Mass Spectrometric Profiling of Microbial Polysaccharides Using Laser Desorption/Ionization – Time-of-Flight (LDI-TOF) and Liquid Chromatography/Mass Spectrometry (LC/MS): A Novel Method for Structural Fingerprinting and Derivatization

33 Over the last two decades, matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-34 TOF MS) has been introduced into the routine diagnostic practice of microbiological laboratories for the rapid 35 taxonomic identification of bacteria and yeasts. However, a method that effectively identifies microbes directly 36 from clinical samples using MALDI-TOF MS has not yet been found. One of the promising targets is microbial 37 polysaccharides, which are abundant structures in bacterial and fungal cells. Their rapid and inexpensive 38 analysis, however, is complicated. This study focused on detecting microbial polysaccharides, such as 39 lipopolysaccharides, using MALDI-TOF MS and liquid chromatography-tandem mass spectrometry (LC-MS). We 40 developed a method for fingerprinting polysaccharides by acid hydrolysis and enzymatic digestion. The mono- 41 and oligosaccharides are then derivatized with a newly developed probe (vanillyl pararosaniline, HD ligand), 42 enabling efficient ionization without the use of the MALDI matrix. The esterification of hydroxyl groups by 43 formic acid was also optimized for precise analysis of the saccharides. The method was validated using several 44 saccharides as well as Escherichia coli lipopolysaccharides (O26:B6, O55:B5, and O111:B4). Derivatization using 45 HD ligand also allows the detection of structures containing amines and phosphate groups in positive ion mode. 46 We also optimize the method using crude bacteria ( Escherichia coli, Salmonella enterica, Shigella dysenteriae, 47 Shigella boydii , Shigella flexneri, and Legionella pneumophila ). This approach opens the possibility of directly 48 detecting microbial polysaccharides from clinical specimens. LDI-TOF MS (without a matrix) also allows specific 49 detection of molecules of interest with suppression of the background signal. 50 51 52 53 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint 3 54 Scheme 1. Analysis of polysaccharides by mass spectrometry after digestion and derivatization using a self-55 ionizable ligand (HD ligand). 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint 4

71 Many innovative applications of matrix-assisted laser desorption/ionization time-of -flight mass spectrometry 72 (MALDI-TOF MS) and liquid chromatography coupled with mass spectrometry (LC-MS) have emerged in clinical 73 diagnostics to detect biomarkers and biomolecules, advancing personalized medicine [1, 2]. 74 In diagnostic microbiology, MALDI-TOF MS became a cornerstone in the rapid taxonomic identification of 75 bacteria and fungi [2, 3]. Similarly, applications for antibiotic resistance determination have also been 76 developed and validated. Among them, beta-lactamase activity determination detecting the molecular mass 77 changes of indicator beta-lactam antibiotics (e.g., carbapenems) has been developed and validated for use in 78 routine laboratory practice [4]. Another method allows the detection of polymyxin resistance using analysis of 79 lipid A of lipopolysaccharides and their structural changes (e.g., the addition of phosphoethanolamine or 4-80 amino-L-arabinose) [5]. 81 As MALDI-TOF MS provides efficient and rapid species identification, there is a key issue whether the method 82 can be used for epidemiological typing directly from spectra obtained directly from a whole-cell lysate. So far, 83 no general typing algorithm has been proposed, but only specific peaks representing significant epidemiological 84 markers have been identified in some species [3]. Artificial intelligence for spectra analysis has recently bee n 85 described as a promising tool for predicting antibiotic resistance and epidemiological typing [6]. 86 Despite the use of artificial intelligence methods for analyzing large-scale data, MALDI-TOF MS should be 87 considered a biochemical tool allowing precise analysis of molecules based on their molecular weight and 88 fragmentation characteristics. Thus, we believe that the scientific community should not resign itself to the 89 exact identification of detected molecules/peaks. For such an analysis, it is usually insufficient to analyze crude 90 bacterial extract without further processing, i.e., specific extraction and enhancement of MALDI-TOF MS-based 91 ionization. 