A bio-coupling approach using a dextran-binding domain to immobilize an engineered streptavidin to Sephadex for easy preparation of affinity matrix

An engineered streptavidin, SAVSBPM18 with reversible biotin binding capability, has been successfully applied to purify biotinylated and streptavidin-binding peptide (SBP) tagged proteins. To simplify the preparation for the SAVSBPM18 affinity matrix without chemical conjugation, two bio-coupling approaches were developed based on a 14-kDa dextran-binding domain (DBD) from a Leuconostoc mesenteroides dextransucrase. The first approach offers simplicity for bio-coupling by creating a direct fusion, SAVSBPM18-Linker-DBD. Purification of the fusion from crude extract and its immobilization to Sephadex can be consolidated in one-step. The second approach aims at flexibility. A SnoopCatcher (SC) was fused to DBD to create SC-Linker-DBD. This fusion can covalently capture any recombinant proteins tagged with a SnoopTag (ST) including SAVSBPM18-Linker-ST via the formation of an isopeptide bond at the interface through the SnoopCatcher-SnoopTag interaction. Although monomeric DBD binds to dextran with nanomolar affinity, DBD tetramerized via streptavidin (SAVSBPM18-Linker-ST·SC-Linker-DBD) showed an even tighter binding to Sephadex. The majority of the fluorescently labelled DBD tetramers were retained on the Sephadex surface even after four months. Affinity columns generated using either approach effectively purified both SBP-tagged and biotinylated proteins. These columns are reusable and functional even after a year of frequent use.

M18-L-DBD retains the biotin binding capability. M18-L-DBD should have both the reversible biotin-binding attribute and the dextran immobilization capability. To evaluate its biotin binding capability, M18-L-DBD was applied to a biotin agarose column. Figure 1d shows that M18-L-DBD was selectively bound to biotin agarose. Large amounts of pure protein could be easily recovered from the intracellular soluble fraction by affinity chromatography using biotin agarose and elution with biotin.
Immobilization of M18-L-DBD to Sephadex G-100. To evaluate the dextran binding functionality and to prepare the M18-L-DBD affinity matrix, M18-L-DBD was applied to a Sephadex G-100 column using two strategies. In the first strategy M18-L-DBD affinity-purified by biotin agarose was loaded to the column. This worked well. However, the purified proteins in a highly concentrated state tend to form precipitates. This results in a significant loss of the starting material. An alternative and more simple approach is to load the crude intracellular soluble fraction containing M18-L-DBD directly to Sephadex-G100 (Fig. 1e). M18-L-DBD in the crude sample was applied under a slightly overloaded condition. Analysis of the boiled matrix (BF in Fig. 1e) showed that M18-L-DBD was selectively retained on the column with over 95% of the bound proteins composed of M18-L-DBD. Dialysis of the crude sample prior to column loading is not crucial. This simple method simultaneously affinity-purifies and immobilizes M18-L-DBD to generate the M18-L-DBD matrix in one step. By overloading the column with M18-L-DBD, around 708 ± 13 µg (5 nmoles of tetramers, Table 1) of M18-L-DBD could be captured per ml of Sephadex G-100.

Recombinant production of SnoopCatcher-L-DBD (SC-L-DBD) and M18-L-SnoopTag (M18-L-ST).
The direct fusion approach requires M18 and DBD to be folded independently without interfering with each other and the fusion to be in the soluble state. As mentioned above, purified M18-L-DBD at high concentrations tends to form insoluble precipitates. It would be of interest to explore other options as a backup.
One possibility is to first produce M18 and DBD individually in a soluble and functional state. These components can then be biologically ligated together via covalent bond formation to generate the final soluble fusion products.
