Mass spectrometric analysis of lipid A obtained from the lipopolysaccharide of Pasteurella multocida

Haemorrhagic septicaemia is mainly caused by an opportunistic pathogen, Pasteurella multocida, a major threat to the livestock dependent economies. The main endotoxins are lipopolysaccharides. The lipid A, a key pathogenic part of lipopolysaccharides, anchors it into the bacterial cell membrane. Hence, profiling of the lipid A is important to understand toxicity of this pathogen. Despite a significant progress made on glycan analyses of core regions of lipopolysaccharides from various P. multocida strains, the structure of lipid A has not been reported yet. The lipid A of P. multocida type B:2 was analyzed using ESI-MS/MS to identify the acylation patterns, number and length of various acyl fatty acids, phosphorylation level and lipid A modifications. The MSn data revealed the existence of multiple lipid A variants, i.e. mono and bisphosphorylated hepta-, hexa-, penta- and tetra-acylated structures, decorated with varied levels of 4-amino-4-deoxy-l-arabinose (Ara4N) on C-1 and/or C-4′ phosphate groups of proximal and distal glucosamine lipid A backbone. The detailed mass spectrometric analyses revealed that even within the same class, lipid A exhibits several sub-variant structures. A primary and secondary myristoylation at C-2, C-3, C-2′ and C-3′ was observed in all variants except hepta-acylated lipid A that carried a secondary palmitate at C-2 position. The lipid A profiling described herein, may contribute in exploring the mechanisms involved in endotoxicity of P. multocida type B:2 in haemorrhagic septicaemia disease.


Introduction
Pasteurella multocida is an opportunistic Gram-negative bacterial pathogen associated with various septic diseases of wild as well as domestic animals. 1 Based on capsular chemistry, P. multocida strains have been classied into 5 sub-groups i.e. A, B, D, E and F; whereas, on the basis of the variations in their lipopolysaccharides (LPS), there are around 16 Heddleston serovars. 2,3 Haemorrhagic septicaemia (HS) disease is mostly associated with serotypes B:2 and E:2 of P. multocida subsp. multocida (Carter and Heddleston system), corresponding to the 6:B and 6:E serotypes in Namioka-Carter system. 4 HS is endemic in ungulates, predominantly water buffalo (Bubalus bubalis) and cattle in most parts of tropical Asia and Africa. 5,6 In fact, HS is classied among list B cattle diseases notiable to the Office International des Epizooties (OIE) with 100% mortality rate in infected animals if not treated at early stage. 7 The diseased animals usually collapse and die within a few hours to a few days aer the onset of the illness. Sudden death with few or no clinical signs can also be seen while the symptomatic animals, especially buffalo, rarely recover. 4,8 This makes HS, the most important disease of bovines in terms of economic impact in various tropical and sub-tropical parts of the world. 9 In Pakistan, having cattle and buffalo populations of 47.8 and 40 million respectively, 8 young buffalo, being the most affected group, 10 has been reported as an important reservoir of P. multocida. 4 The pathogenesis of the P. multocida is poorly understood as compared to other Gram-negative bacteria. The haemagglutinin surface adhesin, 11 capsular polysaccharides and LPS are critical virulence factors. 12 The LPS plays a signicant role in adhesion of P. multocida to the respiratory epithelium of host. 13 The mutants attenuated to acapsular as well as truncated LPS, showed no or signicantly reduced virulence and found to be more prone to the host antimicrobial peptides. [14][15][16] Indeed, intravenous administration of the puried LPS induced the clinical signs resembling haemorrhagic septicaemia in buffalo i.e. endotoxic shock. 17 Notably, in LPS, the lipid A part is considered as the active endotoxic part leading to Hepta-acylated 3-OH-C 14 3-OH-C 14 P 11 1927 05 Ara4N-P 3-OH-C 14  A signicant progress has been made towards characterization of the polysaccharide components of LPS i.e. inner and outer core regions, produced by many P. multocida isolates belonging to serovars 1, 2, 3, 5, 9 and 14. 12,66-68 Moreover, the outer glycan part of LPS obtained from P. multocida genome strain Pm70 has also been structurally characterised. 39 However, despite its strong correlation with endotoxicity, the structure of lipid A part of the LPS of P. multocida has not yet been determined. Thus, in order to ll this knowledge gap, lipid A structure of pathogenic P. multocida type B:2 isolate (termed as PM2) from Pakistan was investigated using mass spectrometric technique that will help in better understanding of pathogenicity of this bacterium.

