Phage antibodies for the immunochemical characterization of Herbaspirillum seropedicae Z78 glycopolymers

Abstract

Purpose

Microbial carbohydrate antigens are targets of the immune systems of hosts. In this context, it is of interest to obtain data that will permit judgment of the degree of heterogeneity, chemical makeup, and localization of the antigenic determinants of the Herbaspirillum surface glycopolymers.

Methods

A sheep single-chain antibody-fragment phage library (Griffin.1, UK) was used to obtain miniantibodies to the exopolysaccharides (EPS-I and EPS-II), capsular polysaccharides (CPS-I and CPS-II) and lipopolysaccharide (LPS) of Herbaspirillum seropedicae Z78. To infer about the presence or absence of common antigenic determinants in the cell-surface polysaccharides of H. seropedicae Z78, we ran a comparative immunoassay using rabbit polyclonal and phage recombinant antibodies to the surface glycopolymers of H. seropedicae Z78.

Results

We isolated and purified the exopolysaccharides (EPS-I and EPS-II), capsular polysaccharides (CPS-I and CPS-II), and lipopolysaccharide (LPS) of Herbaspirillum seropedicae Z78. Using rabbit polyclonal antibodies, we found that these cell-surface polysaccharides were of a complex nature. EPS-I, EPS-II, CPS-I, CPS-II, and LPS contained common antigenic determinants. CPS-I, CPS-II, and LPS also contained individual antigenic determinants composed of rhamnose, N-acetyl-d-glucosamine, and N-acetyl-d-galactosamine—sugars responsible for cross-reactions with miniantibodies.

Conclusions

The anti-LPS miniantibodies were more specific for the core region of the LPS, in which rhamnose was the most abundant sugar, than they were specific for its O portion. The miniantibodies we isolated can be useful reagents not only in basic biochemical research but also in clinical diagnostic and therapeutic applications.

Introduction

Herbaspirillum, a member of the Betaproteobacteria, enjoys intense current interest (Bajerski et al. 2013; Lin et al. 2013; Chemaly et al. 2015; Batista et al. 2018; Correa-Galeote et al. 2018). Except several phytopathogenic strains (Valdameri et al. 2017), herbaspirilla can promote plant growth and development (Pedrosa et al. 2001). Herbaspirillum also colonizes human organs and tissues (Baldani et al. 1996; Michael and Oehler 2005; Tan and Oehler 2005) and has been identified in clinical isolates and human secretions (Coenye et al. 2002; Spilker et al. 2008; Ziga et al. 2010; Chen et al. 2011). Although the ability of these bacteria to colonize their hosts has been proven, the data on the mechanisms of such associations are fragmentary.

The mechanisms of host-bacterium associations involve the glycopolymers of the bacterial surface. Characterization of the structure of glycopolymers is necessary for understanding their properties and functions, including those operating in the interaction of bacteria with other organisms and with the surroundings. The principal macromolecules implicated in the recognition of symbiotic partners are exopolysaccharide (EPS), capsular polysaccharide (CPS), and lipopolysaccharide (LPS), which determine the antigenic specificity of gram-negative bacteria (Kato et al. 1980; Konnova et al. 1994; Skvortsov and Ignatov 1998; Whitfield and Roberts 1999; Newman et al. 2000; Yirmiya et al. 2000; Smol'kina et al. 2010). Much information about the structure of these glycopolymers can be acquired not only by the destructive chemical methods but also by immunochemistry, which enables the study of antigenic determinants in vitro, in vivo, and in planta.

