Introduction

Natural heavy-chain antibodies are first found in camels (Hamers-Casterman et al. 1993) and subsequently also found in llama (Van der Linden et al. 2000), alpaca (Maass et al. 2007), and sharks (Greenberg et al. 1995; Nuttall et al. 2001). This type of antibodies lack light chains and are another form of antibodies existing in these animals, in addition to the traditional antibodies which comprise both heavy and light chains. The various domains of heavy chains of the heavy-chain antibodies (VHHs) are able to bind antigens with high affinity (Muyldermans 2013). The molecular weight of VHHs is 15 kDa in average, and their structure size is about 4 × 2.5 × 3 nm (Desmyter et al. 1996). Therefore, the VHHs are also known as nanobodies.

VHHs have acquired important adaptations to be soluble and functional in the absence of the light chain and possess superior characteristics over various domains of traditional antibodies, which consist of various domains of both heavy chains (VHs) and light chains. These characteristics include 10-fold smaller molecular weight, strict monomeric behavior, higher resistance to proteolysis and thermal denaturation, higher solubility, and simpler structure and are easier to be folded and easier to produce (Muyldermans 2013). Some of VHHs’ advantages are ascribed to their different features from VHs. First, although both VHHs and VHs consist of four framework regions (FRs) and three complementarity-determining regions (CDRs), four hydrophilic amino acids (V37, G44, L45, and W47) in the second FR of VHH substitute the corresponding hydrophobic amino acids (F37, E44, R45, and G47) in the VH, which improves the water solubility of VHHs (Muyldermans 2001). Second, an intermolecular disulfide bridge usually forms between CDR1 and CDR3 in the VHH, which enhances the stability of VHHs (Muyldermans 2001). Third, the CDR3 of VHH is longer than that of VH, which therefore forms larger antigenic determinant and enhances affinity to antigen (Muyldermans 2001; Muyldermans 2013; Riechmann and Muyldermans 1999). These intrinsic biophysical properties of VHHs facilitate their expression in microorganisms such as bacteria and yeast (Rahbarizadeh et al. 2005; Rahbarizadeh et al. 2006; Van der Vaart 2002).

Antigen-specific nanobodies can be selected from constructed VHH libraries by various technologies, such as phage display, cell surface display, and high-throughput DNA sequencing and mass spectrometric identification (Liu et al. 2018). Phage display is the most developed and widely applied. The VHH libraries can be classified into 3 main types: immune library, naïve library, and semisynthetic/synthetic library. The immune library is the most widely used strategy for VHH screening, which provides highly abundant and generates high affinity target-specific binders. However, nanobodies against very conserved or toxic antigens should be selected from naïve library or semisynthetic/synthetic library. Nanobodies specifically against epidermal growth factor receptor, vertebrate nuclear pore complex, and ephrin receptor A4 have been successfully selected from VHH libraries and produced in microorganisms (Noor et al. 2018; Pleiner et al. 2015; Schoonaert et al. 2017). They were then used for immunoprecipitation, super-resolution imaging, and diagnostic and therapeutic purposes (Noor et al. 2018; Pleiner et al. 2015; Schoonaert et al. 2017).

Green fluorescence protein (GFP) is a widely used tag protein. It can be used to observe protein localization in live and in vivo by fusing to the N-terminus or C-terminus of target proteins. It can also be used as a tag to conduct immunoprecipitation. Since natural antibody production requires animal immunization, the production is costly and time-consuming. Moreover, the quality of the traditional antibodies depends on the immune response and the quantity is limited, and different batches of commercial rabbit or mouse antibodies against GFP may not ensure stable performance. Anti-GFP nanobodies were developed by some research groups by phage display or yeast surface display to study modulation of protein properties using nanobodies or super-resolution imaging (Kirchhofer et al. 2010; Platonova et al. 2015; Ryckaert et al. 2010; Szymborska et al. 2013). The structure of GFP:GFP nanobody complex has also been determined, and it is found that CDR1, CDR2, and CDR3 are the main regions responsible for contacting the surface of GFP (Kubala et al. 2010), and different nanobodies may recognize different epitopes on the GFP surface (Kirchhofer et al. 2010). Since nanobodies can be produced in microorganisms limitlessly, we are also interested to screen nanobodies against GFP for the use in basic research.