92 Despite the commonly used taxonomic applications in routine laboratories, methods for identifying microbes 93 directly from clinical specimens have not yet been successfully developed [3]. Such applications, however, are 94 limited by a poor sensitivity of MALDI-TOF MS and by masking the proteins of interest by other abundant 95 molecules with a high ionization ability. These challenges can be addressed by (a) concentrating microbes or 96 microbial proteins in the sample and removing host proteins, (b) detecting other molecules (e.g., lipids or 97 polysaccharides) that can be explicitly extracted from the sample, and (c) using LC-MS. 98 Identification of microbial polysaccharides represents a challenge in clinical microbiology as those structures can 99 be used for (a) rapid and cheap epidemiological typing (e.g., Salmonella enterica, Shigella spp., Escherichia coli), 100 (b) development of polysaccharide vaccines and testing their efficiency, and (c) direct identification of microbes 101 from clinical specimens. The latter option is challenging as microbial saccharides represent relatively stable 102 molecules that can be directly detected in several clinical specimens (e.g., blood, urine), allowing rapid 103 identification of a causative agent of the infection and thus increasing therapeutic efficiency [7, 8, 9]. 104 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint 5 As microbial polysaccharides usually contain >10 basic units, their molecular weight is highly above the efficient 105 ability of MALDI-TOF MS or LC-MS to detect them directly. Therefore, the native molecules must be specifically 106 cleaved into smaller units, providing a specific molecule fingerprint. For that purpose, the glycosidic bond is the 107 common target. Hydrolysis of that bond, however, is challenging due to its stability in some biological 108 structures. Several procedures of polysaccharide fingerprinting, including chemical cleavage by strong acids or 109 Fenton’s reaction and enzymatic disruption, have been developed and optimized [10, 11, 12]. 110 Even if monosaccharides or oligosaccharides are available, their ionization using conventional mass 111 spectrometry techniques is complex. Compared to peptides and proteins, those molecules are generally more 112 complicated to ionize and transfer to the vapor phase [12]. Therefore, the typical approach is to derivatize 113 mono and oligosaccharides before mass spectrometry. For that purpose, several methods have been proposed, 114 including the label-assisted laser desorption/ionization approach [13]. 115 Here, we present a novel method for enzymatic and acidic hydrolysis of bacterial polysaccharides, their 116 derivatization, and analysis, where the sample is ionized through the addition of a specific ligand, enabling the 117 use of LDI-TOF (Laser Desorption/Ionization Time-of -Flight) MS and LC-MS (Trapped Ion Mobility 118 Spectrometry—TIMS). This approach allows the identification of saccharide fingerprints based on their ion 119 mobility and m/z. Additionally, we proposed the esterification of mono- and oligosaccharides by formic acid, 120 which can aid in their identification. 121 122 123

124 Reagents 125 For analysis, all chemicals ( D-glucose, D-glucose-1,2-13C2, glucose-6-phosphate, L-glucose, D-fructose, L-rhamnose, 126 L-fucose, D-xylose, lactose, maltose, sucrose, raffinose, 5-(hydroxymethyl)furfural), starch, agarose, 127 lipopolysaccharides (Escherichia coli O26:B6, E. coli O55:B5, E. coli O111:B4), enzymes (diastase, -amylase, -128 amylase, lysozyme, chitinase), and other reagents were purchased from Merck (Merck Life Science, Prague, 129 Czech Republic). The greater quantity of vanillyl-rosaniline (HD ligand) was synthesized as 2,3-Dichloro- 5,6-130 dicyano-1,4-benzoquinone salt by Ratiochem (Brno, Czech Republic). 131 132 Instrumentation 133 Samples for LC-MS were injected and subsequently desalted online using reverse-phase microtrap column (Neo 134 Trap Cartridge 5mm, ThermoFischer Scientific, Prague, Czech Republic) and separated on the reverse-phase 135 analytical column (Bruker TEN, Bruker Daltonics, Bremen, Germany) at 40 °C using Bruker nanoElute 2 HPLC 136 system at flow rate 0.5 µL/min in water/acetonitrile gradient (mobile phase [A] water with 0.1% formic acid; 137 mobile phase [B] acetonitrile with 0.1% formic acid; the gradient started at 10% [B] and reached 50% [B] in 35 138 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint 6 min). The eluted analytes were directly analyzed by timsTOF Pro 2 mass spectrometer using a CaptiveSpray for 139 ionization (Bruker Daltonics, Bremen, Germany). Measurement was performed in positive ion mode over the 140 m/z range 100 – 1350 in dia-PASEF® mode (0.75 – 1.60 V.s/cm 2, ramp time 100 ms). Data acquisition and data 141 processing were performed using DataAnalysis 6.1. 142 LDI-TOF MS was performed using a MALDI Biotyper® Sirius mass spectrometer and rapifleX® MALDI-TOF/TOF 143 system (Bruker Daltonics, Bremen, Germany). The spectra were analyzed using flexAnalysis 4.0 software. 