Serological and immunochemical investigations have traditionally used polyclonal or monoclonal antibodies raised by animal immunization. If, however, the antigen used is poorly immunogenic or highly toxic, immunization may be difficult to carry out. The problem is solved with antibodies based on variable antigen-binding fragments that are obtained from antibody fragment libraries by phage display (MacCafferty et al. 1990). The most commonly used are Fab and scFv fragment libraries, in which the antigen-binding fragment is present at the surface of bacteriophage M13 as part of its pIII protein (miniantibodies (miniAbs) in phage format) (Asadi-Ghalehni et al. 2015; Petrenko 2018). Phage display technology, proposed by Smith (Smith 1985; MacCafferty et al. 1990; Smith and Petrenko 1997), replaces all work stages with simple manipulations with DNA and bacteria, yielding stable antibody-producing clones within weeks rather than months and decreasing the associated costs.

Here, we use a sheep single-chain antibody-fragment library (Griffin.1, UK) to raise miniAbs against the main surface antigens of Herbaspirillum seropedicae Z78, and we report a comparative immunochemical characterization of these antigens.

Materials and methods

Strain and growth conditions

Herbaspirillum seropedicae Z78 (IBPPM 217) was from the IBPPM RAS Collection of Rhizosphere Microorganisms (http://collection.ibppm.ru). Cells were grown in a vitamin-supplemented liquid synthetic nutrient medium (Smol'kina et al. 2012) at 30 ± 1 °C for 24 h (until the end of the exponential growth phase).

Isolation and purification of bacterial polysaccharides

Cells were sedimented by centrifugation at 3000×g for 40 min. Capsular polysaccharides  were removed from the cell surface by resuspending the cells three times in 0.15 M NaCl, agitating the cells on a magnetic stirrer, and resedimenting the cells. The cells were then degreased with petroleum ether, dried with acetone, and finely dispersed. CPS-I, CPS-II, EPS-I and EPS-II were isolated as described by Smol’kina et al. (Smol'kina et al. 2012). All carbohydrate-containing fractions that did not give absorption between 240 and 260 nm were pooled, concentrated, and lyophilized. LPS was extracted from the acetone-treated cells (10 g) with hot 45% aqueous phenol by a modified Westphal procedure (Velichko et al. 2018), purified by ultracentrifugation two times (each at 105000×g for 4 h), and lyophilized with a Benchtop 2K apparatus (VirTis, USA).

Isolation of O polysaccharide and core oligosaccharides

Lyophilized LPS was heated in 1% acetic acid at 100 °C for 4 h (Müller-Seitz et al. 1968). Lipid A was sedimented by centrifuging the reaction mixture at 12000×g for 20 min. The supernatant liquid was dialyzed against distilled H₂O and fractionated by gel filtration. The high molecular weight O polysaccharide (OPS) fraction was separated from the oligosaccharide fraction on a Sephadex G-50 column (Pharmacia, Sweden). The OPS and core oligosaccharide solutions were concentrated, lyophilized, and analyzed.

Preparation of rabbit antibodies

Antibodies to whole cells of H. seropedicae Z78 were kindly provided by the IBPPM RAS Immunochemistry Laboratory. Rabbits were immunized with whole H. seropedicae Z78 cells treated with 2% glutaraldehyde. Rabbits were also immunized with a mixture of LPS and  Freund’s complete adjuvant into popliteal lymph nodes, three times (0.5, 1.0, and 1.5 mg) at 2-week intervals. The antigen concentration was 1 mg ml⁻¹ in all immunizations. The animals were bled 6 days after the last immunization. Antibody titers were determined by agglutination tests. IgG fractions were isolated from antisera by ammonium sulfate precipitation.

Antibody selection from phage-displayed library

For selection of phage-carrying antibodies to the EPS-I, EPS-II, CPS-I, CPS-II, and LPS of H. seropedicae Z78, an enzyme immunoassay plate was used as a solid support for antigen immobilization. The selection procedure was described in detail elsewhere (Dykman et al. 2012). The concentration of the sheep phage recombinant library (Charlton et al. 2000) was 10¹² phagemids ml⁻¹. The phage specificity was determined by dot and enzyme-linked immunosorbent assays (ELISA). The serum titer was measured by conventional ELISA (Beatty et al. 1987). The titer of the resultant phage antibodies was 1:4000.

