC-knockout mutant 204 and the ykcBstop/ykcCstop mutant, diglucosyl diacylglycerol was not increased. In the 205 absence of YkcC, UDP-glucose might be utilized for molecules other than diglucosyl 206 diacylglycerol, which would prevent the accumulation of diglucosyl diacylglycerol ( Fig.  207   7). 208 The ykcB-knockout mutant was resistant to vancomycin, but sensitive to beta-lactams. 209 In VISA, a vancomycin resistance phenotype is accompanied by a beta-lactam sensitive 210 phenotype, referred to as a "seesaw phenomenon" (22, 23). Thus, the beta-lactam 211 sensitivity of the ykcB-knockout mutant of B. subtilis is consistent with that of VISA. The 212 amounts of penicillin-binding protein 2 or phosphatidylglycerols are proposed to

B. subtilis mutant. 216
This study demonstrated that the ykcB-knockout mutant has increased silkworm killing 217 activity. In our previous study, Escherichia coli mutant strains resistant to vancomycin 218 also showed resistance to antimicrobial peptides and increased silkworm-killing activity 219 plates containing erythromycin (1 µg/ml) and the colonies were aerobically cultured in 237 pDR110 were cultured in LB broth containing tetracycline (30 µg/ml) or spectinomycin 239 (50 µg/ml). Bacterial strains and plasmids used in this study are listed in Table 2. 240 241

Screening of vancomycin-resistant strains 242
The BKE library (18) was cultured in LB broth using a 96-well microplate at 37˚C and 243 the bacterial culture was spotted onto LB plates with or without vancomycin (0.45 µg/ml) 244 using a replicator. The plates were incubated overnight at 37˚C and mutant strains whose trpC2 genomic DNA as a template and oligonucleotide primers ( Table 3). The amplified 264 DNA fragments were inserted into SphI and SalI sites in pDR110, resulting in pDR110-265 ykcB. Double crossover recombination of pDR110 or pDR110-ykcB at the amyE locus 266 was confirmed by PCR using oligonucleotide primers (

2) Construction of pHY300PLK-ykcC-FLAG 270
Two DNA fragments containing the promoter region of the ykcBC and ykcC ORF were 271 amplified by PCR using oligonucleotide primers (Table 3)

4) Construction of chromosome deletion mutant by natural transformation 292
Targeting cassettes were constructed according to the previously described method (18)  293 with minor modification. A DNA fragment containing the erythromycin resistance 294 marker was amplified by PCR using oligonucleotide primers (Table 3) and a template 295 genomic DNA from the ykcB mutant (BKE12880). The upstream and downstream DNA 296 regions of the targeting chromosome locus were amplified by PCR using oligonucleotide 297 primers ( Table 3, Table 4) and a template genomic DNA from 168 trpC2. The 3 DNA 298 fragments comprising the upstream and downstream regions and the erythromycin-299 resistance gene were mixed in an equal molar ratio and connected by PCR overlap 300 extension using KOD FXneo DNA polymerase (Toyobo, Osaka, Japan). The connected 301 DNA fragment was used for transformation without purification. 302 Competent cells for natural transformation were prepared according to the previous 303 method (33) with minor modification. B. subtilis 168 trpC2 overnight culture (50 µl) was 304 inoculated into 5 ml of SPI medium (0.2% ammonium sulfate, 1.4% dipotassium 305 hydrogen phosphate, 0.6% potassium dihydrogen phosphate, 0.1% trisodium citrate 306 dihydrate, 0.02% magnesium sulfate heptahydrate, 0.5% glucose, 0.02% casamino acids, 307 thawed in a 37˚C water bath and a 7.5-fold amount of SPII medium (0.2% ammonium 311 sulfate, 1.4% dipotassium hydrogen phosphate, 0.6% potassium dihydrogen phosphate, 312 0.1% trisodium citrate dihydrate, 0.02% magnesium sulfate heptahydrate, 0.5% glucose, 313 5 mM magnesium chloride, 0.02% yeast extract, 5 µg/ml L-leucine, 5 µg/ml L-314 methionine) was added, and then the cells were aerobically cultured at 37˚C for 90 min. 315 A 50-µl amount of the cells was mixed with a targeting cassette and incubated at 37˚C 316 for 30 min. After adding 100 µl of LB broth to the cells, they were further incubated at 317 37˚C for 60 min. The cells were spread onto LB plates containing 1 µg/ml erythromycin 318 and incubated overnight at 37˚C. The desired chromosomal deletion was confirmed by 319 PCR. 320 321

