Introduction

Immunotherapy is one of the most intensively developing and promising branches of science in the field of neurodegenerative diseases. Scientists are working hard to find a vaccine which is able to prevent pathological changes in central nervous system, and a treatment that could possibly be able to reverse the changes already present in neuron cells. Possibility of personalization is a significant advantage of this type of therapy. It has to be remembered, however, that vaccine preparation can be technically complicated and often requires significant amounts of a pathogen material to obtain the bioactive preparation. Possibility of creating epitope-based vaccines could create an opportunity for immunotherapeutics that can be safer, cheaper and less complicated to produce. Consequently, the most important task is to choose the best and most reliable method for determination of the most promising epitope(s).

In our work, we have focused on one of the neurodegenerative disorders, in which the main role is played by human cystatin C (hCC). This small protein (120 Aa) belongs to cystatins family and its first identified function was inhibition of cysteine proteases. Other functions of cystatin such as antibacterial and antiviral activity have been also reported (Björck et al. 1990; Mussap and Plebani 2004) hCC is present in almost all bodily fluids in significant concentrations and has been reported to interact with other proteins circulating in the human body including the amyloidogenic ones, including serum amyloid A (SAA) (Bokarewa et al. 2007; Spodzieja et al. 2010, 2012), and amyloid beta peptide (Aβ) (Juszczyk et al. 2008). This makes cystatin C an important target for studies focused on a search of new functions and new ligands of this protein.

Cystatin C has also special place in the field of neurodegenerative diseases. Its mutated variant with the point mutation in position 68 (L68Q) is responsible for hereditary amyloid angiopathy Icelandic type (HCCA, hereditary cystatin C amyloidosis). The aggregation of mutated protein is drastically fast, and, when it takes place in the cerebral vessels, it results in recurring strokes and hemorrhages in the brain. The problem affects mostly young adults, usually in their twenties or thirties, and with the lack of any kind of a therapy, people with this genetic mutation are generally sentenced to an early death.

The exact mechanism of hCC aggregation is not known but there is some evidence pointing towards its association with another cystatin C feature namely its propensity to undergo the domain swapping process. It may lead to the dimerization of the protein (Janowski et al. 2001) but also to its more pronounced oligomerization and finally fibrilization (Wahlbom et al. 2007). Initial information about potential use of an antibody to prevent the domain swapping and dimerization of cystatin C was presented in 2004 by Nilsson et al. (2004). Monoclonal antibody IgG2b(κ) (Olafsson et al. 1988) used in the experiments was shown to suppress the dimerization of L68Q mutant of hCC as well as of the native protein. Since then the work on anti-hCC mAbs has begun, and, with time, the number of commercially available antibodies against hCC increased significantly. Östner et al. checked the antiaggregational potential of 12 monoclonal antibodies (Östner et al. 2011), and found that some of them have great potential to stop dimerization of cystatin. The highest inhibition was observed for clones Cyst28 (75 %) and HCC3 (60 %). In contrast to these two, almost no inhibition was observed when clone Cyst10 was used in the experiment (only 4 % of inhibition). Such striking differences in antiaggregational propensities of the studied antibodies prompts the question about the mode of interaction with the antigen (hCC) and, more interestingly, the sites of these interactions. Our work was aimed at the determination of the hCC sequence fragments constituting the epitopes recognized by different antibodies.

To start our work, we have chosen two clones with opposite antiaggregational potential: clone Cyst28 and clone Cyst10. The most significant techniques currently used in epitope identification focus on an immunocomplex X-ray diffraction analysis (Lescar et al. 1997), nuclear magnetic resonance (NMR) (Anglister et al. 1993), enzyme-linked immunosorbent assay (Butler 2000), surface plasmon resonance (SPR) (Mullett et al. 2000), and phage display (Wang and Yu 2004). An alternative for above-mentioned techniques are approaches based on mass spectrometry (MS). Merging high association constants of antigen–antibody complexes (Cerutti et al. 2006), high resistance of antibodies to proteolysis (Parham 1983) with high sensitivities and high resolving power of new mass spectrometers allows for mapping of all kinds of epitopes (linear or conformational) (Papac et al. 1994; Parker et al. 1996; Jeyarajah et al. 1998; Peter and Tomer 2001; Parker and Tomer 2002; Hager-Braun et al. 2006; Williams et al. 2006). One of the strong points of mass spectrometry-based methods is possibility of ion detection at very low sample concentration (fmol-pmole range).

