Ot

Ga

Ub

c

a

ARR1AA

KCOBN

1

fdaatb[cimprtd

P

h0

Journal of Chromatography A, 1376 (2015) 159–166

Contents lists available at ScienceDirect

Journal of Chromatography A

j o ur na l ho me page: www.elsev ier .com/ locate /chroma

n-line capillary electrophoresis-based enzymatic methodology forhe study of polymer-drug conjugates�

. Coussota,∗,1, Y. Ladnera,1, C. Bayarta, C. Fayeb, V. Vigierc, C. Perrina

Institut des Biomolécules Max Mousseron-IBMM, Centre National de la Recherche Scientifique, Université de Montpellier 1, Université de Montpellier 2,nité Mixte de Recherche 5247, Faculté de Pharmacie, F-34093, Montpellier cedex 5, FranceColcom, Cap Alpha, Avenue de l’Europe, F-34830, Clapiers, FranceGemacbio, Lieu-dit Berganton, F-33127, Saint-Jean d’Illac, France

r t i c l e i n f o

rticle history:eceived 30 July 2014eceived in revised form7 November 2014ccepted 8 December 2014vailable online 12 December 2014

eywords:

a b s t r a c t

This work aims at studying the potentialities of an on-line capillary electrophoresis (CE)-based diges-tion methodology for evaluating polymer–drug conjugates degradability in the presence of free trypsin(in-solution digestion). A sandwich plugs injection scheme with transverse diffusion of laminar profile(TDLFP) mode was used to achieve on-line digestions. Electrophoretic separation conditions were estab-lished using poly-l-Lysine (PLL) as reference substrate. Comparison with off-line digestion was carriedout to demonstrate the feasibility of the proposed methodology. The applicability of the on-line CE-baseddigestion methodology was evaluated for two PLL-drug conjugates and for the four first generations of

apillary electrophoresisn-line trypsin digestioniodegradable polymersano-enzyme reactor

dendrigraft of lysine (DGL). Different electrophoretic profiles presenting the formation of di, tri, andtetralysine were observed for PLL-drug and DGL. These findings are in good agreement with the nature ofthe linker used to link the drug to PLL structure and the predicted degradability of DGL. The present on-linemethodology applicability was also successfully proven for protein conjugates hydrolysis. In summary,the described methodology provides a powerful tool for the rapid study of biodegradable polymers.

. Introduction

In the pharmaceutical industry, biodegradable polymers are theocus of attention particularly in the treatment of autoimmuneiseases and cancer [1–4]. Due to the biodegradable polymersrchitecture, these macromolecules exhibit specific drug trappingnd releasing properties with expected few side-effects. Conjuga-ion of a low molecular weight drug to biodegradable polymers haseen also proposed to improve the therapeutic index of small drug4]. The resulting polymeric prodrug (called also polymer-drugonjugate PDC) helps in both enhancing the drug bioavailabil-ty and in preserving the drug activity during circulation, with

inored immunological body response. In this context, PDCs com-osed of �- amino acids (� –AAs) are considered as promising

epresentatives of synthetic biopolymers since after biodegrada-ion the release products are essentially �–AAs, peptides and theirerivatives. Thus, these poly(aminoacid)s-drug conjugates that are

� Presented at the 41st International Symposium on High Performance Liquidhase Separations - HPLC 2014, 10 - 15 May 2014, New Orleans, Louisiana, USA.∗ Corresponding author. Tel.: +00 33 4 11 75 95 93; fax: +00 33 4 11 75 96 38.

E-mail address: gaelle.coussot@univ-montp1.fr (G. Coussot).1 Both authors contributed equally to this work.

ttp://dx.doi.org/10.1016/j.chroma.2014.12.029021-9673/© 2014 Elsevier B.V. All rights reserved.

© 2014 Elsevier B.V. All rights reserved.

1/ synthesized in a relatively simple manner 2/ cost-effective and3/ present entire elimination post-treatment, have a considerablepotential for drug delivery [1–4].

In this article, molecules of study were poly-l-Lysine-drug con-jugates with linear (poly-l-Lysine: PLL) or dendritic architectures.The corresponding PDCs vary with respect to molecular weight,drug grafting, and resulting conformation. In addition, after absorp-tion, the stability of the PDC and the rate of drug release directlyimpact its efficacy and safety. Consequently, it is a challenging taskto manufacture reproducible PDC lots on large scale. Knowledgeabout drug-related impurities, PDC heterogeneity between lots,and percentage of drug grafting are requested to allow market-ing authorization of the PDCs [5]. However, due to the inherentstructural complexity of the PDCs, these characteristics remainchallenging to obtain. There is a need for the development ofvalidated techniques for the analysis of these multi-componentmixtures and for establishing preclinical safety of PDCs. In choos-ing analytical techniques for quality control of PDCs, several factorsare especially important: the methodology should be simple, rapidand developed with commercially available instruments.

