Wadsworth Center, Division of Molecular Medicine, New York State Department of Health and Department of Biomedical Sciences, School of Public Health, The University at Albany, Albany, NY 12201-0509, USA
Wadsworth Center, Division of Molecular Medicine, New York State Department of Health and Department of Biomedical Sciences, School of Public Health, The University at Albany, Albany, NY 12201-0509, USA
This review is part of a series of reviews dealing with different aspects of bacterial S-layers; all these reviews appeared in Volume 20/1–2 (June 1997) of FEMS Microbiology Reviews, thematic issue devoted to bacterial S-layers.
In this chapter we report on the molecular biology of crystalline surface layers of different bacterial groups. The limited information indicates that there are many variations on a common theme. Sequence variety, antigenic diversity, gene expression, rearrangements, influence of environmental factors and applied aspects are addressed. There is considerable variety in the S-layer composition, which was elucidated by sequence analysis of the corresponding genes. In Corynebacterium glutamicum one major cell wall protein is responsible for the formation of a highly ordered, hexagonal array. In contrast, two abundant surface proteins form the S-layer of Bacillus anthracis. Each protein possesses three S-layer homology motifs and one protein could be a virulence factor. The antigenic diversity and ABC transporters are important features, which have been studied in methanogenic archaea. The expression of the S-layer components is controlled by three genes in the case of Thermus thermophilus. One has repressor activity on the S-layer gene promoter, the second codes for the S-layer protein. The rearrangement by reciprocal recombination was investigated in Campylobacter fetus. 7–8 S-layer proteins with a high degree of homology at the 5′ and 3′ ends were found. Environmental changes influence the surface properties of Bacillus stearothermophilus. Depending on oxygen supply, this species produces different S-layer proteins. Finally, the molecular bases for some applications are discussed. Recombinant S-layer fusion proteins have been designed for biotechnology.
Since the discovery of S-layers in 1953 of a Spirillum sp. by Houwink, the structure of hundreds of different S-layer proteins and their morphological properties in a wide range of microorganisms have been studied in detail. In the past years genetic investigations of bacterial S-layer genes have become an important, additional contribution for a better understanding of S-layer protein and gene organization. However, little is still known about the regulatory mechanisms of transport, biosynthesis and the domains which are responsible for intra- and/or intermolecular interactions of layer proteins. Although all S-layers share the identical feature of a two-dimensional (glyco-)protein layer covering the bacterial cell, the analysis of various S-layer proteins revealed the existence of a great diversity in their amino acid composition. The majority of S-layer genes sequenced so far have only little sequence identity, suggesting an analogue rather than a homologue evolutionary process of these structures. An important contribution to S-layer diversity and the understanding of S-layer gene regulation represents the ability of an individual bacterial species to express different S-layer genes under altered environmental conditions. The investigation of S-layer variation at the molecular level will allow us to get a better insight into gene regulation of bacteria. During the last years S-layers have been shown to have considerable application potentials. Beside the chemical modification of S-layers, like the immobilization of functional molecules, the construction of recombinant S-layers by insertional mutagenesis opens a new way in vaccine development. The specific properties of the S-layer protomers in combination with molecular techniques will allow the synthesis of recombinant S-layers with high potential in biotechnical applications.
The following articles will give an overview of the knowledge we have today about S-layer gene expression, variation and the construction of recombinant S-layer genes.
2The crystalline surface layer of Corynebacterium glutamicum
Nicolas Bayan3, Mohamed Chami, Gérard Leblon, Thaddée Gulik-Krzywicki and Emanuel Shechter
C. glutamicum is a Gram-positive bacterium widely used for the industrial production of amino acids such as glutamic acid. Our knowledge concerning the metabolism and the molecular biology of this microorganism is now well developed. In contrast, little is known about the structure and composition of the cell wall. In this article we report on the information available about the crystalline surface layer of corynebacteria.
2.1Freeze-etching electron microscopy of C. glutamicum cells
The various levels of the cell wall of C. glutamicum were visualized by electron microscopy using freeze-fractured and deep-etched preparations of whole untreated cells . Under our experimental conditions, the main fracture plane propagated within the cell wall of the bacterium, close to the cell surface (SL) (Fig. 1A), and produced two fracture surfaces named F1, the convex fracture surface, and F2, the concave fracture surface (Fig. 1B). In some rare instances, the fracture was propagated within the cytoplasmic membrane (CM) (Fig. 1C). This unusual behavior, which was also observed by Richter et al. , suggests the presence of a hydrophobic layer in the cell wall. The cytoplasmic membrane (CM) was densely covered with particles representing integral membrane proteins. The cell surface as well as both fracture surfaces, F1 and F2, were covered with ordered arrays undoubtedly representing the so called S-layers found in nearly all taxonomic groups [3, 4]. Optical diffraction of these lattices (Fig. 1A, inset) showed a hexagonal symmetry with a cell unit dimension of 132 Å. Surprisingly, for some strains, the area occupied by the ordered arrays on the cell surface depended on the growth conditions of the bacteria . While cells grown on solid support were completely covered with ordered arrays, cells grown on a liquid medium displayed ordered arrays separated by smooth, non-etchable regions. Incomplete formation of an S-layer was also reported previously in the case of Bacillus thuringiensis where this phenomenon was shown to be related to growth phase . The discontinuity of the ordered arrays when growth was performed on liquid medium made it possible to correlate the ordered arrays seen on the cell surface and those seen on the fracture surfaces F1 and F2. Indeed, the positions of the ordered regions and smooth regions on the cell surface match exactly the ordered and smooth area of the fracture surfaces F1 and F2. Careful observation of several images suggests that the ordered arrays observed on the F2 fracture surface resulted from the presence of particles while those observed on the F1 fracture surface were clearly the imprint of the former ones on the underlying cell wall material . The ordered arrays seen on the cell surface and those seen on F1, although displaying the same symmetry, have very different aspects. It is not clear from these freeze-etching electron microscopy images if they originate from the same lattice or if they represent two distinct lattices in close contact. Most S-layers are constituted of a single lattice displaying an outer surface (exposed to the external medium) smoother than the inner surface (adhering to the underlying cell wall material). However, the presence of two superimposed layers of ordered molecules at the surface of a bacterial cell has already been described for some organisms such as Bacillus brevis 47 , Corynebacterium diphtheriae C4 , Aquaspirillum serpens MW5 , Aquaspirillum‘ordal’, Nitrosocystis oceanus and Lampropedia hyalina.
2.2Isolation and chemical characterization of the S-layer
The S-layer of C. glutamicum was shown to be very strongly associated with the cell . In contrast to almost all S-layers, denaturing agents such as urea or guanidine hydrochloride were unable to detach the S-layer from the cell. The release of large sheets of S-layer required incubation of the bacteria with 2% SDS indicating that strong hydrophobic interactions are involved. Under these conditions, the lattice was not disrupted unless the preparation was heated to 100°C indicating that the bonds between lattice subunits are stronger than those with the underlying cell wall material. Such stability for an S-layer has also been described in the case of Deinococcus radiodurans (formerly Micrococcus radiodurans) by Baumeister et al. .
The large sheets of S-layer released upon SDS treatment were recovered after differential centrifugation. These sheets display two different faces (Fig. 2A). One resembles the ordered arrays seen on the cell surface; the other resembles those observed on the F2 fracture surface. Biochemical analysis of these sheets indicated the presence of a single protein (molecular mass 63 kDa) corresponding to PS2, the major cell wall protein of C. glutamicum (Fig. 2A, inset). According to our data, PS2 is, in contrast to several S-layer proteins, not glycosylated [14, 15]. Incubation of the cells in the presence of non-specific proteases also led to the release from the cells of large sheets of S-layers. However, these protease sheets displayed symmetrical faces resembling the ordered arrays seen on the surface of the bacterium (Fig. 2B). SDS-PAGE analysis revealed the presence of a slightly truncated form of PS2 (molecular mass 51 kDa) in this preparation (Fig. 2B, inset). If the S-layer of C. glutamicum is composed of two distinct lattices, then the protease may have cleaved the domain of PS2 involved in the association with the second lattice of the S-layer whose organization would be strictly dependent on the presence of the PS2 lattice. In this case, the underlying lattice is probably not a protein since PS2 is the only one found in S-layer sheets isolated with SDS. Its nature remains to be determined. Alternatively, if the S-layer of C. glutamicum is composed of a single PS2 lattice, then one has to assume that the protease cleaved part of the polypeptide chain of PS2, responsible for the formation of the inner face of the S-layer.
2.3Cloning and sequencing of the S-layer protein corresponding gene
The cspB gene encoding PS2 was cloned in λ gt11 by immunological screening . Analysis of the nucleotide sequence revealed an open reading frame of 1533 nucleotides. The deduced 510 amino acid polypeptide has a calculated molecular mass of 55 426. No significant homologies were found with other proteins but PS2 shares several features with surface layer proteins. The similarities include a high content of hydrophobic amino acids (45.2%), a higher proportion of acidic amino acids (17.7%) as compared to basic amino acids (8.3%), a very low content of sulfur-containing amino acids (totally absent in the mature form of PS2) and the presence of putative N-glycosylation sites (seven). The absence of a sequence homology of PS2 with other known surface layer proteins is not unexpected as surface layer proteins display little sequence homology between themselves .
S-layers, which are the outermost envelope component, represent an important interface between the cell and its environment. They probably act as a barrier controlling the exchange of molecules, as protective coats or as ion traps. In some cases, they were identified as virulence factors [18–20]. In order to test the functional role of the S-layer of C. glutamicum, the cspB gene was disrupted by insertion, in its 5′ region, of the aphIII gene of Streptococcus faecalis which confers kanamycin resistance. In such a strain, we could show that PS2 is not expressed and that the S-layer is totally absent. In contrast to the wild-type strain, the cell surface of the mutant displayed a smooth appearance. Under laboratory conditions, the absence of the S-layer did not result in any phenotypic particularity of the strain. The function of the S-layer in C. glutamicum is thus unclear.
According to the predicted amino acid sequence, PS2 is synthesized with a N-terminal segment of 30 amino acid residues reminiscent of eukaryotic and prokaryotic signal peptides (Fig. 3). In contrast to the two other secreted proteins described in C. glutamicum, which possess a large number of positive charges at their N-termini [21, 22], the signal sequence of PS2 is short and displays only two positive charges. Hybrid proteins resulting from the fusion of the putative signal sequence of PS2 and the catalytic domains of the TEM1 β-lactamase, the endoglucanase A of Clostridium thermocellum or the amylase of Streptomyces limosus were shown to be efficiently translocated in C. glutamicum. Moreover, PS2 which can be expressed in E. coli is exported in the periplasm of this organism. All these results confirm that the N-terminal sequence of PS2 is a sequence signal and is recognized by the general secretory pathway which probably exists in C. glutamicum.
2.4Attachment to the cell
PS2 is distributed between the cell surface and the external medium. In the wild-type strain, the protein is mostly present on the cell surface. As we pointed out previously, protease treatment abolishes the interaction of the S-layer with the cell and leads to the release of large S-layer sheets in the medium. These sheets are composed of a truncated form of PS2. More precisely we could determine that the C-terminal part of the protein (the last 79 residues) has been cleaved . This suggests that this portion of the protein is involved in the attachment of the lattice to the cell. According to the predicted amino acid sequence, PS2 possess a very hydrophobic tail of 21 amino acids (Fig. 3) rich in alanine and isoleucine. In order to determine if this sequence is involved in the tight association of the S-layer to the cell we constructed a truncated form of PS2 which does not contain this C-terminal hydrophobic sequence. The truncated form of PS2 was very highly expressed in C. glutamicum, but failed to associate with the cell wall (Fig. 4) and thus does not form any ordered arrays on the surface of the bacterium . Practically all PS2 proteins were recovered in the culture medium as monomers. Clearly, the C-terminal hydrophobic stretch of 21 amino acids is the main anchor of the protein to the cell wall.
Since S-layers are found in almost all taxonomic groups of bacteria, they are associated with a wide variety of cell wall structures. In many cases a consensus sequence (named S-layer homology, SLH) has been shown to be involved in the association of the S-layer with peptidoglycan [25, 26]. In Campylobacter fetus and in Caulobacter crescentus, an interaction of the S-layer protein (via its N-terminal domain) with lipopolysaccharide was demonstrated [27, 28]. In the case of corynebacteria, no SLH domain could be identified and the presence of a C-terminal hydrophobic anchoring signal seems to be very specific since no similar sequences are found in all known S-layers except for Halobacterium halobium , Haloferax volcanii and Rickettsia prowazekii. For R. prowazekii, a Gram-negative bacterium, the C-terminal hydrophobic sequence could be anchored in the outer membrane, while for halobacteria it was proposed that the C-terminal hydrophobic sequence is anchored in the cytoplasmic membrane . In the case of C. glutamicum, it is difficult to assume an insertion of this sequence in the cytoplasmic membrane since the cell wall of corynebacteria is very thick (Fig. 1). The S-layer is thus located far above the plasma membrane. Alternatively, PS2 could be anchored in a hydrophobic layer of the cell wall. The existence of such a layer is suggested by the recent finding of porins in Gram-positive bacteria  and more particularly in corynebacteria . This layer may be composed of mycolic acids known to be present in the cell wall of C. glutamicum.
As we mentioned, PS2 deleted from its hydrophobic anchor is shed in the medium as monomers suggesting that the formation of the S-layer lattice requires the interaction of PS2 with the cell. The association of PS2 with the cell surface by the C-terminal hydrophobic sequence may favor interactions between monomers by lateral diffusion on the surface of the bacterium. The surface of the cell would act as a matrix for crystallization, a phenomenon similar to that leading to the crystallization of proteins on lipid monolayers .