Thus, we created an immune library by immunizing an alpaca with purified GFP. Our screening strategy was developed by combining the processes reported in two previous studies (Nizak et al. 2005; Pardon et al. 2014) with some modifications, such as different phagemids and host strains. Anti-GFP nanobodies were panned against the library by phage display technology and screened by enzyme-linked immunosorbent assay (ELISA). Four specific nanobodies against GFP were obtained and successfully expressed in two prokaryotic expression systems developed in this study. Their binding abilities with GFP were confirmed both in vitro and in vivo.

Materials and methods

Lymphocyte isolation, RNA extraction, and cDNA synthesis

GFP was purified from bacteria and diluted to 2 mg/ml. One milligram of GFP was then mixed with equal volume adjuvant (GERBU). An alpaca was immunized with 1 ml GFP-adjuvant mixture by 5 subcutaneous injections at fortnightly intervals (done by the company of Shenzhen Kangti Life Technology in Shenzhen, China). Three days after the last immunization, 50 ml peripheral blood was collected with anticoagulant tubes from the alpaca and lymphocytes were isolated by Ficoll-Hypaque density gradient centrifugation according to the manufacturer’s instructions. The total RNA was extracted by TRizol Reagent (Invitrogen) and cDNA was synthesized using PrimeScript™ 1st Strand cDNA Synthesis Kit (TAKARA).

VHH library construction

Two rounds of PCR were applied to amplify VHH encoding DNA region (vhh) using the cDNA as template as described previously (Pardon et al. 2014). The first-round PCR was conducted with primers VH-F (5′-GTCCTGGCTGCTCTTCTACAAGG-3′) and VH-R (5′-GGTACGTGCTGTTGAACTGTTCC-3′), and 8 parallel PCR were conducted. The 8 PCR products with a length of about 800 bp were purified individually using the Universal DNA Purification Kit (TIANGEN, China). The second-round PCR was conducted using the 800-bp DNA fragment as the template individually with primers VHH-F (5′-CTCGCGGCCCAGCCGGCCATGGCAGATGTGCAGCTGCAGGAGTCTGGRGGAGG-3′) and VHH-R (5′-GTGTTGGCC TCCCGGGCCACTAGTGCGGCCGCTGGAGACGGTGACCTGGGT-3′). Therefore, 8 parallel PCR were conducted. PCR products with a length of approximately 400 bp were purified and mixed into one tube, and the DNA concentration was determined. Four micrograms of the 400-bp DNA fragments and 10 μg of phage display vector pADL-10b (purchased from Antibody design labs) were then digested individually with Bgl I (NEB) and ligated with T4 DNA ligase (Thermo Scientific). The ligation product was purified and its concentration was determined. Subsequently, 100-ng ligation products were transduced to SS320 competent cells (100 μl) by electrotransformation and 10 parallel transformations were conducted. After recovering for 1 h at 37 °C, the transduced cells were spread onto 6 LB agar plates (245 mm) containing ampicillin and tetracycline and were cultured overnight at 37 °C. Twenty clones were selected from the plates and insertion of vhh fragments in the pADL-10b was confirmed by PCR using primer pairs 10b-F(5′-CAGGAAACAGCTATGACCATGAT-3′) and 10b-R(5′-GCCCTCATAGTTAGCGTAACGAT-3′). All other clones were scraped off with LB medium, which constitute the vhh library. The recombinant plasmids contained in SS320 were named pADL-10b-vhh.

