Derivatization strategies for CE-LIF analysis of biomarkers: Toward a clinical diagnostic of familial transthyretin amyloidosis

1050 Electrophoresis 2014, 35, 1050–1059 Sonia Korchane1,2 Antoine Pallandre1,2 Cédric Przybylski3,4 Christian Poüs5,6 Florence Gonnet3,4 Myriam Taverna1,2 Régis Daniel3,4 Isabelle Le Potier1,2 1Faculté de Pharmacie, Université Paris-Sud, Châtenay-Malabry, France 2CNRS, Institut Galien Paris-Sud, UMR 8612, Châtenay-Malabry, France 3Laboratoire Analyse et Modélisation pour la Biologie et l’Environnement, Université Evry-Val-d’Essonne, Evry, France 4CNRS, Laboratoire Analyse et Modélisation pour la Biologie et l’Environnement, UMR 8587, Evry, France 5Faculté de Pharmacie, Université Paris-Sud, EA4530, Châtenay-Malabry, France 6Hôpital Antoine Béclère AP-HP, Service de Biochimie, Clamart, France Received September 2, 2013 Revised October 31, 2013 Accepted October 31, 2013 Research Article Derivatization strategies for CE-LIF analysis of biomarkers: Toward a clinical diagnostic of familial transthyretin amyloidosis We report three derivatization strategies for CE analysis with LIF detection (CE-LIF) of two synthetic peptides mimicking the wild and mutated fragments of interest for the diagnosis of familial transthyretin amyloidosis. The precapillary derivatization of the peptides with three optical tags, 5-carboxytetramethylrhodamin succinimidyl es- ter (TAMRA-SE), naphtalene-2,3-dicarboxyaldehyde (NDA), and 3-(2-furoyl)quinoline-2- carboxyaldehyde (FQ) has been investigated by CE-LIF detection and MS. Results pro- vide evidence that high reaction yields have been reached whereas the multitagging phe- nomenon has occurred for both NDA and TAMRA-SE labeling procedures. The deriva- tization and electrokinetic separation of a mixture of the two peptides of interest for the pathology diagnosis (22-aa peptides that differ only from one amino acid) were achieved using both approaches. The highest resolution with a value of 2.5 was obtained with TAMRA-SE labeled derivatives whereas NDA gave the best detection sensitivity (LOD of 2.5 �M). The validation of the developed methods showed a good linearity (R � 0.997) between the peak area of the labeled derivatives and the peptide concentration for both NDA and FQ labeling procedures. The intraday RSDs of A and the migration times were less than 3.8 and 2.2%, respectively. Keywords: CE / Familial transthyretin amyloidosis / LIF / MS / Peptide labeling DOI 10.1002/elps.201300426 � Additional supporting information may be found in the online version of thisarticle at the publisher’s web-site 1 Introduction CE-LIF detection represents one of the most sensitive tech- niques for the quantitative analysis of proteins and peptides. In the field of clinical diagnosis and clinical chemical anal- ysis, this technique offers also a good separation power and is therefore particularly adapted to analyze biological sam- ples, where micro- and nanomolar protein concentrations Correspondence: Dr. Isabelle Le Potier, Faculté de Pharmacie, In- stitut Galien Paris-Sud, Université Paris-Sud, CNRS UMR 8612, 5 rue Jean-Baptiste Clément, F-92290 Châtenay-Malabry, France E-mail: [email protected] Fax: +33-1-46-83-59-44 Abbreviations: A, peak area; ATTR, familial transthyretin amyloidosis; FQ, 3-(2-furoyl)quinoline-2-carboxyaldehyde; HCN, cyanhydric acid; I, ionic strength; KCN, potas- sium cyanide; MP, mutated peptide; NDA, naphtalene-2, 3-dicarboxyaldehyde; TAMRA-SE, 5-carboxytetramethylrho- damin succinimidyl ester; Tm, migration time; TTR, trans- thyretin; WP, wild peptide are common [1, 2]. However, many peptides and proteins do not exhibit a sufficient native fluorescence due to the lack or low quantity of amino acids exhibiting UV-absorbing moi- eties (i.e., Trp, Tyr, and/or Phe residues). To carry out such sensitive CE-LIF analysis, the common approach consists in the covalent derivatization of the amino acid residues with a fluorescent or fluorogenic dye. However, the difficulty to achieve such reaction with a high yield and in a reproducible manner is one of the bottlenecks of this approach. Moreover, derivatization can complicate the electrophoregrams because of the production of multiple side products from the dye and also various forms of derivatized analytes bearing a different number of covalently attached tags (multiple labeling) [3]. Among all the covalent dyes reported for proteins and peptides, most of them can react with primary or secondary amines of amino acids or with thiol groups of cysteine residues [3,4]. Fluorogenic dyes such as naphtalene- 2,3-dicarboxyaldehyde (NDA), 3-(2-furoyl)quinoline-2- carboxyaldehyde (FQ), and 3-(4-carboxybenzoyl)quinolone-2- carboxyaldehyde (CBQCA) are widely used for polypeptides LIF detection in the visible region and LODs in the �M–pM range have been reported [5–9]. These nonfluorescent C© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com Electrophoresis 2014, 35, 1050–1059 CE and CEC 1051 reagents form a fluorescent product in the presence of a nucleophile (CN−) after reacting with amino func- tional group [10, 11] and the stability of the resulting fluorescent isoindole derivatives is much greater than ortho- phthaldehyde derivatives [12, 13]. Besides these reagents, 4-fluoro-7-nitrobenzofurazan (NBD-F) and 4-chloro-7- nitrobenzofurazan (NBD-Cl) are also popular fluorogenic labeling reagents reacting within a few minutes under mild conditions with amino functional groups [14–16]. Although they also react slowly to form secondary fluorescent prod- ucts [17, 18], their use in place of a fluorescent dye are attractive because it reduces the background signal by several orders of magnitude. Many reports describe