Classification of Proteus penneri lipopolysaccharides into core region serotypes

The frequency of P. penneri isolation from hospital patients, mostly from urine and wounds, keeps on growing, and numerous isolates are multi-drug resistant. P. penneri rods produce lipopolysaccharide (LPS), which may lead to the septic shock. Until now, O-specific polysaccharide has been the best structurally and serologically characterized region of P. penneri LPS. It is worth having an insight into the serological specificity of both poly- and oligosaccharide parts of P. penneri LPS. The P. penneri core region is less structurally diverse than OPS, but still, among other enterobacterial LPS core regions, it is characterized by structural variability. In the present study, the serological reactivity of 25 P. penneri LPS core regions was analyzed by ELISA, passive immunohemolysis and Western blot technique using five polyclonal P. penneri antisera after or without their adsorption with the respective LPSs. The results allowed the assignment of the tested strains to five new core serotypes, which together with published serological studies led to the creation of the first serotyping scheme based on LPS core reactivities of 35 P. penneri and three P. mirabilis strains. Together with the O types scheme, it will facilitate assigning Proteus LPSs of clinical isolates into appropriate O and R serotypes.

fluid, sputum and the center of struvite bladder stone [2-4, 6, 7]. P. penneri produce many virulence factors which enable them to cause infections, e.g., urease, fimbriae and hemagglutinins, hemolysins, metalloproteases, flagella, siderophores and lipopolysaccharide (LPS) [2,4]. LPS consists of three structurally different regions: lipid A (defined structurally only for one P. mirabilis mutant), core oligosaccharide (OS) and O-specific polysaccharide (OPS) [4,8]. Until now, OPS has been the best structurally and serologically characterized region of P. penneri LPS, which also defines the serospecificity of smooth bacterial cells. Twenty-six different OPS structures have been identified for P. penneri strains so far, among which seven are common also to the other representatives of the genus [4,9,10]. The P. penneri core region is less structurally diverse than OPS but in contrast to other enterobacterial LPS core regions characterized by lager structural variability. Up to date, 12 different structures of the outer core region, accounting for the structural diversity of the P. penneri LPS core regions, were identified ( Fig. 1) [4,11]. The majority of tested P. penneri strains presented one major glycoform of the inner core region [11,12] (Fig. 1). There are only two strains, P. penneri 12 and 42, which present glycoforms of the inner core region not identified in any other Proteus spp. LPSs [4,11,12]. Moreover, the heterogeneity of this LPS part may appear also within one strain, e.g., P. penneri 13 forms ten variants of its core-lipid A backbone [4]. The P. penneri classification scheme is based on the OPSs serospecificity. So far, P. penneri isolates have been classified into 17 Proteus O serogroups, among which 13 consist of these species representatives only [4,9,10,13]. To have an insight into the serological specificity of both polysaccharide and oligosaccharide parts of P. penneri LPS, it is worth creating an additional scheme classifying P. penneri LPSs into serotypes of their core regions. A core types classification scheme which together with the O-types scheme may serve as a diagnostic tool facilitating the assignment of Proteus LPSs of clinical isolates into appropriate O and R serotypes. In the current work, the results of serological studies prove the existence of another five serotypes of P. penneri core regions, which is evidence of further structural variations within this part of Proteus spp. LPS.
The P. penneri 18 LPS was isolated by the phenol-water procedure according to the Westphal and Jann method (1965) and purified with aqueous 50 % trichloroacetic acid [14].

The sera
Rabbit polyclonal sera against the whole cells of P. penneri 17, 28, 60, 103 and 124 came from the collection of the Department of General Microbiology.

Serological assays
LPSs samples were checked with the appropriate antiserum in the enzyme-linked immunosorbent assay (ELISA), in Western blot after sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and/or in PIH according to the previously described procedures [17,18]. For PIH, sheep red blood cells (SRBCs) were sensitized with alkalitreated LPSs (64-200 μg/0.2 ml of SRBCs); 50 ng of LPS per well was used for coating microtiter plates in ELISA. The highest dilution of antiserum giving optical density 405 = 0.2 was assumed as the antibodies titer.