Panning for GFP binders

To pan for GFP binders by phage display, a phage display library was first obtained from the vhh library and titer of the phage display library was determined by preparing serial tenfold dilutions according to the methods described previously (Pardon et al. 2014). We panned for GFP-specific binders according to a method previously described with streptavidin-conjugated Dynabeads (Nizak et al. 2005). Before the panning process, purified GFP was coupled with biotin using the EZ-LinkTM Sulfo-NHS-LC-Biotinylation Kit (Thermo Scientific) according to the instructions. One microgram of GFP-biotin was incubated with 30 μl Dynabeads M-280 Streptavidin (Invitrogen) for 30 min at room temperature (RT). The GFP-streptavidin-Dynabeads were then washed three times with PBST (containing 0.05% tween-20). Meanwhile, the phage display library containing 5 × 1012 phages in 500 μl incubation buffer (PBST with 0.05% BSA) was pretreated by incubating with 30 μl Dynabeads M-280 Streptavidin for 30 min at RT. Then the pretreated phage display library was added to the GFP-streptavidin-Dynabeads, which were incubated with rotation at RT for 2 h. After that, the beads were washed 25 times with PBST to remove free and weak binding phages. Finally, 500 μl trypsin (0.25 mg/ml) was added to the beads and incubated at RT for 30 min to disassociate the binding phages. The eluted phages were neutralized with 10 μl protease inhibitor cocktail (Roche). Three hundred microliters of phages was used to infect 3 ml SS320 in exponential phase at 37 °C for 30 min, then the mixture was added with 7 ml LB medium and cultured at 30 °C overnight to obtain the first sub-vhh library. Then a sub-phage display library was obtained from the sub-vhh library. This panning process was repeated for three times to enrich phages displaying the GFP binders, and the input phage number in the second and the third round was reduced to 1 × 1012.

Selection of GFP binders by ELISA

Subsequently, ELISA was applied to select the positive clones expressing GFP binders. First, 17 clones from the third sub-vhh library were randomly selected and cultured in LB medium at 37 °C. When they grew to exponential phase, IPTG (at a final concentration of 0.2 mM) was added to induce VHH expression and bacteria were then cultured at 30 °C overnight. Next day, the bacteria were collected, suspended with PBS, and lysed with CelLytic™ B Cell Lysis Reagent (Sigma). The supernatants constituting the VHH extract were collected. A 96-well microtiter plate was coated with GFP and blocked with 3% BSA. The plate was then incubated sequentially with the extract from each clone for 1 h, anti-pIII antibodies for 1 h, and alkaline phosphatase–conjugated goat anti-mouse antibodies for 1 h at RT. Between incubation, the plate was washed with PBST five times (1 min/time) to remove unbound VHHs or antibodies. Finally, DNPP (200 μl/well) (Sigma) was added to each well and incubated for no more than 30 min in the dark. The reaction was terminated with 3 N NaOH (50 μl/well) and the absorbance at 405 nm was determined with a microplate spectrophotometer. As negative controls, VHH extracts or anti-pIII antibodies were not added. Wells with the value at 405 nm twice higher than those in negative control wells were defined as positive wells, and the corresponding clones are defined as positive clones.

Expression, purification, and modification of anti-GFP VHHs

Four sequences encoding the potential anti-GFP VHHs were identified and named a12, e6, d5, and b9. These sequences have been deposited into NCBI’s GenBank database and their accession numbers are MN813056, MN813057, MN813058, and MN813059, respectively. To purify these VHHs in prokaryote, two expression vectors, named pADL-10b-6×His and pBAD24-Flag-6×His, were constructed, both of which contain pelB sequences to secret VHHs to the periphery. Specifically, DNA sequence encoding pIII in pADL-10b was replaced with 6×His-encoded sequence by digesting pADL-10b with Hind III and Xho I to generate pADL-10b-6×His. pBAD24-Flag-6×His was constructed by inserting Flag- and 6×His-encoded sequences into pBAD24. DNA sequences encoding E6, D5, and B9 were PCR-amplified individually from corresponding phagemids using prime pairs nb-F (CCGCTCGAGATGGCAGATGTGCAGCTGCAGGAG)/nb-R (ATAAGAATGCGGCCGCGCTGGAGACGGTGACCTGGGTCCC). a12 gene was cut from pADL-10b-a12 with Bgl I and ligated to pADL-10b-6×His to generate pADL-10b-A12-6×His, which was transformed into competent SS320 for expression. e6, d5, and b9 genes were digested with Xho I/Not I and ligated to pBAD24-Flag-6×His individually to generate pBAD24-Flag-D5-6×His, pBAD24-Flag-E6-6×His, and pBAD24-Flag-B9-6×His, which were transformed individually into competent TOP10 for expression.

