Colloids and Surfaces B: Biointerfaces 134 (2015) 8–16
Contents lists available at ScienceDirect
Colloids and Surfaces B: Biointerfaces journal homepage: www.elsevier.com/locate/colsurfb
Physical factors affecting chloroquine binding to melanin R.L. Schroeder, P. Pendleton, J.P. Gerber ∗ School of Pharmacy and Medical Sciences, University of South Australia, North Terrace, Adelaide 5000, Australia
a r t i c l e
i n f o
Article history: Received 26 February 2015 Received in revised form 16 June 2015 Accepted 17 June 2015 Available online 25 June 2015 Keywords: Melanin Sepia Chloroquine Adsorption Isosteric heat pH
a b s t r a c t Chloroquine is an antimalarial drug but is also prescribed for conditions such as rheumatoid arthritis. Long-term users risk toxic side effects, including retinopathy, thought to be caused by chloroquine accumulation on ocular melanin. Although the binding potential of chloroquine to melanin has been investigated previously, our study is the first to demonstrate clear links between chloroquine adsorption by melanin and system factors including temperature, pH, melanin type, and particle size. In the current work, two Sepia melanins were compared with bovine eye as a representative mammalian melanin. Increasing the surface anionic character due to a pH change from 4.7 to 7.4 increased each melanin’s affinity for chloroquine. Although the chloroquine isotherms exhibited an apparently strong interaction with each melanin, isosteric heat analysis indicated a competitive interaction. Buffer solution cations competed effectively at low surface coverage; chloroquine adsorption occurs via buffer cation displacement and is promoted by temperature-influenced secondary structure swelling. © 2015 Elsevier B.V. All rights reserved.
1. Introduction The association between ingested drugs and melanin in the body has been the subject of many studies over the last 50 years. For example, melanin has a strong affinity towards compounds such as phenothiazines [1–6], antibiotics [1], antimalarials [2–6], antirheumatics [2], antifolates [1], as well as various illicit drugs [7–9] and herbicides [10], with each causing various adverse effects. Due to these interactions, various negative associations have been implicated with some compounds, resulting in outcomes unrelated to the main action of the drug. Chloroquine has been used widely for the treatment of malaria and other conditions, including rheumatoid arthritis, discoid lupus erythematosus and amoebic hepatitis, and has also been associated with delayed-onset retinopathy [11,12]. The present study focused on gaining a greater insight into the thermodynamics and mechanism of the melanin–chloroquine interaction. It is anticipated such knowledge would be connected to the associated pathologies in future research. The accumulation of a particular drug and its mode of action ultimately determines its toxicity [13]. An irreversibly bound xenobiotic will not necessarily cause adverse effects, but rather depends on tissue or organ specificity. If no harm arises from
∗ Corresponding author. Tel.: +61 8 8302 2568; fax: +61 8 8302 1087. E-mail addresses: [email protected] (R.L. Schroeder), [email protected] (P. Pendleton), [email protected] (J.P. Gerber). http://dx.doi.org/10.1016/j.colsurfb.2015.06.040 0927-7765/© 2015 Elsevier B.V. All rights reserved.
the interaction, a substance such as melanin would simply act as a deactivating reservoir. However, the injury arises when accumulation (as localised adsorption) occurs in regions of the body where the released drug could cause damage [13]. The correlation between the percentage of tissue accumulation and the amount of pigment in the eye suggests a relationship may exist between retinopathy and melanin. It was proposed that retinopathy caused by chloroquine was either due to its adsorption to ocular melanin [14] or due to another pathophysiological pathway caused by chloroquine itself [15]. A third hypothesis suggested that chloroquine chemisorbs with melanin possibly leading to retinal toxicity through an alteration of the melanin surface, and subsequently undermining the protective role of melanin as a free radical scavenger [16]. Either way, melanin is perceived to play a role. The structural analogue of chloroquine, hydroxychloroquine, is less toxic and has not been linked to retinopathy. We have shown that this reduced toxicity might be interpreted in terms of its adsorption mechanism. Hydroxychloroquine adsorbs through weaker interactions than chloroquine to different types of melanin [17], suggesting that adsorption to melanin may be a deciding factor in determining toxicity. Organic melanin production is highly regulated in melanosomes. The melanogenetic pathway begins with the hydroxylation of tyrosine to dopa through the action of the ratelimiting enzyme tyrosinase. Dopa progresses through a series of reactions from the oxidation to dopaquinone, which then cyclises to dopachrome. At this point, the intracellular conditions dictate the type of melanin produced. Dihydroxyindole (DHI) is formed
R.L. Schroeder et al. / Colloids and Surfaces B: Biointerfaces 134 (2015) 8–16
by the decarboxylation of dopachrome, whereas the presence of dopachrome tautomerase (Dct – a tyrosinase related protein) structurally rearranges dopachrome to produce dihydroxyindole carboxylic acid (DHICA) [18–20]. DHI and DHICA are the primary building blocks (as monomers) of melanin. This process is well described by the Raper-Mason scheme [21–26]. Different classes of melanin are produced through the varied incorporation of the two monomers, DHI and DHICA. As dopachrome tautomerase concentration determines DHICA amounts, the ratio can vary depending on the location of the melanin production within the body, as well as being responsible for interspecies differences. Generally, naturally sourced (organic) melanin is lower molecular weight than synthetically produced, and thus a lower number of ionisable groups [27,28]. Additionally, there are differences between the ratios of DHI and DHICA that make up the primary structure of melanin. Generally, synthetic melanins are either all one monomer type, or the DHI:DHICA ratio is close to 1:1; variations can be apparent in organic melanins [29]. The actual ratio is difficult to ascertain. Most studies base the ratio on oxidative degradation of DHI and DHICA to pyrrole di- and tricarboxylic acid (PDCA and PTCA), respectively. However, interpretations of these outcomes are confused by the compounds also being found in the