Monoclonal antibodies have become important tools in several fields, including molecular biology, pharmaceutical and medical research, as well as in the treatment of diseases such as cancer and infectious diseases 123. Since the advent of antibody technology, antibody production has moved from hybridoma technology to recombinant DNA methodology. The advantages of recombinant antibodies are several folds, (i) antibodies can be produced in bacteria, yeast or plant 456, (ii) immunization is not required and (iii) intrinsic properties such as immunogenicity, affinity, specificity and stability of antibodies can be improved by various mutagenesis technologies 789. In the past two decade, advances in phage display and antibody engineering have led to the development of phage-displayed antibody technology 1011. This technology allows one to isolate antibodies directly from diverse repertoires of antibody genes, generating high-affinity binding sites without the constraint imposed by classical method for generating either polyclonal or monoclonal antibody 1213141516. Since the method does not depend on an animal's immune system, antibodies to a wide variety of antigens, including the molecules that cannot stimulate immune system of the animals such as nonimmunogenic, "self", cell surface or toxic antigens, can be generated 161718. The antibodies can also be engineered to contain in-built features that suit various downstream applications 19 or converted into functional whole immunoglobulin 2021. The antibody genes are expressed and the gene products displayed on the surface of filamentous bacteriophage as fusion proteins 71122232425. This collection of phages is called a phage display antibody library, where each phage particle displays a single antibody. In order to construct a library, antibody genes are fused to phage genes, thus creating a link between antibody phenotype and its encoded genotype. Antibody genes can be isolated from B-lymphocytes of non-immunized donors, rendering a naïve library which is a valuable source of human monoclonal antibodies against various antigens 26. Various formats of antigen-binding fragments, including Fab and scFv have been cloned and displayed on phage 2728. The advantage of smaller antibody fragments is that they have high tissue penetrability, while maintaining their affinity and specificity 293031. They are also easier and faster to produce in recombinant form. However, successful construction of a human antibody phage library has been achieved only by a small number of research groups 102932. One reason may be because of the complexity and cost of generation of the library, even though there have been some reports describing optimized protocols for the generation of efficient libraries 3233.

Here we report a simple and highly efficient method for the construction of a compact and highly useful scFv human library. The library was based on the naïve human re-arranged V-genes and assembled through the use of a gene repertoire derived from 140 non-immunized donors. All possible combinations of heavy and light chains, among all immunoglobulin isotypes, were included by using a mixture of primers and overlapping extension PCR. The resulting variable gene repertoire were cloned to form a moderate size library composed of 1.5 × 10⁸ individual clones from one ligation reaction. This repertoire was used for selection of specific binders to different proteins, a hapten, and complex antigens i.e., viral coat proteins, crude snake venom and cancer cell surface. Binding specificity and sequence diversity among binders were demonstrated.

Phage display antibody technology has become increasing popular for creating binding sites for use in all areas of research, and in medical and industrial applications. There are several examples of successful isolation of antibodies against various antigens from different phage displayed antibody libraries 1117333543, including high throughput selection 44. Unfortunately, the libraries are not available commercially; any laboratories that are interested in using this technique are required to construct a library by themselves, or obtained existing libraries for research with restrictions. Of the two types of phage antibody libraries (immunized and non-immunized libraries), the nonimmune library is of more general use, because it can be applied to generate antibodies to any desired antigen 12283335. These "naive" libraries normally has been constructed from the light-chain and heavy-chain IgM-V-gene pools of B cells isolated from peripheral blood lymphocytes, bone marrow, or spleen cells of nonimmunized healthy donors 131744. However, it has been reported that the diversity of the library could also be maximized by using random hexamers to prime cDNA synthesis, so that all five antibody classes could potentially be represented 35. Since the whole antibodies cannot be functionally expressed in bacteria, only the antibody fragments that contain the binding regions are displayed on the surface of the bacteriophage 10. Most of the phage libraries that have been constructed display the antibody fragment on the surface of the phage minor coat proteins (pIII). It has been shown that both Fab 2745 and scFv 111323 can be expressed in the surface of M13 without apparent loss of the antibody's specificity and affinity. Since the scFv format has been shown to be more efficient for display on phage coat 33 and the aim of this research was to develop the most efficient method to construct a compact human antibody, the scFv format was chosen.