All VHHs were purified by affinity chromatography on nickel column. Specifically, A12 was induced with IPTG (at a final concentration of 0.4 mM) when pADL-10b-A12-6×His-contained SS320 reached exponential phase at 37 °C, and SS320 were then cultured at 30 °C overnight. E6, D5, and B9 were induced with l-arabinose (at a final concentration of 0.02% (w/v)) when pBAD24-Flag-E6-6×His-, pBAD24-Flag-D5-6×His-, or pBAD24-Flag-B9-6×His-contained TOP10 reached exponential phase at 37 °C, and TOP10 were then cultured at 30 °C overnight. Crude extracts of VHHs were obtained by osmometry and then purified by affinity chromatography on nickel column. Purified VHHs were coupled with biotin using the EZ-LinkTM Sulfo-NHS-LC-Biotinylation Kit and VHH-biotin was purified.

Binding ability of anti-GFP VHHs with GFP in vitro by ELISA

ELISA was first performed as described above to determine interaction between GFP and purified anti-GFP VHHs. A 96-well microtiter plate was coated with 1 μg purified GFP and blocked with 3% BSA. This plate was then incubated sequentially with 1 μg purified biotin-labeled VHHs (A12, E6, D5 or B9), HRP-conjugated streptavidin (1:10,000), and finally TMB (200 μl/well) (Thermo Scientific). TMB was incubated for no more than 30 min in the dark. The reaction was terminated with H2SO4 and the absorbance at 450 nm was determined. As negative controls, no GFP, no VHHs, or VHH without biotin was added with otherwise the same. Each VHH was conducted in triplicate and the results represent three independent assays.

Binding ability of anti-GFP VHHs with GFP in vitro by native-PAGE analysis

Interaction between GFP and purified anti-GFP VHHs was further determined by native-PAGE analysis. Native PAGE was developed the same as SDS-PAGE except that SDS was not included. One microgram of purified GFP was incubated with 2 μg A12, E6, D5, or B9 at RT for 30 min. Purified major basic protein (MBP) was incubated with each purified VHH as negative controls. The mixture was then subjected to native PAGE. The gels were observed under a blue light or stained with Coomassie Blue.

VHH binding study by immunoprecipitation

Immunoprecipitation was performed to determine the binding ability of anti-GFP VHHs with GFP in vivo. 293T cells are a mammalian cell line preserved in our lab and used in this study, which is isolated from human embryonic kidney and transformed with large T antigen. 293T cells (3 × 106) were seeded in each of four 10-cm dishes, were transfected with pEGFP-C2 (10 μg for each dish) the next day using lipofectamine 3000 (Thermo Scientific), and scraped at 48 h post-transfection after washing with PBS twice. Cells from each dish were lysed with 1 ml lysis buffer (containing 150 mM NaCl, 10 mM Tris HCl pH 7.5, 5 mM EDTA, 0.5% NP-40, 0.5 mM PMSF) on ice for 1 h. After centrifugation at 12000 rpm for 5 min at 4 °C, the supernatant was transferred and divided equally into 2 parts. Four parts of the supernatants were individually incubated with 30 μl Ni-NTA resin and 5 μg A12; 30 μl Flag antibody–conjugated beads and 5 μg E6; 30 μl Flag antibody–conjugated beads and 5 μg D5; or 30 μl Flag antibody–conjugated beads and 5 μg B9 overnight at 4 °C. As negative controls, one of the other 4 parts of the supernatants was incubated with 30 μl Ni-NTA resin without A12, and the other three parts were individually incubated with 30 μl Flag antibody–conjugated agarose beads without VHHs overnight at 4 °C. Next day, the mixtures were centrifuged at 2000 rpm for 1 min at 4 °C and the supernatants were discarded. All beads were washed with lysis buffer for three times. After the last washing, beads were dissolved in loading buffer and heated at 100 °C for 10 min. The samples were then subjected to Western blot analysis.