A possible approach for controlling PDC composition and druggrafting consists in performing an enzymatic hydrolysis of thePDC to produce a mixture of surrogate peptides and amino acids(named digest) that are further analyzed by appropriate separation

1 atogr

m(otwpaabterlcabCpdtnmbbtq

oliEtaltesmsdsCttop

2

2

lcwh

Gc

ahb(o

60 G. Coussot et al. / J. Chrom

ethods. To date, the enzymatic digestion step of polypeptidesto a lesser extend for poly(amino acids)) has been performedff-line with protease using mostly two-dimensional gel elec-rophoresis and high performance liquid chromatography (HPLC)ith mass spectrometry (MS) in order to structurally characterizeolypeptide biopharmaceuticals or to apply proteomic bottom-uppproaches [6]. However, the off-line digestion step is both timend reactants consuming, and often show cross contaminationsy endogenous compounds that further impair the quality ofhe analysis [7]. One first alternative could be to use capillarylectrophoresis (CE) in combination with immobilized enzymeeactors (IMERs) to achieve on-line enzyme assay [8] but this isimited due to the loss of enzyme activity upon immobilization, theost and the restricted particles size of the support commerciallyvailable [7]. A second alternative approach has been proposedy Li et al. This approach consists in the use of two-dimensionalE system employing a replaceable monolithic reactor for on-linerotein digestion [9]. This two-dimensional CE system with MSetection helped to improve protein identification scores. Unfor-unately, this system has the same disadvantages of IMERs and isot commercially available. Consequently, this cannot be imple-ented for PDCs quality control. All of the above approaches have

een established for qualitative studies aiming to identify proteinsy MS via peptide fragment identification after proteolysis. Onhe opposite, quality control of PDC requires both qualitative anduantitative studies to estimate lot purity and drug grafting.

In this paper, we present an automated on-line CE methodol-gy employing enzyme in solution for hydrolyzing dendrigraft ofysine (DGL) and PLL-drug conjugates, in which the silica capillarys used both as nano-enzymatic reactor and separation support.nzymes in solution are in their most active format and are easyo handle. Moreover, conventional on-line enzyme assays are usu-lly developed for the screening of pharmaceuticals (drugs havingow-molecular-weight), the evaluation of enzyme activity andhe studying of drug–drug interactions. In this paper, the on-linenzyme assays are applied to DGL and PLL-drug conjugates (sub-trates having molecular weights above 5 kDa) using trypsin asodel enzyme to study PDCs proteolysis. The reactants are injected

equentially; the mixing of reactant plugs is done by transverseiffusion, which is the first application in case of proteolysis. Theeparation is done by applying an electric field using conventionalE instrument. The developed methodology is rapid, cost- andime-saving. We found out that DGL and PLL-drug conjugates pro-eolysis is highly repeatable. In addition, we demonstrated that thisn-line trypsin digestion CE methodology can be easily applied torotein conjugates.

. Materials and methods

.1. Chemicals

Native Poly-l-Lysine (PLL, 15–30 kDa), PLL-drug conjugatesinking taurine (Tau), Bovine Serum Albumin (BSA) and itsonjugate bearing indole-3-acetic acid (IAA) named BSA-IAAere supplied in lyophilized form by GemacBio (France,ttp://www.gemacpharma.com).

Dendrigrafts of poly l-Lysine (DGL, generations G1, G2, G3,4), were gratefully provided by Colcom (France, http://www.olcom.eu) and used as received.

Sequencing grade modified trypsin, citric acid (≥ 99.5%), 6-minohexanoic acid (ε-amino caproic acid, ≥ 98.5%), sodium

ydroxide (≥ 98%), hydrochloride acid (≥ 37%), ammonium bicar-onate (≥ 99.5%), phosphoric acid (≥ 85%), dilysine (Lys2), trilysineLys3), tetralysine (Lys4) and pentalysine (Lys5), and poly(ethylenexide) (PEO) of molecular weight 1,000,000 were purchased from

. A 1376 (2015) 159–166

Sigma–Aldrich (France). Imidazole (≥ 99%) was bought from AcrosOrganics (France). All aqueous solutions were prepared with aMilli-Q purification system from Millipore (France). The fused silicacapillaries were purchased from Polymicro Technologies (UnitedStates).

2.2. Capillary electrophoresis (CE)

A PA800 system (Beckman, Fullerton, USA) was used for CE anal-ysis. The capillaries (i.d. 50 �m, total length 50 cm, length to thedetector 40 cm) were coated according to the standard proceduredescribed elsewhere [10]. Briefly, the capillaries were precondi-tioned with 1 M NaOH, for 20 min followed by 10 min flushing withMilli-Q water under 20 psi.

In between runs, the capillary was successively rinsed with1 N NaOH, Milli-Q water, 0.1 M HCl (5 min each under 20 psi), andfinally with coating solution for 10 min under 20 psi. The coatingsolution is composed of acidified polyethylene oxide (PEO) at 0.2%(m/v) in 0.1 M HCl. The capillary was thoroughly flushed (20 psi,5 min) with the background electrolyte (BGE) before introducingnanovolumes (plugs) of enzyme and its substrates. The injectionswere performed by hydrodynamic pressure at the anodic end of thecapillary.