2.5Conclusions and perspectives
The presence of an S-layer at the surface of C. glutamicum has been clearly demonstrated. However, the exact organization of this structure on the surface of the bacterium is still unclear. Indeed, freeze-etching electron microscopy studies did not allow us to discriminate between the presence of one or two superimposed lattices. Additional structural studies are required. Yet, at this stage, ultrathin sections or negatively stained preparations of isolated S-layer sheets have not given additional information.
However, it is clear that PS2 is responsible for the formation of this S-layer. This protein is constituted of two domains. The N-terminal domain involves the major part of the polypeptide and is responsible for monomer interactions. The C-terminal domain is responsible for the attachment of the protein to the cell wall. In this latter domain the terminal hydrophobic stretch of amino acids is absolutely required. Such distinct domains have also been described for some other S-layer proteins [37, 38], but the presence of a hydrophobic terminal sequence is not frequent. Its presence in corynebacteria may be related to that of mycolic acids which are responsible for the formation of a hydrophobic barrier as is the case in mycobacteria [39, 40]. This remains to be determined.
3Bacillus anthracis S-layer
Agnès Fouet3, Stéphane Mesnage, Evelyne Tosi-Couture, Pierre Gounon and Michèle Mock
Bacillus anthracis is the etiologic agent of anthrax, a disease which occurs in many animals including man. It is a Gram-positive spore-forming bacterium. The major virulence factors are two toxins and a poly-γ-d-glutamic acid capsule; their corresponding genes reside on two plasmids [41, 42]. The surface of non-capsulated vegetative bacilli appears layered , and fragments display a highly patterned ultrastructure . S-layers are found ubiquitously , and they may be an important virulence factor . Therefore, we investigated the S-layer of this pathogenic organism in a strain cured for both plasmids.
3.1Genetic analysis of the S-layer components
3.1.1Cloning of the S-layer genes
We initially hypothesized that a major bacterial protein, which is often found in high abundance in the B. anthracis culture supernatant, was an S-layer component. This protein (molecular mass 94 000) is produced by various B. anthracis strains, including strains without plasmids, and must therefore be chromosomally encoded. The N-terminal sequence of this protein (Sap, surface array protein) was determined as well as those of polypeptides obtained by limited chymotrypsin digestion. Two oligonucleotide probes were thus designed and used to isolate sap, which was then sequenced. This gene was disrupted, and no other protein with a molecular mass of 94 000 was observed in the culture supernatant of this mutant. However, another abundant protein with the same molecular mass, EA1 (for extractable antigen 1), was observed in the pellet fraction of the Δsap mutant. Since this mutant still possesses an S-layer, we analyzed EA1. By an approach similar to that used for sap, the eag gene, encoding EA1, was isolated, and then sequenced [46a]. eag was also disrupted, and we then constructed a double deleted mutant. Both genes, sap and eag, are clustered on the chromosome. They are in the same orientation, separated by a non-coding fragment of approximately 700 bp, eag following sap. The presence of this sequence, together with the synthesis of EA1 in the sap disrupted mutant, suggests that these genes are not organized as an operon. This differs from the situation encountered in Bacillus brevis where the genes encoding the S-layer components belong to an operon .
To test whether there were any other putative S-layer component genes, various probes were derived from sap and eag and used to screen B. anthracis chromosomal DNA, by Southern blots at different stringencies. The positive signals could all be explained by hybridization to sap or eag suggesting that there is no other homologous sequence. Also, no other abundant protein appeared in the double mutant, suggesting the absence of another S-layer component gene. Alternatively, in Campylobacter fetus, there are many genes which can rearrange and give rise to protein variation [48, 49]. In this bacterium, the protein diversity is also the consequence of the presence of an invertible DNA segment containing a promoter .
No classical Gram-positive ‘−35’‘−10’ consensus sequences  are found upstream from sap or eag open reading frames. This is similar to a constitutive, highly expressed promoter of the B. brevis cwp operon encoding S-layer proteins . Shine-Dalgarno motifs, consistent with those observed in Gram-positive bacteria , are found before both ATGs. Putative ρ independent transcription sequences are found downstream of each gene, the one behind eag being stronger than the other one.
The deduced amino acid sequences of Sap (814 residues) and EA1 (862 residues) show common features (Fig. 5). Both proteins harbor a classical Gram-positive signal peptide . In each case, the mature protein starts with three SLH motifs, as described by Lupas et al. . In fact, these two proteins are very similar in these 200 first residues, with an identity of 66% and a similarity of 74%. The resemblance is higher than would be required for the mere consensus conservation. For example, each first SLH motif shares more sequence identity with the other first motif than with the second motif of the same protein, and this pattern continues for each motif. After this region, the proteins differ appreciably, the identity decreasing to 22%. Thus these proteins could be composed of two domains; the first one, constituted by the SLH motifs, could be the consequence of a duplication event. This SLH harboring domain could permit the anchoring of these proteins to the peptidoglycan containing sacculus [25, 26, 54].
Further research in the protein data banks showed very little similarity for Sap outside the SLH motifs. This is in contrast to EA1 which is very similar to the sequence of the Bacillus licheniformis RS (OlpA) S-layer protein (accession number U38842) (Fig. 5) [54a]. The first 200 residues share 93% identity and 97% similarity; the scores are still very high, 63% and 76% respectively, for the rest of the proteins [46a]. Since a high level of similarity, such as that found between EA1 and RS, is infrequently encountered in S-layer proteins, it can be hypothesized that they are derived from the same ancestral protein. Also, B. anthracis and B. licheniformis possess not only an S-layer  but also an original poly-γ-d-glutamic capsule . Selective pressure could therefore have occurred to minimize changes in the sequence of the S-layer protein amino acid residues interacting with the capsule. Comparison of these two organisms' cell wall structures would be very informative.
3.2Protein analysis of the S-layer components
The presence of two S-layer components could be due to the simultaneous synthesis of both proteins, or to the activation of a second gene in the deleted mutant. The protein content of the wild-type strain was analyzed by SDS-PAGE, and, after obtaining specific antibodies to EA1 and Sap, by Western blots and by immunoelectron microscopy. The results indicated that both proteins are synthesized in the wild-type strain. EA1 is nearly exclusively cell associated, whereas Sap is equally cell associated and in the supernatant. Thus, both proteins are on the surface of the bacteria, and EA1 is more tightly bound to the bacteria than Sap. These results were supported by the immunoelectron microscopy data.
A similar situation is encountered in some organisms such as Aquaspirillum serpens and B. brevis. In contrast, in Lactobacillus acidophilus, C. fetus, Thermus thermophilus, or Bacillus stearothermophilus, the S-layer proteins are synthesized sequentially. In L. acidophilus, an interchange of the active and silent genes encoding the S-layer proteins occurs by inversion of a chromosomal segment . In C. fetus, reciprocal recombinations also induce expression of other genes . The expression of an S-layer-like array is induced by the deletion of the S-layer protein gene in T. thermophilus. In B. stearothermophilus, only one S-layer protein is synthesized under a given condition; a different gene is expressed when there is an environmental modification .
3.3Phenotypic analysis of wild-type and mutant strains
The main morphological differences seemed Sap-dependent, as indicated by the following. The Δeag and the wild-type bacteria looked alike by various criteria. Alternatively, the Δsap and the ΔsapΔeag double deleted mutants shared common features, which differed from those of the Sap+ strains. The Sap− colonies are larger than the Sap+. In liquid medium, Sap− mutants flocculated and sedimented abnormally when shaking was arrested (Fig. 6). The Δeag mutant bacilli had the wild-type rod-shaped appearance, but long filamentous bacilli were observed in all Δsap mutant strains.
We analyzed the surface array structure of the wild-type and the three mutant strains. B. anthracis cell envelopes were obtained after disruption and negatively stained (Fig. 7). Preliminary optical diffraction analysis suggests that both the wild-type and the Δsap mutant envelope micrographs exhibit diffraction patterns, whereas no diffraction pattern was observed in the double mutant. Thus EA1 is indicated to be an S-layer component. The diffraction patterns obtained with micrographs from wild-type and Δsap envelopes seem slightly different, indicating that Sap is also an S-layer constituent. The slight difference suggests that the EA1 lattice could determine the main pattern of the architecture of the B. anthracis S-layer. However, we failed to visualize an array in the Δeag mutant. Thus, either there are two S-layers, with Sap forming a more fragile structure, requiring the EA1 array to be stable, or there is a single S-layer, with EA1 and Sap forming a protein complex in an array. Although most S-layers result from the assembly of a single protein , certain S-layers contain two abundant surface proteins which are synthesized simultaneously. Some organisms, such as A. serpens, seem to have two superimposed S-layers composed of one protein each . For B. brevis, the arrays are different with either one or both proteins . Alternatively, in Clostridium perfringens, the S-layer is composed of two proteins in equal amounts; no regular array is observed with the self-assembly of each protein alone .
3.4In vivo expression
The in vivo expression of each protein was analyzed by Western blots, using sera from mice injected with an EA1+ Sap+ strain. A strong signal was obtained with EA1, but Sap was barely visible. Thus, EA1 is synthesized in vivo, whereas Sap is either weakly produced or not antigenic. In fact, EA1 appeared as a major antigen. Our data support the conclusion that the previously described extractable antigen 1  and the S-layer component that we have studied are the same protein. The role of the S-layer in B. anthracis is not easy to conceptualize. Fully virulent bacilli are also capsulated, and the capsule could mask the S-layer from the host macromolecules. The B. anthracis S-layer(s) is composed of two simultaneously synthesized proteins with a molecular mass of 94 000. The N-terminal sequences, composed of three SLH motifs, are highly similar. Refining the array structure should enable us to define whether there are one or two S-layers. The in vivo expression of both components can also be further analyzed. The contribution of the relevant proteins in virulence will then be studied. We will assay the possible advantages provided by this structure to the bacterium, such as in vivo survival or development. After injecting mutants constructed during the course of this work and others from a capsulated strain, protection against infection and antigenic responses will be assayed.
4S-layer and ABC transporter genes in methanogenic archaea
Everly Conway de Macario3 and Alberto J.L. Macario
Immunologic analyses have revealed that archaea are immunogenic, and that while their antigenic relationships are coherent within each group and parallel their phylogenetic classification, their antigenic mosaics are quite diverse . This antigenic diversity surely reflects diversity and variability of surface envelope structures. Among these structures, those belonging to the S-layer must contribute significantly to the antigenic mosaic [4, 45, 65, 66]. This brief review will focus on methanogenic archaea (methanogens), and will proceed from the picture uncovered by immunologic analyses to the data currently available on the genes encoding surface structures, with comments on what these data suggest about possible mechanisms responsible for generating molecular (antigenic) diversity. A recently discovered gene cluster, encoding proteins of the ABC transporter system which are important structural and functional components of the cell's envelope, will also be discussed. Due to strict space limitations, the scope will be reduced to essentials.
Methanogens have been found in many different ecosystems. For example, immunologic analyses along with other studies have demonstrated the presence and variety of methanogens in human and animal intestinal tracts [67–69], human dental plaque  and vagina , psychrophilic  and very cold  environments, sea-water column , river sediments , deep subsurface water , deep oil-bearing rocks , copper-mining soil , peat bogs , landfills , and bioreactors processing several kinds of substrates, wastes, and waste waters [81–84]. These data demonstrate that methanogens are globally distributed, and can survive or thrive under different conditions determined by temperature, pressure, pH, and salt concentration. A priori this universality suggests that methanogens will display a variety of cell envelopes (surfaces), each suited to living in the respective habitat.
Universality suggests envelope diversity. Since there are many examples in which the same species has been found in different ecosystems, one may infer that envelope diversity occurs within a single species. Indeed, direct immunologic analyses of new isolates, enrichment cultures, and samples from complex natural and semimanufactured (e.g. bioreactors processing various kinds of wastes or waste waters) microbial communities have revealed that a single species can display different immunotypes. Although immunotypes are often distinctive of habitat, different immunotypes for the same species have also been found within a single niche. Furthermore, these immunologic studies have demonstrated that many of the methanogens found in the ecosystems investigated are different from those used as reference (i.e. those available in culture collections as pure, well-characterized cultures) (see references in Section 2, above). Antigenic diversity within a single species is illustrated by the various immunotypes of Methanobrevibacter smithii isolates from human feces that were identified with antibody probes made against the reference strain PS . Forty-six isolates, all identified as M. smithii by a variety of procedures, could be classified into nine groups (immunotypes) on the basis of their antigenic mosaics (Table 1).
Table 1. Antigenic diversity and antigenic mosaic complexity of Methanobrevibacter smithii isolates determined with a panel of monoclonal antibody probes
aForty-six isolates characterized as Methanobrevibacter smithii by a variety of complementary procedures [68, 85].
bEach antigenic determinant, a–f, was recognized by a single monoclonal antibody probe, A–F, respectively.
cIsolates in a given group possess and lack the same determinants as detected by the antibody probes A–F (e.g. isolates of group II possess determinants a, b, d, e, and f, and lack determinant c).
dThis figure shows how many of the determinants recognized by the panel of monoclonal antibody probes A–F were present on the cells of each group (I–IX).
eFigures in this line show how many times (i.e. frequency) the determinants occurred in the 46 isolates with the corresponding percentage in the line below.
Data in Table 1 show that the antigenic mosaic of M. smithii PS is complex, with at least six different components (antigenic determinants), a–f. The data also show that while some of the isolates (group I) display all these determinants, most do not.
The mosaic of methanogens was dissected further using immunochemical procedures. For example, the mosaic of Methanobacterium thermoautotrophicumΔH was dissected into six distinct components using a panel of monoclonal antibodies and competitive inhibition assays with compounds of known structure . Similarly, the mosaics of several other methanogens were elucidated, and all of them were complex .