bound form in the native melanin [30]. Nevertheless, Sepia melanin appears to contain about double the amount of DHICA compared to bovine eye, while DHI is approximately at similar levels. A recent review suggests that the ratio is closer to 1:1 in most natural melanins [31]. However, this ratio appears unlikely considering the control of melanogenesis in melanocytes by dopachrome tautomerase [32]. When the enzyme is abundant, DHICA formation is favoured, suggesting the location of the melanocyte and the species involved will play a role in ultimately regulating the DHI:DHICA ratio [19]. The secondary structure is also very important as it contains a series of layers in a stacked, planar configuration. Each layer has been suggested to consist of monomers arranged in porphyrin-like structures [33] with the internal arrangement simulating graphene [34,35] (see Supplementary Information Fig. S1). Results from XRD and TEM studies have determined that the pigment has a multilayer 3D arrangement thought to comprise of dihydroxyindole rings and benzothiazine residues. These structure form highly cross-linked oligomers which aggregate through -stacking in layers of 3–4 [31,19]. The spacing ˚ with the variation between each layer is approximately 3.7–4.7 A, due to DHI:DHICA ratio (i.e., the type of melanin) and/or the method of analysis [27,36,37]. The range can be attributed to the specific location on the structure. The closer to the centre of the granule, the wider the spacing between each layer needs to be to accommodate the curvature of the oligomeric plates [37]. Following further hierarchical aggregation, these smaller units assemble to result in spherical structures of approximately 150 nm in diameter. An important consideration for adsorption studies is the available surface area to maximise the capacity for interacting with an adsorptive. Therefore, any porosity in an adsorbent would greatly increase the surface area to volume ratio and thus generate greater binding potential. Nitrogen gas adsorption studies have been conducted on melanin to determine porosity. Melanin is quite commonly compared to graphite in relation to the arrangement of the individual layers within the particle. It was found that Sepia melanin had a specific surface area of approximately 25 m2 /g [38]. To put this in perspective, graphite has a surface area approximating 4.8 m2 /g [39] whereas porous materials such as activated carbons can exceed 1000 m2 /g [40]. Therefore, it can be concluded that melanin is largely non-porous and thus the surface characteristics of the pigment are primarily of concern when conducting binding experiments. To date, the binding potential of melanin for chloroquine has been examined quite extensively [6,41–44]. These previous
9
studies demonstrated that the relationship between melanin and chloroquine could be fitted by the Langmuir isotherm, representing monolayer equivalent adsorption [41]. Interestingly, beyond simple adsorption, other factors associated with chloroquine adsorption such as changes in system pH and temperature or the composition of melanin have not been explored conclusively. From the previous studies, it is clear three main binding interactions occur between melanin and chloroquine: electrostatic, hydrophobic and Van der Waals [6,41–43]. As a general consensus, electrostatic interactions appeared to be primarily involved in chloroquine binding to melanin, occurring at both strong and weak adsorption sites. Relatively weak ionic interactions tended to exist between the carboxyl group on melanin and chloroquine. In addition, acidic solution phase pH caused phenolic group protonation and a consequent reduction of the surface concentration of available strong binding sites [43]. From a pharmacological perspective, Tsuchiya et al. interpreted their mechanism as a decrease in percentage adsorbed with increasing pH [44]. The melanin used in their study was prepared under extreme conditions, by soaking in 11 N HCl, which has been claimed to destroy the adsorption properties of the pigment [45]. In the previous studies examining the binding relationship between melanin and the compound of interest, the influence of the type and source of melanin was ignored. The review of numerous unrelated studies described by Derby [46] demonstrated that melanin properties can differ depending on the isolation method, the means of separation from tissue (when present as an integral part, for example neuromelanin) and, most significantly, its source [46,47]. Therefore, one of the primary aims of this study was to establish the effect of the type of melanin on chloroquine adsorption. Madaras et al. demonstrated how those proteins widely accepted to be part of melanin, are in all probability an artefact from surrounding tissue during isolation [48]. These proteins could therefore interfere with the adsorption of solutes onto the melanin surface [9,41,43,49]. To expand on this set of work, the current study further explored the affinity of melanin for chloroquine by examining adsorption as a function of organic melanin type, and also system pH and temperature. The DHI:DHICA ratios were explored as this is a much debated topic in the literature. No previous studies have investigated the effects of pH or temperature, or even the type of melanin. The pH values selected were 4.7 and 7.4 as these spanned the pKa values of each species to be considered: DHI; DHICA; and, chloroquine. The temperature range included a physiological value to promote applicability of the results. Together, the results provide a novel insight into the sorption characteristics of melanin for chloroquine.
2. Materials and methods 2.1. Melanin sample preparation Two types of cuttlefish melanin were used in this study. The first was purchased as Tinta CalamarTM from SAMTAS, Adelaide, containing melanin from the European Common Cuttlefish, Sepia officinalis and was the result of the blending of multiple cuttlefish ink sacs, including the tissue, prior to packaging. This melanin will be referred to as S. officinalis melanin. A second melanin was extracted from the ink sacs of native Australian cuttlefish, primarily Sepia apama with a small proportion of Sepia novaehollandiae (Valente Seafoods, South Australia, Australia and SARDI, South Australia, Australia) and prepared according to Schroeder and Gerber [50]. Due to the low levels of Sepia novaehollandiae present, it is assumed that this melanin would behave like Sepia apama melanin, and thus will be referred to as Sepia melanin.