Results
The characteristic colonies of P. multocida: grayish, translucent and mucoid having around 1 mm diameter, were obtained on CSY agar plates. The molecular identication of the isolate resulted in amplication of 460 bp fragment of KMT1 gene, 590 bp fragment of 6b gene and 758 bp fragment of bcbD gene (Table 2), conrming the isolate identity as P. multocida belonging to B:2 type (Fig. 1a). No amplication was observed in case of negative controls. Large scale fermentation, resulted in 5 g L À1 yield of wet cell pellet and the LPS extraction yielded 55 mg of puried LPS per gram of bacterial cell pellet. The DOC-PAGE analysis showed the typical heterogenic patterns of the puried rough type LPS (Fig. 1b).

Characterization of lipid A structure
ESI-MS/MS analysis of lipid A. The intact lipid A sample, obtained from the mild acid hydrolysis, was dissolved in the mixture of methanol and chloroform (1 : 1) and injected to ESI-MS using direct syringe pump. The full scan data at negative ion mode is shown in Fig. 2, which revealed that lipid A structure of P. multocida exhibited a high-level of heterogeneity, comprising of different fatty acyl chains and covalent modications. Apparently, lipid A consists of hepta-, hexa-, penta-and a minor variant of tetra-acylation having 7, 6, 5 and 4 fatty acid acyl chains, respectively, attached to the glucosamine disaccharide backbone, which is mono and/or bisphosphorylated at C-1 and C-4 0 positions with extended substitution(s) of one or two 4amino-4-deoxy-L-arabinopyranose (Ara4N) molecules (Fig. 2).
Notably, these lipid A variants show highly capricious endotoxicity. Multiple lipid A structures complicate interpretation and can have variable effects on innate host immune response. The immunogenic properties of lipid A are directly correlated with the number of acyl chains, phosphate groups and covalent modications, i.e., the incorporation of palmitate, the addition of PEtN and/or Ara4N. 38 Usually, bisphosphorylated asymmetric (4/2) hexa-acylated lipid A induces the strongest proinammatory immune reactions aer binding to TLR4, whereas, penta-acylated lipid A exhibits lower binding activity. Overall, bisphosphorylated versions of hepta-acylated (4/3), asymmetric hexa-acylated (4/2) and penta-acylated (3/2) lipid A activate NF-kB pathway and are considered as TLR4 expression agonists. Conversely, bisphosphorylated tetra-acylated (2/2) lipid A, monophosphorylated penta-acylated (3/2) and bisphosphorylated symmetric (3/3) hexa-acylated lipid A antagonize proinammatory NF-kB pathway. 40,41 Overall, endotoxicity of the pathogens i.e. P. multocida, is the collective impression of all these lipid A variants. Therefore, the determination of precise structure of lipid A species is important to visualize the endotoxicity as well as pathogenicity of any Gramnegative bacteria. Hence, the accurate structure of these lipid A variants of the tested P. multocida PM2 strain have been illustrated below in descending order, on the basis of their mass/ charge (m/z) values and is summarized in Table 1. The conjugation sites were conrmed by correlating the acyl group fragmentations with the cross-ring fragmentations that are denoted in the manuscript according to the nomenclature proposed by Domon and Costello 42 and Morrison et al. 43 Proling of hepta-acylated lipid A variants. Hepta-acylated lipid A structures exhibit seven variants in full scan spectrum at m/z 2325, 2194, 2166, 2114, 2063, 2035 and 1983. Their ion abundance, phosphorylation levels, covalent modications and fatty acyl chains have been summarized in descending