An approach using two heterodimeric coiled coil sequences 18,19 to link M18 to DBD has been evaluated. However, the heterodimerization efficiency was low. We here explored another alternative using the SnoopCatcher (SC)-SnoopTag (ST) based bio-ligation system 14  Approaches to generate Sephadex bound M18-L-ST·SC-L-DBD. The initial strategy of producing Sephadex bound M18-L-ST·SC-L-DBD is to first couple SC-L-DBD onto a Sephadex G-100 column and then to saturate the immobilized SC-L-DBD with M18-L-ST. The covalent complex M18-L-ST·SC-L-DBD was obtained but with low efficiency. Even with overnight incubation involving different loading amounts of M18-L-ST, large amounts of free SC-L-DBD and M18-L-ST remained unreacting. Mixing SC-L-DBD bound Sephadex beads in a batch mode (instead of a column format) with excess M18-L-ST did not improve the coupling reaction. The limited diffusion rate of the Sephadex bound SC-L-DBD can be a major factor contributing to the observed inefficient coupling. To address this concern, production of the covalent complex in solution first has to be optimized. The complexes will then be purified and immobilized to Sephadex. To generate the M18-L-ST·SC-L-DBD complexes, the bio-conjugation process could be performed with either SC-L-DBD or M18-L-ST in excess. Generation of M18-L-ST·SC-L-DBD under the condition with excess SC-L-DBD is the preferred choice (Fig. 3a). It can easily generate a homogenous population of M18-L-ST·SC-L-DBD with four SC-L-DBD modules covalently linked to each M18-L-ST tetramer. In contrast, the other condition would generate a heterogeneous www.nature.com/scientificreports www.nature.com/scientificreports/ population M18-L-ST·SC-L-DBD with one to four SC-L-DBD modules per M18-L-ST tetramer (Fig. 3b). Relative to M18-L-ST·SC-L-DBD with four covalently linked SC-L-DBD modules, M18-L-ST·SC-L-DBD with less than four SC-L-DBD modules per M18-L-ST tetramer would have lower binding affinity towards Sephadex. Separation of the high affinity version from other lower affinity versions requires extra purification steps. Furthermore, the yield of the high affinity version would be lower than that generated by using excess SC-L-DBD.
Generation of the M18-L-ST·SC-L-DBD matrix: SC-L-DBD in excess in the reaction mixture. The average production yield of the soluble SC-L-DBD (2.2 nmoles/ml of culture) is about a third of the soluble M18-L-ST (6.6 nmoles of monomer/ml of culture). SC-L-DBD has to be mixed with M18-L-ST at a ratio greater than 3:1 (in terms of culture volume) so that SC-L-DBD will be in excess. Under this condition, M18-L-ST·SC-L-DBD and SC-L-DBD will be the major species in the reaction mix after completion of the bio-conjugation reaction (Fig. 3a). Figure 4a shows the kinetics of the M18-L-ST·SC-L-DBD formation at 4 °C. In this reaction, SC-L-DBD was mixed with M18-L-ST at a ratio of 6:1 (in terms of culture volume). M18-L-ST was gradually depleted and a new covalent complex, M18-L-ST·SC-L-DBD, was formed ( Fig. 4a, lanes 3-6). This complex migrated on the SDS-polyacrylamide gel with an apparent molecular mass of 66 kDa, which agrees closely with the expected apparent molecular mass of 62 kDa (35 kDa + 27 kDa). Some covalent complexes could be observed just five minutes post-mixing. Formation of covalent complex was more than 70% complete at 90 min and over 90% complete at 3 hours post-mixing. Overnight reactions are usually carried out to ensure completion of the reaction.
