Virulence Phase And Lipid A Structural Modifications Biology Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Bordetella bronchiseptica is a respiratory pathogen in mammal species. Its endotoxin is a potent virulence factor. The lipid A moiety of lipopolysaccharides is responsible for most of the toxic properties while the polysaccharide moiety carries the antigenic determinants of the molecule. We have compared the lipid A structures of two human and one rabbit isolates in their virulence and non-virulence phases. The occurrence of palmitate in these structures at one or two sites was correlated with virulence phase, the recently identified glucosamine modifications of Bordetella lipids A were not.


Lipopolysaccharide (LPS), a complex glycolipid, is the major structural component of the Gram-negative bacterial outer membrane. The general structure of the bacterial LPSs consists of three distinct domains: a hydrophobic moiety called lipid A, a core oligosaccharide containing 2-keto-3-deoxy-octulosonic acid (Kdo), and a serospecific O polysaccharide composed of repeating oligosaccharide units (Caroff and Karibian, 2003). Lipids A anchor LPS molecules in the outer leaflet of the external bacterial membrane. The core has only limited structural variability compared to the O antigen, which is present in the LPS of most, but not all, bacterial species. It comprises the most variable part of LPSs and confers serotype specificity on the bacteria.

LPS forms a ligand with the TLR4-MD2-CD14 complex, which is present on many cell types including macrophages and dendritic cells (Poltorak et al., 1998; Hoshino et al., 1999; Qureshi et al., 1999). The stimulation of this receptor complex leads to activation of signaling pathways, resulting in induction of antimicrobial genes and release of cytokines, thereby initiating inflammatory and immune defense responses (Miller et al., 2005). Most of the biological activities of LPS have been associated with their lipid A moiety. Lipid A itself can be modified via mechanisms such as acylation, deacylation, fatty acid hydroxylation, and phosphate group substitution with aminoarabinose, galactosamine, or phosphoethanolamine (Dixon and Darveau, 2005; Trent et al. 2006; Geurtsen et al., 2007). These modifications can play a significant role in modulating host responses to infection. We showed recently that bacteria of the Bordetella genus such as B. bronchiseptica and B. pertussis are capable of modifying their lipids A by substitution of both phosphate groups with glucosamine (GlcN) (Marr et al., 2008). Up to now this substitution had not been observed in any other bacterial genus. Its presence was shown to strongly increase the biological activities of purified LPS and of the whole bacteria (Marr et al., 2010).

The Bordetella genus currently contains nine species, most of which are respiratory tract pathogens, and the most extensively studied ones include B. pertussis, B. parapertussis, and B. bronchiseptica. Although these three pathogens are very closely related genetically (Parkhill et al., 2003), their LPS molecules have significant differences (Caroff et al., 2002). B. petrii is the only known facultative anaerobic species of the genus with an environmental origin; it seems to be more closely related to a common, putative ancestor than the other pathogenic Bordetella species (von Wintzingerode et al., 2001). Interestingly it was recently isolated from humans having different infections, including respiratory ones (Fry et al., 2005; Stark et al., 2007; Spilker et al., 2008; A. Le Coustumier et al., unpublished). Even though other Bordetella species share many aspects of pathogenicity, they have distinct host ranges and cause different pathologies, which could be related to their surface components (Mattoo and Cherry, 2005). Experimentation in the domain is expected to lead to important information on infectious processes specific to bacterial niches.

