DEGREE PROJECT IN BIOTECHNOLOGY, SECOND CYCLE, 30 CREDITS
STOCKHOLM, SWEDEN 2020
Bacterial Display of a Tau- Binding Affibody Construct:
Towards Affinity Maturation
MOIRA EK
KTH ROYAL INSTITUTE OF TECHNOLOGY
SCHOOL OF ENGINEERING SCIENCES IN CHEMISTRY, BIOTECHNOLOGY AND HEALTH
Bacterial Display of a Tau-Binding Affibody Construct:
Towards Affinity Maturation
Moira Ek
Degree project in biotechnology, second cycle, 30 credits
Supervisor: John Löfblom
KTH Royal Institute of Technology Stockholm, Sweden, 2020
i
Abstract
Aggregation of microtubule-associated protein tau is involved in the pathology of several neurodegenerative diseases, including Alzheimer’s disease. The affibody TP4 has been shown to inhibit this aggregation process, and its target-binding positions were previously grafted onto a dimeric affibody scaffold, creating the sequestrin seqTP4. This project constitutes a part of the affinity maturation process of seqTP4, using two different bacterial display methods. An error-prone PCR library was first expressed on Staphylococcus carnosus cells for selection of variants with improved tau-binding properties, resulting in a library of 1.4 × 107 transformants. Flow cytometric tests indicated difficulties in the setup due to nonspecific interactions, and whereas several different approaches to alleviate the problems were investigated, two cell sorting attempts were ultimately unsuccessful. Subcloning of seqTP4 and the library to an Escherichia coli surface display vector resulted in functional surface expression of seqTP4 on E. coli JK321 and BL21 cells, and a BL21 library size of 1.6 × 109 transformants.
An initial flow cytometric test of this library indicates the presence of improved tau-binding variants, paving the way for future cell sorting.
Key words
Microtubule-associated protein tau; Neurodegenerative disorders; Alzheimer’s disease;
Affibody molecules; Affinity maturation; Staphylococcal surface display; Escherichia coli display; Flow cytometry; Cell sorting
ii
Sammanfattning
Aggregering av mikrotubuli-associerat protein tau är involverad i patologin av flera neurodegenerativa sjukdomar, däribland Alzheimers sjukdom. Affibodymolekylen TP4 har visat sig inhibera denna aggregeringsprocess, och överföring av dess målbindande positioner till ett dimeriskt affibodyprotein har tidigare gett upphov till seqTP4, en så kallad sequestrin.
Detta projekt utgör ett led i processen att affinitetsmaturera seqTP4, med hjälp av två olika metoder för presentation av proteiner på ytan av bakterieceller. Ett error-prone PCR-bibliotek uttrycktes först på ytan av Staphylococcus carnosus-celler för selektion av varianter med ökad affinitet för tau, vilket resulterade i ett bibliotek av 1.4 × 107 transformanter.
Flödescytometriska tester tydde på svårigheter i detta upplägg på grund av ospecifika interaktioner, och emedan flera olika angreppssätt för att förmildra dessa problem undersöktes, misslyckades slutligen två cellsorteringsförsök. Omkloning av seqTP4 och biblioteket till en vektor för ytpresentation på Escherichia coli resulterade i funktionellt ytuttryck av seqTP4 på E. coli JK321- och BL21-celler, och ett BL21-bibliotek bestående av 1.6 × 109 transformanter.
Ett första flödescytometriskt test av detta bibliotek tyder på närvaron av varianter med förbättrad förmåga att binda tau, och vägen ligger nu relativt öppen för cellsortering.
iii
Table of Contents
Abstract ... i
Key words ... i
Sammanfattning... ii
1. Introduction ... 5
1.1 Microtubule-associated protein tau ... 5
1.2 The tau-binding affibody construct ... 7
1.3 Display methods ... 7
2. Materials and Methods ... 11
2.1 Electrocompetent S. carnosus cells ... 11
2.2 Electroporation ... 11
2.3 Sequencing of epPCRseqTP4lib... 12
2.4 Initial flow cytometric analyses of S. carnosus ... 12
2.5 Cell sorting of S. carnosus ... 13
2.5.1 Attempt 1 ... 14
2.5.2 Flow cytometric analyses ... 14
2.5.3 Re-cultivation of S. carnosus carrying pScZ1[epPCRseqTP4lib] ... 15
2.5.4 Attempt 2 ... 15
2.6 E. coli display of seqTP4 ... 15
2.6.1 Preparation of the expression system ... 15
2.6.2 Flow cytometric analysis of seqTP4-expressing E. coli ... 16
2.7 E. coli display of epPCRseqTP4lib ... 16
2.7.1 Cloning of epPCRseqTP4lib into pBad-2.2 ... 16
2.7.2 Preparation of pBad-2.2[epPCRseqTP4lib] E. coli library ... 17
2.7.3 Flow cytometric analysis of epPCRseqTP4lib-expressing E. coli BL21 ... 17
3. Results ... 18
3.1 Staphylococcal surface display... 18
3.1.1 Library characteristics ... 18
3.1.2 Initial flow cytometric analyses ... 18
3.1.3 Cell sorting attempt 1 ... 22
3.1.4 Flow cytometric analyses to improve sorting conditions ... 22
3.1.5 Cell sorting attempt 2 ... 24
3.2 E. coli display ... 24
4. Discussion ... 27
5. Future Perspectives ... 31
6. Acknowledgements ... 31
7. References ... 32
iv
Appendix A. S. carnosus flow cytometric population data ... 38
Appendix B. S. carnosus flow cytometric MFI data ... 46
Appendix C. E. coli flow cytometric population data ... 48
Appendix D. E. coli flow cytometric MFI data ... 53
5
1. Introduction
With trends of increased longevity and reduced fertility rates, the proportion of elderly people in the world population is increasing [1]. As a consequence, age-related neurodegenerative diseases are a growing concern, with an estimated 47 million people suffering from dementia alone in 2015, with no disease-course modifying treatments being available [2]. One protein that has garnered interest in the search for more efficient treatments for neurodegenerative diseases is the microtubule-associated protein tau.
Tau was first discovered in 1975 as a protein essential for the assembly of tubulin into microtubule in vitro [3]. Since then, it has been shown to be enriched in brain tissue [4], and roles in axonal development [5,6] and transport [7] have been indicated. In addition, the involvement of protein tau in the pathology of several neurodegenerative diseases, including Alzheimer’s disease (AD) [8,9], frontotemporal dementia and parkinsonism linked to chromosome 17 (FTDP-17) [10,11] and progressive supranuclear palsy (PSP) [12] has been established. These diseases share the common characteristic of abnormal tau aggregates being present in the brains of the afflicted [13].
This Master’s thesis project constitutes a part of the affinity maturation process of a previously developed affibody construct directed towards the aggregation-prone region of tau, using cell-surface display methods. Such a higher-affinity variant could be of potential future use as part of a treatment for diseases involving tau pathology, and as a research reagent.
