European Journal of Pharmaceutics and Biopharmaceutics 83 (2013) 244–252

Contents lists available at SciVerse ScienceDirect

European Journal of Pharmaceutics and Biopharmaceutics

journal homepage: www.elsevier .com/locate /e jpb

Research paper

Plasma distribution of tetraphenylporphyrin derivatives relevant forPhotodynamic Therapy: Importance and limits of hydrophobicity

Benoît Chauvin a,b,⇑, Bogdan I. Iorga c, Pierre Chaminade a, Jean-Louis Paul d,e, Philippe Maillard b,Patrice Prognon a, Athena Kasselouri a

a Faculté de Pharmacie, Univ. Paris-Sud, EA 4041, IFR 141, Châtenay-Malabry, Franceb Institut Curie, UMR 176 CNRS, Centre Universitaire, Univ. Paris-Sud, Orsay, Francec Institut de Chimie des Substances Naturelles, CNRS UPR 2301, Centre de Recherche de Gif-sur-Yvette, Gif-sur-Yvette, Franced Service de Biochimie, Hôpital Européen Georges Pompidou, AP-HP, Paris, Francee Laboratoire de Biochimie appliquée, EA 4529, Univ. Paris-Sud, Châtenay-Malabry, France

a r t i c l e i n f o

Article history:Received 25 October 2011Accepted in revised form 21 September 2012Available online 23 October 2012

Keywords:Meso-tetraphenylporphyrinPhotodynamic TherapyPlasmaLipoproteinAlbuminHydrophobicity

0939-6411/$ - see front matter � 2012 Elsevier B.V. Ahttp://dx.doi.org/10.1016/j.ejpb.2012.09.015

Abbreviations: HDL, high density lipoprotein; LDL,human serum albumin (essentially fatty acid free); H(not fatty acid free); TPP, 5,10,15,20-tetraphenylporpphyrin; MCR-ALS, multivariate curve resolution –TPP(pOH)4, 5,10,15,20-tetra-(para-hydroxyphenyl)p5,10,15-tri(para-O-b-D-galactosyloxyphenyl)-20-phen(pObGalOH)4, 5,10,15,20-tetra-(para-O-b-D-galTPP(pObGluOH)4, 5,10,15,20-tetra-(para-O-b-D-glucorin; TPP(pODEGOaManOH)3, 5,10,15-tri{para-O-[(2-oxy)-ethoxy]-phenyl}-20-phenylporphyrin.⇑ Corresponding author. Laboratoire de Chimie An

Univ. Paris-Sud, 5 rue J-B. Clément, 92296 Châtenay-M83 58 49; fax: +33 1 46 83 53 89.

E-mail addresses: benoit.chauvin@u-psud.fr (@icsn.cnrs-gif.fr (B.I. Iorga), pierre.chaminade@u-psudpaul@u-psud.fr (J.-L. Paul), philippe.maillard@curie.fr@u-psud.fr (P. Prognon), athena.kasselouri@u-psud.fr

a b s t r a c t

In the course of a Photodynamic Therapy (PDT) protocol, disaggregation of the sensitizer upon binding toplasma proteins and lipoproteins is one of the first steps following intravenous administration. This stepgoverns its subsequent biodistribution and has even been evoked as possibly orientating mechanism oftumor destruction. It is currently admitted as being mainly dependent on sensitizer’s hydrophobicity. Inthis context, as far as glycoconjugation of meso-tetraphenylporphyrin (TPP) macrocycle, a promisingstrategy to improve targeting of retinoblastoma cells confers to the sensitizer an amphiphilic character,we have studied the effect of this strategy on binding to plasma proteins and lipoproteins. With theexception of the majoritary protein binding (more than 80%) of more hydrophilic para-tetraglycoconju-gated derivatives, high density lipoproteins (HDL) appear as main plasma carriers of the other amphi-philic glycoconjugated photosensitizers. This HDL-binding is a combined result of binding affinities(log Ka ranging from 4.90 to 8.77 depending on the carrier and the TPP derivative considered) and relativeplasma concentrations of the different carriers. Evaluation of binding affinities shows that if hydropho-bicity can account for LDL- and HDL-affinities, it is not the case for albumin-affinity. Molecular dockingsimulations show that, if interactions are mainly of hydrophobic nature, polar interactions such as hydro-gen bonds are also involved. This combination of interaction modalities should account for the absence ofclear relationship between albumin-affinity and hydrophobicity. Taken together, our findings clarify theimportance, but also the limits, of hydrophobicity’s role in structure–plasma distribution relationship.