Methanosarcinas are methanogens which appear in different forms, uni- and multicellular. For example, Methanosarcina mazei S-6 undergoes morphologic conversions in culture, with three cardinal forms, single cells, laminae and packets (the latter two are multicellular structures) . These forms are also found in natural and semimanufactured ecosystems. It is likely that the cell surface undergoes changes during conversion of one form into another. Thus, one might expect distinctive antigenic determinants, and mosaics, for each cardinal form, and mosaic changes as one form progresses into another. In fact, considerable antigenic diversity has been found for the packet form of M. mazei.
Also, quantitative and topographic changes of surface antigens have been observed during the conversion of packets to laminae . M. mazei S-6 can, therefore, display various antigenic mosaics, depending not only on habitat but also on the growth stage and morphotype. Some of the features of the M. mazei S-6 S-layer genes and proteins to be discussed below provide clues to the mechanism that this species has evolved for generating antigenic (surface) diversity.
4.5S-layer genes and proteins
An S-layer gene (slgB) was cloned from the chromosome of M. mazei S-6 and sequenced . The protein, SlpB, encoded in this gene has most of the characteristics of another two S-layer proteins, SlpA, from the methanogens Methanothermus fervidus and Methanothermus sociabilis. The degree of sequence similarity between SlpB and SlpA is nearly 30% identity for the entire molecules, and reaches 44% when the terminal halves of the proteins are compared . The similarity of SlpA and SlpB to other archaeal S-layer proteins is lower. Both proteins have sequons, N-terminal signal sequence, two Asn-Asn clusters, predominance of Tyr, Trp, Asn, Cys and Ile, and predominance of β-sheet and very little α-helical structure, all features typical of S-layer proteins [4, 65, 91]. While the SlpAs from M. fervidus and M. sociabilis are almost identical, SlpB from M. mazei S-6 shows distinctive characteristics. One of the most remarkable features of SlpB is the occurrence of repeats in its C-terminal half . Two long repeats (125 amino acids) were identified, which are very similar to each other (Fig. 8). Four shorter repeats, named 1–4 from the N- towards the C-terminus, each 56 amino acids long, occur in the same region of the molecule, mostly overlapping the long repeats .
These shorter repeats are quite similar to each other, particularly repeats 2 and 4. Interestingly, repeats also occur in other open reading frames (orfs) adjacent to the gene that encodes SlpB (unpublished results), and in another cluster of genes that also encode envelope proteins (Fig. 9) . In this cluster, three genes, 5′-orf492-orf375-orf783-3′, encode the proteins ORF492, ORF375, and ORF783, respectively. ORF492 (i.e. the protein encoded by the gene at the 5′ end of the cluster) has a putative signal sequence, as SlpA and SlpB do , and the mature protein is made up of repeats (AB), two half repeats (A and B), a segment C (which is a repeat in ORF783), and a hydrophobic stretch. Very similar AB repeats, A and B half repeats, C segment, and hydrophobic stretch are present in ORF375. The protein ORF783 is truncated (no sequence is available for the 3′ end of orf783), and it is constituted of C repeats exclusively, which are very similar to the C segments present in the other two ORFs. The AB and C repeats average 42 and 85 amino acids, respectively.
The AB repeats are quite similar to each other, and so are the C repeats among themselves. However, the latter can be grouped into families or clusters composed of very similar repeats each (unpublished results). The AB repeats contain the SLH domain found in many S-layers . These genes were expressed in vivo [90, 92].
The mechanism(s) involved in generating the observed antigenic (surface) diversity of methanogens has not been elucidated. Perhaps, the organization and structure of the genes described in the preceding section play a role in the generation of surface diversity. It is possible that mechanisms similar to those postulated to operate in Bacteria and/or Eucarya for generating phase and antigenic variation, and other antigenic switches [93, 94], are also operative in archaea. Expression, or absence thereof, of a particular gene coding for a protein antigen, and expression of either one or the other of a series of versions of the same gene-protein antigen, may be part of the general mechanism that generates different envelopes and cell surfaces. Moreover, expression of a protein, or glycoprotein, not always present, may mask one or more antigens or antigenic determinants, and thereby render them unavailable for reaction with antibodies. This masking phenomenon can lead to the erroneous conclusion that a given protein, or determinant in it, is not present in the envelope when in fact it is just hidden. Antigen masking adds to the complexity of the phenotype as detected by antibodies.
The molecular genetic mechanisms leading to phase and antigenic variation in pathogens are believed to operate at the pre- and posttranscription levels, as well as during transcription itself. Phase variation, exemplified by the absence or presence of a protein antigen, may be caused by a pretranscriptional deletion of a gene (or part thereof), or translocation of the gene to a place where there is no functional promoter, or removal (or mutation) of the promoter. Transcriptional ablation or down-regulation may be due to a number of processes involving repressors, absence of activators (lack of them, or of their activation), and failure of the signal transduction pathway (if the gene in question is inducible by outside factors). Posttranscriptional down-regulation may involve changes in mRNA stability-degradation and processing, inhibition of translation (ribosome binding and/or elongation), and alteration of protein (antigen) translocation to the cell's surface. Any one of the above processes can cause the absence of a protein antigen, properly folded and postsynthetically processed, from the cell's surface.
Antigenic variation pertains to a single antigen molecule which is always expressed but in different versions of itself, e.g. longer or shorter than usual, with or without modifications of its amino acid sequence (e.g. amino acid substitutions). One or more of these structural changes will result in antigenic variation detectable by antibodies. The mechanisms that cells use to express on their surface either one or another of several versions of the same protein antigen are varied. For example, a cell may use a single gene with repeats or modules to generate proteins with different lengths depending on how many modules are transcribed, or to generate proteins not only differing in length but also in sequence depending on how the modules are rearranged before transcription. Modules are usually similar but not identical to each other. Thus, depending on which repeats are brought together in the coding region and then transcribed, the final protein product will vary. If, in addition, genes with similar repeats or modules in a cluster (e.g. operon) are used by a cell in a coordinated fashion, the possibilities for generating variability multiply.
The genes encoding the S-layer proteins in M. mazei S-6 described in the preceding section show repeats or modules. These genes are organized in clusters with a putative promoter of the archaeal type in the 5′ flanking region of each cluster. Some of the features of the genes' nucleotide sequences and of the deduced mRNA sequences suggest that regulation at the levels of transcription and translation is possible. The presence of modules in all genes and the sharing of some of these modules between the genes indicate that rearrangements and module exchanges may have occurred during evolution, and probably do still occur frequently enough to generate surface diversity. Future research ought to address these points and elucidate the mechanism(s) involved in gene and module rearrangements, and in the regulation of the rearranged genes.
Two genes, orfD and orfF encoding the proteins OrfD and OrfF, respectively, were discovered in the genome of M. mazei S-6, which are homologues of the nucleotide binding components D and F of bacteria and eukaryotes . The signatures typical of these components are present in the archaeal proteins (Table 2). Shorter motifs were identified that are the most conserved within the signatures or boxes. These short motifs are the amino acid triad GKS for box A, SGG for the ABC transporter family signature, DEP for box B, and HD for box C. Exceptionally, GKT replaces GKS, DDA or DDP occurs instead of DEP, and HR or HK substitutes for HD. SGG is a highly conserved triad, which is present even when the rest of the ABC transporter family signature is hardly discernable at the expected location (see Table 2). Recently, the other components of the archaeal ABC transporter system were cloned and sequenced . The gene cluster is 5′-orfA-orfB-orfC-orfD-orfF-3′ (Fig. 10). This organization is typical but not universal for bacterial systems. In some cases the D and F genes are upstream of B and C, while the A gene is at the 3′ end of the cluster (see Fig. 10). A rather common feature is the overlapping of adjacent genes over stretches encompassing between 1 and 20 bases for example. In eukaryotes, a single protein occurs with domains that are equivalent to the products of the B–C and D–F genes, the former domain being on the N-terminal half of the molecule, and the D–F domain on the C-terminal half (see Table 2). Interestingly, the four genes B, C, D, and F overlap in M. mazei S-6. Assuming that they are cotranscribed to yield a single mRNA, the possibility exists that this mRNA is translated into a single protein, very much like in eukaryotes. If that were the case, the B–F protein would represent another example that shows similarity between Archaea and Eucarya, and difference between the former and eubacteria.
Table 2. Comparison of the sequence and structural features of the archaeal nucleotide binding molecules OrfD and OrfF with those of bacterial and eucaryal homologuesa
ATP/GTP binding box A (8 aa)
ABC transporter (15 aa)
aData from  in which the names of the Bacterial and Eucaryal species compared and the data base accession numbers and references for the proteins are given. Reproduced with permission of the copyright owner.
bMolecular signatures defined by the GCG program Motifs. The amino acid position at which the motif begins in the sequence is indicated.
caa, amino acid(s); I, identity; S, similarity (identities plus conservative substitutions).
dData base accession numbers and literature references are listed in .
fThe sulfonylurea receptor proteins examined have a box A in a different location, beginning at position 713.
gThe sulfonylurea receptor proteins examined do not have this motif.
hThe cystic fibrosis transmembrane conductance regulator and sulfonylurea receptor proteins examined do not have this motif.
Bacterial range (n=9)
Eucaryal range (n=6) N-terminal half
Bacterial range (n=9)
Eucaryal range (n=6) C-terminal half
Perspectives are bright. The data available at the present time suggest a number of potentially very productive research lines. Among these, worth mentioning are: (a) analysis of the mechanisms that control transcription initiation of each gene and/or gene cluster; (b) elucidation of the molecular steps involved in module rearrangement; (c) determination of the transcriptional mode (mono- or polycistronic) for the gene clusters, and of the mRNA processing steps that follow; (d) study whether translational control and postsynthetic protein processing-modification also play a role in generating surface diversity; and (e) elucidation of whether the ABC transporter function is carried by a single protein with specialized domains like in eukaryotes, or by separate molecules like in eubacteria.
5The S-layer of Thermus thermophilus HB8: structure and genetic regulation
Luis Angel Fernández-Herrero, Garbiñe Olabarrı́a and José Berenguer3
T. thermophilus HB8 cells are surrounded by a hexagonal S-layer protein (molecular mass 100 000), named SlpA [97–100]. The SlpA protein frames regular aggregates of hexagonal, trigonal or tetragonal symmetries when recovered from cell envelopes by different treatments [101, 102]. Hexagonal and tetragonal arrays can be transformed into trigonal structures in situ under specific treatments, a trimeric interaction within SlpA subunits being the common motif that remains along this transition . Putatively, an SDS resistant ionic interaction that requires calcium could be responsible for the maintenance of this trimeric interaction . Interestingly, the structure of the tetragonal array was similar to that of bacterial porins, supporting the existence of an ancient phylogenetic relationship between these proteins .
In addition to the strong horizontal interactions within SlpA molecules, there is evidence supporting the existence of a specific binding to the underlying peptidoglycan layer. In fact, solubilization of the S-layer protein by neutral detergents requires the digestion of the peptidoglycan layer . Moreover, the deletion of an SHL domain  from the N-terminus of SlpA resulted in the inability to bind peptidoglycan fragments in a Far-Western .
5.1The expression of SlpA is a well regulated process
There is evidence supporting the existence of tight controls on the expression of the SlpA protein:
(i) There are no stocks of SlpA in the cell, nor can any secreted S-layer protein be detected in the culture media. Moreover, when slpA was cloned in T. thermophilus HB8 using a multicopy vector  there was no overexpression of SlpA.
(ii) GlmS, an essential enzyme in the synthesis of peptidoglycan , is encoded in a divergent transcriptional unit upstream of slpA . The presence of the genes encoding GlmS and SlpA clustered in the chromosome and transcribed from divergent promoters suggests the existence of mechanisms that coordinate their expression, as has been demonstrated in a number of cases of similar genetic architecture .
(iii) The deletion of slpA causes severe defects in cell growth and division , and leads to the overexpression of a regular array built up by the SlpM protein . The overexpression of slpM supports the existence of a genetic control in like manner to what has been described for bacterial porins . The identification of SlpM as a regulator of SlpA added further arguments to these relationships [110, 111].
(iv) A last argument comes from the structure and transcription of the PslpA promoter. The slpA gene is transcribed from a complex promoter which presents inverted repeated sequences separated by two complete helix turn distances within a bent DNA region . As is well known, these structures are usually found as the binding sites for proteins that inhibit or activate the transcription of the target gene . Furthermore, slpA is transcribed with a long leader mRNA which could play a role in its stability, similar to that described for other S-layer genes , and for the OmpA protein of E. coli. However, alternative roles in the control of expression of SlpA are also possible.
5.2Searching for transcriptional repressors in heterologous systems
The functionality of the PslpA in E. coli allowed us to develop a genetic system for the detection of putative repressor genes among an expression library of T. thermophilus HB8 DNA . In the system, we used a transcriptional fusion between the PslpA promoter and a reporter gene (lacZ). Specificity controls were carried out with other transcriptional fusions .
From around 30 000 clones checked, three genes whose expression in E. coli repressed the expression of the reporter gene were reproducibly cloned. One of the genes, termed slrA, was shown to encode a cytoplasmic protein. Surprisingly, the other two genes were identified as 5′ fragments of slpM and slpA [110, 111].
5.3SlrA is a transcriptional repressor of slpA
The slrA gene codes for a cytoplasmic protein (molecular mass 27 000) with no homologous proteins in the gene banks. Nevertheless, the overall properties of SlrA, especially its size and basic isoelectric point, were in agreement with a putative role as DNA binding protein. This possibility was clearly shown by South-Western blots, in which a labeled DNA fragment containing the PslpA promoter was used as probe [110, 111]. The ability of SlrA to bind specifically to the PslpA promoter suggests a role as transcription factor of the slpA gene for this protein.