The phage particle concentration was calculated spectrophotometrically by using a Specord BS-250 UV-vis instrument (Analytic Jena, Germany). The spectrophotometry was done at the Simbioz Center for the Collective Use of Research Equipment in the Field of Physical–Chemical Biology and Nanobiotechnology, Institute of Biochemistry and Physiology of Plants and Microorganisms, Russian Academy of Sciences, Saratov. The calculations were based on the relation (A₂₆₉ − A₃₂₀) × 5 × 10¹⁴/15, where A₃₂₀ is the absorbance of the suspension at 320 nm and A₂₆₉ is the absorbance of the suspension at 269 nm. The virion concentration can be estimated if we know that A₂₆₉ − A₃₂₀ = 30 absorbance units, which correspond to 2 × 10¹⁴ virions ml⁻¹ (Smith & Scot, 1993).

Enzyme-linked immunosorbent assay (ELISA)

For ELISA, 96-well polystyrene plates were used. EPS-I, EPS-II, CPS-I, CPS-II, and LPS with two-fold dilution were immobilized on a plate through simple adsorption. The enzymatic label was horseradish peroxidase conjugated to goat antibodies. The substrate reagent was o-phenylenediamine in the presence of hydrogen peroxide. The absorbance of the samples was measured at 490 nm on a Multiskan Ascent reader (ThermoLabsystems, Finland).

Competitive immunoassay

The inhibition of the immunochemical reactions was assessed by ELISA as described by Kabat and Mayer (1961). Inhibitory monosaccharides were added to miniAbs to 10⁻⁴ M, a concentration purposely chosen in excess of the miniAb concentration. The miniAb concentration was 1.2 × 10¹³ virions ml⁻¹, and the LPS concentration was 0.015 mg ml⁻¹. Solutions of the miniAbs, LPS, OPS, and monosaccharides were made in a buffer of 0.1 M NaCl and 0.01 Tris-HCl (pH 7.2). The miniAbs were mixed with each inhibitor, and the mixture was incubated at 4 °C for 24 h.

Statistical analysis

All experiments were performed in triplicate. Data were analyzed with Excel 2010 software and with standard methods of statistical data processing. Correlation coefficients and unpaired t-tests were used when appropriate. All confidence intervals are for 95% confidence. Differences between means at a confidence level of 5% (P < 0.05) were considered statistically significant. Data are presented as the mean ± the standard deviation (SD).

Results

Preparation of rabbit polyclonal antibodies

Purified preparations of CPS-I, CPS-II, EPS-I, EPS-II and LPS were isolated as described by Smol’kina et al. (2012) and Velichko et. al. (2018).

All preparations were tested by ELISA for their ability to interact with rabbit polyclonal antibodies to glutaraldehyde-treated H. seropedicae cells (Fig. 1). LPS, EPS-II, and CPS-II interacted with the antibodies, with the highest interaction being observed for LPS. Conversely, EPS-I and CPS-I did not interact with the antibodies at all.  Effors to prepare rabbit polyclonal antibodies against purified LPS were unsuccessful.

Fig. 1

a - ELISA of LPS, CPS-II, EPS-II, EPS-I, and CPS-I by using rabbit antibodies to glutaraldehyde-treated H. seropedicae cells. The wells in a polystyrene plate were coated with two-fold dilution of EPS-I, EPS-II, CPS-I, CPS-II, and LPS through simple adsorption; b - ELISA of LPS, CPS-II, EPS-II, EPS-I, and CPS-I by using rabbit antibodies to glutaraldehyde-treated cells with polysaccharide concentration 1.56 μ ml^-1.

