DISPLAY OF HETEROLOGOUS MOLECULES ON BACTERIAL CELLS AND MEMBRANE VESICLES.

The present invention concerns novel Gram-negative bacterial cells and outer membrane vesicles (OMVs) derived from these cells, which display heterologous molecules on their surface. These bacterial cells and OMVs are suitable for use in vaccines or targeted drug delivery.

In Gram-negative bacteria, secretion and surface display of heterologous proteins are difficult to achieve due to the presence of a complex, multi-layered cell envelope consisting of an inner membrane, an outer membrane, and a gel like substance between the membranes called the periplasm. Several secretion systems have evolved to overcome this barrier and deliver proteins in the extracellular environment. Of these, the type V secretion system, also known as the Autotransporter (AT) pathway, appears the best suited for piggy-back transport of cargo proteins, because it combines simplicity with a high transport capacity (van Ulsen et al. 2014 Biochim Biophys Acta, 1843(8): 1592; Nicolai et al. 2015 Crit Rev Microbial 41 (1 ):109). Autotransporters (ATs) are large proteins that are secreted by Gram-negative bacteria, such as E. coli. The autotransporter system is simple in the sense that the autotransporter, as implied by its name, is suggested to carry all information, for translocation across the periplasm and outer membrane within the protein itself. However, the mechanism whereby the autotransporters are secreted is still not completely understood.

Autotransporters are synthesized as large precursor proteins that contain three domains: (i) an

N-terminal signal peptide that targets the protein to the Sec translocon and initiates transfer across the inner membrane, (ii) a passenger domain, which comprises the "cargo" protein that is to be secreted and (iii) a C-terminal β-domain) comprising a beta-barrel structure that integrates into the outer membrane and plays a crucial role in translocation of the passenger domain across the outer membrane into extracellular space.

After translocation, the passenger domain is cleaved from the translocator domain and is released into the extracellular environment. Cleavage can be achieved by the action of an (external) protease or a protease motif situated between the translocator domain and the passenger domain. Alternatively, cleavage takes place through an intramolecular autocatalytic event at a specific site between the translocator domain and the passenger domain. The passenger domain of an autotransporter comprises a beta stem structure and side domains. The beta stem is an elongated structure formed by an extended beta helix. The C-terminus of the passenger domain comprises an autochaperone domain which has been implicated in both passenger folding and translocation across the outer membrane. The autotransporter is first transported across the inner membrane by the Sec machinery, after which its C-terminal domain inserts into the outer membrane (OM) and forms a β-barrel that, together with a central linker domain, mediates transport of the functional N-terminal passenger domain to the cell surface or medium. Based on this secretion system, the inventors set out to develop a display platform by which attenuated bacterial cells or derived Outer Membrane Vesicles (OMVs) can be decorated with multiple heterologous antigens of choice for optimal stimulation of the immune system or with lectins, adhesins or affinity molecules of choice for optimal targeting to certain tissues or immune cells of interest, and other applications such as drug discovery or drug delivery.

Hbp is an autotransporter protein that belongs to the family of serine protease

autotransporters of Enterobacteriaceae (SPATEs). The crystal structure of the passenger domain of Hbp has been determined (Otto et al. 2005 J Biol Chem 280(17): 17339-45). The crystal structure shows that the polypeptide forms a long right-handed beta-helix structure (" beta stem"). Like other classical ATs, Hbp is organized in three domains: (i) the signal peptide at the N-terminus that triggers targeting to and translocation across the inner membrane via the generic protein-conducting Sec- translocon, (ii) the secreted passenger domain that carries the actual effector function of the AT, and (iii) the β-domain at the C-terminus that adopts a β-barrel conformation in the outer membrane (OM) and plays a crucial role in transfer of the passenger domain across the OM (Figure 1 A and D). This latter step is supported by the generic insertase for outer membrane proteins, the Bam complex, in a concerted, poorly understood mechanism (Sauri A et al. Microbiology 2009, 155 (Pt 12), 3982-91 ). The passenger domain of the Hbp comprises two larger side domains, domain d1 and domain d2, of which d1 comprises the serine proteinase activity of the protein, and d2 has an unkown function. There are also three smaller domains, domain d3, domain d4 and domain d5. At the cell surface the passenger domain starts to fold, which provides a pulling force to energize translocation across the outer membrane (Drobnak et al. 2015 J Biol Chem 290(16): 10104-16). Finally, upon completion of folding of the passengers-domain tandem and release from the Bam complex, the passenger is cleaved from its β-domain through an autocatalytic mechanism that takes place in the barrel interior (Barnard et al. 2007, Nat Struc Mol Biol 14(12): 1214-20). The cleaved and released passenger domain is a long β-helical stem structure from which small loops and larger functional domains protrude (Otto et al. 2005 supra). Similar beta stem domains have also been shown for other autotransporters, such as pertacin (Emsley et al. 1996 Nature 381 :90-92), EspP (Khan et al. 201 1 J Mol Biol 413(5):985-1000), Pet (Domingo Meza-Aguilar et al. 2014 Biochem Biophys Res Commun 445(2):439-44)and IgA protease (Johnson et al. 2009 J Mol Biol 389(3): 559-74). Autotransporters have been used as for surface display purposes. Gram-negative bacterial cells or OMVs displaying heterologous molecules on the outer surface are known in the art. Jong et al Microbial Cell Factories 2012, 1 1 :85 and Jong et al. Microbial Cell Factories 2014, 13:162 describe live bacteria which display an adapted Escherichia coli autotransporter, Heamoglobin protease (Hbp), at the surface. The Hbp was engineered into a platform for the secretion and surface display of heterologous polypeptides, by replacement of the passenger side domains with heterologous antigens. By interrupting the cleavage site between passenger and β-domain, the Hbp fusion proteins remain cell associated rather than being released, and facilitate efficient surface exposure of the heterologous proteins at some distance of the cell surface. By using antigens derived from

Mycobacterium tuberculosis, Chlamydia trachomatis and influenza A virus, the Hbp platform was demonstrated to be a useful tool for successfully display of heterologous antigens on the cell surface of live bacteria and OMVs (Jong et al. 2012 supra; Jong et al. 201 supra; Daleke-Schermerhorn et al. 2014 Appl. Environ Microbiol, 80(18):5854-65).

Despite being an efficient and versatile system for bacterial-based display of heterologous proteins, the capacity of Hbp to translocate complex and/or bulky fusion partners across the cell envelop was found to be limited (Jong et al 2014 supra; Daleke-Schermerhorn et al. 2014 supra; Kuipers et al. 2015 Vaccine 33(17): 2022-9; Jong et al. 2007 Mol Microbiol 63(5): 1524-36). Due to inherent limitations in the secretion of Hbp across the outer membrane, translocation of large or structurally complex antigens genetically fused to Hbp generally results in relatively low-level surface exposure. It was found that the number, size and structural complexity of the antigens fused to the Hbp carrier restricts the translocation capacity of the Hbp antigen fusion product. The problem of size can partly be solved by splitting up the antigen and dividing the resulting fragments over the various permissive integration sites in Hbp (Jong et al. 2014 supra; Jong et al. 2012 supra; Kuipers et al. 2015 supra). However, this strategy could result in the loss of conformational epitopes, and is therefore not sufficient for large or structural complex antigens. Since the density of antigens at the OMV surface has shown to be important for protective efficacy, there is a need to increase the display capacity of the Hbp system while maintaining a high expression level.

It was surprisingly found that by covalently linking heterologous molecules to already surface exposed structures, the limitations of the membrane translocation steps could be circumvented. It was found that instead of using genetic fusion, heterologous molecules such as antigens, lectins, nanobodies and antibodies could be linked enzymatically to an autotransporter such as Hbp present at high density at the surface of OMVs using Catcher/Tag coupling technology. Catcher/Tag coupling technology is based on the spontaneous formation of an isopeptide bond between a peptide tag and its peptide binding partner, a so called peptide catcher. The peptide tag and corresponding peptide catcher form a unique Catcher/Tag ligation pair.

Thus, the present invention provides for Gram negative bacterial cells and OMVs derived therefrom, which display on the outer surface an autotransporter fusion protein covalently coupled via isopeptide bonds to a heterologous molecule selected from the group consisting of antigens, lectins, adhesins or affinity molecules, wherein the autotransporter fusion protein comprises at least one moiety of a Catcher/Tag ligation pair, and the heterologous molecule comprises the other moiety of said Catcher/Tag ligation pair.

By genetically fusing the sequence of a moiety of a Catcher/Tag ligation pair to the autotransporter protein, an autotransporter fusion protein could be obtained comprising a Catcher or Tag moiety of a Catcher /Tag ligation pair. By equipping the heterologous molecule with the corresponding moiety of a Catcher/Tag ligation pair, the heterologous molecule could be attached to the autotransporter fusion protein by the formation of an isopeptide bond between the Catcher and Tag moieties of the Catcher/Tag ligation pair. It was surprising to find that heterologous molecules could be covalently linked via an isopeptide bond to the autotransporter fusion carrier expressed on the surface of bacterial cells and OMVs irrespective of the abundant presence of liposaccharides, fimbriae, flagella and other structures present on the surface of the cells and OMVs. Without being bound by theory, it was expected that the presence of other surface structures, in particular liposaccharides would interfere with the formation of the isopeptide bond, thereby hindering the coupling of the heterologous molecule to the displayed autotransporter fusion protein. Thus linking heterologous molecules via Catcher/Tag coupling technology to an autotransporter present at high density at some distance of the surface provides an interesting alternative for displaying heterologous molecules on the surface of bacterial cells and OMVs, especially when complex or large molecules are concerned.

By covalently conjugating heterologous molecules via Catcher/Tag protein ligation to the autotransporter already expressed on the surface, the display capacity of the autotransporter carrier platform could be increased while at the same time a high expression level of the autotransporter on the surface could be maintained. Catcher/Tag protein ligation technology allows for the decoration of bacterial cells and OMVs with heterologous recombinant protein partners without the restrictions observed with genetic fusion of the autotransporter carrier to the heterologous protein and translocation of the autotransporter fusion protein.

Suitable Catcher/Tag ligation pairs for use according to the invention are pairs that have been derived from proteins which are capable of spontaneously forming one or more isopeptide bonds. Isopeptide bonds are amide bonds that are formed between carboxyl/carboxamide and amino groups, whereby at least one of the carboxyl or amino groups is present in the side chain of an amino acid residue that is part of the peptide backbone. The domain of the protein that expresses this spontaneous isopeptide bond can be expressed as separate fragments to give a peptide tag fragment (a "Tag" moiety), and a binding peptide partner for the peptide tag (a "Catcher" moiety). Mixing the two peptide fragments results in the spontaneous formation of an isopeptide bond between the Tag moiety and the Catcher moiety. Of some Catcher/Tag ligation pairs, the Catcher domain could be further split into a small second Tag moiety and a large fragment harboring the Ligase domain to create a tripartite system. In case of a tripartite system, the fragment harboring the Ligase domain mediates the coupling of the Tag- and Catcher moiety, but is not part of the isopeptide bond. Such tripartite systems involving a Catcher/Tag ligation pair and a third peptide fragment harboring the Ligase domain are also encompassed as Catcher/Tag ligation pairs for use according to the invention. Thus, the Catcher/Tag ligation pairs according to the invention include Catcher/Tag ligation pairs of both the bipartite and tripartite systems. The property of spontaneously forming a covalent isopeptide bond enabled the development of Catcher/Tag ligation pairs, wherein the Tag moiety and the Catcher moiety are capable of covalently reconstituting the isopeptide bond, wherein the Tag moiety comprises one of the residues involved in the isopeptide bond (e.g. a lysine), and the Catcher moiety comprises the other residue involved in the isopeptide bond (e.g. an asparagine or aspartate). The binding between a Tag moiety and its corresponding binding partner termed Catcher moiety is unique for each Catcher/Tag pair. In a preferred embodiment, the Catcher moiety comprises the ligase fragment which mediates isopeptide bond formation. The isopeptide bond formed between the Tag moiety and the Catcher moiety is covalent and, hence, very stable. Catcher/Tag ligation pairs can be derived from any protein that is capable of spontaneously forming isopeptide bonds, such as for instance described in Proschel et al PLoS One. 2017.12(6), Tan et al PLoS One. 2016. 1 1 (10) and WO 201 1 /098772. Particular suitable Catcher/Tag ligation pairs are the peptide tag and corresponding peptide catcher pairs described in WO 201 1 /098772, the Spy0128 Catcher/Tag pairs described in Zakeri & Howarth 2010 J Am Chem Soc 132(13), the FbaB Spy Catcher/Tag pair described in Zakeri et al Proc Natl Acad Sci USA. 2012. 109(12), and the Sdy Catcher/Tag pair derived from Fibronectin binding protein

(UniProtKb Q53971 ) described in Tan et al. (2016) supra. Preferably the Catcher/Tag ligation pair is a SpyCatcher/SpyTag pair as described in Zackeri et al. (2012) supra, a SdyCatcher/SdyTag pair described in Tan et al. (2016) supra, or a SnoopCatcher/SnoopTag pair as described in Veggiani et al. Proc Natl Acad Sci USA 2016, 1 13(5): 1202-7, more preferably the SpyCatcher/SpyTag pair (Zackeri et al. (2012) supra) or the SnoopCatcher/SnoopTag pair (Veggiani et al. (2016) supra).

The Spy protein ligation system is based on a domain of fibronectin binding protein FbaB of

Streptococcus pyogenes in which an intramolecular isopeptide bond is formed between an adjacent lysine and aspartic acid residue through an autocatalytic mechanism (Zackeri et al (2012) supra). By splitting this domain into a 13 amino acid peptide, termed SpyTag, and its remaining 138 amino acid fragment, termed SpyCatcher, a Catcher/Tag pair was created for intermolecular coupling. Fused to different proteins the SpyTag and SpyCatcher reconstitute the domain upon mixing and form a stable intermolecular covalent bond under a wide range of conditions (pH, temperature, buffer). In a similar way a peptide tag and corresponding binding partner was developed based on the Streptococcus pneumoniae adhesin RrgA, providing a SnoopTag and SnoopCatcher ligation pair (Veggiani et al. 2016 supra).

The SpyCatcher/SpyTag conjugation system was further adapted into a tripartite system by splitting up the SpyCatcher into the SpyLigase domain and the 10 amino acid peptide KTag (Freirer et al. 2014 Proc Natl Acad Sci USA 1 1 1 (13): E1 176-81 ). The SpyTag and the KTag each contain one of the cognate amino acid residues (Asp and Lys, respectively) that form the isopeptide bond, while the catalytic activity is included in the SpyLigase domain that can be added separately. Preferred Catcher moieties are the SpyCatcher, KTag and Snoop Catcher moieties. Preferred Tag moieties are the SpyTag and SnoopTag.

Suitable autotransporter proteins that can be used to obtain the bacterial cells and OMVs of the present invention are proteins that belong to the pfam autotransporter family ('Autotransporter' PF03797) and which are known or predicted to form a beta stem motif. The BETAWRAPPRO method for sequence analysis can be used to predict if the passenger domain of an autotransporter will form a beta stem motif (Junker et al. 2006 Proc Natl Acad Sci USA 103(13): 4918-23). It was found that an autotransporter protein can be used for improved display of heterologous molecules on the surface of bacterial cells and OMVs. Care should be taken that the beta stem forming sequence of the passenger domain of the autotransporter is essentially intact. The native passenger domain of an autotransporter protein can be considered as comprising several sections of beta stem forming sequence, linked together by non-beta stem forming sequences. The non-beta stem forming sequences are suitable sites for insertion of one or more heterologous sequences, Thus, a Tag- or Catcher moiety can be placed between two parts of beta stem forming sequence. The Tag- or Catcher moiety can also be fused to the N-terminus of the passenger domain.