1.1 Microtubule-associated protein tau
Tau consists of an N-terminal projection domain that does not interact with microtubule [14], a proline-rich region, a microtubule-binding domain (MBD), and a C-terminal tail [15]. Six isoforms of tau are known to be present in the adult human brain, varying in length between 352 and 441 amino acids, arising from alternative splicing of exons 2, 3 and 10, giving rise to isoforms having either zero, one, or two N-terminal inserts (0N, 1N, 2N isoforms), and either three or four microtubule-binding repeats in the MBD (3R, 4R isoforms) [15,16] (Figure 1A).
Tau is an intrinsically disordered protein, showing random coil behaviour in solution [17], and stabilizing microtubule through transient interactions [18]. It has been shown that the same regions that are essential for tau interaction with microtubule [15] are involved in tau aggregation [19]. Intracellular so-called neurofibrillary tangles (NFTs), largely made up of paired helical filaments (PHFs) of tau, constitute one of the hallmarks of AD [20,21], and are known to contain all six tau isoforms [16]. These filaments are made of a β-sheet core, originating from the tau repeat region, surrounded by an outer random coil coat [19,22]. Two hexapeptides with high β-sheet propensity that are crucial for PHF formation have been identified in the microtubule-binding repeats, named PHF6 and PHF6*, situated in R3 and R2, respectively [19] (Figure 1A). This, in combination with the fact that constructs consisting of only the microtubule-binding repeats, such as the four-repeat K18 (Figure 1B) [15], are more aggregation-prone than full-length tau (htau40) [23], has made these constructs especially interesting for studies investigating tau aggregation and how it can be inhibited. One mutation found in FTDP-17 patients, the deletion of K280 (ΔK280) [24,25], has been shown to greatly increase the tau aggregation rate [26]. Whereas polyanions are generally used to induce aggregation of tau constructs, which carry a high positive net charge in the aggregation-prone repeat regions [17], ΔK280 leads to fast aggregation even in the absence of such inducers, by
6 stabilizing the β-strand formed in the region around PHF6*, thereby increasing the proportion of molecules whose repeat regions adopt a pro-aggregant conformation [27,28]. In addition, post-translational modifications, especially phosphorylations, have been shown to have roles in both the function and aggregation of tau [29].
There is accumulating evidence that tau oligomers rather than filaments and NFTs constitute the most neurotoxic species of tau, being responsible for symptoms of neurodegeneration [30–
32]. Tau aggregation has been shown to proceed by a rate-limiting nucleation step followed by a fibril elongation phase [33,34]. The lag phase corresponding to the nucleation step can be eliminated through addition of PHF fragments [33] or tau oligomers [35], acting as seeds for the aggregation process. Similarly, tau aggregation has been shown to spread between cells and brain regions [36,37], in a prion-like manner [38], with tau aggregates arising in different regions of the AD brain in a predictable, sequential manner [39]. Recently, a receptor responsible for the cellular uptake of monomers, oligomers and to a certain extent fibril fragments of tau was identified [40]. The microtubule-binding repeats were indicated as the interaction site [40], making this region of tau an attractive target for therapeutic strategies aiming to inhibit the spread of tau pathology in the brain. In this project, the construct K18ΔK280 with two cysteine to alanine substitutions, C291A and C322A, introduced to avoid the formation of intra- and intermolecular disulphide bonds, was used as a target (K18ΔK280AA, Figure 1B).
Figure 1. Overview of tau domains, isoforms (A), and constructs (B). N1-N2 indicate the N-terminal inserts, whereas R1-R4 indicate the microtubule-binding repeats. Green boxes mark the PHF6* and PHF6 hexapeptides, whereas mutations are marked in red. All numbers refer to amino acid positions in full-length htau40.
Abbreviations: Projection, projection domain; MBD, microtubule-binding domain; Pro, proline-rich region; C, C-terminal tail.
7 1.2 The tau-binding affibody construct
Affibodies are small affinity proteins derived from the B-domain of staphylococcal protein A [41–43]. The IgG-binding B-domain [41], consisting of a bundle of three α-helices [44], was engineered for improved chemical stability and ease of manipulation, resulting in the 58-amino acid, cysteine-free Z-domain [42]. Randomization of 13 of its surface-exposed residues allows the generation of affibodies with affinity for new targets [43]. Affibodies constitute an attractive option as affinity reagents, thanks to general properties such as rapid folding [45], small size, high thermal stability and ease of manufacturing, offering important advantages over antibodies in certain applications [46].
This project is based on the tau-binding affibody TP4, which was developed against K18ΔK280 [47]. TP4 was selected from a phage display library created through randomization of the target-binding positions of the amyloid-β (Aβ)-binding affibody ZAβ3 [47,48]. ZAβ3 is an atypical affibody in at least two ways. Firstly, it contains a cysteine residue in helix 2, leading to the formation of disulphide-linked ZAβ3 dimers, which are necessary for achieving the 17- nM affinity for the target. Secondly, the ZAβ3 subunits adopt an atypical secondary structure in complex with the target. Rather than the normal three-helix bundle of an affibody, the amino acids that would normally constitute helix 1 of each ZAβ3 subunit instead form a β-strand. In the complex between the ZAβ3 dimer and Aβ, the ZAβ3 dimer encloses the Aβ peptide in a barrel- like structure, in which these β-strands give rise to an antiparallel beta sheet together with the β-hairpin of the Aβ target. [49] In light of the role of β-sheet structures within tau aggregates, an affibody with a similar binding mode to tau for prevention of tau aggregation can easily be envisioned.
The dimeric TP4 affibody selected from the ZAβ3–based library has reported equilibrium dissociation constants, KD, of about 260 nM from K18ΔK280AA, and about 580 nM from htau40, respectively. Experiments indicate that TP4 can bind to two alternative conformations of K18ΔK280AA, with the binding site involving PHF6 and either PHF6* or a corresponding hexapeptide within R4, i.e. regions crucial for tau aggregation. Furthermore, in addition to preventing tau aggregation at TP4:tau ratios of 1:1, TP4 was seen to introduce an extended lag phase in tau aggregation at ratios of 1:10, indicating an interference with the nucleation of the aggregation process. [47]
In order to attempt to improve the affinity of TP4 for tau, a novel scaffold based on ZAβ3 is employed. By linking the two ZAβ3 subunits in a head-to-tail dimer using a (S4G)2 linker, and removing the 11 N-terminal-most residues of each original subunit, a new, smaller scaffold composed of a single polypeptide chain has been created [50]. This scaffold was used for successful affinity maturation of the Aβ binder [50], with the so-called ZSYM73 variant having a KD in the lower picomolar range [51]. The target-binding positions of TP4 have been grafted onto this so-called sequestrin scaffold, resulting in the head-to-tail dimer seqTP4. An error- prone PCR (epPCR) library consisting of 8 × 107 variants has been prepared from seqTP4, and the aim of this project was to continue the affinity maturation process of seqTP4.