Antibody selection from phage library

Because it was difficult to interpret the results of the experiment in Fig. 1, we selected phage recombinant anti-LPS, anti-EPS-I, anti-EPS-II, anti-CPS-I, and anti-CPS-II antibodies (miniAbs_LPS, miniAbs_EPS-I, miniAbs_EPS-II, miniAbs_CPS-I, and miniAbs_CPS-II). There were four rounds of selection of phage antibodies. As suggested by Griep et al. (1998), the antigen concentration in each round was reduced twofold to increase miniAb specificity. The starting antigen concentration was 1 mg ml⁻¹. Measuring the UV absorbance allowed calculation of the numbers of phage in the final dialysates. These were 1.2 × 10¹³, 0.6 × 10¹³, 0.5 × 10¹³, and 2.6 × 10¹³ particles ml⁻¹ for LPS, EPS-I, EPS-II, CPS-I, and CPS-II, respectively.

The increase in miniAb specificity was evaluated by ELISA. Figure 2 shows the results obtained after the first and fourth rounds of selection of miniAbs_CPS-II. After round 1, the miniAbs_CPS-II interacted with all preparations included in the figure, and after round 4, they interacted equally intensely with CPS-I and CPS-II but did not interact with EPS-I. The miniAbs_LPS interacted with LPS, CPS-I, and CPS-II. The activity toward CPS-II was the highest, which suggests that the specific antigenic determinants of CPS-II were better surface-exposed. By contrast, the interaction of the miniAbs_LPS with EPS-II and EPS-I was very weak, almost absent. Like the miniAbs_LPS, the miniAbs_CPS-II barely interacted with EPS-I and EPS-II, but the reaction with LPS was weaker and that with CPS-I was stronger than was the reaction with CPS-II (Table 1). The miniAbs_EPS-I and miniAbs_EPS-II interacted with all antigens used. The reactions with EPS-I and EPS-II were weaker than the reaction with CPS-II, the absorbance for which was maximal. The interaction of the miniAbs_EPS-I and miniAbs_EPS-II with LPS was stronger than it was with CPS-I (Table 1).

Fig. 2

ELISA of CPS-I, CPS-II, EPS-I, EPS-II, and LPS by using miniAbs_CPS-II prepared after the first (a) and fourth (b) selection rounds

Table 1 Comparison of the ELISA results for the polysaccharide-containing antigens of H. seropedicae Z78

The miniAbs_CPS-I interacted with all antigens of H. seropedicae Z78, but the interaction peaked for LPS, EPS-II, and CPS-II to an equal degree. The reaction with CPS-II was somewhat stronger than it was with CPS-I.

Analysis of the data indicates that the polysaccharide-containing antigens of the H. seropedicae Z78 surface have a complex nature. Clearly, LPS, CPS-II, and CPS-I have individual antigenic determinants which EPS-I and EPS-II lack and which are recognized by miniAbs_LPS and miniAbs_CPS-II. In LPS, CPS-II, and CPS-I, the miniAbs_EPS-II detect antigenic determinants that were present in EPS-II and EPS-I.

Inhibition of the glycopolymer–miniAb interaction

We examined the inhibition of miniAbs by commercial rhamnose, galactose, glucose, and N-acetyl-d-glucosamine (all sugars part of LPS), as well as by the OPS and core oligosaccharide of H. seropedicae Z78 (Velichko et al. 2018). The negative control was commercial altrose. The monosaccharide concentration (10⁻⁴ M) was purposely chosen in excess of the concentration of the miniAbs. The concentration of miniAbs_LPS was  1.2× 10¹³ virions ml⁻¹, and that of LPS was 0.015 mg ml⁻¹.

The inhibition of the immunochemical reactions decreased in the order core oligosaccharide ˃ rhamnose ˃ OPS ˃ glucose ˃ galactose ˃ N-acetyl-d-glucosamine (Fig. 3). Altrose did not affect the completeness of the LPS–miniAb reaction and did not inhibit the miniAbs. The extent of interaction of altrose-treated miniAbs with LPS was the same as in the nontreated control. The core oligosaccharide inhibited the miniAbs_LPS completely; OPS inhibited them to a lesser extent; and glucose, galactose, and N-acetyl-d-glucosamine had equally intense inhibitory effects, which were greater than the effect of OPS.

