Biochimica et Biophysica Acta 1814 (2011) 1140–1145

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta

j ourna l homepage: www.e lsev ie r.com/ locate /bbapap

Design and characteristics of a stable protein scaffold for specific binding based onvariable lymphocyte receptor sequences

Magdalena Wezner-Ptasińska a, Daniel Krowarsch b, Jacek Otlewski a,⁎a Department of Protein Engineering, Faculty of Biotechnology, University of Wroclaw, Tamka 2, 50-137 Wroclaw, Polandb Department of Protein Biotechnology, Faculty of Biotechnology, University of Wroclaw, Tamka 2, 50-137 Wroclaw, Poland

Abbreviations: VLR, variable lymphocyte receptorreceptor type B; dVLR, designed VLR; LRR, leucine-ricrepeat N-terminal capping region; LRR1, the first leucirich repeat variable; LRRVe, leucine-rich repeat variable econnecting peptide; LRRCT, leucine-rich repeat C-termindichroism; FL, fluorescence spectroscopy; ΔGden

H20, freeGdmCl, guanidinium chloride; TEV, tobacco etch virus⁎ Corresponding author at: Faculty of Biotechnology,

2, 50-137 Wroclaw, Poland. Tel.: +48 71 375 28 24; faxE-mail address: otlewski@protein.pl (J. Otlewski).

1570-9639/$ – see front matter © 2011 Elsevier B.V. Adoi:10.1016/j.bbapap.2011.05.009

a b s t r a c t

a r t i c l e i n f o

Article history:Received 30 December 2010Received in revised form 9 May 2011Accepted 11 May 2011Available online 18 May 2011

Keywords:Variable lymphocyte receptorLeucine-rich repeatConsensus sequence designProtein stability

Variable lymphocyte receptors (VLRs) serve as antigen binding proteins in jawless vertebrates. Theirrelatively low molecular weight makes VLRs an interesting alternative to antibodies in biotechnologicalapplications. A typical VLR comprises several unique motifs called leucine-rich repeats (LRRs). Usingconsensus approach we designed a novel VLR protein (called dVLR) containing six LRR repeats based on a sealamprey receptor sequence. The designed protein was expressed in Escherichia coli in a soluble, native formand showed very favorable biophysical properties. Recombinant dVLR is monomeric in solution and preservesits secondary structure within the pH range 3.0 to 11.0 and tertiary structure between pH 4.0 and 10.0. Itundergoes reversible thermal denaturation in a broad pH range (4.0 to 10.0). The maximal denaturationtemperature of 73.9 °C is observed at pH 6.0, 0.3 M NaCl. Chemical denaturation of dVLR at pH 7.5 is acooperative two-state process with a midpoint at 3.3 M GdmCl and a very high free energy change ofunfolding in the absence of denaturant equal to 14.1 kcal/mol. The biophysical properties of dVLR make ithighly suitable for biotechnological applications such as generation of specific ligand-binding molecules.

; VLRB, variable lymphocyteh repeat; LRRNT, leucine-richne-rich repeat; LRRV, leucine-nd; LRRCP, leucine-rich repeatal capping region; CD, circularenergy change of unfolding;

University of Wroclaw, Tamka: +48 71 375 26 08.

ll rights reserved.

© 2011 Elsevier B.V. All rights reserved.

1. Introduction

Jawless vertebrates (agnathans), such as hagfish and lampreys,have evolved into unique antigen binding proteins—variable lym-phocyte receptors (VLRs). VLRs are the oldest adaptive immunesystem proteins not structurally similar to immunoglobulin-basedantibodies [1–3]. VLRs comprise several unique structural units calledleucine-rich repeats (LRRs). The LRR motifs occur in proteins whichtake part in such process as immune response, signal transduction,cell adhesion, RNA splicing, synapse development and functioning[4,5]. The LRR motif has been found in intracellular, extracellular andtransmembrane proteins, which suggests that it may adapt to manydifferent locations [4].

The high diversity observed for the VLR sequences results fromboth somatic DNA rearrangements of the LRR cassettes and differentnumbers and combinations of the LRR motifs [2,3]. Thus, a vast set ofestimated 1014 unique VLRs is achieved that is equivalent to the

repertoire of jawed vertebrates' antibodies and is sufficiently diverseto recognize any foreign molecule [2,6,7]. Sea lamprey contain threetypes of VLRs, namely VLRA, VLRB, VLRC which are expressed clonallyon lymphocyte-like cells [8]. Only lymphocytes expressing VLRBreceptors after antigenic stimulation differentiate into VLR antibody-secreting cells [9].