10
R.L. Schroeder et al. / Colloids and Surfaces B: Biointerfaces 134 (2015) 8–16
Furthermore, bovine (Bos taurus) eye melanin was used to compare the binding capacity of mammalian melanin with that from cuttlefish. This melanin was extracted from the eyes of freshly slaughtered cattle according to the method of Tsuchiya et al. [44]. Bovine eye melanin was used as a representative mammalian melanin due to its accessibility and ease of comparison with previous studies. Both commercial S. officinalis melanin and the bovine eye melanin were pre-treated to remove all soluble contaminants, including free and bound lipids and protein. The purification entailed the consecutive treatment of approximately 1 g of crude, recovered melanin with 50 mL aliquots of 0.9 wt.% saline solution, 0.01 M HCl, 0.01 M NaOH, Tween 20, acetone, and MilliQ water. Thereafter, adsorbed or bound proteins were removed by reacting the melanin residue over a 24 h period with 20 mL aliquots of 20 mg pepsin (hydrolyses the peptide bond between hydrophobic and aromatic amino acids) at pH 1, followed by 10 mg papain activated with cysteine (cysteine proteinase that breaks peptide bonds at hydrophobic residues) at pH 5.5 and 10 mg trypsin (cleaves the carboxyl side of both the amino acids lysine and arginine residues) at pH 8. Final rinses with saline and MilliQ water and freeze drying of the sample resulted in melanin as a fine, black powder. 2.2. Chemicals Potassium dihydrogen orthophosphate, acetic acid, sodium acetate trihydrate and dichloromethane were obtained from Merck Pty. Ltd (Australia). Sodium chloride and Triton-X 100 were obtained from VWR International Pty. Ltd. (Australia). Chloroquine diphosphate, and the enzymes pepsin, papain and trypsin were sourced from Sigma–Aldrich (USA). All chemicals had a purity greater than 98 wt.% and were used without further purification. 2.3. Preparation of stock solutions Sodium acetate (pH 4.7) and potassium phosphate (pH 7.4) buffer solutions were prepared fresh at 20 mmol/L concentrations prior to use. A stock solution of 3.1 mmol/L chloroquine was prepared in the relevant buffer solution. The acetate and phosphate buffer solutions were used as blanks to correct for background in spectrophotometric measurements. 2.4. Instrumentation A Varian Cary 50 ultraviolet/visible (UV/Vis) spectrophotometer was used to determine chloroquine concentrations. Solution pH was measured using a Hanna potentiometer (Model: pH211 microprocessor pH metre) fitted with a microelectrode and calibrated at pH 4 and 7. 2.5. Calibration curve Chloroquine solutions were prepared ranging from 0.0033 to 0.33 mmol/L by diluting the stock solution with 20 mmol/L acetate buffer at pH 4.7. A similar range was also prepared in a phosphate buffer at pH 7.4 to generate standard curves of absorbance at 255 nm against chloroquine concentration. All samples were diluted with the appropriate buffer for analysis to bring readings within the range of the spectrophotometer. Standard curves were produced in triplicate in order to determine the inter-day reproducibility. 2.6. Equilibration time for adsorption An amount of 5.0 mg of melanin was suspended in 5 mL of 0.63 mmol/L chloroquine at a pH of 4.7. Separate tubes were placed
on an off-vertical (10◦ ), rotary wheel, rotating end-over-end continuously at 40 rpm for the equilibration time periods of 10, 30, 60, 120, and 240 min. After each equilibration time, the samples were centrifuged at 4770 × g for 4 min and the chloroquine concentration in the supernatant was determined by comparison with the standard curve using UV/Vis spectrophotometry at 255 nm. Solutions were measured against a melanin blank solution background, which was centrifuged prior to being added to the cuvette. The same procedure was employed for all subsequent analyses. This process was repeated at pH 7.4. 2.7. Adsorption of chloroquine to Sepia melanin To determine the adsorption capacity of melanin, a series of concentrations of chloroquine were prepared ranging from 0.16 to 2.2 mmol/L. A 5.0 mg portion of melanin was equilibrated with 5 mL of the specified concentration of solute prepared in either 20 mmol/L sodium acetate buffer (pH 4.7) or potassium phosphate buffer (pH 7.4) for 2 h, using an end-over-end mixing method as described above. The samples were centrifuged at 4770 × g for 4 min and the supernatant removed and analysed using UV/Vis spectrophotometry at 255 nm. Results were obtained for chloroquine binding using Sepia and S. officinalis melanin. Adsorption systems were prepared in triplicate by adding the same masses of melanin to the same volume and concentration of chloroquine solution. 2.8. Adsorption of chloroquine to bovine eye melanin A 5.0 mg sample of bovine eye melanin was equilibrated with chloroquine concentrations ranging from 0.078 to 1.6 mmol/L for 2 h, using an end-over-end mixing method as described above. Samples were centrifuged at 2687 × g for 15 min and the supernatant removed. UV/Vis spectrophotometry was used for analysis using a wavelength of 255 nm. The amount of chloroquine adsorbed to bovine eye melanin was compared with Sepia and S. officinalis melanins. Adsorption systems were prepared in triplicate as described above. 2.9. Temperature dependence of adsorption of chloroquine to Sepia melanin To determine system temperature influence on the adsorption capacity of melanin, a series of concentrations of chloroquine were prepared ranging from 0.16–2.2 mmol/L. A 5.0 mg portion of melanin was equilibrated with 5 mL of the specified concentration of solute prepared in potassium phosphate buffer (pH 7.4) for 2 h at 277 K, 291 K and 310 K in addition to the previous result at an ambient temperature of 294 K. After equilibration via an endover-end mixing method as described above, the samples were centrifuged at 4770 × g for 4 min and the supernatant removed and analysed at 255 nm. Adsorption systems were prepared in triplicate as described above. 3. Results and discussion Chloroquine is a compound that accumulates in the eyes of patients using it as a therapeutic agent. We recently demonstrated that the affinity of Sepia melanin for chloroquine is greater than that for hydroxychloroquine [17]. Since retinopathy is one of the main concerns regarding the use of chloroquine, our next aim was to assess whether the type of melanin had an influence on the adsorption behaviour. With melanin isolated from representative mammalian bovine eyes, one cannot rule out some contamination with residual protein or structural changes due to the recovery process. Therefore, Sepia melanin was also examined to establish if
R.L. Schroeder et al. / Colloids and Surfaces B: Biointerfaces 134 (2015) 8–16