also the use of fluorescent dyes such as FITC [19–21] and rhodamine-based dyes [21–23] for LIF detection in the 400–600 nm wavelength range to get near nM LODs. These latter compounds are more reactive than FITC, especially when the dye is activated by the succinimidyl ester group which can react efficiently with unprotonated amines to form a stable amide bond [24,25]. As an example, 5-carboxytetramethylrhodamine succinimidyl ester (TAMRA-SE) labeling of antibodies has been successfully used in SDS-CGE to obtain LOD as sensitive as silver staining in SDS-PAGE slab gels [22]. Familial transthyretin amyloidosis (ATTR) is an auto- somic dominant degenerative disease characterized by the formation of amyloid fibril deposits, mainly composed of transthyretin (TTR), in different organs and tissues such as peripherical nerves, autonomic nervous system, heart, or liver. These amyloid deposits hinder organ functions and ultimately lead to their failure, the disease being usually fa- tal within 7–15 years after the appearance of the symptoms. ATTR is associated with single amino acid substitutions in TTR, a plasma protein (0.2–0.4 g/L) responsible for the trans- port of thyroxine and retinol-binding protein-vitamin A com- plex. TTR is mainly synthetized in the liver and circulates in the plasma as an homotetramer. Each subunit is composed of 127 amino acid residues. More than 80 TTR mutations have been described, the majority of them leading to amyloi- dosis. The main hypothesis for ATTR pathogenesis considers the tetramer instability favoring the dissociation to nonnative monomeric species with the ability to self-associate leading to soluble then insoluble aggregates and amyloid fibrils [26–28]. Several techniques such as IEF [29], radioimmunoas- say [30], ESI-MS [31,32], MALDI-TOF MS [32,33], and SELDI- TOF MS [34] for the analysis of intact TTR or of its proteolytic fragments are currently available to analyze the TTR mu- tations at the protein level. However, no analytical method currently provides a rapid and reliable quantitative assay to measure simultaneously the wild type and mutant-circulating forms. This would be of great value for clinical diagnosis but also for the elucidation of mechanisms of the disease and the monitoring of the evolution of the mutant TTR concentration during therapy protocols, that is, after liver transplantation. In this context, we focused our study on one of the most frequently reported TTR mutation in France, Thr49Ala ATTR [26, 28, 35, 36]. Our goal was to implement a label- ing procedure allowing to address the analytical challenge to separate and quantify by CE-LIF two synthetic peptides, that may be obtained after trypic digestion of the TTR once isolated from human serum. These 22-aa peptides that dif- fer from only one amino acid are representative of the wild (WP) and mutated (MP) forms of TTR (Fig. 1). Based on their attractive properties (described in previously published protocols, i.e., synthesis of fluorescent derivatives with high fluorescence quantum yield, compatibility of the excitation wavelength of the derivatized products with commercially available laser sources, stability of the fluorescent adducts, fast reaction rates, and low background signal for NDA and FQ), three different amine-reactive probes have been selected to label both peptides, two fluorogenic dyes, NDA [5,6,11,12], and FQ [7, 8, 10, 38–40], and one fluorescent dye, TAMRA- SE [21–25]. The performance of these dyes, based on the rate of reaction, the degree of substitution, and the detectability of fluorescent products have been compared using CE-LIF and MS. The labeling procedures were then applied to analyze a mixture of WP and MP by CE-LIF. The influence of various parameters such as nature, pH, and ionic strength (I) of the BGE, capillary length, and applied voltage on the WP/MP res- olution has been investigated. Finally, the optimized CE-LIF methods were validated in terms of linearity, reproducibility, and detection sensitivity. 2 Materials and methods 2.1 Chemicals Synthetic peptides GPS1344 (WP) and GPS1345 (MP) were purchased from Genepep (Prades-Le-Lez, France). NDA, FQ, and TAMRA-SE were purchased from Interchim (Montluçon, France). Potassium cyanide (KCN), HEPES, Tris, boric acid, TFA, and �-cyano-4-hydroxycinnamic acid were purchased from Sigma-Aldrich (Saint-Quentin- Fallavier, France). Methanol (HPLC grade), ACN (HPLC gra- dient grade), sodium hydroxide (NaOH) at 1 M, phosphoric acid, DMSO, and SDS were obtained from VWR (Fontenay- sous-Bois, France). All buffers and solutions were prepared using ultrapure water (Millipore, Bedford, MA). 2.2 Instrumentation CE experiments were performed using a Beckman-Coulter PA800 system equipped with an argon ion laser 488 nm ex- citation (Beckman-Coulter, Fullerton, CA). The CE was also coupled with two other laser sources, a 442 nm and a 532 nm excitation diode-pumped solid state laser (BFI Optilas, Evry, France) depending of the dye optical properties. For that, CE detector was adapted by using two different couples of excitation and emission filters (CVI Melles-Griott, Voisins- le-Bretonneux, France): (441.6 nm, 490 nm) and (532 nm, 575 nm) for the 442 nm and the 532 nm lasers, respectively. MALDI-TOF MS experiments were performed using a Perseptive Biosystems Voyager-DE Pro STR (Applied Biosystems/MDS Sciex, Foster City, CA). This instrument C© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com 1052 S. Korchane et al. Electrophoresis 2014, 35, 1050–1059 Figure 1. CE-LIF analysis of WP sample derivatized with (A) FQ, (B) NDA, and (C) TAMRA-SE. Conditions: fused-silica capillary 57 cm (total length), 50 cm (effective length) × 75 �m id; BGE, 