order in Table 1 (S. # 1-7). The structures of these variants have been determined through tandem mass spectrometric investigations. A detailed fragmentation analysis of one of the variants showing the ion peak at m/z 2062.5, is given here as a representative of this class (Fig. 3). The ESI-MS 2 fragmentation (@CID 24) of the molecular ion at m/z 2062. The MS 2 fragmentation pattern of the precursor ion (m/z 2062.5) and its analysis suggested its structure as bisphosphorylated hepta-acylated disaccharide with 4/3 stoichiometric distribution of acyl chains. GlcN-I : GlcN-II disaccharide carries This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 30917-30933 | 30921 a 3-hydroxymyristate appendage at C-2 (primary position) attached via an amide linkage that further has an ester linkage with a palmitate (secondary acyl chain). The C-3 position also has a 3-hydroxymyristate moiety but through ester linkage. Another 3-hydroxymyristate is anchored at C-2 0 (primary position) utilizing its hydroxyl group for ester linkage with a myristate (secondary). The C-3 0 exhibits yet another 3hydroxymyristate (primary position) through ester linkage with O-myristoylation (secondary) ( Table 1,  The ion peak at m/z 2034.8 emerged due to loss of 28 Da (variation of ethylene group C 2 H 4 ) from m/z 2063 indicating that one of the secondary myristoyl chains is replaced with laurate [C 12:0 ]. Notably, almost all of the lipid A variants i.e. with hepta-, hexa-, penta-and tetra acylation, both in their monophosphorylated and bisphosphorylated forms and having covalent modication with Ara4N, demonstrated the substitution of myristate with laurate (reduction of m/z by 28 Da), suggesting a minor alternative lipid A variant. The exact position of laurate fatty acyl chain was determined to be 2 0 through fragmentation of hexa-acylated precursor ion at m/z 1716.4 (Fig. 7).
The ion peak at m/z 2194 (Fig. 4) represented the addition of 131 Da to the bisphosphorylated hepta-acylated lipid A structure corresponding to the peak at m/z 2063, indicating that one of the phosphate group (more likely at 4 0 -position of GlcN-II) is decorated with Ara4N (D ¼ 131 Da). The presence and position of mono and di-Ara4N on phosphate groups have been further conrmed in the following data analysis of hexa-and pentaacylated lipid A species and is summarized in Table 1 Table 1, S. # 1). In short, the proposed structure for m/z 2325.0 comprises of a diphosphorylated di-glucosamine backbone primarily acylated with 3-OH-C 14:0 at the 2, 3, 2 0 , and 3 0 positions and secondarily acylated with C 16:0 at position 2, and with C 14:0 at the 2 0 as well as 3 0 positions while both phosphate groups decorated with Ara4N.
Proling of hexa-acylated lipid A variants. The current data revealed the presence of hexa-acylated lipid A variants of P. multocida with a high degree of heterogeneity in terms of the degree of acylation, phosphorylation and covalent modication with Ara4N, as has been summarized in Table 1   This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 30917-30933 | 30923 3a cleavage of 3-OH-C 14:0 , respectively (Fig. 5B). The m/z 1272.0 and m/z 1253.9 were produced by the simultaneous loss of phosphate with one of the two pairs of fatty acids i.e. (i) C 14:0 by 3 0 3 cleavage paired with 3-OH-C 14:0 by 3a cleavage and (ii) C 14:0 by 3 0 3 cleavage paired with C 14:0 by 3b cleavage, respectively. The removal of phosphate together with cleavages at 3 0 3, 3 0 b and 3a gave rise to the peak at m/z 1045.7. Similarly, the removal of four fatty acids produced the ion peak at m/z 1027.7.
