Journal of Colloid and Interface Science 431 (2014) 117–124

Contents lists available at ScienceDirect

Journal of Colloid and Interface Science

www.elsevier .com/locate / jc is

A comparison between adsorption mechanism of tricyclicantidepressants on silver nanoparticles and binding modeson receptors. Surface-enhanced Raman spectroscopy studies

http://dx.doi.org/10.1016/j.jcis.2014.05.0600021-9797/� 2014 Elsevier Inc. All rights reserved.

⇑ Corresponding author. Fax: +48 12 634 0515.E-mail address: malek@chemia.uj.edu.pl (K. Malek).

Aleksandra Jaworska, Kamilla Malek ⇑Faculty of Chemistry, Jagiellonian University, Ingardena 3, 30-060 Krakow, Poland

a r t i c l e i n f o

Article history:Received 10 March 2014Accepted 29 May 2014Available online 16 June 2014

Keywords:Tricyclic antidepressantsAdsorption mechanism on Ag nanoparticlesSurface-enhanced Raman spectroscopy(SERS)Drug action

a b s t r a c t

A series of the tricyclic antidepressants known as a surface-active drugs, has been used as a model for anevaluation of their adsorption mechanism on the metal substrate and its relationship to pharmacologicalaction of the chosen drugs. In these studies, six antidepressants were adsorbed on the metal substrate in aform of silver nanoparticles (ca. 30 nm in diameter) and afterwards their interactions have beenexamined in terms of surface-enhanced Raman spectroscopy (SERS). An analysis of SERS spectra hasrevealed that the dibenzopine moiety is a primary site of the adsorption with some differences in theorientation with respect to the metal among the studied molecules. The spectral changes due to the inter-actions with the silver particles also appear in the region typical for vibrations of the side chain. Theseobservations are consistent with a model, in which the tricyclic ring is docked in the outer vestibule ofbiogenic amine transporters whereas the dimethyl-aminopropyl side chain is pointed to the substratebinding site. This work sheds a light on a potential of SERS technique in predicting a key functional groupresponsible for drug action.

� 2014 Elsevier Inc. All rights reserved.

1. Introduction

Tricyclic antidepressants (TCAs) are the drugs commonly usedto treat depression. Chemically, they are cationic, amphiphilic,small-size secondary or tertiary amines with a common core con-sisting of two flanking aromatic rings attached to a seven-mem-bered ring. The latter can be a N- or O-heterocyclic ring (seeFig. 1). The most common TCAs are imipramine (Imi), desipramine(Des), clomipramine (Clo), amitriptyline (Ami), nortriptyline (Nor),and doxepine (Dox). They are categorized as selective serotonin,norepinephrine and dopamine reuptake inhibitors [1]. This typeof the antidepressants still remains in clinical use, especially fortreatment-resistant depression [2]. In toxic doses, however, theycan lead to hypotension and cardiovascular disease and their useis associated with poisoning causing death. Thus, numerous ana-lytical techniques such as fluorimetry, chemiluminometry, chro-matography, enzyme and fluorescence polarization immunoassayand others [3 and therein] have been developed to identify anddetect TCAs in biological matrix. The most often, their use is pro-ceeded by an extraction the drugs from biological materials

through liquid–liquid, solid phase, and microwave-assisted extrac-tion techniques [4,5]. A potential application of surface-enhancedRaman spectroscopy (SERS) without a need of the extraction pro-cess has been also proposed [4].

Although tricyclic antidepressants have been used for years,their orientation within a primary target of their action has beenyet unresolved. Sarker et al. have proposed a model, in which thetricyclic ring of the antidepressant is docked into the outer vesti-bule of serotonin transporter (SERT) whereas the drug’sdimethyl-aminopropyl side chain points to the substrate bindingsite. Consequently, such binding can create a structural change inthe inner and outer vestibule, which precludes docking of the tricy-clic ring [6]. In addition, studies on simultaneous binding of morethan one antidepressant have indicated the presence of a secondbinding site in the inner vestibule, which can be the pseudo-sym-metric fold of monoamine transporters [1]. Imipramine is a ligandthat stabilizes SERT in the outward-facing conformation. Despitethe fact that the flexibility of the methyl-aminopropyl side chainis restricted by a double bond in Ami, Nor and Dox, they adopt adocking pose similar to imipramine [6–8]. In studies of Sinningand co-workers on a series of structural analogues of imipramine,a salt bridge between the tertiary aliphatic amine of TCAs andAsp98 of the human serotonin transporter (hSERT) was identifiedwhile the 7-position of the imipramine ring was found vicinal to

Fig. 1. Structures of the studied tricyclic antidepressants.

118 A. Jaworska, K. Malek / Journal of Colloid and Interface Science 431 (2014) 117–124

Phe335 [2]. A similar docking fashion of clomipramine into a trans-porter has been resolved in crystal structure of a complex betweenClo and a biogenic amine transporter Leu-BAT [9]. In turn, crystalstructure of the Drosophila melanogaster dopamine transporter(DAT) in complex with nortriptyline showed that the TCA alongwith cholesterol molecules stabilize the open conformation ofthe receptor through ionic interaction with one chloride and twosodium ions [8].