The BGE was composed of a mixture of citric acid (27 mM) and�-amino caproic acid (207 mM) mixture at pH 5.0. Its ionic strength(I) was 75 mM. This buffer was used to separate lysine oligomersfrom others hydrolysis products, remaining polymeric substrates,and enzyme. For the study of trypsin digestion of BSA and BSA-IAA,a BGE composed of phosphate buffer (pH 2.5, I 50 mM) was used.

A voltage of 15 kV was applied to separate the substrate, theenzyme and the hydrolysis products. The capillary was ther-mostated at 25 ◦C during the whole procedure. UV detection wasperformed at 214 nm. 5 �L of imidazole solution at 0.25 g/L (or5.0 g/L) was added to 95 �L of the solution to be analyzed.

2.3. Off-line tryptic digestion

The off-line digestion procedure was composed of two steps:the substrate was first incubated with enzyme in a vial for a cer-tain period of time. Then several nL of the sample digest wereinjected at the inlet of the PEO-coated capillary. The substrates PLL,PLL-drug conjugates and DGLs were exactly prepared at 1 g/L inMilli-Q water. A trypsin stock solution was prepared by dissolv-ing 20 �g of trypsin in 100 �L of 5 mM HCl. 20 �L of trypsin stocksolution were added to the PDC solution to have an enzyme-to-substrate ratio (E/S) of 1/20 (w/w) in proteolytic buffer composedof 100 mM ammonium bicarbonate (pH 8.3, I 100 mM). Subsequentto the trypsin addition, the mixture was incubated in a dry bath at37 ◦C. The tryptic digestion was stopped by freezing the samplein liquid nitrogen. The frozen samples were stored at −20 ◦C for ashort period of time before analysis.

2.4. On-line tryptic digestion

On-line tryptic digestion was achieved in a PEO-coated cap-illary. The substrates were exactly prepared at 1 g/L in Milli-Qwater. A trypsin solution was prepared at 0.2 g/L. For proteins,aliquots were denatured at 80 ◦C for 10 min in micro-centrifugetubes prior to injection. All the substrates S (DGL and PLL deriva-tives, BSA and BSA-IAA), the trypsin (enzyme abbreviated by E),and the proteolytic buffer (symbolized by “PB”) were introducedhydrodynamically into the capillary. PB was composed of 100 mM

ammonium bicarbonate (pH 8.3, I 100 mM). Unless stated other-wise, volumes of S, E, and PB plugs were 6.65 nL, 3.56 nL and 1.42 nL,respectively. Reactants mixing and digestion inside the capillarywere achieved by transverse diffusion as described by Krylov group

G. Coussot et al. / J. Chromatogr. A 1376 (2015) 159–166 161

p pro

[s

3

oeodcapbItrt�

ttec

4

4

(tsfoavaud

pt

Fig. 1. Schematic representation of the on-line three-ste

11]. The incubation time was set at 15 min at 25 ◦C. Digests wereeparated by CE as described above.

. Calculations

The observed electrophoretic mobilities (�obs) values for lysineligomers were calculated for each separation using the followingquation: �obs = [(Lc × Ld)/(V × tm)], in where Lc is the total lengthf the capillary, Ld is the capillary length from injection site toetection window in cm, V is the voltage in volts applied across theapillary during the separation, and tm is the migration time of thenalyte in seconds. Imidazole was added as marker into the sam-le to control that the residual EOF was not varying significantlyetween runs and to calculate corrected areas for quantification.n our calculations, we considered �EOF as negligible due to neu-ral polymer wall coating in the capillary. For peak identifications,elative analyte mobilities �rel are then calculated by subtractinghe mobility of analyte �obs from the imidazole mobility value

imidazole.To evaluate dilysine, trilysine and tetralysine formation rate,

he analyte peak area was divided by its corresponding migrationime to obtain the corrected area. Relative corrected areas werexpressed as the ratio of the analyte corrected area to the imidazoleorrected area for the same run.

. Results and discussion

.1. Experimental setup

The on-line methodology is divided into three steps (Fig. 1):1) a filling of the capillary with the BGE followed by a sequen-ial injection of reactants plugs with a substrate (S) plug taken inandwich by two plugs of enzyme (E), the latter’s being separatedrom the BGE by proteolytic buffer (PB) plugs. Finally a last plugf BGE is added. The injection plug sequence can be formalizeds the following: BGE/PB/E/S/E/PB/BGE. Step (2) is based on trans-erse diffusion mixing mode during which no voltage is applied for

defined period of time. Step (3) concerns the digest separationsing conventional capillary zone electrophoresis (CZE) with UV

etection.

