Team:Bielefeld-CeBiTec/Nanobodies
-->
The aim of our project was to determine the fertility of those using our product on the basis of concentrations of the cycle hormones estradiol, progesterone and LH (luteinising hormone). In order to determine the concentrations of these hormones, we use a chip- based technology called SAW (surface accoustic wave). Monoclonal antibodies which bind these hormones were required for a specific detection of these hormones.
The low molecular masses of estradiol (272.38 Da) and progesterone (314.46 Da) make the measurement difficult: Since our measuring method SAW detects differences in mass and conventional antibodies are quite large (around 150 kDa), we needed the smallest possible antibodies - nanobodies. We decided to use this form of antibody due to the reason that nanobodies have only one tenth of mass of conventional antibodies and are also lighter than the alternative single-chain variable fragment (scFvs )[1]. We want to detect mass changes with our chip and due to this nanobodies are more suited for our cause: The mass difference between bound and unbound nanobody is relatively larger (15 kDa to ~15,3 kDa) compared to bound and unbound antibody (150 kDa to ~150,3 kDa). Furthermore, nanobodies are not only lighter in weight, they are also significantly smaller than antibodies: 2-3 nm compared to approximately 10 nm of conventional antibodies [1], [2].
There were no existing nanobodies for our application and a camelid based nanobody library as a new source for new nanobodies is not ethical for our project. Therefore, we decided to creat a synthetic nanobody library. In our research, we identified the two-step approach for the design and generation of nanobodies proposed in 2018 by Wagner et al . [3] called grafting as a suitable approach. The successful grafting of a new nanobody requires to find an antibody with an publicly available sequence in order to use its complementarity-determining regions (CDRs). A comprehensive literature review and discussion with the experts Prof. Dr. Kristian Müller as well as Dr. Hanna J Wagner lead to the selection of monoclonal antibodies 10G6D6 as a CDR donor for targeting estradiol [4], [5] and 15G12C12G11 for targeting progesterone (referred to as “C12G11” hereafter)[6], [7]. We have chosen 3DWT nanobody as scaffhold[8].
But how does grafting of antibodies work? A quick guide how to graft:
In a first step of in silico nanobody grafting, the amino acid sequences of both donor (10G6D6 and C12G11, respectively) and acceptor (3DWT) are numbered. There are different numbering schemes for immunoglobulin variable domains available [9]. We decided to use the AHo numbering scheme, because placement of gaps is based on the spatial alignment of known three dimensional structures of immunoglobulin domains and not on sequence variability [10]. The numbering was performed using ANARCI (SAbPred) (http://opig.stats.ox.ac.uk/webapps/newsabdab/sabpred/anarci/). For further work, the numbered residues were transferred to an excel sheet. Both, the CDRs and the functional important framework residues of the donor (10G6D6 and C12G11) were combined with the acceptor (3DWT) and a new nanobody, the graft, was created.
The grafting results with the exact sequence can be found in table 1 for our estradiol nanobody and in table 2 for our progesterone nanobody.
Using ChimeraX 1.0 (https://www.cgl.ucsf.edu/chimerax/), we manually verified the 3D structure of the grafted nanobodies for steric obstacles to antigen binding: we validated that the grafted residues are sterically impaired residues of the framework.
Subsequently, the amino acid sequences were translated into a complementary DNA sequence using Sequence Manipulation Suite (http://bioinformatics.org/sms2/rev_trans.html).
Furthermore, the sequences were optimized regarding their GC content, number and length of repeats.The created nucleotide sequences were purchased as gene synthesis from IDT and EurofinsGenomics. In general, the nanobody grafting process generates weak binders as the sequence of the scaffold is strongly modified, especially with regard to the CDRs, which can partly form new secondary structures. In addition, core frame work residues are also transferred from the donor to the scaffold, which changes its sequence, perhaps resulting in a changed structure of the scaffold in non-CDR areas. Therefore, the grafted nanobodies have to be improved by suitable affinity maturation measures [3].
As the initial nanobody grafting probably creates nanobodies with a low affinity [3], affinity maturation methods must be applied. maturation methods must be applied to enhance the affinity. Therefore, we applied phage display in combination with error prone amplified nanobody sequences. In order to improve the affinity of our grafted nanobodies, the sequence had to be mutated in a first step, to create diversity as basis for the following selection process. We used error prone PCR (epPCR) according to Cadwell and Joyce from 1992 [11] for random mutagenesis of the nanobody encoding sequences. We did not use the original protocol, but a modified form according to McCullum et al. from 2010 [12].
We tested the epPCR protocol with two polymerases, the OneTaq® DNA polymerase [13] and the Taq DNA polymerase [14] on an mRFP-encoding construct in the vector pJOE. To verify the influence of the generated mutations, we used the Q5® high-fidelity DNA polymerase as positive control. The proof-reading function of this enzymes leads to very low error rates in the amplification process [15]. The mRFP construct was amplified according to the epPCR protocol and the amplification product was then cloned into pJOE via Gibson assembly. The resulting vector pJOE::mRFP was transformed into Escherichia coli DH5α cellsvia heat shock. Selection of successfully transformed cells was performed on LB-Agar plates with Amp50. IPTG was added to induce the promoter which controls the transcription of mRFP. After incubation overnight at 37 °C, the colonies were screened using colony PCR. Plates were incubated for one week at 4 °C to support protein folding
The red colonies carry the mRFP construct with no or only minor mutations. White colonies are either colonies with an empty vector or they carry a vector with a mutated mRFP which encodes a non-functional mRFP. As shown in the figures above, usage of the epPCR protocol resulted in significantly more white colonies than red colonies compared to the control (see Fig. 4). It is also evident that some red colonies are still visible in the OneTaq sample (see Fig. 6). In the Taq Polymerase sample, hardly any red colonies are visible (see Fig. 5).
To check whether the white colonies carried the mRFP insert, we performed a colony PCR. We tested 5 clones per polymerase and the OneTaq® DNA Polymerase showed better results:
Because the epPCR conducted with the OneTaq® DNA Polymerase showed better results regarding positive clones (see Fig. 4) as well as more and larger red colonies after transformation (see Fig. 3) compared to the results of the Taq polymerase, we decided to use OneTaq® DNA Polymerase for our epPCR experiments using the protocol from McCullum et al.with our grafted nanobodies. In this protocol, the natural error rate of the Taq polymerase is increased by unfavorable reaction conditions. This is accomplished by: 1) Increased concentration of Taq DNA polymerase; 2) extended polymerase extension time; 3) increased concentration of MgCl2 ions; 4) unbalanced rate of dNTPs; and 5) addition of MnCl2 ions [12].