144 For precise identification of molecular mass, the samples were analyzed using a 15T solariX XR FT-ICR mass 145 spectrometer (Bruker Daltonics). Mass spectral data were collected in positive broadband mode over the m/z 146 range 150 – 1500, with 1M data points transient and 0.2 s ion accumulation with two averaged scans per 147 spectrum. Data acquisition and data processing were performed using ftmsControl 2.1.0 and DataAnalysis 5.0. 148 149 Synthesis of Vanillyl-Rosaniline Reagent (HD ligand) 150 Pararosaniline hydrochloride (basic fuchsin) and vanillin were dissolved in methanol to a 100 mmol/L 151 concentration (both reagents). After dissolving, glacial acetic acid was added to a final concentration of 2.5 152 mol/L and mixed well. Methylpyridine borane dissolved in methanol as a reducing agent was added to the 153 reaction mixture to a final concentration of 10 mmol/L and incubated at 50 °C with shaking for 2 hrs. 154 After incubation, the reaction was diluted by deionized water to a final concentration of methanol of 10 % and 155 directly applied on the NGC Quest Plus System (Bio-Rad, Prague, Czech Republic) equipped with a XSelect® CSH 156 C18 OBDTM preparative column (Waters, Gesellschaft m.b.H., Prague, Czech Republic) at flow rate 0.1 mL/min. 157 After application of the reaction mixture, the column was washed by 50 mL of 5% acetonitrile with 0.1% formic 158 acid. Then water/acetonitrile gradient (mobile phase [A] water with 0.1% formic acid; mobile phase [B] 159 acetonitrile with 0.1% formic acid; the gradient started at 5% [B] and reached 35% [B] in 60 min) was used for 160 sample purification. Two mL fractions were collected and measured using LC-MS after dilution with water 1:10. 161 Fractions showing purified vanillyl-rosaniline (HD ligand) (Figure 1) were dried by a vacuum concentrator. 162 163 Saccharide’s Derivatization and Analysis 164 Derivatization was performed in a 1% pyridine buffer, and the pH was adjusted to 4.0 using glacial acetic acid. 165 Saccharides were diluted to a concentration of 5 mmol/L with a reaction buffer. Concentrated HD (250 mmol/L) 166 solution diluted in acetonitrile was added to a final 5 mmol/L concentration. The reaction mixture was 167 incubated at 50 °C for 20 minutes. 168 For LC-MS, the samples were diluted 1:5 with a pyridine buffer. For LDI-TOF MS, the samples were purified by 169 C18-reversed phase silica gel to avoid the “sweet spot” formation on the MALDI target. Purification was 170 performed in the Eppendorf tube. To 5 mg of C18 reverse phase resin, 10 μL of 30% acetonitrile with 0.1% 171 formic acid was added. After vortexing, 50 μL of the reaction mixture was added, vortexed for 10 s, and 172 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint 7 centrifuged at 14,000 g. The resin was washed with 1 mL of 5% acetonitrile with 0.1% formic acid. The sample 173 was eluted with 25 μL of 50% acetonitrile with 0.1% formic acid. Two μL of the sam ple was directly applied to 174 the MALDI target, allowed to dry, and then measured without adding any matrix. 175 176 Enzymatic Digestion of Bacterial Lipopolysaccharides 177 Two milligrams of bacterial lipopolysaccharide were dissolved in 100 μL of 25 mM EDTA, pH 7.5, with 1 mg/mL 178 diastase (-amylase and -amylase) and 1 mg/mL lysozyme. The reaction was incubated at 50 °C for 30 minutes. 179 After incubation, the reaction was filtered using an Microcon® - 10 Centrifugal Filters, 10 kDa NMWL (14,000 g , 180 30 minutes). To the fraction containing low molecular mass molecules (filtrate collected in the bottom tube), 50 181 μL of reaction buffer containing 5 mmol/L HD ligand was added, and saccharides were derivatized at 50°C for 20 182 minutes as described above. 183 184 Acidic Digestion of Oligo- and Lipopolysaccharides 185 Saccharides and lipopolysaccharides were digested by formic acid and its isotopic variant ( 13C). One milligram of 186 the sample was dissolved in 10 μL concentrated formic acid or formic acid -13C and incubated at 98 °C for 10 187 minutes in the thermocycler with a heated lid (104 °C). After incubation, 50 μL of 5% pyridine in water was 188 added. The samples were filtered using Microcon® - 10 Centrifugal Filters, 10 kDa NMWL (14,000 g, 30 minutes), 189 and the bottom fraction (50 μL) containing digested saccharides was used for derivatization using HD ligand and 190 analysis as described above. 191 192 Analysis of Lipopolysaccharides from Bacterial Cells 193 The bacteria were obtained from the Czech National Collection of Type Cultures (National Institute of Health, 194 Prague, Czech Republic): E. coli ATCC25922, E. coli O55:B5 CNCTC5874, E. coli O111:B4 CNCTC5650, Salmonella 195 enterica subsp. enterica serovar Enteritidis CNCTC5187, S. enterica subsp. enterica serovar Montevideo 196 CNCTC6279, S. enterica subsp. arizonae CNCTC6478, Shigella dysenteriae CNCTC5204, Shigella boydii 197 CNCTC6338, Shigella flexneri serovar 1a I:2,4 CNCTC6370, Shigella flexneri CNCTC6378 serovar 4a IV:3,4. 