TCAs are also known as agents inducing lipidosis and intracel-lular accumulation of lipids. However, these interactions stronglydepend on the type of the drug as well as a lipid structure andlipid phase transitions [10 and therein]. Optical-trapping confocalRaman spectroscopy has been used in the evaluation of the inter-action of amitriptyline and nortriptyline with various phospho-lipid membranes [10]. Changes in an acyl chain conformation ofthe membranes and their intra- and intermolecular order due tothe drugs action have been determined by monitoring peakintensities and positions of Raman signal originating fromphospholipid vesicles. These studies have proposed that the TCAsinteract via their cyclic rings with the acyl chains of the mem-branes whereas the tertiary aliphatic amine group is located nearthe lipid head groups. These results were found in agreementwith pharmacological studies [10 and therein] and they haveshown that Raman spectroscopy is an effective technique toelucidate drug–membrane interactions using micromolar concen-trations of both lipids and drugs.

The modern modification of Raman spectroscopy, surface-enhanced Raman spectroscopy (SERS), utilizes generation of verystrong electromagnetic field resulting from exciting of the localizedsurface plasmons in the metallic nanoparticles. SERS spectrum isobserved if a molecule is in a close contact with a SERS-activesupport. Nowadays, SERS has been widely used in detection, iden-tification and monitoring various biochemical processes since thistechnique has fast, label-free and non-invasive nature togetherwith its high molecular specificity and sensitivity [11,12]. First ofall, SERS provides valuable information on the adsorptionmechanism of a (bio)molecule on a metallic surface pointing whatfunctional groups or atoms participate in metal–adsorbateinteractions.

In this study, we present SERS studies on six antidepressants:imipramine, desipramine, clomipramine amitriptyline, nortriptyline,and doxepine, whose structures are depicted in Fig. 1. We analyze thetwo groups of the tricyclic antidepressants: the first group containsthe nitrogen atom inserted into the tricyclic ring (Imi, Des, Clo) andthe second group represents the molecules with the double CC bondattached to the ring system (Ami, Nor, Dox). To record surface-enhanced Raman signal of the molecules, colloidal silver particleswere used and spectra were recorded with a laser excitation in thenear-infrared region. The SERS spectra are analyzed in order to getinsight into the adsorption behavior on the metal surface. Finally,we compare the proposed models of the surface adsorption of thechosen molecules with their docking profile into receptors.

2. Experimental

2.1. Chemicals

All chemicals were purchased from Sigma, Germany and wereof analytical grade. The TCAs were in a form of hydrochloride salt.Aqueous solutions of the antidepressants with the concentrationsof 0.2 M (for normal Raman spectra) and 1 � 10�2 M (for SERS)were prepared by dilution of an appropriate amount of the ana-lytes in the 4-fold distilled water. Silver colloid was preparedaccording to procedure described previously [13]. In this synthesis,Ag ions are reduced in the alkaline solution of hydroxylamine.UV–Vis spectrum of the silver colloid shows the presence of aresonant absorption band at ca. 412 nm (Fig. S1, Supporting Infor-mation). This position of the absorption maximum indicates thatthe size of silver nanoparticles varies in the range of 30–40 nm.Next, for SERS measurements, 10 lL of a sample solution wasmixed with 500 lL of a colloid. The final concentration of theanalytes in the mixture was 2 � 10�4 M.

2.2. Instrumentation

The absorption spectra were recorded with a UV–Vis–NIRNicolet spectrophotometer (model Evolution 60) in the range of

A. Jaworska, K. Malek / Journal of Colloid and Interface Science 431 (2014) 117–124 119

190–1100 nm with a resolution of 1 nm. To place a sample, quartzcells of 1 cm width were used.

All Raman spectra were recorded on a MultiRAM FT-RamanSpectrometer (Bruker) equipped with a Nd:YAG laser, emitting at1064 nm, and a germanium detector cooled with liquid nitrogen.The output power of the laser was 150–300 mW while the spectralresolution was 4 cm�1 for all measurements. 128, 4000, and 1024Scans were collected for the solids, solutions and SERS, respec-tively. A few milligrams of solids and 1 mL of solutions were placedon metal discs and in quartz cuvettes, respectively. Integratedintensities of the chosen bands were calculated from the Ramanand SERS spectra employing integration method. To accomplishthis, a linear baseline was drawn through the peak edges, andthe spectrum below this line was integrated over the wavenumberrange of the band. All spectra processing was performed by using aBruker OPUS software (Version 6.5).

2.3. An assignment of Raman bands

Density functional theory (DFT) calculations were carried outfor geometry optimization and simulation of Raman spectra. Thesecalculations were performed with the Gaussian 09 program pack-age [14] by using B3LYP method [15,16] and 6-31+G(d) basis set.No imaginary frequencies were obtained. Next, theoretical Ramanintensities (IR) were obtained from Gaussian Raman scatteringactivities (S) according to the expression in [17]. Then, simulatedspectra were compared to experimental normal Raman spectra ofthe solids (Figs. S2–S7, Supporting Information). To provide theunequivocal assignment of the calculated Raman spectra, thepotential energy distribution (PED) analysis was performed byusing the Gar2ped software [18]. The program first defines a setof non-redundant internal coordinates according to the Pullay’sand Foragasi’s definitions [19,20], and next the percentagecontribution of the internal coordinates to the total energy of eachnormal mode is computed. Definitions of the internal coordinatesused in this work are collected in Table S1 (SupportingInformation).

Fig. 2. Normal (c = 0.2 M) and SERS (c = 2.0 � 10�4 M) spectra of imipramine (A andB, respectively), desipramine (C and D, respectively) and clomipramine (E and F,respectively).