B. pertussis has long been recognised as the pathogen that infects only humans and is the causative agent of whooping cough in infants and in adults (Mattoo and Cherry, 2005). Some B. parapertussis strains are adapted to the human host and cause whooping cough, while others are adapted to the ovine host causing chronic pneumonia (Mattoo and Cherry, 2005). B. bronchiseptica, on the other hand colonizes the respiratory tract of a large number of animal hosts and can be symptomatic (Kennel cough, atrophic rhinitis, bronchopneumonia etc.) or asymptomatic and chronic. Although B. bronchiseptica most often infects animals, it can be collected from humans, most of the time contaminated by infected animals. B. pertussis and B. parapertussis are assumed to have evolved independently from a B. bronchiseptica-like ancestor (Diavatopoulos et al., 2005). The core oligosaccharides from different isolates of B. pertussis and B. bronchiseptica display an almost identical structure of a branched nonasaccharide with several free amino and carboxyl groups (Caroff et al., 2001). This core may carry an antigenic distal trisaccharide, (Caroff et al., 2000) which was shown to be important for bacterial resistance to pulmonary surfactant proteins (Schaeffer et al., 2004). The latter was not found in B. parapertussis "free" core (Caroff et al., 2001). However, nothing is known about LPS polymorphism within the species. The genetic determinants of host specificity and virulence across the Bordetella genus still remain to be explored. Genome reduction, accompanied by proliferation of Insertion Sequence (IS) elements, has played a fundamental role in the evolution of the pathogenic Bordetella species (Parkhill et al., 2003; Cummings et al., 2004; Bouchez et al., 2009).

Bordetella lipid A structures have a common bisphosphorylated β-1,6 glucosamine disaccharide backbone with two amide-linked 3-OH C14 substituents. The nature and distribution of ester-linked fatty acids have so far proved to be species- or strain-specific and highly variable (Caroff et al., 1994; Zarrouk et al., 1997). One of the unusual features of Bordetella lipid A compared to those of most other lipids A is the absence of symmetry at the C-3 and C-3' positions. The Bordetella genus shows a remarkable ability to modify lipid A structures by late-steps in their biosynthesis. It has long been known that lipid A structures differ not only between different genera but also often between species of a genus, as well as among different isolates of a species (Aussel et al., 2000; Therisod et al., 2001). Moreover, a given isolate may express different LPS molecular species simultaneously with varying abundance.

Zarrouk et al., (1997) earlier demonstrated that the lipid A structures of three B. bronchiseptica isolates were highly heterogeneous and different; the differences were found mostly in the nature of the fatty acids. The ability to modify the structure of its lipid A components may allow Bordetella to escape or alter TLR4-dependent host defense mechanisms as well as to decrease the susceptibility to penetration of the bacterial cell wall by antibiotics (Dixon and Darveau, 2005; Miller et al., 2005). It may contribute to adaptation of B. bronchiseptica, a spontaneously mutating animal pathogen, to diverse niches or hosts, leading to the relatively recent appearance of the human pathogens (Parkhill et al., 2003).

B. bronchiseptica, once thought to infect only animals is currently acquiring increased importance as a human pathogen; it colonizes the ciliated epithelium of the respiratory tract of the host and establishes chronic infections (Woolfrey and Moody, 1991; Gueirard et al., 1995). The development of such chronic infections may partially depend on the ability of bacteria to adapt phenotypicaly to variable stimuli.

In order to better characterize structure to virulence relationships, especially in the context of new structural elements found in Bordetella lipids A (Marr et al., 2008) we studied the lipid A structures of B. bronchiseptica isolates from rabbit and human origin as a function of their virulence phase. To our knowledge, this is the first structural comparison of the kind concerning Bordetella clinical isolates.

Results and discussion

Total fatty acid composition

Total fatty acid analysis revealed the presence of 3-hydroxytetradecanoic acid (3-OH C14), tetradecanoic acid (C14), 2-hydroxydodecanoic acid (2-OH C12) and dodecanoic acid (C12) in lipids A extracted from all isolates in their non-virulent (H-) and virulent (H+) phases. In addition to these fatty acids, hexadecanoic acid (C16) was only detected in lipids A extracted from all virulent (H+) isolates.

MALDI-MS analysis of untreated lipids A

The negative-ion matrix-assisted laser-desorption/ionization-mass spectra (MALDI-MS) of 9.73 H- and 9.73 H+ B. bronchiseptica rabbit isolates are shown in Figures 1A and B, while the spectra corresponding to DANG H- and DANG H+ human isolates are shown in figures 1C and D, respectively. The MALDI-MS spectra of ALI H- and ALI H+ human isolates are given as supplementary Figures S1 A and B, respectively. The lipid A preparations of all the isolates examined except for ALI H- (Fig. S1A) were highly heterogeneous. Comparison of lipid A spectra from the isolates in their virulent and non-virulent phases demonstrated that all virulent phase (H+) isolates express noticeably more heterogeneous lipids A than their respective non-virulent (H-) counterparts.