A single LRRmotif consists of 24 amino acid residues and forms aβ-strand followed by diverse secondary structure units connectedby a loop. A functional VLR protein (Fig. 1) consists of the N-terminal motif (LRRNT), the first LRR fragment (LRR1), variablenumber of LRRVmodules (LRRVs), LRRVe fragment, LRRCP part, andthe C-terminal motif (LRRCT) that contains a highly variable insert[10–12]. LRRNT and LRRCT constitute capping regions that preventexposure of hydrophobic residues to a solvent [4,13,14]. A Thr/Pro-rich stalk connects the receptor to a GPI (glycosylphosphatidyl-inositol) anchor [11,12]. The entire VLR assembly forms ahorseshoe-shaped structure with a concave surface involved inantigen binding. This concave surface forms a β-sheet built ofparallelly repeated β-strands [11,12].

Only two crystal structures are available for VLR–antigen com-plexes with completely different types of ligand bound (hen egglysozyme and a trisaccharide) [11,12]. Both ligands are squeezedbetween the β-sheet surface and the highly variable insert of theLRRCT motif. The LRRNT motif is not involved in ligand binding,although it contains four highly variable positions which suggests thatthey may be involved in interactions with other ligands [11]. The totalinterface between an antigen and VLR may reach 2600 Å2. It has also

Fig. 1. dVLR protein model based on variable lymphocyte receptor structure (PDB ID:2R9U) showing domain organization of designed molecule.

1141M. Wezner-Ptasińska et al. / Biochimica et Biophysica Acta 1814 (2011) 1140–1145

been shown that the binding affinities of VLR-based receptors towardligands are comparable to those of antibodies [12]. This high ligandaffinity in combination with a relatively low molecular weight makeVLRs an interesting alternative to antibodies in biotechnology,diagnostics or nanobiotechnology [11].

In this paper we present the designing, cloning, expression,purification and biophysical characterization of a novel consensussequence based on a sea lamprey variable lymphocyte receptor, calleddesigned VLR (dVLR).

2. Materials and methods

2.1. Design of the dVLR consensus sequence and structure modeling

We used SMART [15] and UNIPROT [16,17] databases to collectVLRB sequences from the sea lamprey and to define the number ofindividual LRR modules. Sequence alignment was performed withClustalW [18]. Final sequences of dVLR was defined based on thefrequency of occurrence individual amino acid on each position. Theprotein model was built using VLR crystal structure as a template(PDB ID: 2R9U). Structural alignment was performed with theMUSTER server [19] and final modeling with I-TASSER [20–22].Model quality assessment was performed with the GenesilicoMetaMQAP server [23]. Solvent accessible area analysis was donewith Pymol [24,25].

2.2. Bacterial strains, plasmids and cultivation

The cDNA of consensus dVLR was optimized for Escherichia coliexpression and chemically synthesized by Geneart (Germany). At theN-terminal sequence of dVLR we added a cleavage sequence for TEV(tobacco etch virus) protease: ENLYFQGSS. Recombination sequences(attB) for the Gateway cloning system (Invitrogen, USA) were addedon both sides of the gene. Additionally, we added restriction sites forKpnI and BamHI on the 5′ and 3′ end of the gene respectively. ThecDNA of dVLR was inserted into the pMA, pDONR201 and pETG-60Avectors (Invitrogen, USA). Cloning was carried out in E. coli DH10αand for protein expression E. coli strain Rosetta (DE3) was used.Expression plasmid for tobacco etch virus protease (TEV) wasgenerously provided by J.A. Doudna [26].

2.3. Gateway cloning of dVLR

The pMA vector containing consensus dVLR cDNA was digestedwith KpnI and BamHI. The product purified with CleanUp (A&ABiotechnology, Poland) was used for insertion into the pDONOR201

vector with the BP reaction according to the attached protocol. DH10αcompetent cells were heat shock transformed with pDONR201 vectorcontaining dVLR cDNA and colonies were selected on LB-agar plateswith 50 μg/mL of kanamycin sulfate (Sigma, USA). Then, the gene wastransferred into a destination vector (pETG-60A) in the LR reaction.DH10α competent cells were transformed with the expressionplasmids and selected on LB-agar plates containing 100 μg/mL ofampicillin. The propriety of cloning was confirmed by restrictionanalysis and sequencing.

2.4. Optimization of dVLR expression in E. coli

E. coli Rosetta (DE3) strain was transformed with a pETG-60Avector and colonies were selected on LB-agar plates containingappropriate antibiotics (34 μg/mL chloramphenicol (Sigma, USA) and100 μg/mL ampicillin (Polfa, Poland)). Subsequently, transformantswere grown in 100 mL of LB medium (with antibiotics: 100 μg/mL ofampicillin and 34 μg/mL chloramphenicol) overnight at 37 °C at180 rpm, then 10 mL of the overnight culture was transferred to 1 L ofLB medium and grown as above to an OD600 of 0.8. Next, expressionwas induced with isopropyl β-D-1-thiogalactopyranoside (Roth,Germany) added to 0.5 mM and the bacteria were cultivated furtherat 25 °C. Protein expression level was checked after 4, 6, 16 and 24 h.Cells were harvested by centrifugation (4 °C, 3300×g, 8 min) andstored at −80 °C or immediately used for protein purification.