In-solution, tryptic digestions are commonly performed in aH range 7.5–8.5 using ammonium bicarbonate buffer [12] sincehis buffer is compatible with MALDI and ESI-MS analysis. Based

posed methodology to digest polymer-drug conjugates.

on these considerations, the present on-line study was developedusing 100 mM ammonium bicarbonate at pH 8.3. To demonstratethe feasibility of the on-line digestion of PDC, the concentration ofsubstrate solution was fixed at 1 g L-1, whereas the enzyme was pre-pared at 0.2 g L-1. The plug of substrate (S) was bracketed by plugs ofenzyme (E) to ensure substrate proteolysis whatever the apparentelectrophoretic mobility of substrate. E plugs are, on their oppositeside, in contact with proteolytic buffer (PB) to reach the optimal pHfor enzyme activity. The PB/E/S/E/PB plugs scheme was surroundedby BGE, which present different I and pH that ensures the properseparation of hydrolysis products from the substrate (S) and/or theenzyme (E). Thus, the methodology offers to run subsequently PDCsproteolysis and separation of the products in the same capillary andin an automated manner. To our knowledge, this is the first applica-tion of on-line digestion of polymer-drug conjugates by a proteasein CE by means of PB/E/S/E/PB injection scheme, diffusion mixingmode, and in PEO coated capillaries.

In 2009, Zeisbergerová et al. initiated the on-line trypsin diges-tion of proteins with electrophoretically-mediated microanalysis(EMMA) on fused silica capillary using the application of an electricfield to mix the reactant plugs [13]. The electrophoretic mixing wasnot suitable for PP studies since reactant electrophoretic mobilitiesshould be known to determine the order of plug injections [14]. Inthe present paper, the mixing of reactants by transverse diffusionof laminar profile (TDLFP) was preferred [11] since it offers the pos-sibility to mix reactants independently from their electrophoreticmobility. In addition, the TDLFP mixing using the application of anappropriate pressure allows the mixing of more than two sepa-rately injected reactants, what others mixing modes do not.

Consequently, during the step (2) mixing of S, E and PB plugsin the capillary were done by TDLFP mode. The pressure and injec-tion time used for the consecutive plugs injection were chosen togenerate needle like plugs facilitating penetration into the preced-ing reactants [15]. Sample injection pressures were set as 0.7 psifor S, 0.5 psi for E and 0.3 psi for S plug with 4 s, 3 s, and 2 s injec-tion time for S, E, and PB, respectively. The corresponding injectedreactants volumes were 6.65 nL, 3.56 nL and 1.42 nL for S, E, and PB,respectively.

TDLFP mixing was described to be fast (within few minutes)[11,15]. During our experimental conditions, the incubation time

was set to 15 min. Indeed, no differences in CE profiles of PLL digestshave been observed above 15 min reaction time (data not shown).

During the step (3), a citric (27 mM)/�-amino caproic acid(207 mM) BGE of pH 5.0 and 75 mM I was used to achieve the

162 G. Coussot et al. / J. Chromatogr. A 1376 (2015) 159–166

Fig. 2. A comparison of the electrophoretic profiles of poly-l-Lysine under its native form (a), after on-line tryptic digestion (triplicate analyses) with S( )/E(3.i

spgsosdAcCoatptaea

4

Pmbiu(Ptlp(sottbC

6.65 nL)/E(3.56 nL)/PB(1.42 nL) (b), after on-line tryptic digestion with S (6.65 nLnjected at 5 g/L (Section 2).

eparation of hydrolysis products of PLL, while a phosphate BGE ofH 2.5 and 50 mM I was used for the separation of protein conju-ates digests. In the present work, we also chose to coat the innerurface of the capillary by polyelectrolyte to insure repeatabilityf the injections between runs. Indeed, published data using fusedilica capillaries showed noticeable variations in entire protein andigest profiles [13], probably due to partial adsorption of proteins.

way to circumvent peptide and protein adsorption onto theapillary is to cover it by polymers or polyelectrolytes [16,17].onsequently, neutral polyethylene oxide (PEO) was used to coatur capillary. Electrophoretic (nature, pH and IS of BGE, voltage)nd biochemical (enzyme to substrate ratio, proteolysis buffer,emperature. . .) parameters set for the steps (2) & (3) have been ofarticular interest. In this article, we chose PLL as a model substrateo provide a first view of our methodology performances; thesere presented in the following sub-section. The correspondingxperiments demonstrated proofs of concept for our methodologyre illustrated in Fig. 1.

.2. On-line versus off-line trypsin digestion of PLL

PLL is the most well-known drug carrier. In this study, we usedLL with molecular weights ranging from 15 to 30 kDa. This poly-er has great potential as vector for the in vivo delivery of drugs

ecause it is readily biodegradable [18]. PLL has a polydispersityndex of 1.2, and analyses performed on this native PLL form gavenresolved and diffuse peaks (around 22 min of migration time)Fig. 2A). This should not be the case if analyses are done afterLL proteolysis (enzymatic hydrolysis of peptide bonds). Indeed,he resulting digest mixture should be simpler and composed ofysine oligomers. Moreover, if the digestion process goes com-letely, the digest should be reduced to dilysine (Lys2), trilysineLys3) and tetralysine (Lys4). This is the reason why PLL was cho-en as reference substrate with trypsin as model enzyme. The studyf PLL proteolysis will first permit to evaluate the performances of

he proposed methodology in terms of repeatability of the diges-ion step and, in a second step to prove its ability to discriminateetween PLL and PLL-drug conjugates based on the differences inE profiles.