3.1. Comparison of melanin between different genera – effect of pH To characterise the relationship between melanin and chloroquine, adsorption studies were performed under a variety of conditions using different melanins. Two pH levels were examined to determine how this would affect the affinity between the adsorbent and adsorbate. Standard curves of absorbance against solute concentration were generated for each pH level (4.7 and 7.4). The coefficient of variation (CV) i.e. the quotient of the standard deviation and the mean, or the relative uncertainty, of the slopes for each standard curve (n = 3) was less than 1.05%, with each standard concentration having a CV of less than 3.00%. Each curve gave a linear fit of at least 99.9%. The lower limit of quantification at pH 4.7 was 3.3 mol/L, and 3.1 mol/L at pH 7.4, both with negligible variation. Both of the chloroquine solution adsorption isotherms for each melanin adsorbent were a type L2 [51], suggesting a monolayer equivalent amount adsorbed at equilibrium solution concentrations Ceq /(mmol/L) > 0.5. Consequently, each isotherm was well-defined via the Langmuir isotherm model, Eq. (1). To overcome departure from linearity in a (traditional) linear least-squares analysis at either low or relatively high equilibrium concentrations, the coefficients qm and KL (as the monolayer equivalent amount adsorbed and the Langmuir (equilibrium) coefficient) were determined by fitting the adsorption data to a least squares analysis to fit the entire data set, with the coefficients set to initial values of 0.01. The fitting procedure produced reproducible coefficient values after minimising the average of the sum of the squares of the difference between measured and calculated amounts adsorbed, and maximising the overall Pearson correlation coefficient. An exemplar plot and analysis is given in Fig. S2 and Table S2 (see Supplementary Information). q=
qm KL Ceq 1 + KL Ceq
(1)
Due to the implications of chloroquine adsorption to uveal melanin in retinopathy, bovine eye melanin was used as a representative mammalian melanin. The extent of chloroquine binding was compared with that of S. officinalis melanin to determine if changing to mammalian melanin would give different results. Both melanin types were subjected to a series of chemical ‘washes’ (see Section 2) to aid in purifying the mixture from associated cellular components. The adsorption capacities were determined at pH 4.7 and 7.4, with the results shown in Fig. 1. Although a third system pH value
12000 10000 8000 Amt. ads*
it could be considered as a reference point. As mentioned in the Introduction, melanins from different sources have been used in previous binding experiments, and the effect this difference has on the obtained results has yet to be examined. Two types of Sepia melanin were used in these experiments, as well as a comparison with bovine eye melanin. S. officinalis melanin is readily available as a food additive and was purchased as crudely-extracted ink from the European common cuttlefish. Before unequivocal answers regarding surface properties could be developed via adsorption measurements, it was also essential to determine if differences existed in adsorption properties of melanins extracted from the same genus, cuttlefish, as well as interspecies sources being compared. As with most naturally occurring materials, organically sourced melanins were suspected to contain a relatively high proportion of associated proteins. The aggressive purification procedure described above rendered melanin samples with consistent purity, raising the reproducibility of subsequent measurements and the anticipation that our results would lead to improved interpretation of future pharmacological pathway research.
11
6000 4000 2000 0 0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Ceq/(mmol/L) Fig. 1. Adsorption isotherm of chloroquine for S. officinalis (SO) and bovine eye (BE) SO at pH 4.7, BE at pH 7.4, and melanin. The 䊏 corresponds to SO at pH 7.4, BE at pH 4.7. * Units of amount adsorbed are mol/mole of melanin equivalent to 1 unit of (DHI1 :DHICA40 )n=1 for SO and (DHI1 :DHICA25 )n=1 for BE. The uncertainty bars result from the standard deviation of triplicate measurements. At pH 4.7, the adsorptive capacity of S. officinalis melanin for chloroquine was found to be 7244 mol/mole (0.92 mmol/g), significantly lower (43%) than at physiological pH. In addition, bovine eye melanin also had a significantly lower (44%) capacity for chloroquine at the lower pH with 2788 mol/mole (0.56 mmol/g). This difference may be explained by the ionisation states of DHI and DHICA, as well as that of chloroquine.
would allow plotting selected parameters as f(pH), to maintain the physiological relevance of this work, one would need to consider a value within this range, such as 6.0. The speciation data in Table 1 are reproduced for pH 6.0 in the Supplementary Information; they suggest negligible difference in principal species’ concentration. Comparison of the chloroquine adsorption isotherms for bovine eye melanin with those for S. officinalis shows the former has a reduced affinity under both sets of conditions at any equilibrium solution concentration. To allow this comparison to be made between the two melanin species, each was converted into micromoles per unit area. This enabled the concentration available for adsorption in 5 mg equivalent of particles, per unit area of a single particle, to be calculated. A single particle was used as the density of each melanin species is undefined. At the acidic pH, the carboxyl moiety of the DHICA monomer would be approximately 70% ionised, whereas it would have been completely ionised at physiological pH [52]. DHI would have no charge at both pH values. In addition, at low pH, it has previously been demonstrated that there is a reduction in the number of strong binding sites, thus leading to a decreased affinity seen with these results [43] (Fig. 2). The addition of buffer solute slightly reduced the ionisation percent of both DHI and DHICA. Without buffer, DHICA is 99% ionised at pH 7.4, whereas with buffer this percentage drops to 95% (Fig. 3a). The implication for DHI is less apparent, with a decrease of only 2%. The addition of buffer to chloroquine has little effect on the ionisation state (Fig. 3b). Chloroquine would be triply charged at pH 4.7, but only have two positive charges at pH 7.4. The differences in ionisation states of the melanin surface may result in significant electrostatic attraction between the positively charged chloroquine and anionic DHICA moiety, potentially explaining the sensitivity of chloroquine–melanin adsorption to changes in pH. Reduced aqueous solubility of chloroquine at elevated pH might similarly be expected to favour adsorption. The pH effect was also observed in a rice husk study conducted by Chowdhury et al. [53] which described the changes in the surface characteristics of rice husks with increasing pH. It was noted that the pores on the surface of the rice husks increased in diameter
12
R.L. Schroeder et al. / Colloids and Surfaces B: Biointerfaces 134 (2015) 8–16
Fig. 2. Ionisation states of chloroquine at pH 4.7 (a) and pH 7.4 (b), and ionisation states of DHI (c) and DHICA (d) at pH 7.4.
Fig. 3. Comparison of ionic strength-modulation of fractional dissociation with change in pH for (a) DHICA and (b) chloroquine (CQ). The solid line is without phosphate buffer, whereas the dotted line is with phosphate buffer.