After testing the ideal annealing temperature for our oligonucleotides, we conducted the epPCR experiments and the amplification of the estradiol-nanobody-graft did work:
However, we had serious problems to amplify the sequence of the progesterone nanobody by epPCR. The epPCR was not successful even after several attempts.
Therefore, we decided to purchase and use a commercial error-prone PCR kit from JenaBioScience [16]. With this kit we were able to amplify the progesterone nanobody encoding fragment.
For our antibody-hormone binding process on the SAW chip we need a library of potential binders to find the best possible binder for our system. Therefore, we performed an error prone PCR (epPCR) of our estradiol and progesteron nanobody sequence to grant a high variability of binders. We chose to test two different epPCR systems to evaluate the best system for our approach. To determine the mutation frequency of the two epPCR systems, we decided to use Illumina MiSeq sequencing. For this examination we tested and compared four different samples: The estradiol nanobody amplificated by the epPCR protocol of McCullum et al. from 2010 [12] (sample “E2_S”), the estradiol nanobody (sample “E2_K”) and progesterone nanobody (sample “P_K”) amplified by epPCR using the kit from JenaBioScience [16] and as a reference the estradiol nanobody amplificated by a OneTaq® High Fidelity DNA polymerase.
All four samples were sequenced by Illumina MiSeq technology to assess the diversity of the library as well as the mutation frequency.
The library samples were first checked for fragment length and successful removal of the Illumina adapters. This was verified using the Agilent Bioanalyzer 2100 with the High Sensitivity DNA kit[17]. All samples showed an optimal library size and a complete removal of the unbound adapter fragments and were ready for the sequencing.
Subsequently, the samples were sequenced. We chose a paired end approach with 250 bp per read to get a complete and high coverage of the 410 bp long nanobody sequence.
After the successful sequencing, the quality of the reads was checked with the tool FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/) and processed accordingly. Shown are the results of the sample E2_S.
The first step of read processing is to filter the raw data for high quality reads and drop the reads with low quality. Therefore, we trimmed the reads with the tool Trimmomatic (http://www.usadellab.org/cms/?page=trimmomatic) by cutting off the first and last three bases as well as after four consecutive bases with a quality score lower than 20. If the reads still remained a size of 36 or more, they were kept. In this process 94-96% of the reads survived the trimming process. The trimmed library was analyzed as well with FastQC:
As one can see, it is evident that the quality improved significantly through processing (see Fig. 11 and 12). After trimming, the length of reads, which are in terms of their quality, still in the green area increased from around 200 to 260 bp. Furthermore, the amount of reads in the red quality area decreased drastically and no quantile of reads reaches this area.
After trimming of the raw data, the reads were mapped on the associated reference genome (estradiol nanobody or progesterone nanobody). The Mapping was done with the mapping aligner Bowtie2 (http://bowtie-bio.sourceforge.net/bowtie2/index.shtml). The important mapping statistics are collected with the command flagstat of the tool samtools and are shown in the following table:
Nearly all reads were properly paired.
To visualize our data and check for the coverage of our samples, we chose the Integrative Genomic Viewer [18]. IGV is a tool to visualize genomic data and was used by us to compare the reference sequence (our initially grafted sequences) with the sequenced samples to see for mutations and visualize these.
Figure 15 shows an even distribution of the mutation sites. The same two cytosines between base 200 and 250, which were present throughout all samples from the estradiol samples (E2_O, E2_S and E2_K), are present in the P_K sample as well. These facts strengthen the assumption that an error occurred during gene synthesis.
For the analysis of the mutation rate of each sample, we have to call the base for each read on each position on the reference. For this purpose, we wrote a python script, which used the samtools command “mpileup” to get the coverage informations for each position and afterwards count all A’s, T’s, G’s and C’s. Lastly, it writes the position on the reference with the reference base, the total coverage and all counted bases from the different reads in one single excel document.
As can be seen in Table 4, coverage and reference sequence match. Deviations are only marginal.
With the detailed coverage information, it was possible to calculate the mutation frequency per base pair of all four samples. Therefore, we wrote a python script which imports an excel file, counts coverage for all non-reference bases at every position as well as the total coverage, normalizes the counted mutations by the total coverage of the analyzed sample and calculates the mutation frequency. These mutation frequency values are shown in the following table.
Using the data from table 5, the following graph was created.
The calculated mutation frequency for the control (E2_O) is 3,3 ×10-4 mutations per base pair, which is slightly higher than the expected sequencing error of 1.6×10-4-4 [19]. Since the mutation frequencies are of the same order of magnitude and the control is only slightly higher, the increased error rate cannot be explained by sequencing errors but by the natural error rate of the used OneTaq DNA Polymerase (<140*10-6 bases) [13].
The “selfmade” approach using the epPCR protocol from McCullum et al. from 2010 with the OneTaq® DNA Polymerase yielded the highest rate of mutations per base pair and is six times higher compared to the control not using epPCR conditions. This strongly shows the negative effect of the parameters chosen (higher Mg2+ concentrations, presence of MnCl2+, more time for elongation, imbalanced rate of dNTPs and higher concentrastions of DNA polymerase) on the error rate of the used DNA polymerase.
The mutation rate for the JenaBioScience epPCR kit is expected to be between 0,6 to 2% [16]. In our experiments the kit yielded mutation rates per base pair of 1,101 * 10-3 for the estradiol nanobody and 1,283 * 10-3 for the progesterone nanobody, resulting in an average mutation rate per base pair of 1,192 * 10-3 per run of PCR. Although the “selfmade” epPCR protocol yielded a higher mutation rate per base pair we nevertheless chose the kit to generate our library from the nanobody fragments. This is because we could not achieve amplification of the progesterone nanobody by this protocol. That is why we chose the kit in the first place.
After ensuring that random mutagenesis by epPCR had worked, we proceeded to clone the variable fragments into the phagemid vector pZMB0062. A restriction digest with SfiI [20] was used for cloning. SfiI produces two different sticky ends (see Figures 4 and 5), due to the non-palindromic recognition sequence (GGCCNNNNNGGCC), resulting in different sequences of the 5'- and 3'-cleavage sites. This also prevents a religation of the phagemid vector and enables an easy cloning of the variable sequences.
Following the ligation, we transformed the E. coli strain ER2738 using the electroshock transformation protocol. All resulting clones were screened by colony PCR to verify the integration of the insert into the vector.
Due to the overall difficult situation with COVID-19, we only were able to test our original nanobody grafts on their binding affinity. This examination was performed with a Phage ELISA (enzyme linked immunosorbent assay).
For the detailed phage ELISA protocol click here.