198 Legionella pneumophila serogroups 1 (sequence type (ST) 1, ST23, ST62), and 10 (ST378) were obtained from 199 our collection at the Department of Microbiology, Faculty of Medicine and University Hospital in Pilsen, Czech 200 Republic. All members of the Enterobacterales order were cultivated on Mueller-Hinton agar at 35 °C overnight. 201 Legionella pneumophila was cultivated on BCYE agar in an atmosphere with 5% CO2 for 24 hrs. The full loop of 202 the bacteria was resuspended in 1 mL of 96% ethanol, centrifuged, and the pellet was allowed to dry at 98°C for 203 5 minutes. Twenty microliters of concentrated formic acid were added to the pellet and incubated at 98°C with 204 shaking (1200 rpm) for 15 minutes. After incubation, 100 μL of 5% pyridine was added to adjust the pH to 3.0. 205 The mixtures were then filtered using an Microcon® - 10 Centrifugal Filters, 10 kDa NMWL (14,000 g, 30 206 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint 8 minutes), and the bottom filtrate (50 μL) was used for derivatization using HD ligand as described above. For 207 microbes, further purification of derivatized saccharides is not required. Therefore, one microliter of the 208 reaction mixture was directly applied to the MALDI target. 209 210 211

212 Synthesis of HD Ligand 213 Initially, we tested the derivatization of saccharides using rosaniline. Despite the very efficient binding of 214 aldoses to the rosaniline, the disadvantage of the molecule was the presence of two efficient binding sites (NH 2) 215 and the requirement for a matrix for MALDI-TOF MS analysis. Therefore, we had tested several molecules 216 containing an aldehyde residue to block one of those amines. Using vanillin, we obtained a molecule with an 217 excellent ionization ability that does not require using any matrix in MALDI-TOF MS - an LDI-TOF (laser 218 desorption/ionization time-of -flight) MS approach. It was, however, necessary to enhance the stability of the 219 molecule by reductive amination of Schiff's base (Supplementary Figure 1). Since a conjugated system is 220 essential for efficient molecule ionization, we have tested an appropriate reducing agent (e.g., sodium 221 cyanoborohydride, borane pyridine complex, 2-methylpyridine borane complex). Finally, we have chosen a 2-222 methylpyridine borane complex as the reaction can proceed in one step. It is also important to note that high 223 concentrations of reducing agents, as recommended for conventional reactions, lead to the formation of leuco 224 base in rosaniline. That molecule has very low ionization ability and requires a classical matrix-based MALDI-TOF 225 MS setup. 226 The purity of the HD ligand synthesized in our laboratory was determined using LC-MS, and the structure was 227 verified by solariX magnetic resonance mass spectrometry (Bruker Daltonics, Bremen, Germany) 228 (Supplementary Figure 2). The quality of commercially synthesized HD ligand available as a 2,3-dichloro- 5,6-229 dicyano-1,4-benzoquinone salt (DDQ) was determined by nuclear magnetic resonance (Supplementary Figure 230 3). Both variants showed equal results in the following experiments. 231 232 Derivatization of Mono- and Disaccharides 233 The reaction was initially optimized using glucose and lactose. The best results were obtained in a low pH (3 - 4) 234 and a buffer not containing chlorine ions. Therefore, we selected a pyridine buffer of pH 3.0 for further 235 experiments. Using a MALDI-TOF mass spectrometer, the HD ligand's ionization ability was excellent for 236 detecting relevant signals of the conjugate with saccharides (Figure 1) without using any matrix (LDI-TOF MS). 237 Ionization was performed in a positive ion mode with the analytes detected as [M+H] + ions. All the saccharides 238 tested provide a signal with a mass-to-charge ratio that can be calculated by the following formula (see Table 1): 239 240 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint 9 m/z = [MHD ligand + MSaccharide - MH2O + H]+, 241 242 Where the M HD ligand is 423.195 (monoisotopic mass), M H2O represents the loss of water (-18.015) during the 243 formation of a Schiff base. 244 As demonstrated by glucose-6-phosphate, derivatization with HD ligand enables the detection of molecules with 245 strongly negatively charged groups like phosphates in a positive ion mode as well (Figure 1). 246 A purification step using C18 resin can be performed using the resin itself (as described in the methodology) or 247 by ZipTips®, which enhances ionization ability in LDI-TOF MS analysis of mono- and oligosaccharides. This step 248 removes unbound sugars that otherwise form a "sweet spot" and decreases the method's sensitivity. 249 In LC-MS analysis, derivatized mono-, and di-saccharides have also been detected at the expected mass-to -250 charge ratio (Table 1, Figure 2). The ion mobility of the saccharides allows their further analysis and 251 identification (e.g., discrimination between a disaccharide, lactose, and maltose – see Figure 2). Unfortunately, 252 we could not distinguish between D- and L-glucose by our instrument. 253 The method's sensitivity was tested using diluted glucose and lactose (0.01 mmol/L —100 mmol/L). For both 254