56 nL)/PB(4.75 nL) (c). Trypsin peak appeared about 35 min. Imidazole marker is

As expected, after running our CE on-line methodology on PLL,our first results demonstrated that the tryptic digest mixture wascomposed of Lys2, Lys3, Lys4, and a few amount of Lys5 (Fig. 2B). Toconfirm migration times of E and S, standards have been injectedseparately. For S and E standards, the injection schemes werePB/W/S/W/PB and PB/E/W/E/PB, respectively (W being water plug).Peaks identification of formed lysine oligomers (Lys2 to Lys5) wasdone using co-injection with commercial reference standards.

The relative electrophoretic mobilities �rel were the fol-lowings: 16.58 × 10−5 cm2/V/sec, 17.66 × 10−5 cm2/V/sec, and18.87 × 10−5 cm2/V/sec, and 19.71 × 10−5 cm2/V/sec for Lys2, Lys3,Lys4, and Lys5, respectively. All relative standard deviations (RSD)were not more than 1.5% (n = 3 runs). These experiments indicatedalso that an incubation time of 15 min was sufficient to generatelysine oligomers with excellent repeatability. Lys2, Lys3, Lys4, Lys5corrected area RSD were less than 3% (n = 3 runs). We observedthat increasing PB plugs lengths during step (1) affected productsformation. Higher amounts of Lys2, lower amounts of Lys3 andLys4, and no Lys5 were obtained by tripling the volume of PB plug(Fig. 2C). This suggested that changing PB plug length improvedthe PLL enzymatic degradation.

As proof of concept of the on-line digestion, the PLL was alsosubmitted to off-line cleavage with trypsin to compare enzymatichydrolysis products formation. PLL was incubated with trypsin ata ratio of enzyme to substrate (E/S) of 1/20 in weight over 20 h.The ratio of E/S at 1/20 was used for in-solution digestion since itwas indicated by Hustoft et al. [19] to be sufficient to generate spe-cific bond cleavages and high proteolysis rate. The off-line digestionreaction was continuously monitored by CE injection of digest fordifferent periods of time (Fig. 3A). As shown in Fig. 3B, it has beenestablished that off-line digestion reached a plateau around 8 h ofincubation. The final digest mixture was exclusively composed ofLys2 and Lys3.

An incubation period of 15 min using on-line CE methodologygave relative corrected areas of Lys2/Lys3/Lys4 to be 0.22/0.19/0.02.By comparison, with off-line digestion, the relative corrected areas

of Lys2/Lys3/Lys4 were 0.12/0.09/0 after 15 min incubation and therelative corrected area of Lys2 did not exceed 0.15 after 8 h incu-bation time. This suggests that Lys2 and Lys3 formation rates arebetter when on-line digestion of PLL is carried out. These results

G. Coussot et al. / J. Chromatogr. A 1376 (2015) 159–166 163

F min (ad inject

ca1ma

4

bgεtGttan

ai

F(

ig. 3. Electrophoretic profiles of poly-l-Lysine after off-line tryptic digestion after 5

ilysine and trilysine rate formation (B). No peaks appeared after 20 min. Marker ishe use conventional mode injection.

ould be explained by the fact that a higher ratio of E/S in weight waspplied during on-line digestion. In that case, E/S is approximately/5 if considering total E and S plugs overlapping during diffusionode mixing. This resulted in the presence of a broad trypsin peak

round 35 min in electropherograms referring to on-line digestions.

.3. On-line trypsin digestion of PLL-drug conjugates

Polymer-drug conjugates using PLL and taurine drug (Tau) haveeen investigated. Two coupling agents: glutaraldehyde (G) andlutaric acid anhydride (GA) were chosen to link Tau to the PLL

− amino group. Linkage using G required two steps: imine forma-ion and reduction step to form stable bonds between PLL and Tau.lutarylation of Tau was followed by ethylchloroformiate activa-

ion of the carboxylic group of the glutaryled Tau before couplingo PLL. The resulting PLL-Tau conjugates are named PLL-GA-Taund PLL-G-Tau; they are included in a drug preparation having

europrotection effects [20].

Running the on-line trypsin digestion of PLL-GA-Tau with thebove described methodology gave electrophoretic profiles differ-ng from that of PLL with the presence of peptide intermediates

ig. 4. Electrophoretic profiles of (a) PLL-GA-Tau and (b) PLL-G-Tau after on-line trypticB). Marker is injected at 0.25 g/L.