More convincing evidence was further obtained by in vivo gene replacement . Under the optical microscope, slrA::kat mutants were indistinguishable from wild-type cells. However, lower growth rates were detected in liquid cultures, suggesting some defects in cell metabolism or division. Furthermore, cells started to grow much later than the wild-type when inocula from 2 day old cultures were used. This fact suggested the existence of problems in the survival mechanism in stationary phase. Further analysis will be required to identify which are the mechanisms affected.
Protein analysis of the slrA::kat mutants did not reveal the existence of differences in the amount or the size of S-layer protein synthesized. However, a parallel analysis of the slpA transcription displayed a clear increase in the amount of mRNA synthesized per cell, thus supporting that SlrA functions as a transcriptional repressor of the S-layer gene in vivo.
5.4Coregulation in the expression of regular proteins
The astonishing identification of the slpM gene as repressor of slpA in the E. coli genetic system added further arguments for the close relationship between the expression of both genes. As in the case of SlrA, we could confirm the specific interaction of SlpM with the PslpA promoter in vitro using South-Western blots .
Phenotypically, two kind of slpM::kat mutants were obtained: some of them were identical in cell morphology to wild-type cells (‘normal’ mutants), whilst others formed round multicellular bodies when cells reached the stationary phase (‘round’ mutants). These differences matched those in SlpA size. In ‘normal’slpM::kat mutants, the SlpA protein was also normal in size (molecular mass 100 000), but in ‘round’ mutants, the SlpA protein lacked a fragment from its C-terminus as a consequence of a gene recombination event. This latter result again reinforced the existence of a close relationship between the expressions of slpA and slpM.
Despite differences in size, the amount of SlpA protein in slpM::kat mutants was indistinguishable from that of wild-type cells, thereby supporting again the existence of additional overlapping control mechanisms. However, Northern blots showed that the transcription of slpA was diminished in both slpM::kat mutant types. Consequently, SlpM apparently functions as a transcriptional activator of the slpA gene.
The question of how a membrane protein could function inside the cell as a transcriptional factor could be answered on the basis of the presence of SlpM C-fragments in the soluble fraction after breakage of the cells.
5.5Does SlpA regulate its own translation?
The second intriguing result of the screening was the identification of C-fragments of SlpA as repressors of the expression of the PslpA-lacZ fusion. However, several South-Western blots revealed that the PslpA promoter did not bind to this SlpA fragment making doubtful this result of the screening.
One of the possible alternative pathways through which this SlpA C-fragment could block its own synthesis was a putative effect on the mRNA. North-Western experiments with a labeled mRNA probe demonstrated that the leader region of the slpA mRNA was specifically bound by the C-fragment of the SlpA protein . This result suggested that SlpA could control its own translation through a direct binding to its mRNA, a mechanism frequently found for ribosome proteins.
5.6Conclusions and perspectives
A model and several questions arise from our results. First of all, the ability of the S-layer to bind its own mRNA supports the existence of a mechanism for the autoregulation of its translation (Fig. 11). Such a mechanism could explain how the cell keeps a constant amount of the S-layer protein in the presence of different quantities of S-layer mRNA. In this sense, it is noteworthy that many other S-layer genes are also synthesized with long leader mRNA sequences which have been proposed to increase its half-life . Despite this, the cells are able to rapidly adjust the synthesis of S-layer to external conditions, thus supporting the existence of mechanisms to prevent its oversynthesis from an excess of ‘stable’ mRNA. The ability of SlpA to bind the slpA mRNA leads us to propose the existence of translational controls over the expression of the S-layer of T. thermophilus, which may be applied to other bacteria.
In addition to a probable control of its own translation, the transcription of slpA seems to be controlled by two other proteins. The identification of SlpM as a PslpA binding protein and the analysis of slpM::kat mutants suggested that this protein is required to reach a normal transcriptional level of the S-layer gene (i.e. a transcriptional activator). Likewise, the inactivation of slpA results in the overexpression of SlpM . Thus, a regulatory circuit seems to control the SlpA/SlpM ratio in the cells: SlpA represses the synthesis of SlpM through a still unknown mechanism (?, Fig. 11), and SlpM activates the transcription of slpA (+, Fig. 11). Further work will be required to analyze this circuit.
The SlrA protein functions as a classic transcriptional repressor (−, Fig. 11). However, nothing is known about the signals to which the repressor responds (?, Fig. 11). It is possible that SlrA is connected to other metabolic pathways of the cell, like the cell wall synthesis. Future in vitro work with purified SlrA protein could help us to understand the physiological role of the SlrA protein.
All these results clearly show that the cell has developed complex and overlapping control mechanisms to synthesize the S-layer according to specific physiological requirements, and in coordination with other elements of the cell envelope. Analysis in other systems will determine if this complex regulation seen in T. thermophilus HB8 is an exception, or by contrast a general rule for the control of S-layer synthesis.
6Molecular genetics of variation of S-layer proteins of Campylobacter fetus
Martin J. Blaser
Campylobacter fetus, a Gram-negative, microaerophilic eubacterium, persists on mucosal surfaces in a number of mammals, birds, and reptiles. Its S-layer gives it protection from the alternative pathway of complement, and permits the evasion of antibodies via antigenic variation [20, 115–118]. Thus, the S-layer is a critical virulence factor for this organism. C. fetus strains have been divided into type A and type B, based on their lipopolysaccharide (LPS) characteristics; cells of each type are covered by either one of two parallel families of S-layer proteins (SLPs) that range in molecular mass from 97 000 to 149 000, and form hexagonal or tetragonal crystals [115, 119].
6.1Structure of the sapA homologues
The SLPs of the type A strains are encoded by sapA and seven or eight sapA homologues, each of which encodes a full length SLP open reading frame (ORF) [49, 120, 121]. Each homologue is conserved at the 5′ end [49, 120, 121] and is unique at the 3′ end. The middle region is semiconserved among the homologues . A nearly identical pattern is found for the sapB homologues present in type B strains . The major difference between sapA and sapB is the 5′ conserved region which although substantially different in primary sequence encodes deduced peptides that are similar in secondary structure. These regions are responsible for the type-specific LPS binding of the SLP. Interestingly, sapA and sapB are identical except for the nucleotides encoding the first 190 amino acids of the protein. PCR and hybridization data indicate that sapA1 and sapB1, and sapA2 and sapB2 are homologous to one another as well. These data suggest that such similarities will be present for each of the sapA and sapB homologues .
In addition, non-coding regions both upstream and downstream of the ORFs are identical . This strong conservation of non-coding sequences implies important functions for these regions as well. The conserved features upstream of the ORF include a putative RecBCD recognition (Chi) site, a CTTTT pentamer repeated three times and inverted repeats encompassing the ribosome binding site just upstream of the translation start site . These features suggest importance of the region in S-layer transcriptional and translational regulation of S-layer protein synthesis.
6.2Physical arrangement of the sapA homologues
By pulse field gel electrophoresis and Southern hybridization using the conserved region of the sapA homologues as a probe, it has become clear that all the homologues are clustered in a small part of the genome representing less than 8% of the total . Initial studies indicated that several of the homologues are arranged in tandem, one after another. After each homologue is an inverted repeat sequence that could serve as a ρ-independent transcription terminator, and then a 30–50 bp region that is highly conserved in all homologues studied [49, 120, 121]. However, more recent work has clarified the physical relationship of the homologues to one another . At least one homologue is oriented oppositely to the others and is separated by a 6.2 kb region that does not contain any homologues.
6.3Rearrangement of the sapA homologues
Previous work showed that sapA homologues can rearrange via reciprocal recombination . This is a conservative process, in which cassettes are shuffled but never lost, nor are new cassettes being created, as occurs with gene replacement. We have now shown that SLP expression is governed by a single promoter in which silent gene cassettes are rearranged downstream of this unique promoter . We have found that the promoter is present in the 6.2 kb region flanked by sapA homologues facing in opposite orientations. Using genetic techniques, we have shown that this 6.2 kb region is able to invert, with the consequence that the unique promoter inverts as well, permitting expression of one or the other flanking sapA homologue. The presence of inverted repeats just upstream of the ORFs is consistent with an inversion mechanism because invertases often use such repeats as recognition sites. Each of the homologues possesses such an upstream repeat, a structural feature that could facilitate a general rearrangement process other than a simple switching back and forth only involving the two flanking cassettes. From a teleologic standpoint, why would C. fetus maintain eight highly conserved cassettes from strain to strain, if it was using only two for SLP expression? Recent work has shown that not only can the promoter invert but the gene cassettes can be part of the inversion process, so that SLP expression is not limited to the two flanking cassettes . This appears to be a unique model for invertible DNA elements, and it will be of interest to determine whether such a system governs SLP expression for other bacteria.
The sapA homologue conservation, tractability, and strong phenotypic selection make C. fetus an excellent model to examine the genetics of SLP rearrangements, and also to explore structure-function relationships. The development of animal models of infection [118, 123] further accentuates the utility of this system for biological analysis.
7S-Layer variation in Bacillus stearothermophilus
Holger Scholz3, Beatrix Kuen, Werner Lubitz and Margit Sára
Variation of surface exposed proteins has been reported for many different microorganisms . However, most of them have been described for pathogenic bacteria as a strategy to escape the destruction by the immune system of the infected host. Bacillus stearothermophilus PV72 represents a strictly aerobic, non-pathogenic, thermophilic, S-layer carrying organism, which was isolated from a Slovenian beet sugar factory . The organism can use glucose and maltose as carbon sources [124–126], has a growth optimum of 57°C and was cultivated on complex SVIII medium  before the synthetic PVIII medium was developed by applying the pulse and shift technique in continuous cultures . The S-layer of the wild-type B. stearothermophilus PV72 shows hexagonal symmetry with a center-to-center spacing of the morphological subunits of 22.5 nm and is composed of identical protein subunits with molecular masses of 130 000 each . The gene sbsA encoding the S-layer protein has been cloned and sequenced by Kuen et al. [129, 130]. The S-layer subunits consist of 1198 amino acids and possesses a signal peptide of 29 amino acids. Processing of the S-layer protein occurs before the subunits appear in the peptidoglycan containing layer in which an S-layer protein pool is stored to ensure that the extending cell surface is completely covered with S-layer subunits [131, 132]. When B. stearothermophilus PV72 was cultivated on synthetic PVIII medium at moderate aeration at a dissolved oxygen concentration (DO) of 20–30%, no change of the S-layer was observed. By changing these conditions, however, variation of the S-layer was induced. The amount of S-layer protein synthesized as well as the formation of an S-layer protein pool could be influenced by varying the specific growth rate during continuous cultivation . Feeding of an amino acid mixture consisting of Gly, Ala, Val, Leu, Ile, Glu, Asp, Gln and Asn to the continuous culture significantly stimulated S-layer protein synthesis, whereas the addition of aromatic or basic amino acids led to an irreversible loss of the ability to express the sbsA gene. An S-layer deficient variant, designated T5, could be isolated from batch culture when the wild-type PV72 was grown for at least 10 passages, inoculated in 24 h intervals, at 68°C instead of 57°C [61, 134]. As shown by SDS-PAGE and electron microscopy, the amount of SbsA was significantly reduced after 2–3 passages. However, by lowering the growth temperature back to 57°C, full SbsA expression was retained within the following passage. As shown by hybridization and PCR analyses, the variant T5 exhibits an intact sbsA coding region. However, a DNA rearrangement has occurred within the upstream region of sbsA which probably causes the S-layer negative phenotype. Whether the layer deficient variant induced by adding aromatic or basic amino acids to the medium and the variant arising after temperature upshift have identical changes on DNA level has to be investigated. When the oxygen supply was increased during continuous cultivation, a variant strain, B. stearothermophilus PV72/p2, occurred which was covered by an altered protein (SbsB), forming a p2 ordered surface layer. Protein and sequence analyses revealed that SbsA and SbsB are two different proteins encoded by different genes. The exact mechanism which is used to switch from SbsA to SbsB expression remains to be elucidated. The results of our investigation suggest, however, that multiple recombination events are involved in the switch from SbsA to SbsB expression. During the course of our genetic analysis it could be shown that the wild-type PV72/p6, the variant PV72/p2 and the S-layer deficient variant T5 also harbor huge plasmids which differ in size. Hybridization analysis with specific probes using plasmid DNA from all three strains as templates clearly demonstrated that sbsA homologues are located on the plasmid of the wild-type and the variant T5. However, sbsB homologues could not be detected on any plasmid. Whether recombination between plasmid located sbsA homologues and chromosomal DNA occurs is not clear yet.
7.1Investigation of sbsA in the S-layer deficient variant T5
Our initial goal was to determine whether the sbsA coding region is present in the S− variant T5 as well. For this purpose hybridization experiments with various specific probes against HindIII digested total DNA of the wild-type and the variant T5 were performed. When probes derived from the 3′ end of the gene or from internal regions were used, no difference between the two strains could be detected (Fig. 12A). This observation indicated that the variant which has no layer protein expression harbors an intact sbsA coding region. However, when a probe specific for the very 5′ region of sbsA was used, a signal reduced in size of about 1000 bp was obtained for the variant T5, compared to that of the wild-type (Fig. 12B). To determine whether a deletion or a DNA rearrangement had occurred within the sbsA upstream region, both upstream regions were cloned and sequenced. Cloning of the wild-type sbsA upstream region was hindered by the occurrence of spontaneous deletions. One possible explanation for this phenomenon was that the sbsA promoter is recognized in E. coli, leading to read-through transcripts which might be toxic for the cell. To overcome this problem the promoter probing vector pKK232-8 was used. This vector allows the cloning of strong promoters 5′ to a promoterless CAT reporter gene. Selected positive clones were able to grow on a chloramphenicol concentration of 200 μg/ml, indicating that the sbsA upstream region has a strong promoter activity in E. coli. Sequencing as well as PCR analysis revealed that the size reduced signal of the variant is due to a DNA rearrangement but not to a deletion. The alignment of both sbsA upstream regions showed that they were identical up to position −188 5′ of the ATG start codon, followed by a different nucleotide sequence (Fig. 13). Since the variant T5 has lost the ability to express SbsA, it can be assumed that the sbsA promoter is located further upstream of this position. Characterization of the sbsA promoter region of the wild-type will give further insight into the regulation of the gene.