3. Results and discussion

UV–Vis absorption spectra of the silver colloid and its mixturewith the adsorbate point out what type of SERS mechanism playsa significant role in the adsorption process on a metal colloid.The presence of a single band due to particle plasmon resonanceindicates physisorption of a molecule on the sol whereas theappearance of the second peak in the red/near-infrared spectralregion suggests a contribution of charge transfer mechanism dueto the molecule–metal interaction. UV–Vis spectrum of the silvercolloid used in this work exhibits the presence of the plasmon res-onance band at ca. 412 nm typical for this sol (Fig. S1B, SupportingInformation). After addition of the solution of the drug, absorbanceof this band significantly decreases along with blue-shifting of themaximum (Fig. S1C). Broadening of the band indicates the aggrega-tion of the silver particles due to the interaction of the analyte withthe silver. Fig. S1C shows an exemplary spectrum for imipramine,however, the spectra of the other TCAs exhibit similar features.In addition, the second plasmon resonance band is observed atca. 900 nm. This suggests that charge-transfer mechanism mayalso contribute to surface-enhancement of Raman signal recordedwith the use of the 1064 nm laser excitation. For the analytical pur-pose, we also investigated SERS features of Imi and Des by using alaser excitation at 785 nm [4]. In general, band positions andrelative intensities are very similar for both NIR wavelengths ofexciting light, indicating a similar mechanism of enhancement ofRaman signal.

In our experiment, SERS spectra were recorded in the presenceof potassium chloride as an aggregating agent as well as without it.However, despite the fact that halide ions generally increase SERSintensity due to the formation of Ag+–Cl�-ligand surface complexesor due to an increase of the local electromagnetic field by aggrega-tion of metal colloid particles, we observed a decrease of SERSsignal of the antidepressants by ca. 20–30% after addition of KCl(data not shown). This indicates that SERS phenomenon for TCAsis negatively affected by KCl because of adsorption competitionbetween chlorides ions and the adsorbate due to the formationof Ag–Cl clusters on the metal surface [4,21].

Figs. 2 and 3 display the normal Raman (NR) and SERS spectra ofthe antidepressants containing the nitrogen atom in the 7-mem-bered ring and the C@C bond attached to this ring, respectively.Since Raman bands of the antidepressants are not observed inspectra of the solutions at the concentration of 2 � 10�4 M usedin SERS, the SERS spectra shown in Figs. 2 and 3 are exclusivelydue to surface enhancement of the Raman scattering of adsorbedmolecules by silver nanoparticles. Firstly, a close analysis of NRspectra of solutions (Figs. 2 and 3) and solids (Figs. S2–S7) revealsno substantial differences besides the well-known characteristicsof the solution spectra concerning the broadening and slightchanges in the peak position of the bands. Thus, Tables 1–6summarize the assignment of the Raman and SERS bands of thesolutions on the basis of DFT calculations described above whileTables 7 and 8 summarize the selected relative intensities of bandsoriginating from the phenyl and 7-membered rings, the aliphaticchain and the terminal methyl groups. Instead of a complete com-parison of all modes of all the molecules, we will focus on a fewregions of the Raman and SERS spectra, which can be of a specialinterest in terms of the adsorption mechanism on the silver

Fig. 3. Normal (c = 0.2 M) and SERS (c = 2.0 � 10�4 M) spectra of amitriptyline (Aand B, respectively), nortriptyline (C and D, respectively) and doxepine (E and F,respectively).

Table 1The selected Raman (solution, c = 0.2 M) and SERS (c = 2 � 10�2 M) bands (in cm�1) ofimipramine in the range of 500–3100 cm�1 and their tentative assignments (in %,PED > 7% shown) based on calculations.

Raman SERS Assignment

389 386 xNC/R7 (26), sR7 (15), qNC/aliph (7)450 432 bNC/aliph (35), sR6 (16), xNC/R7 (7)498 500 dCH2/aliph (36), qNC (11)567 567 bR7 (23), bR6 (22), xN1C (9)685 683 bR6 (32), mCC//R7 (19), sR6 (7)973 972 mCC/R7 (32), qCH2/R7 (25)

1044 1042 mCC/aliph (38), dCH2/aliph (21), bNC/met (12)1063 1061 mCC/R61 (41), mNC/aliph (8)1127 1127 mNC/aliph (14), bR7 (10), sCH2/R7 (8), mCC/R62 (7)1207 1206 mCC/R7 (33), bR6 (13)1232 1229 mNC/R7 (13), sCH2/aliph (10), mNC/aliph (7), qCH2/aliph (7)1436 1431 bmet (46), dCH2/aliph (42)1473 1468 dCH2/aliph (38), bmet (31)1597 1597 mCC/R62 (35), mCC/R61 (31)2971 2963 mCH/met (89)3062 3057 mCH/R62 (96)

m – stretching, d – scissoring, q – rocking, x – wagging, s – twisting, R6 – 6-membered ring, R7 – 7-membered ring, aliph – aliphatic chain, Met – methyl group,vs –very strong, s – strong, m – medium, w – weak, vw – very weak. Numbering ofthe functional groups is shown in Fig. 1.

Table 2The selected Raman (solution, c = 0.2 M) and SERS (c = 2 � 10�2 M) bands (in cm�1) ofdesipramine in the range of 350–3100 cm�1 and their tentative assignments (in %,PED > 7% shown) based on calculations.