), 15 min (b), 1 h (c), 6 h (d) and 8 h incubation time at 37 ◦C (A) and the correspondingted at 5 g/L. A small mobility shift of the lysine oligomers was observed because of

(Fig. 2 & Insert of Fig. 4). Indeed, the electrophoretic profile of PLL-GA-Tau showed others peaks of lysine oligomers (bearing or notTau) having different mobilities (Fig. 4A). Nevertheless, as expected,peaks with �rel of 16.66 ×10−5 cm2/V/sec, 17.80 × 10−5 cm2/V/sec,19.08 × 10−5 cm2/V/sec, and 19.97 × 10−5 cm2/V/sec confirmed thepresence of Lys2, Lys3, Lys4, and Lys5. The relative corrected areaswere 1.36/1.58/0.94/0.41 for Lys2, Lys3, Lys4, and Lys5 respectively.Fig. 4B shows the electrophoretic profiles of PLL-G-Tau after on-line digestion with trypsin. In this figure, only three product peakswere identified as Lys2, Lys3, and Lys 4 with relative correctedareas 1.36/2.03/0.93. The electrophoretic profiles of PLL-Tau con-jugates revealed that depending on the nature of the linker usedto covalently bind Tau to PLL, enzymatic degradation behaviors ofPLL-Tau conjugates moved somewhat differently. The results werenot surprising since linkers are known to have different impacton drug release kinetics which will influence the PDCs therapeuticactivity.

By using the present on-line CE-based methodology, we wereable to differentiate two PLL-drug conjugates, whereas analysesof undigested PDC gave diffuse overlapping peaks. Our method-ology shows promising for studying degradability of linear

digestion with S (6.65 nL)/E(3.78 nL)/PB(4.75 nL) (A). PLL-Tau chemical structures

164 G. Coussot et al. / J. Chromatogr. A 1376 (2015) 159–166

F G4) afi times

Pb

4

d(co1ssw[dtdth

FdnpuaasLaaTvImaGmtss

ig. 5. Electropherograms of (a) DGL(G1), (b) DGL(G2), (c) DGL(G3) and (d) DGL(njected at 0.25 g/L. DGLs chemical structures (B). For (c) and (d), analyte migrationtructures.

LL-drug conjugates, and may be extending to others polymersackbones.

.4. On-line trypsin digestion of dendrigrafts poly-l-lysine (DGLs)

In comparison with PLL, dendritic architectures such asendrigraft poly-l-Lysine (DGL) have a tree like structurearborescence) obtained by an iterative condensation of lysine N-arboxyanhydrides (NCA) onto the previous generation. The coref the DGL is composed of about 8 lysine oligomers (generation

is named G1 and is linear) which is used as an initiator for theecond generation (G2) and so on [21]. In this paper, dendrigrafttructures are represented by three generations (G2, G3, G4), forhich physicochemical characterization has already been reported

21–23]. DGL have already been largely used as carrier for gene andrug delivery or for the transport of contrast agent [22,24]. Contraryo dendrimer moiety that did not show degradation over enzymaticegradation study [25], the DGL compounds have been proposed:hey would have the same biocompability as dendrimers but withigher degradability.

On-line digestions of G1, G2, G3, and G4 of DGL, depicted inig. 5, were conducted. As expected, the dendrigraft polymer showsegradation in the presence of trypsine. DGL standards containingo enzyme (E plugs replaced by water plugs) were injected. Broadeaks from native G2, G3, G4 and intermediate hydrolysis prod-cts have been observed with migration times between 22 minnd 30 min. The major products of the hydrolysis are Lys2, Lys3,nd Lys4 independently from the DGL generation. G2 and G3 gaveimilar relative corrected areas: 0.32/0.60/0.35 from Lys2, Lys3, andys4 issued from G2 proteolysis and 0.29/0.58/0.35 from Lys2, Lys3,nd Lys 4 from G3 hydrolysis. However, the relative corrected areasre about five times less compared from those obtained with G1.hese results demonstrate the effect of the topology (linear (G1)ersus dendrigraft (G2 and G3)) on the resulting hydrolysis rate.ndeed, linear structures lead to higher degradation rate due to

uch higher flexibility of the backbone and peptide bond spatialccessibility to enzyme. This point is unambiguously illustrated by4 that gave partial hydrolysis with lower Lys2, Lys3 and Lys 4 for-

ation. Relative corrected areas were reduced by one fifth. Again,

he reduced hydrolysis of peptide bonds could be explained by thetructure of DGL (G4) that may undergo uncontrolled space con-traints such as chain folding towards its core and steric hindrance

ter on-line tryptic digestion with S (6.65 nL)/E(3.78 nL)/PB(4.75 nL) (A). Marker iss between 25 and 30 min correspond to intermediate digested peptides and native

that can reduce accessibility to enzyme. This last comment seemsto be consistent with the results illustrated in Fig. 5 as the rate ofLys2, Lys3, and Lys4 is lower when increasing from one generationof DGL to another.

With these experiments, we demonstrated that CE-on line diges-tion method has the advantage to be applicable to others kinds of lysinepolymers and that, it is possible to apply it in order to demonstrate,for instance, the stability of new polymeric structures. Last step ofthe study was to show that the present CE strategy using on-linedigestion process is reliable and practicable for protein conjugatesdigests.