A MaxiSorp 96-well plate was coated with streptavidin, subsequently biotin-coupled hormone was added to immobilize and present estradiol or progesterone. Incubation with our nanobodies presenting phages was followed by addition of anti-pVIII antibody with coupled HRP (horseradish peroxidase) and further addition of the substrate ABTS. Finally, the absorbance was measured in a Tecan i-control infinite 200 at 419 nm.
The data (see Table 6) show no significant absorption at 419 nm, neither for the estradiol nanobody nor for the progesterone nanobody. Thus, it can be concluded that, as assumed and predicted, the initial grafts of the two antibodies do not bind their respective antigens or bind them poorly and that their specificity needs to improve by appropriate affinity maturation measures.
As the initial grafts are only poor binders, their affinity needs to be improved. Therefore, the phagemid display will be conducted: In a next first step, the cloning and transformation of our created epPCR library of the two nanobodies will be performed. To achieve all possible variants of the nanobody sequences, we will have to pick several colonies and perform the phagemid display. For the application, we will adapt the panning conditions step by step to the conditions prevailing in urine or salvia (pH, other components, etc.) after the first 3 panning rounds, since the nanobodies should ultimately function in urine and detect the hormones in urine samples too. In total, we plan to perform 7 panning rounds to obtain a good and specific affinity of the nanobody to estradiol and progesterone. The remaining candidates in the last round of the phagemid display will be sequenced and amplified. For purification of the nanobodiies, the IMPACT kit will be used: Using Through Gibson assembly, the nanobody sequence of our affinity maturated nanobodies is cloned into the expression vector pTXB1. The vector carries an intein tag, as well as a chitin-binding domain (CBD), which will be used for purification of the nanobody and the off cleavage of the tag from the protein [21]. The resulting nanobodies are going to be immobilized on the gold surface of the chip using their immobilization tag and experiments on durability and sensitivity will be performed.
Due to their small size, scFvs can easily be expressed as unique units or as fusion proteins in bacteria and also be immobilized on a surface with a high density [22]. To obtain our scFvs for binding the hormones progesterone and estradiol, we have linked the sequences of the light and heavy chain of estradiol [23] and progesterone antibodies [24] with a histidine linker with a length of 15 amino acids. The histidine linker between the chains was later used to immobilize the scFv on the SAW chip.
For cloning the scFv fragments into the vector pTXB1 via Gibson Assembly®, 30 bp overhangs were added to the 3' and 5' end of the fragment. The cloning of the scFvs was initially tested successfully in silico.
pTXB1 is an IPTG inducible expression vector with a size of ~ 6700 bp. It mediates an ampicillin resistance and is suitable for the in-frame insertion of a target gene upstream of a mxe intein protein. We choose this vector over using one of the iGEM standard backbones, due to the possibility to express tag-free.
The vector was used to overexpress our scFvs after successful cloning. Subsequently, the proteins can be isolated with the IMPACT™ Protein Purification System [21].
For cloning our scFvs for binding of progsterone into the pTXB1 backbone, three different cloning strategies were applied and tested. The scFv fragment for progesterone binding was ordered as gene synthesis from IDT and Eurofins, already with overhangs complementary to the backbone.
To clone the scFv fragment into the pTXB1 backbone via Gibson Assembly®, the pTXB1 backbone was opened by a PCR upstream of the Mxe Intein and thus linearized. Phusion DNA polymerase was used for the PCR, as it has a proof-reading function. The primers for opening the backbone mediated a 30 bp overhang complementary to the overhang of the scFv fragment.
1.1 After linearization of the backbone, a gel electrophoresis was performed in which the complete PCR product was inserted. The gel was then stained in ethidium bromide. The resulting fragments were visible on the transiluminator afterwards. With a scalpel the band was cut out of the gel at the ~ 6700 bp mark. The DNA was then recovered by using the Gel Clean up Kit from Machery-Nagel. The isolated DNA fragment and the scFv were then used for the Gibson Assemly®.
1.2 A second approach to preserve the opened backbone was the digestion of the PCR product by the restriction enzyme DpnI. Digestion with this restriction enzyme allows methylated template to be removed from the PCR, leaving only synthesized linearized plasmid DNA. After DpnI digestion, a PCR clean up was performed with the Machery-Nagel PCR and DNA clean up kit to remove debrist. The Gibson Assemly® followed.
In addition to the Gibson Assemly®, the Seamless Cloning method was tested. The Seamless Cloning Kit from Thermo Fisher [25] was used. The Invitrogen™ GeneArt™ Seamless Cloning and Assembly Kits allow in vitro cloning in nearly any linearized vector, in around 30 minutes, without additional DNA sequences, restriction endonucleases or ligation. In this method, the backbone was opened as already described for the Gibson assembly. The opened backbone and the scFv fragment are combined. Then, as with the Gibson Assembly®, the enzyme mix is added to the fragments to be cloned. The Invitrogen™ GeneArt™ Seamless Cloning and Assembly Kits allow in vitro cloning of up to 4 DNA fragments simultaneously in virtually any linearized vector, typically in 30 minutes, without additional DNA sequences, restrictions endonucleases or ligation. In this method, the backbone was opened as already described for the Gibson assembly. The opened backbone and the scFv fragment are combined. Then, as with the Gibson Assembly®, the enzyme mix is added to the fragments to be cloned together.
The third cloning approach was a cloning using the restriction enzymes NdeI and SpeI, followed by a ligation with the T4 ligase. The backbone pTXB1 as well as the scFv fragment has a restriction site for both restriction enzymes. After the restriction the PCR mixture was purified with the PCR Clean Up Kit from Monarch. A ligation with the T4 ligase followed.
All cloning approaches were then transformed into E. coli DH5α. The resulting colonies were tested for the fragment by colony PCR. In case of a positive colony, plasmid isolation was performed with the Plasmidprep Kit from Thermo Fischer. The isolated plasmid was then delivered to the sequencing core facility of the CeBiTec for sequencing.
After successful cloning of our scFv fragment into the expression vector pTXB1 with the goal to specifically bind progesterone, our next goal will be its expression, purification and subsequently immobilization of the ScFv protein on the SAW-chip surface.
For the expression of the target fusion protein (scFv-Mxe), the plasmid pTXB1-scFv will be transformed into E. coli ER2566 and the expression is induced by Isopropyl-β-D-thiogalactopyranosid (IPTG). Due to the toxicity of IPTG, the cells are induced while the exponential phase of growth at the OD600 of 0.5. The expression of the protein may require the optimization of the expression conditions. Therefore, different cultivation temperatures can be compared and analyzed on their protein expression and its activity. We would have tested the protein expression at three different temperatures: 20°C, 30°C and 37°C. For further expression analysis, sampling will be performed after 3, 5, 8 and 16 hours of cultivation. After cultivation, the proteins will be purified using the IMPACT kit (NEB) [21]. This method is based on the principle of affinity chromaography.