Major peaks corresponding to molecular-ions [M-H]- common to all isolates were observed at m/z 1345, 1361, 1571, and 1587. In accordance with the gas chromatography (GC) data, these peaks were attributed to tetra- and penta-acyl molecular species in which the di-phosphorylated di-glucosamine backbone was substituted with two 3-OH C14, one C14, and one C12 or 2-OH C12 fatty acids (m/z 1345, 1361) with additional 3-OH C14 in the penta-acyl molecules (m/z 1571 and 1587).

In all spectra, except for that of ALI H- isolate, a series of additional peaks was observed at m/z 161 u or 322 u higher than the mentioned common peaks (m/z 1522, 1732, 1748, 1893, and 1909). These peaks were attributed to lipid A molecular species in which one or both phosphate groups were substituted with GlcN. This Bvg-regulated lipid A modification was recently unveiled in our group in B. bronchiseptica strain 4650 and B. pertussis Tahoma I strain (Marr et al., 2008). We and others have also shown that this modification affects LPS biological activity by strongly increasing proinflammatory cytokine production in cells expressing the human but not the murine TLR4-CD14-MD2 complex (Geurtsen et al., 2009; Marr et al., 2010). We demonstrate here that the presence of these substituents is not correlated with bacterial virulence as it was present in both virulent and non-virulent isolates and was also not related to the human or rabbit origin of the isolate.

On the other hand, the fatty acid compositions revealed differences between LPS of isolates in virulent and non-virulent phases. In contrast to non-virulent phase, all the virulent ones express an LPS with two series of molecular species containing palmitate (C16) in their structures. The first series is represented by molecular-ion peaks at m/z 1809, 1825, 1971, 1987, 2132, and 2148, i.e., 238 u higher to the corresponding penta-acyl molecular species with and without additional GlcNs. The second series of palmitate-containing molecular species was represented by peaks at m/z 1401, 1627, 1788, 1866, 1949, 2027, and 2188. All these peaks were shifted by addition of 56 u with respect to the peaks at m/z 1345, 1571, 1732, 1809, 1893, 1971, and 2132, corresponding to molecular species containing a non-hydroxylated C12 fatty acid. Taking into account the GC analyses and the likelihood that the 56 u shift corresponds to four CH2 units, we concluded that a C16 fatty acid replaced C12 or 2-OH C12. The presence of a C16 fatty acid in the lipid A structures of this genus has been considered to be effected by a late biosynthetic step, using fatty acids taken from the outer membrane phospholipids (Bishop et al., 2000). Bvg regulated palmitoylation at the secondary C-3' position of B. bronchiseptica lipid A by PagP enzyme was described earlier (Preston et al., 2003). This modification is required for persistence of B. bronchiseptica within the mouse respiratory tract and for resistance to antibody-mediated complement lysis during respiratory tract infection (Pilione et al., 2004). PagP mutants in Salmonella typhimurium and Legionella pneumophila increase sensitivity to cationic antimicrobial peptide-mediated killing, suggesting that, in these species, lipid A palmitoylation may increase resistance to these peptides (Guo et al., 1998; Robey et al., 2001).

In all the mass spectra, the major peaks (representing major molecular species) have smaller peaks, on both sides, corresponding to molecular species differing by 2 or 4 CH2 units (plus or minus 28 or 56 u) or by an oxygen atom (plus or minus 16 u) showing the capacity of the bacteria to vary their lipid A structures, possibly by the Bordetellae relaxed enzyme specificity already described (Sweet et al., 2002) or by the late biosynthesis steps. These spectra also showed multiple peaks resulting from a small loss of phosphate groups (80 u) labeled by -P in the figures. This was due to some degree of dephosphorylation during the hydrolysis step.

Lipids A structures

Structures of all the lipid A molecular species mentioned above were established by fragmentation analysis of native samples in MALDI-MS (positive ion-mode) and by sequential fatty-acid liberation using alkaline treatments followed by MALDI-MS analysis (negative-ion mode) of treated samples and taking into account GC data of the released fatty acids.