7.2Oxygen-triggered change in S-layer protein synthesis and isolation of the p2 variant
7.2.1Physiological and morphological changes during variant formation
Continuous cultivation of the wild-type strain on complex medium under glucose and oxygen double limitation (glucose in spent medium <0.1 g/l; DO=0%) or on synthetic PVIII medium when the DO was controlled at 20–30% resulted in a stable synthesis of SbsA. Variation of these conditions (complex medium: DO>0%; synthetic PVIII medium=50%) led to a oxygen triggered synchronized change in S-layer protein synthesis and variant formation [133, 135]. In the following, the typical time course of variant formation on synthetic PVIII medium at a DO of 50%, T=57°C and a dilution rate of 0.3/h is described. As shown in Fig. 14, an apparent steady state concerning the respiratory activity (stirrer speed for constant DO; CO2 emission rate; Fig. 14A), the redox potential (Fig. 14B), the OD and biomass dry weight (Fig. 14C) was observed 2 volume exchanges after starting continuous cultivation representing 10 h after inoculation. The only on-line measured signal which showed a steady increase during this first stage of continuous cultivation was the culture fluorescence determining the intracellular NADH2 level. After 2–3 further volume exchanges (16 h after inoculation) the redox potential and the OD showed a significant increase while the biomass dry weight stayed constant (Fig. 14B,C). At this time point the respiratory activity increased to a maximum (Fig. 14A). The lack of correlation between the OD and the biomass dry weight can be explained by changes in the morphology of the cells, the chain length and a slight sporulation observed during variant formation . Biomass sample analysis by SDS-PAGE revealed that 15–16 h after inoculation, a faint protein band with an apparent molecular mass of 97 000 could be detected on SDS gels. In the course of variant formation the intensity of this protein band became more pronounced while the intensity of the wild-type S-layer protein showed a significant decrease (Fig. 15). As derived from densitometric evaluation of SDS gels, the decrease in the wild-type S-layer protein followed the equation x=xo·e−Dt which reflected the theoretical wash-out of a substance in a continuously stirred tank bioreactor [133, 135]. Since the biomass dry weight stayed constant during variant formation, it was evident that change in S-layer protein synthesis was a synchronous process in most, if not in all individual cells of the culture. As shown by continuous cultivation experiments carried out at different DOs, S-layer variation was triggered by increased oxygen concentrations in the culture [133, 136]. Electron microscopic examination of freeze-etched preparations of biomass samples harvested during variant formation revealed that as soon as the S-layer protein band (molecular mass 97 000) could be detected on SDS gels, the hexagonally ordered S-layer lattice of the wild-type strain (Fig. 16A) became interrupted by amorphous patches. By immunogold labelling, these amorphous patches were found to consist of both types of S-layer proteins which in a mixture were not capable of assembling into a crystal lattice. On a few cells, the hexagonally ordered S-layer lattice of the wild-type strain and the oblique lattice of the variant were arranged as monomolecular layer (Fig. 16B). With increasing intensity of the S-layer protein band (molecular mass 97 000), extended patches with oblique lattice symmetry could be observed (Fig. 16C) which finally completely covered the cell surface (Fig. 16D). The variant strain with the oblique S-layer lattice could be isolated from continuous culture and showed stable growth on complex and synthetic medium, under both oxygen limited and non-oxygen limited conditions. The oblique S-layer lattice of the variant strain was composed of non-glycosylated subunits with a molecular mass of 97 000. The S-layer protein of the wild-type strain PV72 and the p2 variant yielded different cleavage products upon peptide mapping with endoproteinase Glu-C, trypsin or chymotrypsin . Polyclonal antiserum raised against either SbsA or SbsB showed no cross-reaction with the other type of S-layer protein , indicating that SbsA and SbsB were two different proteins.
7.2.2Cloning, sequencing, and expression of sbsA and sbsB
To answer the question whether sbsA and sbsB are two different genes encoding different proteins, they have both been cloned, sequenced and stably expressed in E. coli[129, 130, 138]. The alignment of both sequences revealed an overall low similarity of 48% at the DNA level and only 20% at the protein level . Furthermore no significant stretches of sequence identity could be detected. However, an identical putative transcriptional termination signal is located 50 bp downstream of the TAA stop codon of both genes. The sbsA and sbsB upstream regions are different despite a stretch of 36 bp with a sequence identity of 90% which has high homology to several promoter structures of the genus Bacillus (data not shown). These findings indicate that sbsA and sbsB are two different genes and in addition that sbsA is probably not directly involved in recombination events to generate the sbsB coding sequence. The expression of SbsA and SbsB in E. coli led to the accumulation of stable, recrystallized sheet-like structures in the cytoplasm. These self-assembly products of SbsA and SbsB were arranged in parallel fashion at constant distance to each other following the curvature of the cylindrical part of the cells [130, 138]. As observed for SbsA, SbsB was not translocated through the cell envelope of E. coli[130, 138].
7.2.3Detection of sbsA in the p2 variant
Like the S-layer deficient variant T5, the p2 variant does not express SbsA. To determine whether the same DNA rearrangements within the sbsA upstream region had occurred leading to the loss of the ability to express this protein, PCR analysis with sbsA specific primers was carried out. In contrast to the variant T5, whose S-layer deficient phenotype is caused by DNA rearrangements within the sbsA upstream region, but not by an altered sbsA coding sequence, the sbsA gene in the p2 variant was modified. Only N- and C-terminal regions of the sbsA gene could be amplified with the correct size derived from the sbsA sequence. When internal parts of sbsA were amplified, larger or even multiple PCR products were obtained. The disruption of sbsA in the p2 variant might be responsible for the irreversibility to express this gene.
7.2.4Detection of the second S-layer gene sbsB in the wild-type strain PV72
To detect the second S-layer gene sbsB in the wild-type PV72, PCR analysis was performed with various sbsB specific primers, using wild-type total DNA as template. The length of the PCR products differed from the size calculated from the known sbsB sequence (data not shown). The PCR product obtained from the reaction using total DNA from wild-type strain PV72 and primers homologous to the 5′ and 3′ end of sbsB had a size of 1300 bp, which was 1400 bp smaller than calculated for the full length product from the p2 variant. In contrast to this smaller product, the amplification of internal regions of sbsB using wild-type DNA as template revealed larger PCR products than expected. These observations indicate that sbsB is generated during variant formation by recombination events. Furthermore, the results of various primer combinations gave evidence that sbsB is separated into more than two fragments before it is expressed. To follow the switch from sbsA to sbsB, PCR analysis with DNA isolated from samples taken at different time points during variant formation was performed. Sample 1 was taken before the beginning of SbsB formation, whereas sample 2 corresponded to the stage where the ratio SbsA/SbsB is 1:1 as determined by SDS-PAGE (Fig. 15). The appearance of an additional p2 specific band in sample 2 and the absence of this band in the wild-type as well as in sample 1 demonstrates that sbsB has to be generated during the switch from SbsA to SbsB expression (Fig. 17). These results were confirmed by hybridization analysis (data not shown). The cloning and sequencing of the PCR products obtained with sbsB specific primers and wild-type DNA as template will give further insight into the process generating the sbsB gene.
7.2.5Detection of sbsB in the S-layer deficient variant T5
Neither SbsA nor SbsB is expressed by the variant T5 when it is grown under oxygen limited conditions at 68°C. However, this variant can be induced to a reversible expression of SbsA when the growth temperature is lowered back to 57°C for several passages. Furthermore, this variant expresses SbsB when oxygen limitation is abolished. In contrast to the p2 variant which also does not express SbsA, the variant T5 exhibits an entire sbsA coding region which might be an explanation for the reversible SbsA expression. The presence and the organization of the sbsB gene in the S-variant T5 was analyzed by PCR experiments and hybridization analysis. The results of these experiments revealed that the variant strain exhibits, beside an intact sbsA coding region, an entire sbsB coding region as well. Furthermore, T5 exhibits the altered sbsB sequence observed for the wild-type PV72 (Fig. 18). The PCR product of the sbsB upstream region derived from the sbsB gene of variant T5 differs in size from the fragment obtained using p2 DNA as template. This indicates that the SbsB as well as the SbsA negative phenotype of the variant T5 is the result of changes in the upstream regions, but is not due to reconstructions of the coding regions. The observation that T5 exhibits an entire sbsB coding region as well as an intact sbsA coding region may explain why the variant has the possibility to express either SbsA or SbsB. Further investigation will show whether the variant T5 is an intermediate stage to SbsB expression.
7.3Identification of plasmid located sbsA homologues in B. stearothermophilus PV72 and the variant T5
Isolation of plasmid DNA from strains PV72 and T5 revealed that the wild-type as well as the variant strain harbor huge plasmids. The molecular mass of the plasmids was determined from the size of the DNA fragments obtained by restriction analysis. The calculated size of the wild-type plasmid (pBHp6) is about 80 kb, whereas the molecular mass of the plasmid (pBHT5) derived from the variant T5 has a size of approximately 50 kb. To determine whether the reduction in size contributes to the SbsA negative phenotype, hybridizations with sbsA specific probes against undigested and digested plasmid DNA of both strains were performed. N-terminal sbsA specific probes did not hybridize to plasmid DNA of both strains, whereas probes directed to the C-terminal sbsA region hybridized to multiple bands present in the plasmid DNA of both the wild-type and the variant T5 (Fig. 19A,B). Although the restriction patterns of pBHp6 and pBHT5 were different, the C-terminal probe only hybridized to common bands. Since hybridization occurs only to bands common in size of both plasmids and additionally the restriction patterns of the plasmid from the variant T5 showed only the loss of bands but no additional restriction fragments in comparison to the wild-type it can be assumed that PV72 harbors two plasmids of which one is lost in the variant strain T5. As no full length sbsA gene is localized at the plasmids, it can be assumed that the expression of SbsA occurs from the chromosome but not from the plasmid. Whether recombination events between chromosomal and sbsA homologues located on the plasmids take place has to be investigated.
The mechanisms underlying S-layer variation so far include single recombination events with conserved and variable gene segments , the inversion of whole gene cassettes  or the inversion of promoters . In each case the expressed and the silent gene – or homologues of the gene – share high sequence identity within defined conserved regions. The S-layer genes sbsA and sbsB of B. stearothermophilus, however, have no significant identity. Therefore it can be assumed that sbsA is not directly involved in the generation of sbsB. The results of PCR analysis suggest that sbsB is separated into more than two fragments before it is expressed. On the other hand sbsA is separated into fragments after the switch from SbsA to SbsB expression. These observations indicate that multiple DNA rearrangements are involved in the switch from SbsA to SbsB expression. Interestingly the variant strain T5, which has neither SbsA nor SbsB expression, exhibits the full length sbsA and sbsB coding region. Since this variant can be induced to express either SbsA or SbsB it probably represents an intermediate state of the variant formation. The switch from SbsA to SbsB expression can be followed by PCR where the appearance of the sbsB specific band occurs after induction of variant formation. PCR analysis and sequencing of the PCR products obtained with samples taken at short time intervals will give further insight into the mechanism underlying recombination. Plasmid located genes expressing other surface proteins than S-layers have been described previously [140, 141]. To our knowledge, this is the first report of the presence of plasmid located S-layer sequences. All our findings together suggest a complex mechanism underlying the switch from SbsA to SbsB expression, whose exact course has to be elucidated in further experiments.
8S-layer protein genes of Lactobacillus
Peter H. Pouwels3, Carin P.A.M. Kolen and Hein J. Boot
Lactic acid bacteria are widespread in nature and are generally used in the production and preservation of food and feed products like cheese, meat, yoghurt and silage . Some strains of Lactobacillus are believed to also display health promoting activities for humans and animals when present in the gastrointestinal or female urogenital tract. Several effects have been reported to be associated with the presence of lactobacilli in the gastrointestinal tract, e.g. stimulation of immunoglobulin production , induction of interferon expression in macrophages , hypocholesterolemic effects , and the prevention of pathogenic bacteria like Salmonella typhimurium and Neisseria gonorrhoeae to epithelial cells . These so-called probiotic properties and the potential of lactobacilli as vehicles to target compounds of interest, e.g. antigens or immunomodulators, to the mucosa have stimulated research on the role of surface proteins in adherence and adjuvanticity.
8.1Relationship between the presence of an S-layer and adherence
Some Lactobacillus strains can colonize the gastrointestinal tract of humans and animals. These strains thus must possess the ability to adhere to the exposed surface of the epithelial cells. Adherence of colonizing lactobacilli is mediated by fimbrial adhesins, which interact with the epithelial cells . The presence of an S-layer on the outside of strains of different Lactobacillus species has been reported by several research groups [148–150]. S-layer proteins of lactobacilli have a molecular mass between 40 000 and 55 000 and are, in general, non-glycosylated. The role of the S-layer protein in the interaction of lactobacilli and the epithelial cells of the host is unclear, as several Lactobacillus species which either do or do not contain an S-layer have been recovered from the gastrointestinal and female urogenital tracts of mammalian hosts. The type strain of L. acidophilus binds specifically to intestinal lectins from chicken, although it is not clear whether the S-layer is involved in this binding . Schneitz et al. reported that the S-layer of L. acidophilus strains is involved in the adhesion of these bacteria to avian intestinal epithelial cells . Chemical removal of the S-layers of L. acidophilus, however, does not affect the adhesion to Caco-2 cells .