Whereas the beta stem forming sequence is essential for optimal expression and display, the side domains of the passenger domain of the autotransporter protein are suitable sites for insertion of a moiety of a Catcher/Tag pair. The Catcher- or Tag moiety can be inserted to replace a part of or the whole side domain, or the moiety can be fused to a side domain. By fusing, i.e. inserting, replacing or partly replacing the Catcher- or Tag moiety to one or more side domains of the passenger domain of the autotransporter, while keeping the beta stem structure intact, an efficient display system for the improved display of heterologous molecules on the bacterial cell and OMV was achieved.

Autotransporter proteins that are suitable for producing a fusion product with the at least one Catcher- or Tag moiety are the autotransporter proteins such as described in WO 2012/048199, Jong et al. (2012) supra, Jong et al (2014) supra and Daleke-Schermerhorn et al. (2014 supra, wherein the cleavage site has been disrupted to create display of the passenger domain with the at least one Tag- or Catcher moiety on the surface of the bacterial cell or OMV. The replacement methods described in these references can be used to produce the autotransporter fusion proteins described herein.

Suitable methods for detecting beta stem forming sequence and side domains of passenger domains of autotransporters include biophysical methods such as X-ray crystallography and bioinformatics software such as structure prediction tools. X-ray crystallography is a standard procedure that is highly efficient and automated, and familiar to a person skilled in the art. Examples of high resolution structures of passenger domains and suitable methods for determination of structures of the passenger domain of autotransporters are e.g. found in Otto et al (2005), supra; Emsley et al (1996) Nature 381 :90-92; Johnson et al (2009) J Mol Biol 389(3):559-74; Khan et al. 201 1 J Mol Biol 413(5):985-1000; and Domingo Meza-Aguilar et al. 2014 Biochem Biophys Res Commun 445(2):439- 44). An example of a bioinformatics method that is suitable for determining beta stem structure is the M4T homology modeling method (Rykunov et al 15 2009 J Struct Funct Genomics 10: 95-99). Where a three-dimensional model of the protein is used for the identification of beta stem domains and side domains, it is suitable that the model obtained has a resolution of better than 4 angstrom. Side domains will then be visible as domains that protrude from the beta stem . By observation of the structure of the passenger domains of autotransporters, it can be seen that parts of the sequence are not part of the beta stem but form domains that protrude from the beta stem . Methods such as those described above can be used for determining which domains or amino acids of the native passenger domains are suitable for insertion of heterologous peptide fragment and which should be kept essentially intact. In the fusion protein according to the disclosure, the beta stem forming sequence from the autotransporter passenger domain is essentially intact. Thus, as little as possible of the beta stem forming sequence should be removed. Predicted domain borders are of help to determine where Tag- or Catcher moiety can be inserted . Preferably the autotransporter protein is a serine protease autotransporter of the Enterobacteriacea ('SPATE') such as described in Yihfen et al. 2008 Trends in Microbiol 16(8): 370-9. The serine protease autotransporters of Enterobacteriaceae represent a group of large-sized, multi- domain exoproteins found only in pathogenic enteric bacteria. SPATES form a family of

autotransporters linked by commonality in their structural architecture and mechanism of secretion (Yen et al 2008 Trends Microbiol 16(8):370-9). These proteins contain a highly conserved channel- forming C-terminal domain, which facilitates secretion of the passenger domain to the cell surface. The C-terminal domain also mediates autoproteolytic cleavage, which releases the passenger from the bacterial cell. The passenger folds into a characteristic parallel β-helical stalk-like structure (beta stem structure) with an N-terminal globular domain that performs serine proteolytic activity. SPATES represent a typical example of conventional autotransporters: the beta-domain of each SPATE folds into a single beta stem structure formed by 277 amino acids. Once on the cell surface, the passenger of a SPATE is cleaved from the translocator and released into the extracellular environment. Cleavage occurs in a conserved site located between two consecutive asparagine residues in the linker region joining the passenger and the translocator domains. An aspartate located inside the pore of the translocator domain performs the cleavage task. Following secretion and folding, the mature SPATE passenger becomes functionally active. Although the secreted passenger domain of each SPATE exhibits divergent pathogenic function, the SPATE passenger domains display considerable sequence homology among them, resulting in a similar architecture that is rich in beta-strands, and are predicted to fold into a parallel beta-helical structure. This fold was recently verified by the crystal structure of the SPATE protein Hemoglobin protease Hbp from E. coli (Otto et al 2005 supra) and the crystal structures of the SPATES EspP (Khan et al. 201 1 supra) and Pet (Domingo Meza-Aguilar et al. 2014 supra).

In a preferred embodiment the autotransporter that is used to produce an autotransporter fusion protein for display on the surface is a SPATE protein, more preferably a SPATE protein selected from the group of hemoglobin-binding protease (Hbp; SwissProt 088093), extracellular serine protease (EspC) and temperature-sensitive hemagglutinin (Tsh; SwissProt Q47692) from Escherichia coli, wherein the cleavage site has been disrupted to prevent secretion of the SPATE. Preferably, the cleavage site is disrupted by substituting the conserved consecutive Asn residues with two other amino acid residues, e.g. a Ser - and a Gly residue. The sequence of Tsh is highly homologous to that of Hbp, differing only in amino acid residue from Hbp.

In en even more preferred embodiment the SPATE protein is Hemoglobin protease (Hbp). The Hbp protein which is suitable for use in the present invention is preferably derived from Escherichia bacteria, more preferably E. coli. Preferably a Hbp protein comprising the amino acid sequence depicted in SEQ ID NO:1 is used to make the Hbp fusion proteins of the invention. Suitable Hbp protein constructs that can be used as carrier have a disrupted cleavage site between the β-domain and the passenger domain, which prevents secretion of the passenger domain into the extracellular space, such as the Hbp proteins that have been described in Jong et al (2012) supra, Jong et al. (2014) supra, WO2012/041899 and US2017/0258885. The Hbp protein preferably comprises the amino acid sequence of SEQ ID N0:2 (HbpD), which corresponds to native Hbp wherein the cleavage site between the β-domain and the passenger domain has been disrupted by substituting the conserved Asn1 100 and Asn1 101 residues with a Ser and Gly residue, respectively. Also

encompassed as Hbp protein that is suitable for use in the present invention are Hbp proteins comprising an amino acid sequence that is homologous to SEQ ID NO:2. Homologous sequences are amino acid sequences that have at least 80%, preferably 85%, more preferably 90%, even more preferably 95%, most preferably 97% sequence identity with SEQ ID NO:2. Sequence homology can be readily determined using publicly available software such as BLAST. The Hbp fusion protein according to the invention comprises the at least one Tag- or Catcher moiety in the N-terminus of the Hbp, preferably at the location of one of the side domains of the passenger domain. The Hbp protein comprises at least five side domains located at aa positions d 1 : 53-308 (domain d 1 ), 533-608 (domain d2), 657-697 (domain d3), 735-766 (domain d4) and 898-992 (domain d5) of HbpD (SEQ ID NO:2). The Hbp carrier comprising at least one moiety of a Catcher/Tag pair can be obtained by fusion to at least one side domain or by replacing in whole or in part at least one side domain in the passenger domain with a Tag or Catcher moiety. Each of the side domains 1 , 2, 3, 4 or 5 of the wild-type Hbp (see Figure 1 A) or other loops projecting from the beta-stem domain may be used for inserting one of the moieties of a Catcher/Tag pair. Preferably a moiety of a Catcher/Tag pair is inserted in or is replacing in whole or in part side domain 1 , side domain 2 or side domain 4, more preferably side domain 1. The Hbp fusion protein may comprise at least two, preferably three or more Tag or Catcher moieties. When the Hbp fusion protein comprises two or more Tag- or Catcher moieties, the moieties can be part of the same or orthogonal (i.e. mutually unreactive or non-cognate) Catcher/Tag ligation pairs. In a preferred embodiment, the Tag or Catcher moieties comprised in the Hbp fusion protein comprising two or more Tag or Catcher moieties, correspond to Catcher/Tag pairs that are orthogonal, meaning that they are mutually unreactive, i.e. the Tag- and Catcher moitety of the first pair cannot react with either of the Tag- or Catcher moiety of the second pair. Inserting two or more Catcher- or Tag moieties in the same Hbp carrier will increase flexibility and valency of the Hbp display system. Preferably, the Hbp fusion protein comprises an amino acid sequence selected from the group of SEQ ID NO: 4, SEQ ID NO: 5, SEQ ID NO:6, SEQ ID NO:7, SEQ ID NO:8, SEQ ID NO: 9, SEQ ID NO:10, SEQ ID NO: 1 1 , SEQ ID NO:12 and SEQ ID NO:13. . Also encompassed as Hbp fusion protein according to the invention are Hbp fusion proteins that comprise an amino acid sequence that is homologous to one of the sequences selected from the group of SEQ ID NO: 4, SEQ ID NO: 5, SEQ ID NO:6, SEQ ID NO:7, SEQ ID NO:8, SEQ ID NO: 9, SEQ ID NO:10, SEQ ID NO: 1 1 , SEQ ID NO:12 and SEQ ID NO:13. Homologous sequences in this context are amino acid sequences that have at least 80% sequence identity with their reference amino acid sequence, more preferably 85%, 90%, 95%, 96%, 97%, 98%, or 99%, provided that the beta-stem domain in the Hbp fusion protein remains essentially intact.

An autotransporter fusion protein comprising at least one Catcher- or Tag moiety of a Catcher /Tag ligation pair can be obtained by genetically fusing the sequence of a moiety of a Catcher/Tag ligation pair to the autotransporter protein using standard molecular biology techniques involving restriction enzymes, DNA ligases, PCR, oligonucleotide synthesis, DNA purification, and other methods well known to a person skilled in the art such as Sambrook, J., E.F. Fritsch, and T. Maniatis. 1989. Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York. Using the replacement technology and methods described in Jong et al (2012) supra, Jong et al. (2014) supra, WO2012/041899 and US2017/0258885, an autotransporter fusion protein can be designed comprising at least one moiety of a Catcher/Tag pair. Using these techniques, a nucleic acid comprising the sequence encoding the Hbp carrier according to the invention can be obtained wherein at least one sequence encoding a moiety of a Catcher/Tag pair is inserted into or replacing a side domain of the passenger domain of the autotransporter. By inserting the nucleic acid sequence encoding the autotransporter carrier comprising at least one moiety of a Catcher/Tag pair into an expression vector for bacterial expression, wherein the expression vector carries a promoter that enables expression of said nucleic acid sequence, and by transforming a Gram negative bacteria with said expression vector, an autotransporter fusion protein can be expressed in and subsequently displayed on the surface of Gram negative bacterial cells. Gram negative bacteria, which are suitable for use according to the invention can be any

Gram negative bacterial cell that is capable of expressing and displaying an autotransporter fusion protein comprising at least one moiety of a Catcher/Tag ligation pair on the outer surface. Preferred bacteria are Escherichia or Salmonella bacteria. More preferably, the Gram-negative bacterial cell is a Salmonella spp., e.g. the subspecies S. enterica subsp enterica, more preferably the serovar

Typhimurium, even more preferably S. Typhimurium strain SL3261 . Particularly preferred is S.

Typhimurium strain SL3261 which is ToIRA deficient (MoIRA derivative of S. Typhimurium strain SL3261 ). In a preferred embodiment the bacteria are live, attenuated bacteria. In a further embodiment, the bacteria are live, attenuated auxotrophic bacteria, more preferably live, attenuated, auxotrophic S. Typhymurium bacteria. The bacteria of the present invention can be obtained by transforming the bacteria with said nucleic acid molecule that encodes an autotransporter fusion protein comprising at least one amino acid sequence of a Catcher- or Tag moiety incorporated therein. The Gram-negative bacterial cells preferably display on their surface an autotransporter fusion protein as described herein, more preferably a SPATE fusion protein, even more preferably a Hbp fusion protein. In a particular preferred embodiment the displayed Hbp fusion protein comprises an amino acid sequence selected from the group of SEQ ID NO: 4, SEQ ID NO: 5, SEQ ID NO:6, SEQ ID NO:7, SEQ ID NO:8, SEQ ID NO: 9, SEQ ID NO:10, SEQ ID NO: 1 1 , SEQ ID NO:12 and SEQ ID NO:13. Also encompassed Gram-negative bacterial cells according to the invention are Gram-negative bacterial cells which display a Hbp fusion protein that comprises an amino acid sequence that is homologous to one of the sequences selected from the group of SEQ ID NO: 4, SEQ ID NO: 5, SEQ ID NO:6, SEQ ID NO:7, SEQ ID NO:8, SEQ ID NO: 9, SEQ ID NO:10, SEQ ID NO: 1 1 , SEQ ID NO:12 and SEQ ID NO:13. Homologous sequences in this context are amino acid sequences that have at least 80% sequence identity with their reference amino acid sequence, more preferably 85%, 90%, 95%, 96%, 97%, 98%, or 99%, provided that the beta-stem domain in the Hbp fusion protein remains essentially intact. Besides live attenuated Gram-negative bacterial cells, the present invention also provides for OMVs displaying on their surface an autotransporter fusion protein as described herein, which are suitable for Catcher/Tag coupling. All Gram-negative bacteria produce OMVs that shed from the outer membrane during growth. OMVs are spherical nanostructures (30-200 nm diameter) composed of phospholipids, LPS and proteins derived from the periplasm and outer membrane (Schwechheimer et al. 2015 Nature reviews Microbiol 13(10): 605-19).

OMVs are an attractive vaccine formulation because, though non-replicating and safe to use, they have intrinsic immunostimulatory properties that are a facsimile of their live parental cells (Alaniz et al. 2007) J Immunol 179(1 1 ): 7692-701 ; Gnopo et al. 2017 Adv Drug Deliv Rev). For this reason, OMVs attract increasing interest in the development of vaccines and therapeutic agents. Compared to live attenuated vaccines OMVs raise less safety concerns because of their non-replicative nature. Yet, the intrinsic constituents of OMVs such as outer membrane polypeptides (OMPs), lipopolysacchandes (LPS) and other pathogen associated molecular patterns (PAMPs) activate innate as well as adaptive immune responses leading to the induction of adaptive immunity. This makes the addition of adjuvant unnecessary, which is an important advantage for vaccine approval. Thus OMVs combine antigen presentation and intrinsic adjuvant activity, which in combination with their non-replicative nature makes them extremely suitable and promising candidates for vaccines and therapeutic agents.