Adsorption procedures
• By alkali-treated LPSs Single antiserum diluted 1:50 with a veronal buffer (pH 7.3) was incubated for 30 min on ice with SRBCs (0.2 ml) sensitized with 200 μg of appropriate alkali-treated LPS from one Proteus spp. strain. The antiserum titer was determined by PIH as the last antiserum dilution resulting in 50 % hemolysis [16].
• By the LPS on bacterial cells This adsorption procedure was performed in the case of i.a. the P. penneri 17 antiserum due to the weak activity of cross-reacting alkali-treated LPSs in PIH compared to activity of the native LPSs in ELISA.
A wet mass of bacterial cells, after being washed in phosphate-buffered saline (PBS), was suspended in serum diluted 1:100 in PBS, incubated for 30 min on ice and removed from the serum by centrifugation.
Each serum was adsorbed three-four times to make sure that all antibodies that were able to bind to LPS molecules were removed from the serum.

Results
In previous serological studies, sera specific to appropriate P. penneri strains were tested with a set of 40 P. penneri LPSs and the core region serospecificity for the majority of those antigens was determined [16,17,19,20]. In the present work, the reactivities of the core regions of P. penneri LPSs 2,11,12,13,16,17,18,19,24, [11]), 28, 60 and 124 were analyzed by ELISA and Western blot and classified into core serotypes. Each serum was adsorbed with a single cross-reacting antigen and tested again in PIH or in ELISA (P. penneri 17 antiserum) with all LPSs reacting with the serum to exclude further serological differences within the groups.

P. penneri 28 antiserum
P. penneri 28 antiserum has been obtained by immunizing with whole bacterial cells. O-polysaccharide-specific immunoglobulins were eliminated from the serum by its adsorption with an alkali-treated LPS of P. vulgaris 55/57 (O31a,b) containing the OPS structurally identical to P. penneri 28 OPS and a different core region serotype [21]. A lack of reaction with P. vulgaris 55/57 LPS indicated that all antibodies specific to the OPS were completely removed from the serum (Fig. 2a). This antiserum will be referred to as anti-core serum. Three LPSs, P. penneri 16, 18 and 31, reacted with this serum in ELISA of which P. penneri 31 was distinguished by the lowest reactivity titer (Table 1). In Western blot, P. penneri 28 anti-core serum bound to lowmolecular-mass LPS species (consisting of the core-lipid A moieties) of all the LPSs used, among which P. penneri 18 LPS showed the strongest and the most distinguishing reactions (Fig. 2a). The ladder-like banding pattern, which can be noticed in Fig. 2a, also corresponds to the low-molecular-mass LPS species of P. penneri 18 LPS. This observation was confirmed by the Western blot results obtained for P. penneri 18 LPS and antiserum against the P. penneri 28 strain after its adsorption with P. penneri 18 LPS molecules. The reactions previously observed in Western blot with unadsorbed anti-core serum (Fig. 2a), were abolished. Moreover, P. penneri 28 anti-core serum adsorbed with P. penneri 16, 18 and 31 LPSs did not react in PIH with any of the antigens used (data not showed).

P. penneri 17 antiserum
Although P. penneri 17 antiserum is specific to the whole bacterial strains, P. penneri 17 LPS did not show in Western blot (Fig. 2b,c) the reaction typical for the highmolecular-mass LPS species containing O-polysaccharide, which suggests that core-specific antibodies dominate in the serum. In ELISA, the P. penneri 17 antiserum strongly cross-reacted with P. penneri 2, 11, 19 and 107 LPSs (Table 1), which is in accordance with structural studies which showed that all LPSs present one structural type of the core region ( Fig. 1) [11]. Further, six LPSs from P. penneri 24, 35, 36, 38, 100, and 114 exhibited in ELISA strong cross-reactions, whereas P. penneri 115 LPS showed only weak reactivity (Table 1). In Western blot, the reactions of P. penneri 17 antiserum, concerning the core-lipid A molecules of all used P. penneri LPSs, were similar to the reaction of homologous LPS (Fig. 2b, c). The adsorption of the P. penneri 17 antiserum by each tested antigen completely abolished its reactivity with all the LPSs used (data not showed).

