All the bacterial strains and plasmids used here are summarized in Table 1. Bacteria Escherichia coli and Bacillus subtilis strains were grown in Luria Bertani (LB) medium (Sigma, St. Louis, MO, the United States) at 37°C and 180 rpm. Bacterial plasmid vectors were transformed into E. coli DH5α or E. coli BL21 (DE3) for protein expression. Pichia pastoris was grown in Yeast Extract–Peptone–Dextrose (YPD) medium at 28°C and 180 rpm. HEK293F cells were maintained at 37°C in a 5% CO2 atmosphere in Dulbecco modified Eagle medium (DMEM) supplemented with 5% fetal bovine serum and streptomycin (50 μg/ml)–penicillin (50 U/ml). The SARS-CoV-2 Spike ectodomain Hexa-pro construction (Table 1) was a gift from Jason McLellan (Addgene # 154754; http://n2t.net/addgene:154754; RRID: addgene_154754) (Hsieh et al., 2020).

Lysozyme gene was amplified from the pLysS plasmid (Millipore Sigma, Novagen) with the oligonucleotide primers Fw-lys-NdeI (CCCATATGG CTCGTGTACAGTTTAAACAACGTG) and Rv-lys-SalI (CGGTCGACTCCACGGTCAGAAGTGACCAGTTCG). The PCR product was digested with the restriction enzymes NdeI and SalI and then cloned into the pLC3-EITH7-SlpA vector in the same restriction sites, and the recombinant plasmid was transformed into E. coli DH5α. In addition, the lysozyme gene was subcloned into pET28-e GFP-SlpA at the NdeI and NotI restriction sites and transformed into E. coli BL21 by electroporation.

EspA gene was amplified with the following primers: pRvEspAXbaI (GCTCTAGATTTACCAAGGGATATTGCTG) and pFwEITSalI (ACGCGTCGACGATATGAATGAGGCATCTAAA). After digestion with XbaI and SalI, the gene was cloned into pGEX-SlpA (Table 1), which was previously digested with the same enzymes. The plasmid was introduced by electroporation into E. coli BL21.

The gene encoding the human rhinovirus (HRV) type 14 3C protease fused to SLAP_TAG was synthesized and cloned in pMAL-c5X and pPICZ alpha B for protein expression in E. coli and P. pastoris, respectively (Gene Universal, Delaware—USA).

For P. pastoris GS115 transformation, 5 μg of pPICZ alpha B—hrv-3c plasmid was linearized with SacI restriction enzyme and transformed into cells through electroporation using 2 mm gap cuvettes (1,500 V, 125 X, 50 lF). Transformed cells were selected by plating on a YPD medium containing Zeocin 100 ug/ml for resistance selection. Isolated colonies were further grown in test tubes containing YPD broth, and after 24 h, protein expression was induced by adding 1% (v/v) pure methanol every 24 h. The supernatants were sampled after 72 h of induction, and the best producer clones were chosen for further experiments.

The pET28-eGFP-SLAP was digested with BamHI and XhoI restriction enzymes, and the SLAP_TAG-containing band was subcloned into SARS-CoV-2 S HexaPro plasmid (Addgene#154754) with the same restriction sites.

E. coli BL21 were grown at 37°C and 180 rpm in liquid LB supplemented with antibiotics until they reached OD₆₀₀ = 0.6. Then, 0.1 mM IPTG was added, and bacteria were grown for another 20 h at 18°C and 180 rpm. Finally, bacteria were harvested and lysed by sonication. The bacterial lysates were clarified by centrifugation.

For Pichia, pastoris protein expression cells were induced with methanol. In brief, P. pastoris cultures were grown in 50 ml of YPD for 48 h until the dextrose was consumed. Then, methanol pulses (200 ul) were supplied every 24 h for 5 days. Finally, P. pastoris were harvested, and the supernatants were collected.