VHH binding study by immunofluorescence

Immunofluorescence was also performed to determine the binding ability of anti-GFP VHHs with GFP in vivo. 293T cells (2 × 104/well) in 8 wells in a 24-well plate were transfected with 1 μg pEGFP-C2 using lipofectamine 3000 and subjected to immunofluorescence at 36 h post-transfection. Specifically, cells were fixed in 4% paraformaldehyde for 10 min, washed three times with PBS for 1 min/time, and permeabilized in 0.25% Triton X-100 in PBS for 10 min at RT. The cells were then blocked with 1% BSA in PBS for 30 min at RT and subsequently probed with purified biotin-labeled A12, E6, D5, or B9 (at a final concentration of 10 μg/ml), together with a mouse anti-biotin antibody overnight. Next day, cells were washed twice with PBS for 1 min/time, incubated with Alexa Fluor 594–conjugated goat anti-mouse antibody (1:500, Thermo Scientific) for 1 h at RT, and then stained with Hoechst 33342 (1:3000, Thermo Scientific). As negative controls, the anti-biotin antibodies were not added with otherwise the same. Finally, cells were observed with a Leica TCS SP5 confocal microscope.

Western blot analysis

Protein samples were resolved in SDS-12% PAGE and transferred to PVDF membrane (Millipore). Western blot analysis was performed as follows. The membrane was blocked with 5% non-fat milk (in TBST) and incubated with a homemade rabbit polyclonal antibody against GFP (1:1000) (in 5% milk) as the primary antibody and then with a horseradish peroxidase (HRP)–conjugated donkey anti-rabbit antibody (1: 3000, Thermo Scientific) (in TBST) as the secondary antibody. The signals were detected using the electrochemical luminescence system (Thermo Fisher Scientific) according to the manufacturer’s instructions.

Results

Construction of vhh library

PCR amplification of vh sequences resulted in two products with length of nearly 1300 bp and 800 bp (Fig. 1a). Amplification of vhh sequences using the 800-bp DNA fragment as a template resulted in a nearly 400-bp product, which was the expected size (Fig. 1b). Bacteria library containing vhh fragments was constructed as described in the Materials and methods. The capacity of the library was 6.7 × 107. Quality and diversity of the library was evaluated by PCR. As shown in Fig. 1c, all 20 random selected clones contained vhh fragments, suggesting a 100% insertion rate of vhh into pADL-10b. These vhh fragments were then sequenced and the results showed different CDR sequences in different clones, suggesting a high diversity of the library (Fig. 1d).

Fig. 1
figure 1

Construction of VHH library. a PCR amplification of vh sequence. b PCR amplification of vhh sequence. c Insertion rate of vhh into pADL-10b. 20 clones were randomly selected and vhh sequences were amplified by PCR. d Diversity of the VHH library. Amplified vhh sequences were sequenced and the FR sequences and CDR sequences of VHHs are showed

Full size image

Enrichment and identification of GFP binders

The numbers of input phages in the 1st- to 3rd-round panning were 5 × 1012, 1 × 1012, and 1 × 1012, while the numbers of output phages were 2 × 107, 6 × 108, and 5 × 1010, respectively. These results showed an increasing trend in the amount of output phages (Fig. 2a), indicating the successful enrichment of GFP binders. Subsequently, positive clones expressing GFP binders were selected by ELISA from the third sub-vhh library. The results showed that 14 out of 17 clones showed positive reactions compared with the negative controls. All 14 clones were sequenced and 4 sequences with different CDR3 were identified, named A12, E6, D5, and B9 (Fig. 2b). Among these sequences, A12 is the most abundant and contained by 9 clones (64.29%), followed by E6 contained by 3 clones (21.43%), and D5 and B9 are each contained by one colony (7.14%) (Fig. 2c).