We have used the epitope extraction/excision mass spectrometry approach as a basic technique for the epitope identification (Dhungana et al. 2009). It has been proved that the limited proteolysis of the hCC-Ab immune complex allows identification of both continuous (linear) and discontinuous (conformational) epitopes (Hager-Braun et al. 2006; Williams et al. 2006; Dhungana et al. 2009). Linear epitopes consist usually of approximately 9–12 amino acid residues and represent amphipathic properties (Gershoni et al. 2007). In the case of a discontinuous type the situation is more complicated. Most of conformational epitopes consist of 15–22 residues derived from 2 to 3 different protein segments (domains), which are brought together in the three-dimensional space by the folding of the protein and an antibody recognizes the surface created by all of epitopic fragments (Ponomarenko and Van Regenmortel 2009; Gershoni et al. 2007).

In this work, we present the results of epitope identification for two clones of anti-cystatin C antibodies: mAb Cyst28 and mAb Cyst10. With the use of limited proteolysis followed by MS studies, the binding sites for antibodies have been localized. They present discontinuous (structural) type. Epitopic sequences for the highly potent dimerization inhibitor clone 28 are located on N-terminus, loop 1 and 2. In the case of clone 10 only one loop (loop 2) together with a fragment of an Appending Structure (AS) is engaged in interactions with the antibody. The results clearly explain low potential of clone 10 for inhibition of cystatin C domain swapping process. Only mAb Cyst28 interacts with hCC via the loop 1, which has been proved to play an important role in the domain swapping process (Szymańska et al. 2009; Orlikowska et al. 2011).

Materials and methods

Materials

Mouse monoclonal antibodies Cyst10 and Cyst28 were purchased from HyTest Company (Turku, Finland) (4CC1).

Peptide synthesis and purification

Fragments of hCC were synthesized according to published methods using standard solid-phase peptide synthesizer with continuous flow (Millipore 9050 Plus PepSynthesizer) and a strategy compatible with Fmoc-chemistry (Fields and Noble 1990).

All reagents and chemicals required for the synthesis were purchased from Peptide International, Fluka, and Sigma-Aldrich. The TentaGel R RAM resin (Rapp Polymers; capacity of 0.19 mmol/g, 1 g) was used as a solid support. Peptidylresin, after automatic synthesis and removal of the N-terminal Fmoc-group was treated with 10 ml of the cleavage cocktail [trifluoroacetic acid/water/phenol/triisopropylsilane (8.8/0.5/0.5/0.2, v/v)] at room temperature for 2 h with gentle shaking. After filtration and evaporation of the cleavage mixture, cold diethyl ether was added to the filtrate and the precipitated peptide was isolated by centrifugation. The crude product was purified by a semi-preparative RP-HPLC using C8 column (10 × 250 mm, 5 µm). The aqueous system consisted of 0.1 % (v/v) TFA in water, while the organic solution was 80 % acetonitrile in water, containing 0.08 % (v/v) TFA. The elution was carried out using a linear gradient from 0 to 80 % of the organic phase in 60 min and the separation was monitored by UV absorbance at 220 nm.

The peptide purity and identity was verified by an analytical HPLC system using C8 column (4.4 × 250 mm, 5 µm) on the Varian chromatograph and by ESI-IT-TOF MS mass spectrometer (Shimadzu, Shimpol, Warsaw, Poland).

Synthesis of biotinylated peptides

The hCC fragments used in the ELISA-like tests were elongated by five glycine residues followed by biotin. The peptides were synthesized automatically, and biotin was attached manually using O-(benzotriazol-1-yl)-N,N,N,N-tetramethyluronium tetrafluoroborate (2.3 eq. to resin capacity), N,N-diisopropylethylamine (2 eq. to biotin) and biotin (2.5 eq. to resin capacity). The biotinylated peptides were detached from the support and purified as described above. Fractions containing pure peptides were lyophilized, and the purity was assessed by ESI-IT-TOF MS (Shimadzu, Shimpol, Warsaw, Poland).

hCC expression and purification

The protein was produced in E.coli strain C41(DE3) and purified by ion-exchange chromatography on S-Sepharose as described previously (Szymańska et al. 2009). The pure hCC fractions were collected, dialyzed extensively against 10 mM ammonium bicarbonate (pH 8.0) and lyophilized. The protein purity was characterized by SDS-PAGE electrophoresis and size exclusion chromatography.