40 mM HEPES buffer pH 8 for (A), 40 mM borate buffer pH 8 for (B), 50 mM phosphate buffer pH 7 for (C); applied voltage, +25 kV; temperature, 25°C; LIF detection �ex, 488 nm for (A), 442 nm for (B), 532 nm for (C); injection, 10 s at 3.5 kPa; sample, derivatized WP and dye blank (sample containing the derivatization mixture except WP) diluted fourfold prior to CE analysis for (B, C); derivatization conditions: peptide concentration, 0.2 mM for (A, B) and 0.6 mM for (C); WP/dye molar ratio, 1:100 for (A) and 1:40 for (B, C); incubation buffer, methanol/pH 9 borate buffer for (A, B) and DMSO/pH 10 borate buffer for (C); temperature, 55°C for (A) and 25°C for (B, C); time, 15 min for (A, B) and 60 min for (C); see Section 2.3 for other labeling conditions. was equipped with a nitrogen UV laser (� = 337 nm) pulsed at a 20 Hz frequency. The mass spectrometer was operated in the positive ion reflector mode with an accelerating po- tential of +20 kV and a grid percentage equal to 70%. Mass spectra were recorded with the laser intensity set just above the ionization threshold (2800 in arbitrary units), extraction delay was set to 450 ns and mass spectra were obtained by accumulation of 300 laser shots and processed using Data Explorer 4.0 software (Applied Biosystems). The instrument was calibrated using standard peptides mixtures provided by manufacturer. 2.2.1 CE CE experiments were performed using a fused-silica capil- lary, purchased from Phymep (Paris, France), with a total length varying from 57 to 97 cm and 75 �m of id. Samples were introduced into the capillary by hydrodynamic injection for 10 s under 3.5 kPa. Separation voltages were in the range of 20–30 kV. The capillary was thermostated between 15 and 25°C depending from the application. Each new fused-silica capillary was preconditioned with 0.1 M NaOH for 5 min, 1 M NaOH for 5 min, and ultrapure water for 5 min under 138 kPa. Before each electrophoretic run, the capillary was sequentially flushed with 0.1 M NaOH (3 min), then with wa- ter (3 min) and then equilibrated with the separation buffer for 3 min. A special rinsing method was optimized for NDA- labeled peptides separation: between each run, the capillary was flushed with 30 mM SDS (5 min), 1 M NaOH (2 min), and water (2 min). For TAMRA-SE-labeled derivatives sep- aration, the capillary was flushed between each run with a DMSO/water 50:50 v/v solution for 5 min. The working electrolytes (phosphate, borate, Tris-HCl, and HEPES buffers with pHs and Is ranging from 7 to 10 and from 20 to 100 mM, respectively) were prepared according to the PHoEBuS software’s procedures (Analis, Belgium) and filtered through a 0.22 �m pore size nylon membrane filter (VWR). 2.2.2 MS In order to remove the excess of salts and organic solvent, sample issued from incubation with NDA and TAMRA- SE was cleaned up by using ZipTipC18 tips (Millipore, C© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com Electrophoresis 2014, 35, 1050–1059 CE and CEC 1053 Billerica, MA) prior to MS analysis. MALDI-TOF MS analysis was achieved by mixing 1 �L of �-cyano-4-hydroxycinnamic acid matrix at 10 mg/mL in ACN/water 50:50 v/v 0.1% TFA with 1 �L of sample. 2.3 Derivatization procedures Stock solutions of individual peptides at a concentration of 5 × 10−2 M were prepared in ultrapure water. Working solu- tions of WP and MP (with concentrations ranging from 2.5 to 600 �M) were prepared by appropriate dilutions of stock solutions in sodium borate buffer with pHs and I ranging from 7.9 to 10 and from 40 to 70 mM, respectively. Stock solutions of FQ (10 mM) and NDA (4 mM) were prepared in methanol. TAMRA-SE was prepared in DMSO at a con- centration of 19 mM. The dye solutions were aliquoted (100 �L) and kept in the dark at −20°C. FQ aliquots containing 1 �mol of dye were dried then stored at −20°C. 2.3.1 NDA labeling As described previously [37], 50 �L of peptide solution in 70 mM borate buffer pH 9 were mixed with 100 �L of KCN solution prepared in 70 mM borate buffer pH 9 at a concen- tration of 4 mM, 100 �L of NDA stock solution, and 150 �L of 70 mM borate buffer pH 9. The mixture was shaken and kept at room temperature in the dark for 15 min. 2.3.2 FQ labeling Derivatization was performed by adding 100 �L of methanol to the dried FQ solution, 50 �L of peptide solution in 40 mM borate buffer pH 9, and 50 �L of KCN solution prepared in 40 mM borate buffer pH 9 at a concentration of 10 mM. The mixture was shaken and kept in the dark at 55°C for 15 min. 2.3.3 TAMRA-SE labeling A total of 17.2 �L of TAMRA-SE stock solution were added to 40 �L of peptide solution prepared in 40 mM borate buffer pH 10. The mixture was shaken and kept at room temperature in the dark for 60 min. 3 Results and discussion 3.1 Labeling investigation Three commercially tags reacting with amino groups (FQ, NDA, and TAMRA-SE) were selected to label the synthetic peptides WP and MP representative of the wild and mutated TTR (see Supporting Information Fig. S1 and Table S1 for re- action and dye physicochemical characteristics, respectively). As shown in Fig. 1, these peptides differ only by the nature of the N-term amino acid (Ala vs. Thr residue). Both pep- tides exhibit two NH2 groups (aliphatic ε-NH2 of C-term Lys and N-terminal Thr for WP or N-term Ala for MP) that can potentially be tagged. We first studied the ability of FQ to tag the two pep- tides of interest. Based on previously reported works dealing with amino acid and protein derivatization by FQ [38,39], we selected a methanolic/pH 9 borate buffer (1:1 v/v) as deriva- tization medium. This pH value appeared to be optimal to provide the best the labeling reaction yield. Indeed it has been reported