R.L. Schroeder et al. / Colloids and Surfaces B: Biointerfaces 134 (2015) 8–16
13
Table 1 Mole fraction charge for DHI and DHICA at pH 7.4 and 4.7. Monomer
Dissociation Fraction**
Charge
S. officinalis moles/unit area (m2 )*
Bovine Eye moles/unit area (m2 )*
pH 7.4 DHI DHI− DHICA− DHICA2−
0.99 0.01 0.99 0.01
0 −1 −1 −2
36.4 0.4 94.5 0.8
2.9 0.03 3.8 0.03
pH 4.7 DHI+ DHI DHICA DHICA−
0.02 0.98 0.21 0.78
1 0 0 −1
0.6 36.2 20.3 74.7
0.05 2.9 0.8 3.0
* **
The relative uncertainty in surface area concentration = 4.1% [57]. All other species were evaluated and were significantly
Contents lists available at ScienceDirect
Colloids and Surfaces B: Biointerfaces journal homepage: www.elsevier.com/locate/colsurfb
Physical factors affecting chloroquine binding to melanin R.L. Schroeder, P. Pendleton, J.P. Gerber ∗ School of Pharmacy and Medical Sciences, University of South Australia, North Terrace, Adelaide 5000, Australia
a r t i c l e
i n f o
Article history: Received 26 February 2015 Received in revised form 16 June 2015 Accepted 17 June 2015 Available online 25 June 2015 Keywords: Melanin Sepia Chloroquine Adsorption Isosteric heat pH
a b s t r a c t Chloroquine is an antimalarial drug but is also prescribed for conditions such as rheumatoid arthritis. Long-term users risk toxic side effects, including retinopathy, thought to be caused by chloroquine accumulation on ocular melanin. Although the binding potential of chloroquine to melanin has been investigated previously, our study is the first to demonstrate clear links between chloroquine adsorption by melanin and system factors including temperature, pH, melanin type, and particle size. In the current work, two Sepia melanins were compared with bovine eye as a representative mammalian melanin. Increasing the surface anionic character due to a pH change from 4.7 to 7.4 increased each melanin’s affinity for chloroquine. Although the chloroquine isotherms exhibited an apparently strong interaction with each melanin, isosteric heat analysis indicated a competitive interaction. Buffer solution cations competed effectively at low surface coverage; chloroquine adsorption occurs via buffer cation displacement and is promoted by temperature-influenced secondary structure swelling. © 2015 Elsevier B.V. All rights reserved.
1. Introduction The association between ingested drugs and melanin in the body has been the subject of many studies over the last 50 years. For example, melanin has a strong affinity towards compounds such as phenothiazines [1–6], antibiotics [1], antimalarials [2–6], antirheumatics [2], antifolates [1], as well as various illicit drugs [7–9] and herbicides [10], with each causing various adverse effects. Due to these interactions, various negative associations have been implicated with some compounds, resulting in outcomes unrelated to the main action of the drug. Chloroquine has been used widely for the treatment of malaria and other conditions, including rheumatoid arthritis, discoid lupus erythematosus and amoebic hepatitis, and has also been associated with delayed-onset retinopathy [11,12]. The present study focused on gaining a greater insight into the thermodynamics and mechanism of the melanin–chloroquine interaction. It is anticipated such knowledge would be connected to the associated pathologies in future research. The accumulation of a particular drug and its mode of action ultimately determines its toxicity [13]. An irreversibly bound xenobiotic will not necessarily cause adverse effects, but rather depends on tissue or organ specificity. If no harm arises from
∗ Corresponding author. Tel.: +61 8 8302 2568; fax: +61 8 8302 1087. E-mail addresses: [email protected] (R.L. Schroeder), [email protected] (P. Pendleton), [email protected] (J.P. Gerber). http://dx.doi.org/10.1016/j.colsurfb.2015.06.040 0927-7765/© 2015 Elsevier B.V. All rights reserved.
the interaction, a substance such as melanin would simply act as a deactivating reservoir. However, the injury arises when accumulation (as localised adsorption) occurs in regions of the body where the released drug could cause damage [13]. The correlation between the percentage of tissue accumulation and the amount of pigment in the eye suggests a relationship may exist between retinopathy and melanin. It was proposed that retinopathy caused by chloroquine was either due to its adsorption to ocular melanin [14] or due to another pathophysiological pathway caused by chloroquine itself [15]. A third hypothesis suggested that chloroquine chemisorbs with melanin possibly leading to retinal toxicity through an alteration of the melanin surface, and subsequently undermining the protective role of melanin as a free radical scavenger [16]. Either way, melanin is perceived to play a role. The structural analogue of chloroquine, hydroxychloroquine, is less toxic and has not been linked to retinopathy. We have shown that this reduced toxicity might be interpreted in terms of its adsorption mechanism. Hydroxychloroquine adsorbs through weaker interactions than chloroquine to different types of melanin [17], suggesting that adsorption to melanin may be a deciding factor in determining toxicity. Organic melanin production is highly regulated in melanosomes. The melanogenetic pathway begins with the hydroxylation of tyrosine to dopa through the action of the ratelimiting enzyme tyrosinase. Dopa progresses through a series of reactions from the oxidation to dopaquinone, which then cyclises to dopachrome. At this point, the intracellular conditions dictate the type of melanin produced. Dihydroxyindole (DHI) is formed
R.L. Schroeder et al. / Colloids and Surfaces B: Biointerfaces 134 (2015) 8–16
by the decarboxylation of dopachrome, whereas the presence of dopachrome tautomerase (Dct – a tyrosinase related protein) structurally rearranges dopachrome to produce dihydroxyindole carboxylic acid (DHICA) [18–20]. DHI and DHICA are the primary building blocks (as monomers) of melanin. This process is well described by the Raper-Mason scheme [21–26]. Different classes of melanin are produced through the varied incorporation of the two monomers, DHI and DHICA. As dopachrome tautomerase concentration determines DHICA amounts, the ratio can vary depending on the location of the melanin production within the body, as well as being responsible for interspecies differences. Generally, naturally sourced (organic) melanin is lower molecular weight than synthetically produced, and thus a lower number of ionisable groups [27,28]. Additionally, there are differences between the ratios of DHI and DHICA that make up the primary structure of melanin. Generally, synthetic melanins are either all one monomer type, or the DHI:DHICA ratio is close to 1:1; variations can be apparent in organic melanins [29]. The actual ratio is