In the ion peak at m/z 1716.4, one of the secondary myristate was found to have been replaced with laurate (D ¼ 28 Da). The exact position of C 12:0 was determined by MS 2 of m/z 1716.4, which yielded daughter ions by losing one of the fatty acyl chains at m/z 1516 (removal of laurate at 2 0 3 cleavage site), m/z 1488 (3 0 3 cleavage of myristate) and m/z 1472 (3a cleavage of 3hydroxymyristate) (Fig. 7). The product ions arising from the elimination of two acyl chains were also detected at m/z 1288 [elimination of C 12:0 (2 0 3 cleavage) together with C 14:0 (3 0 3 cleavage)], m/z 1272 [elimination of C 12:0 (2 0 3 cleavage) and b-OH-C 14:0 (3a cleavage) as a ketene derivative], m/z 1261.8 [elimination of C 14:0 (3 0 3 cleavage) and b-OH-C 14:0 (3b cleavage)], and at m/z 1243.8 [elimination of C 14:0 (3 0 3 cleavage) and 3-OH-C 14:0 (3a cleavage)]. The plausible elimination of C 14:1 both by charge-remote (loss of a free fatty acid) and chargedriven processes (loss of a ketene derivative) suggested its location at O-3 0 as well as its secondary substitution by C 14:0 (ion peak of m/z 1243.8). 44 The product ions at m/z 1431.4, 1230.7 and m/z 1203.1 represented 0,2 A 2 fragments formed from the ions at m/z 1716.2, 1516.1 and 1488.0, respectively. Whereas, cross-ring fragments 0,4 A 2 at m/z 944.6 and m/z 916.6 were produced by m/z 1516 and 1488, respectively. These ions yielded by the in-source cross-ring fragmentation further established the distribution of the iden-tied fatty acids on specic positions. Aer interpreting the types of fragment ions, it was possible to dene fatty acid distribution on both GlcN residues. From this data analysis, it is suggested that the ion peak at m/z 1716.4 can be attributed to mono-phosphorylated, hexa-acylated lipid A form consisting of two GlcN, four primary acyl chains of b-OH-C 14:0 at 2, 3, 2 0 & 3 0 positions and two secondary acyl chains C 12:0 and C 14:0 at 2 0 and 3 0 positions, respectively.
Proling of hexa-acylated lipid A variant m/z 1756.3. In the full scan MS data (Fig. 3), the ion at m/z 1756. 3 (Fig. 8A). The loss of 228 mass units at from the parent m/z 1756.3, obtaining a stable base peak at m/z 1528.1, indicated the removal of a secondary acyl group C 14:0 from the b-hydroxy acyl chain at 3 0 3 position rendering the remaining structure unsaturated. The cleavage of both primary and secondary acyl groups from the C-3 0 of GlcN-II at 3 0 b position through charge-driven process (leaving OH behind) produced a fragment at m/z 1320.0, while further fragmentation at 3 0 a cleavage site through charge-remote process generates a product ion of m/z 1302 with unsaturation between C-3 0 and C-4 0 of GlcN-II 44 (Fig. 8A).
The MS 3 of the base peak m/z 1528.1, obtained from m/z 1756.3, generated product ions at m/z 1510, 1320, 1302, 1272, 1254, 1064, 1046, 1005, 946, 796, 778 and 718 (Fig. 8B). The ion at m/z 1510 was generated due to the water loss presumably C-1 on GlcN-I, resulting in an unsaturation between C-1 and C- Similarly, the generation of a peak at m/z 1253.8 can be spotted due to the loss of water from C-1 position along with the secondary deacylation from C-2 (23 cleavage) (Fig. 8B). A less frequent 2z cleavage was also observed in combination with 3 0 b cleavage and water loss from C-1 leading to a product ion at m/z 1063.7.