Suitable OMVs which can be used according to the invention are preferably derived from Gram-negative bacterial cells as described herein. Preferably the OMVs are derived from Gram- negative bacterial cells as described herein, which are hypervesiculating. These cells have been modified to shed large amounts of OMVs. More preferably, the OMVs are derived from Salmonella spp., e.g. the subspecies S. enterica subsp enterica, more preferably the serovar Typhimurium, even more preferably S. Typhimurium strain SL3261 . In a preferred embodiment, the OMVs are derived from a hypervesiculating derivative of S. Typhimurium strain SL3261 , in particular a ToIRA deficient derivative wherein the toIRA genes are deleted or inactivated (MoIRA derivative). This strain is particularly interesting in that it is capable of shedding high amounts of OMVs displaying on their surface an autotransporter fusion protein comprising at least one moiety of a Catcher/Tag ligation pair. The OMVs preferably display on their surface an autotransporter fusion protein as described herein, more preferably a SPATE fusion protein, even more preferably a Hbp fusion protein. In a particular preferred embodiment the OMVs display on their surface a Hbp fusion protein, which comprises an amino acid sequence selected from the group of SEQ ID NO: 4, SEQ ID NO: 5, SEQ ID NO:6, SEQ ID NO:7, SEQ ID NO:8, SEQ ID NO: 9, SEQ ID NO: 10, SEQ ID NO: 1 1 , SEQ ID NO: 12 and SEQ ID NO: 13. Also encompassed as OMVs according to the invention are OMVs that display on their surface a Hbp fusion proteins that comprise an amino acid sequence that is homologous to one of the sequences selected from the group of SEQ ID NO: 4, SEQ ID NO: 5, SEQ ID NO:6, SEQ ID NO:7, SEQ ID NO:8, SEQ ID NO: 9, SEQ ID NO: 10, SEQ ID NO: 1 1 , SEQ ID NO: 12 and SEQ ID

NO: 13. Homologous sequences in this context are amino acid sequences that have at least 80% sequence identity with their reference amino acid sequence, more preferably 85%, 90%, 95%, 96%, 97%, 98%, or 99%, provided that the beta-stem domain in the Hbp fusion protein remains essentially intact. Methods for OMV production and isolation are known in the art and have been described in e.g. Gnopo et (2017 supra, Chen et al (2010) PNAS 107:3099-3104; Bernadac et al (1998) J Bacteriol 180: 4872-4878; Kesty and Kuehn (2004) J Biol Chem 279: 2069-2076; 10 Kolling and Matthews (1999) App Env Microbiol 65:1843-1848; Kitagawa et al (2010) J Bacteriol 192:5645-5656). The OMVs of the present invention can be prepared from the Gram-negative bacterial cells described herein using any of these methods.

Heterologous molecules that can be coupled via isopeptide bonding technology to Gram- negative bacterial cells or OMVs displaying on their surface an AT fusion protein as described herein can be any molecule that can be produced as a Tag- or Catcher fusion product. Preferably the heterologous molecules are selected from the group consisting of antigens, lectins, adhesins and affinity molecules.

Suitable antigens are any antigen that can be produced as a Tag- or Catcher fusion construct, including proteins that have been modified such as glycoproteins, lipoproteins and the like, oligomeric proteins, , and antigenic fragments thereof. Suitable antigens also comprise oligosaccharide that can be produced in fusion with a Tag- or Catcher-moiety.

Preferred antigens that can be coupled via isopeptide bonding technology according to the invention are antigens from infectious microorganisms, including but not limited to Streptococcal antigenic proteins, Mycobacterial antigenic proteins, Chlamydial antigenic proteins, Influenza antigenic proteins, and the like as described in Jong et al (2012) supra, Jong et al. (201 ) supra, Daleke- Schermerhorn et al. (2014) supra, WO2012/041899 and US2017/0258885. Preferably the antigen is a Pneumococcal antigen from Streptococcal pneumoniae, more preferably pneumococcal surface protein A (PspA) or SP1690. Even more preferably the antigen is PspAa', which is the N-terminal domain of PspA, or full-length SP1690 lacking the N-terminal Cys residue. In a specific embodiment, the antigen is PspAa comprising the amino acid sequence corresponding to amino acid residues 1 12- 324 of SEQ ID NO:17 (SpC-PspAa-SnT). In another specific embodiment the SP1690 antigen comprises the amino acid sequence corresponding to amino acid residues 1 12-533 of Seq ID NO:24 (SpC-SP1690-SnT) or amino acid residues 135-556 of SEQ ID NO:25 (SnC-SP1690-SpT).

The present invention also provides for the ligation of multiple antigens or antigenic fragments to a Hbp fusion protein as described herein that is displayed on the surface of Gram-negative bacteria and OMVs. The ligation of multiple antigens can be obtained by using produced fusion constructs consisting of a string of multiple antigens fused to each other, wherein the construct is attached to a Catcher- or Tag moiety of a Catcher/Tag ligation pair. Thus, in a further embodiment, the Gram- negative bacterial cells and OMVs as described herein display on their surface an autotransporter fusion protein covalently linked to a string of multiple antigens.

The string of multiple antigens can be obtained by genetic fusion of the multiple antigens to each other and to a Catcher- or Tag moiety of a Catcher/Tag ligation pair. The string of multiple antigens can be a repetition of the same antigen, or may comprise different antigens or antigenic fragments thereof, provided that the antigens or antigenic fragments thereof are compatible.

Compatible antigens or fragments are understood to be antigens or antigenic fragments that do not interfere with the immune response induced by each antigen or fragment. Preferably the string of multiple antigens is a fusion protein comprising PspAa and SP1690, fused to a Catcher moiety at one end and a Tag moiety at the other end, wherein the respective Catcher- and Tag moiety are derived from orthogonal Catcher/Tag pairs to prevent circularization. Preferably the PspAo SP1690 fusion protein is fused to a Catcher moiety at the N-terminus and a Tag moiety at the C-terminus, more preferably to a SpyCatcher moiety at the N-terminus, and a SnoopTag at the C-terminus. More preferably the string of multiple antigens has the amino acid sequence of SEQ ID NO:18 (SpC-PspAa- SP1690-SnT).

In an alternative embodiment, a string of multiple antigens can be obtained by iterative extension of multiple different antigens by using recombinant antigens, wherein each antigen is flanked at the C- and N-terminal end with a moiety of a Catcher/Tag ligation pair. Care should be taken to use moieties of orthogonal Catcher/Tag pairs for the N- and C-terminal end of each the antigen respectively, to prevent circularization of the C-terminal end to the N-terminal end, thereby creating a circular antigen. Preferably, the antigens are flanked by a Catcher moiety of Catcher/Tag pair 1 , and a Tag moiety of Catcher/Tag pair 2, or a Tag-moiety of Catcher/Tag pair 1 and a Catcher moiety of Catcher/Tag pair 2, wherein pair 1 and pair 2 are orthogonal. More preferably each antigen is flanked by a SnoopCatcher- and SpyTag moiety, or a SnoopTag- and SpyCatcher moiety respectively. By iterative extension on the displayed Hbp fusion protein comprising the corresponding Catcher or Tag moiety, a string of compatible antigen modules can be stacked on the Hbp fusion protein. Ligation of multiple antigens via isopeptide bonding technology can be carried out as described in e.g. WO 2016/193746. Using these methods, a SpyCatcher- PspAa-SnoopTag fusion protein could be successfully ligated to a SnoopCatcher-SP1690-SpyTag fusion protein, thereby creating a string of multiple antigens suitable for ligation to cell surface expressed Hbp fusion protein.

Suitable lectins which can be coupled via isopeptide bonding technology to Gram-negative bacterial cells or OMVs displaying on their surface an autotransporter fusion protein as described herein can be HIV-1 envelope gp41 (Alfsen & Bomsel; J Biol Chem. 2002 Jul 12;277(28):25649-59), Ulex europaeus 1 (Giannasca et al; Am. J. Physiol., 267 (1994), pp. G1 108-G1 121 ), Wheat germ agglutinate (Ertl et al; J. Drug Targeting, 8 (2000), pp. 173-184), Tomato lectin (Hussain et al; Pharm. Res., 14 (1997), pp. 613-618 ), peanut agglutinin, asparagus pea lectin (Gupta et al; Int. J. Pharm., 318 (2006), pp. 163-173 ), Ateuria aurantia lectin (Roth-Walter et al; J. Allergy Clin. Immunol., 1 14 (2004), pp. 1362-1368), C-type lectin receptors, DC-SIGN, CD23, CLEC4G, DEC205 (Apostolopoulos et al; J Drug Deliv. 2013;2013:869718 ).

Suitable adhesins or affinity molecules which can be coupled via isopeptide bonding technology to Gram-negative bacterial cells or OMVs displaying on their surface a Hbp fusion protein as described herein can be any adhesin or affinity molecule that is capable of being attached to a moiety of a Catcher/Tag ligation pair. Suitable adhesin include bacterial adhesins such as FimH, PapG, PrsG, SfaS, FocH, FimD, MrpH, FaeG, FanC, CfaB, MrkD, pilin protein, PilC, CsgA, Afa-IE, DraA, NfaA, AIDA-I (Klemm & Schembri, Int J Med Microbiol. 2000 Mar;290(1 ):27-35), and also Claudin-4 (Rajapaksa et al; J Biol Chem (2010) 285(31 ):23739-46), Invasin (Marra & Isburg; Infect. Immun., 65 (1997), pp. 3412-3421 ), lnvasin-C192 (Hussain & Florence; Pharm. Res., 15 (1998), pp. 153-156), Long polar fimbria (Baumler et al; Proc. Natl. Acad. Sci. USA, 93 (1996), pp. 279-283), viral haemagglutinin σ1 protein (Rubas et al; J. Microencapsul., 7 (1990), pp. 385-395), or affinity molecules are antibodies, nanobodies, and the like. Other suitable affinity molecules are

immunoglobulin-binding bacterial proteins such as Streptococcal Protein A, Protein G or

Peptostreptococcus protein L, or immunoglobulin-binding fragments of these proteins. Preferably, the affinity molecule is protein A G, which is a recombinant fusion protein that combines IgG binding domains of Streptococcal aureus Protein A and G with a mass of 50 kDa.

The heterologous molecules of the invention can be coupled to the autotransporter fusion protein displayed on the surface of the Gram-negative bacterial cells or OMVs according to the invention by contacting the bacterial cell or the OMV with a heterologous molecule comprising the other binding moiety of said Catcher/Tag ligation pair to allow the formation of an isopeptide bond between the autotransporter fusion protein and the heterologous molecule. Thus, in a further aspect, the invention provides for a method to prepare Gram-negative bacterial cells or Outer Membrane Vesicles (OMVs) which display on their outer surface an autotransporter (AT) fusion protein covalently coupled via an isopeptide bond to a heterologous molecule, said method comprising;

- providing Gram-negative bacterial cells or Outer Membrane Vesicles (OMVs) which

display on their outer surface an autotransporter (AT) fusion protein comprising at least one moiety of a Catcher/Tag ligation pair, contacting the bacterial cells or the OMVs with a heterologous molecule comprising the other binding moiety of said Catcher/Tag ligation pair under conditions which allow the formation of an isopeptide bond between the autotransporter fusion protein and the heterologous molecule, and

recovering the bacterial cells or the OMVs .

To allow the formation of an isopeptide bond, the bacterial cells or OMVs can be suspended in isotonic buffers well known in the art, for example phosphate buffered saline (PBS) (pH 7.4), optionally supplemented with a cryopreservative such as glycerol (10-15%). Cell or OMV suspensions can be contacted with the heterologous molecules by mixing the cell or OMV suspensions with preparations of heterologous molecules solubilized in an isotonic buffer such as PBS, possibly supplemented with glycerol (10-15%), and incubated at temperatures ranging from 4°C to 37°C, preferably 4°C. A molar excess of heterologous molecules over Hbp-fusion protein may be used to achieve optimal decoration of cells or OMVs. Incubation times may be as short as 1 hour but preferably are longer, such as overnight to allow the formation of the isopeptide bond. Following incubation, the cells and OMVs can be recovered by separating the cells and OMVs from the incubation mixture using suitable separation techniques known in the art. The cell and OMV mixtures may be subjected low-speed (e.g. 14,000 x g) or high-speed (e.g. 208,000 x g) centrifugation, respectively, to separate the Cells and OMVs from the soluble mixture content. To remove non-covalently associated heterologous protein material, Cells and OMVs may be resuspended in fresh PBS, or PBS supplemented with 0.5M NaCI to break non- covalent electrostatic interactions, before being reisolated with low-speed or high-speed centrifugation. Covalently conjugating the heterologous molecules according to the invention via Catcher/Tag protein ligation to an autotransporter fusion protein already expressed on the surface of Gram- negative bacterial cells or OMVs allows development of a display platform by which attenuated bacterial cells or derived Outer Membrane Vesicles (OMVs) can be decorated with multiple heterologous antigens of choice for optimal stimulation of the immune system or with lectins, adhesins or affinity molecules of choice for optimal targeting to certain tissues or immune cells of interest. Thus, in a further embodiment the invention provides for a pharmaceutical composition comprising the Gram-negative bacterial cells or OMVs that display on their surface an autotransporter fusion protein covalently coupled to a heterologous molecule via isopeptide bond formation, wherein the

autotransporter protein comprises at least one moiety of a Catcher/Tag ligation pair, and the heterologous molecule comprises the other moiety of said Catcher/Tag ligation pair.

If the heterologous molecule is an antigen, the Gram-negative bacterial cells and OMVs described herein can be used as candidates for a vaccine. The present invention provides in particular for a vaccine comprising Gram-negative bacterial cells or OMVs that display on their surface a SPATE fusion protein, more particularly a Haemoglobin protease (Hbp) fusion protein, which is covalently coupled to a heterologous antigen via isopeptide bond formation, wherein the SPATE fusion protein, in particular the Hbp fusion protein comprises at least one moiety of a Catcher/Tag ligation pair, and the heterologous molecule comprises the other moiety of said Catcher/Tag ligation pair.

If the heterologous molecule is a lectin, adhesin or affinity molecule, such as an antibody. nanobody or the protein A/G fusion protein and the like, the Gram-negative bacterial cells and OMVs described herein can be used as a drug delivery system for optimal targeting of antigens or therapeutic agents to certain tissues or immune cells of interest. Of particular interest is the use of the bacterial cells and OMVs described herein for targeting of a therapeutic agent. Thus in yet another particular embodiment the present invention provides for a drug delivery system comprising Gram- negative bacterial cells or OMVs that display on their surface a SPATE fusion protein, more particularly a Haemoglobin protease (Hbp) fusion protein coupled to a heterologous adhesin or affinity molecule via isopeptide bond formation, wherein the SPATE fusion protein, in particular the Hbp fusion protein comprises at least one moiety of a Catcher/Tag ligation pair, and the heterologous adhesin or affinity molecule comprises the other moiety of said Catcher/Tag ligation pair.

It will be evident to the skilled person that the various embodiments and aspects of the invention which are described above in different paragraphs may be combined. The invention and certain embodiments are illustrated by the following examples and/or embodiments, without being limited thereto or thereby. Legend of the Figures

Figure 1. Schematic representations of the proteins used in this study. (A) Hbp fusions. Wild-type Hbp is synthesized with an N-terminal signal sequence (ss) that is cleaved off after translocation across the inner membrane. The C-terminal β-domain (black) integrates into the outer membrane facilitating translocation of the passenger domain. After translocation, autocatalytic cleavage separates the passenger and the β-domain (after Asn1 100). The passenger domain contains five subdomains (white, numbered 1 -5) protruding from a β-helical stem structure (dark grey). The derived HbpD(Adl ) display platform lacks subdomain 1 and the autocatalytic cleavage site (Jong et al (2012) supra). Ligation tags and catchers were integrated at the site of subdomain 1. (B) Catcher-fused model proteins. The SpyCatcher-SnoopCatcher fusion protein (SpC-SnC; SEQ ID NO: 14) was constructed by Veggiani et al. containing a 34 amino acid a-helical linker (Veggiani et al. 2016, supra).