difficult to ascertain. Most studies base the ratio on oxidative degradation of DHI and DHICA to pyrrole di- and tricarboxylic acid (PDCA and PTCA), respectively. However, interpretations of these outcomes are confused by the compounds also being found in the bound form in the native melanin [30]. Nevertheless, Sepia melanin appears to contain about double the amount of DHICA compared to bovine eye, while DHI is approximately at similar levels. A recent review suggests that the ratio is closer to 1:1 in most natural melanins [31]. However, this ratio appears unlikely considering the control of melanogenesis in melanocytes by dopachrome tautomerase [32]. When the enzyme is abundant, DHICA formation is favoured, suggesting the location of the melanocyte and the species involved will play a role in ultimately regulating the DHI:DHICA ratio [19]. The secondary structure is also very important as it contains a series of layers in a stacked, planar configuration. Each layer has been suggested to consist of monomers arranged in porphyrin-like structures [33] with the internal arrangement simulating graphene [34,35] (see Supplementary Information Fig. S1). Results from XRD and TEM studies have determined that the pigment has a multilayer 3D arrangement thought to comprise of dihydroxyindole rings and benzothiazine residues. These structure form highly cross-linked oligomers which aggregate through -stacking in layers of 3–4 [31,19]. The spacing ˚ with the variation between each layer is approximately 3.7–4.7 A, due to DHI:DHICA ratio (i.e., the type of melanin) and/or the method of analysis [27,36,37]. The range can be attributed to the specific location on the structure. The closer to the centre of the granule, the wider the spacing between each layer needs to be to accommodate the curvature of the oligomeric plates [37]. Following further hierarchical aggregation, these smaller units assemble to result in spherical structures of approximately 150 nm in diameter. An important consideration for adsorption studies is the available surface area to maximise the capacity for interacting with an adsorptive. Therefore, any porosity in an adsorbent would greatly increase the surface area to volume ratio and thus generate greater binding potential. Nitrogen gas adsorption studies have been conducted on melanin to determine porosity. Melanin is quite commonly compared to graphite in relation to the arrangement of the individual layers within the particle. It was found that Sepia melanin had a specific surface area of approximately 25 m2 /g [38]. To put this in perspective, graphite has a surface area approximating 4.8 m2 /g [39] whereas porous materials such as activated carbons can exceed 1000 m2 /g [40]. Therefore, it can be concluded that melanin is largely non-porous and thus the surface characteristics of the pigment are primarily of concern when conducting binding experiments. To date, the binding potential of melanin for chloroquine has been examined quite extensively [6,41–44]. These previous
9
studies demonstrated that the relationship between melanin and chloroquine could be fitted by the Langmuir isotherm, representing monolayer equivalent adsorption [41]. Interestingly, beyond simple adsorption, other factors associated with chloroquine adsorption such as changes in system pH and temperature or the composition of melanin have not been explored conclusively. From the previous studies, it is clear three main binding interactions occur between melanin and chloroquine: electrostatic, hydrophobic and Van der Waals [6,41–43]. As a general consensus, electrostatic interactions appeared to be primarily involved in chloroquine binding to melanin, occurring at both strong and weak adsorption sites. Relatively weak ionic interactions tended to exist between the carboxyl group on melanin and chloroquine. In addition, acidic solution phase pH caused phenolic group protonation and a consequent reduction of the surface concentration of available strong binding sites [43]. From a pharmacological perspective, Tsuchiya et al. interpreted their mechanism as a decrease in percentage adsorbed with increasing pH [44]. The melanin used in their study was prepared under extreme conditions, by soaking in 11 N HCl, which has been claimed to destroy the adsorption properties of the pigment [45]. In the previous studies examining the binding relationship between melanin and the compound of interest, the influence of the type and source of melanin was ignored. The review of numerous unrelated studies described by Derby [46] demonstrated that melanin properties can differ depending on the isolation method, the means of separation from tissue (when present as an integral part, for example neuromelanin) and, most significantly, its source [46,47]. Therefore, one of the primary aims of this study was to establish the effect of the type of melanin on chloroquine adsorption. Madaras et al. demonstrated how those proteins widely accepted to be part of melanin, are in all probability an artefact from surrounding tissue during isolation [48]. These proteins could therefore interfere with the adsorption of solutes onto the melanin surface [9,41,43,49]. To expand on this set of work, the current study further explored the affinity of melanin for chloroquine by examining adsorption as a function of organic melanin type, and also system pH and temperature. The DHI:DHICA ratios were explored as this is a much debated topic in the literature. No previous studies have investigated the effects of pH or temperature, or even the type of melanin. The pH values selected were 4.7 and 7.4 as these spanned the pKa values of each species to be considered: DHI; DHICA; and, chloroquine. The temperature range included a physiological value to promote applicability of the results. Together, the results provide a novel insight into the sorption characteristics of melanin for chloroquine.
2. Materials and methods 2.1. Melanin sample preparation Two types of cuttlefish melanin were used in this study. The first was purchased as Tinta CalamarTM from SAMTAS, Adelaide, containing melanin from the European Common Cuttlefish, Sepia officinalis and was the result of the blending of multiple cuttlefish ink sacs, including the tissue, prior to packaging. This melanin will be referred to as S. officinalis melanin. A second melanin was extracted from the ink sacs of native Australian cuttlefish, primarily Sepia apama with a small proportion of Sepia novaehollandiae (Valente Seafoods, South Australia, Australia and SARDI, South Australia, Australia) and prepared according to Schroeder and Gerber [50]. Due to the low levels of Sepia novaehollandiae present, it is assumed that this melanin would behave like Sepia apama melanin, and thus will be referred to as Sepia melanin.