In order to verify the expression of the ScFv and its purification, protein concentration will be determined by Bradford assay and the protein expression and purification success of the different cultivation samples will be analyzed by SDS PAGE. Thereafter, the purified ScFv protein will be immobilized on the gold surface of the SAW-chip. For this purpose, we would have connected the isolated protein with the histidine in the linker to the gold surface [26]. After verification of the ScFv immobilization, the binding of our synthesized scFv to an horseradish peroxidase (HRP)-conjugated progesterone for improved photometrically detectability. Thereby, HRP oxidizes various organic and chromogenic substrates in the presence of hydrogen peroxide enabling the spectrophotometrically detection of the protein binding [27]. Thus, the binding ability of ScFv to progesterone is analyzed. The specific progesterone binding ScFv will be used for the SAW-chip based detection.
[1] „Nanobodies: natural single-domain antibodies - PubMed“. https://pubmed.ncbi.nlm.nih.gov/23495938/ (zugegriffen Okt. 28, 2020).
[2] „Molekular- und Zellbiologische Grundlagen | SpringerLink“. https://link.springer.com/book/10.1007/978-3-662-08822-7 (zugegriffen Okt. 28, 2020).
[3] H. J. Wagner, S. Wehrle, E. Weiss, M. Cavallari, und W. Weber, „A Two-Step Approach for the Design and Generation of Nanobodies“, Int. J. Mol. Sci., Bd. 19, Nr. 11, Nov. 2018, doi: 10.3390/ijms19113444. [4] „RCSB PDB - 1JNH: Crystal Structure of Fab-Estradiol Complexes“. https://www.rcsb.org/structure/1JNH (zugegriffen Okt. 28, 2020).
[5] „Highly specific anti-estradiol antibodies: structural characterisation and binding diversity - PubMed“. https://pubmed.ncbi.nlm.nih.gov/11812141/ (zugegriffen Okt. 28, 2020).
[6] „monoclonal anti-progesterone P15G12C12G11 immunoglobulin VHCH1 heavy c - Protein - NCBI“. https://www.ncbi.nlm.nih.gov/protein/AAX47169.1 (zugegriffen Okt. 28, 2020).
[7] O. Dubreuil u. a., „Fine Tuning of the Specificity of an Anti-progesterone Antibody by First and Second Sphere Residue Engineering“, J. Biol. Chem., Bd. 280, Nr. 26, S. 24880–24887, Juli 2005, doi: 10.1074/jbc.M500048200.
[8] R. P. D. Bank, „RCSB PDB - 3DWT: Structure of CabBCII-10 nanobody“. https://www.rcsb.org/structure/3DWT (zugegriffen Okt. 28, 2020).
[9] M. Dondelinger u. a., „Understanding the Significance and Implications of Antibody Numbering and Antigen-Binding Surface/Residue Definition“, Front. Immunol., Bd. 9, 2018, doi: 10.3389/fimmu.2018.02278.
[10] A. Honegger und A. Plückthun, „Yet Another Numbering Scheme for Immunoglobulin Variable Domains: An Automatic Modeling and Analysis Tool“, J. Mol. Biol., Bd. 309, Nr. 3, S. 657–670, Juni 2001, doi: 10.1006/jmbi.2001.4662.
[11] R. C. Cadwell und G. F. Joyce, „Randomization of genes by PCR mutagenesis“, PCR Methods Appl., Bd. 2, Nr. 1, S. 28–33, Aug. 1992, doi: 10.1101/gr.2.1.28.
[12] E. O. McCullum, B. A. R. Williams, J. Zhang, und J. C. Chaput, „Random mutagenesis by error-prone PCR“, Methods Mol. Biol. Clifton NJ, Bd. 634, S. 103–109, 2010, doi: 10.1007/978-1-60761-652-8_7.
[13] N. E. Biolabs, „OneTaq® DNA Polymerase | NEB“. https://international.neb.com/products/m0480-onetaq-dna-polymerase (zugegriffen Okt. 28, 2020).
[14] N. E. Biolabs, „Hot Start Taq DNA Polymerase | NEB“. https://international.neb.com/products/m0495-hot-start-taq-dna-polymerase (zugegriffen Okt. 28, 2020).
[15] N. E. Biolabs, „Q5® High-Fidelity DNA Polymerase | NEB“. https://international.neb.com/products/m0491-q5-high-fidelity-dna-polymerase (zugegriffen Okt. 28, 2020).
[16] J. B. GmbH, „JBS Error-Prone Kit, Molecular Evolution: Random Mutagenesis“, Jena Bioscience. https://www.jenabioscience.com/molecular-biology/cloning-and-mutagenesis/random-mutagenesis-kits/pp-102-jbs-error-prone-kit (zugegriffen Okt. 28, 2020).
[17] „Quality control NGS libraries, Bioanalyzer High Sensitivity DNA | Agilent“. https://www.agilent.com/en/product/automated-electrophoresis/bioanalyzer-systems/bioanalyzer-dna-kits-reagents/bioanalyzer-high-sensitivity-dna-analysis-228249#productdetails (zugegriffen Okt. 28, 2020).
[18] „igvtools | Integrative Genomics Viewer“. https://software.broadinstitute.org/software/igv/igvtools (zugegriffen Okt. 28, 2020).
[19] „Understanding Illumina Quality Scores“, S. 2.
[20] N. E. Biolabs, „SfiI | NEB“. https://international.neb.com/products/r0123-sfii (zugegriffen Okt. 28, 2020).
[21] N. E. Biolabs, „Impact | NEB“. https://international.neb.com/products/protein-expression-and-purification-technologies/e-coli/impact/impact (zugegriffen Okt. 28, 2020).
[22] „Engineered Recombinant Single-Chain Fragment Variable Antibody for Immunosensors | Analytical Chemistry“. https://pubs.acs.org/doi/pdf/10.1021/ac0507690 (zugegriffen Okt. 28, 2020).
[23] R. P. D. Bank, „1D PFV: 1JNH“. https://www.rcsb.org/sequence/1JNH (zugegriffen Okt. 28, 2020).
[24] R. P. D. Bank, „1D PFV: 1DBB“. https://www.rcsb.org/sequence/1DBB (zugegriffen Okt. 28, 2020).