Fig. 3

Inhibitor effects on miniAbs_LPS, as evaluated by ELISA

Discussion

Knowledge concerning surface polymers is conducive to a better understanding of bacterial interactions with macrosymbionts. Studies of biopolymer structure are important from both basic and applied perspectives. The major, highly conserved structures, which are important for the immunochemical behavior of microorganisms, are LPS, CPS, and EPS (Holst et al. 1996; Ovodov 2006; Weidenmaier et al. Weidenmaier and Peschel 2008). Besides being implicated in the mechanical attachment of bacteria to the root surface, CPS and EPS ensure the error-free recognition of the plant host. The unique chemical structure of LPS, formed from three structurally different portions (lipid A, core oligosaccharide, and O polysaccharide), determines its broad biological activity (Holst et al. 1996). Under certain conditions, some bacteria can produce LPS extracellularly. The capsular glycans of some bacteria are represented by LPS (Konnova et al. 1994; Whitfield and Roberts 1999; Smol'kina et al. 2010).

The fine mechanisms of Herbaspirillum interactions with host organisms have been understudied. Silva-Froufe et al. (2009) used polyclonal antibodies raised against whole bacterial cells to detect bacteria in the plant tissue interior. Antibodies to component II (nifH or Fe protein) of the nitrogenase complex from Rhodospirillum rubrum were used to evaluate the nitrogenase activity of endophytic herbaspirilla (Reinhold et al. 1987; James et al. 1997; Olivares et al. 1997; James et al. 2002).

In this context, it is of interest to obtain data that will permit judgment of the degree of heterogeneity, chemical makeup, and localization of the antigenic determinants of the Herbaspirillum surface glycopolymers. In addition, results from comparisons of the immunochemical specificities of LPS, CPS, and EPS may serve the needs of microbiology and immunology. Because the CPS of H. seropedicae Z78 is an extracellular form of LPS [ Smol´kina et al. 2012], the removal of the capsules from the cell surface was absolutely necessary to prevent contamination of the LPS preparations with the surface glycopolymers. The monosaccharide composition of H. seropedicae Z78 CPS-I, CPS-II, EPS-I, and EPS-II was described by us earlier [Smol´kina et al. 2012], as were the monosaccharide composition and the structure of H. seropedicae Z78 LPS [Velichko et al. 2018]. In this study, we used the following preparations: EPS-I, EPS-II, CPS-I, CPS-II, and LPS.

To infer about the presence or absence of common antigenic determinants in the EPS-I, EPS-II, CPS-I, CPS-II, and LPS of H. seropedicae Z78, we ran a comparative immunoassay with rabbit polyclonal and phagerecombinant antibodies to the surface glycopolymers of H. seropedicae Z78. Experiments with rabbit polyclonal antibodies to glutaraldehyde-treated whole cells of strain Z78 showed that the CPS-I, CPS-II, EPS-I, EPS-II, and LPS, had some antigenic differences. Glutaraldehyde modifies protein epitopes and makes impossible an immune response to native membrane proteins, enabling the preparation of antibodies against bacterial LPS. The antibodies so prepared specifically interact with the carbohydrate components of the bacterial surface glycopolymers. These results indicate that although the polysaccharide components of EPS-I and CPS-I containe the same sugars [Smol´kina et al. 2012], they lack common antigenic determinants. Similar findings have been reported elsewhere for structurally similar glycoconjugates of other bacteria in which serological differences were found. For instance, the K and O antigens of Proteus mirabilis O40 are structured similarly but differ serologically [Kenne & Lindberg, 1983]. Rabbit anti-LPS antibodies did not cross-react with any of the antigens used, including LPS.