(LDI-TOF MS and LC-MS), the sensitivity was determined to be 0.1 mmol/L. 255 256 Acidic Digestion and Fischer-Speier Esterification 257 Using acidic digestion of oligosaccharides, we identified peaks that did not correspond with simple derivatized 258 mono- and oligosaccharides. Using monosaccharides ( D-glucose, L-fucose, D-xylose), formic acid, and isotopic 259 formic acid-13C, the peaks corresponding to esters formed in hydroxyl groups of saccharides can be detected. In 260 this reaction (Fischer-Speier esterification), the peaks are shifted by 28 g/mol, corresponding to formic acid-261 derived esters (Figure 3). 262 In glucose, the hydroxyl group at position 6 is almost completely esterified ( m/z 614) with a very low signal of 263 derivatized native glucose (m/z 586) (Figure 3). All except one hydroxyl group are also esterified with a different 264 ratio, showing signals at m/z 642, 670, 698. Similar results were obtained in D-xylose and L-fucose, showing no 265 signal with a non-esterified derivatized saccharide ( m/z 556 and 570). As observed in glucose, variants of all 266 esterified hydroxyl groups except one could be detected in LDI-TOF spectra (Figure 3). Based on those 267 characteristics, we hypothesize that only one hydroxyl group of vicinal diols can be esterified efficiently. The 268 esterification mechanism on D-glucose, D-xylose, L-fucose, and lactose was confirmed by precise molecular 269 mass determination (<1 ppm) using solariX XR FT-ICR mass spectrometer (Supplementary Figure 4). 270 As hexoses in acidic conditions and high temperatures can be dehydrated to form 5-(hydroxymethyl)furfural, we 271 also focused on identifying this molecule in the reaction. The 5-(hydroxymethyl)furfural can also be recognized 272 as a signal at m/z 532 in acidic conditions. In glucose and fructose, the 5-(hydroxymethyl)furfural intermediate 273 formed during saccharide dehydration [14] was identified in the spectra at m/z 550 (Figure 4, Figure 5). 274 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint 10 275 Analysis of Bacterial Polysaccharides by Acidic Digestion 276 Initially, formic acid was tested for non-specific fingerprinting of polysaccharides. As demonstrated in Figure 4 , 277 concentrated hot formic acid (98 °C) can efficiently hydrolyze glycosidic bonds. The same results were obtained 278 in raffinose and starch (data not shown). The method also allows the detection of common saccharides of the 279 bacterial cell wall, a muramic acid, containing an amine and carboxyl group (m/z 657) (Figure 6). 280 Based on those mono- and oligosaccharide results, the method was tested for fingerprinting bacterial 281 lipopolysaccharides. Comparing three different bacterial polysaccharides of Escherichia coli, the results 282 demonstrated different patterns. The same results were obtained for Escherichia coli O26:B6, E. coli O55:B5, E. 283 coli O111:B4 purified lipopolysaccharides, and crude bacteria (Figure 7). Similarly, all tested bacteria, including 284 Salmonella spp., Shigella spp., and Legionella pneumophila, provided specific patterns showing that the method 285 can be used for bacterial typing. 286 287 Analysis of Polysaccharides by Enzymatic Digestion 288 To analyze polysaccharides and lipopolysaccharides, -amylase and -amylase were tested using starch as a 289 positive control. In starch, mono —(m/z 586), di—(m/z 748), and trisaccharide ( m/z 910) can be observed using 290 both methods (LDI-TOF MS and LC-MS). We also detected distinct lipopolysaccharide profiles ( E. coli O26:B6, E. 291 coli O55:B5, E. coli O111:B4) similar to acidic digestion. 292 293 294