Raman SERS Assignment

411a 423 sR6 (29), bR6 (16), bflR62/R7 (9)495 498 sR6 (18), sR7 (15), bR6 (8), xN1C/aliph (8)565 565 bR7 (33), bR6 (22), xNC/aliph (9)604 621 sR6 (42), bR7 (13)685 683 bR6 (34), mCC/R7 (19), sR6 (7)779 774 sR6 (15), xN2H (8), bR6 (7), mCC/R7 (7)973 970 mCC/R7 (32), qCH2/R7 (25)

1044 1042 mCC/aliph (32), mCC/R62 (21),1063 1059 mCC/R61 (53), qCH/R61 (12)1127 1127 mNC/aliph (13), bR7 (9), sCH2/R7 (7)1165 1161 qCH/R61 (13), qCH/R62 (9)1208 1206 mCC/R7 (33), bR6 (13), mCC/R62 (7)1233 1231 mNC/R7 (18), qCH/R61 (12), sCH2/R7 (11), mNC/aliph (7)1438 1433 dCH2/aliph (83)1475 1470 dCH2/R7 (80)1598 1595 mCC/R61 (31), mCC/R62 (23)

2957a 2963 mCH/met (89)3051a 3055 mCH/R62 (86)

m – stretching, d – scissoring, q – rocking, x – wagging, s – twisting, b – in-planebending, btf – butterfly, R6 – 6-membered ring, R7 – 7-membered ring, Met –methyl group, aliph – aliphatic chain, vs – very strong, s – strong, m – medium, w –weak, vw – very weak. Numbering of the functional groups is shown in Fig. 1.

a Observed in the Raman spectrum of the solid sample.

Table 3The selected Raman (solution, c = 0.2 M) and SERS (c = 2 � 10�2 M) bands (in cm�1) ofclomipramine in the range of 350–3100 cm�1 and their tentative assignments (in %,PED > 7% shown) based on calculations.

Raman SERS Assignment

370 372 sR6 (36), bR7 (16), bR6 (10)399 399 bR7 (30), bR6 (21), mCCl (10)462 463 bCN/met (22), sR6 (13)507 506 dCH2/aliph (34), sR6 (15), qNC (11), bR7 (7)557 556 bR6 (17), bR7 (14), sR7 (10), xNC/aliph (9)631 629 sR6 (33), xCCl (21), bR6 (8)692 692 bR6 (39), mCC/R7 (18)769 770 xCH/R62 (33), bR6 (10)974 972 mCC/R7 (32), qCH2/R7 (26)

1044 1044 mCC/aliph (36), dCH2/aliph (21), bNC/met (15)1065 1061 mCC/R61 (36), qCH/R61 (17)1105 1101 mCC/R62 (28), b/R6 (7)1127 1128 mCC/R61 (16), qCH/R61 (15), mCC/R62 (11)1165 1161 qCH/R61 (85)1208 1206 mCC/R7 (33), b/R6 (13)1231 1229 mNC/R7 (20), mNC/aliph (11), sCH2/R7 (9)1437 1431 dCH2/aliph (89)1592 1591 mCC/R62 (41), mCC/R61 (12), mCC/R7 (7)2974 2960 mCH/met (86), mCH/aliph (8)

3064a 3054 mCH/R62 (97)

m – stretching, d – scissoring, q – rocking, x – wagging, s – twisting, b – in-planebending, R6 – 6-membered ring, R7 – 7-membered ring, Met – methyl group, aliph –aliphatic chain, vs – very strong, s – strong, m – medium, w – weak, vw – very weak.Numbering of the functional groups is shown in Fig. 1.

a Observed in the Raman spectrum of the solid sample.

120 A. Jaworska, K. Malek / Journal of Colloid and Interface Science 431 (2014) 117–124

nanoparticles. At first glance, the SERS spectra exhibit the presenceof all bands observed in the corresponding NR spectra. No shift andsignificant broadening of SERS bands are observed, thus electro-magnetic (long-distance) mechanism of surface enhancement ofRaman signal rather plays a leading role in SERS of TCAs. However,the most pronounced feature of all the SERS spectra is a strongenhancement of a band at ca. 680–690 cm�1 attributed to the in-plane bending mode of both the flanking phenyl rings (bR6) coupled

with the stretches of the CAC bonds of the 7-membered ring. Thismode is a simultaneous in-phase breathing vibration of the dib-enzazepine ring. We compare the enhancement of this mode withSERS intensity of the phenyl ring CC stretching vibration (mCCR6)observed at 1600 cm�1 and often denoted as the 8a mode (accord-ing to the Wilson’s notation, [22], see PED in Tables 1–6). Since theformer is a symmetric vibration, its large enhancement due to theadsorption on the metal surface confirms its origin from electro-magnetic rather than from charge-transfer mechanism of the SERSphenomenon [23]. First of all, the enhancement of the 680 cm�1

band gives an evidence for a strong interaction of the conjugated

Table 6The selected Raman (solution, c = 0.2 M) and SERS (c = 2 � 10�2 M) bands (in cm�1) ofdoxepine in the range of 350–3100 cm�1 and their tentative assignments (in %,PED > 7% shown) based on calculations.

Raman SERS Assignment

343 343 sR6 (32), bR7 (22)455 457 qNC/met (23), bR61(10)696 696 bR6 (44), mCC/R7 (17), mCC/R7 (8)838 837 mNC/met (48), mNC/aliph (20)

1041 1040 mCC/aliph (42), mOC/R7 (24),1050 1051 mOC (31), mCC/R62 (24), qCH2/R7 (8)1165 1157 qCH/R62 (45), mCC/R62 (10)1221 1221 mOC (28), qCH/R62 (10), mCC/R7 (10)1369 1366 qCH2/aliph (46), mCC/R7 (14)1461 1459 bCH3 (82)1604 1601 mCC/R61 (31), mCC/R62 (19)1638 1636 mC@C (63), qCH/aliph (9)2960 2961 mCH/met (96)3060 3056 mCH/R61 (95)

m – stretching, d – scissoring, q – rocking, x – wagging, s – twisting, b – in-planebending, R6 – 6-membered ring, R7 – 7-membered ring, Met – methyl group, aliph –aliphatic chain, vs –very strong, s – strong, m – medium, w – weak, vw – very weak.Numbering of the functional groups is shown in Fig. 1.