The positive-ion MALDI mass spectra (not shown) of all samples had a prominent fragment peak at m/z 904 and a minor, but unambiguous one at m/z 678. The lipid A samples extracted from virulent strains had an additional prominent fragment peak at m/z 1142. In accordance with a well established fragmentation pattern (Karibian et al., 1999) these peaks were attributed to lipid A fragments composed by the distal glucosamine (GlcN II) substituted with one phosphate group and different numbers of fatty acids: one 3-OH C14 and one C14 for tetraacyl molecular species, two 3-OH C14 and one C14 for pentaacyl ones and two 3-OH C14, one C14 and one C16 for hexaacyl ones, the latter being characteristic of virulent strains. Fragments containing additional GlcN substituting the phosphate group were not observed because of the loss of this element during the fragmentation process. Therefore, it was concluded that in all molecular species the reducing glucosamine (GlcN I) was substituted with two fatty acids: one 3-OH C14 and one C12 or 2-OH C12 or C16.

The exact position of each fatty acid was determined by its liberation pattern upon alkaline treatments. The transformation of native lipid A negative-ion MALDI mass spectra under different conditions of alkaline treatment is summarized in Table 1. Under the conditions of the method (Tirsoaga et al., 2007) all substituents at C-3 and C-3' positions are liberated by ammonium hydroxide treatment. The B. pertussis lipid A deacylation pattern was used as a reference (not shown). In this case, there was total release of the fatty acid at position C-3 (C10-OH) in the first 15 min of treatment. The first 15 min of treatment applied to B. bronchiseptica lipids A did not change the general aspect of the spectrum, which could be explained by an unsubstituted C-3 position. After 2 h of treatment, C14-OH and C14-O-C16 were almost completely released, which clearly indicated that these fatty acids were at the C-3'position. Penta- (m/z: 1587, 1748, 1909) and hexa-acylated (m/z: 1825, 1986, 2148) molecular species, respectively, released 3-OH C14 and C14-O-C16 fatty acids to generate tetra-acylated species at m/z 1361, 1522, and 1683 substituted with two 3-OH C14, one C14 and one 2-OH C12. Similarly, the second series of penta- ( m/z: 1627, 1788, 1949) and hexa-acylated (m/z: 1865, 2026, 12188) species released 3-OH C14 and C14-O-C16, respectively, giving rise to tetra-acylated species with m/z 1401, 1562, and 1723 substituted with two 3-OH C14, one C14 and one C16. After 5 h of ammonium hydroxide treatment, these peaks were still present. After treatment with methylamine, aimed at liberating secondary ester-linked fatty acids, all that remained were three peaks at m/z 952, 1113, and 1274 corresponding to diacyl molecules with and without additional GlcNs and substituted at the C-2 and C-2' positions via amide bonds with 3-OH C14 fatty acids. Release of C14, 2-OH C12 and C16 fatty acids under these conditions showed that they substituted acyloxyacyl at C-2 or C-2' positions. According to the fragmentation data, C-2' secondary position was always substituted by C14 fatty acid. As for the C-2 secondary position it was substituted by C12 or 2-OH C12 or by C16.

Taken together, these data establish the lipid A structures presented in Figure 2. All the isolates studied shared a common structural basis consisting of a bis-phosphorylated di-glucosamine backbone substituted at the C-2 and C-2' positions with 3-OH C14 and 3-C14-O-C14 fatty acids, respectively. In the penta- and hexa-acyl molecular species, the C-3' position is substituted with a 3-OH C14 fatty acid. The C-3 position is always free because of the activity of the PagL enzyme (Geurtsen et al., 2005). This feature is common to the Pseudomonas aeruginosa (Ernst et al., 2006) and B. parapertussis lipid A structures (El Hamidi et al., 2009) but not to that of B. pertussis, in which PagL is not active and this position was always found to be substituted by 3-OH C10 fatty acid in the strains analyzed (Geurtsen et al., 2005). In lipid A structures common to virulent and non-virulent phases of isolates, the secondary C-2 position is substituted with either a C12 or a 2-OH C12 fatty acid. This substitution, which is most probably due to the action of a late acyltransferase, LpxL ortholog, adds one more structural similarity between lipids A of B. bronchiseptica and P. aeruginosa (Hancock et al., 1970). LpxL homologues were recently discovered in the genome of B. pertussis, one of which (LpxL1) mediates the addition of C12 to the secondary C-2 position. Increased LpxL1 expression in B. pertussis was shown to lead to more endotoxic lipid A structures and favors infection of human macrophages (Geurtsen et al., 2007). Hydroxylation of this secondary fatty acid is known to be due to the hydroxylase LpxO (Gibbons et al., 2000).