M18-L-ST·SC-L-DBD could be separated from the excess SC-L-DBD and other background proteins by affinity chromatography using biotin agarose (Fig. 4b). While all the free SC-L-DBD molecules and background contaminants were out in the flow-through and wash fractions, highly pure M18-L-ST·SC-L-DBD was captured on the column and could be recovered in the elution fractions (Fig. 4b, E1 and E2) using biotin as the eluent. The absence of M18-L-ST in the elution fractions indicates that all M18-L-ST moieties have reacted to form the covalent complexes. When the purified M18-L-ST·SC-L-DBD complexes were not boiled before loading onto the polyacrylamide gel, these complexes migrated with an apparent molecular mass over 200 kDa (Fig. 4c, lanes 1 and 2) which corresponds to the expected apparent molecular mass of 264 kDa (66 kDa × 4) of the assembled tetramer. This assembly process is relatively biotin independent (Fig. 4c, lane 1 vs lane 2). The faint band observed in lane 1 represents small quantities of disassembled M18-L-ST·SC-L-DBD monomers. The partial disassembly of tetrameric M18-L-ST·SC-L-DBD in the absence of biotin can be triggered by the presence of SDS. In contrast, biotin binding is known to strengthen the subunit interactions and can minimize the disassembly of tetrameric streptavidin even in the presence of SDS 21 (Fig. 4c, lane 2).
Purified M18-L-ST·SC-L-DBD could be effectively captured by Sephadex G-100 to generate the M18-L-ST·SC-L-DBD matrix (Fig. 4d). Analysis of the bound fraction (BF in Fig. 4d) by boiling a bead sample suggests that the covalent complex captured was composed purely of M18-L-ST·SC-L-DBD. By overloading the column with M18-L-ST·SC-L-DBD, the amount of the immobilized M18-L-ST·SC-L-DBD was estimated to be around 610 ± 6 µg (2.75 nmoles of tetramers, Table 1) per ml of Sephadex G-100.

Dynamic spatiotemporal distribution of fluorescently labelled His-SC-L-DBD(Cys) and
M18-L-ST·His-SC-L-DBD(Cys) to Sephadex G-100 beads. In this study, DBD is applied as an immobilization domain to anchor capturing proteins to chromatographic matrices such as Sephadex to generate the   www.nature.com/scientificreports www.nature.com/scientificreports/ affinity matrices. It is interesting to monitor the retention ability of the bound DBD fusions in their monomeric and tetrameric states to Sephadex using fluorescent microscopy 4,22 . A His-Tagged version of SC-L-DBD with  www.nature.com/scientificreports www.nature.com/scientificreports/ a cysteine in the C-terminal region was constructed (Fig. 5a, Supplementary Sequence 4). The His-Tag enables purification of His-SC-L-DBD(Cys) by IMAC affinity chromatography 23 . Presence of a unique cysteine allows the fluorescent labelling of this protein through thiol coupling 4 . Pure M18-L-ST·His-SC-L-DBD(Cys) complexes with four DBD domains per tetrameric M18 (Fig. 5b) were generated by the approach using excess His-SC-L-DBD(Cys). The use of the direct fusion version of M18-L-DBD(Cys) for the fluorescent study is not possible because this fusion was found to become insoluble after fluorescent labelling. DBD fusions in either the monomeric or tetrameric state are mainly localized on the Sephadex bead surface from day 0 to day 62 (Fig. 5c). Events of dissociation, diffusion and rebinding of DBD to dextran strands result in the diffusion of the molecules into the centre of the beads. This effect is much more dramatic with monomeric DBD. The disappearance of most of the surface bound DBD moieties and the increase in brightness at the centre of the bead at day 120 were obvious. In contrast, the majority of tetrameric DBD complexes are retained on the bead surface even after four months. This illustrates that avidity is an effective means to strengthen the retention of tetrameric DBD on the Sephadex surface.
Since the binding affinity of DBD to dextran is in the nmolar range 6 while the binding strength of agarose-binding domain to agarose is in the µmolar range 4,22 , the retention ability of the bound monomeric or tetrameric DBD to the Sephadex surface is much better than that of the monomeric or tetrameric agarose-binding domain to agarose. This makes DBD an attractive immobilization tool for affinity matrix preparation.