Monoclonal antibody immobilization on Sepharose

The coupling reaction was performed in a microcolumn (MoBiTec, Göttingen, Germany) for 1 h at 25 °C with thorough mixing. The solution containing approximately 100 μg of monoclonal antibody (Cyst10 or Cyst28) in coupling buffer (0.2 M NaHCO3, 0.5 M NaCl, pH 8.3), was added to 66 mg of dry NHS-activated Sepharose (4B, Sigma-Aldrich). The next steps consisted of alternate washing with blocking solution (0.1 M aminoethanol, 0.5 M NaCl, pH 8.3) and washing solution (0.2 M CH3COONa, 0.5 M NaCl, pH 4.0). Next, the resin was incubated in blocking solution for 1 h at 25 °C and washed again as described above. Finally the column was washed with PBS buffer, pH 7.5 and stored at 4 °C (Juszczyk et al. 2009).

Binding of hCC/hCC fragments to the immobilized antibody (microcolumn affinity assay)

A solution of hCC or hCC fragment (10 μg/column) was loaded onto anti-hCC mAb-Sepharose column equilibrated in 50 mM NH4HCO3 (pH 8.3) and incubated for 2 h at 25 °C with gentle shaking. The excess of the unbound protein/peptide was washed off with 50 mM NH4HCO3. After washing, the complex was dissociated, the bound material was eluted under acidic conditions (0.1 % TFA/H2O, pH 2.5) and the results were analyzed with SDS-PAGE electrophoresis or MALDI-MS (Juszczyk et al. 2009).

Epitope excision and extraction

The formation of a complex between an immobilized antibody (mAb-Sepharose) and its antigen (hCC) was the basis of the epitope excision method. The complex between immobilized antibody and full length cystatin C was formed as described above and after extensive washing with 50 mM NH4HCO3 was digested with appropriate enzymes; trypsin (2 h at 37 °C), Asp-N (18 h at 37 °C) and pronase (2 h at 40 °C). The proteolysis with trypsin was stopped by addition of the inhibitor, and with pronase by extensive washing of the microcolumn with the buffer. The unbound digestion products were removed by washing with 50 mM NH4HCO3, whereas the bound hCC fragments were dissociated by addition of 0.1 % TFA/H2O and analyzed by mass spectrometry. In the second version of the experiment (epitope extraction) the complex of an antibody and hCC fragment(s) was prepared by addition of proteolytic digest of hCC to the immobilized mAb and subsequent incubation for 2 h at 25 °C. The excess of the peptides was removed by washing and the eluate containing interacting fragments was obtained by elution with 0.1 % TFA/H2O as described above (Juszczyk et al. 2009).

ELISA test for binding of hCC fragments to mAbs

ELISA assay for the antibodies interactions with hCC fragments was started by covering a 96-well microtiter plate (ImmunoGrade, BrandTech Scientific) with 50 µl of the antibody solution (2 µg/ml) in a buffer containing 0.09 M Na2CO3, 0.012 M NaHCO3 (pH 10.4). After an overnight incubation at 4 °C and washing using PBS buffer containing 0.05 % Tween-20 (PBST), the plate was coated with serial dilutions of N-biotinylated hCC fragments (N-biotinyl-(Gly)5-hCC peptide) and incubated for 1 h. After washing, the streptavidin-HRP conjugate in PBST (dilution 1:5000) was added and incubated for 1 h. Finally, after scrupulous washing of the plate, the binding affinities were determined by addition of 50 µl of the substrate (3,3′,5,5′ tetramethylbenzidine, TMB) and measurement of the absorbance at 650 nm with microplate reader (TECAN).

Microscale Thermophoresis (MST)

Thermophoresis binding experiments were carried out with Monolith NT.115 system (NanoTemper Technologies GmbH), according to the manufacturer’s protocol. Briefly, the sample solution containing 15 µM human cystatin C was labelled with fluorescent dye NT-647 (Protein Labeling Kit RED-NHS, Amine Reactive, NanoTemper Technologies GmbH). The labelling procedure and the subsequent removal of the free dye were performed within 45 min according to the applied protocol (Wienken et al. 2010). The concentration of the NT-647-labelled hCC was 4.7 µM. A serial dilutions of the non-labelled titrant: mAb Cyst28 (3.87 µM–0.12 nM) or Cyst10 (8.00 µM–2.44 nM), were prepared using MST optimized buffer (50 mM Tris, pH 7.6, 150 mM NaCl, 10 mM MgCl2, 0.05 % Tween-20). 10 µl of the serial dilution of the non-labelled molecule were mixed with 10 µl of the diluted fluorescently labelled molecule. Mixed samples were loaded into glass capillaries and the MST analysis was performed using Monolith.NT115 and Monolith software.