P. penneri 60 antiserum
In ELISA, anti-P. penneri 60 serum cross-reacted with one LPS, P. penneri 63, but more weakly than homologous LPS (Table 1). In Western blot, P. penneri 60 recognized only fast migrating bands of P. penneri 63 LPS (Fig. 2d), which explains the weak reaction observed for this system in ELISA (Table 1). After the adsorption of P. penneri 60 antiserum by P. penneri 63 LPS, the titer of its reactivity with homologous LPS was slightly lower compared to the titer of non-adsorbed serum (Fig. 3a). In order to find out which part of LPS was recognized by the antibodies that remained in the adsorbed serum, the serum was checked with both tested LPSs in Western blotting (Fig. 3a). As expected, the adsorption procedure removed the anti-core-specific antibodies from the serum, which indicates that P. penneri 60 and 63 LPSs present one serotype of the core regions.

P. penneri 103 antiserum
P. penneri 103 antiserum cross-reacted in ELISA only with one LPS and P. penneri 75 to the titer equal to that of homologous LPS (Table 1). In Western blot, similar strong Fig. 2 Western blot data of Proteus spp. LPSs with antisera against: a P. penneri 28 (anticore serum), b, c P. penneri 17, d P. penneri 60, e P. penneri 103 and f P. penneri 124 reactions were observed for the serum with both high-and low-molecular-mass LPS species of the homologous strain and of the cross-reacting P. penneri 75 strain (Fig. 2e). After the adsorption of P. penneri 103 antiserum by P. penneri 75 LPS, only a small fraction of anti-LPS antibodies, reacting to the titer 1:400 in PIH and recognizing slowmigrating bands of P. penneri 103 LPS in Western blotting, remained in the serum (Fig. 3b).

P. penneri 124 antiserum
P. penneri 124 antiserum is specific to the clinical rough strain, i.e., it contains the core-specific antibodies only. It was selected for the studies to check whether P. penneri 124 LPS presents the same core region serotype as the type strain of the species, P. penneri 12 (American Type Culture Collection 33519) with the known core region structure (Fig. 4) [11]. The previous serological studies conducted with the use of the P. penneri 13 antiserum showed its strong cross-reactivity with P. penneri 12, 124, 26 and 112 LPSs, reacting to the same reactivity titer (1:16.000) as P. penneri 13 LPS [16]. To confirm the previously obtained results, the P. penneri 13, 26 and 112 LPSs were additionally selected to be included in the present work. As was expected, the P. penneri 124 antiserum reacted in ELISA identically with P. penneri 12, 124, 13 and 112 LPSs, but the reaction with P. penneri 26 LPS was characterized by the lowest intensity (Table 1). In Western blot, the P. penneri 124 antiserum recognized the fast migrating bands of all tested LPSs (Fig. 2f). In contrast to ELISA results, P. penneri 112 LPS reacted more weakly than homologous LPS. The adsorption procedure of the tested serum completely abolished all reactions previously observed for nonadsorbed serum (data not showed).