The fluorescence intensity on the Sephadex bead surface is uneven (Fig. 5c). This is very different from the relatively even distribution of the bound agarose-binding domains on the agarose bead surface 4,22 . Scanning electron microscopic studies illustrate that agarose beads [24][25][26] have a relatively smooth surface. In contrast, Sephadex beads have a crinkled surface 24,25 . Presence of an uneven bead surface can explain the observed uneven distribution of the surface bound fluorescently labelled DBD molecules.  Table 1) or M18-L-ST·SC-L-DBD (1.43 nmoles per ml of Sephadex) was applied to purify SBP-tagged β-lactamase 9,11 and several biotinylated model proteins with different degrees of biotinylation. The rationale for the use of the half-saturated matrices is to ensure that any M18-DBD or M18-L-ST·SC-L-DBD detached from the matrix can easily rebind to the matrix again.
For purification of SBP-tagged β-lactamase, the sample was applied to the affinity column under a non-overloading condition (i.e. not to exceed the column binding capacity of SBP-tagged protein) 9 . SDS-PAGE ( Fig. 6 and Supplementary Fig. S1a) shows that the SBP-tagged protein could be selectively captured and affinity-purified in one step on either column. Recovery in both cases was around 80% within a few column volumes of elution. Both columns were functional even after one year of frequent use.
For capture of biotinylated proteins, chemically biotinylated bovine serum albumin (BSA) with an average of 12 biotin moieties per protein was first selected for testing with both matrices. BSA should bind tightly to the affinity matrix because of the avidity effect (i.e. one BSA can potentially interact with multiple M18). As expected, BSA binds well to both matrices (Fig. 7a for the M18-L-DBD matrix and Supplementary Fig. S1b for the M18-L-ST·SC-L-DBD matrix). To ensure that the bound protein could be efficiently eluted off the column, 10 mM (instead of 5 mM) of biotin was used as the eluent and the column was incubated with the eluent for 1-2 hours before collecting the eluted fractions. BSA could be eluted from either matrix with no difficulties (Fig. 7a and Supplementary Fig. S1b). To confirm that all bound BSA could be eluted from the column, a sample of the affinity matrix post-elution was boiled to analyze the bound fraction. As shown in Fig. 7a (lane BF) only M18-L-DBD was detected. The same is true for the M18-L-ST·SC-L-DBD matrix (lane BF of Supplementary Fig. S1b).
To estimate the binding capacity of both matrices, biotinylated BSA was applied at an overloading amount. Analysis of the protein distribution in the flow-through, wash and eluted fractions showed that at saturation, 1 ml of M18-L-DBD affinity matrix (containing 2.5 nmoles of tetrameric M18-L-DBD) could bind around 252 µg ± 3 µg (3.8 nmoles, Table 1) of biotinylated BSA. In a tetrameric streptavidin, two biotin-binding sites are localized on one side while the other two are on the opposite side ( Fig. 1b and Supplementary Fig. S2) 13,27 . Since BSA is relatively bulky with a molecular mass of 66.43 kDa, binding of one BSA molecule to a biotin-binding pocket will likely hinder the binding of another BSA to the biotin-binding pocket on the same side ( Supplementary Fig. S2). Therefore, one tetrameric streptavidin will likely bind a maximum of two BSA molecules, one on each side of the tetrameric streptavidin. Capture of a maximum of 5 nmoles of biotinylated BSA/ml matrix would be expected in this case. As one biotinylated BSA (12 biotin moieties/BSA) can interact with two or more tetrameric streptavidin, capture of 3.8 nmoles of BSA per ml matrix seems to be reasonable. This translates to a capture efficiency of 76% (Table 1). Similar binding efficiency was observed with the M18-L-ST·SC-L-DBD affinity matrix. In another study, the Sephadex matrix was fully saturated with M18-L-DBD (i.e. 5 nmoles of M18-L-DBD immobilized per ml of Sephadex). This matrix was found to capture biotinylated BSA with a capture efficiency of 73% (Table 1).