MALDI analysis

The MS measurements were performed with the use of MALDI TOF/TOF 5800 (ABSciex, Germany). As a matrix α-cyano-4-hydroxycinnamic acid (CHCA, Sigma-Aldrich) was used. The measurements were done in reflector positive mass mode with previous mass calibration with commercial standard peptide mixture (The Peptide Mass Standards Kit for Calibration of AB SCIEX MALDI-TOF™ Instruments). Samples were prepared using the dried droplet preparation method by mixing 0.6 µl of an analyte solution with 0.6 ml of matrix solution (directly on a plate). After air drying the plate was introduced directly into the instrument. MS spectra were acquired from 500 to 5000 m/z for a total of 1000 laser shots by an 1 kHz OptiBeam laser (YAG, 349 nm). Laser intensity remained fixed for all the analyses. Registered spectra were analyzed with Data Explorer software (Sciex).

Results

Affinity of mAb Cyst10 and 28 to human cystatin C

Microscale thermophoresis. To determine the affinity of the studied antibody clones to human cystatin C, microscale thermophoresis (MST) was applied. This technique creates the opportunity to work under close to native conditions, and is based on detection of small changes in mobility of the molecules in microscopic temperature gradients (Jerabek-Willemsen et al. 2011; Seidel et al. 2013). MST also offers advantages over the most common nonfluorescent methods such as isothermal titration calorimetry (ITC) or surface plasmon resonance (SPR), by avoiding surface immobilization and high sample consumption (Jerabek-Willemsen et al. 2014).

In our experiments, human cystatin C was labelled by amine reactive red dye NT-647 (excitation 625 nm, emission 680 nm). NHS-activated ester of the dye reacts efficiently with primary amine groups of cystatin and forms highly stable conjugates. In the experiments for both clones, concentration of hCC was kept constant on a level of 4.7 µM for the entire duration of the experiments. Concentration of the antibodies varied for the clone 10 between 2.44 nM to 8 microM, and for the clone 28—between 3.87 µM to 0.12 nM. After short incubation the samples were loaded into MST hydrophilic glass capillaries and the analysis was performed using Monolith NT.115. The results of the experiments are presented in Fig. 1. In the case of mAb Cyst28 a clear and strong interaction was observed with KD value for this immunocomplex 20.2 ± 1.85 nM. For the second complex the interaction was also observed, but the KD was found much higher (141 ± 13.2 nM). Comparison of the results for both mAbs clearly shows that the affinity (1/KD) to human cystatin C for clone 28 is ca. six times higher than for clone 10.

Fig. 1
figure 1

The MST experiments for a mAb Cyst28, K D = 20.2 ± 1.85 nM; and b mAb Cyst10 K D = 141 ± 13.2 nM