4.5. On-line trypsin digestion of protein conjugates

Protein conjugates are used for antibody production againsthaptens. Therefore, carrier proteins, such as BSA, are currently usedto stimulate immune response. In this section, BSA-IAA conjugatewas used as a model of BSA-drug conjugate.

The on-line digestions of BSA and BSA-IAA were accomplishedas described in Section 2. We demonstrated that to achieve efficientdigestion, those proteins required an additional denaturation stepprior to their tryptic digestion. Indeed, few proteolysis productswere detected if no denaturation step was done. Thermal denatur-ation was then carried out at 80 ◦C for 10 min. The results for on-and off-line digestions of BSA and BSA-IAA are shown in Fig. 6, usingphosphate buffer (pH 2.5, I 50 mM) as BGE. Again, as expected, theCE profiles of intact BSA and BSA-IAA gave broad overlapping peaks(see insert in Fig. 6).

At pH 2.5, most of the peptides/proteins have global cationiccharge. Separation of tryptic peptides of BSA were facilitated withthis BGE of low pH. BGE at pH 5.0 was not suitable since it is nearthe isoelectric point of BSA (4.7). Indeed, our aim was to have anoverall CE profile of digests including undigested initial conjugate.

The on-line methodology proved its ability to clearly differenti-ate BSA from BSA-IAA conjugate (Fig. 6(c) and (e)) only on the basisof the comparison of CE fingerprints. Some differences are noticedbetween on- and off-line CE profiles (principally the presence of

trypsin peak at 14.3 min due to higher amount of E/S ratio for on-line methodology). Studies are under investigation to widen the useof the on-line methodology to others PDCs with the optimizationof PB and E/S ratio for instance. Nevertheless, these data are very

G. Coussot et al. / J. Chromatogr. A 1376 (2015) 159–166 165

F -IAA (I nL), r1

et

5

iPCteh

uoaotspmt

A

t

ig. 6. Comparison of electrophoretic profiles of BSA (a, continuous line) and BSAAA (e,f) after on- and off-line tryptic digestions with S (6.65 nL)/E(3.56 nL)/PB(4.754.3 min.

ncouraging for protein conjugate analysis, and more especially forhe drug quantitation loading on protein carrier.

. Conclusion

As a conclusion, the originality of our methodology resides firstn the combination of “sandwich” injections/Diffusion-mediatedroteolysis/ Electrophoretic Separation in a single coated capillary.onsequently, we named this methodology as “D-PES”. Secondly,he D-PES methodology has the advantages to be totally automatedliminating the risk of interferences, saving time, and providingighly repeatable electrophoretic profiles of digests.

In this paper, various PDC synthesized from synthetic or nat-ral polypeptides were used to demonstrate the feasibility ofur methodology. Applied to PLL, the D-PES methodology haslso proved its ability to directly detect and estimate the lysineligomers formation. D-PES potentialities may also have applica-ions in development of new poly(aminoacids)-drug conjugatesince both the study of their degradability and the estimation ofercentage of grafting could be considered (with some improve-ents). Studies are still under investigations to extend the use of

he D-PES methodology to quality control of PDCs.

cknowledgments

We would like to thank the GemacBio company (conven-ion number: CT 050886) for financial support. The authors are

b, dashed line) in their intact form (undigested). CE profiles of BSA (c,d) and BSA-espectively. Imidazole marker is injected at 0.25 g/L. Trypsin peak appeared about

grateful to Rana Hassini, Coline Spinau and master students fortheir assistance in the on-line CE-based digestion methodologyexperiments. The authors gratefully acknowledge all staff mem-bers of the Colcom and GemacBio companies for their helpfulcomments.

References

[1] H. Tian, Z. Tang, X. Zhuang, X. Chen, X. Jing, Biodegradable synthetic polymers:Preparation, functionalization and biomedical application, Prog. Polym. Sci. 37(2012) 237–280.

[2] W.B. Liechty, D. Kryscio, B.V. Slaughter, N.A. Peppas, Polymers for drug deliverysystems, Annu. Rev. Chem. Biomol. Eng. 1 (2010) 149–173.

[3] R. Swami, D. Kumar, W. Khan, R. Sistla, N. Shastri, Polymer–drug conjugate infocal drug delivery, in: A.J. Domb, W. Khan (Eds.), Advances in Delivery Scienceand Technology, Part I, 2014, pp. 117–147.

[4] D. Polyak, A. Eldar-Boock, H. Baabur-Cohen, R. Satchi-Fainaro, Polymer conju-gates for focal and targeted delivery of drugs, Polym. Adv. Technol. 24 (2013)777–790.

[5] R. Gaspar, R. Duncan, Polymeric carriers: Preclinical safety and the regulatoryimplications for design and development of polymer therapeutics, Adv. Drug.Deliv. Rev. 61 (2009) 1220–1231.

[6] L. Switzar, M. Giera, W.M.A. Niessen, Protein digestion: an overview of theavailable techniques and recent developments, J. Proteome Res. 12 (2013)1067–1077.