[25] „geneart-seamless-cloning-assembly-mutagenesis-tools-brochure.pdf“. Zugegriffen: Okt. 28, 2020. [Online]. Verfügbar unter: https://tools.thermofisher.com/content/sfs/brochures/geneart-seamless-cloning-assembly-mutagenesis-tools-brochure.pdf.
[26] Z. Shen, G. A. Stryker, R. L. Mernaugh, L. Yu, H. Yan, und X. Zeng, „Single-Chain Fragment Variable Antibody Piezoimmunosensors“, Anal. Chem., Bd. 77, Nr. 3, S. 797–805, Feb. 2005, doi: 10.1021/ac048655w.
[27] A. M. Azevedo, V. C. Martins, D. M. Prazeres, V. Vojinović, J. M. Cabral, und L. P. Fonseca, „Horseradish peroxidase: a valuable tool in biotechnology“, Biotechnol. Annu. Rev., Bd. 9, S. 199–247, 2003, doi: 10.1016/s1387-2656(03)09003-3.
Nanobodies & scFv's
Nanobody Grafting
The low molecular masses of estradiol (272.38 Da) and progesterone (314.46 Da) make the measurement difficult: Since our measuring method SAW detects differences in mass and conventional antibodies are quite large (around 150 kDa), we needed the smallest possible antibodies - nanobodies. We decided to use this form of antibody due to the reason that nanobodies have only one tenth of mass of conventional antibodies and are also lighter than the alternative single-chain variable fragment (scFvs )[1]. We want to detect mass changes with our chip and due to this nanobodies are more suited for our cause: The mass difference between bound and unbound nanobody is relatively larger (15 kDa to ~15,3 kDa) compared to bound and unbound antibody (150 kDa to ~150,3 kDa). Furthermore, nanobodies are not only lighter in weight, they are also significantly smaller than antibodies: 2-3 nm compared to approximately 10 nm of conventional antibodies [1], [2].
In a first step of in silico nanobody grafting, the amino acid sequences of both donor (10G6D6 and C12G11, respectively) and acceptor (3DWT) are numbered. There are different numbering schemes for immunoglobulin variable domains available [9]. We decided to use the AHo numbering scheme, because placement of gaps is based on the spatial alignment of known three dimensional structures of immunoglobulin domains and not on sequence variability [10]. The numbering was performed using ANARCI (SAbPred) (http://opig.stats.ox.ac.uk/webapps/newsabdab/sabpred/anarci/). For further work, the numbered residues were transferred to an excel sheet. Both, the CDRs and the functional important framework residues of the donor (10G6D6 and C12G11) were combined with the acceptor (3DWT) and a new nanobody, the graft, was created.
Figure 1: Map of the grafted estradiol nanobody. The CDRs were taken from the mAb anti-estradiol antibody 10G6D6 and were subsequently grafted onto the scaffold – the nanobody 3DWT. For immobiliziation, a 2xHis tag was added. In accordance with our cloning strategy SfiI recognition sites were added at 5´ and 3´ ends as well as a 6 nt long appendix, because SfiI is an endonuclease which has low efficiency at terminal recognition sites.
Figure 2: Map of the grafted progesterone nanobody. The CDRs were taken from the mAb anti-progesterone antibody 15G12C12G11 and were subsequently grafted onto the scaffold – the nanobody 3DWT. For immobiliziation, a His tag was added. In accordance with our cloning strategy, SfiI recognition sites were added at 5´ and 3´ ends as well as a 6 nt long appendix, because SfiI is an endonuclease which has low efficiency at terminal recognition sites.
Table 1: Design of the estradiol nanobody. The first 44 amino acid residues (AHo numbering) of the donor antibody (10G6), the accepting antibody (3DWT) and the grafted estradiol nanobody are shown. The complementarity determining regions are highlighted in yellow.
Table 2: Design of the progesterone nanobody. The first 44 amino acid residues (AHo numbering) of the donor antibody (C12G11), the accepting antibody (3DWT) and the grafted progesterone nanobody are shown. The complementarity determining regions are highlighted in yellow.
Figure 3: Screenshot taken from ChimeraX analysing the estradiol graft. An alignment between the scaffold 3DWT and CDR donor (10G6D6) was performed to check for sterical impairments with the amino acid residues. Estradiol is depicted in yellow, the scaffold 3DWT in cyan and 10G6D6 is marked in grey. Furthermore, three grafting candidate CDRs of 10G6D6 are shown in red. Searching for impairments, we checked every amino acid residue for sterical clashes between the grafted amino acids and the remaining residues from the scaffold which are not neutralized by the grafting process.
Figure 4: Screenshot taken from ChimeraX analyzing the progesterone graft. An alignment between the scaffold 3DWT and CDR donor 15G12C12G11 was performed to check for sterical impairments with the amino acid residues. Depicted in white is progesterone. Highlighted in blue is the scaffold 3DWT and in red 15G12C12G11. Furthermore, shown in green are the three grafting candidate CDRs of 15G12C12G11. Searching for impairments, we checked every amino acid residue for sterical clashes between the grafted amino acids and the remaining residues from the scaffold which are not neutralized by the grafting process.
Furthermore, the sequences were optimized regarding their GC content, number and length of repeats.The created nucleotide sequences were purchased as gene synthesis from IDT and EurofinsGenomics. In general, the nanobody grafting process generates weak binders as the sequence of the scaffold is strongly modified, especially with regard to the CDRs, which can partly form new secondary structures. In addition, core frame work residues are also transferred from the donor to the scaffold, which changes its sequence, perhaps resulting in a changed structure of the scaffold in non-CDR areas. Therefore, the grafted nanobodies have to be improved by suitable affinity maturation measures [3].
Error-prone PCR (epPCR)
We tested the epPCR protocol with two polymerases, the OneTaq® DNA polymerase [13] and the Taq DNA polymerase [14] on an mRFP-encoding construct in the vector pJOE. To verify the influence of the generated mutations, we used the Q5® high-fidelity DNA polymerase as positive control. The proof-reading function of this enzymes leads to very low error rates in the amplification process [15]. The mRFP construct was amplified according to the epPCR protocol and the amplification product was then cloned into pJOE via Gibson assembly. The resulting vector pJOE::mRFP was transformed into Escherichia coli DH5α cellsvia heat shock. Selection of successfully transformed cells was performed on LB-Agar plates with Amp50. IPTG was added to induce the promoter which controls the transcription of mRFP. After incubation overnight at 37 °C, the colonies were screened using colony PCR. Plates were incubated for one week at 4 °C to support protein folding
Figure 5: LB-Agar plate with E. coli DH5alpha pJOE::mRFP after incubation over night at 37°C and one week of incubation at 4°C. The mRFP was amplified using the Q5® high-fidelity DNA polymerase (NEB) using the standard protocol for Q5 DNA polymerase.