Efforts to raise antibodies against purified LPS were unsuccessful, possibly owing to the structural peculiarities of H. seropedicae Z78 LPS. Previous work by us (Velichko et al. 2018) has found that the OPSrepeating unit in H. seropedicae Z78 consists of glycerol-1-phosphate substituted by residues of N-acetyl-d-glucosamine. Structures of this kind are typical of the teichoic acids of gram-positive bacteria (Naumova et al. 2001) and are rare in gram-negative bacteria (Kondakova et al. 2005; Zdorovenko et al. 2011; Shashkov et al. 2015). Many studies of teichoic acids in a range of microorganisms, including staphylococci, bacilli, pneumococci, lactobacilli, and listeria, have shown that these acids have antigenic properties and can induce immune responses (Baddiley and Davison 1961; Clark et al. 2000; Wicken and Knox 2016). An important condition for an immune response is the natural surroundings of teichoic acids in an intact cell or cell wall, because purified teichoic acids are nonimmunogenic. Antigenic properties may also be determined by the nature of the polyol and glycosyl substituents in a biopolymer (Naumova et al. 2001).

Very few reports have used scFv antibodies for immunochemical studies of bacterial glycopolymers. However, those reports show that such antibodies prove more sensitive that traditional monoclonal antibodies. Thus, Griep et al. (1998), using recombinant antibodies against the LPS of Ralstonia solanacerum (biovar 2, race 3), recorded 5 × 10³ microbial cells in potato tuber extracts. For Herbaspirillum, this is the first time that miniAbs to the exopolysaccharides (EPS-I and EPS-II), capsular polysaccharides (CPS-I and CPS-II) and lipopolysaccharide (LPS) have been obtained.

The procedure that we used to increase miniAb specificity proved highly effective. From round to round, there were increases in the number of phage carrying specific variable domains to the corresponding antigens (Fig. 2). To look into the structure of the antigenic determinants, we inhibited the formation of antigen–antibody complexes with various competitive components of known chemical composition (Kabat and Mayer 1961). The results for the inhibition of LPS–miniAb precipitation with mono- and disaccharides suggest that these carbohydrates are part of the immunodominant sites of the O antigens.

The results obtained correlate well with the data on the monosaccharide composition of the antigens examined (Smol'kina et al. 2012; Velichko et al. 2018). LPS, CPS-II, and CPS-I contain rhamnose, N-acetyl-d-glucosamine, and N-acetyl-d-galactosamine, which may form part of their antigenic determinants. This explains why the miniAbs_LPS, miniAbs_CPS-I, and miniAbs_CPS-II not only interacted but also cross-reacted with LPS, CPS-II, and CPS-I. We speculate that EPS-II and EPS-I did not react with miniAbs_LPS and miniAbs_CPS-II because they contained no rhamnose and only trace amounts of N-acetyl-d-glucosamine and N-acetyl-d-galactosamine.

The interaction of miniAbs_CPS-I and miniAbs_CPS-II with LPS, CPS-II, and CPS-I could be explained by the presence in them of N-acetyl-d-glucosamine and galactose. On the basis of the foregoing, we speculate that miniAbs_LPS and miniAbs_CPS-II should be more specific for the determinants containing rhamnose, N-acetyl-d-glucosamine, and N-acetyl-d-galactosamine; miniAbs_CPS-I and miniAbs_CPS-II should be more specific for the determinants containing N-acetyl-d-glucosamine and galactose; and miniAbs_CPS-I should be specific for the determinants including all the above sugars.

The miniAbs_LPS were more specific for the core region of the LPS, in which rhamnose was the most abundant sugar, than they were specific for its O portion. These results are in harmony with our speculation that miniAbs_LPS have the highest specificity for rhamnose-carrying determinants. The inhibition of miniAbs_LPS by a wide range of sugars possibly indicates that the core oligosaccharide is highly branched and heterogeneous.

The use of highly specific miniAbs makes it possible to detect identical antigenic determinants in samples whose structures have not yet been examined, to compare those samples with polysaccharides of known structure, and to detect microorganisms and bioactive molecules.