2.5. Purification of NusA-His6-dVLR fusion protein

Cell pellet from a 2 L culture was resuspended in 80 mL of ice-coldlysis buffer (50 mM Tris–HCl, pH 8.0, 300 mM NaCl, 0.5% Tween 20)containing Roche protease inhibitor cocktail, sonicated and centri-fuged at 12,000×g for 1 h at 4 °C. The supernatant was applied on anNi-NTA column (Qiagen) and rocked for 1 h at 4 °C. The resin waswashed with 2 L of washing buffer (50 mM Tris–HCl, pH 8.0, 200 mMNaCl). Bound proteins were eluted with 50 mL of 50 mM Tris–HCl, pH8.0, 150 mM NaCl, 300 mM imidazole. Protein samples were analyzedby 12% SDS–PAGE and fractions with the highest protein concentra-tions were pooled.

2.6. Cleavage of NusA-His6-dVLR fusion protein

The NusA-His6 tag was cleaved off from the protein with TEVprotease. The fusion protein was incubatedwith TEV protease at roomtemperature for 24 h. The protease (containing His6 tag) and NusA-His6 tag were removed on an Ni–NTA column. dVLR was eluted fromthe column with 40 mL of 50 mM Tris–HCl, pH 8.0, 150 mM NaCl andfractions were analyzed by 12% SDS–PAGE. dVLR-containing fractionswere pooled and concentrated eightfold with Centriprep YM-3(Millipore, USA).

The concentrated fractions were applied to a Superdex 75 prep-grade column (GE Healthcare) equilibrated with 50 mM Tris-HCl, pH8.0, 2 M NaCl and eluted at 1 mL/min at 4 °C. The homogeneity of thepurified protein was analyzed by 12% SDS-PAGE.

2.7. Circular dichroism measurements

CD spectra were acquired in a 1 mm cuvette at 195–265 nm and ina 1 cm cuvette at 250–350 nm on a Jasco J-715 spectropolarimeter at21 °C with the protein concentration of 1.9×10− 6 M and1.9×10−5 M, respectively. The protein was dissolved in the followingbuffers supplemented with 0.3 M NaCl: 5 mM glycine at pH between2.0 and 3.0, 5 mM citrate, pH 4.0, or 10 mM sodium phosphate at pHbetween 5.0 and 12.0. Spectra were averaged from three separatescans with a slit width set to 2 nm and a response time of 1 s.Estimation of secondary structure content was performed with K2D2software [27].

1142 M. Wezner-Ptasińska et al. / Biochimica et Biophysica Acta 1814 (2011) 1140–1145

CD-monitored thermal denaturation was carried out in 5 mMglycine, pH 3.0, 5 mM citrate, pH 4.0 and 10 mM sodium phosphate atpH between 5.0 and 12.0. All buffers contained 0.3 M NaCl to increasereversibility of thermal transitions. Thermal scans were performed ina 1 cm cuvette following ellipticity at 214 nm using a response time of16 s. An automatic Peltier accessory PFD 350S allowed continuousmonitoring of the thermal transition at a constant rate of 0.25 °C/min.The data were analyzed using PeakFit software (Jandel ScientificSoftware) assuming a two-state reversible equilibrium transition asdescribed previously [28].

2.8. Fluorescence spectroscopy measurements

Fluorescence spectra were measured in a 1 cm cuvette using an FP-750 spectrofluorimeter (Jasco) equipped with an ETC 272T Peltieraccessory. An excitation wavelength of 280 nm was used and emissionspectrawere collected from300 to 450 nm. Spectrawere acquired at theprotein concentration of 1.9×10−6 M in buffers described for CD-monitored denaturation.

Stability of dVLR was assessed by thermal denaturation using anFP-750 spectrofluorimeter (Jasco), measuring the fluorescence at365 nm on excitation at 280 nm with 5 nm emission and excitationbandwidths and response time of 10 s. Measurements were carriedout at the protein concentration of 1.9×10−6 M in a 1 cm cuvette.

An automatic Peltier accessory ETC 272T allowed continuousmonitoring of the thermal transition at a constant rate of 0.25 °C/min.Denaturation data were fitted assuming a two-state reversibleequilibrium transition, as described above.

2.9. Isothermal equilibrium denaturation studies

Chemical denaturation of dVLR with GdmCl was monitored with aJasco J-715 spectropolarimeter and FP-750 spectrofluorimeter (Jasco)at 21 °C. Protein samples (1.9×10−6 M) were incubated in variousconcentrations of GdmCl in 10 mM sodium phosphate, pH 7.5 for 24 hat 21 °C. The transitions were monitored by the decrease of the CDsignal at 214 nm with a 2 nm bandwidth or by the decrease ofintrinsic fluorescence at 365 nm on excitation at 280 nm with 5 nmemission and excitation bandwidths. The free energy change ofunfolding (ΔGden

H20) was determined by fitting fluorescence or CDintensity changes as a function of GdmCl concentration as describedpreviously [29]. Analysis of the data was performed by the PeakFitsoftware.