[7] H. Kolsrud Hustoft, H. Malerød, S. Ray Wilson, L. Reubsaet, E. Lundanes, T.Greibrokk, A critical review of trypsin digestion for LC-MS based proteomics,

in: H.-C. Leung (Ed.), Integrative Proteomics, InTech, 2012, pp. 72–92, Chapter4.

[8] H. Yamaguchi, M. Miyazaki, T. Honda, M.P. Briones-Nagata, K. Arima, H. Maeda,Rapid and efficient proteolysis for proteomic analysis by protease-immobilizedmicroreactor, Electrophoresis 30 (2009) 3257–3264.

1 atogr

[

[

[

[

[

[

[

[

[

[

[

[

[

[

66 G. Coussot et al. / J. Chrom

[9] Y. Li, R. Wojcik, N.J. Dovichi, A replaceable microreactor for on-line proteindigestion in a two-dimensional capillary electrophoresis system with tandemmass spectrometry detection, J. Chromatogr. A 1218 (2011) 2007–2011.

10] N. Iki, E.S. Yeung, Non-bonded poly(ethylene oxide) polymer-coated columnfor protein separation by capillary electrophoresis, J. Chromatogr. A 731 (1996)273–282.

11] V. Okhonin, X. Liu, S.N. Krylov, Transverse diffusion of laminar flow profiles toproduce capillary nanoreactors, Anal. Chem. 77 (2005) 5925–5929.

12] D. López-Ferrer, A. Ramos-Fernández, S. Martínez-Bartolomé, P. García-Ruiz,J. Vázquez, Quantitative proteomics using 16O/18O labeling and linear ion trapmass spectrometry, Proteomics 6 (2006) S4–S11.

13] M. Zeisbergerová, A. Adámková, Z. Glatz, Integration of on-line protein diges-tion by trypsin in CZE by means of electrophoretically mediated microanalysis,Electrophoresis 30 (2009) 2378–2384.

14] X. Wang, K. Li, E. Adams, A.V. Schepdael, Recent advances in CE-mediated microanalysis for enzyme study, Electrophoresis 35 (2014)119–127.

15] V. Okhonin, E. Wong, S.N. Krylov, Mathematical model for mixing reactants ina capillary microreactor by transverse diffusion of laminar flow profiles, Anal.Chem. 80 (2008) 7482–7486.

16] R. Nehmé, C. Perrin, H. Cottet, M.D. Blanchin, H. Fabre, Influence of poly-

electrolyte coating conditions on capillary coating stability and separationefficiency in CE, Electrophoresis 29 (2008) 3013–3023.

17] R. Nehmé, C.C. Perrin, H. Cottet, M.D. Blanchin, H. Fabre, Influence of polyelec-trolyte capillary coating conditions on protein analysis in CE, Electrophoresis30 (2009) 1888–1898.

[

[

. A 1376 (2015) 159–166

18] S.C. Shukla, A. Singh, A.K. Pandey, A. Mishra, Review on production and medicalapplication of �-polylysine, Biochem. Eng. J. 65 (2012) 70–81.

19] H.K. Hustoft, L. Reubsaet, T. Greibrokk, E. Lundanes, H. Malerød, Critical assess-ment of accelerating trypsination methods, J. Pharmaceut. Biomed. 56 (2011)1069–1078.

20] M. Geffard, Use of monofunctional and/or polyfonctionnal derivates of thepoly-L-lysine for the preparation of useful drugs in the treatment of the neu-rodegeneration, contagious, traumatic and toxic neuropathies, degenerativeauto immune diseases and proliferative diseases, Patent registered on the11/18/1994 under no. 9413861, international extension no. PCTR, 95011517on the 11/17/1995.

21] H. Collet, E. Souaid, H. Cottet, A. Deratani, L. Boiteau, G. Dessalces, J.-C. Rossi, A.Commeyras, R. Pascal, An expeditious multigram-scale synthesis of lysine den-drigraft (DGL) polymers by aqueous N-carboxyanhydride polycondensation,Chem. Eur. J. 16 (2010) 2309–2316.

22] H. Cottet, M. Martin, A. Papillaud, E. Souaid, H. Collet, A. Commeyras, Determi-nation of dendrigraft poly-L-lysine diffusion coefficients by taylor dispersionanalysis, Biomacromolecules 8 (2007) 3235–3243.

23] G. Coussot, E. Nicol, A. Commeyras, I. Desvignes, R. Pascal, O. Vandenabeele-Trambouze, Colorimetric quantification of amino groups in linear and dendriticstructures, Polym. Int. 58 (2009) 511–518.

24] R. Huang, S. Liu, K. Shao, L. Han, W. Ke, Y. Liu, J. Li, S. Huang, C. Jiang, Evaluationand mechanism studies of PEGylated dendrigraft poly-L-lysines as novel genedelivery vectors, Nanotechnology 21 (2010) 265101.

25] K.A. Boduch-Lee, T.M. Chapman, S.E. Petricca, K.G. Marra, P. Kumta, Macro-molecules 37 (2004) 8959–8966.