GFPnanobody-SpyCatcher (GFPnb-SpC; SEQ ID NO: 15) contains an N-terminal PelB signal sequence (ss) (Kubala et al. 2010 Protein Science: a publication of Protein Society 19(12): 2389-401 ) for translocation into the periplasm and a hexa-histidine tag (H6) for metal affinity purification. (C) Catcher-fused antigens. Fusion proteins containing pneumococcal antigens PspAa and SP1690,

SpyCatcher (SpC) and SnoopTag (SnT) for protein ligation, an HA tag (HA) for detection and a hexa- histidine tag (H6) for metal affinity purification (SEQ ID NO: 17; SEQ ID NO: 24, resp). (D) Cartoon of Hbp-mediated Spy-ligation to the surface of outer membrane vesicles. HbpD(Ad1 )-SpT (colouring and numbering as in A) is embedded in the membrane of an outer membrane vesicle. The SpyTag has covalently bound to a chimera of the SpyCatcher and a cargo protein (model structure of SP1690) through the formation of an isopeptide bond between the SpyTag and the SpyCatcher. The protein structures were generated using PyMOL.

Figure 2. Ligation of OMVs displaying SpyTag using increasing amounts of SpyCatcher- SnoopCatcher (SpC-SnC). (A) SDS-PAGE/Coomassie staining analysis of reaction mixes of OMVs harbouring HbpD(Ad1 )-SpT and SpC-SnC incubated for 21 h at 4 °C. Reactant proteins, the adduct and the major outer membrane protein OmpA are indicated on the right-hand side of the panel. The sizes of the molecular weight markers are indicated on the left side of the panel. (B) Quantification of ligation efficiency to HbpD(Ad1 )-SpT. The intensities of the adduct and HbpD(Ad1 )-SpT in the gel shown in panel A were determined by densitometry. The percentage of HbpD(Ad1 )-SpT ligated with SpyCatcher-SnoopCatcher is plotted as a function of the molar ratio SpyCatcher- SnoopCatcher:HbpD(Ad1 )-SpT.

Figure 3. Ligation to OMVs displaying various Tag and Catcher moieties. Protein ligation of soluble model proteins to OMVs harbouring HbpD(Adl ) fusions containing SpyCatcher (SpC), SnoopCatcher (SnC), SpyTag (SpT), SnoopTag (SnT) or KTag ( T) was tested (lanes 1 -6). The soluble model proteins used are indicated on the left side of the -/+ matrix. Ligation of soluble protein partners independent of the OMV context was tested for comparison (lanes 7-9). Reaction mixes were incubated for 24 h at 4 °C and analysed by SDS-PAGE with Coomassie staining. Reactant proteins are indicated with dots in the image and on the right-hand side of the panel (H, HbpD(Adl ) fusion; M, SpT/SnT-MBP; G, SnT-mEGFP-SpT; CC, SpC-SnC; S, SUMO-KT; L, SpyLigase), adducts are indicated with arrowheads. The ligation efficiencies (percentage of HbpD[Ad1] or SpT/SnT-MBP converted) are shown below the lanes. The sizes of the molecular weight markers are indicated on the left side of the panel.

Figure 4. Spy-ligation of pneumococcal antigens to OMVs. OMVs harbouring HbpD(Ad1 )-SpT were mixed with a 5- to 10-fold molar excess of SpC-PspAa'-SnT (α'), SpC-SP1690-SnT (1690) or SpC- PspAa'-SP1690-SnT (a'-1690). The reaction mixes were incubated for 24 h at 4 °C and analysed by SDS-PAGE with Coomassie staining. Reactant proteins are indicated on the right-hand side of the panel. Adducts are indicated with arrowheads. The sizes of the molecular weight markers are indicated on the left side of the figure.

Figure 5. Adduct extension by alternating Spy-ligation and Snoop-ligation. OMVs harbouring

HbpD(Ad1 )-SpT were mixed with a 1.5-fold molar excess of SpC-PspAa'-SnT and incubated for 2 h at 4 °C. Subsequently a similar amount of SnC-SP1690-SpT was added. After 2 h at 4 °C again the same amount of SpC-PspAa'-SnT was added and after another 2 h at 4 °C same amount of SnC- SP1690-SpT was added. Finally, the mix was incubated for a further 15 h at 4 °C to allow completion of the reaction. The OMVs and suspending buffer were separated by centrifugation. After

centrifugation the supernatant (ligation— sol) and the pellet fraction (ligation— OMV) and were analysed by SDS-PAGE with Coomassie staining (A, lanes 4 and 5 respectively). (B) Immunoblotting analysis of OMV pellet fraction after ligation and reference OMVs as described under A. Monoclonal antibodies were used for immunodetection of the HA-tag in SpC-PspAa'-SnT and of the FLAG-tag in SnC-SP1690-SpT. The adducts are indicated with arrowheads (HbpD-P, HbpD(Ad1 )-SpT-SpC- PspAa'-SnT; HbpD-P-S, HbpD(Ad1 )-SpT-SpC-PspAa'-SnT-SnC-SP1690-SpT; HbpD-P-S-P, HbpD(Ad1 )-SpT-SpC-PspAa'-SnT-SnC-SP1690-SpT-SpC-PspAa'-SnT). The sizes of the molecular weight markers are indicated on the left side of the panels.

Figure 6. Binding of GFP to bacterial cells upon ligation and display of a GFP nanobody. (A) S.

Typhimurium cells expressing HbpD(Ad1 )-SpT were incubated with GFPnb-SpC or with the unreactive GFPnb-SpC EQ for 60 min at 25 °C. The cells were subsequently incubated for 5 min at 4 °C in the presence of GFP and analysed by SDS-PAGE with Coomassie staining (A). HbpD(Ad1 )-SpT is indicated on the right hand side of the panel, the adduct is indicated with an arrowhead. The sizes of the molecular weight markers are indicated on the left side of the panel. (B) Fluorescence microscopy analysis of cells produced under A. Phase-contrast images (upper panels) and the corresponding fluorescence images (lower panels) are shown. The lower right panel, showing the fluorescence of the cells incubated with GFPnb-SpC EQ, is split into two parts. In the upper left half the contrast is enhanced to visualize the weak fluorescence signal (autofluorescence of the cells), while the lower right half shows the (absence of) signal at the same settings as the left panel.

Figure 7. Analysis of OMVs and purified proteins by SDS-PAGE with Coomassie staining. A) S.

Typhimurium derived OMVs harboring HbpD(Ad1 )-SpyCatcher (SpC), -SnoopCatcher (SnC), -SpyTag (SpT), -SnoopTag (SnT) or -Ktag (KT). The HbpD(Adl ) variants and the major outer membrane proteins OmpA, OmpC and OmpF are indicated on the right-hand side of the panel. B) SUMO-KTag (S-KT), SnoopTag-mEGFP-SpyTag (GFP), SpyLigase (SpL), SpyCatcher-SnoopCatcher (SpC-SnC), SpyTag-MBP (SpT-MBP) and SnoopTag-MBP (SnT-MBP) as purified from E. coli BL21 (DE3) cells. The sizes of the molecular weight markers are indicated on the left side of the panels.

Figure 8. Surface exposure analysis of HbpD(Adl ) fusion proteins carrying Tag and Catcher moieties. The exposure of the HbpD(Adl ) fusion proteins containing SpyCatcher (SpC), SnoopCatcher (SnC), SpyTag (SpT), SnoopTag (SnT) or KTag (KT) on the exterior surface of the OMVs was analysed by Proteinase K treatment. OMVs were diluted in 50 mM Tris.CI, pH 7.5, 1 mM CaC to obtain a concentration of approximately 2 pmol of HbpD(Adl ) fusion protein per μΙ_. The suspension was split into three aliquots. To one aliquot Triton X-100 was added to a concentration of 1 % to dissolve the OMVs. After incubating all aliquots on ice for 15 min, Proteinase K was added (0.1 mg/mL final concentration) to the dissolved membranes and to one of the other two aliquots. After 30 min of incubation at 37 °C, PMSF was added to 0.2 mM. The reaction was stopped by the addition of phenylmethanesulfonylfluoride (PMSF) to 0.2 mM and incubation on ice for 10 min. The samples were analysed by SDS-PAGE with Coomassie staining. Successful exposure of HbpD(Adl ) fusion proteins follows from their degradation upon treatment of OMVs with Proteinase K. A Proteinase K sensitive intracellular loop of OmpA is not accessible under these conditions, unless the OMVs were dissolved with Triton X-100 first, indicating that OMVs remained intact throughout the procedure. Full-length HbpD(Adl ) fusion proteins, OmpA and Proteinase K, as well as degradation products of the

HbpD(Adl ) fusion proteins (Δ1 ) and of OmpA (Δ2) are indicated on the right side of the panel. The sizes of the molecular weight markers are indicated on the left.

Figure 9. Presence of the Hise-tag in the HbpD(Ad1 )-SpT-SpC-GFPnb adduct. S. Typhimurium

SL3261 AtoIRA derived OMVs harbouring HbpD(Ad1 )-SpT were incubated with GFPnb-SpC or with the unreactive GFPnb-SpC EQ for 60 min at 25 °C and analysed by SDS-PAGE followed by

Coomassie staining (A) or by electroblotting onto a nitrocellulose membrane, Ponceau S staining and immunodetection using anti-polyHistidine antibody (B). The C-terminal His6-tag of GFPnb-SpC was detected in HbpD(Ad1 )-SpT-SpC-GFPnb. HbpD(Ad1 )-SpT and the adduct are indicated on the right- hand side of the panels. The sizes of the molecular weight markers are indicated on the left.

Figure 10. Adduct extension by alternating Spy-ligation and Snoop-ligation in the absence of OMVs harboring HbpD(Ad1 )-SpT. Equimolar amounts of SpC-PspAa'-SnT and SnC-SP1690-SpT were mixed and incubated for 2 hours at 25°C. The reaction mix and the purified proteins separately were analyzed by SDS-PAGE with Coomassie staining. For comparison, to show the similarity, the "sol" sample of Figure 5A is shown in lane 4. The "sol" sample is the supernatant after centrifugation of the reaction mix containing OMVs harboring HbpD(Ad1 )-SpT and SpC-PspAa'-SnT and SnC-SP1690-SpT (for details see Figure 5A). The sizes of the molecular weight markers are indicated on the left.

Figure 11. Coupling of rat antibody to the surface of bacterial cells or OMVs harboring HbpD(Adl )- SpC. (A) cartoon of anti-mouse CD180 antibody comprising heavy chain Fc domains carrying a SpyTag at the extreme C-terminus. (B) SDS-PAGE analysis of cells displaying SL3261A/o/R l- HbpD(Ad1 )-SpC before addition of aCD180-SpT (lane 1 ), of the crude SL3261Afo/R/4-HbpD(Ad1 )- SpC + aCD180-SpT mix (lane 2) and pellet (p) and supernatant (s) after centrifugation (lanes 3 and 4) and PBS washing (lanes 5 and 6). Molecular weight markers (kDa) are indicated at the left side of the panel. (C) SDS-PAGE analysis of OMVs of SL3261Afo/R>4-HbpD(Ad1 )-SpC before addition of aCD180-SpT of aCD180-SpT (lane 1 ). Samples of the crude SL3261Afo/R>A-HbpD(Ad1 )-SpC OMVs + aCD180-SpT mix (lane 2) and pellet (p) and supernatant (s) after centrifugation (lanes 3 and 4) and Salt washing (lanes 5 and 6) were analyzed in parallel by Coomassie stained SDS-PAGE. Molecular weight markers (kDa) are indicated at the left side of the panel. Figure 12. Median Fluorescence intensity per HeLa cell. Mammalian HeLa cells expressing murine CD180 (Hela/CD180) were incubated with bacterial cells harboring SL3261A/o/R \-HbpD(Ad1 )-SpC carrying covalently coupled aCD180-SpT. HeLa cells not expressing murine CD180 (HeLa) were used as a control. The HeLa cells, grown on IBIDI microscopy slides, were incubated with various loads of SL3261Aio/R/ -HbpD(Ad1 )-SpC-aCD180-SpT (multiplicity of infection, MOI: 5000, 500, 50 or 5) (2h, 37°C, 8% C0 ₂ ).

Figure 13. Coupling of Catcher-equipped model proteins to Tag sequences inserted at position d2 and d4 of HbpD. (A) Schematic representation of wild-type Hbp lacking the autocatalytic cleavage site (HbpD) and Hbp fusion proteins carrying a SpyTag (SpT; flanked by GSGSS and GSGSG linkers), a SpyTag with extended linkers (EL.SpT; flanked by GSGSSGSASG and GEGTGGSGSG linkers), or a SnoopTag (SnT; flanked by GSGSS and GSGSG linkers) at either position d2 (HbpD(Ad2)) or d4 (HbpD(Ad4) of the passenger domain. (B) SDS-PAGE/Coomassie blue analysis of covalent coupling of the model protein comprising a SpyCatcher-SnoopCatcher fusion (SpC-SnC) to OMVs displaying the respective HbpD variants on the surface. The adduct (>) corresponds with a covalent fusion product between the (EL.)SpT-carrying HbpD-variants and the SpyCatcher containing model protein. (C) SDS-PAGE/Coomassie blue analysis of covalent coupling of the model protein comprising a SpyCatcher-SnoopCatcher fusion (SpC-SnC) to OMVs displaying HbpD(Ad2)-SnT. The adduct (*) corresponds with a covalent fusion product between the SnT-carrying HbpD-variant and the

SnoopCatcher containing model protein.

Figure 14. Improved display of a complex protein using isopeptide bonding technology (A)

Swissmodel (https://swissmodel.expasy.org/interactive) structure of S. pneumoniae TIGR4 antigen SP1690 (UniProt KB A0A0H2URB7). (B). SDS-PAGE and Coomassie blue staining analysis of S. Typhimurium SL3261 AtoIRA OMVs displaying HbpD(Adl ), HbpD-SP1690-F1-F2 or HbpD-SP1690-F3- F4. (C) SDS-PAGE and Coomassie blue staining analysis of coupling of SP1690 carrying an N- terminal SpyCatcher (SpC-SP1690-SnT) to S. Typhimurium 513261^0^ OMVs displaying

HbpD(Adl ) equipped with an N-terminal SpyTag (HbpD(Adl )-SpT). Molecular weight markers (kDa) are indicated at the left side of the panels. Bands corresponding with endogenous Salmonella outer membrane proteins (OMPs) are indicated. Of note, the folding state of bacterial OMPs is known to be sensitive to subtle changes in SDS concentration and heat encountered during SDS-PAGE analysis (Burgess et al; J Biol Chem. 2008 Sep 26;283(39):26748-58). This explains the varying migration behaviors of the indicated OMPs in panels B and C, respectively.

Figure 15. Coupling of an adhesin to the surface of Gram-negative bacteria. SDS-PAGE and

Coomassie blue staining analysis of E. coli TOP10F' cells displaying HbpD(Adl ) with a SpyTag (HbpD(Adl )-SpT) following incubation in the absence (-) or the presence (+) of FIITIHL equipped with a SpyCatcher (FimH _L -SpC).

Examples

General methods and material Strains. E. coli BL21 (DE3), S. Typhimurium SL3261 (Hoiseth et al. 1981 Nature 291 (5812): 238-9) and the isogenic SL3261 AtoIRA (Daleke-Schermerhorn et al 201 , supra) were grown in Lysogeny Broth (LB; 10 g/L tryptone, 5 g/L yeast extract, 10 g/L NaCI).