The ESI-MS 2 spectrum of m/z 1756.3 followed by MS 3 of its daughter ion at m/z 1528.1 revealed its structure as hexaacylated mono-phosphorylated variant with three primary acylations of 3-OH-C 14:0 at C-2, C-2 0 and C-3 0 positions. The rst two of these primary acyl chains are amide linked while the remaining one is ester linked. The secondary acylation of C 14:0 was also observed at C-2 0 and C-3 0 while a C 16  Proling of penta-and tetra-acylated lipid A variants. The full scan MS data also revealed the presence of penta-acylated lipid A variants in LPS of P. multocida. Out of these, the prominent precursor ion at m/z 1518.1 was subjected to MS 2 fragmentation, which produced the daughter ions at m/z 1500.  (Fig. 9). The fragment ions at m/ z 1081.7 and 1063.7 appeared aer the 3 0 b and 3 0 a cleavages of both primary and secondary acyl chains through charge-driven and charge-remote processes. 44 The fragments at m/z 1272.0 and 1045.7 are the dehydrated versions of the fragments with m/ z 1290.0 and 1081.7, respectively, as depicted in Fig. 9. The fragmentation data revealed the structure corresponding to the precursor at m/z 1518.1 as a mono-phosphorylated pentaacylated lipid A having 3-hydroxymyristate as primary acyl chains at C-2, C-2 0 and C-3 0 and C 14:0 as secondary acyl chains at C-2 0 and C-3 0 positions ( Fig. 9 and Table 1, S. # 35). The distribution of fatty acids was corroborated by cross-ring fragments 0,2 A 2 (m/z 1232.9, 1004.6 and 796.5) and 0,4 A 2 (m/z 1172.8, 946.6 and 736.4). 24 Successful proling of m/z 1518.1 contributed in correlation of other penta-acylated lipid A variants (Fig. 10). The ion at m/z 1490.2 (D ¼ 28 Da) indicated a variation in fatty acid ruler such that a laurate (C 12:0 ) is present at secondary acyl position of C-2 0 with 4/1 fatty acyl distribution as described in the analysis of m/z 1518.1 (Table 1, S. # 36). There could also be a dehydrated moiety at m/z 1499.4 owing to the water loss from the anomeric C-1 position of the ion corresponding to the base peak at m/z 1518.0 (   This journal is © The Royal Society of Chemistry 2020 RSC Adv., 2020, 10, 30917-30933 | 30927 The ion peak at m/z 1533.4 corresponded to another set of mono-phosphorylated penta-acylated lipid A molecules having 3-OH-C 14:0 appendages at C-2, C-3, C-2 0 and C-3 0 positions while only one secondary C 14:0 acyl chain at C-2 0 . Its bisphosphorylated (D ¼ 80 Da) and Ara4N (D ¼ 80 + 131 Da) analogues were spotted at m/z 1614 and 1745, respectively (Table 1, S. # 31 & 28). Similarly, there was another ion peak at m/z 1757 (Table 1, S. # 27), which indicated penta-acylation with 3/2 pattern and an Ara4N moiety on C-4 0 phosphate group and a palmitate at secondary position of C-2. By the addition of 131 Da, a second Ara4N on C-1 phosphate group was also observed at m/z 1887.6 (Table 1, S. # 24). Another 3/2 penta-acylated lipid A appeared at m/z 1546 having only one phosphate group at C-4 0 . The ESI-MS/MS data also depicted a small proportion of tetra-acylated lipid A variants at m/z 1439, 1308, 1290 and 1262 (

Discussion
Haemorrhagic septicaemia (HS) is associated with P. multocida (serotype B:2 and E:2) infections in cloven-hoofed ungulates. 45 The disease occurs in acute, sub-acute and chronic forms. Its initial phase includes temperature elevation, a stage of respiratory involvement and the terminal phase involves septicaemia and recumbence leading to death. 9 Although the HS disease etiology is strongly related to the endotoxicity of lipid A component of the bacterial LPS, the structure elucidation of the lipid A has not been reported yet.
ESI-MS analysis of lipid A obtained from P. multocida isolate PM2 in current study, showed a high degree of complexity and heterogeneity. The overall heterogeneity of lipid A varies from hepta-acylated to tetra-acylated variants decorated with varied level of Ara4N moiety on C-1 and/or C-4 0 phosphate groups of proximal and distal glucosamine lipid A backbone. The full scan mass spectrometric data revealed that even within the same class of lipid A, several sub-variant structures exist, which have been summarized in Table 1.