Proteolysis with trypsin and Asp-N: MS-assisted preliminary epitope assessment

Structural information about monoclonal antibodies and their interactions with antigens are a prerequisite for their potential use in therapeutic applications. There are many strategies providing information of epitopic sequences characteristic for mAbs and their target proteins and many of them are mass spectrometry-based (Zhang et al. 2014). One of the experimental approaches, known as epitope extraction/excision procedures was successfully used in the past by our group for binding site identification in the complexes of hCC with different protein/peptide ligands (e.g., amyloid β peptide, the monoclonal antibody Cyst13 or serum amyloid A) (Juszczyk et al. 2009; Sladewska et al. 2011; Spodzieja et al. 2012, 2013). To start the experiments, microcolumn with the appropriate antibody attached to the N-activated Sepharose was prepared and checked for the affinity using simple binding assay with the intact human cystatin C molecule. Formation of the hCC-mAb immunocomplex was proved by the observation of hCC signal in the fraction obtained after acidic dissociation of the immunocomplex (see—electrophoretogram and MS spectra of elution fraction in the Supplementary Material, Fig. 1). Next, the system was used for the preliminary epitope assessment. To start the search for hCC sequences involved in binding to the studied antibodies, two basic enzymes, trypsin (cleaving peptide chains mainly at the carboxyl side of lysine or arginine), and Asp-N (an endoproteinase that hydrolyzes peptide bonds on the N-terminal side of aspartic acid), were chosen. The enzymatic digestion was carried out for cystatin C in the complex with immobilized antibody (epitope excision approach) or free in solution (epitope extraction). In the latter case the enzymatic digest was applied on the column with immobilized antibody. Each proteolytic experiment was repeated at least three times and the results of epitope excision and extraction are presented in Table 1. The use of trypsin in the epitope extraction approach in the case of mAb Cyst28 revealed seven hCC regions potentially engaged in the immunocomplex formation. The fragments from N-terminal part of the protein, such as the sequence 1–15 and 16–42 were not observed, unlike the middle and C-terminal part of hCC. When Asp-N was used for hCC digestion, the elution fractions from the extraction experiment gave only two hCC fragments covering residues from 40 to 118 (with small gap from Asp65 to His86) whereas the N- and C-terminal hCC fragments were not present. The same results were obtained when the epitope excision approach was applied. In the case of trypsin digestion of the immunocomplex between human cystatin C and mAb Cyst28 in the elution fraction (after the complex dissociation) we have not observed any peptide fragments resulting from expected hCC digestion (range 800–5000 m/z). Therefore, we additionally checked the higher masses region (from 5000 to 20 000 m/z), and in this range we observed some weak signals that could be associated with very long hCC fragments, for example, one resulting from the cleavage of the first eight amino acid residues in N-terminal part (first possible digestion after Arg8 in the hCC sequence). Since the complex formation for Cyst28 and hCC was confirmed, and the cleavage of hCC with trypsin in solution proceeds unequivocally, the observed result has to be associated with an unexpected influence of the antibody/immunocomplex on trypsin activity.

Table 1 Summary of hCC fragments obtained after epitope extraction procedure with trypsin and extraction/excision with Asp-N for clone 10 and 28

For the clone 10 the extraction experiment using tryptic digest of hCC revealed peptides covering residues from 37 to 120. Digestion with Asp-N reduced the number of peptides to two fragments, representing middle and C-terminal parts of hCC (fragments 40–64 and 87–120, respectively). For this enzyme, the results from the epitope extraction were confirmed by the epitope excision approach. However, similarly as in the experiments with the clone 28, the cleavage of the immunocomplex hCC-Cyst10 by trypsin in the excision approach followed by the complex dissociation, in our hands did not bring any informative results (no peptides in the elution fraction observed in the range 800–5000 m/z).

At this point of the analysis the basic difference between the two clones can be summed up to the presence of N-terminal hCC fragments in the elution fractions in the case of clone 28, and only C-terminal fragments of hCC in the case of clone 10. Generally, first experiments suggested that the epitopes for the studied anti-hCC mAb clones are located in slightly different parts of cystatin C and that in both cases they are of discontinuous nature.

Peptide affinity to the immobilized antibody

Affinity of selected peptides to the studied antibodies was assessed in binding experiments to get more detailed information on sequences involved in the immunocomplexes formation. The microtiter plate was covered with the appropriate mAb (2 µg/ml) and solutions of the biotinylated hCC fragments after serial dilution were added. Streptavidin-HRP conjugate was used as a detection reagent. The affinity of the peptides was measured by means of the TMB-derived diimine absorbance intensity at 650 nm and the results are presented in Fig. 2. In the case of clone 28, the differences in binding are clearly visible. The highest affinity was observed for middle and C-terminal fragments of hCC (43–72, 85–101 and 93–120). Poor interactions were observed for the fragment 16–42, especially at lower concentration of the ligand. The immunocomplex concentration increased significantly when the ligand was present in higher amounts and the equilibrium was shifted to the complex side. N-terminal fragment of hCC(1–15) displayed the weakest affinity to Cyst28 that is rather unexpected result because this part of the protein was present in the epitope extraction results with trypsin.

Fig. 2
figure 2

Interaction of synthetic human cystatin C fragments with two monoclonal antibodies Cyst28 and Cyst10, determined by binding assay. TMB-derived diimine absorbance at 650 nm was read out for the studied peptides. Non-binding fragments of hCC were used as a control

In the case of Cyst10, the results were similar. Strongest binding was observed for the fragment 43–72, as well as for the C-terminal fragments 85–101 and 93–120. In the case of the last peptide, dropping of the line (signal) after achieving the maximal absorption (binding) was observed. The same effect was also noted for the clone 28 and in our previous studies (Juszczyk et al. 2009). This effect can be associated with relatively low solubility of this peptide and its high tendency to self-association leading to precipitation at higher concentrations.