295 We describe here a novel method for derivatizing mono- and oligosaccharides that can be used for the 296 identification and analysis of those molecules, not only restricted to microbial origin. Initially, we focused on 297 developing a technique to identify bacterial cell-wall polysaccharides (i.e., lipopolysaccharides). We tested 298 several derivatizing agents, as saccharides cannot be easily ionized compared to peptides/proteins by MALDI-299 TOF mass spectrometry. Inspired by a detection of lactose fermentation using a basic fuchsin in diagnostic 300 bacteriology (e.g., Endo agar), we tested that molecule to form a Shiff base between the aldehyde group of 301 reducing sugars and the amine of the fuchsin. This complex could be derivatized by MALDI-TOF MS using a 302 standard matrix (e.g., 2,5-dihydroxybenzoic acid). The disadvantage of this process is the presence of two 303 efficient amine residues in the molecule, responsible for the polyvalent binding of tested saccharides. However, 304 the ability to modify saccharides with fuchsin and subsequent ionization showed excellent results. Therefore, 305 we decided to focus our research further on modifying this molecule. 306 After modifying the fuchsin by adding vanillin to one of the amines, a designated HD ligand, allowed the 307 ionization of the complex with saccharides without using the matrix (LDI-TOF MS). The complex can also be 308 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint 11 analyzed using LC-MS with a separation on the C18 reverse phase column. As demonstrated in the results 309 (Figure 1), some isomeric saccharides can also be distinguished by their ion mobility using trapped ion mobility 310 technology. 311 The HD ligand allows the derivatization and analysis of saccharides with different substituents, including 312 phosphates, in a positive ion mode. This makes the method universal to detect microbial ol igo- and 313 polysaccharides of different origins. However, it is necessary to use optimal reaction conditions for 314 derivatization, including a pH between 3 and 5. We also found that a high concentration of chlorine ions inhibits 315 the formation of a Schiff base (data not shown). Among the buffers tested, pyridine in a concentration between 316 1 and 10% with a pH adjusted by formic or acetic acid provided the highest binding efficiency. 317 Although we expected that the fuchsin-based system would also allow the detection of ketones (ketoses), we 318 could not find conditions that would enable efficient, stable binding of the HD ligand. This may be because the 319 HD ligand requires specific binding conditions or is unstable upon ionization during mass spectrometry. On the 320 other hand, however, ketoses are not common structures in bacterial cells. For analysis, they can be modified to 321 furfurals in acidic conditions that are analyzable by our system as well (see Figure 5). 322 When different options for the hydrolysis of the glycosidic bond of polysaccharides were tested, we found that 323 using formic acid, spectra containing many ions of different m/z with regular repetitions (+28) were detected. By 324 detailed analysis, including the reaction in 13C formic acid, we identified the signals as Fischer-Speier esters 325 formed in the hydroxyl groups of saccharide molecules. This behavior can be further used to identify and 326 analyze saccharides more accurately. For future experiments, the relative intensity of signals representing 327 esterified hydroxyl groups can be further verified to determine their position in the molecule. We also tested 328 acetic acid for esterification. However, its efficiency was very low compared with formic acid. In glucose , an 329 acetic acid-derived ester was formed at position six only (data not shown). Those findings were crucial for 330 further microbial polysaccharide experiments to understand the reaction. 331 An essential step in the analysis of microbial polysaccharides is to digest the molecule specifically, which usually 332 possesses a very high molecular weight. Similarly to derivatization methods, we tested many possibilities, 333 including specific lipopolysaccharide isolation, e.g., Bligh-Dyer solution and its modifications (data not shown). 334 Interestingly, common amylases (  and ) can efficiently digest bacterial polysaccharides to mono-, di-, and 335 trisaccharides. Peptidoglycan-specific enzymes (e.g., lysozyme) can also be used for specific analysis. A similar 336 approach can be optimized to analyze other microbial polysaccharides (e.g., galactomannan and glucan in molds 337 and yeasts) that are clinically relevant for rapidly diagnosing invasive fungal infection or detecting resistance to 338 antifungal drugs. 339 Finally, we selected a straightforward method that does not require any specific extraction of cell-wall 340 polysaccharides: incubating microbes in concentrated formic acid at 98 °C. This procedure blocks reactive 341 amines in the crude microbes, and the polysaccharides can be simultaneously digested into mono and 342 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint 12 oligosaccharides in a one-step process without previous specific extraction of the cell walls, which is usually 343 laborious. That approach is a typical example of applying Ockham’s razor and can be easily used in routine 344 diagnostic laboratories. 345 In all applications, however, it is crucial to filter the reaction mixture before the derivatization of saccharides to 346 remove undigested polysaccharides (> 3 – 10 kDa). Without this step, the spectra show an insufficient noise-to -347 signal ratio. 348 349 350