that, if FQ reacts only with nonionized amino groups, secondary reactions between FQ and the nucleophile reagent KCN occur at pH values higher than 9.3 leading to several side products which complicate the CE profile [40]. The derivatization of WP was performed in methanol/40 mM borate buffer pH 9 with a peptide/FQ molar ratio of 1:100 and with an incubation time and temperature of 15 min and 55°C, respectively. The CE-LIF analysis of the sample showed a peak corresponding to the FQ-labeled WP derivative at 7.5 min using 40 mM HEPES buffer pH 8 as BGE. The assignment of the peak was done after the analysis of a FQ blank electropherogram. As expected and because of the fluorogenic character of the probe, only very few peaks corresponding to hydrolysis products and/or impurities were observed in the FQ blank electropherogram (Fig. 1A). We investigated several peptide/FQ molar ratio from 1:10 to 1:100. No significant difference between FQ-labeled WP peak areas (As) as a function of the dye concentration was observed. So, as it is known that a large excess of labeling reagent accelerates the reaction kinetics, the molar ratio of 1:100 was used for further experiments. Finally, the reaction was performed by varying the tem- perature and time incubation from 45 to 65°C and from 5 to 30 min, respectively, while keeping the other parameters constant. As expected, these experiments indicated that the higher is the temperature, the shorter is the reaction time. However, we also noticed at 65°C some degradation of the fluorescent isoindole derivative illustrated by the apparition of new peaks and a decrease of tagged peptide A (data not shown). Therefore, optimal conditions for the highest deriva- tization efficiency were achieved for the temperature reaction at 55°C for 15 min. To check the effectiveness of the reaction, the peptide was analyzed before and after derivatization with MALDI-TOF-MS (Supporting Information Fig. S2). MALDI mass spectra of both unlabeled peptides exhibit a major peak corresponding to the [M+H]+ ion at m/z 2456.0 and m/z 2426.2, for WP and MP, respectively. After FQ derivatization, an increment of 242.3 mass units was detected, resulting in a [M+H]+ ion at m/z 2698.3 and m/z 2668.5, corresponding to the mono-tagged derivative of WP and MP, respectively. In the same time, these results showed that the labeling of WP is complete in the optimized conditions. Moreover, no peak corresponding to a di-tagged derivative was present, in- dicating that one of the two available NH2 groups on peptide is probably less reactive in our experimental conditions. ESI- MS2 experiments revealed that reagent react exclusively with ε-NH2 of C-term Lys (data not shown). C© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com 1054 S. Korchane et al. Electrophoresis 2014, 35, 1050–1059 Figure 2. MALDI-TOF MS analysis of WP and MP before (A, C, respectively) and after (B, D, respec- tively) TAMRA-SE-labeling reaction. Derivatization condi- tions: peptide concentration, 0.6 mM; peptide/TAMRA-SE molar ratio, 1:40; buffer, DMSO/40 mM borate buffer pH 10; temperature, 25°C; time, 60 min; see Section 2.3 for other labeling condi- tions. MS conditions: 1 �L of �-cyano-4-hydroxycinnamic acid matrix at 10 mg/mL in ACN/water 50:50 v/v 0.1% TFA with 1 �L of sample cleaned up by using ZipTipC18 tips for (B, D). We then studied the NDA labeling of WP and MP. We employed the optimal derivatization conditions (pep- tide/NDA ratio of 1:40 with an incubation time of 15 min in 70 mM borate buffer pH 9 at room temperature) defined in a previous work [37]. Figure 1B shows the corresponding CE-LIF profile where the major peak observed at 7.3 min was attributed to NDA-labeled WP. MALDI-TOF analysis was used to check the efficiency of the labeling and to determine the labeling stoichiometry (Supporting Information Fig. S3). No peak corresponding to the unreacted peptides was observed in the mass spectrum of WP and MP solutions incubated with NDA, indicating that the labeling reaction was complete. Among the observed peaks in mass spectra, we quoted out peaks corresponding to the [M+H]+ of the mono-tagged derivative at m/z 2630.9 and m/z 2601.4 for WP and MP, respectively. Nevertheless, in- tense peaks corresponding to common H+/Na+ and Na+/K+ exchanges were also observed. In addition, extra peaks at m/z 2603.9 and 2574.4 were also observed, likely due to the loss of cyanhydric acid (HCN), according to the N-2-substituted-1- cyanobez-[f]-isoindole derivative structure (see Supporting In- formation Fig. S1). A second group of peaks that was difficult to assign for WP was detected whereas for MP, it may corre- spond to the di-tagged derivative and associated adducts. We noticed a difference between the experimental and expected m/z values. This discrepancy could be due to peptides pre- senting a loss of HCN (m/z 2749.4) without or with additional loss of H2O (m/z 2731.3). All together, these MS data indi- cate that the fluorescent derivatives are instable and rapidly lose an HCN molecule. Those conclusions were supported by ESI-IT MS analysis where the same forms were detected (data not shown). Accordingly, one might postulate that the minor peaks observed in the electropherogram of the NDA-labeled WP sample (Fig. 1B) could be attributed to these multiple side products from hydrolysis reactions and/or impurities of both reagents. The fact that di-tagged derivatives was not clearly C© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com Electrophoresis 2014, 35, 1050–1059 CE and CEC 1055 shown by MS for WP in contrast with MP might come from the nature of the N-term amino acid (Ala vs. Thr). Thus, the reactivity of this N-term amino group