Table 4The selected Raman (solution, c = 0.2 M) and SERS (c = 2 � 10�2 M) bands (in cm�1) ofamitriptyline in the range of 350–3100 cm�1 and their tentative assignments (in %,PED > 7% shown) based on calculations.

Raman SERS Assignment

334 336 sR6 (28), bR7 (25)534 534 bR6 (28), bR7 (12), sR6 (7), mCC/R7 (7)689 689 bR6 (48), mCC/R7 (7)966 968 mCC/R7 (29), qCH2/R7 (25), bR61(7)

1040 1038 mCC/aliph (56), qCH/R62 (13)1055 1053 mCC/R61 (56), qCH/R61 (12)1163 1163 mCC/R7 (16), mCC/R62 (12), bR6 (11), qCH/R61 (9)1208 1208 mCC/R7 (25), qCH/R62 (12), mCC/R62 (10), bR6 (7)1368 1366 qCH2/aliph (52), mCC/R7 (9)1431 1427 bNC/met (95)1599 1597 mCC/R61 (53), qCH/R61 (13), bR6 (8)1638 1638 mC@C (64), qCH/aliph (9), mCC/aliph (7)2972 2963 mCH/met (94)3047 3042 mCH/R62 (99)

m – stretching, d – scissoring, q – rocking, x – wagging, s – twisting, b – in-planebending, R6 – 6-membered ring, R7 – 7-membered ring, Met – methyl group, aliph –aliphatic chain, vs – very strong, s – strong, m – medium, w – weak, vw – very weak.Numbering of the functional groups is shown in Fig. 1.

Table 5The selected Raman (solution, c = 0.2 M) and SERS (c = 2 � 10�2 M) bands (in cm�1) ofnortriptyline in the range of 350–3100 cm�1 and their tentative assignments (in %,PED > 7% shown) based on calculations.

Raman SERS Assignment

334 336 sR6 (32), bR7 (26)534 535 bR6 (27), sR6 (8), mCC/R7 (7), bR7 (7)690 689 bR6 (46), mCC/R7 (8)970 968 mNC/aliph (29), mNC/met (21), bNC/met (11)

1040 1038 mCC/aliph (68), qCH/R62 (15)1055 1057 mCC/R61 (60), qCH/R61 (12)1165 1163 bNC/met (40), mNC/aliph (27)1208 1206 mCC/R7 (30), qCH/R62 (11), mCC/R62 (10), bR6 (7)1369 1370 qCH2/aliph (45), mCC/R7 (10), xCH2/R7 (8)1425 1427 bNC/met (86)1599 1597 mCC/R62 (54), qCH/R62 (13)1641 1641 mC@C (64), qCH/aliph (9), mCC/aliph (7)2963 2963 mCH/R7 (97)3052 3044 mCH/R62 (95)

m – stretching, d – scissoring, q – rocking, x – wagging, s – twisting, b – in-planebending, R6 – 6-membered ring, R7 – 7-membered ring, Met – methyl group, aliph –aliphatic chain, vs –very strong, s – strong, m – medium, w – weak, vw – very weak.Numbering of the functional groups is shown in Fig. 1.

A. Jaworska, K. Malek / Journal of Colloid and Interface Science 431 (2014) 117–124 121

system of the rings and the silver surface instead of the most com-mon adsorption on the silver nanoparticles via the p-electron sys-tem of the aromatic rings [24–26]. In addition, weakening of the 8amode and the presence of very weak bands originating from thestretching vibration of the aromatic CH bonds (about 3055 cm�1,cf. Figs. 2 and 3) confirm that the phenyl rings are not oriented per-pendicularly to the metal surface [27]. Thus, we conclude that theconjugated rings are tilted with the respect to the metal adoptingan edge-on geometry.

The detailed examination of the SERS spectra of the drugs con-taining the nitrogen atom in the 7-membered ring reveals thattheir SERS features differ even due to small structural modifica-tions such as substitution of the methyl group by the hydrogenatom in imipramine and desipramine (Imi, Des, Clo; cf. Figs. 1and 2). The mentioned-above ratio of bands at 685 and1600 cm�1 (I683/I1600) in these SERS spectra increases noticeablyrelative to the corresponding normal Raman bands by a factor of2–8, confirming the contribution of the dibenzazepine system inthe interaction with the silver (Table 7). This ratio in the SERS spec-tra is similar for imipramine and clomipramine whereas it is aslightly lower for desipramine (see Table 7). Since spatial arrange-ment of the ring systems in all the antidepressants is similarwhereas the I683/I1600 ratio is similar for Imi and Clo containingthe chlorine atom in the phenyl ring, we assumed that the loss ofthe methyl group in the aliphatic chain of desipramine may resultin the competition of the other fragment of this antidepressant inthe adsorption process and/or a different orientation of its ringswith respect to the silver surface. In the case of the latter, we noticevariation in the ratio between bands at 683 and 1060 cm�1 (cf.Table 7). According to PED calculations for this group of antide-pressants, the 1060 cm�1 band is assigned to the in-phasestretches of the CC bonds of the R61 ring that are non-conjugatedto the 7-membered ring (see Fig. 1 for numbering). The I683/I1060

