Development of a yeast cell surface display method using the SpyTag/SpyCatcher system

Yeast cell surface display (YSD) has been used to engineer various proteins, including antibodies. Directed evolution, which subjects a gene to iterative rounds of mutagenesis, selection and amplification, is useful for protein engineering. In vivo continuous mutagenesis, which continuously diversifies target genes in the host cell, is a promising tool for accelerating directed evolution. However, combining in vivo continuous evolution and YSD is difficult because mutations in the gene encoding the anchor proteins may inhibit the display of target proteins on the cell surface. In this study, we have developed a modified YSD method that utilises SpyTag/SpyCatcher-based in vivo protein ligation. A nanobody fused with a SpyTag of 16 amino acids and an anchor protein fused with a SpyCatcher of 113 amino acids are encoded by separate gene cassettes and then assembled via isopeptide bond formation. This system achieved a high display efficiency of more than 90%, no intercellular protein ligation events, and the enrichment of target cells by cell sorting. These results suggested that our system demonstrates comparable performance with conventional YSD methods; therefore, it can be an appropriate platform to be integrated with in vivo continuous evolution.


Results
Cell surface display of nanobodies using SpyTag/SpyCatcher-based protein ligation. We utilised the SpyTag/SpyCatcher system to ligate the nanobodies and anchor proteins in yeast. We designed two gene cassettes to fuse the synthetic anchor protein 649-stalk with SpyCatcher (113 amino acids) and a nanobody with SpyTag (16 amino acids). The two fusion proteins can be bound covalently in yeast via the post-translational isopeptide formation between SpyTag and SpyCatcher. The final fusion product is expected to be transported via the secretion pathway and anchored to the cell wall via a GPI attachment signal (Fig. 1b). We introduced the constructed gene cassettes in yeast cells to test our scheme's feasibility ( Supplementary  Information Fig. S2). We initially used a multicopy pRS425 vector to produce the SpyCatcher-anchor protein and a multicopy pRS423 vector for the nanobody-SpyTag. Flow cytometry analysis demonstrated a successful display of nanobodies on the yeast cell surface when the yeast cells produced both the SpyCatcher-anchor protein and Lys Nb-SpyTag (Fig. 1c). Protein ligation also occurred when we added purified Lys Nb-SpyTag to yeast cells producing only the SpyCatcher-anchor protein (Fig. 1c, Supplementary Information Fig. S2 and Fig. S3). These results indicated that SpyTag/SpyCatcher-based protein ligation worked well to display the separately produced and then ligated nanobodies and anchor proteins. Improvement of display efficiency of SpyTag/SpyCatcher-mediated YSD. We sought to improve the display efficiency of the SpyCatcher-anchor protein because the number of yeast cells displaying the Spy-Catcher-anchor protein was very low in our initial experiment (Fig. 1c). We designed a follow-up experiment to identify an optimal backbone vector to produce the SpyCatcher-anchor protein. The multicopy pRS425 vector, a high copy pULD1 vector 33 , a centromeric pRS415 vector and an integrative pRS403 vector were compared. We changed only backbone sequences, and used the same gene cassette as the initial construct ( Supplementary  Information Fig. S2). Also, the initial multicopy vector used to produce nanobody-SpyTag was changed to centromeric vector. This is because too many copies of nanobody-SpyTag vectors in a cell could make it challenging to identify the strong binder sequences in future studies with in vivo continuous evolution. We investigated the effect of the backbone sequences on the display efficiency and the expression level ( Fig. 2 and Supplementary Information Table S1). The pRS403 strain showed the highest ratio of double-positive yeast cells; thus, it had the highest display efficiency of functional nanobodies (Fig. 2). This is probably because the pRS403 vector was transformed into yeast by stable single-copy integration 34,35 . The expression level of the integrative pRS403 vector was lower than that of the multi-copy pRS421 vector and the high-copy pULD1 vector, indicating that high expression levels did not improve display efficiencies (Supplementary Information Table S1) 35 . We used the pRS403 vector to produce the SpyCatcher-anchor protein and the p415 vector to produce nanobody-SpyTag in the subsequent experiments because the display efficiency is more important than the expression level for library screening.
Next, we investigated the universal applicability of our YSD method using the SpyTag/SpyCatcher system by examining whether we could display various nanobodies randomly isolated from a synthetic library. We designed a synthetic nanobody library based on a consensus framework derived from an anti-β-lactamase nanobody (cAbBCII10) 36 . The framework was combined with randomised complementary determining regions (CDRs) that recapitulate amino acid position-specific variation in natural llama immunological repertoires 5 . Using the Sanger sequencing, we confirmed that the five nanobody clones isolated from the synthetic library had randomised CDRs (Supplementary Information Fig. S4 and Fig. S5). These nanobodies were used to investigate whether our platform can display randomized nanobodies and to show the feasibility of de novo nanobody screening in future studies. We introduced the vectors encoding each synthetic nanobody into Saccharomyces cerevisiae. We found that both HA tag fused to the anchor protein and FLAG tag fused to the synthetic nanobodies were detected with high display efficiencies (Fig. 3). These results confirmed the universal applicability of our system for displaying various nanobodies on the yeast cell surface.