and/or the stability of the fluorescent derivatives could be different. One can thus hypothesize that the peptide conformation could be slightly affected by the replacement of an apolar amino acid (Ala in MP) by a polar one (Thr in WP) that decreases the accessibil- ity of the NH2 group of Thr residue toward the fluorescent tag. We finally studied TAMRA-SE as a tagging reagent. The derivatization of WP was first performed in 40 mM borate buffer pH 10 with a 40-fold molar excess of TAMRA-SE at room temperature for 1 h. In order to separate the dye in ex- cess and possible fluorescent TAMRA-SE side products from the TAMRA-SE-labeled WP, a 50 mM phosphate buffer pH 7 was employed as BGE for the CE-LIF analysis. Electro- pherogram displayed in Fig. 1C show a peak correspond- ing to the tagged WP at 6.2 min besides the TAMRA-SE peaks. However, minor peaks that could not be attributed to TAMRA-SE and dye impurities were also detected after 6.2 min suggesting that several fluorescent adducts bear- ing a different number of covalently attached tags were also produced. MALDI-TOF MS analysis confirmed the deriva- tization reaction efficiency since no peak corresponding to unreacted peptide was detected (Fig. 2). Moreover, for WP and MP, MS data indicated that not only mono- and di- tagged derivatives were obtained but also tri-tagged species (m/z 3693.5 and m/z 3663.4 for WP and WP, respectively). Since WP exhibits only two primary amine groups (aliphatic ε-NH2 of C-term Lys and N-terminal Thr), we assumed that TAMRA-SE reacted also with the imidazole moiety of the His residue even if, to best of our knowledge, this reaction was not previously reported in the literature. It is worth not- ing that the labeling of this residue is usually performed using reagents with an epoxy group which is a good leaving group [41–43] and it is also the case of the succinimidyl ester group. To further understand the reactions that take place dur- ing the TAMRA-SE derivatization, the kinetic of reaction in the experimental conditions described above was studied by MALDI-TOF MS. The proportion of mono-, di-, and tri-tagged derivatives as a function of time was monitored assuming a similar ionization efficiency of all the present species in the same conditions [44]. As shown in Fig. 3A, with WP as exam- ple, the amount of multitagged derivatives (di- and tri-tagged) increased up to 30 min and remained constant thereafter. The proportion of di-tagged derivative reached around 50% and was almost similar to the mono-tagged derivative one, while the proportion of tri-tagged derivative remained very low. The underivatized peptide was not detected by MS at any time of reaction. Therefore, considering that the stability of the derivatization reaction was assessed at around 30 min, an incubation time of 1 h, similar to previously reported val- ues [24], was selected as a sufficient time to ensure optimal reaction conditions. Since TAMRA-SE is expected to react with amino group also under mild basic pH [24], conditions known to prevent its hydrolysis, WP was incubated with the dye in 40 mM borate buffer 7.9. The proportion of the reaction products was monitored as a function of incubation time by MALDI- TOF MS. As shown in Fig. 3B, the amount of multitagged derivatives followed the same trend but the proportion of multitagged species was found higher in these slightly basic pH conditions than that obtained at pH 10. Indeed, only 25% of mono-tagged derivative was obtained whereas the propor- tion of di-tagged derivative was almost doubled. Moreover, we observed that the percentage of tri-tagged derivative was around 10% in these conditions. From these data, we hypothe- sized that a change of peptide conformation occurred between pH 7.9 and 10 facilitating the accessibility of the secondary amine of His residue at pH 7.9. As expected, the elec- trophoretic pattern of the WP sample derivatized at pH 7.9 showed, besides the TAMRA-SE peaks, numerous peaks with low intensity corresponding to the multiple labeled Figure 3. Percentage of mono-, di-, and tri-tagged TAMRA-SE derivatives normalized over the various forms of WP as a function of time obtained in 40 mM borate buffer (A) pH 10 and (B) pH 7.9 monitored by MALDI-TOF MS. (Without taking into account the unreacted peptide that has not been detected by MS for all incubation conditions.) MS and derivatization conditions are shown in Fig. 2 except pH buffer for (B). C© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com 1056 S. Korchane et al. Electrophoresis 2014, 35, 1050–1059 derivatives (data not shown). We therefore concluded that the best analytical procedure to limit the number of tagged deriva- tives and therefore to simplify the CE-LIF electropherograms was to perform the WP labeling at pH 10. Similar results were obtained with the MP peptide. The overall results, summarized in Supporting Infor- mation Table S1, showed that the targeted peptides can be successfully tagged by either FQ, NDA, and TAMRA-SE dyes in a mixture of organic solvent/borate buffer pH 9–10 as incubation medium and with a large excess of dye (at least 40-fold). High reaction yields can be reached in less than 15 min with FQ and NDA whereas 60 min is required for TAMRA-SE. NDA- and TAMRA-SE-labeling procedures are efficient at room temperature while a temperature of 55°C is required for FQ-labeling one. As expected, the degree of substitution of the targeted peptides depends on the reactivity and on the steric hindrance of the fluorescent tag but also on the peptides conformation. The ε-NH2 of C-term Lys of both peptides is a good target for the derivatization whereas the reactivity of N-term amino group and imidazole moiety of His residue is more sensitive to the reaction medium and the chemical nature of the tag. The reactivities of NDA and FQ are quite similar but the steric hindrance of the furyl substituent of the FQ probably prevents the synthesis of di-tagged species that have not been detected here. NDA can also react with the N-term amino acid leading to di-tagged derivatives. Based on the assumption that TAMRA-SE has also a good reactivity with the secondary amine of the imida- zole moiety of His, through the activation of the rhodamine dye by the succinimidyl ester group, a tri-tagged derivative was observed under these experimental conditions. 