ratio is much lower for desipramine than for imipramine and clo-mipramine indicating a larger enhancement of the 1060 cm�1 bandfor Des than for Imi and Clo due to the adsorption on the silver sol.This can suggest that one of the phenyl rings of Des is directedtowards the silver nanoparticles whereas Imi and Clo interactthrough the entire dibenzazepine moiety. Additionally, amongthe molecules discussed here, the out-of-plane bending modes ofthe rings observed below 650 cm�1 are enhanced in the SERS spec-tra of Imi and Des, see Fig. 2 and Tables 1 and 2. They are also pres-ent in the SERS spectrum of clomipramine but their intensities aresimilar to those in the normal Raman spectrum of the solution.This may suggest that the interaction of the Imi and Des aromaticrings with the metal via the p-electron system also appears. Hence,the presence of these modes and the in-plane bending vibrations ofthe rings indicates a slightly tilted orientation of the dibenzazepinerings of Imi and Des in comparison to Clo. We also examine thechange in the integral intensity of the marker bands of the aliphaticchain, i.e. mCC and dCH2 observed at 1042 and 1431 cm�1, respec-tively, with respect to the 683 cm�1 band (see Table 7). Among thethree antidepressants, the SERS spectrum of desipramine exhibitsthe largest surface enhancement of the mCCaliph mode along witha very low increase in the intensity of the scissoring vibration ofthe methylene groups. The opposite trend is found for clomipra-mine. Hence we concluded that the aliphatic chain of Des is ori-ented almost perpendicularly with respect to the metal whereasit is strongly tilted in clomipramine. In the case of imipramine,the aliphatic chain adopts a intermediate geometry between thoseproposed for Des and Clo. These orientations of the side chain arealso confirmed by the SERS ratio of the 1042 and 1431 cm�1 bands,see Table 7.

The second group of the tricyclic antidepressants studied here,i.e. amitriptyline, nortriptyline and doxepine, characterize the pres-ence of the double CC bond attached to the dibenzazepine system,

Fig. 4. The proposed model of the adsorption mechanism of the antidepressants (colours of atoms correspond to: gray – carbon, white – hydrogen, blue – nitrogen, green –chlorine, red – oxygen). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 7Relative intensities of selected bands in NR and SERS spectra of imipramine (Imi), desipramine (Des) and clomipramine (Clo).

I683/I1600a I683/I1060

b I689/I1042c I689/I1431

d I1042/I1431

Imi NR 0.5 2.9 1.6 7.0 4.4SERS 2.2 11.5 5.8 3.4 0.6

Des NR 0.7 2.6 1.2 5.7 4.7SERS 1.4 7.8 4.0 5.0 1.2

Clo NR 0.3 4.2 3.1 2.4 0.8SERS 1.9 11.9 8.8 2.0 0.2

a 683 cm�1: bR6/mCCR7, 1600 cm�1: mCCR6.b 1060 cm�1: mCCR61.c 1042 cm�1: mCCaliph.d 1431 cm�1: dCH2.

Table 8Relative intensities of bands in NR and SERS spectra of amitriptyline (Ami), nortriptyline (Nor) and doxepine (Dox).

I690/I1600a I690/I1640

b I690/I1042c I690/I1369

d I1042/I1369

Ami NR 0.6 0.4 2.0 2.9 1.5SERS 4.7 1.9 6.1 8.5 1.4

Nor NR 0.5 0.3 2.0 2.3 1.2SERS 3.1 1.2 4.5 3.8 0.8

Dox NR 0.6 0.4 3.7 1.7 0.5SERS 3.5 2.0 5.5 4.0 0.7

a 690 cm�1: bR6; 1600 cm�1: mCCR6.b 1640 cm�1: mC@C.c 1042 cm�1: mCCaliph.d 1369 cm�1: qCH2/aliph.

122 A. Jaworska, K. Malek / Journal of Colloid and Interface Science 431 (2014) 117–124

cf. Fig. 1. The detailed assignment of Raman and SERS bands is col-lected in Tables 4–6. Similarly to the group discussed above, theSERS spectra of those antidepressants are similar to the NR spectrain means of bands positions and they also exhibit the enormousintensification of a band at 689 and 696 cm�1 for Ami/Nor andDox, respectively (Fig. 3). First of all, the I690/I1600 ratio increases6–8 times in the SERS spectra in comparison to the Raman spectraof the solutions like for the previous group of the molecules,confirming that the dibenzazepine moiety is a primary site of the

interaction with the surface. Interestingly, this ratio is similar forSERS of nortriptyline and doxepine, despite the presence of the oxy-gen atom in the 7-membered ring of Dox, but the largest value ofthe ratio is found for amitryptiline. Thus, nortryptiline and doxe-pine adopt a less tilted orientation of the ring system in comparisonto amitryptiline since the 8a mode of the phenyl ring is moreenhanced than the bending motion of the three rings together.Moreover, according to our PED calculations, the out-of-planebending vibration of the phenyl rings conjugated with the in-plane