Lipids A molecules specific to virulent phase of isolates are thus characterized by the presence of one or two C16 fatty acid in their structures. We demonstrate that this fatty acid can occupy two different positions: the secondary C-3' position acylated by the action of the PagP enzyme and earlier associated with B. bronchiseptica infection in a mouse model (Preston et al., 2003). The other position is in the secondary C-2 substitution. The latter is demonstrated for the first time in B. bronchiseptica. It was earlier found in B. parapertussis (El Hamidi et al., 2009). The sequence of the biosynthetic steps remains to be established. However it is interesting to note that in Enterobacteria such as Escherichia coli and S. typhimurium, C16 is added at this position by the PagP enzyme (Guo et al., 1998; Bishop et al., 2000).

Finally, these data confirmed that lipids A of all the B. bronchiseptica isolates tested (except ALI H-) had one or two extra GlcNs, a characteristic modification due to the activity of the ArnT ortholog glycosyl transferase in Bordetellae (Marr et al., 2008). The importance of phosphate groups in LPS biological activity interactions has been well documented in this group (Caroff et al., 1986; Lebbar et al., 1986; Cavaillon et al., 1989; Haeffner-Cavaillon et al., 1989). Lipid A modification is not unusual in Bordetella: substantial transcriptional and genetic diversity among different isolates of the same Bordetella species have been demonstrated recently (Cummings et al., 2004; Diavatopoulos et al., 2005; Cummings et al., 2006). Little is known on how this affects virulence, but neutralization of phosphate groups is known to strengthen bacterial resistance to antibacterial peptides (Gunn et al., 1998). In this work, "extra" GlcN being present in lipid A structures from isolates from both virulent and non-virulent phases, no correlation was established with the virulence criteria.

We demonstrate here that one important modification of lipid A associated with virulence phase was palmitoylation. We earlier hypothesized (Preston et al., 2003) that such a structural modification is harbored by LPS of isolates in their virulent phase within the host and not by the isolation in their non-virulent phase adopted in the environment. In this report, lipid A palmitoylation would occur in clinical human and rabbit isolates. The other difference is the degree of palmitoylation with the presence at C-2 of a second palmitate to the one described in the mouse model. The molecular species presented in Fig. 2B is identical to the structure previously reported for one B. parapertussis lipid A isolate (El Hamidi et al., 2009). Park et al., (2009) recently presented an interesting molecular model illustrating the role of additional fatty acids on a lipid A structure for modifying the presentation of the LPS-MD2 complex to TLR4. The reason why some Bordetella species, but not others can regulate their LPS structure is at present unknown, but may be linked, in B. bronchiseptica, to its capacity to persist several years inside its human or animal host (Woolfrey and Moody, 1991; Gueirard et al., 1995). The molecular heterogeneity of B. bronchiseptica lipids A may explain how these pathogens adapt to new host ranges and to changes in host dynamics by manipulating the host immunity-pathogen interactions. Again, lipid A structure can be considered as a good tool to illustrate evolution of the bacterial Bordetella species, the human pathogen B. pertussis being assumed to be derived from a common ancestor of B. bronchiseptica.