� 2012 Elsevier B.V. All rights reserved.

ll rights reserved.

low density lipoprotein; HSA,SAlip, human serum albuminhyrin, meso-tetraphenylpor-

alternating least squares;orphyrin; TPP(pObGalOH)3,ylporphyrin; TPPactosyloxyphenyl)porphyrin;pyranosyloxyphenyl)porphy-(2-O-a-D-mannosyloxy)-eth-

alytique, EA 4041, IFR 141,alabry, France. Tel.: +33 1 46

B. Chauvin), bogdan.iorga.fr (P. Chaminade), jean-louis.(P. Maillard), patrice.prognon(A. Kasselouri).

1. Introduction

Photodynamic Therapy (PDT) is an emerging technique whichcombines administration of a drug, called photosensitizer, andexposure of targeted tissue to light of appropriate wavelength.Treatment effect results from the potency of the photosensitizeronce activated by light to generate singlet oxygen and radical spe-cies responsible for cellular death. PDT has already proven its effi-cacy in the field of oncology for the treatment of lung,gastrointestinal or cutaneous tumors. It has also been applied tonon malignant diseases such as age-related macular degeneration[1]. In that case, transparency of ocular tissues to light makesPDT of particular interest. This property should also be exploitedfor the treatment of malignant ocular pathologies, such as retino-blastoma, the most frequent intraocular tumor in childhood. In-deed, besides poor efficiency for advanced tumors, currentlyavailable conservative treatments expose patients to a risk of

B. Chauvin et al. / European Journal of Pharmaceutics and Biopharmaceutics 83 (2013) 244–252 245

developing secondary tumors [2]. PDT appears as promising, com-bining a physical selectivity (tissular volume illuminated) and achemical one (tissular volume containing the photosensitizer).When applied to retinoblastoma tumors, photosensitizers devel-oped for other pathologies have shown poor efficiencies and selec-tivities, leading to side-effects such as long lastingphotosensitization of normal tissues. Design of new photosensitiz-ers adapted to retinoblastoma appears necessary [3].

Our group is involved in the evaluation of glycoconjugation oftetrapyrrolic macrocycles (Fig. 1). This strategy combines targetingof cellular sugar receptors and improvement of photosensitizer sol-ubility. The former promotes selective destruction of malignantcells, the latter favors rapid elimination from healthy tissues. In vi-tro photocytotoxicity and in vivo pharmacokinetics studies haveconfirmed the potential interest of this approach [4,5]. Efficacy of

Fig. 1. Structure of meso-tetraphenylporphyrin derivatives. (For interpretation of the refearticle.)

a glycoconjugated TPP, TPP(pODEGOaManOH)3, has been attestedin vivo, especially with a particular administration protocol (dou-ble drug dose with a 3 h interval), which combines targeting ofcancer cells and of blood vessels. Indeed, at the time of illumina-tion, drug administered 10 min before is still present in the vicinityof blood vessels, whereas drug administered 3 h before has reachedtumor cells [6]. Destruction of blood vessels indirectly kills tumortissue, through deprivation of oxygen and nutriments [7].

Photo-induced destruction of blood vessels is of particularinterest in the case of an application of PDT to retinoblastoma asfar as this tumor is considered as extremely sensitive to vascularinsufficiency [8]. Besides being a candidate target to PDT treat-ment, tumor vasculature has also been involved in sensitizersequestration in tumor tissue [9]. Indeed, tumor angiogenesis leadsto the formation of permeable neo-vessels, a specificity which has

rences to color in this figure legend, the reader is referred to the web version of this

246 B. Chauvin et al. / European Journal of Pharmaceutics and Biopharmaceutics 83 (2013) 244–252

been exploited for drug targeting via the Enhanced Retention andPermeation Effect (EPR Effect) [10]. However, in the particular caseof PDT, Roberts et al. have shown that this particular permeabilityis insufficient to account for selective retention of photosensitizersand presumed an implication of drug carriers, such as albumin andlipoproteins [11]. Binding to the latter has retained particularattention since the hypothesis formulated by Jori et al. that LDL-binding should favor tumor cell delivery [12]. Overexpression ofLDL-receptors by tumor cells and also by endothelial cells rein-forces this hypothesis [13]. On the opposite, binding of sensitizerto high density lipoprotein (HDL) or albumin has been associatedwith vascular sequestration of photosensitizer [14]. Besides possi-ble effect on target tissue, nature of plasma carrier should alsomodulate pharmacokinetics as suggested by the delayed tumoraccumulation described for BPD-MA when conjugated to HDL [15].