Fig. 2
figure 2

Screening and selection of four anti-GFP VHHs. a The input and output phage number in the three-round panning processes. b Percentage of the four different anti-GFP VHHs in all positive clones. c Sequence alignment of the four anti-GFP VHHs. FR and CDR3 sequences of the VHHs are indicated. The numbers on the right indicate the number of amino acids

Full size image

Purification of anti-GFP VHHs in prokaryotes

Prokaryotic expression vectors containing A12, E6, D5, or B9 were generated as described in the Materials and methods. Schematic diagrams of expression vectors are shown in Fig. 3a. A12 was cloned to the pADL-10b-6×His and the other three were cloned to the pBAD24-Flag-6×His for expression (Fig. 3a). All VHHs were purified by affinity chromatography on nickel column, and purified VHHs were shown in Fig. 3b. The molecular weight values of purified A12, E6, D5, and B9 were 15.7, 18.4, 16.6, and 17.3 kDa, respectively (Fig. 3b).

Fig. 3
figure 3

Expression and purification of anti-GFP VHHs. a Schematic diagram of vectors expressing A12, E6, D5, and B9. A12 was cloned into modified pADL-10b for expression under the lac promoter, and E6, D5, and B9 were cloned individually into modified pBAD24 for expression under the araC promoter. b Purified A12, E6, D5, and B9 analyzed on SDS-PAGE. Purified VHHs were analyzed on 15% (w/v) SDS-PAGE and stained by Coomassie Blue. The table below shows the molecular weight of each VHHs

Full size image

Anti-GFP VHHs bind GFP in vitro

ELISA was performed to determine the binding ability between purified anti-GFP VHHs and purified GFP. As negative controls, no GFP, no VHH, or VHH without biotin label was added with otherwise the same. As shown in Fig. 4, the absorbance value at OD450 of A12, E6, D5, and B9 was all significantly higher than those of negative controls. Specifically, the value at OD450 of A12, E6, and D5 were more than 3-fold higher than those of negative controls, while the value at OD450 of B9 was more than 3-fold higher than those of negative controls without VHH or with no biotin-labeled VHH and more than 2-fold higher than that of the negative control without GFP (Fig. 4). These results indicate that A12, E6, D5, and B9 can bind GFP specifically.

Fig. 4
figure 4

Binding analysis of purified anti-GFP VHHs to purified GFP by ELISA. The assay was performed as described in materials and methods. A12, E6, D5, and B9 indicate wells added with VHHs. NC, negative control. NC1, negative control well with D5 (without biotin) added. NC2, negative control well without VHH. NC3, negative control well without GFP. Student’s t test was employed to determine the significant difference between samples. ***p < 0.001

Full size image

Subsequently, the interaction between anti-GFP VHHs and GFP was further confirmed using native PAGE. During incubation, adequate VHHs were added to bind all GFP. The principle of the native-PAGE analysis is that the migration of GFP is shifted when it binds VHHs. Moreover, GFP can emit green fluorescence in the excitation of blue light in native condition. As shown in Fig. 5a, b, in blue light, compared with GFP alone, migration of GFP was shifted after incubation with A12, E6, D5, or B9, indicating that A12, E6, D5, and B9 all can bind GFP. It was speculated that GFP might be degraded as two bands were observed in Fig. 5a. The same gels in Fig. 5a, b were then stained with Coomassie Blue and the results are shown in Fig. 5c, d, respectively. In Fig. 5c, GFP and A12 are indicated by circle and square, respectively, while GFP-A12 complex is indicated by triangle. A12 could not enter the separation gel and retained in the interface of spacer and separation gel. As a negative control, GFP was replaced with MBP and the same assay was conducted. Migration of MBP (indicated by circle) was not shifted after incubation with A12 (Fig. 5c), indicating that A12 does not bind MBP. Similarly, in Fig. 5d, compared with GFP alone (indicated by circle), migration of GFP-B9, GFP-D5, and GFP-E6 complexes was shifted and indicated by triangles. Moreover, B9 and D5 (indicated by squares) can be seen clearly in the gel. In negative controls, MBP (indicated by a circle) was observed while no MBP-VHH complexes were observed (Fig. 5d). Free B9 and D5 (indicated by squares) can be seen at the same positions with those in the left (Fig. 5d). Free E6 (indicated by square) was also clearly seen in a position closed to the GFP-E6 complex, which may explain that free E6 was not observed in the lane of GFP/E6. Taken together, these results demonstrate that purified A12, E6, D5, and B9 can bind GFP specifically.