S. Typhimurium SL3261 AtoIRA AmsbB (Kuipers et al. 2017 Infect Immun 85(10)) was grown in TYMC (10 g/L tryptone, 5 g/L yeast extract, 2 mM MgS04, 2 mM CaC ). The growth medium was supplemented with 0.2% of glucose. Where appropriate antibiotics were added to the following concentrations: ampicillin, 100 μg/mL; kanamycin, 50 g/mL; chloramphenicol 30, μg/mL. Unless stated otherwise cultures were incubated at 37 °C with shaking.

Construction of expression plasmids: pET28a SpyCatcher-SnoopCatcher, pET28a-SpyTagMBP, pET28a SnoopTag-MBP, pET28a SnoopTag-mEGFP-SpyTag, pET28a SUMO-KTag and pDEST14 SpyLigase were obtained from Addgene. Table 1 gives an overview of the expression plasmids used in the examples. DNA primers which have been used are listed in Table 2.

Table 1. Overview of the plasmids used in the examples.

Name Protein expressed Hise Reference tag*

pET28a SpyCatcher- SpyCatcher-SnoopCatcher fusion linked via a L9 N Veggiani ef SnoopCatcher alpha-helix linker al. supra,

Addgene

#72324 pET28a-SpyTagMBP SpyTag-MBP Zakeri ef al. supra,

Addgene

#35050 pET28a SnoopTag-MBP SnoopTag-MBP Veggiani ef supra,

Addgene

#72323 pET28a SnoopTag- SnoopTag-mEGFP-SpyTag Veggiani ef mEGFP-SpyTag supra,

Addgene

#72325 pET28a SUMO-KTag SUMO-KTag Fierer ef al.

supra,

Addgene

#51723 pDEST14 SpyLigase SpyLigase Fierer et al.

supra,

Addgene

#51722 pHbpD(Ad1 )-SpT Hbp display platform with the SpyTag at the site of this work domain 1 (amino-terminus)

pHbpD(Ad1 )-SnT Hbp display platform with the SnoopTag at the site this work of domain 1 (amino-terminus)

pHbpD(Ad1 )-SpC Hbp display platform with the SpyCatcher at the this work site of domain 1 (amino-terminus)

pHbpD(Ad1 )-SnC Hbp display platform with the SnoopCatcher at the this work site of domain 1 (amino-terminus)

pHbpD(Ad1 )-KT Hbp display platform with the KTag at the site of this work domain 1 (amino-terminus) pET22b GFPnb PelB signal sequence (PelBss)-GFP-binding Kubala et al.

nanobody (GFPnb) supra pET22b GFPnb-SpC PelBss-GFPnb-SpyCatcher fusion C this work pET22b GFPnb-SpC EQ PelBss-GFPnb fused to SpyCatcher E77Q mutant C this work

(E77 is a key residue for isopeptide bond

formation)

pET22b GFPnb-SpT PelBss-GFPnb-SpyTag fusion c this work pET22b GFPnb-SnT PelBss-GFPnb-SnoopTag fusion c this work pET28 SpC-PspAa-SnT S pyCatch er- H A- Ps p A32-244-S n oopTag c this work pET28 SnC-PspAa-SpT SnoopCatcher-FLAG-PspA32-244-SpyTag c this work pET28 SpC-SP1690-SnT SpyCatcher-HA-SP1690-SnoopTag c this work pET28 SnC-SP1690-SpT SnoopCatcher-FLAG-SP1690-SpyTag c this work pET20b GFP-His6 GFP-Hise c kind gift from

J.W. de Gier pET22b FimHL-SpC PelBss-FimHL-SpyCatcher This work pcDNA3.3 aCD180Hc aCD180 heavy chain-SortaseTag-SpyTag Gift C. Kuijl pcDNA3.3 aCD180 _L c aCD180 light chain Gift C. Kuijl pHbpD(Ad2)-SnT Hbp display platform with the SnoopTag at the site This work of domain 2

pHbpD(Ad2)-SpT Hbp display platform with the SpyTag at the site of This work domain 2

pHbpD(Ad4)-SpT Hbp display platform with the SpyTag at the site of This work domain 4

pHbpD(Ad2)-EL.SpT Hbp display platform with the SpyTag at the site of This work domain 2, flanked by extended linkers

pHbpD(Ad4)-EL.SpT Hbp display platform with the SpyTag at the site of This work domain 4, flanked by extended linkers

* N, Hise-tag on the amino-terminus; C, His6-tag on the carboxy-terminus

To construct pHbpD(Ad1 )-SpC, a SpyCatcher-encoding fragment was amplified by PCR using pET28a-SpyCatcher-SnoopCatcher as a template and the primers SpyCat fw and SpyCat rv. Using the In-Fusion method (Clontech), the resulting fragment was cloned into Sacl/BamHI digested pHbpD(Ad1 )-Ag85B, a derivative of pHbpD(Ad1 )-ESAT6 carrying mycobacterial ag85B between the Sacl/BamHI restriction sites (Jong et al. 2012 supra). This yielded pHbpD(Ad1 )-SpC. The same strategy was followed for construction of pEH3-HbpD(Ad1 )-SnC but now the primers used were

SnoopCat fw and SnoopCat rv.

For construction of pHbpD(Ad1 )-SpT, a PCR fragment was generated encoding a fusion product between an N-terminal segment of Hbp and a down-stream SpyTag. Plasmid pEH3-Hbp (Jong et al. 2007 supra) was the template and the primers used were pEH_Xba_lnFu fw and Hbp-SpyTag rv. . The product was cloned into the Xbal/BamHI sites of pHbpD(Ad1 )-Ag85B using In-Fusion

methodology to yield pHbpD(Ad1 )-SpT. The same strategy was used to create pHbpD(Ad1 )-KT but now using the primer pair pEH_Xba_lnFu fw and Hbp-KTag rv. To construct pHbpD(Ad1 )-SnT, a SnoopTag-encoding fragment was generated by annealing the long oligo's SnoopTag S/B fw and

SnoopTag S/B rv. The resulting product was ligated into the Sacl/BamH sites of pHbpD(Ad1 )-Ag85B, yielding pHbpD(Ad1 )-SnT. The same annealed oligo product was ligated into the Sacl/BamH sites of pHbpD(Ad2)-ESAT6 (Jong et al., Microb Cell Fact. 2012 Jun 18;1 1 :85) to yield pHbpD(Ad2)-SnT. To construct pHbpD(Ad2)-SpT and pHbpD(Ad4)-SpT, a SpyTag-encoding fragment was generated by annealing the long oligo's SpyTag S/B fw and SpyTag S/B rv. The resulting fragment was ligated into the Sacl/BamHI sites of pHbpD(Ad2)-ESAT6 and pHbpD(d4in) (Jong et al., Microb Cell Fact. 2014 Nov 25; 13: 162), yielding pHbpD(Ad2)-SpT and pHbpD(Ad4)-SpT, respectively. To construct pHbpD(Ad2)-EL.SpT and pHbpD(Ad4)-EL.SpT the same strategy was followed using the long primers EL.SpyTag S/B fw and EL.SpyTag S/B rv.

pET22b GFPnb (kind gift from Hansjbrg Gbtzke, Center for Biomembrane Research, Department of Biochemistry and Biophysics, Stockholm University, Stockholm, Sweden).

pET22b GFPnb-SpC: the gene encoding a GFP-nanobody (GFPnb) was PCR amplified from pET22b GFPnb using primers ACYC Duet UP1 and GFPnb rv. The SpyCatcher-encoding sequence was PCR amplified from pET28a SpyCatcher-SnoopCatcher using primers SpyC fw and SpyC rv introducing a carboxy-terminal HiS6-tag. After overlap PCR the GFPnb-SpC sequence was cloned Xbal-Hindlll into pET22b (Novagen) yielding pET22b GFPnb-SpC.

pET22b GFPnb-SpC EQ: The E77Q codon mutation was introduced by PCR amplification from pET28a SpyCatcher-SnoopCatcher using either primers SpyC fw and SpyC EQ rv or SpyC EQ fw and SpyC rv followed by overlap PCR using both PCR products. The final product was cloned KpnI-Hindlll into pET22b GFPnb-SpC, replacing SpC, yielding pET22b GFPnb-SpC EQ. pET22b GFPnb-SpT and pET22b GFPnb-SnT: Oligonucleotides SpT-H6 us and SpT-H6 Is or SnT-H6 us and SnT-H6 Is were mixed at 0.25 mM concentration in 100 mM potassium acetate, 30 mM HEPES, pH7.5, heated to 100 °C and allowed to gradually cool down to ambient temperature. The annealed oligonucleotides were cloned into KpnI-Hindlll opened pET22b GFPnb-SpC yielding pET22b GFPnb-SpT and pET22b GFPnb-SnT respectively. pcDNA3.3 aCD180 _H c and pcDNA3.3 aCD180 _L c: a synthetic DNA fragment was obtained (IDT) encoding the heavy chain of mouse CD180-directed rat lgG2a antibodies carrying a SortaseTag and a SpyTag at the C-terminus (SEQ ID NO: 27). Similarly, a synthetic DNA fragment was obtained (IDT) encoding the light chain of mouse CD180-directed rat lgG2a antibodies. To allow antibody production according to the method by Vink et al. (Methods. 2014 Jan 1 ;65(1 ):5-10), both fragments were ligated under control of a CMV promoter into a derivative of mammalian expression vector pcDNA3.3, yielding pcDNA3.3 CICD180HC and pcDNA3.3 aCD180i_c, respectively pcDNA3.3 aCD180LC: a synthetic DNA fragment was obtained (IDT) encoding the light chain of mouse CD180-directed rat lgG2a antibodies. To allow antibody production according to method by Vink et al. (Methods. 2014 Jan 1 ;65(1 ):5-10), the fragment was ligated into mammalian expression vector pcDNA3.3, yielding pcDNA3.3 O,CD180HC pET22b FimHL-SpC: a synthetic DNA fragment was obtained (IDT) encoding a FimHi_ (residues 23- 181 of SEQ ID NO: 30). The fragment was ligated into pET22b GFPnb-SpC using In-Fusion cloning, yielding pET22b- FimH_-SpC, which encodes a translational fusion between PelBss, FiimH_,

SpyCatcher and a Hise tag (SEQ ID NO: 30). pET28 SpC-PspAa'-SnT, pET28 SnC-PspAa'-SpT, pET28 SpC-SP1690-SnT and pET28 SnC- SP1690-SpT: The SpyCatcher-encoding sequence was PCR amplified from pET28a SpyCatcher- SnoopCatcher using primers Ncol SpyC fw and EcoRI HA SpyC rv introducing an HA tag. The product was cloned Ncol-EcoRI into pET28a yielding pET28 SpC-HA. Next, a SnoopTag-HiS6-encoding sequence was PCR amplified from pET22b GFPnb-SnT using primers Sail SpT fw and T7 Terminator. The product was cloned Sall-Xhol into pET28 SpC-HA yielding pET28 SpC-HA-SnT-Hise. The

SnoopCatcher-encoding sequence was PCR amplified from pET28a SpyCatcher-SnoopCatcher using primers Ncol SnC fw and EcoRI FLAG SnC rv introducing a FLAG tag. The product was cloned Ncol- EcoRI into pET28a yielding pET28 SnC-FLAG. Next, a SnoopTag-His6-encoding sequence and a SpyTag-HiS6-encoding sequence were PCR amplified from pET22b GFPnb-SnT and pET22b GFPnb- SpT respectively using primers Sail SpT fw and T7 Terminator. The products were cloned Sall-Xhol into pET28 SpC-HA and pET28 SnC-FLAG yielding pET28 SpC-HA-SnT-Hise and pET28 SnC-FLAG- SpT-Hise respectively. Last, DNA encoding amino acids 32-244 of PspA or encoding amino acids 24- 445 of SP1690 (both Streptococcus pneumoniae serotype 4 [TIGR4] sequences), were PCR amplified using primers EcoRI aTIGR4 fw and Sail aTIGR4 rv or EcoRI SP1690 fw and Sail SP1690 rv

respectively. The products were cloned EcoRI-Sall into pET28 SpC-HA-SnT-His6 and pET28 SnC- FLAG-SpT-Hise yielding pET28 SpC-alpha-SnT, pET28 SnC-alpha-SpT, pET28 SpC-SP1690-SnT and pET28 SnC-SP1690-SpT.

Table 2. DNA primers used in the examples.

Name 5'-3' sequence*

ACYC Duet UP1 GGATCTCGACGCTCTCCCT

GFPnb rv cqqtaccaccqctqcctTTGCTGCTAACGGTAACC

SpyC fw qcaqcqqtqqtaccqqcGATAGTGCTACCCATATTAAATTCTC

SpyC rv GCGGCCGCAAGCTTTTACTAATGATGGTGATGATGATGaccgctgccAATAT

GAGCGTCACC

SpyC EQ fw GAAAATATACATTTGTCcAAACCGCAGC

SpyC EQ rv GCTGCGGTTTgGACAAATGTATA I I M C

SpT-H6 us cggcgcccacatcgtgatggtggacgcctacaagccgacgaagggtagtggtgaaagtgg tCATCAT

CATCACCATCATTAGTAAA

SpT-H6 Is AGC I I I I ACTAATGATGGTGATGATGATGaccactttcaccactacccttcgtcggcttgta ggcgtccaccatcacgatgtgggcgccggtac

SnT-H6 us cggcAAACTGGGCGATATTGAA I I I ATTAAAGTGAACAAAggtagtggtgaaagtgg tCATCATCATCACCATCATTAGTAAA

SnT-H6 Is AGC I I I I ACTAATGATGGTGATGATGATGaccactttcaccactaccTTTGTTCACT

TTAATAAATTCAATATCGCCCAGTTTgccggtac

Ncol SpyC fw atatatccatgggCGATAGTGCTACCCATATTAAATTCTC

EcoRI HA SpyC rv atatatgaattcGCCGGACCCCGCATAGTCAGGAACATCGTATGGGTATCCCG

AACCAATATGAGCGTCACC I I I AG

Sail SpT fw atatatgtcgacggcagcggtggtacc

T7 Terminator GCTAGTTATTGCTCAGCGG

Ncol SnC fw atatatccatgggCAAGCCGCTGCGTGG

EcoRI FLAG SnC rv atatatgaattccCCGCCGCTACCGCCTTTATCGTCATCATCCTTATAGTCACCG

CCGCTACCGCCTTTCGGCGGTATCGG

Sail SpT fw atatatgtcgacggcagcggtggtacc

EcoRI QTIGR4 fw atatatgaattcGAAGAAAGTCCGCAGGTTG

Sail QTIGR4 rv atatatgtcgacGGTGCCATCATCCGG

EcoRI SP1690 fw atatatgaattcTCAG G AAAAAAAG AAG CT ACAACTAGTAC

Sail SP1690 rv atatatgtcgacCTGAACAGCCTCAAATAAATCA I I I AATTG

SpyCat fw ggaagfcffgcggggagcfccGATAGTGCTACCCATATTAAATTC

SpyCat rv taccgctgccggatccAAJ ATGAGCGTCACCTTTAGTtg SnoopCat fw ggaagfcffgcggggagcfccAAGCCGCTGCGTGGTGCC

SnoopCat rv faccgcfgccggafccTTTCGGCGGTATCGGTTCATTG

pEH_Xba_lnFu fw ttgctaactttctagaitacaaaac

Hbp-SpyTag rv faccgcfgccggafcccttcgtcggcttgtaggcgtccaccatcacgatgtgggcggagc tccccgcaagac ttc

Hbp-KTag rv faccgcfgccggafccatcacgttttgagaatttaatatgggtagcggagctccccgcaa gacttc SnoopTag S/B fw CCAAACTGGGCGATATTGAATTTATTAAAGTGAACAAAG

SnoopTag S/B rv GATCCTTTGTTCACTTTAATAAATTCAATATCGCCCAGTTTGGAGCT SpyTag S/B fw CCGCCCACATCGTGATGGTGGACGCCTACAAGCCGACGAAGG

SpyTag S/B rv GATCCCTTCGTCGGCTTGTAGGCGTCCACCATCACGATGTGGGCGGAGC

T

EL.SpyTag S/B fw CCGGCTCGGCTAGCGGTGCCCACATCGTGATGGTGGACGCCTACAAGC

CGACGAAGGGTGAGGGAACCGGCG

EL.SpyTag S/B rv GATCCGCCGGTTCCCTCACCCTTCGTCGGCTTGTAGGCGTCCACCATCA

CGATGTGGGCACCGCTAGCCGAGCCGGAGCT

* Overlapping regions for overlap PCR are underlined; restriction enzyme recognition sites used for cloning are in bold and overhangs used for In-Fusion cloning are in italics; mutagenic nucleotides are in lower case.