A. Jaworska, K. Malek / Journal of Colloid and Interface Science 431 (2014) 117–124 123

bending mode of the R7 ring is manifested by the presence of a weakband at ca. 340 cm�1 (Tables 4–6), which becomes a medium-inten-sity band in the SERS spectra of all the molecules. Therefore, one canassume that, for this group of the antidepressants, the moleculesare oriented in such a way that the benzene rings are preponder-antly tilted with respect to the surface. Along with decreasing theintensity of the 8a mode of the phenyl rings due to adsorption onthe silver nanoparticles, we also observed lowering intensity ofthe stretching vibration of the C@C bond (at 1640 cm�1) comparingto the corresponding normal Raman spectra. The intensity ratio ofthese bands (I690/I1640) in the NR spectra is similar for all moleculesindicating significantly larger intensity of the C@C stretching modethan the breathing vibration of the dibenzazepine group. Afteradsorption on the metal, the latter has an intensity twice as largeas the mC@C band in the spectra of Ami and Dox and slightly largerfor Nor (cf. Table 8). The molecular structures of the antidepressantsstudied here rather exclude a simultaneous perpendicular orienta-tion of the dibenzazepine moiety and the C@C bond with respect tothe surface (cf. Fig. 4). Thus the ratio between the 690 and1640 cm�1 bands can indicate to what extend both functionalgroups are directed perpendicularly to the metal. This ratio is sim-ilar for Ami and Dox (ca. 2) whereas it decreases to 1.2 for nortrip-tyline indicating that the double CC bond is reoriented from a tiltedto an upright geometry, respectively. An analysis of the SERS bandsattributed to vibrations of the aliphatic chain, mCC (1042 cm�1) andqCH2 (1369 cm�1), exhibits a similar enhancement of the mCC modewith respect to the 690 cm�1 band for all the antidepressantswhereas a 2-fold decrease of the I690/I1369 ratio is found for Norand Dox in comparison to the ratio values calculated for amitripty-line. In turn, the relative intensities of the 1042 and 1369 cm�1

bands clearly indicate an orientation of the side on the silver nano-particles (see Table 8). Since the I1042/I1369 ratio is similar in the NRand SERS spectra of Ami and Dox, the aliphatic chain is likely ori-ented in a way similar to the DFT-optimized structures (cf. Fig. 4and Table 8). In turn, the SERS intensity of the dCH2 mode is higherthan the intensity of the mCC vibration for Nor, suggesting that thealiphatic chain of Nor is more tilted to the surface than for Ami andDox. The reason for that could be the absence of one methyl groupin Nor and in consequence a stronger interaction of the lone elec-tron pair of the terminal N atom than in the other molecules. How-ever, the opposite effect is observed for desipramine as discussedabove and this clearly results from the different orientations ofthe dibenzazepine moiety on the silver surface.

Summarizing the proposed adsorption models for all antide-pressants studied here are illustrated in Fig. 4.

4. Conclusions

In this work, we present the orientation of the six tricyclicantidepressants on the silver nanoparticles determined by sur-face-enhanced Raman spectroscopy. We identified the primaryadsorption site on the silver nanoparticles via the p-electron sys-tem of the ring moiety and the methyl-aminopropyl side chain.The orientation of the dibenzazepine ring slightly differs betweenthe molecules pointing an effect of the complex structure of themolecules on the subtle changes in their surface behavior. Thesechanges in the orientation of the ring systems are consistent withthe interactions of TCAs within the central substrate site of theserotonin and leucine transporters in that the ring system is lyingalmost perpendicularly in the membrane plane. The alkylamineside chain is oriented similarly to the ring, however, according tomolecular dynamics simulations its spatial arrangement variesfor the particular antidepressant [2]. Several works cited in Sec-tion 1 [2,6,8,9,28] suggest that the position of the particular TCAis flexible in the binding site and in some cases it is difficult to

identified the most favorable location. However, all of them sug-gest that the tricyclic ring adopts a tilted orientation with respectto the hydrophilic pocket of the serotonin transporter similarly tothe TCAs interaction with the metal surface. The possible rotationalreorientation of 3-substituted antidepressants like clomipraminefrom the symmetric docking of imipramine in hSERT could be inthe agreement with our SERS spectra that revealed a less tilted ori-entation of the tricyclic moiety of Clo. A similar orientation of thering system of nortriptyline was identified in its SERS spectrumthat is congruent with the fact that Nor is wedged between trans-membrane helices of the dopamine transporter [8]. No detaileddata on locations of amitriptyline and doxepin within a transporterhave been reported so far in the literature, thus a direct compari-son between the SERS model of the adsorption on the metal surfaceand pharmacological action of the two antidepressants is limited.On the other hand, the proposed models of the surface adsorptionfor Ami and Dox (Fig. 4) suggest a greater preferential interactionbetween the p-electron system of the azepine ring and silver nano-particles in comparison to that observed for the ring system of nor-triptyline. Moreover, SERS features of Ami and Dox can implicatethat the presence of the oxygen atom in the azepine ring enforcesslightly different strength and/or binding fashion with the metallicsurface through this moiety. It is also worth mentioning that thebinding affinity studies of TCAs have suggested a contribution ofchloride ions into the stabilization of their docking pose for a givenreceptor conformation and they have assumed the presence of onechloride ion in the close vicinity of the antidepressant molecule[2,9]. In turn, in our SERS investigations, we observed weakeningof the adsorbate–metal interactions when an excess of chlorideions was added to the mixture of antidepressants and silver sol,taking into account that the TCA:Cl� molar ratio of 1:1 is alreadyintroduced to the solution.

Our results demonstrate an analogy between the molecularstructure and orientation of the investigated drugs when interact-ing with the colloidal silver nanoparticles and their orientationwithin the central substrate site of the transporters. This suggests,despite the simplicity of the model, that SERS can serve as a prob-ing and complimentary method for experimental pharmacologicalstudies and molecular dynamics simulations. Although, it shouldbe also pointed out that surface-enhanced Raman spectroscopydoes not allow identifying the specific amino acid residues asbinding sites in the receptors and the type of interaction like p–pstacking, hydrogen bonding or water-mediated salt bridge, but thismay rather mimic a docking pose of a ligand within a receptor.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.jcis.2014.05.060.