Plasma distribution studies have evidenced the major role oflipoproteins in photosensitizer transport, compared with the albu-min binding of most drugs [14,16]. This particularity is attributedto the high hydrophobic character of sensitizers. This propertyseems to govern plasma distribution, as it is frequently consideredthat hydrophilic compounds bind to proteins (especially albumin)and lipophilic ones to LDL. Amphiphilic derivatives present a ten-dency to bind mainly to HDL [17]. In this point of view, glycocon-jugation, which increases the solubility of the sensitizer anddecreases its hydrophobicity, should affect interactions with plas-ma proteins and lipoproteins. Thus, it appears essential to focus onthe impact of the glycoconjugation on drug distribution betweenplasma components. This study covers ten meso-tetraphenylpor-phyrin derivatives, six of which are glycoconjugated according todifferent modalities, and thus different lipophilicities. The aim is,beyond a description of the relationship between structure andplasma distribution, to better understand factors governing inter-actions of TPP sensitizers with plasma proteins and lipoproteins.

2. Materials and methods

2.1. Chemicals

TPP(pOH)4 was purchased from Sigma–Aldrich (Germany) andTPP(mOH)4 from Frontier Scientific (USA). All other porphyrinswere synthesized according to previously published protocols[18–21]. Stock solutions were prepared in DMSO and kept in thedark at +4 �C.

Theophylline, 5-phenyl-1H-tetrazole, indole, propiophenone,and valerophenone were provided by Acros Organics (USA), benz-imidazole, butyrophenone, colchicine, potassium bromide andammonium acetate by Merck (Germany), acetophenone by CarloErba (Italia), 0.9% sodium chloride solution by Aguettant (France).HPLC grade acetonitrile, methanol and dimethylsulfoxyde camefrom VWR (Germany), pH 7.4 PBS and human serum albumin fromSigma–Aldrich (Germany). Two different references of the latter(corresponding to different purification levels) were used, one isessentially fatty acid free (HSA), and the other is not fatty acid free(HSAlip). Ultrapure water was provided by an Alpha-Q device (Mil-lipore�, France). Human plasma was taken from normolipidemichemochromatosis patients.

2.2. Determination of chromatographic hydrophobicity index (CHI)

The procedure proposed by Valko et al. has been applied to theTPP derivatives [22]. CHI values of the two parent tri-hydroxylatedcompounds are not evaluable with this protocol. Calibration setcovered the logP range from �0.02 to 3.26: theophylline, 5-phenyl-1H-tetrazole, benzimidazole, colchicine, 8-phenyltheophylline,indole, acetophenone, propiophenone, butyrophenone, and valero-

phenone. HPLC measurements were performed on a Biotek Kon-tron system, operated with Geminyx (version 1.91) software.Experiments were carried out on a Modulo-cart QS uptisphereODB column (Interchim, France), with the dimensions of150 � 4.6 mm. The mobile phase, a gradient between of 50 mMammonium acetate (pH ranging from 7.0 to 7.3) and acetonitrile,was delivered at the flow rate of 1.0 mL min�1 according to the fol-lowing program: 0–1.5 min, 0% acetonitrile; 1.5–10.5 min, 0–100%acetonitrile; 10.5–11.5 min, 100% acetonitrile; 11.5–12.0 min, 0%acetonitrile; 12.0–20.0 min, 0% acetonitrile. For every TPP studied,reference data set was injected simultaneously with the photosen-sitizer in a mixture of 50% acetonitrile and 50% aqueous ammo-nium acetate buffer. Elution of the standards and of thephotosensitizer was monitored, respectively, at 254 nm and416 nm. Final CHI values for TPPs were the mean of three experi-ments, using CHI values determined by Valko for reference dataset.