Confirmation of intracellular SpyTag/SpyCatcher-based protein ligation.
During a screening with our system, intercellular protein ligation events, in which the nanobody-SpyTag secreted from a yeast cell is ligated to the SpyCatcher-anchor protein of neighbouring cells, will be unfavourable. We searched for potential intercellular protein ligation using microscopy. We prepared two yeast strains which produced different nanobody-SpyTag proteins (Lys Nb-SpyTag or Syn Nb_e-SpyTag) and SpyCatcher-anchor proteins (strep-tagged or HA-tagged). First, we cultured the two yeast strains separately and stained them with AF488-labelled lysozyme and anti-HA-AF546 antibody. The yeast strain producing Lys Nb-SpyTag or Syn Nb_e-SpyTag was successfully labelled by AF488-labelled lysozyme or AF546-labelled anti-HA antibody, respectively (Fig. 4ab). Next, we cocultured the two strains in one-pot. The co-cultured cells producing Lys Nb-SpyTag or Syn Nb_e-SpyTag were positive for either AF488 or AF546; there were no double-positive cells that showed fluorescence signals of both AF488 and AF546 (Fig. 4c). These results indicated that the isopeptide bond formation between SpyTag and SpyCatcher mostly occurred intracellularly.

Enrichment of target binders by cell sorting.
Cell sorting is a powerful strategy to isolate antigenspecific nanobodies from a library 37 . We further tested the feasibility of SpyTag/SpyCatcher-mediated YSD as a screening platform by enriching yeast cells producing Lys Nb-SpyTag. Lys Nb-SpyTag and Syn Nb_e-SpyTag were co-cultured from 1:999 ratio and the 0.1% mixed sample was generated. The 0.1% mixed sample was incubated with AF647-labelled lysozyme and analysed with flow cytometry (Fig. 5a). The AF647-positive singlet cells (gates A and B) were sorted and recovered. The singlet cells (gate A) were also sorted as a control. These recovered samples were cultured and then incubated again with AF647-labelled lysozyme. Consequently, the Scientific Reports | (2021) 11:11059 | https://doi.org/10.1038/s41598-021-90593-w www.nature.com/scientificreports/ percentage of AF647-positive cells increased from 0.12% to 92.4% during a single enrichment process against the AF647-positive singlet cells (Fig. 5b).