Fig. 5
figure 5

Binding analysis of purified anti-GFP VHHs to purified GFP by native-PAGE analysis. a A12 was incubated with GFP and migration of GFP was observed under blue light. b E6, D5, or B9 was incubated with GFP and migration of GFP was observed under blue light. c The gel in a was stained with Coomassie Blue. Incubation of A12 with MBP was used as a negative control. d The gel in b was stained with Coomassie Blue. Incubation of VHHs with MBP was used as negative controls. Triangles indicate VHH-GFP complex. Circles indicate free GFP or free MBP. Squares indicate free VHHs

Full size image

Anti-GFP VHHs binds GFP in vivo

The binding ability of purified anti-GFP VHHs with GFP expressed in vivo was determined by immunoprecipitation. 293T cells transfected with pEGFP-C2 were subjected to immunoprecipitation. As shown in Fig. 6, GFP could be precipitated in the presence of A12 (Fig. 6a), E6 (Fig. 6b), D5 (Fig. 6c), or B9 (Fig. 6d), but could not be precipitated without VHHs. These results indicate that A12, E6, D5, and B6 can bind GFP in vivo.

Fig. 6
figure 6

Binding analysis of purified anti-GFP VHHs to GFP in vivo by immunoprecipitation. 293T cells were transfected with pEGFP-C2, collected at 48 h post-transfection, and subjected to immunoprecipitation (IP). Cell lysates were incubated with each VHHs and corresponding beads. Cell lysate samples and immunoprecipitation samples were subjected to Western blot analysis. GFP was detected with a rabbit polyclonal antibody against GFP, A12 was detected with an anti-His antibody, and the other three VHHs were detected with an anti-Flag antibody. Cell lysates incubated with corresponding beads alone were used as negative controls. Input indicates cell lysate

Full size image

We further confirmed the binding ability of purified anti-GFP VHHs with GFP in vivo by immunofluorescence. As shown in Fig. 7, GFP was localized in both the nucleus and the cytoplasm after cells were probed with A12, E6, D5, and B9, which was similar to the phenomena observed under self-luminescence of GFP. As negative controls, no fluorescence was observed as expected in the absence of VHHs. These results indicate that A12, E6, D5, and B6 can bind GFP in vivo specifically.

Fig. 7
figure 7

Binding analysis of purified anti-GFP VHHs to GFP in vivo by immunofluorescence. 293T cells were transfected with pEGFP-C2, collected at 36 h post-transfection, and subjected to immunofluorescence. The cells were probed with purified VHHs, incubated with anti-biotin antibodies and Alexa Fluor 594–conjugated goat anti-mouse antibodies and stained with Hoechst 33342. Cells that were not probed with VHHs were used as the negative controls. GFP localization was then observed with a confocal microscope

Full size image

Discussion

In this study, we generated a VHH library from an alpaca immunized with GFP and obtained four VHHs against GFP from this library. Further studies indicate that all four VHHs purified from bacteria are able to bind GFP both in vivo and in vitro specifically. These results demonstrate that the platform for VHH selection is feasible and the four anti-GFP VHHs can be purified from bacteria and are functional.

During selection process by ELISA, 14 out of 17 clones (82.35%) are positive to GFP, which suggests that the panning strategy using biotin-streptavidin Dynabeads system is highly efficient and three-round panning is enough for specific VHH enrichment. The efficiency is comparable with that of a Nature Protocol which uses the 96-well microtiter plate for panning and 129 out of 138 clones are positive (93.48%) (Pardon et al. 2014). Moreover, among the 129 colonies, 30 unique sequences are identified. But the strategy using biotin-streptavidin Dynabeads system is better for fully displaying antigen and avoids the denature of antigen. Among the 14 clones, 9 clones contain the A12 sequence, indicating that A12 is the most enriched. These results suggest that A12 may have the highest affinity to GFP among the four VHHs. Indeed, more fluorescent cells were observed in the fluorescence excitation of 594 nm when transfected 293T cells were incubated with A12 during immunofluorescence assay. Many previous studies always do a large-scale selection for the positive clones during ELISA, such as screening from 96 to 192 clones, which is laborious and time-consuming. In this study, our results suggest that 20 clones may be enough for positive VHH selection. Moreover, among all sequenced clones, VHH present in the highest frequency is supposed to have a high affinity to antigen. Since three-round panning is highly efficient for enrichment of antigen-specific VHHs, it may be also feasible by randomly sequencing at least 20 clones to identify the most abundant vhh sequence, or by analyzing the library with high-throughput sequencing. The most enriched VHH is supposed to bind the antigen efficiently.