The hepta-acylated lipid A consists of seven structures with 4/3 acyl distribution on GlcN-II : GlcN-I sugar moieties of lipid A backbone, respectively (Table 1, S. # 1-7). All of these structures have four b-hydroxy C 14:0 primary acyl chain, palmitate secondary acyl chain at C-2 position and myristate at C-3 0 position. The secondary acyl chain at C-2 0 position in majority of the hepta-acylated variants is C 14:0 , whereas, C 12:0 is also found as a minor alternative ( Similar to other Gram-negative bacteria, the palmitoylation as well as regulated addition of Ara4N in PM2 isolate may be environment-dependent and can contribute in protecting the bacterium from certain host immune defenses. 38 Addition of palmitate fatty acyl chain to lipid A is catalyzed by PagP, a palmitoyl transferase enzyme, which is the component of PhoP-PhoQ regulatory system. 46 The pagP gene and its functional homologs are distributed among a narrow group of primary pathogenic bacteria (Salmonella species, 47,48 E. coli, 33 Proteus mirabilis, 49 Acinetobacter baumannii, 50 Pseudomonas aeruginosa, 51 L. pneumophila, B. bronchiseptica, Y. enterocolitica, and Y. pseudotuberculosis 52 ) and its regulation is oen correlated with their pathogenic lifestyle. 53 The addition of palmitate not only strengthens the outer membrane of the bacterial cell wall through the hydrophobic and van der Waals interactions, which prevents translocation of the cationic antimicrobial peptides across the bilayer, but also contributes in evading the host immune response by lowering TLR4 related signal transduction pathway as compared to hexa-acylated lipid A variants. 53 Similarly, insertion of Ara4N is catalyzed by ArnT through activation of PmrA transcription factor. Being positively charged at neutral pH, Ara4N can neutralize the phosphate groups, consequently, reducing bacterial susceptibility to the cationic host antimicrobial peptides and polymyxin antibiotics. 38 The hexa-acylated lipid A of the PM2 isolate exhibited a high degree of variation with most of the structures bisphosphorylated at C-1 and C-4 0 positions. In ve structures, phosphates were decorated with Ara4N at C-4 0 positions and only one structure demonstrated Ara4N sugars on both of its phosphate groups (Table 1: S. # 8-23). There was a signicant heterogeneity of fatty acyl patterns with respect to their relative positions and variation in chain length of fatty acids. Most of the hexaacylated structures exhibited 4/2 fatty acyl stoichiometry with C 14 carbon ruler at primary and secondary acyl chains. The presence of laurate (C 12:0 ) at secondary acyl chain at C-2 0 position was also spotted in minor proportion (Table 1: S. # 11, 17 and 23).
The bisphosphorylated 4/2 structures have been reported to warrant maximal immunostimulatory activities of LPS. 54,55 In P. multocida infections these structures presumably induce high inammatory response of the innate immune system, thus playing a pivotal role in causing HS. However, hexa-acylated structures having 4/2 fatty acyls also exist in which palmitate occupy secondary position at C-2 with absence of 3-OH-C 14:0 at C-3 positions (Table 1: S. # 9, 12, 15, 19 and 21). The outer membrane-bound lipid A 3-O-deacylase, encoded by the pagL gene removed the fatty acyl chain from C-3 position of heptaacylated lipid A. The PagL enzyme was initially reported in Salmonella, but later its homologs were found to be widely distributed among pathogenic and non-pathogenic Gramnegative bacterial species. 33 The presence of hexa-acylated lipid A variants without fatty acyl chains at C-2 position hints the presence of PagL in P. multocida as well. The addition of palmitate at C-2 by PagP and the removal of primary acyl chain at C-3 by PagL have been reported to get activated by cationic antimicrobial peptides induced through the PhoP transcription factor. 46 Hence, these structures are likely to contribute in evading the host immune defense in P. multocida infections as well. Moreover, hexa-acylated lipid A structures with 3/3 stoichiometry (Table 1: S. # 14 and 18), were also spotted in a minor proportion. Rather than acting as agonist, the 3/3 stoichiometric structures have been found to antagonize TLR4 receptor. 