2.10. Ellman's assay

Samples of dVLR at the protein concentration of 3.8×10−6 Mwereprepared in 100 mM HEPES, pH 7.5 with or without 6 M GdmCl. Aftera 20-min incubation DTNB (5,5′-dithiobis(2-nitrobenzoic acid)) wasadded to the final concentration of 1 mM, and 10 min later theabsorbance of NTB (2-nitro-5-thiobenzoate) formed was read at412 nm (ε412=13 700 M−1 cm−1 in 6 M GdmCl, ε412=14 150 M−1

cm−1 in its absence).

Fig. 2. Amino acid sequence of dVLR protein. Positions that are more than 35%conserved in 222 aligned sequences of VLR proteins are indicated with normal letters,and variable positions (b35% conserved) are marked with grey background.

2.11. ANS binding

The binding of 250×10−6 M 1-anilinonaphtalene-8-sulfonic acid(ANS) to recombinant dVLR at the protein concentration of2×10−6 M was measured on a FP-750 spectrofluorimeter (Jasco) at21 °C in 10 mM sodium phosphate, 300 mM NaCl, pH 7.0. Theemission spectra on excitation at 375 nm were collected after 20-minof preincubation.

3. Results

3.1. Definition of consensus sequence

Before designing a consensus dVLR we had to decide on thenumber of LRR modules within the designed polypeptide. VLRs arebuilt of different types of LRRs, namely a single LRR1, up to nine LRRV,a single LRRVe and a single LRRCP. These repeats are flanked bycapping modules—LRRNT and LRRCT at the N- and C-terminus,respectively [11]. Since the average number of the LRRV repeats perreceptor is 1.31 [3] we decided to design a dVLR consensus containinga single LRRV module.

The amino acid sequence of dVLR was determined by alignmentwith ClustalW 222 unique sequences of VLR diversity region from thesea lamprey deposited at the SMART and UNIPROT databases. Thetotal length of the designed protein was 168 amino acids. We chosethe most frequently encountered amino acids at a given position(Fig. 2) except position 19, where we introduced serine instead ofarginine (the secondmost frequent amino acid), to avoid two residuesof the same charge next to each other.

3.2. Expression and purification of dVLR

dVLR was expressed as a NusA-His6 tag fusion in several growthconditions. The highest overexpression level was achieved after 16 hcultivation at 25 °C. All of the produced protein was soluble.Purification of NusA-His6-dVLR was performed in two chromato-graphic steps. First, the protein was purified by nickel affinitychromatography. The purified protein showed approximately 70%purity (based on SDS-PAGE). The fusion protein was then cleaved byTEV. Following cleavage, TEV protease was removed by reapplying thecleaved sample to the Ni–NTA column. The final purification step wasa size exclusion chromatography on a Superdex 75 prep-gradecolumn. dVLR was eluted in a single peak. SDS-PAGE analysis of thepurified protein confirmed its high purity (N97%). The homogeneity ofthe protein was verified by analytical size-exclusion chromatographyin the concentration range 0.1 to 300 μM. The analytical Superdex 75elution profile showed only a single symmetric peak, independentlyof the protein concentration, at an elution volume of 11.71 mLcorresponding to the dVLR monomer (Fig. 3). The final yield of thepurified protein was about 6 mg per 1 L culture.

3.3. Quantitation of free thiols

A conserved four-cysteine motif is present in both cappingmodules (LRRNT and LRRCT) and forms two sets of disulfide bridges[12]. The Ellman's assay for free thiol groups was used to assess thestatus of the potential disulfide bonds in dVLR. Ellman's assay resultsfor the native protein and that denatured in 6 M GdmCl showed that

Fig. 3. Size-exclusion chromatography of dVLR on analytical Superdex 75.

1143M. Wezner-Ptasińska et al. / Biochimica et Biophysica Acta 1814 (2011) 1140–1145

all cysteines in dVLR form disulfide bridges both in the native andGdmCl-denatured states.

Fig. 4. Circular dichroism and fluorescence spectra of dVLR. (A) CD spectra collected atpH 7.5 (−), pH 11.0 (− −), pH 3.0 (−∙−) and in 6 M GdmCl, pH 7.5 (∙∙∙).(B) fluorescence spectra collected at pH 7.5 (−), pH 10.0 (− −), pH 4.0 (−∙−) andin 6 M GdmCl, pH 7.5 (∙∙∙). Inset in panel A shows mean residue molar elipticity of dVLRin near-UV at pH 7.5. Inset in panel B shows ANS binding to dVLR at pH 7.5 (−), pH 10.0(− −), pH 4.0 (−∙−) and fluorescence of free ANS (∙∙∙).