hCC fragments affinity to the immobilized antibody

To cross-check the above-described results and confirm that peptide fragments extracted from the hCC sequence in extraction/excision procedures display significant binding to the studied monoclonal antibodies, affinity experiments with the use of a microcolumn with the immobilized mAbs were performed. Synthetic fragments covering the sequences obtained in the extraction/excision experiments as well as some shorter sequences were obtained. For the synthetic purposes the cysteine residues at positions 73, 83, 97, and 117 were exchanged for serine. We have verified in our previous studies that this exchange does not influence affinity of a peptide to its interacting partner, but makes a significant difference during peptide synthesis and purification (He and Quiocho 1991; Spodzieja et al. 2013). The ionization of serine derivatives of hCC is also better thus facilitating mass spectrometry analysis (unpublished data).

The results of the binding assay of the synthetic peptides to mAb clone 28 presented in Table 2 confirmed the results from the epitope extraction/excision studies for majority of the studied hCC fragments. The absence of some peptides such as fragments: 33–51 or 65–83 may suggest unspecific binding of these fragments either to the antibody or to peptides already bound to mAb, which may occur when working with multicomponent enzymatic digest mixture used in the epitope extraction experiment. Therefore, the results presented above, in our opinion, do not contradict the data obtained in the first approach. For example, the non-binding fragment 33–51 is a part of helical structure which can interact via its side chains with the beta strands of hCC. Two other nonbinding peptides represent aforementioned beta strands which, due to their hydrophobic character, can easily bind to other cystatin fragments. Fragments of middle part with the sequence covering hCC fragment 43–72, and for C-terminal part of hCC fragments 85–101, 93–120 were selected for further studies.

Table 2 Summary of the affinity binding experiments of hCC fragments to clone Cyst28 (elution fraction)

For the clone 10 similar binding experiments were performed and the results are summarized in Table 3. They are also in agreement with the data obtained in the epitope extraction and excision experiments.

Table 3 Summary of the affinity binding experiments of hCC fragments to clone Cyst10 (elution fraction)

Based on all obtained results it may be concluded that the hCC epitope for mAb clone 10 is not located in N-terminal fragment of hCC. Therefore, we focused our research on middle (43–72) and C-terminal part (85–101, 93–120) of the protein and these three fragments were chosen for further experiments. Fragments 33–51, and 65–83 were considered as nonbinding ones.

Proteolysis of the immunocomplexes with pronase

Aiming at narrowing down the number of hCC residues potentially involved in the immunocomplex formation and refine the epitopic sequences, we used the enzyme-based MS approach using pronase. Pronase is a mixture of several nonspecific endo- and exopeptidases that digest proteins to individual amino acids. It can be used for precise determination of sequences involved in intermolecular complex formation (Simmons 1988; Marzilli et al. 2000; Dodds et al. 2009). This method was previously used in our studies of complexes of hCC with amyloid beta peptide (Juszczyk et al. 2009) and serum amyloid A (Spodzieja et al. 2012) to give precise information about the protein binding sites. In the present study the immunocomplex between the immobilized mAbs and synthetic fragments of cystatin C was formed and subsequently exposed to limited proteolysis with pronase. The results received for both clones are presented in Table 4 and in Supplementary materials (Figs. 2, 3).

Table 4 The results of the epitope excision procedure with pronase for the immunocomplexes between mAb Cyst28/10 and the synthetic hCC fragments and the microcolumn affinity assay results for the determined hCC fragments
Fig. 3
figure 3

Interaction of synthetic human cystatin C short epitopic fragments with two monoclonal antibodies Cyst28 and Cyst10, determined by binding efficiency test. TMB-derived diimine absorbance at 650 nm was read out

The results of the excision procedure for the immobilized immunocomplexes with pronase for the clone 28 revealed several short 6–7 amino acid long peptides which can be considered as epitopes or parts of a discontinuous epitope. The microcolumn affinity experiments (see: Experimental section) with synthetic equivalents of the identified sequences revealed that all of them bind to the antibody by itself or after elongation by two/three Aa residues from C- and N-terminus. The results point out five major cystatin C fragments interacting with the antibody (Table 4).

For the clone 10 the results from the epitope excision procedure with pronase point to sequences of three, seven to eleven amino acid long, peptides. Some differences between both antibodies can be seen on this level. In the case of Cyst 10, only one peptide from middle part, namely 60–70 fragment was revealed after digestion with pronase. In the case of C-terminus we can point out two peptides covering the sequence 96–102 and 101–111, whereas clone 28 shows affinity to fragments 85–91, 92–99, and 101–111.