40 Like hexa-acylated lipid A, penta-acylated class also exhibited several variants and most of those were bisphosphorylated with Ara4N placement only at C-4 0 position in four structures, while three structures displayed Ara4N sugars on both of its phosphate groups (Table 1: S. # [24][25][26][27][28][29][30][31][32][33][34][35][36][37]. Signicant heterogeneity was found in the positions and types of fatty acyl chains with predominant 4/1 stoichiometry (Table 1: S. # 26, 29, 30, 32, 35-37) as demonstrated by a base peak at m/z 1518. Fatty acid distribution with 3/2 stoichiometry demonstrated two subcategories, one with secondary palmitate at C-2 (Table 1: S. # 24, 27 and 33), while the second contained two primary acyl chains at C-2 and C-3 positions (Table 1: S. # 25, 28, 31 and 34). In small proportion, mono-phosphorylated tetra-acylated variants having 3/1 stoichiometry were also spotted ( Table 1: S. # [38][39][40][41] in the lipid A of the PM2 isolate. TLR4 activation has been markedly decreased by tetra-and penta-acylated lipid A structures as compared to hexa-acylated variants. Even antagonistic trend to the TLR4 activity has been found in pentaacylated lipid A having 3/2 fatty acyl distribution stoichiometry lacking phosphate at C-1 position, as well as in bisphosphorylated tetra-acylated lipid A structures. 40,56 Gram-negative bacteria modulate lipid A structures to evade host immune response e.g. H. pylori synthesizes a hexa-acylated lipid A intrinsically, but trims it down to tetra-acylated by the action of deacylase enzymes and also removes phosphates (by LpxE & LpxF) to switch the biological activity from TLR4 agonistic to antagonistic, which can turn its infection to a chronic one for the duration of host life. 57 The lipid A of PM2 depicts a variety of acylation, and other modes of modications e.g. phosphorylation and Ara4N decoration. The presence of both phosphate and Ara4N moieties can be located on C-1 and/or C-4 0 positions of the basic glucosamine backbone. The ESI-MS n further reveals the C-2, C-3, C-2 0 and C-3 0 bonded with primary and secondary myristoylation except few lipid A variants, where there is palmitoylation at C-2 position (Table 1, S # 1 -7, 9, 12, 14, 15, 18, 19, 21, 24, 27 and 33) and lauric acid at secondary positions of C-2 0 carbon ( Table 1, S # 3 , 4, 6, 7, 11, 17, 23, 37 and 41). There could be a strong correlation between the heterogeneous lipid A modication capabilities of P. multocida and its sporadic HS epidemics, which needs further investigations.
In addition to the number and length of the acyl groups, the stereochemistry and degree of their unsaturation may also have important biological implications. In our study, all the acyl groups bonded to the disaccharide backbone of lipid A happened to be saturated (Table 1), but in many other microbial strains, the unsaturated fatty acids may also exist. 58 The carboncarbon double bond on these unsaturated acyl groups can be located via a recently developed technique using visible-light activated [2 + 2] cycloaddition reaction and tandem mass spectrometry. 59

Experimental
Material and method Bacterial identication. P. multocida isolate, named PM2 was taken from Veterinary Research Institute, Peshawar and was revived in mice at Animal House Facility, National Institute for Biotechnology and Genetic Engineering (NIBGE), Faisalabad, Pakistan aer approval from Institutional Bioethics Committee, ref. no: NIBGE/Bioethics/2014/ 02. The mice blood was directly streaked on casein sucrose yeast (CSY) agar plate and kept overnight at 37 C. A well isolated colony having typical morphology of P. multocida was inoculated into 3 mL of CSY broth and incubated at 37 C for overnight and preceded for molecular conrmation by polymerase chain reaction (PCR). Briey, the DNA of overnight bacterial culture was extracted using genomic DNA purication kit (Thermo Scientic # K0512). Agarose gel electrophoresis (1%) and Nano-Drop (Thermo Scientic Nano-Drop 20000c) were used to check the integrity and quantication of the DNA respectively. The PCR of the extracted DNA was performed targeting two genes: KMT1 gene fragment of 460 base pairs (bp), specic for P. multocida and a 590 bp fragment of 6b gene, specic for its serotype. To identify its sub serogroup B:2 type, the 758 bp bcbD gene was also targeted. 60,69 The composition of PCR mixture and thermal cycler conditions were kept same as reported earlier. 61 The sequences of the primers used are mentioned in Table 2.