Table 3. List of sequences

name SEQ ID No

Hbp (wild-type) 1

HbpD 2

HbpD(Adl ) 3

HbpD(Ad1 )-SpT 4

HbpD(Ad1 )-SnT 5

HbpD(Ad1 )-KT 6

HbpD(Ad1 )-SpC 7

HbpD(Ad1 )-SnC 8

HbpD(Ad2)-SpT 9

HbpD(Ad2)-EL.SpT 10

HbpD(Ad2)-SnT 1 1

HbpD(Ad4)-SpT 12

HbpD(Ad4)-EL.SpT 13

SpC-SnC 14

GFPnanobody-SpC 15

GFPnanobody-SpC_EQ 16

SpC-PSpAa-SnT 17

SpC-PspAa-SP1690-SnT 18

SpT-MBP 19

SnT-MBP 20

SnT-EGFP-SpT 21

SUMO-KT 22

SpyLigase 23

SpC-SP1690-SnT 24

SnC-SP1690-SpT 25

GFP-Hise 26

Rat-lgG2A_heavy_chain_constant_region_aCD180- 27

SpT HbpD-SP1690-F1 -F2 28

HbpD-SP1790-F3-F4 29

FimHL-SpC 30

ACYC Duet UP1 31

GFPnb rv 32

SpyC fw 33

SpyC rv 34

SpyC EQ fw 35

SpyC EQ rv 36

SpT-H6 us 37

SpT-H6 Is 38

SnT-H6 us 39

SnT-H6 Is 40

Ncol SpyC fw 41

EcoRI HA SpyC rv 42

Sail SpT fw 43

T7 Terminator 44

Ncol SnC fw 45

EcoRI FLAG SnC rv 46

Sail SpT fw 47

EcoRI QTIGR4 fw 48

Sail aTIGR4 rv 49

EcoRI SP1690 fw 50

Sail SP1690 rv 51

SpyCat fw 52

SpyCat rv 53

SnoopCat fw 54

SnoopCat rv 55

pEH_Xba_lnFu fw 56

Hbp-SpyTag rv 57

Hbp-KTag rv 58

SnoopTag S/B fw 59

SnoopTag S/B rv 60

SpyTag S/B fw 61

SpyTag S/B rv 62

EL.SpyTag S/B fw 63

EL.SpyTag S/B rv 64

Protein purification - cytoplasmic proteins. E. coli BL21 (DE3) cells harboring a pET28 or pDEST14 expression plasmid were grown in LB containing 0.2% glucose and kanamycin to an OD600 of 0.4-0.5. Protein expression was induced by the addition of isopropyl β-D-thiogalactopyranoside (IPTG) to a final concentration of 0.5 mM and the cells were incubated for a further two hours.

Subsequently the cells were washed with PBS (pH 7.4) and stored at -20 °C. The cells were resuspended in buffer A (50 mM NaPC , 300 mM NaCI, pH 7.4) and phenylmethylsulfonyl fluoride (PMSF) was added to a concentration of 125 μΜ. The cells were disrupted by two passages through a One Shot cell disruptor (Constant Systems Ltd.) at 1.2 kbar. Cell debris and membranes were removed by centrifugation at 10,000g and 293,000ο; respectively at 4 °C. His6-tagged proteins were isolated from the cleared lysate using TALON Superflow medium (GE Healthcare Life Sciences) according to the manufacturer's instructions. Eluates were dialyzed overnight at 4°C against 500 to 1000 volumes of PBS (pH 7.4). After dialysis glycerol was added to 10% and aliquots were stored at - 80 °C.

E. coli BL21 (DE3) cells harboring a pET20b GFP-His6 expression plasmid were grown at 30 °C in LB containing 0.2% glucose and ampicillin to an ODeoo of 0.2-0.3. Protein expression was induced by the addition of IPTG to a final concentration of 0.4 mM and the cells were incubated for a further four hours. GFP-His6 was purified from the cells as described above.

Protein purification - periplasmic proteins. A procedure based on that described by Pardon ef a/, in Nature Protocols was used. ³⁸ E. coli BL21 (DE3) cells harboring a pET22b GFPnb-SpC or pET22b GFPnb-SpC EQ expression plasmid were grown in LB containing 0.2% glucose and ampicillin to an ODeoo of 0.8-0.9. Both fusion proteins were equipped with a cleavable PelB signal sequence for translocation into the periplasmic space. Isolation of the GFPnb-SpC fusions from the oxidizing periplasmic environment allowed formation of an intramolecular disulphide bond ensuring nanobody functionality. Protein expression was induced by the addition of IPTG to a final concentration of 0.5 mM and the cells were incubated for a further 20 hours at 12 °C with shaking. The cells were harvested by centrifugation and stored at -20 °C. The cells were resuspended in PBS (pH 7.4) and incubated for 30 min at 21 °C in a Thermomixer (Eppendorf) at 1400 rpm. Next, the cells were removed by centrifugation (10,000g) and the His6-tagged proteins were isolated as described above (Protein purification - cytoplasmic proteins).

OMV isolation. S. Typhimurium SL3261 toIRA cells harboring the HbpD(Adl ) expression plasmid were grown in LB containing glucose and kanamycin to an ODeoo of 0.2-0.3. Protein expression was induced by the addition of IPTG to a final concentration of 0.1 mM and the cells were incubated for a further two hours. Cells were removed by two successive centrifugation steps at 5000g. OMVs were isolated from the second supernatant by centrifugation at 235,000g, resuspended in PBS (pH 7.4) containing 15% glycerol and stored at -80 °C. The same strategy was followed for the production of OMVs from S. Typhimurium SL3261 AtoIRA cells harboring plasmids for the expression of

HbpD(Ad2)-SpT, HbpD(Ad2)-EL.SpT HbpD(Ad2)-SnT HbpD(Ad4)-SpT and HbpD(Ad4)-EL.SpT, HbpD(Ad1 )-SpC, HbpD(Ad1 )-SnT, HbpD(Ad1 )-SnC and HbpD(Ad1 )-KT.

Alternatively, in view of future therapeutic purposes, OMVs were isolated from S. Typhimurium SL3261 Δίο/RiA msbB cells harboring the pHbpD(Ad1 )-SpT expression plasmid. These cells were grown at 30 °C in TYMC containing glucose and kanamycin. Fresh medium containing 1 mM of IPTG was inoculated to an Οϋεοο of 0.02 and incubated at 30 °C, shaking, for 17 hours. Cells were removed by two successive centrifugation steps at 5000g. The supernatant was passed through 0.45 m pore size filters (Millipore) and centrifuged at 235,000g for 75 min to sediment the OMVs. The vesicles were resuspended in PBS (pH 7.4) containing 15% glycerol. Spy-ligation with OMVs containing Hbp(Ad1)-SpT and increasing amounts of SpC-SnC. To

OMVs (isolated from S. Typhimurium SL3261 AtoIRA cells) containing approximately 62 pmol of Hbp(Ad1 )-SpT purified SpC-SnC was added in the ratios indicated in Figure 2A. The mixes were supplemented with PBS (pH 7.4) to reach a final volume of 20 μΙ_. After 21 hours of incubation at 4 °C the reaction mixes were analyzed by SDS-PAGE with Coomassie staining.

Protein ligation with Hbp(Ad1)-SpC, -SnC, -SpT, -SnT or -KT in OMVs. To OMVs (isolated from S. Typhimurium SL3261 AtoIRA cells) containing approximately 200 pmol of Hbp(Ad1 )-SpC, -SnC, -SpT, -SnT or -KT a 2-fold molar excess of purified SnT-mEGFP-SpT, SpC-SnC, or SUMO-KT was added (for the specific combinations see Figure 3). Where appropriate, 800 pmol of purified SpyLigase was added and the mixes were supplemented with PBS (pH 7.4) to reach a final volume of 50 μΙ_. After 24 hours of incubation at 4 °C the reaction mixes were analyzed by SDS-PAGE with Coomassie staining.

Spy-ligation of antigens to Hbp in OMVs. To 500 μΙ_ of OMVs (isolated from S. Typhimurium SL3261 AtoIRA AmsbB cells) containing HbpD(Ad1 )-SpyTag, obtained from a 1000 ODeoo equivalent (ODeoo x mL) of cell culture 900 μΙ_ of purified SpC-PspAa'-SnT (140 μΜ), SpC-SP1690-SnT (176 μΜ) or SpC-PspAa'-SP1690-SnT (72 μΜ), was added and incubated for 24 h at 4 °C. The reaction mixes were analyzed by SDS-PAGE with Coomassie staining and with immunodetection after western blotting. For immunodetection anti-polyHistidine antibody (H1029, Sigma) was used. Using similar strategy Spy- and Snoop ligation was achieved of the model fusion protein SpyCatcher- SnoopCatcher (Spc-SnC) to S. Typhimurium SL3261 AtoIRA OMVs displaying HbpD(Ad2)- or HbpD(Ad4)- variants, respectively, carrying either a SpyTag (SpT), SpyTag with extended linkers (EL. SpT), or a SnoopTag (SnT). In these cases OMVs obtained from a 1 OD600 equivalent (OD600 x mL) of cell culture were incubated with 15.47 μΜ of Spc-SnC in a total reaction volume of 50 μΙ.

Adduct extension by alternating Spy-ligation and Snoop-ligation. To 100 μί of OMVs (isolated from S. Typhimurium SL3261 AtoIRA cells) containing HbpD(Ad1 )-SpyTag, obtained from a 100 OD600 equivalent (ODeoo x mL) of cell culture 80 μί of purified SpC-PspAa'-SnT (86 μΜ) was added. After two hours of incubation at 4 °C, 120 μί of purified SnC-SP1690-SpT (57 μΜ) was added. After two hours of incubation at 4 °C another 80 μί of purified SpC-PspAa'-SnT. After a further two hours of incubation at 4 °C another 120 μΙ_ of purified SnC-SP1690-SpT was added and the mix was incubated for 15 h at 4 °C. The OMVs were isolated by centrifugation at 293,000g for 45 min at 4 °C. The OMVs were resuspended in PBS (pH 7.4) containing 15% glycerol and analyzed by SDS-PAGE with Coomassie staining and with immunodetection after western blotting. For immunodetection monoclonal anti-FLAG M2 antibody (F3165, Sigma), HA-tag monoclonal antibody (2-2.2.14)

(ThermoFisher Scientific) and anti-polyHistidine antibody (H1029, Sigma) were used.

Spy-ligation of GFPnb-SpC to Hbp in cells. S. Typhimurium SL3261 cells harboring pHbpD(Adl )- SpyTag were grown in LB containing 0.4% glucose and chloramphenicol to an OD600 of 0.3-0.4.

Protein expression was induced by the addition of IPTG to a final concentration of 1 mM and the cells were incubated for a further two hours. Subsequently the cells were harvested, resuspended to an ODeoo of 10 in PBS (pH 7.4) containing 3 μΜ of GFPnb-SpC or GFPnb-SpC EQ and incubated for 60 min at 25 °C in a Thermomixer (Eppendorf) at 1000 rpm. After incubation with the nanobody- SpyCatcher protein the cells were resuspended to an OD600 of 10 in PBS (pH 7.4) containing 3 μΜ of GFP and incubated at 4 °C for 5 min. Last, the cells were resuspended to an OD600 of 10 in PBS (pH 7.4) and analyzed by SDS-PAGE with Coomassie staining and microscopy.

Spy-ligation of GFPnb-SpC to Hbp in OMVs. 1 μΙ_ of OMVs containing HbpD(Ad1 )-SpyTag, obtained from a 1 OD600 equivalent (OD600 x ml_) of cell culture, were diluted in 100 μΙ_ of PBS (pH 7.4) containing 3 μΜ of GFPnb-SpC or GFPnb-SpC EQ. The mixes were incubated for 60 min at 25 °C in a Thermomixer (Eppendorf) at 1000 rpm and analyzed by SDS-PAGE with Coomassie staining and with immunodetection after western blotting. For immunodetection anti-polyHistidine antibody (H1029, Sigma) was used.

Spy-ligation of FimH _L to Hbp on E. coli cells. To allow coupling, 1.0 OD660 unit of E. co// TOP10F' cell material displaying HbpD(Ad1 )-SpT was incubated overnight with 50 pg of FimHL-SpC at 4°C. Successful coupling was assessed by SDS-PAGE /Coomassie staining analysis.

Spy-ligation of aCD180 to Hbp on OMVs. To couple SpyTagged anti-mouse CD180 (aCD180-SpT) to bacterial cells, 2 ODe6o units of Salmonella SL3261Afo/R/l cells expressing HbpD(Ad1 )-SpC at the surface were overnight incubated with 200 μΙ of aCD180-SpT (0.5 mg/ml) at 4°C). Next day, to separate cell-coupled and non-coupled material, the mixture was centrifuged at low speed (5,000 rpm, 5 min) and the bacterial pellet and supernatant fraction were recovered. Subsequently, the bacterial pellet was resuspended in PBS to wash off loosely associated material. The cell suspension was subjected to low speed centrifugation (5,000 rpm, 5 min) and the bacterial pellet and supernatant were recovered. Coupling of aCD180-SpT to HbpD(Ad1 )-SpC was analyzed by Coomassie stained SDS- PAGE.

Quantitation of ligation efficiency. After SDS-PAGE proteins and staining Coomassie Brilliant Blue G250 gels were scanned on a GS-800 Calibrated Densitometer (BioRad). The intensities of protein bands were determined using ImageJ (http://imaoiei.nih.gQv/ij/). After correction for the difference in molecular weight the fractions of Hbp (unmodified and ligated) were calculated.

Microscopy. Cells were photographed using an F-View II CCD camera (Olympus) mounted on a BH2 microscope with a RFCA fluorescence attachment using a DApo100UV PL 1.30 oil 160/0.17 objective and a GFP fluorescence filter cube (BP495/DM505) (Olympus). Image files were processed using ImageJ (http://imagej.nih.gov/ij/).

Example 1 : Spy- and Snoop-ligation to Hbp on the surface of OMVs.