8.2S-layer protein of L. acidophilus
To investigate the role of the S-layer or S-layer protein of lactobacilli in adhesion to epithelial tissues, we have started a research project aimed at the characterization of the S-protein of L. acidophilus and its mode of expression. The L. acidophilus ATCC 4356 type strain, which originates from human pharynx, was chosen because of its presumed adhering properties and because it had been shown to contain an S-layer. The amino acid composition of purified S-layer protein L. acidophilus is typical for S-layer proteins, i.e. a relative abundance of threonine, serine and hydrophobic amino acids, and absence of cysteine and methionine residues . The molecular mass determined by electrospray ionization mass spectroscopy yielded a value of 43 639±6. This result taken together with that of N-terminal sequence determination of the purified, mature protein and nucleotide sequence analysis (see below) allowed the conclusion that the S-layer protein of L. acidophilus is not glycosylated . The isoelectric point of the L. acidophilus S-protein, calculated from the deduced amino acid sequence, is 9.4. The S-proteins of L. brevis and L. helveticus display similarly high values. These high values are in contrast with those of S-layer protein genes of other species which are in general negatively charged (pI 3–4).
8.3S-layer protein genes of L. acidophilus
The nucleotide sequence of the cloned S-protein gene showed 79% similarity to that of the L. helveticus slpA gene (GenBank X91199) but very little if any similarity to the L. brevis slpA gene . Also the deduced amino acid sequences of the L. acidophilus and L. helveticus S-layer proteins are very similar (75% identical amino acids). The closer similarity of the S-proteins of L. acidophilus and L. helveticus compared to L. brevis is in agreement with the closer evolutionary relationship. The S-layer protein gene of L. acidophilus, like that of L. helveticus and L. brevis, encodes a pre-protein comprising a signal sequence with a predicted cleavage site after amino acid 24. Several direct repeats have been found in the DNA sequence of the slp genes of L. acidophilus and L. brevis which result in amino acid repeats. However, no experimental data exist supporting the suggested structural function of the amino acid repeats . S-proteins are expressed at a high level. As for many other species, a biased codon usage is also observed for Lactobacillus mRNAs that are translated at a high rate . Over 50% of the S-protein of these organisms are encoded by seven triplets only. Random rearrangement of the S-protein encoding triplets of L. acidophilus and L. brevis yields about the same number of direct repeats, again resulting in amino acid residue repeats, arguing against a functional role for these repeats. All L. acidophilus strains investigated contain two slp genes, since two hybridizing bands of the same size are found, when the slpA gene is used to probe the chromosome . The nucleotide sequence of the second gene (slpB) is highly similar to that of slpA. The 5′ untranslated region and the sequence encoding the signal sequence up to the start of the mature S-layer protein are highly similar. The same holds for the 3′ region of the two genes but the middle regions differ . The two proteins encoded by the slp genes are also very similar. The C-terminal one-third regions are the same, except for 1 amino acid substitution, while the N-terminal and middle parts of the proteins are different. The conclusion that slpA encodes the S-layer protein whereas slpB is a silent gene is based on the following considerations. (i) slpA rather than slpB was isolated from an expression library probed with antibodies against the S-protein, (ii) the amino acid sequences of the N-terminal region and of tryptic peptides of the S-layer protein correspond to that of slpA, (iii) RNA is transcribed from splA but not from slpB, and (iv) a promoter sequence is present before the slpA gene but is lacking before slpB .
8.4Genetic organization of S-layer protein region of L. acidophilus
By a combination of restriction enzyme analysis and PCR experiments slpA and slpB were found to be located, facing each other, on a 6 kb DNA fragment [59, 113]. The nucleotide sequence of the region between the two slp genes has been determined. Four ORFs (>150 nt) are present in the region (3.0 kb) between slpA and slpB (Fig. 20). No transcriptional regulatory sequences (e.g. promoter, operator, terminator) could be detected in this region. Preliminary results indicate that ORF-1 and ORF-2 are transcribed at a very low level, if at all, suggesting that ORF-1 and ORF-2 are part of a silent operon, together with slpB. The amino acid sequence encoded by ORF-1 shows considerable homology with IcaI of Staphylococcus epidermis and with HmsR of Yersinia pestis. IcaI has been implicated in the formation of the polysaccharide intercellular adhesin PIA of Staphylococcus epidermidis. HmsR was postulated to be a transacting, positive regulator for expression of the hemin storage locus, hms. The hms locus has been shown to be involved in transmission of Yersinia by fleas. Interestingly, expression of the hms genes is accompanied by a cell surface change of Yersinia, facilitating autoaggregation and blockage of the foregut of fleas . The blockage is dependent on adherence to a specific region of the fleas' gut. It is therefore tempting to assume that the silent SB operon of L. acidophilus may be involved in adhesion and/or heme binding. A homology search of the protein sequence encoded by ORF-2 yielded no significant homologies. Despite this lack of general homology, we found in this protein a region of amino acid residues which has a high degree of similarity with the active site of proteins belonging to the family of Din invertases (see below). The putative protein (79 amino acids) encoded by ORF-3 shows for the 3/5 C-terminal part homology with an ATP binding transporter protein, whose gene has been identified in the genome of Mycoplasma genitalium. For Aeromonas salmonicida it has been shown that an ATP binding domain was present in a protein which is involved in regulation of S-layer protein expression. This protein (AbcA) acts, in a heterologous system, as a positive regulator of one of the two promoters present in front of the S-protein encoding gene . No homology with a known protein was found for ORF-4.
8.5Occurrence of two slp genes in other lactobacilli
Previously it had been observed that strains of L. acidophilus group A (L. acidophilus, L. crispatus, L. amylovorus and L. gallinarum) possess an S-layer, whereas group B (L. gasseri and L. johnsonii) strains lack such a structure . We have confirmed these results and have shown that, in contrast to some reports, L. bulgaricus and L. fermentum strains do not harbor S-layer proteins. Antibodies raised against L. acidophilus ATCC 4356 strongly cross-react with S-layer proteins from L. amylovorus and L. helveticus, react weakly with S-layer protein from L. gallinarum, but do not cross-react with S-layer protein from L. crispatus. Apparently, all L. acidophilus group A bacteria, except L. crispatus, have epitopes in common. The observation that L. crispatus S-layer protein does not cross-react with L. acidophilus antibodies, while other group A bacterial S-proteins and even L. helveticus S-layer protein do, is striking as L. crispatus is evolutionarily more closely related to L. acidophilus than L. helveticus. Moreover, the C-terminal regions of the L. helveticus and L. crispatus S-layer proteins are almost identical to that of the L. acidophilus S-layer protein. A possible explanation might be that the C-terminal part is buried inside the S-protein or is much less immunogenic than the less conserved N-terminal regions.
By using probes that are specific for different regions of the slp genes as well as probes that can differentiate between slpA and slpB, we could establish that all A group bacteria contain two slp genes with a structure very similar to that of the L. acidophilus slp genes. L. acidophilus group B bacteria and L. bulgaricus lacked slp genes, whereas L. helveticus contained a second gene showing strong homology with the L. acidophilus genes at the 3′ end only . Apparently, L. helveticus contains a truncated gene lacking the 5′ region or harbors a second gene with a non-homologous 5′ region. A silent S-layer protein gene which lacks the 5′ part of the coding region has also been described for Bacillus sphaericus. The strong conservation of both 5′ and 3′ regions among different lactobacilli living in different environments suggests that these regions have important functions. The C-terminal region encoded by the 3′ end might for instance be involved in interaction of S-proteins with the peptidoglycan layer. Peptidoglycan binding regions, also called SLH regions, have been found at either the N-terminal or the C-terminal end of several S-layer proteins . Conservation of the 5′ untranslated region probably has to be explained by constraints imposed on the nucleotide sequence regarding mRNA stabilization and as a target site for recombination (see below). Differences in cross-reactivity with antibodies between group A bacteria and the complete lack of reaction of group B bacteria offer attractive possibilities for rapid identification of these organisms. An even more powerful method can be devised for identification of these species which is based on differences in the structure of the slp genes. The presence of nearly identical 5′ and 3′ regions interspaced by more variable regions offers the opportunity to unambiguously differentiate between different isolates by PCR. Once the variable regions have been (partly) sequenced specific PCR primers can be designed to amplify specific DNA fragments. These rapid and cheap methods could replace the rather time consuming and thus costly methods currently in use for strain identification.
8.6Regulation of expression of slpA gene
As in many other bacteria, S-layer proteins of Lactobacillus are very efficiently expressed and secreted. Expression and/or secretion are probably tightly controlled. Bacteria are completely covered with S-layer protein indicating that S-layer protein expression and secretion keep pace with bacterial growth. The efficient expression is most probably due to a combination of expression signals that allow high rates of transcription, translation and secretion. The promoter of the L. acidophilus slpA gene is two times more efficient than that of the lactate dehydrogenase gene (ldh) of L. casei, considered to be one of the strongest promoters in many bacteria . A high rate of translation of slp genes is achieved not only because of the use of a biased set of codons but also because slp mRNA is highly stable (L. acidophilus mRNA, half-life 15 min) . The 5′ untranslated leader of L. acidophilus mRNA can fold into a stem-loop structure with an energy of −191 kJ/mol, which most likely will protect mRNA from 5′-3′ degradation. The role of secondary structure of the 5′ untranslated region in regulation of gene expression was established by introduction of a deletion which removed approximately one half of the untranslated region. The inability to form a stem-loop structure resulted in a 2-fold reduction of the level of gene expression . Two promoters are present in the region upstream of the slpA gene of L. acidophilus yet only the promoter closest to the slpA gene is being used under all growth conditions tested . In L. brevis both promoters are equally efficiently used during the exponential phase of growth . The sequence of the upstream promoter of L. acidophilus (P2) is the same as that of the downstream located promoter (P1) but the spacing between the −35 and −10 regions is 20 nucleotides, whereas for P1 the optimal spacing is 17 nucleotides. Although P2 does not conform to the optimal promoter sequence, promoters with a spacing of 20 nucleotides have been described before . The suboptimal spacing may reflect the need of an activator protein that binds to a nearby site in the DNA and facilitates the entry of RNA polymerase. A sequence of dyad symmetry, which is located immediately upstream of the −35 region of P-2, might play a role in binding of such a hypothetical activator. Regulation involving a transacting factor (AbcA) and a potential inverted DNA sequence near one of the promoters of an S-layer protein gene has recently been described for Aeromonas salmonicida.
8.7Expression of L. acidophilus S-layer protein in a heterologous host
To study the role of the S-layer in the interaction of probiotic lactobacilli with receptors of the epithelial tissue of the host, we expressed the SA protein gene of L. acidophilus in L. casei, a bacterium which does not have an S-layer. Transformed L. casei bacteria produced and secreted the SA protein efficiently. However, secreted SA protein does not interact with the cell wall and can therefore not form an S-layer on the outside of this bacterium . Our findings corroborate the results of earlier studies showing that in vitro reconstitution of S-layers is only possible on the cell wall of bacteria that normally carry an S-layer . A comparison of the structure of the cell walls of L. acidophilus and L. casei may reveal which essential component is missing in the latter organism.
8.8Antigenic variation in L. acidophilus?
The presence of regions of near-identity at the 5′ and 3′ regions of L. acidophilus and of related strains suggests that these regions may be involved in genetic recombination. By a series of PCR experiments we have shown that inversion of the slp segment takes place, replacing the silent gene slpB by the active one and vice versa, in 0.3% of the bacteria . Inversion of the slp segment occurs through homologous recombination which takes place in the 5′ untranslated region of the S-layer protein genes. Interestingly, a stretch of 15 bp which has a high degree of similarity with the consensus recognition site for invertases of the Din family is present in the middle of this region, suggesting that inversion of the slp segment is catalyzed by a member of this family. A region in the hypothetical protein encoded by ORF-2 of the slp segment (Fig. 20) shows a striking conservation of amino acid residues which were shown to be involved in recognition of the DNA recognition site by the Hin invertase . Because of the similarity in structure of the slp genes in bacteria that are evolutionarily related, we assume that inversion, leading to the interchange of the expressed and silent S-protein genes, will also take place in these organisms. These data strongly suggest that L. acidophilus and related organisms show antigenic variation. What the role of antigenic variation might be is at present not known. The S-layer protein of L. crispatus, which is closely related to L. acidophilus, has recently been shown to be involved in adherence to the extracellular matrix proteins, suggesting a role of antigenic variation in adherence .
8.9Potential applications of lactobacilli as vehicles for antigen presentation
Lactobacillus strains have a number of properties which make them attractive candidates for oral vaccination purposes . Lactobacilli have been used for centuries in food and feed, and are considered to be safe organisms, this in contrast to other live vaccine carriers used so far (e.g. Salmonella, Escherichia coli, Vaccinia) which cannot be classified as safe. Furthermore, in contrast to lactobacilli, the latter type of carriers are themselves highly immunogenic, possibly preventing repetitive use of the carriers with the same or other antigens. Forming the outermost layer of bacteria and being present in vast quantities (up to 105 molecules per bacterium), the S-layer protein is, in principle, an attractive candidate for fusion with antigenic determinants. The possibility to genetically manipulate lactobacilli and the availability of multiple S-layer genes of Lactobacillus opens new avenues for research aimed at presenting foreign proteins (antigens, ScFv, enzymes) at the surface of these bacteria. Further studies on the mechanisms triggering antigenic variation should shed light on the role of S-proteins in adherence. Such knowledge will facilitate screening of new isolates with probiotic properties.
9Molecular biology of the Lactobacillus brevis S-layer gene (slpA) and utility of the slpA signals in heterologous protein secretion in lactic acid bacteria
The Lactobacillus brevis S-layer protein is the major protein of the cell with a molecular mass of 46 000 in SDS-PAGE. Genetic characterization of the slpA gene has revealed two adjacent promoters (P1, P2) with the conserved hexanucleotide −10 and −35 regions typical of prokaryotic promoters, the consensus ribosome binding site, the ATG start codon and a signal sequence encoding a region of 90 nucleotides. The structural slpA gene has a coding capacity for a mature S-layer protein (molecular mass 45 000) followed by a strong transcription terminator sequence downstream of two translation stop codons. According to the mRNA size and the 5′ end analyses the slpA gene forms a monocistronic transcriptional unit where both promoters are functional. Further characterization of in vivo expression of the L. brevis S-layer protein and determination of the usage of the two slpA promoters as a function of growth have been performed. The half-life of slpA transcripts has been shown to be 14 min.