HEK293 cells were transfected using polyethyleneimine (PEI) for protein expression. In brief, 30,000 cells per well were seeded in 24 well plates and incubated at 37°C, 5% CO₂ for 24 h. For transfection, 4 μl of PEI was diluted in 40 μl of DMEM. 500 ng of DNA was added, and the transfection mix was incubated for 20 min at room temperature. Then, the transfection mix was transferred to the cells and incubated for 48 h. Finally, the supernatant was collected to check protein expression.

Purification of 6xHis-tagged proteins was carried out according to the manufacturer's protocol. In brief, benchtop columns were equilibrated with the binding buffer (200 mM NaCl, 50 mM Tris-HCl pH 7.5). Columns were loaded with the bacterial lysate. After being washed, columns were eluted in steps with 10-, 50-, 100-, 300-, and 500-mM imidazole in the equilibration buffer. When compared with the SAC, the IMAC batch format was adopted. Thus, Ni-NTA superflow resin (Qiagen) was equilibrated with buffer (200 mM NaCl, 50 mM Tris-HCl pH 7.5) in 1.5 ml tubes. Cleared bacterial lysates were loaded and mixed in an orbital shaker for 1 h at 4°C. The samples were centrifugated, and, after being washed, the resin was eluted with 500-mM imidazole in an equilibration buffer.

B. subtilis natto was grown in 200 mL of Luria Bertani broth at 37°C and 180 rpm for 48 h. Then, the culture was centrifugated, and the bacteria were washed twice with PBS. The culture was resuspended in PBS with 2% glutaraldehyde and incubated overnight with soft agitation. Next, fixed bacteria were washed twice with PBS and stored in 20 % ethanol. 1 mL of BM correspond to DO600 = 30 of B. subtilis.

BM was weighted in a drying scale (KERN MLS-D), and a calibration curve for dry weight vs. optical density was performed.

To establish an optimized SAC protocol, BM was equilibrated in an initial binding buffer (50 mM Tris-HCl, pH 7.5). Then, the samples containing SLAP-tagged proteins were incubated with the BM at different times and temperatures. After incubation, the samples were washed three times by centrifugation (8,000 rpm) using the binding buffer. Finally, as shown in the results section, elution was evaluated by incubating at different times and temperatures with different elution buffers. Based on our experimental optimization studies, we have determined that the optimal binding conditions are 200 mM NaCl and 50 mM Tris-HCl at pH 7.5 and a temperature of 4°C for 5 min. For the elution step, we found that using a carbonate-bicarbonate buffer (0.1M) with a pH of 10 and a concentration of 200 mM NaCl for 5 min was optimal.

Once the protocol was set, proteins expressed in Escherichia coli, HEK293, and Pichia pastoris were purified from bacterial lysate or cellular supernatant using the final protocol described here.

To analyze stability in time, aliquots of BM were frozen at −20°C. At different times, the samples were unfrozen and used for purifying GFP-SLAP following the protocol developed in this study. To analyze reusage, an aliquot of BM was used for purifying GFP-SLAP following the protocol developed in this study. After elution, the BM was washed with two volumes of the elution Buffer (0.1M carbonate-bicarbonate buffer) with a pH of 10 and a concentration of 200 mM NaCl and then two volumes of binding buffer (200 mM NaCl, 50 mM Tris-HCl pH 7.5). The cleaning process was repeated each time after elution.

Protein samples were dissolved in a cracking buffer and incubated for 5 min at 100°C. Protein electrophoresis was performed at 120 V on 12% SDS-PAGE gel. Gels were stained in a Coomassie Blue solution (20% methanol and 10% acetic acid).