The E. coli Hemoglobin protease (Hbp) was used as a carrier. The autocatalytic cleavage site was mutated to create an Hbp carrier that is displayed at the cell surface rather than being released, providing permanent exposure of fused antigens at some distance of the cell surface.

As host for the Hbp antigen display carrier an attenuated Salmonella Typhimurium strain was used that provokes strong mucosal and systemic responses of both cellular and humoral nature (Jong et al (2014) supra; Hoiseth et al. (1981 ) supra; Moreno et al. 2010 Curr Gene Ther 10(1 ): 56-76).

Furthermore, as an alternative non-living platform, we have explored OMVs derived from the same strain. To increase the production of OMVs a AtoIRA derivative of the attenuated S. Typhimurium strain was used.

To investigate the suitability of the SpyCatcher/SpyTag system for ligation of proteins to Hbp fixed on the surface of OMVs, we fused the SpyTag (SpT) to the N-terminus of the Hbp display construct HbpD(Adl ) (Figure 1A). In our experience, this position in Hbp is the most tolerant towards fusion of heterologous sequences. To test display in the context of OMVs, the fusion construct was expressed in the hypervesiculating S. Typhimurium AtoIRA strain and OMVs were collected by

ultracentrifugation. As expected, the small (13 aa long) tag did not impede expression and surface display of Hbp (Figure 2A, compare lane 1 and 8 and Figure 8, lanes 7-9 providing a density of the SpyTag at the OMV surface that is similar to the major outer membrane protein OmpA. Importantly, the SpyTag is located at the tip of the β-helical stem structure of Hbp providing considerable distance to the membrane surface and consequent accessibility towards cognate coupling partners.

As proof of concept for Spy-mediated ligation, OMVs decorated with HbpD(Ad1 )-SpT (SEQ ID NO: 4) were incubated with purified SpyCatcher-SnoopCatcher hybrid protein (SpC-SnC, Figure 1 B) (SEQ ID NO: 14) in incremental molar ratios and subsequently the samples were analyzed by SDS-PAGE and total protein staining. An adduct with an electrophoretic mobility corresponding to that expected of the HbpD(Ad1 )-SpT-SpyCatcher-SnoopCatcher ligation product, 145.7 kDa, appeared at the expense of HbpD(Ad1 )-SpT (Figure 2A, lanes 2-7). In a control reaction containing OMVs displaying the carrier HbpD(Adl ) without a SpyTag (SEQ ID NO: 3) this ligation product was not detected (Figure 2A, lanes 8 and 9). The amount of unligated and ligated Hbp protein was determined by densitometric scanning. Saturation of coupling, at a maximum of -85% of the Hbp molecules ligated, was reached at an approximate 4-fold molar excess of cargo protein relative to the Hbp carrier (Figure 2B). This level of ligation efficiency comes close to the levels observed upon co-incubation of purified cognate ligation partners (e.g. Figure 3, lanes 7 and 8). These results indicate that the SpyTag fused to displayed Hbp is well accessible for coupling at the surface of OMVs.

To investigate if the Spy and Snoop ligation systems are more generally compatible with the Hbp display platform, the SpyCatcher (SpC), KTag (KT), SnoopCatcher (SnC) and SnoopTag (SnT) were also individually introduced at the N-terminus of the Hbp display construct HbpD(Adl ) (SEQ ID NO: 7; SEQ ID NO: 6; SEQ ID NO: 8; and SEQ ID NO: 5, resp) and expressed on OMVs. All fusion proteins were present in the OMV fraction in a density similar to the major OMPs (Figure 7). For all fusion proteins, protease accessibility experiments confirmed their localization at the cell surface indicating that the Tags and Catchers did not impede translocation of Hbp across the cell envelope (Figure 8). To further substantiate surface localization and suitability for ligation of cognate partner proteins, ligation tests were carried out. OMVs decorated with the Hbp fusion proteins (Figure 7 A) were incubated with purified recombinant proteins that were used by Howarth and co-workers (Figure 1 1 B) as tools to demonstrate bipartite and tripartite ligation: SpyCatcher-SnoopCatcher (SEQ ID NO: 14), SpyTag-MBP (SEQ ID NO: 19), SnoopTag-MBP (SEQ ID NO: 20), SnoopTag-mEGFP-SpyTag (SEQ ID NO: 21 ), SUMO-KTag (SEQ ID NO: 22) and SpyLigase (SEQ ID NO: 23) (Veggiani et al. 2016 supra; Zackeri et al. 2012 supra; Feirer et al. 2014 supra). Summarizing, in the context of OMVs the highest efficiency of display of ligated product was obtained with the HbpD(Ad1 )-SpT construct (Figure 3, lane 3).

Comparing the display efficiencies of (bipartite) ligated adducts, the constructs rank HbpD(Ad1 )-SpT> HbpD(Ad1 )-SpC> HbpD(Ad1 )-SnT> HbpD(Ad1 )-SnC (Figure 3, lanes 1 -4). Apparently, the 10.0 kDa SpyCatcher and the 12.6 kDa SnoopCatcher had no major impact on the display of the Hbp carrier and folded into their catalytic structure at the OMV surface enabling binding to Spy or Snoop-tagged partner proteins.

In tripartite ligation the SpyCatcher domain is further split up in the so-called KTag harboring the lysine residue that is part of the isopeptide bond and the larger catalytic "SpyLigase" domain. The isolated SpyLigase is able to couple a KTagged protein and a SpyTagged protein in trans expanding the potential applications of this technology (Fierer et al. 2014 supra). In contrast to the bipartite system, the tripartite system did not yield any appreciable ligation product in the OMV context (Figure 3, lanes 5 and 6). To verify the activity of the SpyLigase used, purified SUMO-KTag and SpT-MBP were shown to be coupled upon co-incubation, though not very efficiently (Figure 3, lane 9).

Example 2: Ligation of antigens to Hbp

Encouraged by the efficient enzymatic coupling of the various cargo proteins in Example 1 to surface- exposed Hbp using Spy/Snoop technology, this approach was tested for the decoration of OMVs with complex antigens. As proof of concept, coupling of the large N-terminal a1 a2-domain (213 amino acid, 24.4 kDa, further referred to here as the PspA a'-domain) of the surface exposed pneumococcal PspA antigen (from strain TIGR4) to OMVs pre-decorated with Hbp using Spy technology was explored. Previously, efficient display required splitting up the PspA a'-domain and incorporating the two resulting fragments at different sites in Hbp (HbpD-PspA[a1 -a2]) (Kuipers et al. 2015 supra). Yet, the highest achievable concentration of HbpD-PspA[a1 -a2] in OMVs was still only one third of that of antigen-free HbpD. For coupling we fused the SpyCatcher to the complete PspA a'-domain. For purposes described in the next section we further included an HA-tag and a SnoopTag at the C- terminus of the fusion construct (Figure 1 C, SpC-PspAa'-SnT; SEQ ID NO: 17).

Incubation of S. Typhimurium OMVs decorated with HbpD(Ad1 )-SpT (SEQ ID NO: 4) with a 10-fold molar excess of SpC-PspAa'-SnT (SEQ ID NO: 17) resulted in efficient (-85%) ligation as judged by SDS-PAGE analysis (Figure 4, lane 2). For further proof of principle we followed the same strategy using the conserved pneumococcal antigen SP1690 (WO/2008/127094). SP1690 is a 47.3 kDa component of an ABC-type oligosaccharide transport system belonging to the type 2 periplasmic- binding fold superfamily (Marchler-Bauer et al. 2017 Nucl Acid Res 45(D1 ): D200-D203). Initial attempts to display intact SP1690 as a fusion to Hbp were unsuccessful necessitating the splitting of SP1690 in four fragments that were inserted in two Hbp carriers (two fragments per carrier). However, even upon splitting, only moderate amounts of SP190 were displayed at the surface (data not shown). For Spy-ligation we created a fusion protein of SpyCatcher and full-length SP1690 (only omitting the N-terminal Cys residue that is most likely acylated in the native protein as predicted from its lipoprotein-like signal peptide) (Juncker et al. 2003 Proten Science: a publication of the Protein Society, 12(8): 1652-62). A FLAG-tag and a SnoopTag were included at the C-terminus of the fusion construct (Figure 1 C, SpC-SP1690-SnT; SEQ ID NO: 24). Similar to SpC-PspAa'-SnT, efficient (-85% coupling) was observed upon incubation with OMVs decorated with HbpD(Ad 1 )-SpT (Figure 4, lane 3). A further increase in complexity of the fusion partner was achieved by fusing the complete PspAa' domain and full-length SP1690 in one construct. The resultant -87 kDa bivalent antigen (SEQ ID NO: 18) was coupled through Spy technology to Hbp-decorated OMVs at equally high efficiency (Figure 4, lane 4).

Together the data indicate that large complex antigens and even chimeras of complex antigens can be displayed at high density the combined Hbp and Spy technologies. By combining the Hbp display technology with protein ligation via isopeptide bond formation, such as the Spy - and Snoop

Catcher/Tag technology, antigens and even chimeras of complex antigens can be displayed at high density on the surface of bacterial cells and OMVs.

Example 3: Ligation of multiple antigens to Hbp.

The combined use of the Spy and Snoop ligation systems was investigated to see whether they allow a more attractive iterative approach to couple multiple antigens to the Hbp platform while maintaining high antigen concentrations per OMV. First, HbpD-PspAa' formed by the ligation of SpC-PspAa'-SnT (SEQ ID NO: 17) to HbpD(Ad1 )-SpT (SEQ ID NO: 4) was tested to see whether it could be further extended by adding SP1690 fused to SnoopCatcher. To allow subsequent coupling of purified SpC- PspAa'-SnT, the SP1690 fusion protein also contained a SpyTag at its C-terminus (SEQ ID NO: 25). A similar iterative extension approach was shown by Veggiani ef a/, to efficiently polymerize affibodies into multimeric chains (Veggiani et al. 2016 supra). Importantly, the process was started with an affibody fused to a SpyCatcher immobilized on a solid support. Subsequently the material was incubated with affibody containing both a SpyTag and a SnoopTag at the N- and C-terminus, respectively. After the unreacted affibody was washed away the material was incubated with a SpyCatcher-SnoopCatcher chimera to extend the chain. By iterating this process chains of affibodies were formed. Since immobilizing and washing the OMVs is technically more challenging we explored the principle of the strategy in one reaction mix. This approach therefore has the disadvantage that SpC-PspAa'-SnT and SnC-SP1690-SpT may link to each other, but not to HbpD(Ad1 )-SpT. Moreover, polymers of SpC-PspAa'-SnT and SnC-SP1690-SpT might cyclize and form unreactive end products that are not linked to the Hbp platform. To limit undesirable product formation OMVs containing HbpD(Ad 1 )-SpT were incubated sequentially with SpC-PspAa'-SnT and SnC-SP1690-SpT in two rounds, followed by centrifugation and analysis of the OMVs and the spent medium by SDS-PAGE. In both fractions a complex pattern of adducts was observed, which likely resulted from polymer formation combined with limited proteolysis (Figure 5A, lanes 4 and 5). Importantly, in the OMV fraction (Figure 5A, lane 5) three major adducts that probably correspond to single, double and triple ligation events dominate. Omitting the Spy/Snoop-ligation terms from the nomenclature for clarity, these probably represent HbpD-PspAa' with an expected molecular mass of 157 kDa, HbpD-PspAa'-SP1690 of 222 kDa and HbpD-PspAa'-SP1690-PspAa' of 262 kDa, respectively. The identity of these three adducts was confirmed by immunodetection using anti-HA and anti-FLAG antibodies that recognize the HA- and FLAG epitopes integrated in SpC-PspAa'-SnT and SnC-SP1690-SpT, respectively (Figure 5B). Quantification of HbpD(Ad1 )-SpT and the three major adducts (Figure 5A, lanes 3 and 5) indicated a total ligation efficiency of -85% and a ratio of -5:1 :1 (157 kDa/222 kDa/262 kDa) for the adducts. The data thus show that the principle of polymerization worked, and that isopeptide formation can be used for iterative extension of two or more antigens to the Hbp fusion protein.

Example 4: Ligation of nanobodies to Hbp

Decoration of bacteria or OMVs with lectins, adhesins or affinity molecule handles, such as antibodies or nanobodies, could be used as a tool for OMV targeting to certain tissues and immune cells, thereby increasing the efficacy of antigen delivery. Nanobodies are particularly attractive for this purpose because of their versatility and ease of recombinant production. They are much smaller (15 kDa) than regular antibodies since they consist only of a single variable domain fragment (VHH) as found in heavy-chain-only antibodies that occur naturally in camelids and sharks. Previous attempts to incorporate nanobodies in Hbp by gene fusion have met with variable and generally low display efficiencies (data not shown).

To provide proof of concept for coupling of nanobodies to Hbp on OMVs through Spy technology, we produced a fusion construct in which SpyCatcher and a GFP-specific nanobody (GFPnb; Kubala et al 2010, supra) are connected through a flexible linker to enable independent folding (SEQ ID NO: 15). Sequential incubation of HbpD(Ad1 )-SpT decorated bacteria with purified GFPnb-SpC and GFP should lead to covalent coupling of GFPnb-SpC and binding of GFP provided that GFPnb is correctly folded in the context of the fusion construct. Consequently, the bacteria should get a green coat, which can be visualized by whole cell immunofluorescence. S. Typhimurium cells expressing HbpD(Adl )- SpT (SEQ ID NO: 4) were first incubated with GFPnb-SpC (residues 25-245 of SEQ ID NO: 15) to allow coupling, which was confirmed by SDS-PAGE. As expected, an adduct of -140 kDa was detected consistent with the combined mass of Hbp-SpT (1 17.1 kDa) and GFPnb-SpC (24.2 kDa) (Figure 6A, lane 1 ). The cells were harvested and briefly incubated in PBS containing purified GFP (SEQ ID NO: 26). Whole cell immunofluorescence revealed a clear circumferential staining as expected for cells decorated with GFP (Figure 6B).

As a control for non-specific interaction of GFP with the bacterial surface in this procedure, the same sequential labelling was carried out using a catalytically inactive SpyCatcher fused to GFPnb (SEQ ID NO: 16). The catalytic GIU77 in SpyCatcher was substituted by a Gin (SpC EQ) which has been shown to abolish covalent bond formation (Zackeri et al. 2012, supra). Using purified GFPnb-SpC EQ (residues 25-245 of SEQ ID NO: 16), the -140 kDa adduct was not detected (Figure 6A, lane 2) and surface labelling with GFP did not occur (Figure 6B) confirming the specificity of the labelling procedure. Although GFP labelling is better suited for visualization using whole cells, the GFPnb-SpC coupling step was also shown to occur on the surface of OMVs decorated with Hbp-SpT again depending on the catalytic integrity of the SpyCatcher part of the GFPnb-SpC hybrid (Figure 9). Previously, a nanobody was inserted into Hbp to achieve display, but with limited success (Jong et al., PLoS One. 2018 Feb 7; 13(2):e0191622) indicating that the nanobody is largely translocation- incompetent. In contrast, the observed efficient coupling of Catcher-nanobody chimeras to vesicles, shown in this example, will permit for instance targeted drug delivery in tumor tissue (Kijanka et al. 2015 Nanomedicine (London) 10(1 ): 161 -74). For OMV vaccine development it will be interesting to include surface exposed nanobodies that recognize receptors specific for certain dendritic cell types. In this way, specific T-cell responses can be induced towards the presented antigens (Goyvaerts et al. 2015 J Immunol Res 2015:785634).