1.3 Display methods
Whereas phage display is the most commonly used method for selecting high-affinity variants from libraries, the cell surface display methods used here carry several advantages. Most importantly, the greater size of cells enables the use of flow cytometric methods for analysis
8 and sorting. Thereby the target-binding characteristics of the individual variants in the library can be analysed pre-sorting, and the enrichment of high-affinity variants can be followed, enabling optimization of sorting conditions. [52]
Flow cytometry is based on the principle of enabling analysis of the particles in a sample by forcing them to pass individually in front of one or several lasers. Scattered light provides information about particle size (forward scatter, FSC) and internal structure (side scatter, SSC), whereas fluorescence emission can provide additional information. In addition to analysis of cell characteristics, cell sorting can be performed by breaking up the sample stream into droplets, which are charged if deemed to contain cells of interest according to predefined criteria in terms of FSC, SSC and fluorescence characteristics, enabling isolation by deflection in an electrical field. [53,54]
The two methods employed here are staphylococcal surface display and Escherichia coli display. Staphylococcus carnosus is a gram-positive, non-sporulating bacterium used in the fermentation of dry sausages [55–57]. Display of recombinant proteins on the S. carnosus surface is achieved by means of the shuttle vector pScZ1 [52,58], containing origins of replication for both staphylococcal and E. coli hosts, as well as genes for ampicillin resistance in E. coli and chloramphenicol resistance in S. carnosus [58]. The affibody-encoding sequence is preceded by a secretion signal and propeptide sequence derived from the Staphylococcus hyicus lipase gene [58–60], and followed by a sequence encoding a streptococcal protein G- derived albumin-binding protein (ABP) [58,61]. This is followed by the X and M regions of staphylococcal protein A, enabling covalent linkage of the protein construct to the cell wall’s peptidoglycan [58,62]. (Figure 2A) Thus, a construct consisting of an N-terminal propeptide, which is not processed in S. carnosus [57,60], an affibody variant, and a C-terminal ABP is attached to the cell wall (Figure 2B). Whereas the propeptide is required for efficient secretion [59], the ABP serves both as a spacer to increase the accessibility of the affibody to its target, and as a tag for surface expression, thereby enabling normalization of target-binding signal to surface expression level [63]. While a surface expression level of around 104 molecules per cell is reported for the system in question [58,64], the level can vary between cells, making normalization important in order to disfavour selection based on expression level rather than target-binding properties [65].
A major drawback of the staphylococcal surface display system lies in the fact that gram- positive bacteria tend to have lower transformation efficiencies compared to gram-negative bacteria [66], limiting the library sizes that can be obtained. E. coli display provides a way of obtaining a larger library size, at the cost of the added complexity of a double-membrane system. Here, surface display is achieved by means of a system based on the autotransporter Adhesin Involved in Diffuse Adherence (AIDA-I) [67,68]. An arabinose-inducible promoter controls the expression of a construct whose N-terminal signal peptide enables translocation of the unfolded polypeptide chain across the inner membrane [67–69]. This signal peptide is followed by the affibody variant, an albumin-binding domain (ABD) [68,70], and a C-terminal domain derived from AIDA-I, which forms a β-barrel in the outer membrane, through which the passenger affibody and ABD are translocated [69], resulting in surface display. (Figure 2C- D) It has been proposed [71] that this kind of system is incompatible with proteins forming an extensive tertiary structure in the periplasm, as has been indicated for a disulphide-containing protein and a different autotransporter system [72]. This problem has been reported to be
9 alleviated when using the E. coli strain JK321, lacking functional DsbA [73], a periplasmic enzyme important for disulphide bond formation [74].
Figure 2. Overview of the staphylococcal (A-B) and E. coli (C-D) display systems. A, C: The genetic structure of the open reading frames of the expression cassettes of both display vectors, shown in a direction corresponding to translation from left to right. B, D: A simplified schematic view of both constructs on the cell surface. The affibody or affibody derivative is illustrated by seqTP4, with the disulphide bond shown in yellow. The dimeric nature of seqTP4 is highlighted, whereas the other domains of the fusion proteins are not depicted to illustrate their structure. Note that the S. carnosus surface consists of a peptidoglycan mesh, to which the XM region (excluded for simplicity) provides covalent linkage, whereas the E. coli construct is anchored in the outer membrane by AIDA-I (excluded for simplicity). Abbreviations: S, signal peptide; PP, propeptide.
For both display systems, the general workflow consists of expressing a single variant from the seqTP4 library on each of a large number of cells. Incubation with biotinylated tau molecules is followed by addition of streptavidin-R-phycoerythrin (SAPE), enabling fluorescent quantification of the target-binding properties of each variant in the library. Addition of fluorophore-conjugated human serum albumin (HSA) enables normalization by surface expression level. Thus, variants with high target-binding capacity can be sorted out, amplified by cell growth, and used as input to another cycle of cell sorting. (Figure 3) The sorting itself can either be performed under equilibrium conditions, with target concentrations optimised to favour preferential labelling of high-affinity variants, or be based on target dissociation rate [75]. Staphylococcal surface display has previously been successfully used for selection of the high-affinity sequestrin ZSYM73, and its ability to discriminate between fine differences in affinity makes it highly suitable for affinity maturation efforts [50,63]. In contrast, E. coli display has not been previously used for display of sequestrins. However, very high surface expression levels, in the range of 105 molecules per cell, have been reported for AIDA-I-based E. coli display systems [76], which provides a means of increasing the separation between variants of interest and background. This is especially relevant in the case of a highly positively charged target such as K18ΔK280AA, since this tends to entail a high level of unspecific binding, potentially complicating the sorting process [65].
10 Figure 3. Overview of the cell sorting process. The plasmid library is transformed into electrocompetent cells through electroporation, generating a cell library from which high-affinity variants can be sorted out. The cell library is first incubated with target (tau-biotin, black dots), followed by SAPE (yellow stars) and HSA- AlexaFluor647 (red stars), for detection of tau-biotin and monitoring of the surface expression level through albumin-binding capacity, respectively. Following flow cytometric analysis, sorting gates are drawn to define the cell population of interest, after which sorting is performed. The sorted-out variants are amplified by cell growth (overnight, O/N), and are used as input to another sorting cycle. Insets on the right depict the sequence of events on the cell surface, showing the two parts of the fusion protein that are common to both cell display systems used in the project. A seqTP4 variant is indicated in green, binding tau-biotin, to which SAPE subsequently binds.
Notably, the varying affinity of different variants in the library for the target will result in varying proportions of the fusion proteins on the cell surface binding to the target. Thus, fluorescence intensity for a cell correlates with the affinity of the variant it is expressing. An albumin-binding region (ABP in the staphylococcal system, ABD in the E. coli system) is indicated in blue, binding to HSA-AlexaFluor647, enabling normalization by surface expression level. In the case of flow cytometric analysis, the workflow is the same, but stops prior to cell sorting.