Experimental procedures

Bacterial isolates and growth conditions

Three B. bronchiseptica isolates were used in this study: 9.73 H+ from rabbit origin (Gueirard and Guiso, 1993) and DANG H+ and ALI H+ from humans (Le Blay et al., 1997). From these three hemolytic isolates, spontaneous non-hemolytic (9.73 H-, DANG H- and ALI H-) variants were obtained. These variants were shown by western blotting and specific antibodies not to express adenylate cyclase-hemolysin (AC-Hly), filamentous hemagglutinin (FHA), and pertactin (PRN) according to Khelef et al. (1993). These variants are considered to be in non-virulent phases since they do not express the virulence factors and are not lethal in the murine model (Gueirard et al., 1995) while the parental isolates are considered to be virulent. Bacteria were grown as described in Gueirard and Guiso (1993).

LPS and lipid A preparation

LPSs from B. bronchiseptica isolates were extracted by the enzyme-phenol-water method (Johnson and Perry, 1976) and sedimentation by ultracentrifugation (105,000 g, 4 °C, 12 h).

Lipid A was isolated from LPS using mild, detergent-facilitated hydrolysis as described previously (Caroff et al., 1988). Alternatively, when samples of less than 1mg were used, lipid A was obtained after hydrolysis of LPS in 1 % acetic acid at 100 °C for 1 h at a concentration of 10 mg/ml. With both hydrolysis, applied to 1 mg LPS samples, lipid A was recovered from the lyophilized residue by three extractions with 100 ml of a choloroform-methanol-water mixture (3:1.5:0.25, v:v:v) and analyzed by MALDI-MS.

Sequential liberation of ester-linked fatty acids by mild alkali treatment

Sequential liberation of ester-linked fatty acids by mild alkali treatment was used to establish the lipid A acylation patterns (Tirsoaga et al., 2007). Briefly, for the first-step liberation of primary ester-linked fatty acids, lipid A (200 µg) was suspended at 1 mg/ml in 28 % ammonium hydroxide solution in 1.5 ml Eppendorf tube and stirred in a thermomixer (Eppendorf, Germany) for 5 h at 50 °C and 1000 rpm. Kinetics of fatty acid liberation was followed after 15 min, 30 min, 1 h, 2 h and 5 h. To liberate the secondary ester-linked fatty acids, 50 µg of lipid A was suspended in 50 µl of 41 % methylamine solution and stirred for 5 h at 37 °C. At this stage all sample suspensions were dried under a stream of nitrogen, the residues taken up in a mixture of chloroform-methanol-water (3:1.5:0.25, v:v:v) and analysed by MALDI-MS. B. pertussis lipid A fatty acid liberation kinetics (15 min, 30 min, 1 h, 2 h, 5 h) was used as a reference.


Matrix-assisted laser-desorption/ionization mass spectra were obtained in the linear mode with delayed extraction using a Perseptive Voyager STR (PE Biosystem, France) time-of-flight mass spectrometer (I.B.B.M.C., University of Paris Sud, France). A suspension of native or treated lipid A in chloroform-methanol-water (3:1.5:0.25; v:v:v) (1 mg/ml) was desalted with a few grains of Dowex 50W-X8 (H+), one µl was deposited on the target, mixed with one µl of the matrix solution and dried. Matrix: 2, 5-dihydroxybenzoic acid (DHB) was dissolved at 10 µg/µl in the same solvent or in 0.1 M citric acid in chloroform-methanol-water (3:1.5:0.25; v:v:v) (Therisod et al., 2001). Analyte ions were desorbed with pulses of a nitrogen laser (337 nm). Spectra were recorded in the negative- and positive-ion modes with ion acceleration set at 20 kV.

Fatty acid analysis

To determine the overall fatty acids content in lipids A, all fatty acids were released after hydrolysis of the LPS or lipids A with 4 M HCl for 2 h at 100 °C, extracted with ethyl acetate and methylated with a mixture of anhydrous methanol and acetyl chloride (10:1.5; v:v) (Wollenweber and Rietschel, 1990). Fatty acids were identified by gas chromatography. An HP 5890 gas chromatograph was equipped with an HP5 capillary column (30 m x 0.32 mm) and a temperature gradient from 150 to 300 °C, 2 °C/min was used. Fatty acids were characterized by comparison of their retention times with reference samples (methyl esters of 2-OH and 3-OH C10 to C16 fatty acids as well as of non-hydroxylated C10, C12, C14 and C16 fatty acids). C20 was used as an internal standard.