Discussion
The serological results of the present study have allowed classifying 22 LPSs of P. penneri 2, 11, 12, 16, 17, 18, 19, 24, 28, 31, 35, 36, 38, 60, 63, 75, 100, 103, 107, 114, 115 and 124 into one of the five new core serotypes (Table 2, below the middle line). Four of the antigen groups contain LPSs with determined core region structures [11] (in Table 2 marked with*), and one contains only LPSs (P. penneri 60 and 63) with structurally unknown core regions. Table 2 includes all serotypes of LPS core regions formed on the basis of the results (Table 1; Figs. 2, 3) and previous serological studies [16,17,19,20]. The studies were possible to perform owing to a unique set of anti-core sera obtained by: (1) the adsorption of serum against the bacterial whole cells by the LPS possessing the same O-polysaccharide as the strain homologous to the serum and different core region serotypes, (2) the immunization of rabbit with the conjugate of core oligosaccharide with diphtheria toxoid and (3) the immunization of rabbit with the whole cells of rough strains or of smooth strains but having the majority of LPS molecules unsubstituted with O-polysaccharide chains.
Assigning the P. penneri LPSs to the appropriate core serotype has not always been easy to perform. In some cases, one LPS cross-reacted with the appropriate serum similarly to homologous LPS in one method and weaker in another technique, e.g., P. penneri 112 LPS reacted with P. penneri 124 antiserum to the same titer as homologous LPS (Table 1) and in Western blot more weakly than homologous LPS (Fig. 2f). The differences in the antisera reactivity titers within one group of LPSs may also result from the Although P. penneri 28 LPS possesses the core region with an unknown structure, it was selected as a representative of serotype no. 7 (Table 2) because it was possible to obtain the serum specific to its core region. The reactions observed for this serum in Western blot (Fig. 2a) with P. penneri 28, 16, 18 and 31 LPSs were the most diverse in their intensity. The LPS exhibiting the most distinguished reaction was P. penneri 18 LPS, for which the ladder-like banding pattern was observed at the level corresponding to the higher-molecular LPS species. Smearing which appeared for P. penneri 28 and 16 LPSs at the level, typical for higher-molecular mass species, concerned unspecific reactions since both LPSs have no common fragments in the OPS parts [9]. What is more, P. vulgaris 55/57 OPS which is known to be structurally identical to the O-antigen of the P. penneri 28 LPS has not reacted with the tested serum (Fig. 2a). The method, which occurred to be very useful for proving that all tested antigens present a common core region serotype, was the serum adsorption. It allowed abolishing the reactions observed previously with the unadsorbed serum (Table 1; Fig. 2a). The procedure of antiserum adsorption, by LPS molecules coated on SRBC as well as on the bacterial cells, is a proven method for assessing serological similarities between antigens. Even though slight differences in reactivities have been observed for a group of LPSs with unadsorbed serum, the adsorption procedure usually appears to be helpful in explaining disputable results [16,17].
The serum adsorption procedure also enabled classifying 11 LPS to the core serotype presented by P. penneri 17 LPS. It is the most numerous group among the analyzed LPSs, having one serotype of the core region, including five LPSs of the determined core region structures (P. penneri 2, 11, 17, 19 and 107). The structural studies showed that the core regions of the mentioned LPSs possess a unique type of linkage [(1S)-GaloNAc-(1 → 4,6)-α-GalN] occurring also in OSs of a few Proteus spp. LPSs, which had not previously been reported in natural glycopolymers [11]. This linkage probably occurs in OSs of the remaining P. penneri LPS presenting the same serotype as the P. penneri 17 core region.
LPSs P. penneri 60 and 63 present a core region serotype new for P. penneri strains tested so far [16,17,19,20]. P. penneri 60 antiserum cross-reacted only with the low-molecular-mass species of P. penneri 63 (Fig. 2d). This result also stays in agreement with the structural data showing that in OPSs of both LPSs there is no common fragment except for the N-acetyl-fucosamine residue [-3)-α-l-FucpNAc-(1-] [9,22]. It should be remembered that, in contrast to the anti-core sera, O-specific immunoglobulins predominate in the sera specific to the whole bacterial cells, including P. penneri 60 antiserum. It was reflected in ELISA by the higher titer of the serum with homologous LPS than that observed for P. penneri 63 LPS (Table 1). P. penneri 63 LPS was assigned to the core region serotype presented