3.2 Optimization of wild and mutated fluorescent derivatives CE-LIF separation Separation of WP and MP fluorescent derivatives was quite challenging since the untagged peptides share the same pI (4.2) and they only differ by the N-term amino acid resulting in a mass difference of 30 Da (Fig. 1). The covalent attachment of a dye, and especially a neutral one such as FQ and NDA, leading to a mass increment of the analyte with corresponding suppression of the positively charge of tagged amino groups was not expected to facilitate the separation of the derivatives. Therefore, the electrophoretic difference between these species could rely only on their different hydrodynamic volume. A mixture of WP and MP was subjected to the three opti- mized labeling procedures then analyzed by CE-LIF. Several parameters including the pH and composition of the BGE, the length of the capillary, and the applied voltage were in- vestigated in order to optimize the separation of WP and MP fluorescent derivatives. Since the pH of the running buffer not only affects the EOF and the elecrophoretic mobility of the analytes but also the fluorescence emission, the influence of BGE pH and nature was first studied. In order to ensure a good buffer capacity, BGEs with I varying from 20 to 70 mM Figure 4. CE-LIF separation of a mixture of WP and MP after la- beling with (A) FQ, (B) NDA, and (C) TAMRA-SE. CE conditions: fused-silica capillary (75 �m id) with an effective length of 50 cm (A, C) and 90 cm for (B); BGE, 40 mM HEPES buffer pH 8 for (A), 40 mM borate buffer pH 8 for (B), 50 mM phosphate buffer pH 7 for (C); applied voltage, +20 kV for (A, C) and +30 kV for (B); temperature, 25°C for (A, C) and 15°C for (B); LIF detection �ex, 488 nm for (A), 442 nm for (B), 532 nm for (C); injection, 10 s at 3.5 kPa; sample, equimolar mixture of WP and MP at 0.2 mM for (A, B) and 0.6 mM for (C) diluted fourfold prior to CE injection for (B, C); derivatization conditions are shown in Fig. 1. were prepared either with phosphate (pH 7–7.5), borate (pH 8–10), Tris-HCl, and HEPES (pH 8–8.5) buffers. It was found that FQ-labeled WP and MP were com- pletely unresolved at pHs 7–7.5. BGE containing only borate with pH ranging from 8 to 10 or Tris-HCl pH 8–8.5 led also to complete co-migration of the analytes of interest. In con- trast, the use of HEPES buffer as BGE allowed the separation of the two derivatives, the resolution (Rs) being markedly higher at pH 8 than at pH 8.5. I of HEPES buffer pH 8 was then varied from 40 to 70 mM to determine the optimal con- ditions for separating the two derivatives. As expected, the migration time (Tm) increased with the increase of I but Rs remained constant. As illustrated in Fig. 4A, under optimal BGE conditions (40 mM HEPES buffer pH 8), the separation of the two derivatives could be achieved within 11 min with a C© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com Electrophoresis 2014, 35, 1050–1059 CE and CEC 1057 satisfactory resolution of 1.5 and a theoretical plate number of ca. 27 500. The electrophoretic behavior of NDA-labeled derivatives was quite similar to FQ derivatives ones: a co-migration of the wild and mutated derivatives was observed in both pH ranges 7–7.5 and 9–10, whatever the nature of the electrolyte buffer, whereas the analyte peaks could be partially resolved at pH 8–8.5. However, in this case, only borate buffer allowed the separation of the two derivatives while HEPES buffer failed to discriminate them. Again, these results point out the importance of the chemical nature of the buffer beyond its pH and I on the electrophoretic behavior of the analytes through its impact on both the EOF and the analyte solvation and therefore on the resolution between the derivatives owing very close structure. Attempts were made to optimize Rs by varying the pH and I from 8 to 8.5 and from 20 to 70 mM, respectively. However, under optimal conditions (pH 8 and I = 40 mM), the resolution remained quite poor (Rs � 0.9). Thus, in order to achieve a full separation, the effective capillary length was increased from 50 to 90 cm. Rs was improved to 1.7 and then to 1.9 by varying simultaneously the applied voltage and the temperature (Supporting Information Table S2). As displayed in Fig. 4B, the optimized conditions enabled the complete separation of the two derivatives, and despite the long analysis time, the number of theoretical plates (ca. 27 200) was similar to that obtained for the CE-LIF method with FQ labeling. Compared to previous optimizations described above, the analysis of TAMRA-SE derivatives was easier and baseline separation could be achieved regardless of run buffer pH. Moreover, it was found that 50 mM phosphate buffer pH 7 was the best BGE to separate the derivatives peaks from the two major peaks produced by the dye in excess. As illustrated in Fig. 4C, under optimal BGE conditions, the separation of the two derivatives could be achieved within 10 min with a complete baseline resolution of 2.5 and a theoretical plate number of ca. 250 000. 