5) Construction of mutant strains carrying the stop codon mutation 322
A DNA fragment carrying the mhqA-ykcBC region and the erythromycin resistance gene 323 was amplified by PCR using primer pairs ( Table 3, ykcB-F2-SalI, 5pR-BglII) and 324 template genomic DNA from the delA mutant. The DNA fragment was inserted into SalI 325 and BglII sites of pGEM-3Z. Using the plasmid as a template, a thermal cycling reaction 326 was performed using oligonucleotide primers to introduce the ykcB stop codon (Table 3). 327 The reaction solution was digested with DpnI, and then used to transform the E. coli 328 JM109 strain. A plasmid carrying the ykcB stop codon was purified from the E. coli 329 colonies. Using the plasmid as a template, a thermal cycling reaction was performed using 330 oligonucleotide primers to introduce the ykcC stop codon (Table 3) NaCl in a glass tube and incubated for 2 days at 37˚C. The bacterial culture containing 347 biofilms was poured onto a Kimwipe placed on a Kimtowel (Nippon Paper Cresia, Tokyo, 348 Japan). MilliQ water (2 ml) was added to the KimWipe, which was vortexed to detach 349 the biofilms. The OD600 value of the solution was measured. 350 cultured at 37˚C for 24 h; then, 40 ml of the bacterial culture was centrifuged at 10,400 g 354 for 10 min at 4˚C. The bacterial pellet was suspended with 1 ml of milliQ water and the 355 lipids were extracted using the Bligh and Dyer method (34). The lipid fraction was 356 evaporated by a centrifuge evaporator and the lipids were dissolved with 500 µl of (Merck) and the plate was developed in chloroform:methanol:water (65:25:4 v/v). Sugars 359 were visualized by spraying a coloring agent (10.5 ml 15% 1-naphthol in ethanol, 40.5 360 ml ethanol, 6.5 ml sulfuric acid, and 4 ml water) and heating at 115˚C. 361 362 Western blot analysis 363 FLAG-tagged YkcC was detected according to a previous method (28) with minor 364 modifications. B. subtilis overnight cultures was centrifuged at 21,400 g for 2 min and 365 the bacterial pellet was frozen in liquid nitrogen. The bacterial pellet was thawed in buffer 366 (50 mM Tris-HCl pH 7.8, 2 mM EDTA, 0.5 mM dithiothreitol, 0.4 mg/ml lysozyme) and 367 subjected to freeze-thawing 2 times. TritonX-100 was added to the sample to produce a 368 final concentration of 0.1% and the sample was incubated at 37˚C for 30 min. An equal 369 volume of 2x Laemmli sample buffer with 350 mM dithiothreitol was added to the sample 370 and the sample was heated at 95˚C for 3 min. The sample was centrifuged at 21500 g for 371 15 min, and the supernatant was electrophoresed in a 12% sodium dodecyl sulfate-372 polyacrylamide gel. Anti-DYKDDDDK (anti-FLAG) antibody (Wako, Japan) diluted 373 1:3000 in Canget signal solution 1 (Toyobo, Japan) was used as a first antibody solution. 374 was boiled for 40 min and centrifuged at 10,400g for 10 min. The supernatants were 382 electrophoresed in a 15% polyacrylamide gel and transferred to a nitrocellulose 383 membrane (0.2 µm, Trans-Blot Transfer Medium, BioRad). The membrane was treated 384 with 1:1000 anti-lipoteichoic acid antibody (clone 55, Hycult Biotech, Uden, The 385 Netherlands) and washed 3 times with phosphate buffered saline. The membrane was 386 treated with anti-mouse IgG HRP conjugate (Promega) and washed 3 times with 387 phosphate buffered saline. The membrane was reacted with HRP substrate (Western 388 Lightning, Perkin Elmer) and the signals were detected using ImageQuant LAS 4000 389 (Fujifilm, Tokyo, Japan). The band intensity was measured by Image J software (36). 390