Functionality of slpA expression and secretion signals in heterologous protein secretion has been studied by using the E. coliβ-lactamase (bla) as the reporter in a slpA based secretion cassette. Secretion studies performed with L. lactis, L. brevis, L. plantarum, L. gasseri and L. casei have shown that in all hosts tested the bla gene was expressed under the slpA signals and all detectable Bla activity was secreted into the culture medium. The production of Bla was mainly restricted to the exponential phase of growth. The highest yield of Bla was obtained with L. lactis and L. brevis. Without pH control, substantial degradation of Bla occurred during prolonged cultivation with all lactic acid bacteria tested. When growing L. lactis and L. brevis under pH control, the Bla activity could also be stabilized at the stationary phase. L. lactis produced up to 80 mg/l of Bla which represents the highest amount of a heterologous protein secreted by lactobacilli so far.
Among over 300 S-layer harboring eu- and archaebacterial species , also many Lactobacillus species are present, which are either widely used in food fermentations or found as part of the normal flora of humans and animals. The DNA sequence of the lactobacillar S-protein gene (slp) has been published only from L. brevis, L. acidophilus and L. helveticus. Functions of the S-layer proteins in Lactobacillus are unknown, but they can be expected to play an essential role since inactivation of these genes has repeatedly failed . Recently, the L. crispatus S-layer protein has been demonstrated to mediate adhesion to type IV collagen  and preliminary results indicate that also L. brevis S-layer protein can mediate binding to intestinal epithelial cells.
Since for average sized cells 5×105 S-layer subunits per cell generation have to be synthesized in order to cover the entire cell surface with the S-layer proteins , the expression of a slp gene and the secretion machinery of a S-layer harboring cell may be expected to be very efficient. These properties are obvious targets for utilization of S-layers in biotechnological applications. To study heterologous protein production in lactobacilli, the L. brevis slpA has been chosen as a model. L. brevis is a heterofermentative lactic acid bacterium commonly found in vegetable fermentations, sour dough, silage and in the intestine of humans and animals [171, 172].
In this review, the characterization of the L. brevis S-layer protein, gene, mRNA and in vivo expression are discussed along with the demonstration of heterologous protein secretion with the aid of the slpA signals.
9.1Characterization of the L. brevis S-layer protein and gene
In L. brevis (ATCC 8287/GRL1), the S-layer has previously been shown to consist of tetragonally arranged subunits, which are composed of a protein with a molecular mass of about 51 000 . The subunits can be dissociated from the cell wall, e.g. with guanidine hydrochloride, and they can be reassembled into a native-like array in vitro . From the intact L. brevis cells, boiled in Laemmli sample buffer and analyzed in SDS-PAGE, only one major band with an apparent molecular mass of 46 000 was detected . To confirm its identity as the S-layer protein of L. brevis, the cells were treated with an antiserum raised against the isolated protein (molecular mass 46 000) and analyzed by immunogold electron microscopy. The post embedding immunoelectron microscopy clearly showed that the protein (molecular mass 46 000) was heavily enriched in the outermost part of the cell wall of L. brevis cells . The antiserum also recognized a major protein (molecular mass 55 000) of L. buchneri (DSM 20557) released from the cell similarly to that of L. brevis, thus indicating antigenic relatedness of these two surface proteins (molecular masses of 46 000 and 55 000) .
The N-terminal sequence of the intact S-layer protein was determined to be NH2-Lys-Ser-Tyr-Ala-Thr-Ala-Gly-Ala-Tyr-Ser. N-terminal analyses were also performed for a few tryptic peptides of the S-layer protein and the amino acid sequence information was used to design degenerated oligonucleotides for the isolation of the slpA gene .
DNA sequencing of the slpA gene from L. brevis was performed with PCR fragments since the gene was unstably maintained in L. brevis gene libraries in E. coli and Bacillus subtilis. Briefly, the slpA gene is 1395 bp in size with a coding capacity for a protein with a molecular mass of 48 159. The first 90 nucleotides of the structural gene encode a signal peptide of 30 amino acid residues with features of typical Gram-positive type signal peptides [156, 175]. The size of the mature polypeptide is 435 amino acid which is in good agreement with the molecular mass of 46 000 of the S-layer protein analyzed by SDS-PAGE. The slpA gene is preceded by a well conserved ribosome binding site (RBS) and two putative promoter regions, P1 and P2. The −35 and −10 regions of P1 and P2 resemble the conserved prokaryotic −35 and −10 consensus sequence . Furthermore, a transcription termination sequence is found downstream of the two translation stop codons of the slpA gene, indicating that the slpA gene is monocistronic . Furthermore, in the coding region of slpA 10 partly overlapping direct repeats of 10–12 nucleotides are present  possibly partly explaining the instability of the gene in heterologous hosts [168, 156].
Computer analyses of the predicted amino acid sequence of S-layer protein revealed that the codon usage was clearly biased  and most resembled that reported for highly expressed B. subtilis proteins . The typical features of the amino acid composition of the S-layer protein deduced from the DNA sequence were the high number of hydrophobic amino acids, amino acids with hydroxyl groups (Thr (17.8%), Tyr (6.4%) and Ser (9.2%)) and the absence of cysteine. The amount of basic amino acids (lysine (8.1%) and arginine (1.7%)) was higher than that of acidic amino acids which is a notable exception to the general features of the other non-lactobacillar S-layer proteins . A similar characteristic has also been described for L. acidophilus S-layer protein  and can be analyzed from the L. helveticus slpA sequence. The pI values predicted for the L. brevis, L. acidophilus and L. helveticus S-layers are 9.88, 9.84 and 10.08, respectively [154, 156].
The first search for L. brevis S-layer protein homologues from data bases revealed no genuinely related sequences . The predicted amino acid sequences of the recently described L. acidophilus and L. helveticus (EMBL: X91199 and X92752) slpA genes, however show 35.7% and 28.8% similarity to that of the L. brevis S-layer protein , respectively. Hybridization of a L. brevis slpA specific fragment with chromosomal DNA of L. buchneri, L. helveticus and L. acidophilus resulted in signals from clearly positive to weakly positive, respectively , indicating, according to phylogenetic relatednesses, that the slpA gene of L. buchneri is more closely related to that of L. brevis than the other two. In L. acidophilus, two slp genes with phase variation and in L. helveticus one functional slp gene and a truncated slp 3′ end have been found . In L. brevis, only one slp gene is present instead .
9.2In vivo expression of the slpA gene
The size of the S-layer transcripts determined by Northern blot analysis is 1.5 kb, confirming that slpA is a monocistronic transcriptional unit . Mapping of the transcription start site of slpA revealed two 5′ ends located immediately downstream of the two −10 regions deduced from the DNA sequence, and thus confirming the functionality of the promoters .
Determination of the stability of the slpA mRNA showed that the half-life of the slpA transcripts was 14 min . When compared to typical half-lives of prokaryotic mRNAs, the slpA transcripts are exceptionally stable. Recently, the half-life of the L. acidophilus slpA mRNA has been determined showing a value of 15 min . Furthermore, the transcripts of the Aeromonas salmonicida vapA gene have also been shown to be very stable (11–22 min) . The long half-lives of these three S-layer mRNAs from three different species may indicate that a high mRNA stability is a general feature of S-layer mRNAs. As they mediate the synthesis of a major structural component of the cell, the high stability of S-layer mRNAs is not unexpected.
A study of the usage of the two L. brevis slpA promoters (P1, P2) in different stages of growth by Northern blot analysis has revealed that the P2 promoter, located closer to the start codon, is efficiently used during both the exponential and the early stationary phase whereas slpA mRNA derived from P1 was only weakly detectable . Further quantitative analysis by dot blot hybridization showed that transcripts derived from both promoters are present throughout the entire growth phase but the level of transcripts derived from promoter P2 is 10 times higher than that of P1 . Some other S-layer genes carrying multiple promoters have also been described [47, 181]. The usage of three promoters has been described in the expression of the cwp operon of Bacillus brevis. Of these P2 is constitutively used and P3 preferentially at the exponential phase of growth .
In vivo expression studies of L. brevis have shown that the kinetics of the accumulations of the slpA mRNA and protein correlates well up to the onset of the stationary phase followed by a sharp decrease at the level of slpA mRNA. The rate of mRNA decay is, however, slower than expected from the half-life of slpA transcripts suggesting that residual transcription continues even though the total amount of the S-layer protein does not further increase at the stationary phase . In addition to the in vivo transcription studies of L. brevis slpA, growth dependent S-layer mRNA synthesis has been described only for A. salmonicida. The level of transcripts derived from the A. salmonicida S-layer protein gene, vapA, was found to be highest at the mid-exponential phase of growth, whereas a relatively sharp decline of vapA transcripts already occurred at the late exponential phase .
The L. brevis S-layer protein is not released into the supernatant fractions at any of the growth phases studied, confirming earlier observations  and suggesting a tight regulation of S-layer synthesis and assembly . Breitwieser et al.  have demonstrated the presence of substantial amounts of S-layer subunits on the inner surface or within the peptidoglycan layer in B. stearothermophilus suggesting an intermediate phase between the synthesis and final location of the S-layer protein. This has also been quite commonly observed in S-layers of other Gram-positive eubacteria . However, in L. brevis over 95% of the S-layer subunits could be released with the SDS-PAGE sample buffer from intact cells, as Western blot analyses of intact and disrupted cells indicated. Thus, it appears that essentially no accumulation of the L. brevis S-layer subunits took place inside the peptidoglycan layer prior to translocation to the outer surface.
9.3Heterologous protein secretion with the L. brevis S-layer signals
9.3.1Construction of a secretion vector based on the slpA signals
A derivative (pKTH2095) of the shuttle vector pGK12  has been utilized as the carrier of the secretion cassette constructed to contain the two promoters (P1, P2), signal sequence (SS) and transcription terminator (tslpA) of the L. brevis slpA and another terminator (t) upstream of splA. As reporter the β-lactamase gene (bla) of pUC19 was used. The secretion vector (pKTH2121, see Fig. 21) was constructed stepwise using PCR technology .
9.3.2Expression and secretion of Bla by the slpA secretion cassette
L. lactis (MG1614), transformed with pKTH2121, efficiently secreted β-lactamase into the culture medium and DNA sequencing of the cassette confirmed its stability and correctness. To test the utility of the cassette in lactobacilli, L. brevis (ATCC 8287), L. plantarum (NCDO 1193), L. gasseri (NCK 334) and L. casei (ATCC 393) hosts were also transformed with pKTH2121 and expression and secretion of β-lactamase was determined as the function of growth in flask cultivations (Fig. 22). In each strain carrying pKTH2121 all detectable Bla activity was in the growth medium. The highest yield (10 240 U/ml; 50 mg Bla/l) in the culture supernatants was obtained with L. lactis at the early stationary phase. The highest production levels of Bla in the early stationary phase L. brevis cells and in the exponential phase L. plantarum cells were 60% and 30%, respectively, of that in L. lactis. However, the rate of Bla production was roughly equal in L. lactis and L. plantarum, whereas that of L. brevis was somewhat slower. In all strains studied, degradation of β-lactamase due to proteolysis was observed. With L. plantarum, L. gasseri and L. casei the Bla activity rapidly decreased already at the early stationary phase, suggesting higher protease activity in these strains .
The comparison of the activity and amount of Bla protein by Western blots revealed a good correlation and lack of cell associated β-lactamase. The size of Bla secreted to the culture medium was equal to that of the mature Bla of E. coli, suggesting that the enzyme was correctly processed . Small amounts of steady state Bla precursor found in cell fractions may suggest that with the reporter construct in question the translocation machinery of each host strain was approaching its saturation level or processing of the slpA-bla fusion was not optimal. In the case of L. brevis the translocation of Bla was probably affected by the competition of the very efficient export of S-layer subunits .
The L. brevis slpA promoters were very efficiently recognized in L. lactis, L. brevis and L. plantarum, whereas in L. gasseri, the slpA promoter region appeared to be recognized at a lower level and in L. casei the level of transcripts was below the detection limit . Furthermore, high integrity and a correct size of bla mRNA was demonstrated, except of L. casei, in all these species.
9.3.3pH controlled cultivation
In order to determine whether pH control improves the stability and production of β-lactamase, L. lactis and L. brevis were grown in a bioreactor under constant pH and glucose feeding (Fig. 23; 214). In L. lactis, the yield of Bla could be increased up to 80 mg/l (Fig. 23). After reaching the maximum activity, 10 h after the cell density OD600=1, the level of β-lactamase was stably maintained at the stationary phase of growth, indicating stabilization of Bla activity with the pH control and glucose. Comparison of yields of β-lactamase production in lactococci revealed that a hybrid expression-secretion unit consisting of a strong cytoplasmic lactococcal promoter and an indigenous lactococcal signal sequence [184, 185] resulted in only 14% of the Bla activity obtained with the slpA based secretion system. This supports the high efficiency of the slpA expression and secretion signals. The kinetics of β-lactamase accumulation shows that the Bla production is essentially restricted to the exponential phase of growth. 65% of the maximum Bla activity was already reached within 2 h after the cell density of OD600=1 (Fig. 23), implying a very high rate of secretion with a calculated value of 5×105 molecules/cell/h. This value is comparable to or exceeds the best exoenzyme producing laboratory strains of Bacillus, and thus suggests utility of lactococci as production hosts. However, lengthening of the effective duration of the production phase, in order to improve product yields, requires optimization of growth or alternatively the use of e.g. immobilized cell systems. Similarly to lactococcal Bla production, the pH control resulted in the stabilization of β-lactamase formation in L. brevis (Fig. 23). The rate of Bla production was, however, clearly lower in L. brevis than in L. lactis. Thus, it is evident that the presumed competition caused by the translocation of S-layer protein has to be overcome before L. brevis can be efficiently utilized as a production host .