10
R.L. Schroeder et al. / Colloids and Surfaces B: Biointerfaces 134 (2015) 8–16
Furthermore, bovine (Bos taurus) eye melanin was used to compare the binding capacity of mammalian melanin with that from cuttlefish. This melanin was extracted from the eyes of freshly slaughtered cattle according to the method of Tsuchiya et al. [44]. Bovine eye melanin was used as a representative mammalian melanin due to its accessibility and ease of comparison with previous studies. Both commercial S. officinalis melanin and the bovine eye melanin were pre-treated to remove all soluble contaminants, including free and bound lipids and protein. The purification entailed the consecutive treatment of approximately 1 g of crude, recovered melanin with 50 mL aliquots of 0.9 wt.% saline solution, 0.01 M HCl, 0.01 M NaOH, Tween 20, acetone, and MilliQ water. Thereafter, adsorbed or bound proteins were removed by reacting the melanin residue over a 24 h period with 20 mL aliquots of 20 mg pepsin (hydrolyses the peptide bond between hydrophobic and aromatic amino acids) at pH 1, followed by 10 mg papain activated with cysteine (cysteine proteinase that breaks peptide bonds at hydrophobic residues) at pH 5.5 and 10 mg trypsin (cleaves the carboxyl side of both the amino acids lysine and arginine residues) at pH 8. Final rinses with saline and MilliQ water and freeze drying of the sample resulted in melanin as a fine, black powder. 2.2. Chemicals Potassium dihydrogen orthophosphate, acetic acid, sodium acetate trihydrate and dichloromethane were obtained from Merck Pty. Ltd (Australia). Sodium chloride and Triton-X 100 were obtained from VWR International Pty. Ltd. (Australia). Chloroquine diphosphate, and the enzymes pepsin, papain and trypsin were sourced from Sigma–Aldrich (USA). All chemicals had a purity greater than 98 wt.% and were used without further purification. 2.3. Preparation of stock solutions Sodium acetate (pH 4.7) and potassium phosphate (pH 7.4) buffer solutions were prepared fresh at 20 mmol/L concentrations prior to use. A stock solution of 3.1 mmol/L chloroquine was prepared in the relevant buffer solution. The acetate and phosphate buffer solutions were used as blanks to correct for background in spectrophotometric measurements. 2.4. Instrumentation A Varian Cary 50 ultraviolet/visible (UV/Vis) spectrophotometer was used to determine chloroquine concentrations. Solution pH was measured using a Hanna potentiometer (Model: pH211 microprocessor pH metre) fitted with a microelectrode and calibrated at pH 4 and 7. 2.5. Calibration curve Chloroquine solutions were prepared ranging from 0.0033 to 0.33 mmol/L by diluting the stock solution with 20 mmol/L acetate buffer at pH 4.7. A similar range was also prepared in a phosphate buffer at pH 7.4 to generate standard curves of absorbance at 255 nm against chloroquine concentration. All samples were diluted with the appropriate buffer for analysis to bring readings within the range of the spectrophotometer. Standard curves were produced in triplicate in order to determine the inter-day reproducibility. 2.6. Equilibration time for adsorption An amount of 5.0 mg of melanin was suspended in 5 mL of 0.63 mmol/L chloroquine at a pH of 4.7. Separate tubes were placed
on an off-vertical (10◦ ), rotary wheel, rotating end-over-end continuously at 40 rpm for the equilibration time periods of 10, 30, 60, 120, and 240 min. After each equilibration time, the samples were centrifuged at 4770 × g for 4 min and the chloroquine concentration in the supernatant was determined by comparison with the standard curve using UV/Vis spectrophotometry at 255 nm. Solutions were measured against a melanin blank solution background, which was centrifuged prior to being added to the cuvette. The same procedure was employed for all subsequent analyses. This process was repeated at pH 7.4. 2.7. Adsorption of chloroquine to Sepia melanin To determine the adsorption capacity of melanin, a series of concentrations of chloroquine were prepared ranging from 0.16 to 2.2 mmol/L. A 5.0 mg portion of melanin was equilibrated with 5 mL of the specified concentration of solute prepared in either 20 mmol/L sodium acetate buffer (pH 4.7) or potassium phosphate buffer (pH 7.4) for 2 h, using an end-over-end mixing method as described above. The samples were centrifuged at 4770 × g for 4 min and the supernatant removed and analysed using UV/Vis spectrophotometry at 255 nm. Results were obtained for chloroquine binding using Sepia and S. officinalis melanin. Adsorption systems were prepared in triplicate by adding the same masses of melanin to the same volume and concentration of chloroquine solution. 2.8. Adsorption of chloroquine to bovine eye melanin A 5.0 mg sample of bovine eye melanin was equilibrated with chloroquine concentrations ranging from 0.078 to 1.6 mmol/L for 2 h, using an end-over-end mixing method as described above. Samples were centrifuged at 2687 × g for 15 min and the supernatant removed. UV/Vis spectrophotometry was used for analysis using a wavelength of 255 nm. The amount of chloroquine adsorbed to bovine eye melanin was compared with Sepia and S. officinalis melanins. Adsorption systems were prepared in triplicate as described above. 2.9. Temperature dependence of adsorption of chloroquine to Sepia melanin To determine system temperature influence on the adsorption capacity of melanin, a series of concentrations of chloroquine were prepared ranging from 0.16–2.2 mmol/L. A 5.0 mg portion of melanin was equilibrated with 5 mL of the specified concentration of solute prepared in potassium phosphate buffer (pH 7.4) for 2 h at 277 K, 291 K and 310 K in addition to the previous result at an ambient temperature of 294 K. After equilibration via an endover-end mixing method as described above, the samples were centrifuged at 4770 × g for 4 min and the supernatant removed and analysed at 255 nm. Adsorption systems were prepared in triplicate as described above. 3. Results and discussion Chloroquine is a compound that accumulates in the eyes of patients using it as a therapeutic agent. We recently demonstrated that the affinity of Sepia melanin for chloroquine is greater than that for hydroxychloroquine [17]. Since retinopathy is one of the main concerns regarding the use of chloroquine, our next aim was to assess whether the type of melanin had an influence on the adsorption behaviour. With melanin isolated from representative mammalian bovine eyes, one cannot rule out some contamination with residual protein or structural changes due to the recovery process. Therefore, Sepia melanin was also examined to establish if
R.L. Schroeder et al. / Colloids and Surfaces B: Biointerfaces 134 (2015) 8–16