Figure 6: LB-Agar plate with Amp50 and IPTG of transformants of pJOE::mRFP after incubation over night at 37°C and one week of incubation at 4°C. The mRFP was amplified by epPCR using the Taq DNA polymerase for the random mutagenesis.
Figure 7: LB-Agar plate with Amp50 and IPTG of transformants of pJOE::mRFP after incubation over night at 37°C and one week of incubation at 4°C. The mRFP was amplified by epPCR using the OneTaq® DNA polymerase for the random mutagenesis.
The red colonies carry the mRFP construct with no or only minor mutations. White colonies are either colonies with an empty vector or they carry a vector with a mutated mRFP which encodes a non-functional mRFP. As shown in the figures above, usage of the epPCR protocol resulted in significantly more white colonies than red colonies compared to the control (see Fig. 4). It is also evident that some red colonies are still visible in the OneTaq sample (see Fig. 6). In the Taq Polymerase sample, hardly any red colonies are visible (see Fig. 5).
Figure 8: Gel image of the colony PCR. For each polymerase eight samples (5 µl each) were loaded, 100bp plus ladder used. The gel (1.2% agarose, 1x TAE) was run for 25 minutes at 100 V. The amplified mRFP has a size of 800 bp.
Figure 9: Gel image of the estradiol nanobody encoding fragment amplified by error-prone PCR. The numbers 1 to 20 depict the dilution round from which the sample was derived . A 100 bp plus ladder was used for size comparison. The gel (1.2% agarose, 1x TAE) was run for 25 minutes at 100 V. The amplified estradiol nanobody fragment is 400 bp large. The samples were run on two gels under the same conditions.
However, we had serious problems to amplify the sequence of the progesterone nanobody by epPCR. The epPCR was not successful even after several attempts.
Figure 10: Gel image of the progesterone nanobody encoding fragment amplified by error-prone PCR. The numbers 1 to 20 depict the dilution round the sample stems from. A 100bp plus ladder was used for size comparison. The gel (1.2% agarose, 1x TAE) was run for 25 minutes at 100 V. The amplified progesterone nanobody fragment is 400 bp large. The samples were run on two gels under the same conditions. A strong signal is only visible in the first two dilution steps. A weak signal can be seen until dilution 5, afterwards no bands can be seen anymore.
Therefore, we decided to purchase and use a commercial error-prone PCR kit from JenaBioScience [16]. With this kit we were able to amplify the progesterone nanobody encoding fragment.
Figure 11: Gel image of the nanobody encoding fragments amplified by error-prone PCR (error-prone PCR kit). Five samples per nanobody were analyzed. A 100 bp plus ladder was used for size comparison. The gel (1.2% agarose, 1x TAE) was run for 25 minutes at 100 V. The amplified nanobody fragments are ~400 bp large.
Sequencing
Figure 12: Quality distribution of the reads from E2_S pre trimming. Analyzed with FastQC. Quality is measured with Phred score. Red area depicts a quality range from 0 to 20, yellow area from 21 to 28 and green area is a quality range from 29 to 38. The reads are represented by box plots: The average quality of the reads at each position is shown in the blue line. The median quality is represented by the red line. The yellow boxes indicate the quantiles and the extreme values are displayed by the black lines. It can be seen that the quality of the reads decreases with increasing length.
Figure 13: Quality distribution of the reads from E2_S post trimming. Analyzed with FastQC. Quality is measured with Phred score. The red area depicts a quality range from 0 to 20, yellow area from 21 to 28 and green area is a quality range from 29 to 38. The reads are represented by box plots: The average quality of the reads at each position is shown in the blue line. The median quality is represented by the red line. The yellow boxes indicate the quantiles and the extreme values are displayed by the black lines. It is evident that the quality improved significantly through processing.
After trimming of the raw data, the reads were mapped on the associated reference genome (estradiol nanobody or progesterone nanobody). The Mapping was done with the mapping aligner Bowtie2 (http://bowtie-bio.sourceforge.net/bowtie2/index.shtml). The important mapping statistics are collected with the command flagstat of the tool samtools and are shown in the following table:
Table 3: Flagstats for the four sequenced samples E2_O, E2_S, E2_K and P_K. The data was retrieved using the “flagstat” command from samtools analysing the with Bowtie2 mapped data.
| Sample | ||||
|---|---|---|---|---|
| attribute | E2_O | E2_S | E2_K | P_K |
| Reads in total | 733474 | 882572 | 886832 | 948928 |
| Read 1 | 364161 | 437139 | 418312 | 458932 |
| Read 2 | 363718 | 436823 | 416450 | 457526 |
| Properly paired | 726754 (99,85%) | 872830 (99,87%) | 829972 (99,43%) | 912058 (99,52%) |
| Mapped reads | 727879 (99,24%) | 873962 (99,02%) | 834762 (94,1%) | 916458 (96,58%) |
Figure 14: Screenshot from the IGV file of the estradiol samples´ bam files, showing the coverage of the samples E2_O, E2_S and E2_K. Furthermore, deviations from the reference sequence (the initially grafted sequence of the estradiol nanobody) are shown and each base has a distinct colour: Adenine is shown in green, cytosine in blue, guanine in orange and thymine in red. The first paragraph depicts sample E2_O, second paragraph E2_S and the third paragraph E2_K.
As shown in Figure 14, the two epPCR approaches show significantly more mutations than the control. Furthermore, an even distribution of the mutation sites is shown. The most noticeable feature of the IGV visualization are the two cytosines between base 200 and 250, which are present throughout all samples. These two mutations are also present in the control sample E2_O, where otherwise significantly fewer mutations occurred. This indicates that an error occurred during gene synthesis.
Figure 15: Screenshot from the IGV file of the progesterone sample bam file, showing the coverage of the samples P_K. Furthermore, deviations from the reference sequence (initial sequence of the progesterone graft) are shown and each base has a distinct colour: Adenine is shown in green, cytosine in blue, guanine in orange and thymine in red. The first paragraph depicts the sample P_K.
Figure 15 shows an even distribution of the mutation sites. The same two cytosines between base 200 and 250, which were present throughout all samples from the estradiol samples (E2_O, E2_S and E2_K), are present in the P_K sample as well. These facts strengthen the assumption that an error occurred during gene synthesis.