The results from ELISA indicate that all four purified VHHs can bind purified GFP specifically (Fig. 4), but our results cannot deduce the affinity of VHHs with GFP. Many factors can influence the result of the ELISA. For example, since the signal was amplified using the biotin-streptavidin system, the number of biotins labeled to VHHs can affect the value of absorbance at OD450. Although D5 has the highest absorbance value at OD450, it does not suggest a highest binding ability of D5 with GFP. As the average number of biotin labeled to A12, E6, D5, and B9 was not assessed in our study, it is possible that D5 was labeled with more biotin than the other three VHHs.

Prokaryotic expression and purification analysis of VHHs showed that the production and purity of A12 is superior to other three VHHs (Fig. 3). These results suggest that pADL-10b-SS320 expression system is more efficient than pBAD24-TOP10 expression system. Cloning E6, D5, and B9 into the modified pBAD24 vector may affect VHH expression or the araC promoter in pBAD24 is less efficient than the lac promoter in pADL-10b. Therefore, expression of VHHs in the original system used in the panning and screening processes is a better choice. Since there is no peptide tag attached to expressed VHHs in the pADL-10b, a modified pADL-10b or other vectors with peptide tag sequences fused with vhhs may be used to construct library in future studies and would be more convenient. For a common application in basic research, the expression system of GFP-specific VHHs can be further optimized and purified VHHs can be chemically modified, and then their affinities to GFP can be assessed.

Although the four VHHs are able to bind with both purified native GFP in vitro and native GFP expressed in 293T cells, all VHHs against GFP are not applicable to Western blot analysis. This may be because the native GFP was used for panning and selection processes. Since VHHs are strongly specific, all positive VHHs recognize native GFP instead of denatured GFP. A VHH recognizing denatured GFP is supposed to be obtained by panning the library with denatured GFP.

Since there are some other GFP nanobodies reported by other research groups, we made amino acid alignment of GFP nanobodies from our study and two other studies (Kirchhofer et al. 2010; Kubala et al. 2010) (data not shown). The major differences are in the CDRs. It is indicated that CDRs are responsible for contacting the surface of GFP (Kubala et al. 2010) and different nanobodies recognize different epitopes on the GFP surface (Kirchhofer et al. 2010). Therefore, it is possible that the four GFP nanobodies in our study may recognize epitopes of GFP different from other GFP nanobodies. A recent study develops a functionalized GFP nanobody (termed “dongle”) for knock-sideways experiments and shows effective knock-sideways of GFP-tagged proteins in human cells (Kuey et al. 2019). However, to some GFP-tagged proteins, such as dynamin-2-GFP and TPD54-GFP, dongles caused inhibition of target protein function prior to knock-sideways (Kuey et al. 2019). They suggest that alternative gfp nanobodies may be used to decrease the perturbation and solve such issues. The four GFP nanobodies developed in our study contribute to the diversity of GFP nanobodies and may enhance the use of GFP in scientific research.

In conclusion, four novel anti-GFP nanobodies were successfully selected from an immune library and expressed in bacteria in this study. All four anti-GFP nanobodies bind GFP specifically and can be applied in basic research, such as immunoprecipitation of GFP-tagged proteins. Our study also demonstrates a feasible platform for nanobody selection and production. In the future, the nanobody selection and production platform developed in this study can be applied for other antigen-specific nanobody selection.