11
2. Materials and Methods
Unless otherwise stated, all enzymes and accompanying buffers were from New England Biolabs, and all employed kits were from Qiagen and were used according to the manufacturer’s instructions. All electroporations were performed using a MicroPulser (BioRad), and 1-mm cuvettes (Cell Projects, cat. no. EP-101).
2.1 Electrocompetent S. carnosus cells
S. carnosus TM300 [57] cells were grown overnight (O/N, 37⁰C, 150 rpm) in B2 medium (1%
(w/v) casein hydrolysate, 2.5% yeast extract (Merck), 0.5% D-glucose, 2.5% NaCl, 0.1%
K2HPO4 dihydrate, pH 7.5 [77]) to an absorbance at 600 nm (OD600) of 9. B2 medium was inoculated to a starting OD of 0.5, and incubated (37⁰C, 150 rpm) until an OD600 of 4 was reached. The subsequent steps were performed at 4⁰C, with all centrifugations at 4⁰C, 10 min.
The cultures were put on ice to stop growth (15 min), followed by centrifugation (3000xg). The supernatant was discarded, and the pellets were resuspended in residual supernatant, followed by resuspension in autoclaved MilliQ water (Staq, 100% of initial culture volume). This procedure was repeated five more times, with pairs of rcf values and resuspension volumes as specified in Table 1. The cell slurry was then divided into aliquots for storage (-80⁰C). After
>24 h, cell density was estimated through spreading duplicate dilutions on tryptose blood agar base (TBAB) plates, and a contamination test was performed by spreading cells on TBAB plates supplemented with either kanamycin (Km, 50 μg/mL), chloramphenicol (Cml, 20 μg/mL) or ampicillin (Amp, 200 μg/mL), followed by growth at 37⁰C (O/N).
Table 1. Summary of the centrifugal and resuspension specifications used for the washing steps and final resuspension in the preparation of electrocompetent S. carnosus TM300 cells.
Step Relative centrifugal
force (xg)
Resuspension
volume (% of initial culture volume)
Resuspension liquid
Wash 2 4000 100 Staq
Wash 3 5000 54 Staq
Wash 4 5000 54 10% glycerol
Wash 5 5000 18 10% glycerol
Final resuspension 5500 0.6 10% glycerol
2.2 Electroporation
Transformation frequency was estimated by performing test electroporations, using the previously prepared plasmid library pScZ1[epPCRseqTP4lib]. Thawed electrocompetent S.
carnosus TM300 cells were incubated at room temperature (RT, 25 min), followed by heating (56⁰C, 1.5 min). 1 mL of 0.5 M sucrose + 10% glycerol (in MilliQ) was then added per 240 μL of cells, followed by centrifugation (4500xg, 15 min, RT) and resuspension of the pellet in 0.5 M sucrose + 10% glycerol (160 μL). Aliquots (60 μL) were incubated with 2.5 μg or 5 μg of pScZ1[epPCRseqTP4lib] (RT, 10 min), followed by transfer to electroporation cuvettes and electroporation (2.3 kV, 1.1 ms). Immediate addition of B2 medium (940 μL) was followed by
12 incubation (37⁰C, 150 rpm, 2 h) and spreading of duplicate dilutions on Cml (10 μg/mL) plates for growth O/N (37⁰C).
Based on the estimated transformation frequency, five rounds of 20 electroporations were performed, following the procedure described above, using 2.5 μg of plasmid per electroporation reaction (55 μL). Rather than incubating each electroporation reaction individually, 20 electroporations were performed, pooled and incubated. Library size was assessed by spreading duplicate dilutions of each round of electroporations on Cml (10 μg/mL) plates, which were incubated O/N (37⁰C). The remaining cell pool was inoculated to 180 mL tryptic soy broth supplemented with yeast extract (TSB+Y, Merck) and Cml (10 μg/mL) and incubated (37⁰C, 150 rpm, 16-18 h). The cells were then harvested by centrifugation (4570xg, 10 min, 4⁰C), discarding the supernatant, and gently dissolving the pellet in residual medium.
Glycerol was added to a final concentration of 17%, and aliquots were prepared for storage (- 80⁰C). After >24 h, the cell density was assessed by spreading duplicate dilutions of each glycerol stock on Cml (10 μg/mL) plates, followed by incubation O/N (37⁰C).
2.3 Sequencing of epPCRseqTP4lib
In order to assess library characteristics and confirm unbiased transformation to S. carnosus, sequencing of the original and S. carnosus libraries was performed. Dilutions of a previously prepared glycerol stock of E. coli Top10 cells carrying pScZ1[epPCRseqTP4lib] were spread on Amp (200 μg/mL) plates and incubated O/N (37⁰C). Single colonies were picked and sent for Sanger sequencing, using primer SAPA23 (5’-GGCTCCTAAAGAAAATACAACGGC- 3’).
Dilutions of a glycerol stock of S. carnosus TM300 cells containing pScZ1[epPCRseqTP4lib] were spread on Cml (10 μg/mL) plates and incubated O/N (37⁰C). A colony PCR was performed by picking single colonies into MilliQ (30 μL), for cell lysis in a thermocycler (GeneAmp PCR System 9700, Applied Biosystems) (95⁰C, 10 min). PCR reactions were set up using DreamTaq polymerase (Thermo Scientific, Ref EP0702) and primers SAPA23 and SAPA24 (5’-TGTTGAATTCTTTAAGGGCATCTGC-3’), according to the manufacturer’s recommendations. Following gel electrophoretic verification (1% agarose gel), the PCR products were sent for Sanger sequencing, using primer SAPA23.
2.4 Initial flow cytometric analyses of S. carnosus
The general procedure employed (with variations as specified below) consisted of inoculating an aliquot of S. carnosus TM300 cells carrying the plasmid of interest from a glycerol stock into TSB+Y (5 mL) supplemented with Cml (10 μg/mL), followed by O/N cultivation (37⁰C, 150 rpm). All centrifugation steps were carried out at 3500xg, 4⁰C, 6 min. At least 106 cells from the O/N culture were added to phosphate-buffered saline supplemented with Pluronic (PBSP, 137 mM NaCl, 2.68 mM KCl, 1.47 mM KH2PO4, 8.1 mM Na2HPO4, 0.1% (w/v) Pluronic F108 NF Surfactant (BASF Corporation), pH 7.4), followed by centrifugation. The pellet was washed twice with PBSP, followed by resuspension in K18ΔK280AA-biotin [47]
(courtesy of Wolfgang Hoyer, Universität Düsseldorf, mean biotinylation degree 1.7 mol biotin/mol K18ΔK280AA). After incubation (45 min, RT, 15 rpm), the cells were washed twice with ice-cold PBSP, and subsequently resuspended in an excess of ice-cold HSA- AlexaFluor647 (310 nM, in-house prepared) + SAPE (2 μg/mL, Invitrogen, Ref S866) in PBSP.