3.4. Biophysical characterization of dVLR

Conformation of dVLR protein was analyzed by CD and fluores-cence spectroscopy. CD spectra recorded in the far-ultraviolet regionshowed a double minimum at 204 nm and 212 nm and a maximumaround 192 nm (Fig. 4A). The spectrum is typical for polypeptideswith a significant β-sheet content. The minimum at 204 nm mayindicate the occurrence of a 310-helix. The α-helix and β-sheetcontent was 13% and 28%, respectively, while the random coil wasestimated at the level of 59%, as calculated by K2D2 software [27].These values are close to observed in the crystal structures of VLRsdeposited in PDB, 13% helical and 17% β-sheet content in the VLR 2913Ectodomain (PDB ID: 2R9U) or 14% helical and 15% β-sheet content inthe VLR (PDB ID: 3G39). CD spectra were also collected in the near-UVrange 250–350 nm. The positive signal suggests that the protein has awell-defined tertiary structure and is not in a molten globule state(Fig. 4A, inset).

The fluorescence spectrum of the native protein had an emissionmaximum at 334 nm (Fig. 4B). In the presence of 6 M GdmCl theemission maximum was shifted to longer wavelengths (maximum at353 nm), reflecting denaturation of the protein.

We also determined the secondary and tertiary structure contentof dVLR under different pH conditions. CD and fluorescence spectra ofdVLR samples in the pH range between 2.0 and 12.0 were collected.The protein showed the highest secondary structure content withinthe pH range 3.0 to 11.0, as determined by CD spectra analysis(Fig. 4A). At a pH below 3.0 or above 11.0 the ellipticity at 214 nmdropped by more than 10% compared with the spectrum collected atpH 7.5.

The fluorescence spectra of dVLR had a prominent maximum at334 nm in the pH range 4.0 to 10.0 (Fig. 4B). At a pH below 4.0 orabove 10.0 the emission maximum was shifted toward longerwavelengths and was weaker.

Moreover, we used 8-anilino-1-naphtalene-sulfonate (ANS) tocheckwhether dVLR binds this dye. The fluorescence spectra of ANS inpresence of 2×10−6 M dVLR at a pH 4.0, 7.5 and 10 show only

marginal increase of dye fluorescence as compared to the spectra offree ANS in the same conditions [Fig. 4B, inset].

The thermal stability of dVLR was measured within a broad pHrange (pH 3.0 to 12.0). Within the pH range 4.0 to 10.0 we observed ahighly cooperative and reversible two-state denaturation (Fig. 5A). Asmall population of intermediates in the predenaturation region wasobserved at pH 3.0. Moreover, the denaturation showed only 40%reversibility at this pH. At pH 11.0 or 12.0 the transitions werereversible only in 30% and 10%, respectively. The highest stability wasobserved at pH 6.0 with denaturation temperature 73.9 °C anddropped to about 50 °C and 59 °C at pH 3.0 and 12.0, respectively(Table 1).

We also investigated chemical denaturation of dVLR byGdmCl usingCD and FL (Fig. 5B). Both denaturation curves were similar in shape andshowed highly cooperative transitions (m values 4.27 kcal/mol·M (CD)and 4.37 kcal/mol·M (FL)) with a midpoint at 3.3 M GdmCl (CD) and3.2 M GdmCl (FL) at pH 7.5. According to the two-state model, thecalculated free energy change of unfolding in the absence of denaturant(ΔGden

H20) was 14.1 kcal/mol (CD) and 14 kcal/mol (FL).

4. Discussion

Immunoglobulin-based antibodies recognize an enormous varietyof antigens. However, these very large glycosylated proteins aredifficult to produce in standard prokaryotic expression systems.Exploring novel, stable scaffolds suitable for specific target recogni-tion is one of the current objectives of protein engineering. Suchscaffolds should have fairly large binding surfaces and accept changesof amino acid residues at positions taking part in ligand recognitionwithout disturbing the protein structure. For practical purposes it is

Fig. 5.Denaturation transitions of dVLR. (A) Normalized thermal denaturation curves ofdVLR in 10 mM phosphate buffer, 0.3 M NaCl, pH 10.0. (B) Chemical denaturationcurves of dVLR in 10 mM phosphate buffer pH 7.5. Transitions were monitored bychanges of fluorescence (grey symbols) or ellipticity (black symbols).

1144 M. Wezner-Ptasińska et al. / Biochimica et Biophysica Acta 1814 (2011) 1140–1145

also critical that the proteins are well expressed, stable over a wide pHrange, can be stored for a long period of time and remain soluble andmonomeric even at high concentration.

A number of proteins composed of several repeats are known thatrecognize ligands in a manner alternative to the antigen recognitionby immunoglobulins. Successful design of consensus sequences forTPR [30], ankyrin [31] and LRR [14] repeats has been reported. Thestarting point of the design was sequence alignment allowingidentification of conserved residues crucial for protein stability.