351 The method described here can be used to analyze microbial polysaccharides not only for epidemiological 352 typing and polysaccharide vaccine development but also to open the possibility of detecting those structures 353 directly in clinical specimens. Derivatization of saccharides and related molecules (e.g., furfurals) containing an 354 aldehyde group using the HD ligand is challenging for further analysis of those important biological structures by 355 mass spectrometry (LDI-TOF MS, LC-MS). 356 357 358

359 1. Son, A.; Kim, W.; Park, J.; Park, Y.; Lee, W.; Lee, S.; Kim, H. Mass Spectrometry Advancements and 360 Applications for Biomarker Discovery, Diagnostic Innovations, and Personalized Medicine. Int. J. Mol. Sci. 361 2024, 25, 9880. DOI: 10.3390/ijms25189880 362 2. Clark, A.E.; Kaleta E.J.; Arora A.; Wolk D.M. Matrix-assisted laser desorption ionization-time of flight 363 mass spectrometry: a fundamental shift in the routine practice of clinical microbiology. Clin. Microbiol. 364 Rev. 2013, 26, 547-603. DOI: 10.1128/CMR.00072-12 365 3. Kostrzewa, M. Application of the MALDI Biotyper to clinical microbiology: progress and potential. 366 Expert. Rev. Proteomics. 2018, 15, 193-202. DOI: 10.1080/14789450.2018.1438193 367 4. Hrabák, J.; Chudácková, E.; Walková, R. Matrix-assisted laser desorption ionization-time of flight (maldi-368 tof) mass spectrometry for detection of antibiotic resistance mechanisms: from research to routine 369 diagnosis. Clin. Microbiol. Rev. 2013, 26, 103-114. DOI: 10.1128/CMR.00058-12 370 5. Dortet, L.; Broda, A.; Bernabeu, S.; Glupczynski, Y.; Bogaerts, P.; Bonnin, R.; Naas, T.; Filloux, A.; Larrouy-371 Maumus, G. Optimization of the MALDIxin test for the rapid identification of colistin resistance in 372 Klebsiella pneumoniae using MALDI-TOF MS. J. Antimicrob. Chemother. 2020, 75, 110-116. DOI: 373 10.1093/jac/dkz405 374 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint 13 6. Weis, C.; Cuénod, A.; Rieck, B.; Dubuis, O.; Graf, S.; Lang, C.; Oberle, M.; Brackmann, M.; Søgaard, K.K.; 375 Osthoff, M.; Borgwardt, K.; Egli, A. Direct antimicrobial resistance prediction from clinical MALDI-TOF 376 mass spectra using machine learning. Nat. Med. 2022, 28, 164-174. DOI: 10.1038/s41591-021-01619-9 377 7. Farrington, M.; Rubenstein, D. Antigen detection in pneumococcal pneumonia. J. Infect. 1991, 23, 109-378 116. DOI: 10.1016/0163-4453(91)91900-i 379 8. Viasus, D.; Gaia, V.; Manzur-Barbur, C.; Carratalà, J. Legionnaires' Disease: Update on Diagnosis and 380 Treatment. Infect. Dis. Ther. 2022, 11, 973-986. DOI: 10.1007/s40121-022-00635-7 381 9. White, P.L. Progress on nonculture based diagnostic tests for invasive mould infection. Curr. Opin. 382 Infect. Dis. 2024, 37, 451-463. DOI: 10.1097/QCO.0000000000001060 383 10. Amicucci, M.J.; Nandita, E.; Galermo, A.G.; Castillo, J.J.; Chen, S.; Park, D.; Smilowitz, J.T.; German, J.B.; 384 Mills, D.A.; Lebrilla, C.B. A nonenzymatic method for cleaving polysaccharides to yield oligosaccharides 385 for structural analysis. Nat. Commun. 2020, 11, 3963. DOI: 10.1038/s41467-020-17778-1 386 11. Urakami, S.; Hinou, H. MALDI glycotyping of O-antigens from a single colony of gram-negative bacteria. 387 Sci Rep. 2024, 14, 12719. DOI: 10.1038/s41598-024-62729-1. 388 12. Han, L.; Costello, C.E. Mass spectrometry of glycans. Biochemistry (Mosc). 2013, 78, 710-720. DOI: 389 10.1134/S0006297913070031 390 13. Lageveen-Kammeijer, G.S.M.; Kuster, B.; Reusch, D.; Wuhrer, M. High sensitivity glycomics in 391 biomedicine. Mass. Spectrom. Rev. 2022, 41, 1014-1039. DOI: 10.1002/mas.21730 392 14. Kognou ALM, Shrestha S, Jiang Z, Xu C, Sun F, Qin W. High-fructose corn syrup production and its new 393 applications for 5-hydroxymethylfurfural and value-added furan derivatives: promises and challenges. J. 394 Bioresour. Bioprod. 2022, 7, 148 – 160. DOI: 10.1016/j.jobab.2022.03.004. 395 396 397 TABLE 398 Table 1. LDI-TOF MS and LC-MS analysis of mono-, disaccharides, and 5-(hydroxymethyl)furfural showing m/z 399 and ion mobility of derivatized molecules. 