by P. penneri 60 LPS on the grounds of the Western blot results after the serum adsorption by P. penneri 63 LPS molecules (Fig. 3a). Abolishing the bands corresponding to the reactions with low-molecular-mass species of Table 2 First typing scheme based on the core region serological specificity of 35 P. penneri and three P. mirabilis LPSs LPS of the structurally defined core regions The first representatives of each core region serotype are homologous to the respective antiserum used in the studies The core serotype presented by the LPSs Other representatives of core region serotypes both tested LPSs and leaving the ladder-like banding pattern typical for the serum reaction with high-molecularmass species of P. penneri 60 LPS (Fig. 3a, in a frame) clearly confirmed the suggestion that these LPSs present a similar serotype of the core region. P. penneri 103 LPS was found to present the same core region serotype as P. penneri 75 LPS on the basis of the Western blot results obtained after the adsorption of P. penneri 103 antiserum by P. penneri 75 LPS (Fig. 3b). Only a slight reaction with slow-migrating bands of P. penneri 103 LPS was observed. Comparing the OPSs structures of both LPSs, it can be noticed that they differ only in the lateral substituents of the glucose residue and probably the lateral group (EtnP in P. penneri 103 OPS) was recognized by the immunoglobulins remaining in anti-P. penneri 103 serum after its adsorption by P. penneri 75 LPS (d-Glc residue except EtnP) (Fig. 3b, the reaction in a frame) [23].
The P. penneri 124 strain possesses LPS with a structurally unidentified core region, but its rough form decided about its selection for the studies. All serological data, the previous [16] and present ones (Table 1; Fig. 2f), indicate the P. penneri 124 LPS core region as serologically identical to the OS of P. penneri 12 LPS with a known core region structure (Fig. 4). Analyzing the results of P. penneri 124 antiserum and the previous results obtained for P. penneri 13 antiserum [16] in relation to the core region structures of P. penneri 12, 13 and 26 LPSs (Fig. 4) [11], it was confirmed that the common fragment of the core regions: α-GalN-(1 → 4)-α-GalA may be responsible for the observed cross-reactions. The results obtained for P. penneri 124 antiserum, after its adsorption by the tested P. penneri LPSs (data not showed), differ insignificantly from those obtained for P. penneri 13 antiserum [16]. Adsorption of P. penneri 13 antiserum with P. penneri 12 and 124 LPSs left a small fraction of antibodies (reciprocal titer 1.600) reacting with P. penneri 13, 112 and 26 LPSs, which was not observed in the present studies (all antibodies against P. penneri 12 and 124 LPSs were removed). It had been shown previously that those antibodies probably recognized the α-Hep-(1 → 2)-α-dd-Hep fragment, which is present in the core region of P. penneri 13 (probably of 112) and 26 LPSs but absent from P. penneri 12 LPS (probably from P. penneri 124) (Fig. 4). The weaker reaction of P. penneri 124 antiserum with P. penneri 26 LPS (Table 1; Fig. 2f) can be explained by: (1) the lack in the serum of the antibodies binding to the heptose disaccharide, (2) the additional terminal Glc residue (present only in the OS of P. penneri 26 LPS), absent from the outer core region of the remaining LPSs tested (Fig. 4), which may hamper binding the antibodies to the α-GalN-(1 → 4)-α-GalA fragment. However, this small structural difference in the P. penneri 13 and 26 core regions had not been shown to result in decreasing the titer of P. penneri 13 antiserum reactivity with P. penneri 26 LPS, compared to the titer in the homologous system [16]. Thus, the P. penneri 26 LPS was classified into the group of LPSs (with one core region serotype) represented by P. penneri 13 LPS as its subgroup 4a (Table 2).
Serological classification of Proteus spp. strains is not a typing method commonly used in clinical laboratories. However, the serological classification scheme based on the P. penneri core region provides useful information necessary for assigning the clinical isolates, not only from P. penneri species, to an appropriate core region serotype. Further extension of the scheme with other representatives may form a database, which, together with the scheme of Proteus LPS O-types, will show which core region serotype dominates within an area where the tested isolates come from. This may help in the selection of antigens presenting the most common O and R serotypes and serve as a tool to identify common epitopes among strains which may upon immunisation lead to the formation of cross-reactive and cross-protective antibodies against the core region as shown in one example for E. coli [24].