Table 1Thermodynamic parameters of thermal denaturation of dVLR.

pH Circular dichroism Fluorescence

Tden [°C] ΔHden [kcal/mol] Tden [°C] ΔHden [kcal/mol]

3.0a 50.2 103.0 49.8 126.04.0 60.1 110.0 60.1 121.95.0 69.5 111.0 69.6 121.96.0 73.9 112.5 73.2 124.07.5 70.6 105.9 71.6 96.09.0 71.7 115.5 73.2 108.010.0 73.5 99.0 72.4 110.011.0a 63.4 100.9 64.5 92.712.0a 59.3 72.5 57.6 67.0

Data shown are mean values of two independent denaturation experiments. Standarderrors in Tden and ΔH were estimated to be±0.1 °C and ±7.0 kcal/mol, respectively.

a Transition poorly reversible.

Conversely, positions with the highest variability were consideredto be responsible for antigen recognition. Typically, a tandem ofrepeats was used and capped by designed N- and C-terminal regionsto increase the protein's stability and solubility [13,14,30–32]. Weapplied a similar strategy to design a stable dVLR protein.

One group within repeat proteins is built of LRR motifs. This groupincludes variable lymphocyte receptors of the adaptive immunesystem in jawless vertebrates [1–3,8,9,11,12,33]. These proteinseffectively recognize different antigens and thus could serve as aconvenient alternative to immunoglobulins. Selection of proteinsspecifically recognizing a target molecule could be achieved throughthe well-established phage display technology [34]. The first objectivein the construction of a library for this purpose is to design a suitableVLR template sequence. After analyzing 222 unique sequences of theVLRB diversity region from the sea lamprey we designed a consensusprotein containing a single LRRV module flanked by LRRNT and LRR1motifs on the N-terminal side and by LRRVe, LRRCP and LRRCT on theC-terminus (Fig. 1). Among the 168 positions in the polypeptidesequence of dVLR 86%were considered as conserved (Fig. 2), while forthe remaining 14% of positions (23 residues) the most frequent aminoacid was present in less than 35% of sequences. In our designedsequence all but one position were occupied with the most frequentamino acid. For potentially destabilizing position 19 we selectedserine instead of the most frequent arginine.

The designed VLR sequence showed very favorable biophysicalproperties. The protein could be easily produced in a bacterialexpression system in a soluble form, despite the presence of eightcysteines that in natural VLR proteins form two sets of disulfidebridges (in both capping modules, LRRNT and LRRCT), but potentiallycould engage in intermolecular bridges in the heterologouslyexpressed dVLR and thus cause its aggregation. Regardless of thispossibility the dVLR expressed in E. coli was soluble, monomeric(Fig. 3) and, as determined by Ellman's assay, had all eight cysteines inthe oxidized form.

The far-UV CD spectrum of the protein revealed features of an α/βprotein indicating its proper folding. The designed protein was highlystable in a broad pH range and retained its secondary and tertiarystructure at the pH range of 4.0 to 10.0 (Fig. 4). To further investigatethe structure of dVLR, we used 8-anilino-1-naphtalene-sulfonate(ANS), a probe widely used to study partially folded and moltenglobule states [35]. In an aqueous environment the fluorescence ofthis dye is quenched, but in contact with hydrophobic surfacesexposed in a molten globule state its intensity is significantlyenhanced. Only marginal ANS binding was detected for the dVLRprotein [Fig. 4B, inset], which confirms native state (and properfolding) of designed sequence.

The highest denaturation temperature of 73.9 °C was observed atpH 6.0 (denaturation monitored by CD) and exceeded 60 °Cthroughout the pH range 4.0 to 10.0 (Table 1). The denaturationprocess was reversible and highly cooperative, with the denaturationenthalpy (ΔHden) within the range of 99 to 115.5 kcal/mol. Chemicaldenaturation of dVLR at pH 7.5 was a reversible, highly cooperativetwo-state process (Fig. 5B) with the free energy change of unfolding(ΔGden

H20) of 14.1 kcal/mol or 14.0 kcal/mol and midpoints at 3.3 MGdmCl or 3.2 M GdmCl, as monitored by CD or fluorescence,respectively. Similar cooperative two-state denaturation was ob-served for ankyrin repeat proteins (ΔGden

H20=9.5 to 21.1 kcal/mol) [31].Compared with other leucine-rich repeat proteins, dVLR shows

much better biophysical properties. Repeat proteins based onribonuclease inhibitor sequence have been designed with one to sixdouble LRRs, flanked by capping repeats, each double LRR composedof two types of fairly long modules A and B of 28 and 29 amino acids,respectively. These polypeptides produced in E. coli were mostlyinsoluble and the largest varieties tended to aggregate. Chemicalunfolding with urea monitored by CD and fluorescence spectroscopy,gave midpoints transition between 2 and 4 M urea. The CD

1145M. Wezner-Ptasińska et al. / Biochimica et Biophysica Acta 1814 (2011) 1140–1145

denaturation data of that designed protein built of 12 modules couldbe fitted to a two-state model with a ΔGden

H20 of only 0.72 kcal/mol [14].Two other proteins composed of LRR repeats similar in size to

dVLR have been analyzed in terms of their stability. The first N-terminal half of internalin B from Listeria monocytogenes containingseven LRRs flanked by an α-helical cap. Its chemical denaturation wasdescribed by a two-state model with a midpoint at 0.7 M GdmCl andΔGden

H20=5.42 kcal/mol, and the denaturation temperature at pH 7.5was 42 °C [36], i.e. almost 30 °C less than that of our designed VLR. Thesecond protein was a C-terminal deletion construct of the YopMprotein from Yersinia pestis built of an N-α-cap fallowed by six LRRs.Urea-induced denaturation of this protein exhibited a transition witha midpoint at 0.7 M urea and ΔGden

H20 was 1.7 kcal/mol [37]. Thus, thosetwo protein fragments were much less stable than our designedsequence.