13 After incubation (30 min, in the dark on ice), the cells were washed twice with ice-cold PBSP, followed by resuspension in ice-cold PBSP and analysis in a Gallios flow cytometer (Beckman Coulter), using blue (488 nm) and red (640 nm) lasers for excitation of phycoerythrin and AlexaFluor647, respectively, and fluorescence sensors FL2 (575/30 nm BP) and FL6 (660/20 nm BP) for detection, counting 100 000 events per sample. Results were analysed using Kaluza Analysis (Beckman Coulter).
An initial test of the library was performed using pScZ1[seqTP4] (previously prepared glycerol stock), pScZ1[TP4-G4S-TP4] (a TP4 dimer with a GS4 linker, previously prepared glycerol stock), pScZ1[epPCRseqTP4lib], and pScZ1[Zwt] (the Z domain, previously prepared glycerol stock) cultures. Each washing step consisted of only one wash, and 200 nM K18ΔK280AA-biotin was used.
In order to investigate the proportion of untransformed cells in the library, pScZ1[seqTP4], pScZ1[TP4-G4S-TP4] and pScZ1[Zwt] were used as controls, whereas both 5-mL and 200-mL cultures with pScZ1[epPCRseqTP4lib] were analysed. 200 nM K18ΔK280AA-biotin was used.
Two tests of the influence of the target concentration were performed on separate days, using pScZ1[Zwt] and either pScZ1[TP4-G4S-TP4] or pScZ1[seqTP4]. Samples from the Zwt culture were incubated with 100 nM and 900 nM K18ΔK280AA-biotin on the first day, and 100 nM, 900 nM and 1.5 μM on the second day. On the first day, samples from the TP4-G4S-TP4 culture were incubated with 100 nM, 300 nM and 900 nM K18ΔK280AA-biotin, whereas samples from the seqTP4 culture were incubated with 100 nM, 500 nM, 900 nM, 1.2 μM and 1.5 μM K18ΔK280AA-biotin on the second day.
In order to assess the equivalence of K18ΔK280AA and the non-mutated 4R microtubule- binding repeat region, a test was performed using pScZ1[seqTP4], pScZ1[TP4-G4S-TP4], and pScZ1[Zwt]. Samples from the Zwt and seqTP4 cultures were incubated with 250 nM K18ΔK280AA-biotin, and samples from the Zwt, seqTP4 and TP4-G4S-TP4 cultures were incubated with 250 nM biotinylated tau-441(244-372) (amino acids 244-372 of htau40, SignalChem, cat. no. T08-55NB-20, hereafter termed tau-441).
In order to investigate the rate of tau dissociation, pScZ1[seqTP4] and a modified procedure was used. Five times the amount of cells was used, with resuspension being performed in 900 nM K18ΔK280AA-biotin (125 μL). After washing post-tau incubation, the pellet was resuspended in PBSP, followed by splitting to five tubes, centrifugation, and resuspension in PBSP (1.4 mL). The samples were incubated at RT (15 rpm), and one by one centrifuged and incubated with HSA-AlexaFluor647 + SAPE (20 min) after varying amounts of time, followed by centrifugation, resuspension and analysis in the Gallios flow cytometer.
2.5 Cell sorting of S. carnosus
Two attempts at sorting out high-affinity variants from the epPCRseqTP4lib were performed, with additional flow cytometric analyses and preparation of new library glycerol stocks preceding the second sorting attempt. The general procedure employed (with variations as specified below) consisted of inoculating 100 times the library size to TSB+Y supplemented with Cml (10 μg/mL), and incubating O/N (37⁰C, 150 rpm). Based on OD600 measurements, 100 times the library size was then taken out and added to PBSP. This was followed by centrifugation and 2x washing with PBSP. The cells were then resuspended n K18ΔK280AA- biotin and incubated at RT (15 rpm, 1 h), followed by 3x washing with ice-cold PBSP.
14 Subsequent resuspension and incubation with an excess of HSA-AlexaFluor647 (310 nM) + SAPE (2 μg/mL) in PBSP (in the dark on ice, 30 min) was followed by 2x washing with PBSP, changing tubes in between washings. A final resuspension in PBSP was followed by analysis in a MoFlo AstriosEQ high speed cell sorter (Beckman Coulter), using 561-585/40 height-log for detection of SAPE, and 640-671/30 height-log for detection of HSA-AlexaFluor647.
Sorting was performed using two gates based on 488-SSC-height-log versus 488-FSC1-height- log and 561-585/40 height-log versus 640-671/30 height-log, respectively, by screening at least 10-fold the library size and sorting cells into TSB+Y. The sorted-out cells were incubated (37⁰C, 15 rpm, 1 h) prior to inoculation into TSB+Y supplemented with Cml (10 μg/mL final concentration). Following O/N incubation (37⁰C, 150 rpm), glycerol stocks were prepared to a final concentration of 15% glycerol, for long-term storage (-80⁰C). At least 10-fold the number of cells that were sorted out in the previous round were taken out from the O/N culture, added to PBSP, followed by repetition of the procedure described above.
2.5.1 Attempt 1
100 times the library size of each electroporation pool was used for inoculation of the O/N culture (50 mL). For the first round of sorting, 300 nM K18ΔK280AA-biotin (200 μL), enrich sorting mode, a drop envelope of 1-2 and a screening rate of about 20 000-40 000 events per second (EPS) was used. In the second round of sorting, 300 nM K18ΔK280AA-biotin (50 μL), purify sorting mode, a drop envelope of 1-2, and about 1000 EPS was used. For the third round of sorting, 300 nM K18ΔK280AA-biotin (50 μL) was used, but no sorting was performed, based on the analysis results. In all rounds, Premium-grade SAPE (Invitrogen, Ref S21388) was used.
2.5.2 Flow cytometric analyses
To investigate the effects of Premium-grade SAPE, as well as the presence of wild-type seqTP4 in the library, flow cytometry was performed as described (section 2.4), using 200-mL O/N cultures with pScZ1[seqTP4], pScZ1[epPCRseqTP4lib], and pScZ1[Zwt], respectively. In addition to analysing one sample from each culture, one sample containing about 15% cells carrying pScZ1[seqTP4] in a background of cells carrying pScZ1[epPCRseqTP4lib] was analysed. 400 nM K18ΔK280AA-biotin and Premium-grade SAPE was used, with three washes being performed post-tau incubation. 300 000 events were counted for each sample.