3.1. Comparison of melanin between different genera – effect of pH To characterise the relationship between melanin and chloroquine, adsorption studies were performed under a variety of conditions using different melanins. Two pH levels were examined to determine how this would affect the affinity between the adsorbent and adsorbate. Standard curves of absorbance against solute concentration were generated for each pH level (4.7 and 7.4). The coefficient of variation (CV) i.e. the quotient of the standard deviation and the mean, or the relative uncertainty, of the slopes for each standard curve (n = 3) was less than 1.05%, with each standard concentration having a CV of less than 3.00%. Each curve gave a linear fit of at least 99.9%. The lower limit of quantification at pH 4.7 was 3.3 mol/L, and 3.1 mol/L at pH 7.4, both with negligible variation. Both of the chloroquine solution adsorption isotherms for each melanin adsorbent were a type L2 [51], suggesting a monolayer equivalent amount adsorbed at equilibrium solution concentrations Ceq /(mmol/L) > 0.5. Consequently, each isotherm was well-defined via the Langmuir isotherm model, Eq. (1). To overcome departure from linearity in a (traditional) linear least-squares analysis at either low or relatively high equilibrium concentrations, the coefficients qm and KL (as the monolayer equivalent amount adsorbed and the Langmuir (equilibrium) coefficient) were determined by fitting the adsorption data to a least squares analysis to fit the entire data set, with the coefficients set to initial values of 0.01. The fitting procedure produced reproducible coefficient values after minimising the average of the sum of the squares of the difference between measured and calculated amounts adsorbed, and maximising the overall Pearson correlation coefficient. An exemplar plot and analysis is given in Fig. S2 and Table S2 (see Supplementary Information). q=
qm KL Ceq 1 + KL Ceq
(1)
Due to the implications of chloroquine adsorption to uveal melanin in retinopathy, bovine eye melanin was used as a representative mammalian melanin. The extent of chloroquine binding was compared with that of S. officinalis melanin to determine if changing to mammalian melanin would give different results. Both melanin types were subjected to a series of chemical ‘washes’ (see Section 2) to aid in purifying the mixture from associated cellular components. The adsorption capacities were determined at pH 4.7 and 7.4, with the results shown in Fig. 1. Although a third system pH value
12000 10000 8000 Amt. ads*
it could be considered as a reference point. As mentioned in the Introduction, melanins from different sources have been used in previous binding experiments, and the effect this difference has on the obtained results has yet to be examined. Two types of Sepia melanin were used in these experiments, as well as a comparison with bovine eye melanin. S. officinalis melanin is readily available as a food additive and was purchased as crudely-extracted ink from the European common cuttlefish. Before unequivocal answers regarding surface properties could be developed via adsorption measurements, it was also essential to determine if differences existed in adsorption properties of melanins extracted from the same genus, cuttlefish, as well as interspecies sources being compared. As with most naturally occurring materials, organically sourced melanins were suspected to contain a relatively high proportion of associated proteins. The aggressive purification procedure described above rendered melanin samples with consistent purity, raising the reproducibility of subsequent measurements and the anticipation that our results would lead to improved interpretation of future pharmacological pathway research.
11
6000 4000 2000 0 0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Ceq/(mmol/L) Fig. 1. Adsorption isotherm of chloroquine for S. officinalis (SO) and bovine eye (BE) SO at pH 4.7, BE at pH 7.4, and melanin. The 䊏 corresponds to SO at pH 7.4, BE at pH 4.7. * Units of amount adsorbed are mol/mole of melanin equivalent to 1 unit of (DHI1 :DHICA40 )n=1 for SO and (DHI1 :DHICA25 )n=1 for BE. The uncertainty bars result from the standard deviation of triplicate measurements. At pH 4.7, the adsorptive capacity of S. officinalis melanin for chloroquine was found to be 7244 mol/mole (0.92 mmol/g), significantly lower (43%) than at physiological pH. In addition, bovine eye melanin also had a significantly lower (44%) capacity for chloroquine at the lower pH with 2788 mol/mole (0.56 mmol/g). This difference may be explained by the ionisation states of DHI and DHICA, as well as that of chloroquine.
would allow plotting selected parameters as f(pH), to maintain the physiological relevance of this work, one would need to consider a value within this range, such as 6.0. The speciation data in Table 1 are reproduced for pH 6.0 in the Supplementary Information; they suggest negligible difference in principal species’ concentration. Comparison of the chloroquine adsorption isotherms for bovine eye melanin with those for S. officinalis shows the former has a reduced affinity under both sets of conditions at any equilibrium solution concentration. To allow this comparison to be made between the two melanin species, each was converted into micromoles per unit area. This enabled the concentration available for adsorption in 5 mg equivalent of particles, per unit area of a single particle, to be calculated. A single particle was used as the density of each melanin species is undefined. At the acidic pH, the carboxyl moiety of the DHICA monomer would be approximately 70% ionised, whereas it would have been completely ionised at physiological pH [52]. DHI would have no charge at both pH values. In addition, at low pH, it has previously been demonstrated that there is a reduction in the number of strong binding sites, thus leading to a decreased affinity seen with these results [43] (Fig. 2). The addition of buffer solute slightly reduced the ionisation percent of both DHI and DHICA. Without buffer, DHICA is 99% ionised at pH 7.4, whereas with buffer this percentage drops to 95% (Fig. 3a). The implication for DHI is less apparent, with a decrease of only 2%. The addition of buffer to chloroquine has little effect on the ionisation state (Fig. 3b). Chloroquine would be triply charged at pH 4.7, but only have two positive charges at pH 7.4. The differences in ionisation states of the melanin surface may result in significant electrostatic attraction between the positively charged chloroquine and anionic DHICA moiety, potentially explaining the sensitivity of chloroquine–melanin adsorption to changes in pH. Reduced aqueous solubility of chloroquine at elevated pH might similarly be expected to favour adsorption. The pH effect was also observed in a rice husk study conducted by Chowdhury et al. [53] which described the changes in the surface characteristics of rice husks with increasing pH. It was noted that the pores on the surface of the rice husks increased in diameter
12
R.L. Schroeder et al. / Colloids and Surfaces B: Biointerfaces 134 (2015) 8–16
Fig. 2. Ionisation states of chloroquine at pH 4.7 (a) and pH 7.4 (b), and ionisation states of DHI (c) and DHICA (d) at pH 7.4.
Fig. 3. Comparison of ionic strength-modulation of fractional dissociation with change in pH for (a) DHICA and (b) chloroquine (CQ). The solid line is without phosphate buffer, whereas the dotted line is with phosphate buffer.
R.L. Schroeder et al. / Colloids and Surfaces B: Biointerfaces 134 (2015) 8–16
13
Table 1 Mole fraction charge for DHI and DHICA at pH 7.4 and 4.7. Monomer
Dissociation Fraction**
Charge
S. officinalis moles/unit area (m2 )*
Bovine Eye moles/unit area (m2 )*
pH 7.4 DHI DHI− DHICA− DHICA2−
0.99 0.01 0.99 0.01
0 −1 −1 −2
36.4 0.4 94.5 0.8
2.9 0.03 3.8 0.03
pH 4.7 DHI+ DHI DHICA DHICA−
0.02 0.98 0.21 0.78
1 0 0 −1
0.6 36.2 20.3 74.7
0.05 2.9 0.8 3.0
* **
The relative uncertainty in surface area concentration = 4.1% [57]. All other species were evaluated and were significantly