In summary, using a consensus approach we design a scaffoldbased on sequences of variable lymphocyte receptor that could beefficiently expressed in a soluble form in E. coli. The designed dVLRretained its secondary and tertiary structure in a wide range of pH,was highly stable and had a very high free energy of unfolding. Thebiophysical properties of dVLR make this novel protein perfectlysuitable for biotechnological applications as ligand-binding mole-cules, which could be selected by phage display.

Acknowledgments

The work was supported by the Ministry of Science and HigherEducation, Grant NN302 252233.

References

[1] S. Tasumi, C.A. Velikovsky, G. Xu, S.A. Gai, K.D. Wittrup, M.F. Flajnik, M.A.Mariuzza, Z. Pancer, High-affinity lamprey VLRA and VLRBmonoclonal antibodies,PNAS 106 (2009) 12891–12896.

[2] Z. Pancer, C.T. Amemiya, G.R.A. Ehrhardt, J. Ceitlin, G.L. Gartland, M.D. Cooper,Somatic diversification of variable lymphocyte receptors in the agnathan sealamprey, Nature 430 (2004) 174–180.

[3] M.N. Alder, I.B. Rogozin, L.M. Iyer, G.V. Glazko, M.D. Cooper, Z. Pancer, Diversityand function of adaptive immune receptors in a jawless vertebrate, Science 310(2005) 1970–1973.

[4] J. Bella, K.L. Hindle, P.A. McEwan, S.C. Lovell, The leucine-rich repeat structure,Cell. Mol. Life. Sci. 65 (2008) 2307–2333.

[5] A.V. Kajava, Structural diversity of leucine-rich repeat protein, J. Mol. Biol. 277(1998) 519–527.

[6] M.N. Alder, B.R. Herrin, A. Sadlonova, C.R. Stockard, W.E. Grizzle, L.A. Gartland, G.L.Gartland, J.A. Boydston, C.L. Turnbough Jr.,M.D. Cooper, Antibody responses of variablelymphocyte receptors in the lamprey, Nat. Immunol. 9 (2008) 319–327.

[7] I.B. Rogozin, L.M. Iyer, L. Liang, G.V. Glazko, V.G. Liston, Y.I. Pavlov, L. Aravind, Z. Pancer,Evolution and diversification of lamprey antigen receptors: evidence for involvementof an AID-APOBEC family cytosine deaminase, Nat. Immunol. 8 (2007) 647–656.

[8] J. Kasamatsu, Y. Sutoh, K. Fugo, N. Otsuka, K. Iwabuchi, M. Kasahara, Identification of athird variable lymphocyte receptor in lamprey, PNAS 107 (2010) 14304–14308.

[9] P. Guo, M. Hirano, B.R. Herrin, J. Li, C. Yu, A. Sadlonova, M.D. Cooper, Dual nature of theadaptive immune system in lampreys, Nature 459 (2009) 769–801.

[10] B.R. Herrin, M.N. Alder, K.H. Roux, C. Sindal, G.R.A. Ehrhardt, J.A. Boydstron, C.L.Turnbough Jr., M.D. Cooper, Structure and specificity of lamprey monoclonalantibodies, PNAS 105 (2007) 2040–2045.

[11] C.A. Velikovsky, L. Deng, S. Tasumi, L.M. Iyer, M.C. Kerzic, L. Aravind, Z. Pancer, R.A.Mariuzza, Structure of a lamprey variable lymphocyte receptor in complex with aprotein antigen, Nat. Struc. Mol. Biol. 16 (2010) 725–731.

[12] B.W. Han, B.R. Herrin, M.D. Cooper, I.A. Wilson, Antigen recognition by variablelymphocyte receptors, Science 321 (2009) 1834–1837.

[13] E.R.G. Main, A.R. Lowe, S.G.J. Mochrie, S.E. Jackson, L. Regan, A recurring theme inprotein engineering: the design, stability and folding of repeat proteins, Curr. Opin.Struct. Biol. 15 (2005) 464–471.

[14] M.T. Stumpp, P. Forrer, H.K. Binz, A. Pluckthun, Designing repeat proteins: modularleucine-rich repeat protein libraries based on the mammalian ribonuclease inhibitorfamily, J. Mol. Biol. 332 (2003) 471–487.