Discussion
Here, we developed a SpyTag/SpyCatcher-based nanobody display system in which the nanobodies and anchor proteins produced by different gene cassettes were ligated in yeast cells. Utilising this system, more than 90% of the yeast cells successfully displayed functional nanobodies on their cell surface (Fig. 2); the display efficiency was comparable with that of conventional yeast cell surface display methods 32 . The high display efficiency was achieved using an integrative pRS403 vector for the stable production of the anchor proteins 34,35 and the improved SpyTag003/SpyCatcher003 system, which facilitated the formation of an irreversible covalent bond at a rate approaching the diffusion limit 23 .
Since the anchor proteins and passengers, for example, nanobodies, are produced separately in our system, the selective, directed evolution of the passengers can be achieved when combined with the in vivo continuous evolution of target genes in future studies. Recently, Liu et al. conducted a similar study that utilised Aga1 as an anchor protein and Aga2-fused nanobodies as passengers 20 . SpyTag, comprising only 16 amino acids, is smaller than Aga2 of 87 amino acids; hence, our system may be less likely to generate detrimental nonsense and missense mutations that inhibit the cell surface display of nanobodies.
Our system may be useful not only as a screening platform but also as a display platform for complex proteins that are difficult to display using conventional methods. For example, a YSD of full-length IgG with secretion and capture strategies has been reported; however, the system requires complex processes, such as the addition In summary, we have successfully developed a SpyTag/SpyCatcher-mediated yeast cell surface display system with a display efficiency of more than 90% and no observed intercellular protein ligation events. Furthermore, the target cells can be easily enriched by cell sorting. Therefore, our system will be a powerful tool for screening libraries constructed with in vivo continuous evolution. Table S2 describes all antibodies used in this study and the dilution ratio.  Yeast cell surface display of various nanobodies derived from a synthetic nanobody library. Five nanobody-encoding nucleotide sequences were isolated from the synthetic nanobody library (see "Methods") and cloned into pRS415 with a FLAG tag and a SpyTag. The multiple alignments of the five nanobodies were also performed ( Supplementary Information Fig. S4). The plasmids were transformed into the BY4741 strain harbouring a plasmid encoding the 649-stalk anchor protein fused with an HA tag. The negative control strain, used to indicate the nonspecific absorption during immunostaining for flow cytometry, did not contain an HA nor a FLAG tag. Each sample was stained using mouse anti-HA tag antibody and AF488-conjugated anti-mouse antibody, or mouse anti-FLAG tag antibody and AF647-conjugated anti-mouse antibody. The data of three independent experiments were represented as means ± standard deviations. Syn Nb nanobody derived from the synthetic library, N.C. negative control, RFI relative fluorescence intensity. This figure was created using Illustrator CS2 (https:// www. adobe. com/).     24 h. Afterwards, the BMMY culture was centrifuged to collect the supernatant and analyse it with SDS-PAGE. For in vitro isopeptide formation, the supernatants were filtered using 0.5 µm filters and concentrated using Amicon Ultra-15 Centrifugal Filters Ultracel-3K (Merck Millipore, Burlington, MA, USA) at 8000×g for 60 min. Then, 10 mL of phosphate-buffered saline (PBS) was added to the Amicon Ultracel-3K unit and centrifuged at 8000×g for 60 min. The buffer replacement procedure was repeated twice.