3.3 Analytical performances of CE-LIF methods The validation of the developed CE-LIF methods was done using the optimized derivatization and separation conditions as described above. The linearity of the methods based on NDA and FQ labeling was studied by performing the derivati- zation of WP solutions in a micromolar concentration range (all WP concentrations reported in this study were the initial peptide concentration in the sample solution prior to deriva- tization). The results of linear regression analysis, calculated by plotting the A of the tagged derivative versus the molar concentration, are presented in Table 1. For both methods, the correlation coefficients were higher than 0.997 indicating good linearity. The intraday repeatability was studied through five consecutive analyses of a 200 �M WP solution incubated with FQ and NDA probes and a 600 �M WP solution for TAMRA-SE. The interday reproducibility was determined by injecting three WP working solutions over a period of 3 con- secutive days. For FQ, Tm and As were highly reproducible between runs allowing a straightforward identification of peaks (Table 1). Conversely, the Tm and A RSD values ob- tained for the NDA- and TAMRA-labeled WP were quite high. These poor performances were attributed to the detrimental effects of the residual adsorption of the dyes or their fluores- cent adducts on the inner fused-silica capillary wall resulting in a highly variable EOF. To overcome this problem, a special washing procedure including a capillary flush with a 30 mM SDS solution or a DMSO/ultrapure water 50:50 v/v solution has been done before each analysis of NDA- and TAMRA- SE-labeled derivatives, respectively. In these conditions, an improvement of the RSDs was observed for both procedures (Table 1). In all cases, the interday reproducibility was relatively good (Tm and A RSD � 4.8 and 7.6%, respectively). The accuracy was studied by comparing the mean of the observed values, calculated from three assays of a 200 �M so- lution, with the theoretical value. The average concentrations observed are included in the range 200 �M ± 5% for both Table 1. Calibration, reproducibility, accuracy, and detection sensitivity of the labeled WP Probe Linear Calibration Correlation Intraday Interday Accuracy LOQ LOD range curve coefficient repeatability reproducibility (%)d) (�M) (�M) (�M) y = ax + ba) (R) (%)b) (%)c) a (×104) b (×104) A Tm A Tm FQ 100–500 0.5 6.3 0.997 3.7 0.3 7.1 4.8 −1.5 100 50 NDAe) 5–600 59.1 −391.5 0.999 2.6 0.8 7.6 3.7 0.2 5 2.5 TAMRA-SEf) – – – – 3.8 2.2 7.4 4.0 – 600 300 CE conditions as in Fig. 4. – : not determined; TAMRA-SE data concerning the linear regression analysis are not provided due to the poor LOD observed. a) Y: peak area (RFU.min); X: concentration of WP (�M). b) RSDs of peak areas (A) and Tms were calculated from five consecutive runs. c) RSDs of A and Tm were calculated by injecting three working solutions over a three day period (n = 9). d) Accuracy was calculated from three assays of a 200 �M solution according to the following equation: (mean of calculated concentration− theoretical concentration value)/(theoretical concentration value)× 100. e) Between each run, the capillary was flushed with 30 mM SDS (5 min), 1 M NaOH (2 min), and water (2 min). f) Between each run, the capillary was flushed with a 50:50 v/v DMSO/water solution for 5 min. C© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com 1058 S. Korchane et al. Electrophoresis 2014, 35, 1050–1059 FQ- and NDA-labeling methods. The detection sensitivity was studied by calculating the LOQ and LOD corresponding to a S/N ratio of 10 and 3, respectively (Table 1). The lowest LOD value was obtained for NDA-labeled WP (2.5 �M). Derivati- zation of WP with this dye results in a 20-fold improvement compared to FQ. The LOD and LOQ values of TAMRA-SE labeled WP are by far higher than those obtained with the two other probes. Since fluorescence emission is known to be dependent on the medium (polarity, pH . . . ), we checked that this poor LOD was not linked to the neutral pH of the BGE employed in the case of TAMRA-SE derivatives analysis. Similar LOD and LOQ values were calculated by performing the analysis of TAMRA-SE-labeled WP in borate buffer pH 8. Therefore, since MS data showed that the derivatiza- tion reaction was complete and since rhodamine derivatives are generally reported to be highly fluorescent, it cannot be ruled out that this unexpected poor sensitivity may be linked to a poor technical performance of the 532 nm external laser source which was coupled to the commercial CE instrument. These results show that both methods based on FQ and NDA labeling are able to provide reliable quantitative assay of WP. Although the detection sensitivity was not as high as expected with LIF detection, these methods could be fully compatible with analyses of real samples. Indeed, given the high physiological level of TTR in human serum [26], a few hundred microliters of serum sample would be sufficient to perform the sample pretreatment (SDS-PAGE purification of TTR followed by in-gel tryptic digestion of TTR band) and subsequent CE-LIF analysis. Moreover, the detection sensi- tivity of the CE-LIF methods could be enhanced if required by increasing the hydrodynamic injection time or using elec- trokinetic preconcentration. 