400 Saccharide m/z Ion Mobility [V.s/cm2] D-glucose 586 1.242 L-glucose 586 1.242 D-glucose-1,2-13C2 588 1.242 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint 14 glucose-6-phosphate 666 1.219 L-rhamnose 570 1.224 L-fucose 570 1.228 D-xylose 556 1.210 lactose 748 1.503 maltose 748 1.490 5-(hydroxymethyl)furfural 532 1.099 401 402 FIGURES 403 Figure 1 . Vanillyl-rosaniline (HD ligand) mechanism of saccharide’s derivatization, and LDI -TOF MS analysis of 404 selected saccharides. Mass spectra of glucose, D-glucose-1,2-13C2, and glucose-6-phosphate derivatized by HD 405 ligand measured on a rapifleX mass spectrometer in linear positive ion mode. Derivatized glucose is visible at 406 m/z 586; D-glucose-1,2-13C is visible as a signal at m/z 588; and glucose-6-phosphate as a signal at m/z 666. 407 408 Figure 2 . LC-MS spectra showing differentiation of disaccharide s’ isoforms by ion mobility determination. 409 Disaccharides maltose and lactose appear as signals at m/z 612 (HD ligand with the lost vanillin molecule) and 410 m/z 748 (intact HD ligand). Maltose complex (m/z 748) shows an ion mobility of 1.490 V.s/cm 2 and lactose 1.503 411 V.s/cm2. 412 413 Figure 3. Fischer-Speier esterification of glucose using concentrated formic acid at 98 °C and its derivatization by 414 HD ligand. Esterification of the hydroxyl group at position 6 (green color) was immediately detected. 415 Esterification of other positions was identified in different ratios. LDI-TOF MS spectra of monosaccharides 416 (negative control, L-fucose, D-glucose, D-xylose) esterified by concentrated formic acid (+m/z 28). 417 418 Figure 4 . LDI-TOF MS spectra of glucose (A), 5-(hydroxymethyl)furfural (B), lactose (C), and sucrose (D) after 419 digestion (lactose, sucrose) and esterification using concentrated formic acid at 98 °C for 10 minutes and 420 derivatization using HD ligand. 421 422 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint 15 Figure 5. LDI-TOF MS spectra show the formation of 5-(hydroxymethyl)furfural intermediate from fructose and 423 glucose after heating at 98 °C for 30 minutes ( m/z 550). The molecule was derivatized by HD ligand and 424 measured using LDI-TOF. 425 426 Figure 6. LDI-TOF MS spectra of glucose (A), muramic acid (B), Escherichia coli O26:B6 lipopolysaccharide (C), E. 427 coli O55:B5 lipopolysaccharide (D), and E. coli O111:B4 lipopolysaccharide (E) after digestion and esterification 428 using concentrated formic acid at 98 °C for 10 minutes and derivatization using HD ligand. 429 430 Figure 7. LDI-TOF MS spectra of Escherichia coli ATCC25922 (A), E. coli O111:B4 CNCTC5650 (B), E. coli O55:B5 431 CNCTC5874 (C), Salmonella enterica subsp. enterica serovar Enteritidis CNCTC5187 (D), Shigella dysenteriae 432 CNCTC5204 (E), and negative control (F) after digestion and esterification using concentrated formic acid at 98 433 °C for 10 minutes and derivatization using HD ligand. 434 435 436 SUPPLEMENTARY MATERIAL 437  Supplementary Figure 1 - Principle of low-scale preparation of Vanillyl-Rosaniline (HD) ligand using 2-438 methylpyridine borane complex as a reduction agent and purification by HPLC system. 439  Supplementary Figure 2 - Confirmation of Vanillyl-Rosaniline (HD) ligand structure using 15T solariX XR 440 FT-ICR mass spectrometer (Bruker Daltonics). 441  Supplementary Figure 3 - Quality control protocol (NMR spectra) of commercially prepared Vanillyl-442 Rosaniline (HD) ligand. 443  Supplementary Figure 4 - Confirmation of esterified D-glucose, D-xylose, L-fucose, and lactose using 15T 444 solariX XR FT-ICR mass spectrometer (Bruker Daltonics). 445 446 AUTHOR CONTRIBUTIONS 447 L.D. was responsible for laboratory work, interpretation of the results, designing novel derivatization agents, 448 and writing the manuscript. V.P. was responsible for laboratory work, the result interpretation, and the 449 manuscript's writing. P.N. was responsible for confirming HD ligand structure and analysis of esterified products. 450 J.H. was responsible for conceptualization, methodology, designing new derivatization agents, data 451 interpretation, and writing the manuscript. 452 453 ACKNOWLEDGMENT 454 We thank Dana Kralova and Andrea Pospechova for their excellent technical assistance. The study was 455 supported by the Czech Health Research Council grant Nr. NW24-09 -00464, the Charles University Grant Agency 456 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint 16 (GA UK) project Nr. 550225, and the National Institute of Virology and Bacteriology (Programme EXCELES, ID 457 Project No. LX22NPO5103) —funded by the European Union —Next Generation EU. The method has been 458 patented, includin g the HD reagent molecule and its variants (Czech National Patent PV 2024-48, 459 PCT/CZ2025/050014). 460 .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint .CC-BY 4.0 International licenseavailable under a was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint (whichthis version posted May 14, 2025. ; https://doi.org/10.1101/2025.05.13.653920doi: bioRxiv preprint