Several strategies have been described for producing antibody phage display library in scFv format but it is difficult to define the minimal repertoire size needed to retrieve good binders to antigens. Many approaches have been proposed to improve phage displayed human antibody repertoires, mainly by increasing the library sizes 1735, by sophisticated in vivo (Cre-lox) recombination method 17, and by improving cloning steps 3335. In this study, we describe a simple and highly efficient method to construct a compact human scFv phage library, by modification of several previously published protocols 153246. Our method needs the least number of oligonucleotide primers (33 primers). The lab of J.D. Marks used 66 primers to construct a naive library 13 while the lab of J. McCafferty used 89 primers 44. In these reports, two sets of primers were used to construct the library. The first set allow the amplification of the V_H and V_L gene repertoires from cDNA, and the second set is used to introduce restriction sites for sub-cloning the genes into phagemid. In our work, we optimized the protocol such that only one set of primers are used to both amplify the genes and introduction of restriction sites, thus reducing the number of primers by half. Moreover, in all of the previous reports on the construction of large (10^9–10) naïve human libraries, smaller libraries were first created and then combined together to obtain the larger libraries. For example, Vaughan's library of 1.4 × 10¹⁰ was generated from three sub-libraries 35, Sheet's library of 6.7 × 10⁹ was generated from two sub-libraries 33, and McCafferty's library of 10¹⁰ was generated from sixteen sub-libraries 44. In our work, because of the compact size, we used only a couple of ligation reactions and electroporations, compare to thirty six 33 to several hundred electroporations 35, and more than twelve ligation reactions 33 used by others to construct the libraries. Thus, one can use our method to create a compact phage display antibody library that is better than other libraries of the same size, and equally well to larger libraries, with less time and budget.

The scFv phage library was created by focusing on amplification of highly diverse re-arranged V genes, as it has been formerly suggested that the most useful scFv library are constructed from V-genes rearranged in vivo 33. In order to reduce amplification bias we performed 75 independent PCR reactions using all possible combinations within a primer set, modified from previous reports 14. The high diversity of this compact library came from the high diversity of re-arranged variable region genes, which were isolated from one hundred and forty non-immunized healthy donors. This is the highest number of re-arranged V-gene templates that has ever been reported. Other reports of large and complex human scFv libraries, for example, used five 33, forty two 44, forty three 35, or fifty 47 donors.

The cDNAs from peripheral blood lymphocytes of the 140 non-immunized donors were synthesized using a mix of random hexamers and oligo dTs, so that all five antibody classes could be represented. Amplification using these non-specific primers has been shown to be as efficient as using germline, V_H, IgG or IgM, specific primers for the construction of efficient naïve repertoires 1348. In order to create scFv fragment genes as a V_H-linker-V_L type, the 3' ends of V_H genes were made complementary to 5' ends of V_L genes through a (Gly₄Ser)₃ linker peptide, and the V_H and V_L genes were assembled and amplified by overlap extension followed by pull-through PCR. Moreover, to ensure the correct overlaps during assembly PCR, the three (Gly₄Ser)₃ repeat in the single chain linker region were encoded by different codons (Figure 2). A proof reading DNA polymerase was used in the overlap extension step to ensure the accurate joining of blunt ends of V_H and V_L genes segments and in-frame formation of scFv repertoire. For other steps, TAQ DNA polymerase was used due to a more efficient amplification using this polymerase. In addition, the mutations created by amplification errors could add more diversity to the library. The primer sets used for the amplification of V gene repertoires were designed based on information retrieved from V-BASE 34 and previous publications 1432. The 5' and 3' ends of each primer incorporated sequences which could be cleaved by restrictions enzymes, avoiding a two step amplification procedure as previously reported for a murine scFv library 32. It has been previously suggested that several hundred bps of DNA sequence 5' and 3' of the scFv cloning sites is required for efficient digestion of the insert 3233; however we found that this is not necessary, provided that the inserts were incubated with the restriction enzyme for at least 10 hour. In addition to the shortest list of primer used, the fewest steps were needed for the generation of the library. This is because only one ligation step and two electroporation were required, without the creation of sub-libraries as performed by other strategies 333544.

Utilization of different modified helper phages to improve the bio-panning efficiency has been reported 37495051525354. Here, the helper phage KM13 37, which is a protease sensitive helper phage, was used during the bio-panning. This modified helper phage encodes a modified gene III, encoding a peptide sequence which will be cleaved by trypsin, between domain D2 and D3. By applying trypsin cleavage following selection of the phage antibody library, the phage particles not carrying an antibody-pIII fusion would lose all there domain D1 and D2 of protein 3, and thus lose their ability to infect bacteria; whereas phage particles carrying an antibody-pIII fusion would retain their infectivity, due to the absence of the protease sensitive site in the gene III of the fusion. Thus, in the bio-panning procedure described here, elution was done by a combination of low pH treatment with glycine buffer (pH2.2) and trypsin treatment. Any phagemid encoded phage display library contain a large proportion of phage particles devoid of pIII fusion on their surface, however these can still be retrieved following selection due to non-specific binding of phage particles. Removal of phage particles carrying no scFv by trypsin digestion, greatly improve the efficiency of the selection. We found that we obtained better result when using KM13 helper phage than regular helper phage, M13K07 (data not shown). Utilization of KM13 helper phage also limits the number of eluted phages after the first round of section 55, which would be a logic consequence of decreasing the number of background binders of non-displaying phage particles and taken together this explain why only one round of panning is sufficient for most selections done in this study.