9.4Conclusions and perspectives
S-layers are common surface structures in lactobacilli and are present in many industrially used Lactobacillus species. The DNA sequences available for a few Lactobacillus S-layer protein genes suggest that lactobacillar S-proteins share common characteristics different from other S-layers in possessing a large amount of basic amino acids and thus very high predicted pI values, even though the consequences of that property are unknown. Functions of lactobacillar S-layers are unknown but recent observations of the adhesive properties of some S-proteins may suggest that they may have an important role in involvement in different gastrointestinal ecosystems. The L. brevis slpA gene is the first lactobacillar S-layer protein gene characterized. Also its in vivo expression at the level of transcription and translation is well studied. Furthermore, the utility of the slpA expression and secretion signals in heterologous protein secretion has been demonstrated. It is evident that the L. brevis slpA signals can be efficiently used for protein secretion in a variety of lactic acid bacteria even though the recognition of slpA promoters is host dependent. At present, the slpA based secretion cassette functions most efficiently in Lactococcus, giving a yield of the secreted model protein, β-lactamase, at the highest level of heterologous protein production described for lactic acid bacteria so far.
On the basis of the knowledge gathered, the future perspectives include further developments of lactobacillar S-layers for different biotechnological applications. An evident application of L. brevis S-layer protein is its development as a carrier vehicle for foreign antigenic epitopes. Further elucidation of the adhesive properties of L. brevis S-layer protein may also allow and extend its utilization as a carrier of oral vaccines and other substances for animal and human use. Further developments also include the extension of the use of L. brevis slpA signal both for secretion and for intracellular protein production to improve fermentation processes.
10Biotechnological applications of recombinant S-layer proteins rSbsA and rSbsB from Bacillus stearothermophilus PV72
Michaela Truppe, Stefan Howorka, Gerhard Schroll, Sonja Lechleitner, Beatrix Kuen, Stephanie Resch3 and Werner Lubitz
As several S-layer sequences have been elucidated, a great potential for the biotechnological use of recombinant S-layers exists . The possibility to build two-dimensional crystalline arrays of identical protein or glycoprotein subunits on different surfaces or interfaces opens appealing possibilities for functioning surfaces and to build supramolecular structures in the third dimension. Most importantly, S-layer lattices are highly anisotropic structures exhibiting remarkable differences in the topography and physicochemical properties of their surfaces [187–189]. In nature, the inner surface often interacts either with peptidoglycan or with membrane surfaces . As no detailed information on the arrangement of single amino acids or atoms within the tertiary structure of any S-layers with known sequences is available, recombinant modifications by exchange or introduction of single amino acids, introduction of epitopes within surface loops, weakening or enforcing intra- and intermolecular interactions remain so far the most practical tools to functionally describe the molecular architecture of specific S-layers . In this section examples of the latter approach will be given for the two different S-layer proteins SbsA and SbsB from B. stearothermophilus PV72.
B. stearothermophilus PV72 is an aerobic thermophilic Gram-positive bacterium. Under oxygen limited growth conditions, the cell surface layer is covered with a hexagonally (p6) ordered S-layer lattice [126, 136]. The crystalline p6 surface protein array consists of a self-assembly structure of a single protein species (SbsA) with a molecular mass of 130 000. SbsA is synthesized with a N-terminal leader peptide of 30 amino acids (aa) that is subsequently cleaved during the secretion process .
As shown in the accompanying communication by Scholz et al., under non-limiting oxygen conditions (oxidative stress) strain PV72 replaces the S-layer protein SbsA rapidly by a new type of S-layer protein (SbsB) with a molecular mass of 98 000 exhibiting an oblique (p2) lattice .
10.1Heterologous expression of sbsA and sbsB in Escherichia coli
The sequences of the S-layer genes sbsA and sbsB from strain PV72, encoding the two proteins with molecular masses of 130 000 and 98 000, were determined from PCR cloned gene fragments [129, 138]. The open reading frame of sbsA (3684 nucleotides) is predicted to encode a protein of 1228 aa including a leader sequence of 30 aa  and the reading frame of sbsB (2760 nucleotides) is predicted to encode a protein of 920 aa including a signal sequence of 31 aa. Amino acid sequence comparison of SbsA and SbsB did not reveal any significant homology .
The SbsA coding region including the signal sequence was cloned under thermosensitive transcriptional control of the λpL promoter into a low copy number vector (pBK4). Thermal inducible expression of sbsA from pBK4 in a strain of E. coli expressing the λcI857 from the chromosome (Table 3) led to accumulation of sheet-like self-assembling products of the protein in the cytoplasm as shown by ultrathin sectioning of whole cells and immunogold labelling using SbsA specific antibodies . SbsA protein was not detected in the periplasm or in the supernatant fractions of E. coli and long-term expression of sbsA did not lead to degradation of SbsA (12).
Table 3. Heterologous expression systems of sbsA and sbsB
Recently we also cloned and expressed the S-layer gene sbsA without signal sequence in E. coli and thus could confirm that the N-terminus of SbsA is rather flexible and does not affect the self-assembly capability of the S-layer in E. coli. Expression of the cloned sbsB gene in E. coli from plasmid pAK26 (Table 3) under transcriptional control of the lacpo system also resulted in SbsB formation in the cytoplasm of the cells .
10.2Heterologous expression of sbsA in Bacillus subtilis
By heterologous expression of sbsA and sbsB in E. coli it was observed that the signal sequences of both SbsA and SbsB were not recognized by the host cells. In order to determine whether the sbsA specific signal sequence is recognized in B. subtilis leading to the export of the S-layer protein to the cell surface or the supernatant SbsA was expressed in this heterologous Gram-positive bacterium. For this purpose the integration vector pX (13) was used, thereby circumventing the structural and segregational instabilities of common, autonomous replicating plasmids. In the resulting vector pST1 (Table 3) sbsA is under the transcriptional control of the xyl promoter. Transformants of B. subtilis with sbsA integrated in the chromosome could be induced for expression of sbsA by addition of xylose to the growth medium and large amounts of SbsA were found in the supernatant of the cells .
10.3Structural/functional analysis of SbsA and SbsB
Partial digestion of the cloned S-layer genes sbsA and sbsB allowed the insertion of a kanamycin resistance cassette  at defined sites of the genes. After selection for antibiotic resistance the mutagenized clones were isolated and the cassettes were removed by excision leaving behind inserts in the two genes with six nucleotides coding for the rare restriction site ApaI. These insertions resulted in the addition of two amino acids within the open reading frame of the genes. The heterologous expression of the mutagenized genes sbsA and sbsB in E. coli and electron microscopic examinations proved whether the different inserts interfered with inter- and intramolecular interactions involved in the self-assembly process (Table 4).
Table 4. Insertion of two amino acids at various positions of SbsA and SbsB
aa position in SbsA
Effect on self-assembly of rSbsA
aa position in SbsB
Effect on self-assembly of rSbsB
Positions of the two amino acids introduced in SbsA and SbsB are given in columns 1 and 3, respectively. Columns 2 and 4 show the effect of insertion of two amino acids on the self-assembly of rSbsA and rSbsB determined by electron microscopy.
10.4Construction of SbsA/SbsB fusion proteins
10.4.1Insertion of streptavidin in SbsA/SbsB
Whether the extension of two amino acid insertions by sequences coding for epitopes or larger protein sequences can be tolerated was investigated by insertion of streptavidin at aa positions 197, 296, 309 and 886 of SbsA. Expression of all fusion proteins could be detected by SDS-PAGE and immunoblots; however, electron microscopic analysis revealed that insertion of streptavidin at positions 309 and 886 disturbed rSbsA self-assembly (Table 5). The isolation of rSbsA-StrpA/rSbsB-StrpA and reassembly of dissolved monomers into homogeneous rSbsA-StrpA or mixed S-layers of the rSbsA-StrpA/SbsA type is one of the aims of these investigations. The binding of biotinylated substances as well as the determination of binding capacities for enzymes or other attached molecules is in progress.
Table 5. Effect of insertion of streptavidin, PRV epitopes, Bet v 1 and PHB synthase in SbsA and SbsB on S-layer formation
Insert of sequences (aa)
aa position of insert
Specific antibody recognition of S-layer protein and/or insert
Column 1 gives the lengths (aa) of the inserts. Positions of introduced sequences in SbsA and SbsB are given in column 2. Columns 3 and 4 show the results of expression and self-assembly of rSbsA and rSbsB by electron microscopy and Western blot using polyclonal antiserum directed against SbsA, SbsB, StrpA or PRV, respectively.
Bet v 1
10.4.2Insertion of Bet v 1 in SbsA/SbsB
Bet v 1 is the major pollen allergen of the birch and is responsible for atopic (IgE mediated) allergies in an increasing percentage of the population . Sequencing and cloning of Bet v 1 by others  made it possible to insert the open reading frame of Bet v 1 into nucleotide position 878 (aa 296) of sbsA (Table 5). This Bet v 1-S-layer fusion protein is being studied to determine whether it is capable of converting a (TH2 directed) IgE antibody response into a TH1 mediated response against Bet v 1. If so, it would indicate an ability to suppress the manifestations of allergy in patients susceptible to pollen allergies. Furthermore, SbsA-Bet v 1 fusion proteins are investigated for the ability to assay anti-Bet v 1 antibody concentrations and/or to reduce high levels of anti-Bet v 1 IgE.
10.4.3Insertion of pseudorabies virus antigens into SbsA/SbsB
Insertion of the pseudorabies virus  gB epitope SmaBB (489–1224 nucleotide fragment coordinates, refer to EMBL HEHSSGP2) into sbsA/sbsB (Table 5) was performed to test gB specific immune responses in experimental animals. Western blot analysis with a monoclonal antibody corresponding to the inserted sequence showed the accessibility of the virus polypeptide within the SbsA/rSbsB proteins. An extended study with other PCR fragments inserted into sbsA and/or sbsB is in progress.
10.4.4Insertion of PHB synthase (PhbC) of Alcaligenes eutrophus H16 into SbsA
A regular arrangement of polypeptide structures with enzymatic activities on the surface of S-layers is an ambitious goal for the construction of immobilized enzymes within a living cell and in the case of PHB synthase (590 aa) for the construction of a molecular machine for biopolymer synthesis.
The phbC gene was amplified by PCR from plasmid p4A  and inserted into the 878 ApaI site (Table 4) in frame with sbsA giving rise to plasmid pSbsA-PhbC. For the functional test of the enzymatic activity of the SbsA-PhbC construct E. coli cells harboring the corresponding plasmid had to be cotransformed with plasmid pUMS harboring the β-ketothiolase (PhbA) and acetoacetyl-CoA reductase (PhbB) of A. eutrophus. Poly-β-hydroxybutyrate (PHB) formation in E. coli (pSbsA-PhbC, pUMS) cells was indicated by staining with Sudan black, gas chromatography and electron microscopy. These findings indicate that the SbsA-PhbC construct is enzymatically active and could serve as an example for immobilizing enzymes on intracellular S-layer matrices.
These few examples show that the architectural principles using S-layer structures as rigid matrices can be applied to modulate their surfaces, e.g. to build multiple, multifunctional recombinant S-layers with various functions. It was demonstrated that the expression of recombinant SbsA/SbsB constructs in various cell systems combined with future ultrastructural approaches is useful to develop new tools in biotechnology. For example, recent data obtained from the comparison of protein and prostaglandin mediators suggest that even in extremely high doses, S-layer proteins do not exhibit endotoxic properties. They enhance immune cell functions and are therefore a promising basis for the construction of new types of vaccines or diagnostic systems. In addition, multivaccine components can be derived from multiple, different rSbsA subunits, each carrying relevant immunodeterminants of pathogens, when mixed together. Molecular S-layer specific machines can contribute to new ways of metabolic designs by S-layer immobilized enzymes.
Nicolas Bayan et al. thank J.C. Dedieu for his skilful technical assistance. Their article is dedicated to the memory of J.L. Peyret, who tragically died in 1993 and who initiated this work.
Agnès Fouet et al. wish to thank Gervaise Mosser (Institut Curie, Paris) for scientific advice, and Caroline Frolet for the Southern experiments.
Everly Conway de Macario and Alberto J.L. Macario thank the members of their laboratory who over the years participated in the study of S-layer and ABC transporter genes, particularly Linda E. Mayerhofer, Rong Yao, Charles B. Dugan, Robert J. Jovell, and Wayne Decatur. They thank Javier Cordero and Rosemarie A. Jack for assistance, and Tracy L. Godfrey for word processing and the members of the Photo Unit for the artwork and illustrations. Work in their laboratory was partially supported by a grant from NYSERDA (706-RIER-BEA).
Luis A. Fernández-Herrero et al. appreciate the excellent technical assistance of J. de la Rosa. Their work has been supported by Project BIO94-9789 from the CICYT and by an institutional grant from the ‘Fundación Ramón Areces’. During the development of their work L.A. Fernández Herrero and G. Olabarrı́a were the holders of fellowships from the Comunidad Autónoma de Madrid and the Gobierno Vasco, respectively.
The work of Michaela Truppe et al. was supported by the Fonds zur Förderung der Wissenschaftlichen Forschung (FWF, Projects S72/02 and S72/08). Stefan Howorka holds a fellowship from the Austrian Academy of Sciences.
We thank Monika Timm for editorial help.
The work of Martin J. Blaser and colleagues was supported by RO1 24145 from the National Institutes of Health.