[15] I. Letunic, L. Goodstadt, N.J. Dickens, T. Doerks, J. Schultz, R. Mott, F. Ciccarelli, R.R.Copley, C.P. Ponting, P. Bork, Recent improvements to the SMART domain-basedsequence annotation resource, Nucleic Acids Res. 30 (2002) 242–244.

[16] The UniProt Consortium, The Universal Protein Resource (UniProt) in 2010, NucleicAcids Res. 38 (2010) D142–D148.

[17] E. Jain, A. Bairoch, S. Duvaud, I. Phan, N. Redaschi, B.E. Suzek, M.J. Martin, P. McGarvey,E. Gasteiger, Infrastructure for the life sciences: design and implementation of theUniProt website, BMC Bioinforma. 10 (2009) 136.

[18] M.A. Larkin, G. Blackshields, N.P. Brown, R. Chenna, P.A. McGettigan, H.McWilliam, F. Valentin, I.M. Wallace, A. Wilm, R. Lopez, J.D. Thompson, T.J.Gibson, D.G. Higgins, Clustal W and Clustal X version 2.0, Bioinformatics 23(2007) 2947–2948.

[19] S. Wu, Y. Zhang, MUSTER: Improving protein sequence profile–profile alignments byusing multiple sources of structure information, Proteins 72 (2008) 547–556.

[20] Y. Zhang, I-TASSER server for protein 3D structure prediction, BMC Bioinforma. 9(2008) 40.

[21] Y. Zhang, Template-basedmodeling and freemodeling by I-TASSER in CASP7, Proteins69 (2007) 108–117.

[22] S. Wu, J. Skolnick, Y. Zhang, Ab initio modeling of small proteins by iterative TASSERsimulation, BMC Biol. 5 (2007) 17.

[23] M. Pawlowski, M.J. Gajda, R. Matlak, J.M. Bujnicki, MetaMQAP: a meta-server for thequality assessment of protein models, BMC Bioinforma. 9 (2008) 403.

[24] W.L. DeLano, The PyMOL Molecular Graphics System, DeLano Scientific, Palo Alto, CA,2002. http://www.pymol.org.

[25] W.L. DeLano, The PyMOL User's Manual, DeLano Scientific, Palo Alto, CA, USA,2002.

[26] L.J. Lucast, R.T. Batey, J.A. Doudna, Large-scale purification of a stable form ofrecommbinant tobacco etch virus protease, Biotechniques 30 (2001) 544–546.

[27] C. Perz-Iratxeta, M.A. Andrade-Navarro, K2D2: estimation of protein secondarystructure from circular dichroism spectra, BMC Struct. Biol. 8 (2008) 25.

[28] M. Zakrzewska, D. Krowarsch, A. Wiedlocha, S. Olsnes, J. Otlewski, Highly stablemutants of human fibroblast growth factor-1exhibit prolonged biological action,J. Mol. Biol. 352 (2005) 860–875.

[29] M. Zakrzewska, D. Krowarsch, A. Wiedlocha, J. Otlewski, Design of fully active FGF-1variants with increased stability, PEDS 17 (2004) 603–611.

[30] E.R.G.Main, Y. Xiong,M.J. Cocco, L. D'Andrea, L. Regan,Design of stableα-helical arraysfrom an idealized TPR motif, Structure 11 (2003) 497–508.

[31] H.K. Binz, M.T. Stumpp, P. Forrer, P. Amstutz, A. Pluckthun, Designing repeat protein:well-expressed, soluble and stable proteins from combinatorial libraries of consensusankyrin repeat proteins, J. Mol. Biol. 332 (2003) 489–503.

[32] E.R.G.Main, S.E. Jackson, L. Regan, The folding anddesign of repeat proteins: reaching aconsensus, Curr. Opin. Struct. Biol. 13 (2003) 482–489.

[33] H.M. Kim, S.C. Oh, K.J. Lim, J. Kasamatsu, J.Y. Heo, B.S. Park, H. Lee, O.J. Yoo,M. Kasahara,J. Lee, Structural diversity of the hagfish variable lymphocyte recptors, J. Biol. Chem.282 (2007) 6726–6732.

[34] R.J. Hosse, A. Rothe, B.E. Power, A new generation of protein display scaffolds formolecular recognition, Protein Sci. 15 (2006) 14–27.

[35] G.V. Semisotnov, N.A. Rodionova, O.I. Razgulyaev, V.N. Uversky, A.F. Gripas, R.I.Gilmanshin, Study of the “molten globule” intermediate state in protein folding by ahydrophobic fluorescent probe, Biopolymers 31 (1991) 119–128.

[36] A. Freiberg, M.P. Machner, W. Pfeil, W. Schubert, D.W. Heinz, R. Sockler, Folding andstability of the leucine-rich repeat domain of internalin B from Listeria monocytogenes,J. Mol. Biol. 337 (2004) 453–461.

[37] E. Kloss, D. Barrick, C-terminal deletion of leucine-rich repeats from YopM reveals aheterogeneous distribution of stability in a cooperatively folded protein, Protein Sci. 18(2009) 1948–1960.