LPS extraction and purication
The PCR conrmed PM2 isolate was grown in sterile CSY broth using 6 Â 2 L asks at 37 C and 180 rpm. Aer overnight growth, the cells were treated with formalin (0.5% v/v) for 2 hours and then harvested by 7000 rpm centrifugation at 10 C. The wet cell biomass was subjected to LPS extraction by hotphenol method. 62 The extracted LPS were dialyzed for 3 consecutive days against deionized water using ZelluTrans/Roth dialysis membrane (T2 MWCO 6000-8000 cat # E665.1), with 2 water changes daily and lyophilized. The LPS were analyzed by deoxycholate-polyacrylamide gel electrophoresis (DOC-PAGE) followed by silver staining. 63,64 Acid hydrolysis and extraction of lipid A For lipid A extraction, the puried LPS was subjected to mild acid hydrolysis. 65 Briey, 10 mg L À1 of the puried LPS was dissolved in 1% glacial acetic acid and heated at 95 C using water bath for 90 minutes. The solution was cooled and ultracentrifuged at 96 000g at 4 C for 5 hours with a drop of 1 M CaCl 2 . The pellet was dissolved in chloroform : methanol :water solution as 2 : 1 : 1 (v/v/v) ratio and kept overnight. The chloroform extract was dried under nitrogen and redissolved in chloroform : methanol : acetonitrile solution as 3 : 1 : 1 (v/v/v) ratio and used for Electron Spray Ionization (ESI) mass spectrometric analysis.

Mass spectrometry of lipid A
The ESI-MS/MS was conducted using LTQ XL mass spectrometer (Thermo Electron Corporation, USA). The sample was analyzed through direct injection to the ESI probe set at negative ionization mode by syringe pump at a ow rate of 0.3 mL min À1 . The temperature of capillary was kept at 198 C. Aer getting the full scan mass spectrum data in the range of m/ z 100 to m/z 3000, the generated ions were isolated in the ion trap and fragmented by collision induced dissociation (CID) energies ranging from 10 to 40 according to the stability of the target precursor ions selected for tandem mass spectrometry.

Data analysis
The data was analysed using Xcalibur 2.2 and Chemdraw sowares. The identication was conducted using MS/MS fragmentation patterns of various precursor ions. The details for each specic fragmentation can be seen in the footnotes of all the corresponding gures.

Conclusion
The pathogenic P. multocida PM2 isolate was found to exhibit a variety of lipid A variants, i.e. mono and bisphosphorylated hepta-, hexa-, penta-and tetra-acylated versions with the 4/3, 4/ 2, 3/3, 3/2, 4/1 and 3/1 patterns on the Glc-II and Glc-I backbone, respectively. It can also decorate its lipid A with the Ara4N sugar moiety, either at one or both of the phosphate moieties at C-1 and C-4 0 positions. Majority of the lipid A variants are hexaacylated with a total of 16 observed variants, followed by 14 penta, 7 hepta and 4 tetra-acylated variants. Based on the relative abundance, the penta-acylated version at m/z 1518.1 is the most abundant, followed by hexa-acylated lipid A variant having an m/z 1955.5. The presence of highly endotoxic hexa-acylated lipid A variants may have a leading role in inducing the disease, the haemorrhagic septicaemia. While covalently modied hexa-acylated lipid A variants (palmitoylation at C-2 position and Ara4N placement at C-4 0 and/or C-1 phosphate positions) can contribute in evading the host immune responses.

Conflicts of interest
The authors declare no conict of interests and no permission is required for publication.