The second chemically biotinylated protein studied is the human fibroblast growth factor (FGF) with a biotin to protein ratio of 4.8. In comparison with BSA, this protein should have less avidity effect in binding. To demonstrate the selective purification of biotinylated FGF from a mixture of proteins, pure biotinylated FGF was mixed www.nature.com/scientificreports www.nature.com/scientificreports/ with a crude E. coli cell extract and the sample was loaded onto either column. Figure 7b and Supplementary  Fig. S1c show that despite the lower avidity effect, FGF was effectively captured and selectively purified in one step by either matrix. No leakage of the protein was observed in the wash fractions. Protein recovery was in the range of at least 80% within a few column volumes.
Finally, two enzymatically biotinylated proteins [pure maltose binding protein (MBP-AviTag from Avidity) and a pure staphylokinase with biotinylation tag 28 ] with one biotin per protein were used as the model proteins for purification. The binding of these proteins to either column was not tight enough to allow effective purification. The inability of these matrices to bind proteins with single biotin per protein is unexpected since the M18 affinity matrix (M18 Affi-gel) generated by chemical coupling has been successfully applied to purify monobiotinylated MBP-AviTag 9 . Replacement of M18 in M18-L-DBD with either another mutein that has a higher (yet still reversible) biotin binding affinity or a different linker may address this limitation. Should purification of monobiotinylated proteins be the major application, the use of M18 Affi-gel matrix 9 is recommended.
A remarkable feature for the M18-DBD matrices is that the affinity capture, elution and regeneration processes can be operated under mild, protein-friendly conditions (i.e. PBS ± biotin). SBP tagged or biotinylated proteins  www.nature.com/scientificreports www.nature.com/scientificreports/ are selectively and highly purified (Figs 6-7 and Supplementary Fig. S1). If more stringent conditions are needed, PBS or Tris-buffered saline containing mild detergents (Tween 20 or Triton X-100 to 0.1%) with different salt concentrations (0.1 M to 0.25 M NaCl) and different pH (7.0 to 8.0) have been tested. The experiments worked well under room temperature or at 4 °C. All these operating conditions do not trigger any detachment of the bound M18-DBD from the matrix.

Comparison of the streptavidin affinity matrices generated by different coupling methods. Affinity matrices generated by bio-coupling (using the direct fusion and bio-ligation approaches) and
chemical coupling were compared in three aspects. The first aspect is the coupling capacity of streptavidin per ml of settled beads. For the bio-coupling approach, each streptavidin subunit is connected to DBD via a linker sequence (Fig. 1b). A tetrameric streptavidin-DBD fusion can occupy a larger surface area of a Sephadex bead during the binding event. Under a fully saturated condition, 5 nmoles of tetrameric M18-L-DBD can be coupled to 1 ml of Sephadex (Table 1). With the presence of SnoopCatcher/SnoopTag in the linker region as the connection domains, it is not surprising that 1 ml of Sephadex can capture only 2.85 nmoles of M18-L-ST·SC-L-DBD. Chemical coupling of M18 to affi-gel offers a higher coupling capacity and can reach over 15 nmoles per ml of gel. In comparison, a commercially available streptavidin-agarose was reported to have 18-24 nmoles of biotin binding sites per ml of matrix. Therefore, the number of biotin binding sites (20 nmoles) available from 1 ml of the M18-L-DBD Sephadex is fairly comparable.
The second aspect is the capture efficiency of the matrices. With the ability to chemically couple more M18 to affi-gel, this matrix can capture more biotinylated BSA (Table 1). However, on molar basis, one mole of chemically coupled tetrameric M18 could only capture 0.98 mole of biotinylated BSA (Table 1). In contrast, one mole of tetrameric M18-L-DBD generated by bio-coupling could capture approximately 1.46 moles of biotinylated BSA. Thus, the effective column binding capacity (on molar basis) is considerably lower with the matrix generated by chemical coupling. Assuming one tetrameric streptavidin can bind two biotinylated BSA molecules with one BSA on either side of the tetrameric streptavidin ( Supplementary Fig. 2), the capture efficiency (Table 1) of biotinylated BSA by M18-affi-gel is only 48.9% while the capture efficiency for columns generated by bio-coupling can reach 73-76%. Among other factors, the lack of orientation-specific capturing molecules and the indiscriminating activation of lysine residues (some of which might be critical for the capturing function) of M18 during chemical coupling might be contributing to this lower effective binding capacity of the M18 Affi-gel.