4 Concluding remarks We propose analytical approaches for a simultaneous and fully quantitative analysis of two peptides of interest for the diagnosis of Thr49Ala ATTR based on precapillary derivatiza- tion of peptides and subsequent CE-LIF analysis. For achiev- ing good derivatization efficiency, good resolution, and high reproducibility, we chose and optimized labeling incubation conditions, BGE conditions, and in-between-run rinsings. The study of the analytical performances and in particular detection sensitivity showed that both NDA- and FQ-based methods can be in principle applied to authentic biological samples such as human serum. FQ derivatization brings the possibility to have a satisfactory resolution in much shorter analysis time while NDA allows to achieve the higher sen- sitivity. Current efforts are focused on sample pretreatment (TTR extraction from serum and in-gel proteolytic digestion) and preliminary results are quite promising. We are indebted to David Adams and Anne-Marie Haghiri- Gosnet for fruitful discussions. This research was supported by french ANR (Agence Nationale de la Recherche) in the context of P2N “DIMIPOLE” project. The authors have declared no conflict of interest. 5 References [1] Mesbah, K., Verpillot, R., De l’Escaille, F., Falmagne, J. B., Taverna, M., Methods Mol. Biol. 2013, 984, 167–190. [2] Szökő, E., Tabi, T., J. Pharm. Biomed. Anal. 2010, 53, 1180–1192. [3] Garcı́a-Campaña, A. M., Taverna, M., Fabre, H., Elec- trophoresis 2007, 28, 208–232. [4] Lacroix, M., Poinsot, V., Fournier, C., Couderc, F., Elec- trophoresis 2005, 26, 2608–2621. [5] Frerichs, V. A., Herrmann, J. K., Aguirre, A., Colon, L. A., Microchem. J. 2004, 78, 135–142. [6] Sauvinet, V., Parrot, S., Benturquia, N., Bravo-Moraton, E., Renaud, B., Denoroy, L., Electrophoresis 2003, 24, 3187–3196. [7] Pinto, D. M., Arriaga, E. A., Craig, D., Angelova, J., Sharma, N., Ahmadzadeh, H., Dovichi, N. J., Anal. Chem. 1997, 69, 3015–3021. [8] Michels, D. A., Brady, L. J., Guo, A., Balland, A., Anal. Chem. 2007, 79, 5963–5971. [9] Novatchev, N., Holzgrabe, U., J. Pharm. Biomed. Anal. 2002, 28, 475–486. [10] Beale, S. C., Hsieh, Y. Z., Wiesler, D., Novotny, M., J. Chromatogr. 1990, 499, 579–587. [11] Roach, M. C., Harmony, M. D., Anal. Chem. 1987, 59, 411–415. [12] de Montigny, P., Stobaugh, J. F., Givens, R. S., Carlson, R. G., Srinivasachar, K., Sternson, L. A., Higuchi, T., Anal. Chem. 1987, 59, 1096–1101. [13] Ndube, N., van der Westhuizen, L., Green, I. R., Shephard, G. S., J. Chromatogr. B 2011, 879, 2239– 2243. [14] Solinova, V., Kasicka, V., Koval, D., Barth, T., Ciencialová, A, Záková, L., J. Chromatogr. B 2004, 808, 75–82. [15] Wise, E. T., Singh, N., Hogan, B. L., J. Chromatogr. A 1996, 746, 109–121. [16] Zhang, H., Le Potier, I., Smadja, C., Zhang, J., Taverna, M., Anal. Bioanal. Chem. 2006, 386, 1387–1394. [17] Arriaga, E. A., Zhang, Y., Dovichi, N. J., Anal. Chim. Acta 1995, 299, 319–326. [18] Hu, S., Li, P. C. H., J. Chromatogr. A 2000, 876, 183–191. [19] Huang, Y., Duan, J., Chen, Q., Chen, G., Electrophoresis 2004, 25, 1051–1057. [20] Little, M. J., Paquette, D. M., Harvey, M. D., Banks, P. R., Anal. Chim. Acta 1997, 339, 279–288. [21] Rodat, A., Kalck, F., Poinsot, V., Feurer, B., Couderc, F., Electrophoresis 2008, 29, 740–746. [22] Hunt, G., Nashabeh, W., Anal. Chem. 1999, 71, 2390–2397. [23] Verbeck, G. F., Beale, S. C., J. Microcol. Sep. 1999, 11, 708–715. [24] Brunner, A., Minamitake, Y., Göpferich, A., Eur. J. Pharm. Biopharm. 1998, 45, 265–273. C© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com Electrophoresis 2014, 35, 1050–1059 CE and CEC 1059 [25] Liu, K.,Wang, H., Bai, J.,Wang, L.,Anal. Chim. Acta 2008, 622, 169–174. [26] Connors, L. H., Lim, A., Prokaeva, T., Roskens, V. A., Costello, C. E., Amyloid: J. Protein Folding Disord. 2003, 10, 160–184. [27] Ando, Y., Ueda, M., Front. Biosci. 2008, 13, 5548– 5558. [28] Planté-Bordeneuve, V., Carayol, J., Ferreira, A., Adams, D., Clerget-Derpoux, F., Misrahi, M., Said, G., Bonaı̈ti- Pellié, C., J. Med. Genet. 2003, 40, 120e. [29] Connors, L. H., Ericsson, T., Skare, J., Jones, L. A., Lewis, W. D., Skinner, M., Biochim. Biophys. Acta 1998, 1407, 185–192. [30] Nakazato, M., Kurihara, T., Matsukura, S., Kangawa, K., Matsuo, H., J. Clin. Invest. 1986, 77, 1699–1703. [31] Yamashita, T., Ando, Y., Bernt-Suhr, O., Nakamura, M., Sakashita, N., Ohlson, P. I., Terazaki, H., Obayashi, K., Uchino, M., Ando, M., J. Neurol. Sci. 2000, 173, 154–159. [32] Théberge, R., Connors, L., Skinner, M., Skare, J., Costello, C. E., Anal. Chem. 1999, 71, 452–459. [33] Haraoka, K., Ando, Y., Ando, E., Sun, X., Nakamura, M., Terazaki, H., Misumi, S., Tanoue, Y., Tajiri, T., Shoji, S., Ishizaki, T., Okabe, H., Tanihara, H., Amyloid 2002, 9, 247–251. [34] Ueda, M., Misumi, Y., Mizuguchi, M., Nakamura, M., Ya- mashita, T., Sekijima, Y, Ota, K., Shinriki, S., Jono, H., Ikeda, S., Suhr, O. B., Ando, Y., Clin. Chemi. 2009, 55, 1223–1227. [35] Almeida, M. R., Ferlini, A., Forabosco, A., Gawinow- icz, M., Costa, P. P., Salvi, F., Plasmati, R., Tassinari, C. A., Atland, K., Saraiva, M. J., Hum. Mutat. 1992, 1, 211–215. [36] Benson, M. D. II, Julien, J., Leipnieks, J., Zelden- rust, S., Benson, M. D., J. Med. Genet. 1993, 30, 117–119. [37] Faure, M., Korchane, S., Le Potier, I., Pallandre, A., Deslouis, C., Haghiri-Gosnet A. M., Gamby, J., Talanta 2013, 116, 8–13. [38] Wojcik, R., Swearingen, K. E., Dickerson, J. A., Turner, E. H., Ramsay, L. M., Dovichi, N. J., J. Chromatogr. A 2008, 1194, 243–248. [39] Warren, C. R., Soil Biol. Biochem. 2008, 40, 916–923. [40] Veledo, M. T., de Frutos, M., Diez-Masa, J. C., J. Chro- matogr. A 2005, 1079, 335–343. [41] Li, X., Ma, H., Dong, S., Duan, X., Liang, S., Talanta 2004, 62, 367–371. [42] Li, X., Ma, H., Nie, L., Sun, M., Xiong, S., Anal. Chim. Acta. 2004, 515, 255–260. [43] Zamora, R., Alaiz, M., Hidalgo, F. J., Chem. Res. Toxicol. 1999, 12, 654–660. [44] Leitner, A., Lindner, W., J. Mass Spectrom. 2003, 38, 891–899. C© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.electrophoresis-journal.com