Table 4: Excel table created by the used python script, which shows the coverage of all reads per position. It also shows the percentage of each of the four bases per position in relation to the total number of reads. Shown are the first 10 of 410 lines of the excel sheet of the sample Prog_K.
| Position | Reference Sequence | A | C | G | T | coverage |
|---|---|---|---|---|---|---|
| 1 | A | 339539 | 1294 | 1015 | 602 | 339129 |
| 2 | T | 1385 | 134 | 78 | 343302 | 344756 |
| 3 | T | 5490 | 291 | 157 | 361494 | 366775 |
| 4 | A | 367264 | 316 | 271 | 1410 | 367969 |
| 5 | T | 525 | 141 | 77 | 372295 | 372631 |
| 6 | A | 382438 | 357 | 378 | 2953 | 385930 |
| 7 | G | 494 | 435 | 395276 | 829 | 397093 |
| 8 | G | 627 | 1991 | 397089 | 1907 | 401218 |
| 9 | C | 191 | 413368 | 161 | 799 | 415012 |
| 10 | C | 267 | 414865 | 61 | 594 | 415740 |
Table 5: Mutation frequencies [bp-1] for the four sequenced samples.
| Sample | Mutation frequency [bp]-1>/sup>
| E2_O | 3,302 * 10-4
| E2_S | 1,992 * 10-3
| E2_K | 1,101 * 10-3
| E2_P | 1,283 * 10-3
| |
|---|
Figure 16: Graphical display of the calculated mutation frequencies [bp-1] of the “selfmade” and the kit epPCR amplification. Shown are the estradiol nanobody control sample, amplified by OneTaq DNA polymerase (E2_O, in grey); the estradiol sample generated by epPCR amplification using the protocol of McCullum et al. from 2010 (E2_S, in dark red) and the two samples of estradiol (E2_K, in light red) and progesterone (P_K, in green) amplified using the epPCR kit bought from JenaBioScience.
The mutation rate for the JenaBioScience epPCR kit is expected to be between 0,6 to 2% [16]. In our experiments the kit yielded mutation rates per base pair of 1,101 * 10-3 for the estradiol nanobody and 1,283 * 10-3 for the progesterone nanobody, resulting in an average mutation rate per base pair of 1,192 * 10-3 per run of PCR. Although the “selfmade” epPCR protocol yielded a higher mutation rate per base pair we nevertheless chose the kit to generate our library from the nanobody fragments. This is because we could not achieve amplification of the progesterone nanobody by this protocol. That is why we chose the kit in the first place.
Cloning
Figure 17: Overview of SfiI restriction sites. The gene of interest will be inserted between the pelB signal peptide (light pink coloured) and the linker (dark purple coloured), connecting it to the M13 geneIII protein. As insert is the grafted estradiol nanobody (grey coloured) and the immobilisation tag (green coloured) shown. A) Sequence of the 5´ SfiI restriction site with indicated cutting pattern. B) Sequence of the 3´ SfiI restriction site with indicated cutting pattern.
Figure 18: Map of the phagemid vector pZMB0062 with the sequence of the nanobody progesterone inserted.
Figure 19: Colony PCR of ER2738 clones after one day of incubation at 37 °C. Colonies 2, 3, 4, 7, 8, 9, 11 and 13 are positive for the progesterone nanobody insert. Colonies 1, 5, 10, 14 and 15 carry the religated vector without an insert. The gel (1,2% agarose with 1x TAE) was run at 100 V for 20 minutes. A 100 bp ladder was used for size reference.
Figure 20: Colony PCR of ER2738 clones after one day of incubation at 37 °C. The colony tested was positive for the estradiol nanobody insert. The gel (1,2% agarose with 1x TAE) was run at 100 V for 20 minutes. A 100 bp ladder was used for size reference.
Phage ELISA
For the detailed phage ELISA protocol click here.
A MaxiSorp 96-well plate was coated with streptavidin, subsequently biotin-coupled hormone was added to immobilize and present estradiol or progesterone. Incubation with our nanobodies presenting phages was followed by addition of anti-pVIII antibody with coupled HRP (horseradish peroxidase) and further addition of the substrate ABTS. Finally, the absorbance was measured in a Tecan i-control infinite 200 at 419 nm.
Table 6: Raw data (top) and corrected data (bottom) of the phage ELISA. The estradiol nanobody was tested against biotin conjugated estradiol in wells B4 to B6 and C4 to C6, well B7 served as control. The progesterone nanobody was tested in wells D4 to D6 and E4 to E6 against biotin conjugated progesterone, well C7 served as a control. The detection reaction was performed with the HRP conjugated anti pVIII antibody and ABTS substrate. Absorbance was measured at 419 nm in the plate reader Tecan i-control infinite 200.
| Raw data | 4 | 5 | 6 | 7 |
|---|---|---|---|---|
| B | 0,0523 | 0,0533 | 0,0521 | 0,0521 |
| C | 0,0525 | 0,0522 | 0,0512 | 0,051 |
| D | 0,049 | 0,0515 | 0,0518 | |
| E | 0,0518 | 0,0524 | 0,0429 |
| Corrected data | 4 | 5 | 6 | 7 |
|---|---|---|---|---|
| B | 0,0002 | 0,0012 | 0 | 0 |
| C | 0,0004 | 0,0001 | -0,0009 | 0 |
| D | -0,002 | 0,0005 | 0,0008 | |
| E | 0,0008 | 0,0014 | -0,0081 |
The data (see Table 6) show no significant absorption at 419 nm, neither for the estradiol nanobody nor for the progesterone nanobody. Thus, it can be concluded that, as assumed and predicted, the initial grafts of the two antibodies do not bind their respective antigens or bind them poorly and that their specificity needs to improve by appropriate affinity maturation measures.
Outlook
scFv Grafting
Figure 21: Visualization of the scFv Fragment for progesterone binding. The figure shows the overlap for Gibson Assembly at the 3’ and 5’ end of the fragment. Also the restrictionsites of SpeI and NdeI are shown.
Figure 22: Map of the pTXB1 vector with inserted scFv progesterone fragment.
The vector was used to overexpress our scFvs after successful cloning. Subsequently, the proteins can be isolated with the IMPACT™ Protein Purification System [21].
scFv Cloning Strategies
Cloning via Gibson Assembly®
1.1 After linearization of the backbone, a gel electrophoresis was performed in which the complete PCR product was inserted. The gel was then stained in ethidium bromide. The resulting fragments were visible on the transiluminator afterwards. With a scalpel the band was cut out of the gel at the ~ 6700 bp mark. The DNA was then recovered by using the Gel Clean up Kit from Machery-Nagel. The isolated DNA fragment and the scFv were then used for the Gibson Assemly®.