The third aspect is the simplicity of matrix preparation. The chemical coupling approach is the most tedious and labor intensive process, requiring three steps including purification, concentration of the capturing agent (e.g. M18), and coupling to the activated matrix. The coupling efficiency is also more difficult to be precisely controlled. In contrast, the direct fusion approach of bio-coupling is the simplest since purification, concentration and immobilization of M18-L-DBD can be consolidated into one-step. For the M18-L-ST·SC-L-DBD approach, although the use of the biotin agarose chromatography is required to remove the excess SC-L-DBD before loading onto the Sephadex, it is not more tedious than the chemical coupling approach. Furthermore, it does not require the use of the expensive, chemically activated matrix for coupling. It is important to note that the extra operation steps evoked in the M18-L-ST·SC-L-DBD system are peculiar in this case because it involved a tetrameric protein. For other monomeric proteins (e.g. monomeric streptavidin), the generation of affinity matrices by this SnoopCatcher-SnoopTag system can be as straightforward and efficient as the direct fusion approach. Crude extracts containing SC-L-DBD and excess SnoopTagged protein can be mixed in a proper ratio for several hours. The bio-ligated sample can then be applied directly to Sephadex to generate the affinity column. This system offers a high degree of flexibility since any SnoopTagged proteins can bio-ligate to SC-L-DBD for affinity matrix preparation. potential applications. Although this study focuses on the simplified preparation of the M18-Sephadex matrix, many other dextran-based matrices including Sephacryl, Superdex and dextran coated silica monolithic media 28 can serve as the support matrices.
Since a wide variety of engineered streptavidin 13,29,30 is available with the biotin binding affinity ranged from lower to higher than wild-type streptavidin, both the wild-type streptavidin and the engineered versions such as streptactin 31 , SAVSBPM32 30 and traptavidin 29 can be fused to DBD. The resulting fusions can be applied for affinity purification and reversible/irreversible immobilization of target proteins. With the SnoopCatcher-L-DBD module described in this study, any proteins of interest fused to a SnoopTag can be easily coupled to the dextran matrices. Furthermore, technologies for preparing dextran-coated biosensor chips 32 , quantum dots 33,34 , magnetic nanoparticles 35,36 and biodegradable hydrogels 17,37 are well established. Coupling of fusions of engineered streptavidin and other proteins of interest to these platforms can be applied for biomolecular interaction studies, single-cell imaging, affinity protein purification, cell isolation, tumor targeting, tissue repair and drug delivery.
Two other potential applications of this system are worth attention. The first is for the easy isolation of genetically engineered therapeutic T cells for immunotherapy. Based on the chimeric antigen receptor (CAR)-T cell technology 38,39 , an antigen specific binding domain (e.g. a tumor specific single-chain antibody fragment) is fused to the transmembrane domain and the intracellular signalling domain of a T cell receptor to create a chimeric antigen receptor. The gene of this rationally designed receptor is then transduced to the T cells isolated from the patient using a viral vector. Reintroduction of these engineered cells to the patient usually shows promising efficacy for the treatment of a specific disease. Since the transduction efficiency is not 100%, a mixture of transduced and non-transduced T cells will be introduced to the patient. This creates variation in efficacy for the treatment. To address this concern, an affinity tag has been introduced to CAR 40 . Transduced cells can then be isolated using the affinity beads coated with the tag specific capturing agents. Based on the same principle, an SBP tag can be introduced to CARs. The transduced cells can be selectively isolated using dextran beads coupled with M18-L-DBD. Enriched cells are then introduced to the patient. www.nature.com/scientificreports www.nature.com/scientificreports/ The second application is the easy generation of artificial antigen presenting cells to stimulate specific T cells 41 . This can be easily achieved by coupling both the biotinylated MHC-peptide complexes and co-stimulatory proteins to the dextran-coated nanoparticles displaying streptavidin-DBD fusions. The resulting nanoparticles can subsequently serve as artificial antigen presenting cells.