In order to investigate the notion that tau aggregation might be taking place on the S.
carnosus cell surface, the possibility that this might occur at a different rate than the interaction between the affibodies and tau, as well as the effect of bovine serum albumin (BSA) on the unspecific interactions, flow cytometry was performed as described (section 2.4), using pScZ1[seqTP4]. Three tau incubation variations were tested: 400 nM K18ΔK280AA-biotin in PBSP (20 min), 400 nM K18ΔK280AA-biotin in PBSP (45 min), and 400 nM K18ΔK280AA- biotin in PBSP + BSA (3% (w/v), MP Biomedicals, LLC, cat. no. 160069) (45 min), respectively. Three washes were performed post-tau incubation, and 300 000 events were counted for each sample.
15 2.5.3 Re-cultivation of S. carnosus carrying pScZ1[epPCRseqTP4lib]
In order to decrease the proportion of untransformed cells, 100 times the library size of each electroporation pool was inoculated to 200 mL TSB+Y supplemented with Cml (10 μg/mL), and cultured (37⁰C, 150 rpm, O/N). The cells were harvested as described (Section 2.2). After
>30 min at -80⁰C, duplicate dilutions were plated on both TBAB plates and Cml (10 μg/mL) plates in order to asses cell density and proportion of untransformed cells. The glycerol stock was used to inoculate an O/N culture of TSB+Y (5 mL) supplemented with Cml (10 μg/mL) (37⁰C, 150 rpm). A sample was prepared for flow cytometry as described (section 2.4), using 400 nM K18ΔK280AA-biotin, and analysed both on the Gallios flow cytometer and the Astrios cell sorter (counting 100 000 events).
2.5.4 Attempt 2
A second sorting attempt was performed using the newly prepared library glycerol stock for inoculation, and using non-premium grade SAPE (Invitrogen, Ref S866). 100 times the library size was used for inoculation of the O/N culture (200 mL). For the first round of sorting, 400 nM K18ΔK280AA-biotin (200 μL), enrich sorting mode, a drop envelope of 1-2, and a screening rate of about 20 000 EPS was used. The gates used during the first FACS attempt were used as a basis, with the 561-585/40 height-log versus 640-671/30 height-log gate being extended to include somewhat lower 561-585/40 height-log values. For the second round of sorting, 400 nM K18ΔK280AA-biotin (50 μL), purify sorting mode, a drop envelope of 1, and a screening rate of about 2000 EPS was used. After the second round of sorting, a contamination test was performed by plating dilutions of both the pre- and post-sorted cells on both TBAB and Cml (10 μg/mL) plates. For the third round of sorting, 400 nM K18ΔK280AA-biotin (50 μL) was used, but no sorting was performed, based on the analysis results.
2.6 E. coli display of seqTP4
2.6.1 Preparation of the expression system
The vector pBad-2.2 had previously been linearised through PCR and DpnI-treated in order to remove circular plasmid. The seqTP4 sequence was amplified in a PCR reaction using primers
seqTP4-FWD (5’-GCAGTAAATCTCGAGGCAGGCGGTGAAATGGCAAG-3’) and
seqTP4-REV (5’-GTGATGATGCCATGGTTTCGGTGCTTGGGCATCAC-3’), carrying overhangs complementary to the ends of the linearised vector, template pScZ1[seqTP4] and Q5 High-fidelity DNA polymerase (cat. no. M0491L), according to the manufacturer’s recommendations. Following gel electrophoretic verification, gel extraction of the PCR product from an agarose gel (1%) was performed using a QIAquick Gel Extraction Kit (Ref 28706), followed by gel electrophoretic purity assessment.
The seqTP4 insert was then cloned into the linearised vector using a NEBuilder HiFi DNA Assembly Cloning Kit (New England BioLabs, E5520S). A negative control containing vector and MilliQ was included. The products were used for heat-shock transformations into NEB 5- alpha Competent E. coli, using Amp (200 μg/mL) plates for O/N growth (37⁰C). A colony PCR was performed in order to screen for colonies containing the desired insert, by picking colonies into MilliQ (30 μL). Cell lysis was performed in a thermocycler (95⁰C, 10 min), followed by a PCR reaction using primers FLFI55 (5’-TTCTGTAACAAAGCGGGACCAAAGC-3’) and ANKE55 (5’-CATGGCCCTGAATTGCTTACG-3’), and DreamTaq polymerase, according to
16 the manufacturer’s recommendations. Following gel electrophoretic verification of the insert size, two colonies were chosen, and inoculated into TSB supplemented with Amp (100 μg/mL) for O/N growth (37⁰C, 150 rpm). Glycerol stocks were prepared, to a final glycerol concentration of 15%, for long-term storage (-80⁰C), and plasmids were purified using a QIAprep Spin Miniprep Kit. The purified plasmids were used for heat shock transformation into chemically competent E. coli JK321 (in house-prepared), and BL21 Star DE3 (Invitrogen) cells using the following procedure. Plasmid (50 ng, or MilliQ (negative control)) was mixed with 5xKCM (1x final concentration) and chilled on ice (5 min). Cells (10 μL) were thawed on ice and added to the DNA, followed by incubation on ice (20 min). The subsequent heat shock (42⁰C, 45 sec) was followed by incubation on ice (2 min) and addition of TSB medium (300 μL). Following incubation (37⁰C, 15 rpm, 1 h) the cells were spread on Amp (200 μg/mL) plates and incubated O/N (37⁰C). A colony PCR was performed for verification, according to the FLFI55/ANKE55 protocol described above. These colonies were subsequently sent for Sanger sequencing using primer EKMO1 (5’-AGCCAGAGGACATGGTTTTGTCC-3’).
2.6.2 Flow cytometric analysis of seqTP4-expressing E. coli
10 mL of Difco LB Broth, Lennox (BD, cat. no. 240230) supplemented with Amp (100 μg/mL) was inoculated with a single E. coli colony carrying pBad-2.2[seqTP4] or pBad-2.2[Zwt] (previously prepared) and incubated ON (37⁰C, 150 rpm). The cultures were then diluted 1:100 and grown (37⁰C, 150 rpm) until reaching an OD600 of 0.5-0.8, at which point L(+)Arabinose was added to a final concentration of 0.60% (w/v). This was followed by incubation O/N (25⁰C, 150 rpm). Flow cytometric analysis was then performed as described (section 2.4), using 300 nM K18ΔK280AA-biotin. Analysis of BL21 Star and duplicate experiments of JK321 were performed on separate days, using two seqTP4 cultures and one Zwt culture.
In order to investigate the target concentration dependence and whether saturation could be achieved, the procedure was repeated using BL21 Star cells, incubating seqTP4 samples with 100 nM, 300 nM, 500 nM, 900 nM and 1.2 μM, respectively, and Zwt samples with 100 nM, 500 nM and 1.2 μM, respectively.
2.7 E. coli display of epPCRseqTP4lib
2.7.1 Cloning of epPCRseqTP4lib into pBad-2.2
An NcoI restriction site was introduced downstream of the seqTP4 variant genes from the original pScZ1[epPCRseqTP4lib] through PCR, using 100-fold the library size as template.