Secretary production of nanobody-SpyTag in
Generation of a synthetic nanobody library. A synthetic DNA library encoding diversified nanobodies was constructed using two-step overlap-extension PCR (OE-PCR). A custom trimer mix was created to synthesise randomised primers (Glen Research, Sterling, VA, USA). A set of eight primers was synthesised at a concentration of 100 μM (Supplementary Information Table S3). Mixed pools A, B and C containing 2 μM of each primer were prepared. The three mixed pools used different P7_for primers (mixed pools A, B and C used P7a_for, P7b_for and P7c_for, respectively) to create corresponding CDR3 synthetic sequences of 12, 16, or 20 amino acids. One microlitre of each mixed pool was used in a 50 μL PCR reaction that also included 1 × GC buffer, 200 µM dNTPs, 0.04 µM each primer mix, 1.5 mM MgCl 2 and 1.0 unit of Phusion DNA Polymerase (New England Biolabs, Ipswich, MA, USA). The OE-PCR reactions were performed for 35 cycles. The full-length nanobody DNA fragments from each pool were purified using a FastGene Gel/PCR Extraction Kit. The purified nanobody DNA fragments and pULD1-based plasmid for surface display (pYSD) were each digested with SpeI (Toyobo) and SfiI (Takara Bio).
The digested nanobody DNA fragments from mixes A, B and C were mixed at a molar ratio of 1:2:1; the mixture would be referred to as the nanobody DNA library hereafter. The nanobody DNA library and the digested backbone plasmid were mixed at an equimolar ratio and ligated using the Ligation high (Toyobo). We transformed the ligated DNA into E. coli DH5α using chemical transformation and cultured the transformed www.nature.com/scientificreports/ E. coli in LBA media. After incubation, the plasmids from five colonies were extracted and sequenced using the Sanger sequencing method.
Immunofluorescence labelling of yeast cells for flow cytometry. Immunofluorescence labelling was performed to detect nanobody display on the yeast cell surface and measure the percentage of the yeast cells displaying nanobodies or an anchor protein, or both. The fluorescence intensity of the labelled yeast cells was evaluated via flow cytometry. The cell density of each sample was measured at OD600. Approximately 4.5 × 10 6 cells were subjected to immunofluorescence labelling. After centrifugation at 1000×g for 5 min, the cells were washed with PBS (pH 7.2), resuspended in PBS containing 1% bovine serum albumin, and incubated for 30 min at room temperature. At corresponding dilutions (Supplementary Information Table S2), primary antibodies were added for incubation at room temperature with gentle shaking on a rotary shaker for 1 h. The cells were then washed with PBS and incubated with secondary antibodies at room temperature with gentle shaking on a rotary shaker for 1.5 h. To evaluate functional display of anti-lysozyme nanobody, Alexa Flour 647 (AF647)-labeled lysozyme was also added with secondary antibodies. Lysozyme was labelled with fluorescence using an Alexa Fluor 647 Microscale Protein Labelling Kit (Invitrogen Corporation, Carlsbad, CA, USA) to produce Alexa Flour 647 (AF647)-labelled lysozyme. During immunofluorescence labelling, the AF647-labelled lysozyme was added at a dilution ratio of 1:500 with the secondary antibodies. Afterwards, the cells were washed with PBS, suspended in PBS and analysed with a flow cytometer (JSAN; Bay Bioscience, Kobe, Japan). The fluorescence of AF647 was detected with an excitation at 640 nm and emission at 661 ± 10 nm, and AF488 was detected with an excitation at 488 nm and emission at 535 ± 23 nm. Then, the fluorescence intensity of 20,000 yeast cells was displayed in a density plot or a histogram. Data were analysed using the Kaluza software (Beckman Coulter, Brea, CA, USA). In the density plot, the ratio of the upper right (UR) corner of the plot, which represented both AF488-and AF647-positive cells, was quantified. In the histogram, the ratio of the right region which represented strong fluorescent intensity cells were quantified.
Confocal laser scanning fluorescence microscopy to detect potential intercellular protein ligation. In addition to the immunofluorescent labelling described the previous section, Alexa Flour 488 (AF488)-labelled lysozyme was also added with secondary antibodies. The fluorescence labelling of the lysozyme was performed using an Alexa Fluor 488 Microscale Protein Labelling Kit (Invitrogen). During immunofluorescence labelling, the AF488-labelled lysozyme was added at the dilution of 1:100 with the secondary antibodies. The cells were observed by confocal laser scanning fluorescence microscopy (LSM700; Carl Zeiss, Oberkochen, Germany). Fluorescence of AF488 and AF546 was observed using 488 and 561 nm lasers, respectively. The acquired images were processed using the Zen lite software.
Statistical analysis. In Figs. 2 and 3, the data from three independent experiments were represented as means ± standard deviations. Dunnett's test was used to calculate p value in Fig. 2.

Data availability
All relevant data are within the manuscript and its supplementary information file. License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.