For Western blot analysis, proteins were transferred to a nitrocellulose membrane for 55 min at 15 V using a semi-dry electroblotting transfer unit (Bio-Rad, Hercules, CA, the United States). Membranes were incubated for 1 h with blocking buffer (1% dry skim milk and 0.1% Tween in PBS). Then, membranes were incubated for 1 h with primary antibody diluted in blocking buffer (1/500). After washing with PBS-0.1% Tween, membranes were incubated for 1 h with IRDye fluorophore-labeled secondary antibodies (LI-COR, Lincoln, NE, the United States) diluted in a blocking buffer (1/20,000). Finally, membranes were scanned using the Odyssey Imaging System (LI-COR).

GFP fluorescence measurements were performed at 485/535 nm excitation–emission wavelength, respectively, using FilterMax F5 Microplate Reader in Black 96 Well Plates (Thermo).

(H₆)-GFP-SLAP_TAG structure was predicted by AlphaFold2 (Jumper et al., 2021). ColabFold web interface was employed using standard settings (five models and no templates).

Adsorption isotherms for (H₆)-GFP-SLAP_TAG on BM were performed using batch experiments. BM was equilibrated with binding buffer (50 mM Tris-HCl, 200 mM NaCl, and pH 7.6). Purified (H₆)-GFP-SLAP_TAG protein at 3 mg/ml in binding buffer was used as a stock solution. Different concentrations of protein were incubated with 10 μl of BM. After reaching equilibrium, the samples were centrifugated and GFP fluorescence from the supernatant was measured. Unbound protein in equilibrium with the BM was calculated using GFP fluorescence. Bound protein was estimated by the difference between input and unbound protein. The adsorption isotherm data were then fitted to the Langmuir isotherm equation to calculate the parameters Qmax and K_D.

The samples were incubated for 30 min in plates treated with poly-L-Lysine (Sigma, St. Louis, MO, the United State). After treatment with PFA (4% in PBS), the samples were washed twice with PBS. Finally, the samples were observed with a confocal laser-scanning microscope Olympus FV1000 using a PlanApo N (60 × 1.42 NA) oil objective.

Cleavage buffer recommended for commercial HRV3c protease was prepared: 50 mM Tris-HCl, pH 7.0, 150 mM NaCl, 1 mM EDTA, and 1 mM dithiothreitol. GFP-LEVLFQGP-SLAP_TAG protein was bound to the BM. After binding, BM was washed with the same buffer thrice at 4°C. SLAP_ASE was added and incubated at 4°C for 60 min. Percolate containing GFP protein without SLAP_TAG was recovered. As a control, the protease and the cut SLAP_TAG were eluted with Bio-Matrix elution buffer after cleavage using 2M LiCl.

Iron nanoparticles were synthesized by reverse co-precipitation as described by Nadi et al. (2019). In summary, the precursor was prepared by dissolving 0.89 g of FeCl2.4H2O in 90 mL of water. The sample was incubated in stirring for 15 min and sonicated for 10 min to ensure complete dissolution of the salt. The solution was then poured over a solution of ammonium hydroxide diluted 1:2 with water and stirred for an hour. Finally, the sample was washed repeatedly with deionized water.

In brief, 5 ml of Bio-Matrix were resuspended in PBS and iron nanoparticles were added at a final concentration of 40 g/L. The sample was mixed with gentle stirring for 30 min. Then, it was washed repeatedly with PBS 1X and preserved in 20% ethanol.

BALB/c mice were immunized intraperitoneally with 10 μg of the different purified recombinant proteins (GST-SLAP_TAG or GFP) using aluminum hydroxide as an adjuvant. Boosters with 5 μg of protein were further performed at 2 and 4 weeks. One week after the last immunization, the mice were bled, and the serum was stored at −20°C for later use.

Statistical analyses were performed using GraphPad Prism 9 software. Statistical significance was analyzed by one-way ANOVA with Bonferroni.