1.2 A second approach to preserve the opened backbone was the digestion of the PCR product by the restriction enzyme DpnI. Digestion with this restriction enzyme allows methylated template to be removed from the PCR, leaving only synthesized linearized plasmid DNA. After DpnI digestion, a PCR clean up was performed with the Machery-Nagel PCR and DNA clean up kit to remove debrist. The Gibson Assemly® followed.
Cloning via Seamless Cloning
Cloning via Restriction Enzymes
Outlook
Referenzes
[2] „Molekular- und Zellbiologische Grundlagen | SpringerLink“. https://link.springer.com/book/10.1007/978-3-662-08822-7 (zugegriffen Okt. 28, 2020).
[3] H. J. Wagner, S. Wehrle, E. Weiss, M. Cavallari, und W. Weber, „A Two-Step Approach for the Design and Generation of Nanobodies“, Int. J. Mol. Sci., Bd. 19, Nr. 11, Nov. 2018, doi: 10.3390/ijms19113444. [4] „RCSB PDB - 1JNH: Crystal Structure of Fab-Estradiol Complexes“. https://www.rcsb.org/structure/1JNH (zugegriffen Okt. 28, 2020).
[5] „Highly specific anti-estradiol antibodies: structural characterisation and binding diversity - PubMed“. https://pubmed.ncbi.nlm.nih.gov/11812141/ (zugegriffen Okt. 28, 2020).
[6] „monoclonal anti-progesterone P15G12C12G11 immunoglobulin VHCH1 heavy c - Protein - NCBI“. https://www.ncbi.nlm.nih.gov/protein/AAX47169.1 (zugegriffen Okt. 28, 2020).
[7] O. Dubreuil u. a., „Fine Tuning of the Specificity of an Anti-progesterone Antibody by First and Second Sphere Residue Engineering“, J. Biol. Chem., Bd. 280, Nr. 26, S. 24880–24887, Juli 2005, doi: 10.1074/jbc.M500048200.
[8] R. P. D. Bank, „RCSB PDB - 3DWT: Structure of CabBCII-10 nanobody“. https://www.rcsb.org/structure/3DWT (zugegriffen Okt. 28, 2020).
[9] M. Dondelinger u. a., „Understanding the Significance and Implications of Antibody Numbering and Antigen-Binding Surface/Residue Definition“, Front. Immunol., Bd. 9, 2018, doi: 10.3389/fimmu.2018.02278.
[10] A. Honegger und A. Plückthun, „Yet Another Numbering Scheme for Immunoglobulin Variable Domains: An Automatic Modeling and Analysis Tool“, J. Mol. Biol., Bd. 309, Nr. 3, S. 657–670, Juni 2001, doi: 10.1006/jmbi.2001.4662.
[11] R. C. Cadwell und G. F. Joyce, „Randomization of genes by PCR mutagenesis“, PCR Methods Appl., Bd. 2, Nr. 1, S. 28–33, Aug. 1992, doi: 10.1101/gr.2.1.28.
[12] E. O. McCullum, B. A. R. Williams, J. Zhang, und J. C. Chaput, „Random mutagenesis by error-prone PCR“, Methods Mol. Biol. Clifton NJ, Bd. 634, S. 103–109, 2010, doi: 10.1007/978-1-60761-652-8_7.
[13] N. E. Biolabs, „OneTaq® DNA Polymerase | NEB“. https://international.neb.com/products/m0480-onetaq-dna-polymerase (zugegriffen Okt. 28, 2020).
[14] N. E. Biolabs, „Hot Start Taq DNA Polymerase | NEB“. https://international.neb.com/products/m0495-hot-start-taq-dna-polymerase (zugegriffen Okt. 28, 2020).
[15] N. E. Biolabs, „Q5® High-Fidelity DNA Polymerase | NEB“. https://international.neb.com/products/m0491-q5-high-fidelity-dna-polymerase (zugegriffen Okt. 28, 2020).
[16] J. B. GmbH, „JBS Error-Prone Kit, Molecular Evolution: Random Mutagenesis“, Jena Bioscience. https://www.jenabioscience.com/molecular-biology/cloning-and-mutagenesis/random-mutagenesis-kits/pp-102-jbs-error-prone-kit (zugegriffen Okt. 28, 2020).
[17] „Quality control NGS libraries, Bioanalyzer High Sensitivity DNA | Agilent“. https://www.agilent.com/en/product/automated-electrophoresis/bioanalyzer-systems/bioanalyzer-dna-kits-reagents/bioanalyzer-high-sensitivity-dna-analysis-228249#productdetails (zugegriffen Okt. 28, 2020).
[18] „igvtools | Integrative Genomics Viewer“. https://software.broadinstitute.org/software/igv/igvtools (zugegriffen Okt. 28, 2020).
[19] „Understanding Illumina Quality Scores“, S. 2.
[20] N. E. Biolabs, „SfiI | NEB“. https://international.neb.com/products/r0123-sfii (zugegriffen Okt. 28, 2020).
[21] N. E. Biolabs, „Impact | NEB“. https://international.neb.com/products/protein-expression-and-purification-technologies/e-coli/impact/impact (zugegriffen Okt. 28, 2020).
[22] „Engineered Recombinant Single-Chain Fragment Variable Antibody for Immunosensors | Analytical Chemistry“. https://pubs.acs.org/doi/pdf/10.1021/ac0507690 (zugegriffen Okt. 28, 2020).
[23] R. P. D. Bank, „1D PFV: 1JNH“. https://www.rcsb.org/sequence/1JNH (zugegriffen Okt. 28, 2020).
[24] R. P. D. Bank, „1D PFV: 1DBB“. https://www.rcsb.org/sequence/1DBB (zugegriffen Okt. 28, 2020).
[25] „geneart-seamless-cloning-assembly-mutagenesis-tools-brochure.pdf“. Zugegriffen: Okt. 28, 2020. [Online]. Verfügbar unter: https://tools.thermofisher.com/content/sfs/brochures/geneart-seamless-cloning-assembly-mutagenesis-tools-brochure.pdf.
[26] Z. Shen, G. A. Stryker, R. L. Mernaugh, L. Yu, H. Yan, und X. Zeng, „Single-Chain Fragment Variable Antibody Piezoimmunosensors“, Anal. Chem., Bd. 77, Nr. 3, S. 797–805, Feb. 2005, doi: 10.1021/ac048655w.
[27] A. M. Azevedo, V. C. Martins, D. M. Prazeres, V. Vojinović, J. M. Cabral, und L. P. Fonseca, „Horseradish peroxidase: a valuable tool in biotechnology“, Biotechnol. Annu. Rev., Bd. 9, S. 199–247, 2003, doi: 10.1016/s1387-2656(03)09003-3.