Our compact scFv antibody phage library of 1.5 × 10⁸ scFv repertoire rendered binders to seven different antigens. The number of different antibody fragments selected with each antigen was in the same range as from other naïve scFv phage display library, even though those libraries are 93 14 and 44 times 56 larger than ours. Since we aimed to amplify as many different variable regions as possible, sequence diversity was confirmed in both the primary library and the binders retrieved after antigen selection. The high diversity of our compact scFv library was confirmed by DNA sequence analysis of ten randomly picked clones from unselected library and seven selected clones. Sequence analysis of the primary library indicated that variable regions were derived from all six V_H and seven V_L gene families. The variable regions derived from V_H1, V_H3, V_H4, V_κ1, V_κ3, V_λ1, V_λ6 families had been commonly observed among antibody fragments from phage display libraries 3357, and human hybridomas 58. Besides variable regions derived from these preferences gene families, other less frequently used V gene families i.e., V_H2, V_H5, V_H6, V_κ2, V_κ4, V_λ2, V_λ3 and V_λ5 were also found in our library. These clones were derived from a number of germ line segments with varied number of amino acid substitutions, ranging from 0–24, indicating excellent diversity in our scFv phage library. We observed a preference usage of certain V gene segments in unselected library (V_H3DP47, V_λDPL16, and V_κ3DPK22) as well as in the selected clones (V_H3DP47, and V_κ1DPK9). This is in accordance with the bias of gene segments most often found in nature 173335 The length (5–19) and amino acid sequence of CDR3 regions varied considerably, even when the genes came from the same germ line segments. None of these clones shared the same CDR3 sequence. These are all characteristic of the creation of naturally occurring immunoglobulin repertoire 1735, which involve somatic recombination during the development of B cells in the central lymphoid organs, and somatic hypermutation that operates on B cell in peripheral lymphoid organs, resulting in an affinity maturation of the antibody population 59. Thus, a highly diverse and compact antibody library similar to nature could be successfully constructed from a large number of V gene repertoire templates. It has been estimated that the human antibody repertoire is at least 10¹¹ 59; however, the number of antibody specificities present at any one time is limited by the total number of B cells in an individual, as well as by each individual's encounters with antigens, which is approximately 10⁸ different specificities at any one time 59. This is the same diversity of the compact library described in this paper. Therefore, by performing one-round of bio-panning with this library, one can mimic natural selection and obtain a diverse pattern of binding antibodies.

Even if the size of the library is essential for successful isolation of high affinity antibodies, a very large library is difficult to maintain 10 and in many case, high affinity is not necessary the pre-requisite for successful application of antibody. This is because other properties such as specificity, expression level and stability are also important. The affinity of the antibody can be further enhanced by affinity maturation as previously reported 7606162636465. These methods involve introduction of diversity into the antibody genes by various methods of mutagenesis such as error-prone PCR, DNA or chain shuffling, or oligonucleotide-directed PCR, while affinity selection of the variants with decreasing amounts of antigens 16. The affinity maturation process is useful either for improving the affinity of antibody that is selected from naive phage library of medium size (10^7–8 different clones) that has lower binding affinity, or for generating "super" antibody to be used in certain applications, such as diagnostic or immunotherapy. It has been reported that phage antibodies with approximately tenfold higher affinities (10¹¹ M^-1) 636667 than the ceiling affinity that can be obtained from in vivo selection of B-cell (10¹⁰ M^-1) 65 have been made by in vitro affinity maturation.

The ability to isolate an antibody that can recognize free Aflatoxin from a simple bio-panning procedure underscores the high quality of our compact library. It has been well known that it is difficult to obtain antibody against free haptens 65, especially from the naïve library. Recombinant scFv against soluble alfatoxin that has been generated so far could only be obtained from either an optimized bio-panning method by elution with soluble haptens 39, or isolated from hybridoma 41. The ability to isolate antibodies that are specific to both conjugated and free Aflatoxin demonstrated the high quality of this compact library. It is possible that the population of donors have been previously exposed to Aflatoxin as it is a very common contaminant in the region.

In addition to hapten, we also demonstrated successful selections of specific antibodies against complex antigens, i.e., crude snake venom, cancer cell surface, and rabies virus, which are more difficult to obtain. This is because of the limited amount of target antigen present in the mixture, the background binding, and enrichment of phage antibodies specific for non-relevance antigens 10. However, one cannot rule out the possibility that the apparent specific binding of antibody to complex target is actually resulted from the different level of expression of the recognized antigen. Thus, characterization of these selected antibodies, as well as more investigation to determine the identities of their specific partners, are needed to be done.

In conclusion, a simple method for the construction of a compact and efficient human scFv phage library has been reported. The key to the success of this method is a very high diversity V-gene repertoire template that was obtained from 140 non-immunized donors in combination with optimized primer set and cloning technique. This method is very practical and should be able to be carried out in any academic institutions, small research facilities, and non-profit organizations world-wide, for the generation of both naïve and immunized human phage display scFv library.

PP is the PhD student at school of Biotechnology, Suranaree University of Technology, under the supervision of MY, and co-supervised by PK. NJ and KR are graduate students under the supervision of MY. PP constructed the library. PP, NJ and KR participated in affinity selections and characterizations of scFv to different targets. All authors read and approved the final manuscript.