Example 5: Coupling of a rat antibody to the surface of bacterial cells or outer membrane vesicles. This example illustrates successful and high-density coupling of a rat antibody (anti-mouse CD180) to the surface of bacterial cells or derived OMVs using isopeptide bonding technology. Furthermore, the example shows that the coupled antibodies are functional and, as a consequence, allows targeting of bacterial cells decorated with the antibodies to human cells expressing the antibody ligand.

To allow coupling of antibodies using isopeptide bonding technology, SpyTagged rat lgG2a antibodies were isolated directed against mouse CD180. To this end, plasmid constructs encoding anti-mouse CD180 antibodies comprising heavy chain Fc domains carrying a SpyTag at the extreme C-terminus (Figure 1 1 A) (SEQ ID NO: 27) were transfected to mammalian HEK cells and antibodies were produced and isolated as described (Vink et al., Methods. 2014 Jan 1 ;65(1 ):5-10).

To couple SpyTagged anti-mouse CD180 (aCD180-SpT) to bacterial cells, 2 OD660 units of

Salmonella SL3261Afo//¾4 cells expressing HbpD(Ad1 )-SpC (SEQ ID NO: 7) at the surface were overnight incubated with 200 μΙ of aCD180-SpT (0.5 mg/ml) at 4°C (crude mix). Next day, to separate cell-coupled and non-coupled material, the mixture was centrifuged (centrifuge.) at low speed (5,000 rpm, 5 min) and the bacterial pellet and supernatant fraction were recovered. Subsequently, the bacterial pellet was resuspended in PBS to wash off loosely associated material (PBS wash). The cell suspension was subjected to low speed centrifugation (5,000 rpm, 5 min) and the bacterial pellet and supernatant were recovered. A sample of SL3261 Afo/R l-HbpD(Ad 1 )-SpC before addition of aCD180- SpT of aCD180-SpT was analyzed by Coomassie stained SDS-PAGE (lane 1 ). Samples of the crude SL3261 Afo/R/¾-HbpD(Ad 1 )-SpC + aCD180-SpT mix (lane 2) and pellet (p) and supernatant (s) after centrifugation (lanes 3 and 4) and PBS washing (lanes 5 and 6) were analyzed in parallel (Figure 1 1 , B). Molecular weight markers (kDa) are indicated at the left side of the panel.

The SDS-PAGE analysis shows that aCD180-SpT is covalently coupled to the 125 kDa cell surface expressed HbpD(Ad 1 )-SpC. This is illustrated by the appearance of a higher molecular weight adduct in the crude mix sample (lane 2), corresponding with a covalent fusion product comprising HbpD(Ad l )- SpC and the aCD180-SpT heavy chain. The appearance of the adduct coincides with the

disappearance of non-coupled HbpD(Ad 1 )-SpC (c.f. lane 2 and 1 ). The adduct remains detectable in the bacterial pellet following centrifugation and PBS washing, underscoring the covalent association of aCD180-SpT with HbpD(Ad1 )-SpC and association with the bacterial cell. Moreover, a band corresponding with the aCD180-SpT light chain remains detectable in the bacterial pellet following centrifugation and PBS washing. The non-covalently associated light chain is dissociated from the coupled HbpD(Ad1 )-SpC aCD180-SpT fusion in denaturing conditions analysis, explaining its existence as a separate lower molecular weight protein entity upon SDS-PAGE analysis. Together, the data indicate that complete aCD180-SpT antibodies were successfully coupled to cell surface- exposed HbpD(Adl ) using isopeptide bonding technology.

To couple SpyTagged anti-mouse CD180 (aCD180-SpT) to OMVs, 20 μΙ of Salmonella SL3261A/o/R/4 OMVs expressing HbpD(Ad1 )-SpC at the surface, obtained from a 20 Οϋβοο equivalent (Οϋβοο x ml_) of cell culture, were overnight incubated with 100 μΙ of aCD180-SpT (0.5 mg/ml) at 4°C {crude mix). Next day, to separate cell-coupled and non-coupled material, the mixture was centrifuged (centrifuge.) at high speed (200.000 x g, 30 min, 4°C) and the OMV pellet and supernatant fraction were recovered. Subsequently, the OMV pellet was resuspended in PBS supplemented with 0.5 M NaCI to wash off material associated by electrostatic protein-protein interactions (Salt wash). The OMV suspension was subjected to high speed centrifugation (200.000 x g, 30 min, 4°C) and the bacterial pellet and supernatant were recovered. A sample of SL3261Afo/R l-HbpD(Ad1 )-SpC OMVs before addition of aCD180-SpT of aCD180-SpT was analyzed by Coomassie stained SDS-PAGE (lane 1 ). Samples of the crude SL3261Afo/R/4-HbpD(Ad1 )-SpC OMVs + aCD180-SpT mix (lane 2) and pellet (p) and supernatant (s) after centrifugation (lanes 3 and 4) and Salt washing (lanes 5 and 6) were analyzed in parallel (Figure 1 1 ). Molecular weight markers (kDa) are indicated at the left side of the panel.

The SDS-PAGE analysis shows that aCD180-SpT is covalently coupled to the 125 kDa HbpD(Adl )- SpC expressed at the surface of OMVs. This is illustrated by the appearance of a higher molecular weight adduct in the crude mix sample (lane 2), corresponding with a covalent fusion product comprising HbpD(Ad1 )-SpC and the aCD180-SpT heavy chain. The appearance of the adduct coincides with the disappearance of non-coupled HbpD(Ad1 )-SpC (c.f. lane 2 and 1 ). The adduct remains detectable in the OMV pellet following centrifugation and Salt wash, underscoring the covalent association of aCD180-SpT with HbpD(Ad1 )-SpC and association with the OMV particle. Moreover, a band corresponding with the aCD180-SpT light chain remains detectable in the OMV pellet following centrifugation and Salt washing. Together, the data indicate that complete aCD180- SpT antibodies were successfully coupled to HbpD(Adl ) at the OMV surface using isopeptide bonding technology.

To demonstrate the functionality of surface-coupled aCD180-SpT for the targeting of bacterial antigen delivery platforms to specific ligands, mammalian HeLa cells expressing murine CD180 were incubated with above-described SL3261Afo/R/4-HbpD(Ad1 )-SpC carrying covalently coupled aCD180- SpT. HeLa cells not expressing murine CD180 were used as a control. The HeLa cells, grown on IBIDI microscopy slides, were incubated with various loads of SL3261Afo/R \-HbpD(Ad1 )-SpC-aCD180-SpT (multiplicity of infection, MOI: 5000, 500, 50 or 5) (2h, 37°C, 8% C0 ₂ ). Upon incubation, the slides were washed with cell culture medium (DMEM + GlutaMAX, 10% FBS, penicillin/streptavidin) and fixed with 4% PFA in PBS (30 minutes, RT) before immersion into fresh PBS. HeLa cells were permeabilized with PBS + 0.1 % Triton-X100 (10 minutes, RT) before surface localized and internalized SL3261Afo/R \-HbpD(Ad1 )-SpC- aCD180-SpT were detected by indirect immunofluorescence. In this procedure rabbit polyclonal aHbp (J40) was used as the primary antibody and goat-a-rabbit Alexa 488 as the secondary antibody. Hoechst stain and Phalloidin Alexa 568 were used to stain the DNA and actin content of the HeLa cells, respectively. Post-staining with 0.7% PFA was performed before analysis of the samples at an Olympus 1X83 microscope. The median of the green fluorescence intensity per HeLa cell was determined making use of Cell profiler software (Figure 12).

The data show significantly enhanced fluorescence intensity for HeLa cells expressing murine CD180 compared to HeLa cells not expressing the protein at MOI 5000 and 500. This result demonstrates that more SL3261Afo//¾4-HbpD(Ad1 )-SpC-aCD180-SpT bacteria were localized to the HeLa cells when CD180 was present at the cell surface. In turn, this demonstrates the functionality of aCD180- SpT coupled to SL3261Afo/R l-HbpD(Ad1 )-SpC using isopeptide bonding technology. Moreover, the data indicate that covalent coupling of affinity molecule handles such as antibodies to bacterial cells and derived OMVs by isopeptide bonding technology is a valid approach to mediate the targeting of these bacterial cells or derived OMVs to specific ligands, cell types or tissues.

Example 6: Coupling of Catcher-equipped model proteins to Tag sequences inserted at position d2 and d4 of HbpD.

This example illustrates that ligation Tags expressed internally in OMV-surface exposed HbpD are functional for isopeptide bonding of model proteins carrying a cognate Catcher moiety.

To demonstrate the functionality of internally inserted ligation tags, HbpD variants carrying a SpyTag (SpT; flanked by GSGSS and GSGSG linkers corresponding to residues 534-556 of SEQ ID NO: 9 and residues 760-782 of SEQ ID NO: 12, respectively) or SpyTag with extended linkers (EL. SpT; flanked by GSGSSGSASG and GEGTGGSGSG linkers corresponding to residues 534-566 of SEQ ID NO: 10 and residues 760-792 of SEQ ID NO: 13, respectively) at either position d2 (SEQ ID NO: 9; SEQ ID NO: 10, respectively) or d4 (SEQ ID NO: 12; SEQ ID NO: 13, respectively) of the passenger domain (Figure 13 A) were expressed in Salmonella Typhymurium SL3261 AtolRA. The resulting

OMVs displaying the respective HbpD variants on the surface were harvested and incubated with a model protein comprising a SpyCatcher-SnoopCatcher fusion (SpC-SnC). Along same lines OMVs displaying HbpD carrying SnoopTag (SnT; flanked by GSGSS and GSGSG linkers corresponding to residues 534-555 of SEQ ID NO: 1 1 ) at position d2 of the passenger domain (SEQ ID NO: 1 1 ) (Figure 13 A) were prepared and incubated with SnoopCatcher-containing model protein. Following overnight incubations at 4°C, OMV-model protein mixtures were subjected to SDS-PAGE/Coomassie analysis to assess covalent coupling of the model protein to the respective HbpD variants. Successful isopeptide bonding of the model protein to internal SpT and EL. SpT inserted at either position d2 or d4 of HbpD was observed (Figure 13 B). This is illustrated by the appearance of a higher molecular weight adduct (>) in the OMV samples upon incubation with the model protein. The adduct corresponds with a covalent fusion product between the (EL.)SpT-carrying HbpD-variants and the SpyCatcher containing model protein. Similarly, successful coupling to SnT inserted at position d2 was observed, as illustrated by the appearance of a higher molecular weight adduct (*) upon incubation of OMVs expressing HbpD(Ad2)-SnT with SnC-model protein (Figure 13 C). Example 7: Improved display of a complex protein using isopeptide bonding technology

This example illustrates vastly improved display of complex proteins on the OMV surface when using isopeptide technology compared to conventional translational fusion to the passenger of an autotransporter.

The 47.3-kDa conserved pneumococcal antigen SP1690 is a bulky and complex protein as shown by 3D structural modelling (Swissmodel of S. pneumoniae TIGR4 SP1690; Figure 14 A). Attempts to display intact TIGR4 SP1690 as a translational fusion to the passenger of HbpD were unsuccessful, necessitating the splitting of SP1690 into four fragments F1 (residues 30-130), F2 (residues 161 -257), F3 (246-386) and F4 (residues 370-445) that were inserted into the passenger of two separate Hbp carriers (two fragments per carrier). The resulting fusion proteins HbpD-SP1690-F1 -F2 (SEQ ID NO: 28) and HbpD-SP1690-F3-F4 (SEQ ID NO: 29) were displayed at the surface of OMVs but only in moderate amounts. This follows from SDS-PAGE/Coomassie analysis of Salmonella Typhimurium SL3161 -derived OMVs displaying either HbpD-SP1690-F1 -F2 or HbpD-SP1690-F3-F4 showing that both fusions are present in significantly lower amounts than the endogenous outer membrane proteins (OMPs) of the OMVs (Figure 14 B, lanes 2 and 3). In contrast, antigen-less HbpD(Adl ) (SEQ ID NO: 3) expressed in parallel under identical expression conditions (Figure 14 B, lane 1 ) is detected at a levels vastly exceeding those of the OMPs, indicating that the integration of antigens adversely affects the display capacity of the Hbp platform. In comparison to the translational fusion approach, clearly more favorable results are obtained when isopeptide bonding technology was used to mediate display of SP1690 at the OMV surface. To achieve this, Salmonella Typhimurium SL3161-derived OMVs displaying SpyTagged HbpD variant HbpD(Ad1 )-SpT (SEQ ID NO: 4) at the surface were incubated with purified SP1690 carrying a SpyCatcher moiety (SpC-SP1690-SnT; SEQ ID NO: 24). Following a washing step to remove non-bound SpC-SP1690-SnT, the OMVs were analyzed by SDS-PAGE and Coomassie staining (Figure 14 C). An adduct was detected comprising a covalent fusion between HbpD(Ad1 )-SpT and SpC-SP1690-SnT. The amounts of the adduct detected far exceeded those of the endogenous OMPs and were similar to those of antigen-less HbpD(Adl ) (Figure 14 B, lane 1 ). These data clearly indicate that an isopeptide bonding approach favorably compares to translational fusion to Hbp because it allows for: 1 ) high-density display of complex proteins at bacterial-derived surfaces, and 2) surface display of complete complex proteins.

Example 8: Coupling of an adhesin to the surface of Gram-negative bacteria

This example illustrates that isopeptide bonding technology can be used to couple adhesins to the surface of Gram-negative bacteria. Furthermore it demonstrates that molecules that are incompatible with surface display upon integration into the autotransporter passengers can be displayed at the Gram-negative bacterial cell surface using isopeptide-bonding technology.

The fimbrial adhesin FimH is the best characterized adhesin derived from bacteria. FimH is responsible for D-mannose sensitive adhesion, which is mediated by its lectin domain FimHL. The lectin domain contains two cysteine residues that form a disulphide bond in the bacterial periplasmic space. Integration of the domain into the passenger of an autotransporter can therefore not be used as a strategy to translocate the adhesin domain across at the bacterial cell envelope. As an alternative approach, we used isopeptide bond-mediated coupling to display FimHi_ at the surface of Escherichia coli cells. To this end, a chimeric protein was constructed and purified comprising FimHi_, a C- terminally attached SpyCatcher and a HisTag for purification purposes (FimHL-SpC; residues 23-300 of SEQ ID NO: 30). Furthermore, E. coli TOP10F' cells displaying HbpD(Ad1 )-SpT (SEQ ID NO: 4) were prepared. To allow coupling, 1 .0 OD660 unit of cell material was incubated overnight with 50 pg of FimHL-SpC at 4°C. Successful coupling was assessed by SDS-PAGE /Coomassie staining analysis of the sample in parallel to a TOP10F7HbpD(Ad1 )-SpT mock sample lacking FimHL-SpC (Figure 15).

Successful covalent coupling of SpyCatchered FimHL SpyTagged HbpD at the bacterial cell surface was demonstrated by the occurrence of higher molecular weight adduct upon incubation of

TOP10F7HbpD(Ad1 )-SpT with FimHL-SpC, comprising a covalent fusion between the two proteins. As expected, the appearance of the adduct coincides with the disappearance of significant amounts of non-coupled HbpD(Ad1 )-SpT material compared to the situation in which no FimHL-SpC was added to the cells. As a further control, no higher molecular weight adduct appeared under these latter conditions. Molecular weight markers (kDa) are indicated at the left side of the panel.