To get more details about binding the above-determined, short hCC fragments to the antibodies, a second ELISA-like assay was performed. This time all short peptides found in the elution fraction after pronase digestion and the immunocomplex dissociation, and their prolonged versions were synthesized as biotinylated derivatives. The results are presented in Fig. 3 and Table 4. According to the obtained data, for mAb Cyst28 the strongest binding can be observed for the peptide hCC(92–99). The other studied peptides: 53–62, 85–91 and 101–111 present moderate interactions with the examined antibody. Only one of the studied hCC fragments—hCC(64–70)—was found as not interacting with Cyst28.

In the case of Cyst10, the binding of short hCC fragments in the ELISA-like test revealed that best results can be observed for C-terminal part of human cystatin C, namely for hCC(101–111). For this peptide the absorbance decline was also observed after achieving the maximum, similarly as it was noted before for longer peptide 93–120. Two other studied fragments represent low binding to the Cyst10 antibody. This can be explained by solubility problems with biotin-5Gly derivatives of the peptides.

Discussion

Seeking an answer to the still-open question about a mechanism of inhibition of cystatin C dimerization by some of monoclonal anti-hCC antibodies, we engaged methods which rely on functional binding of mAbs to the antigen or its derivatives. In the presented studies human cystatin C constitutes the antigen and two of the commercial monoclonal antibodies, Cyst28 and Cyst10, are the ligands for the immunocomplex formation. Epitope extraction/excision methodology, which was previously successfully used in our studies, directly combines affinity binding, limited enzymatic proteolysis and mass spectrometry analysis. This multidisciplinary approach, together with assays like ELISA, can provide valuable information about the location of epitopes.

To get an idea about possible epitopes for both studied clones, it is necessary to analyze simultaneously the results from enzymatic digestion experiments and affinity tests in the context of the three-dimensional structure of human cystatin C. The summary of the identified short epitopic sequences is presented in Fig. 4. For mAb Cyst28 four fragments scattered on middle and C-terminal amino acid sequence of hCC were identified. Based on our studies we determined that sequences located in the N-terminal fragment, comprising residues 1–42 are not responsible for antigen binding. The physiological role of the first 12 residues of hCC is not clear; however, it is known that they are an important part of the active sequence involved in the interactions with cysteine proteases (hCC is their inhibitor) (Abrahamson et al. 1987). The loss of these amino acids, which is physiologically observed, triggers the pathological accumulation and aggregation of hCC molecules (Janowski et al. 2004). Residues comprising the fragment of α-helical structure, which is an important element of the 3D domain swapping process, also do not belong to the epitope part. The strongest affinity to examined mAb is exhibited by fragment 92–99 (KRKAFSSF) which encompasses the end of the Appending Structure (AS). The forth β-strand shows the second strongest affinity. Another hCC fragment exhibiting modern interaction with Cyst28 is the fragment 85–91 (FHDQPHL) which is also a part of the AS (Janowski et al. 2001). AS is the hCC part unrelated to the compact core of the molecule and is positioned on the opposite end of β-sheet relative to N-terminus and loops L1 and L2. From the physiological point of view the fragment is located in close proximity to the site responsible for binding and inhibiting the activity of legumain, the key enzyme in antigen presentation (Alvarez-Fernandez et al. 1999). Location of the two latter fragments is very interesting and intriguing. Their sequences correspond exactly to two loop structures present in human cystatin C: fragment 54–60 (KQIVAGV) covers loop L1 and fragment 101–111 (IYAVPWQGTMT)—loop L2. This is of particular importance due to inhibitory properties of hCC, which is the main regulator for cysteine proteases activity. The above-mentioned loops together with four residues from the N-terminus (8RLVG11) are responsible for the papain-like enzyme binding and it appears worthwhile to further research cystatin C inhibitory properties in the presence of clone 28 (studies in progress). Additionally, loop L1 plays the important role in cystatin C dimerization and aggregation (Szymanska et al. 2012). We have proved in our previous studies that even small changes in loop 1, especially in position 57, can lead either to stabilization of the monomeric form of the molecule or relaxation of the structure and triggering the 3D domain swapping process and dimer formation. If this fragment constitutes also a part of a discontinuous epitope, the explanation of inhibitory properties of clone 28 observed in the dimerization experiments (Östner et al. 2011) can be easily reasoned.