Essentially, many other potential applications of the M18-DBD system are possible and can be tailored depending on the needs of the users. The sample (in a total volume of 600 μl) was rotated by using an end-over-end rotator for 15 minutes at room temperature. The beads were then washed 5 times with 1 ml Tris-Buffered Saline (TBS: 50 mM Tris-HCl, pH 7.5; 150 mM NaCl) per wash to remove any unbound proteins and stored in TBS as a 50% suspension at 4 °C. At each time point, a sample of 10-μl suspended beads was mounted onto slides and examined under an Axio Image Z.1 microscope equipped with AxioVision (version 4.9.1.0) software (Zeiss). Images were taken with constant exposure between samples. Optical stacks were taken at 0.5 μm per optical section.

Methods
Purification of SBP-tagged protein and biotinylated proteins using the matrices with either immobilized M18-L-DBD or M18-L-ST·SC-L-DBD. The model SBP-tagged protein was SBP-tagged β-lactamase overproduced by B. subtilis WB800 via secretion 9,11 . The culture supernatant of B. subtilis secreting the recombinant SBP-tagged β-lactamase was concentrated and dialysed against PBS to constitute the crude sample. Biotinylated BSA (12 biotin moieties/protein) was from BioVision. Biotinylated human FGF was from AcroBioSystems. Biotinylated maltose binding protein and biotinylated staphylokinase were generated by enzymatic biotinylation of MBP-AviTag (Avidity) and staphylokinase-biotinylation tag 28 respectively by E. coli BirA 28 .
For all applications except where indicated, a 1-ml Sephadex G-100 column half-saturated with the respective M18-DBD fusion protein was used. This will ensure that any detached M18-DBD can easily rebind to the matrix. As the column capacity is around 708 µg for M18-L-DBD matrix and 610 µg for M18-L-ST·SC-L-DBD matrix, the M18-L-DBD column has around 354 µg of bound M18-L-DBD per ml of matrix while the M18-L-ST·SC-L-DBD column has around 305 µg of bound M18-L-ST·SC-L-DBD per ml of matrix. For each application, the contact time between the loaded sample and the matrix was 30 min. Column equilibration and wash buffers were PBS. Elution buffer was PBS with 5 mM biotin except where indicated otherwise. After sample loading, column was washed with 5-10 column volumes of wash buffer before the elution step. other methods. Chemical coupling of M18 to the matrix was prepared by coupling M18 mutein purified from culture supernatant of B. subtilis WB800[pSAVSBPM18] 9 onto Affi-gel 15 gel (BioRad) at a concentration of 1 mg protein/ml of gel according to the manufacturer's instruction. Column regeneration of the M18-L-DBD matrix through extended buffer wash was detailed in Supplementary Discussion. Schematic drawing of various streptavidin, dextran-binding domain, SnoopTag and SnoopCatcher fusions with their size proportional to the sequence length was prepared using the Illustrator for Biological Sequences program 42 . The three-dimensional model of M18-L-DBD was homology modelled using the Yasara structure package 43 and the ribbon structure was presented using Chimera 44 .

Data Availability
Both the protein and nucleic acid sequence data for the fusions described in this manuscript are available in the Supplementary File. Since SAVSBPM18 is patented, any plasmids containing the sequence of SAVSBPM18 will be released after signing the material transfer agreement.