PCR reactions were set up using primers LIHA88 (5’-GCAGGATCCCCTCGAG-3’) and seqTP4-REV, and Phusion DNA polymerase (Thermo Scientific, cat. no. F530S), according to the manufacturer’s recommendations, with an optimised annealing temperature of 66⁰C. The desired product was purified through gel extraction from a SeaKem GTG agarose gel (2%, Lonza, cat. no. 50070) using a QIAquick Gel Extraction Kit. The product was digested with XhoI (cat. no R0146S) and NcoI-HF (cat. no. R3193S) restriction enzymes, according to the manufacturer’s recommendations, with the addition of an initial heating step of DNA and MilliQ (70⁰C, 10 min), cooling to RT prior to adding buffer and enzymes, and using an extended incubation (2 h 45 min). This was followed by gel electrophoretic verification and PCR cleanup using a QIAquick Gel Extraction Kit.
17 The E. coli display expression vector was prepared from a glycerol stock of Top10 cells carrying pBad-2.2[dummy] (previously prepared), using an EndoFree Plasmid Maxi Kit (cat.
no. 12362). Following gel electrophoretic verification, NcoI-HF/XhoI digestion was performed as described above, but at an increased DNA concentration (64 ng/μL) and with an additionally extended incubation time (5 h). Gel verification was followed by extraction of the vector from a GTG agarose gel (2%), using a QIAquick Gel Extraction Kit. For comparison, a single digestion of pBad-2.2[dummy] using only NcoI-HF was performed, with a 37⁰C incubation of 1 h.
Ligation of epPCRseqTP4lib into pBad-2.2 was performed at a molar ratio of 1:3 (vector:insert), using T4 DNA ligase (cat. no M0202L), according to the manufacturer’s recommendations, with the addition of an initial heating step of DNA and MilliQ (50⁰C, 10 min), cooling to RT, and using one fifth of the recommended amount of enzyme with an extended incubation (RT, O/N). This was followed by PCR cleanup using a QIAquick Gel Extraction Kit, eluting in MilliQ. Successful ligation was verified by gel electrophoresis of the product and by FLFI55/ANKE55 PCR, as described (section 2.6.1). A background ligation without insert was performed according to the same procedure.
2.7.2 Preparation of pBad-2.2[epPCRseqTP4lib] E. coli library
The pBad-2.2[epPCRSeqTP4lib] ligation product was transformed to TG1 cells (Lucigen, cat.
no. 60502-2) according to the manufacturer’s instructions, to obtain a roughly 100-fold library coverage, based on test electroporations of the library and background ligation products.
Dilutions of the transformed cells were spread on Amp (200 μg/mL) plates, and the remaining cells were inoculated to TSB+Y supplemented with Amp (100 μg/mL) for O/N cultivation (37⁰C, 150 rpm). A small PCR screen of the plated cells was performed using primers FLFI55 and ANKE55 (section 2.6.1) to verify correct insert size. Half of the O/N culture was harvested as described (section 2.2), whereas the other half was used to prepare a plasmid library, using an EndoFree Plasmid Maxi Kit. Following gel electrophoretic verification of plasmid purity, the library was transformed to E. cloni EXPRESS BL21(DE3) cells (Lucigen, cat. no. 60300- 2), according to the manufacturer’s instructions, to obtain a roughly 20-fold coverage of the epPCR library, based on test electroporations. Dilutions of the transformed cells were spread on Amp (200 μg/mL) plates, and the remaining cells were inoculated to TSB+Y supplemented with Amp (100 μg/mL) for O/N cultivation (37⁰C, 150 rpm). The library was harvested as described (section 2.2), followed by plating of dilutions on Amp (200 μg/mL) plates after freezing (>2 h), in order to assess the cell density. 83 colonies were sent for Sanger sequencing, using primer EKMO1.
2.7.3 Flow cytometric analysis of epPCRseqTP4lib-expressing E. coli BL21
An initial analysis of the BL21 library was performed as described (section 2.6.2), inoculating an aliquot from a glycerol stock of BL21 cells carrying pBad-2.2[epPCRseqTP4lib], as well as single colonies of BL21 Star cells carrying pBad-2.2[seqTP4] and pBad-2.2[Zwt], respectively.
300 nM K18ΔK280AA-biotin was used.
18
3. Results
3.1 Staphylococcal surface display 3.1.1 Library characteristics
As a compromise between practical considerations and the desire to maximise the library size, pScZ1[epPCRseqTP4lib] was transformed into S. carnosus TM300 cells by means of 100 electroporations. This resulted in a library size of 1.4 × 107 transformants, corresponding to 17.5% of the original library size. Sequence characteristics of the seqTP4 variants in the original Top10 E. coli library and the new S. carnosus library are summarised in Table 2.
Table 2. Overview of sequence characteristics of the original epPCRseqTP4lib in E. coli Top10 cells and the new epPCRseqTP4lib in S. carnosus TM300 cells, based on the set of sequences from which those of poor quality had been excluded, with the total number of sequences prior to exclusion in brackets. The statistics refer to mutations within the seqTP4 sequence. Truncated variants refers to variants with a stop codon within the seqTP4 sequence, whereas frameshifted variants refers to those in which the introduction of a frameshift prohibits functional surface expression through its effect on downstream regions.
S. carnosus TM300 E. coli Top10
Number of sequences 178 (186) 182 (190)
Proportion of wild-type (wt) seqTP4 sequences
15% 5%
Proportion of truncated or frameshifted variants
17% 11%
Mean number of mutations per variant
2.3 2.6
Mean number of mutations per non-wt variant
2.8 2.7
Mean number of mutations per non-wt and non- truncated/-frameshifted variant
2.7 2.6
Mean number of non-silent mutations in non-wt and non- truncated/-frameshifted variants
2.1 1.9
3.1.2 Initial flow cytometric analyses
For all S. carnosus flow cytometry experiments, scatter plots and fluorescence intensity distributions, including FL6 gates, are collected in Appendix A, whereas mean fluorescence intensity (MFI) data for all single clones is collected in Appendix B. The average proportion of analysed cells with surface expression of the construct of interest, as judged by FL6 intensity, was 88%, 88%, and 87% for pScZ1[seqTP4], pScZ1[TP4-G4S-TP4] and pScZ1[Zwt], respectively, whereas the values for pScZ1[epPCRseqTP4lib] were 25% for 5-mL cultures and 50% for 200-mL cultures.
Initial flow cytometric tests indicated successful surface expression of all controls, with the TP4 dimer and seqTP4 displaying similar normalised target binding levels, whereas the negative control Zwt displayed a lower mean target binding. (Figure 4) Unexpectedly high FL2
19 values for some cells indicated the presence of a certain extent of target interactions that were not mediated solely by the affibody construct (hereafter termed nonspecific interactions). These could be observed in the negative control population (Figure 4C), contributing to a certain overlap in signal distribution with the positive controls (Figures A1 and A2). Initial library tests showed signs of target binding (Figure 4D-E), and the presence of variants with improved affinity could not be ruled out.