As described in the Methods section, the SLAP_TAG (Supplementary Figure 1) was fused in-frame to the carboxy terminus of the green fluorescent protein (GFP) to generate the chimeric protein (H₆)-GFP-SLAP_TAG adapted here as a SLAP_TAG reporter protein. As shown in Figure 1A, the structure of the chimeric protein (H₆)-GFP-SLAP_TAG was predicted based on the AlphaFold2 method (Jumper et al., 2021). The beta-barrel corresponding to GFP and the two globular domains corresponding to SLAP_TAG were predicted with high performance, and as expected, linkers and 6xHis-tag, which are flexible regions, showed low prediction values.

As a working chromatographic matrix, a culture of Bacillus subtilis natto was processed as described in the Materials and Methods section to generate a bacterial-derived affinity chromatography matrix, named here Bio-Matrix (BM). Interestingly, although B. subtilis has no S-layer, we have demonstrated previously that this bacterium can be externally covered with SLAP-tagged proteins to generate a recombinant S-layer on its bacterial cell wall, a process that we called decoration (Uriza et al., 2020). Decoration of B. subtilis with SLAP-tagged proteins is possible because teichoic acid and lipoteichoic acid (which are the molecules responsible for the SLAP_TAG association) have the same chemical composition as Lactobacillus acidophilus. Interestingly, in B. subtilis and L. acidophilus, teichoic acid and lipoteichoic acid are distributed homogeneously on their cell wall.

To characterize the binding properties of the SLAP_TAG, the reporter protein (H₆)-GFP-SLAP_TAG was recombinantly expressed in Escherichia coli, purified by affinity chromatography mediated by its Hisx6 tag (H₆), and further incubated with BM under a variety of conditions. Initially, as shown in Figure 1B, the optimal binding temperature of (H₆)-GFP-SLAP_TAG to the BM was evaluated, finding that the best binding efficiency was reached when the incubation was performed at 4°C. To get insights into the SLAP_TAG association dynamics, (H₆)-GFP-SLAP_TAG was incubated for 5, 30, and 60 min with BM, and as shown in Figure 1C, the maximal binding of (H₆)-GFP-SLAP_TAG to BM was reached very rapidly (5 min), showing no significant increments in its association at longer time points. These results indicated that the SLAP_TAG binds very rapidly and with an apparent high affinity to the BM. To quantify the affinity of the interaction between the SLAP_TAG with the BM, an adsorption isotherm was performed to determine the apparent equilibrium dissociation constant as described in the Materials and Methods section. As shown in Figure 1D, the apparent K_D was estimated as 4.7 μM. Interestingly, this dissociation constant value was in the same order as other K_D described for different microbial S-layer proteins and their respective bacterial cell walls (Garduno et al., 1995; Mader et al., 2004; Li et al., 2009). In addition, the maximum adsorption capacity (Bmax) of BM was estimated as 1.152 mmol of (H6)-GFP-SLAPTAG or 53.9 mg of protein per milliliter of BM or 0.0815 g of SLAPTAG protein/g of wet BM.