Fig. 4
figure 4

Comparison of epitopic sequences location in monomeric cystatin C structure for Cyst28. Epitope fragments are marked in colors 54–60 (red), 85–91 (orange), 92–99 (blue), and 101–111 (green). Colors in spherical structures correspond to the ones in ribbon presentations. a and b represent views from beta strand side, and c, d represent views where the AS structure is in the front (color figure online)

Binding of an antibody to L1 can exert the stabilization effect on the structure of human cystatin C, rendering the domain exchange simply impossible. This hypothesis would explain the data published by the Swedish group (Östner et al. 2011). According to their data dimerization of cystatin C was reduced up to 75 % in the presence of mAb Cyst28.

The visualization of the identified epitope segments on ribbon and spherical models of hCC (Fig. 4) clearly shows that two of the sequences are located close to each other (54–60, 101–111), and are very well exposed to the environment, thus can be available for any ligand. Two other fragments (85–91 and 92–99), which were identified as potentially binding to mAb Cyst28, are located on the opposite side of the cystatin structure. They belong to Appending Structure which structure is not well defined and can fit to antibody CDR’s pockets.

For the second antibody, clone 10, the suppression of hCC dimerization was described on the level of only 4 % (Östner et al. 2011). For this antibody three sequences possibly constituting a discontinuous epitope were identified. All of the sequences are located in the middle and C-terminal parts of cystatin C. For mAb Cyst10 the N-terminal fragment as well as the crucial L1 loop sequence seem not to be involved in the antibody binding. Epitope sequence starts from residues 60–70 (VNYFLDVELGR) which are located on the second beta strand. Last two fragments found in the excision procedure are overlapping but in elution fractions obtained after digestion with pronase they were found as the separate peptides. The sequence starts with residue 96 and continues to 111. First fragment, 96–102 (FCSFQIY), is a part of third beta strand and the second one, 101–111, is located on loop 2 (the same sequence as found for the clone 28). The spherical presentation of the identified fragments visualized in Fig. 5 clearly shows that two sequences representing residues 60–70 and 96–102 are buried inside the cystatin structure. The only one fragment with an access for any ligand corresponds to the sequence fragment covering residues 101–111.

Fig. 5
figure 5

Comparison of epitopic sequences location in monomeric cystatin C structure for Cyst10, epitope fragments are marked in colors 60–70 (cyan), 96–102 (yellow) and 101–111 (green). Colors in spherical structures correspond to the ones in ribbon presentations. a and b represent views from L2 side, and c, d represent views from βstrand side

Taking into account all of the obtained results it can be concluded that for mAb clone 28 the hCC epitope is conformational and incorporates residues from the hCC middle and C-terminal fragments. The most promising location is between loop 1 and loop 2. They are located close to each other, are exposed to the environment, and potentially can fit to CDRs of the antibody. Two other sequences from C-terminus, part of beta-3 strand and AS structure, can work as secondary (supporting) binding site by the antibody. Especially fragment 85–91 which is well accessible for interactions and can fit to the antibody pocket together with the fragment 92–99.

For the clone 10 only one linear epitope can be proposed, and it comprises residues 101–111 from the L2 loop. Two beta strand fragments present as potential epitopic sequences of studied antibodies and, based on binding efficiency test, show only week affinity to the antibody. In hCC structure they are located next to each other, building the beta structure.

Lack of interactions between the N-terminal and the loop L1 region of cystatin C in the case of the Cyst10 clone is the most striking difference observed for the studied antibodies. This fact may explain significantly lowered potential of Cyst10 antibody for hCC dimerization suppression. It is very likely that effective interactions with the flexible hinge loop L1 is a key factor for inhibition of dimerization by hCC-directed monoclonal antibodies as well as by other hCC ligands. On the other hand, it is possible that the difference between the studied mAbs in suppression of hCC dimerization process is also related to their different affinity towards hCC. As we have demonstrated using microscale thermophoresis the interactions in the immunocomplex with hCC are much stronger for clone Cyst28 than for the clone Cyst10.

We are currently working to verify our conclusions concerning the epitope sequence location by studying the epitope mapping using a complementary approach of amide hydrogen/deuterium exchange (HDX) followed by pepsin digestion and MS analysis. Preliminary data obtained from this new study (manuscript under preparation) seem to be in good agreement with the ones we present in this work.