References

[1] M. Gupta, A. Jain, K.K. Verma, J. Sep. Sci. 33 (2010) 3774–3780.[2] S. Sinning, M. Musgaard, M. Jensen, K. Severinsen, L. Celik, H. Koldsø, T. Meyer,

M. Bols, H.H. Jensen, B. Schiøtt, O. Wiborg, J. Biol. Chem. 285 (2010) 8363–8374.

[3] K. Madej, P. Koscielniak, Crit. Rev. Anal. Chem. 38 (2) (2008) 50–66.[4] A. Jaworska, R. Wietecha-Posluszny, M. Wozniakiewicz, P. Koscielniak, K.

Malek, Analyst 136 (2011) 4704–4709.[5] G. Trachta, B. Schwarze, B. Sagmuller, G. Brehm, S. Schneider, J. Mol. Struct. 693

(2004) 175–185.[6] S. Sarker, R. Weissensteiner, I. Steiner, H.H. Sitte, G.F. Ecker, M. Freissmuth, S.

Sucic, Mol. Pharm. 78 (2010) 1026–1035.[7] S. Apparsundaram, D.J. Stockdale, R.A. Henningsen, M.E. Milla, R.S. Martin, J.

Pharmacol. Exp. Ther. 327 (2008) 982–990.[8] A. Penmatsa, K.H. Wang, E. Gouaux, Nature 503 (2013) 85–91.[9] H. Wang, A. Goehring, K.H. Wang, A. Penmatsa, R. Ressler, E. Gouaux, Nature

503 (2013) 141–145.[10] C.B. Fox, J.M. Harris, J. Raman Spectrosc. 41 (2010) 498–507.

124 A. Jaworska, K. Malek / Journal of Colloid and Interface Science 431 (2014) 117–124

[11] S. Schlücker, Surface Enhanced Raman Spectroscopy. Analytical, Biophysicaland Life Science Applications, Wiley-VCH, Weinheim, Germany, 2011.

[12] J. Bukowska, P. Piotrowski, Optical Spectroscopy and Computational Methodsin Biology and Medicine, in: M. Baranska (Ed.), Springer Science, Dordrecht,2014, pp. 29–60.

[13] N. Leopold, B. Lendl, J. Phys. Chem. B 107 (2003) 5723–5727.[14] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman,

G. Scalmani, V. Barone, B. Mennucci, G.A. Petersson, H. Nakatsuji, M. Caricato,X. Li, H.P. Hratchian, A.F. Izmaylov, J. Bloino, G. Zheng, J.L. Sonnenberg, M.Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y.Honda, O. Kitao, H. Nakai, T. Vreven, J.A. Montgomery Jr., J.E. Peralta, F. Ogliaro,M. Bearpark, J.J. Heyd, E. Brothers, K.N. Kudin, V.N. Staroverov, R. Kobayashi, J.Normand, K. Raghavachari, A. Rendell, J.C. Burant, S.S. Iyengar, J. Tomasi, M.Cossi, N. Rega, J.M. Millam, M. Klene, J.E. Knox, J.B. Cross, V. Bakken, C. Adamo,J. Jaramillo, R. Gomperts, R.E. Stratmann, O. Yazyev, A.J. Austin, R. Cammi, C.Pomelli, J.W. Ochterski, R.L. Martin, K. Morokuma, V.G. Zakrzewski, G.A. Voth,P. Salvador, J.J. Dannenberg, S. Dapprich, A.D. Daniels, Ö. Farkas, J.B. Foresman,J.V. Ortiz, J. Cioslowski, D.J. Fox, Gaussian 09, Revision A.1, Gaussian Inc.,Wallingford, CT, 2009.

[15] C. Lee, W. Yang, R.G. Parr, Phys. Rev. B 37 (1988) 785–789.

[16] A.D. Becke, J. Chem. Phys. 98 (1993) 5648–5652.[17] D. Michalska, R. Wysokinski, Chem. Phys. Lett. 403 (2005) 211–217.[18] J.M.L. Martin, C. Van Alsenoy, Gar2ped, University of Antwerp, Antwerp,

Belgium, 1995.[19] G. Fogarasi, X. Zhou, P.W. Taylor, P. Pulay, J. Am. Chem. Soc. 114 (1992) 8191–

8201.[20] P. Pulay, G. Fogarasi, F. Pang, J.E. Boggs, J. Am. Chem. Soc. 101 (1979) 2550–

2560.[21] S. Sanchez-Cortes, J.V. Garcia-Ramos, Surf. Sci. 473 (2001) 133–142.[22] G. Varsanyi, Assignments for Vibrational Spectra of 700 Benzene Derivatives,

Akademiai Kiado, Budapest, 1973.[23] J.R. Lombardi, R.L. Birke, Acc. Chem. Res. 42 (6) (2009) 734–742.[24] K. Malek, M. Makowski, A. Królikowska, J. Bukowska, J. Phys. Chem. B 116

(2012) 1414–1425.[25] K.M. Marzec, A. Jaworska, K. Malek, A. Kaczor, M. Baranska, J. Raman Spectrosc.

44 (1) (2013) 155–165.[26] M. Larsson, J. Lindgren, J. Raman Spectrosc. 36 (5) (2005) 394–399.[27] X. Gao, J.P. Davies, M.J. Weaver, J. Phys. Chem. 94 (1990) 6858–6864.[28] S. Apparsundaram, D.J. Stockdale, R.A. Henningsen, M.E. Milla, R.S. Martin, J.

Pharm. Exp. Ther. 327 (2008) 982–990.