Figure 4. Overview of the initial flow cytometric analyses of S. carnosus TM300 cells carrying either the seqTP4 library or control constructs, after incubation with 200 nM K18ΔK280AA-biotin. A-E: Representative scatter plots showing the distribution of the population in terms of target-binding signal (FL2) and surface expression level signal (FL6), for cells carrying pScZ1[seqTP4] (A), pScZ1[TP4-G4S-TP4] (B), pScZ1[Zwt] (C) or pScZ1[epPCRseqTP4lib] (D-E), with each washing step consisting of two washes. The samples analysed in A-D originated from 5-mL O/N cultures, whereas the sample analysed in E originated from a 200-mL O/N culture. F:
Target binding normalised by surface expression level, as measured by the ratio of mean fluorescence intensities of FL2 and FL6, gated for surface expression (Appendix A), for the different controls. The Zwt values correspond to 39% and 30% of the seqTP4 values for one and two washes, respectively. Experiments with one and two wash steps were performed on separate days.
In order to inform the cell sorting strategy, the effect of the target concentration on the normalised target binding signal and the possibility of achieving saturation was investigated.
Increasing target concentrations in the range of 100 nM to 1.5 μM led to increased target binding signals for the positive controls, with no signs of saturation. (Figure 5) Whereas the mean target binding signal remained low for Zwt with increasing target concentrations, an increase in the extent of nonspecific interactions could be observed, with a certain overlap in signal distribution remaining at high target concentrations (Figure 5C-F).
20 Figure 5. Summary of the influence of the K18ΔK280AA-biotin concentration on the normalised target-binding signal in S. carnosus cells carrying pScZ1[seqTP4], pScZ1[TP4-G4S-TP4], and pScZ1[Zwt], respectively. A-B, D-E: Representative scatter plots giving an overview of the influence of the target concentration on seqTP4- (A- B) and Zwt- (D-E) expressing cells, showing the lowest and highest tested target concentrations, 100 nM (A, D) and 1.5 μM (B, E), respectively. C: Summary of the distribution of target-binding signal (FL2) intensities of all seqTP4 and Zwt samples analysed on the same day, gated to include only expressing cells, as judged by FL6 intensity. F: Target concentration dependence of the ratio of gated mean fluorescence FL2 and FL6 intensities, i.e. normalised target-binding signal, for the three different constructs. TP4-G4S-TP4 and seqTP4 experiments were conducted on separate days, whereas Zwt samples with 100 nM and 900 nM were analysed on both days, with the displayed values being the average +/- one standard deviation.
To investigate whether better signal to background separation could be achieved, and to assess the influence of the mutations in K18ΔK280AA on seqTP4’s binding ability, a comparison with the microtubule-binding repeat region of htau40 was performed. Higher FL2 signals and a wider distribution were observed for tau-441 compared to K18ΔK280AA, for both seqTP4 and Zwt, not resulting in an improved signal separation (Figure 6).
21 Figure 6. Summary of the results of the flow cytometric comparison of K18ΔK280AA and the microtubule-binding repeats of htau40, here termed tau-441. A-B: Scatter plots showing the distribution of surface expression level and K18ΔK280AA-biotin binding of cells carrying pScZ1[seqTP4] (A) and pScZ1[Zwt] (B). C: Histogram showing the distribution of target-binding (FL2) intensities in each of the samples, gated by FL6 intensity. D-E: Scatter plots showing the distribution of surface expression level and biotinylated tau-441 binding of cells carrying pScZ1[seqTP4] (D) and pScZ1[Zwt] (E). F: Normalised target binding, calculated as the ratio of gated mean fluorescence intensity of FL2 and FL6, for seqTP4-, TP4-G4S-TP4- and Zwt-expressing cells incubated with either of the two target constructs. The Zwt values correspond to 39.5% and 46.5% of the seqTP4 values for K18ΔK280AA and tau-441, respectively.
In order to obtain information for the design of a potential off-rate selection sorting cycle, the tau dissociation rate from seqTP4 was investigated. All time points after t=0 showed similar target-binding signal distributions (Figure 7), indicating that equilibrium was reached within the first 40 minutes of RT incubation.
Figure 7. Summary of the influence of dissociation time on seqTP4 K18ΔK280AA-biotin binding, as measured by flow cytometry after varying amounts of incubation time at room temperature in an excess of PBSP post-tau incubation and washing. A: Target binding signal normalised by surface expression level, i.e. ratio of gated mean FL2 and FL6 intensity, as a function of dissociation time. B: Histogram showing the distribution of FL2 intensities in the populations (gated for surface expression, as judged by FL6 intensity) after varying amounts of time.
22 3.1.3 Cell sorting attempt 1
Two attempts at sorting out high-affinity variants from the S. carnosus library were performed, with the rationale of first removing the bulk of low-affinity and non-binding candidates, after which - upon observing an enrichment - the target concentration was to be decreased, possibly followed by an off-rate selection round. During the first sorting attempt, 0.10% and 0.30% of the screened cells were sorted out in round 1 and 2, respectively. During round 3, 0.12% of the screened cells fell into the sorting gates, and no sorting was performed. This first sorting process, summarised in Figure 8, was followed by additional flow cytometric tests in order to improve the sorting conditions prior to performing a second attempt.
Figure 8. Overview of the sorting process. 561-585/40 height-log is described as K18ΔK280AA binding, whereas 640-671/30 height-log is described as Surface expression level, after the fluorophores they are used to detect. Note that FSC/SSC gates were used during each round, in addition to the gates shown here. 300 nM K18ΔK280AA- biotin was used in each round.
3.1.4 Flow cytometric analyses to improve sorting conditions
Sorting conditions were emulated in a flow cytometric test, investigating both the effect of the Premium-grade SAPE used during the first sorting attempt, and a seqTP4 spike-in in the library.
Collecting 300 000 events, a large amount of nonspecific interactions was observed (Figure 9), compared to when non-premium-grade SAPE was used (Figure 10B). Furthermore, spike-in of seqTP4-carrying cells in the library resulted in a reinforcement of a previously observed binding population in the library (Figure 9D-E), reinforcing the notion of a non-negligible seqTP4wt
population being present in the S. carnosus library (Table 2), whereas no obvious binding population at higher FL2 signals was observed (Figure 9C-E).
In search of an explanation for the nonspecific interactions, it was hypothesised that aggregation of tau might be occurring on the staphylococcal surface. If so, the rate of such a process might differ from the association rate of tau to the sequestrin variants, or it might be possible to affect the nonspecific interactions through addition of BSA. However, an initial test showed very similar FL2 intensity distributions for incubations of 20 min, 45 min and 45 min with 3% BSA, with similar extents of nonspecific interactions and similar normalised target- binding signal values (Figure 10), and this line of thought was abandoned.