As it was mentioned, the SLAP_TAG contains the protein region responsible for the association of SlpA to the L. acidophilus cell wall. As reported, L. acidophilus SlpA has the natural ability to self-assemble on the bacterial surface to generate a proteinaceous layer named S-layer (Lortal et al., 1992), a highly ordered wall structure that functions as a protective barrier against bacteriophages, resistance to low pH and proteases, and bacterial adhesion. It was characterized that removal of the S-layer can be performed efficiently by the addition of chaotropic agents like LiCl, a compound that disrupts hydrogen bonds leading to a partial denaturation of proteins and the consequent detachment of the S-layer (Lortal et al., 1992). To study (H₆)-GFP-SLAP_TAG elution from BM, LiCl solution was selected as the positive control. As shown in Figure 2A while PBS has no effect on protein elution, the detachment of BM-associated (H₆)-GFP-SLAP_TAG was achieved with high efficiency with a 2M LiCl solution at room temperature, in a short lapse of 5 min (Figure 2B). As it was mentioned above, lithium solutions have certain deleterious effects on proteins, which is an undesired effect for protein purification, especially for enzyme purification. Therefore, to elute SLAP_TAG-tagged proteins preserving protein structure and therefore their function, different milder alternatives of buffers were explored. As shown in Figure 2C, a series of sugars (monosaccharide and disaccharides) described previously (Fina Martin et al., 2019) were tested for the elution of the SLAP reporter observing only a partial elution efficiency when compared with the LiCl elution control (Figure 2C). Considering that the cationic nature of the SLAP_TAG at neutral pH is critical for the interaction of SLAP_TAG with its membrane-bound ligand, the teichoic acid, a set of cationic compounds were tested for the elution. As shown in Figure 2C, 0.3M of CTAB was able to elute (H₆)-GFP-SLAP_TAG with similar efficiency as the LiCl solutions. Although CTAB was very efficient in this step, we found that this compound was difficult to remove downstream from the elution fraction. Then, taking into consideration that SLAP_TAG has a theoretical isoelectric point (pI) value of 9.92, consequently, we explored whether the modification of pH can be adapted as an elution method. As shown in Figure 2C, when the pH of the carbonate-bicarbonate buffer was close to the theoretical SLAP_TAG isoelectric point (pI), (H₆)-GFP-SLAP_TAG was efficiently eluted from BM. In addition, after establishing the carbonate-bicarbonate buffer at pH 10 as an elution buffer, we explore whether the addition of NaCl can improve the elution process. As shown in Figure 2C, the addition of 200 mM of NaCl was able to maximize the yield of recovery of (H₆)-GFP-SLAP_TAG in the eluate. Interestingly, at pH 7.6, the same concentration of NaCl (present in the binding/washing buffer) has no effect on detaching SLAP-tagged proteins from BM.

With all the experimental information obtained, an optimized SAC protocol for the purification of SLAP-tagged proteins was established as shown in Figure 3A. Remarkably, the entire process of protein purification can be performed in 15 min. To have a direct observation of the purification process of (H₆)-GFP-SLAP_TAG, the BM was fixed and immobilized on coverslips at different steps of the process of protein purification to be observed by confocal microscopy. As shown in Figure 3B, before the incubation of BM with (H₆)-GFP-SLAP_TAG, only a red native autofluorescence was observed from fixed Bacillus subtilis natto cells present in BM. Remarkably, after the incubation of BM with (H₆)-GFP-SLAP_TAG, it was possible to detect the adhesion of the reporter protein to the BM establishing a recombinant fluorescent S-layer that covers completely the bacterial surface (Figure 3C). As expected, after the addition of the elution buffer, the (H₆)-GFP-SLAP_TAG was completely removed from the bacterial surface (Figure 3D). In addition, the whole process of purification can also be monitored by direct observation under UV light (Figure 3E).

Since SAC was efficient in protein purification, we compared this new technique against an established affinity purification protocol. For this, we took advantage of (H₆)-GFP-SLAP_TAG which also has a His-tag that can be purified by a metal affinity chromatography or IMAC. As described in Materials and Methods, a bacterial lysate (H₆)-GFP-SLAP_TAG was divided into two fractions, and both protocols were performed accordingly in parallel, confirming that SAC optimized protocol was able to purify (H₆)-GFP-SLAP_TAG with high efficiency (Supplementary Table 1; Figure 3F) and with a similar yield to the one obtained with the commercial IMAC (Figure 3G).

To evaluate whether SLAP_TAG and SAC can be adapted as a universal protein purification system, a set of the selected proteins were fused to the SLAP_TAG (Figure 4).

Thus, bacterial proteins such as Salmonella flagellin (Figure 5A), the STEC proteins, EspA (Figure 5B), the chimeric EIT (Figure 5C), the B. abortus Omp19 (Figure 5D), the bacteriophage protein T7 lysozyme (Figure 5E), the human viral proteins Rhinovirus 3C Protease (Figure 5E), the mouse Galectin-8 (Figure 5G), and the commercial molecular tag GST (Figure 5H) were fused in-frame with the SLAP_TAG.

As shown in Figure 5, all the selected proteins were purified efficiently with SAC. In addition to the bacterial expression system, different protein expressions systems (yeast and mammalian cells) were also evaluated (Figure 6). As shown in Figure 6A, the viral HRV-3C protease fused to the SLAP_TAG was expressed and purified from Pichia pastoris supernatant. In addition, the SARS-CoV-2 SPIKE fused to the SLAP_TAG was purified from the supernatant of transfected HEK293 (Figure 6B). These results confirmed that the SLAP_TAG can be widely adopted for affinity chromatography purification. It is interesting to note that the protein SPIKE-SLAP_TAG was efficiently purified from a sample with a high protein content, such as a culture medium containing FBS, using SAC (Supplementary Figure 2). These findings indicate that SAC can be potentially employed as a first step to enrich a protein of interest directly from an unclarified feedstock derived from a host cell culture medium in downstream protein production processes.

As described in the Materials and Methods section, a batch of BM was produced, and several aliquots were frozen at −20°C to study time stability. As shown in Figure 7A, at different times, a few aliquots were unfrozen and tested for protein purification of the reporter protein (H₆)-GFP-SLAP_TAG being the BM stable for more than a year that was tested (14 months). In addition, BM reusability capacity was evaluated determining that the matrix can be reused five times with no modification of the protein purification yield (Figure 7B).

As shown in Figures 4, 5, the gene sequence of the viral protease HRV-3C was fused to the SLAP_TAG (named SLAP_ASE), recombinantly expressed, and purified (Figures 8B, D, lane SLAPase) to evaluate its activity. A reporter protein for SLAP_ASE activity was generated (GFP-LEVLFQGP-SLAP_TAG) (Table 1), and a protocol for tag removal by SLAPase was set as described in the Methods section. As shown in Figure 8A, the GFP-LEVLFQGP-SLAP_TAG was expressed (Figures 8B, C, lane Input or IN) and mixed with the BM for 5 min allowing the binding process. After binding, a purified SLAP_ASE was added to the mix and incubated for 60 min at 4°C, releasing a tag-less GFP (Figures 8B, C, E, lanes flow-through or FT) by proteolysis. Cut SLAP_TAG is not observed in the flow-through indicating that was retained by the BM along with the SLAPase. To confirm these steps, post-proteolysis BM was incubated with LiCl solution to recover any residual bound SLAP-tagged protein. As shown in Figures 8B, D (lane Elution or E), the LiCl solution released the SLAPase and the cut SLAP_TAG.

Since we confirmed the efficiency and universality of SAC for the purification of recombinant proteins, we explored a magnetic affinity chromatography alternative for SAC (Figure 9A). As shown in Figures 9B, C, and Supplementary Video 1, the (H₆)-GFP-SLAP_TAG was purified in a few and easy steps. As shown in the SDS-PAGE (Figure 9B), the adapted SAC protocol for magnetic chromatography (BM_mag) was able to purify the reporter protein (H₆)-GFP-SLAP_TAG very efficiently (Figure 9B, lane Elution or E). As shown in Supplementary Table 2, BMmag is identical to BM in terms of purification performance. These results indicate that the binding capacity of BM is not modified by its association with magnetic particles.

Interestingly, this protocol can be adapted to commercial devices that use magnetic force for protein or DNA purification (like King Fisher Flex from Thermo Fisher). In Figure 9C, the entire process of purification was monitored by direct observation under UV light (Supplementary Video 1). Noteworthy, the binding of the BM_mag to GFP quenched the fluorescence of this protein, an effect that was described for transition metal binding to GFP or binding of iron cations to fluorescent proteins (Figure 9C, tube IN+BM_mag) (Richmond et al., 2000; Kim et al., 2021).