2.3. Distribution in human plasma

Plasma was recovered from healthy normolipidemic blood do-nors with prior consent. One percent of a porphyrin solution indimethylsulfoxide was incubated 24 h in plasma. The porphyrin fi-nal molar concentration (1 lM) was in the order of magnitude ofwhat should be expected in vivo with an effective dose. To separateprotein and total lipoprotein fractions, plasma was adjusted at thedensity of 1.21 g mL�1 with potassium bromide and centrifuged8 h (90,000 rpm, 4 �C) using a Beckman NVT 90 rotor in a BeckmanXL 90 ultracentrifuge. Separation of lipoproteins fractions was thenperformed with a density gradient ultracentrifugation using a five-step KBr/NaCl gradient (densities of 1.063, 1.042, 1.019, and1.006 g mL�1 on top of plasma and a 1.21 g mL�1 KBr solution)and centrifuging for 24 h (38,000 rpm, 4 �C) using a Beckman SW41 rotor in a Beckman XL 90 ultracentrifuge. After ultracentrifuga-tion, fractions were collected using a system including a DensityGradient Fractionator ISCO Model 185, a collector LKB Bromma –2212 HELIRAC and a detector LKB Bromma – 2238 UVICORD S II(continuous absorbance monitoring at 280 nm). An extractionwas performed on the samples according to the method proposedby Wang [23]. 1900 lL of a mixture dimethylsulfoxide–methanol1:4 (v/v) was added to 100 lL of each fraction collected. After cen-trifugation (10 min, 4000 rpm), fluorescence intensity was read onthe supernatant with a Perkin-Elmer LS-50B spectrofluorimeter,with an excitation wavelength set at 420 nm. Plasma distributionbetween the different fractions was calculated on the basis of thosefluorescence intensities.

2.4. Spectroscopic study of interactions with plasma proteins andlipoproteins

2.4.1. Preparation of LDL and HDL fractionsTotal plasma was obtained from healthy normolipidemic blood

donors with prior consent. LDL and HDL were isolated from freshplasma by sequential ultracentrifugation in the Beckman 50.2 Tirotor (Beckman Coulter, Fullerton, CA, USA) at the densities of1.063 and 1.210 g mL�1, respectively. Molar concentrations ofLDL and HDL particles were determined on the basis of apoproteinquantitation according to the method proposed by Ohnishi [24].

2.4.2. Sample preparation and conditions of spectra recordingAn intermediate dilution of TPP stock solutions in pH 7.4 phos-

phate buffer saline (PBS) was used to prepare mixtures of a TPPwith the studied plasma carrier (HSA, HSAlip, HDL ou LDL).Dimethylsulfoxide final proportion in this solution was 0.5%. TPPfinal concentration was 1 � 10�7 M for fluorescence measurementsand 5 � 10�7 M for absorption study. Transporter concentration

B. Chauvin et al. / European Journal of Pharmaceutics and Biopharmaceutics 83 (2013) 244–252 247

varied from 0 to 1 � 10�4 M. The mixtures were kept in darkness at37 �C for 24 h. UV–Visible absorption spectra were recorded on aVarian� Cary Bio 100 spectrophotometer (Australia), with an opti-cal path of 10 mm and a slit width of 2 nm. Fluorescence emissionspectra were recorded with a Perkin-Elmer LS-50B spectrofluorim-eter, with an excitation wavelength set at 420 nm (excitation andemission slits equal to 7 nm).

2.4.3. Determination of binding constantsWhen compared with absorption spectroscopy, determination

of binding constants by fluorimetry presents two advantages: thepossibility of working with lower TPP concentrations (�10�7 M)than with absorption spectroscopy (�5 � 10�7 M) and the lowerdiffusion due to plasma carriers. Combined together, those twoadvantages widen the TPP–carrier ratio range possible to study.Classical binding of drugs to plasma proteins and lipoproteins isdescribed by an equilibrium involving the free drug, the free carrieron the one side and the drug–carrier complex on the other side.Thus, if binding involves a change in drug fluorescence intensityat one wavelength, affinity constants can be determined throughmonitoring of fluorescence at this wavelength:

F ¼ Ffree þ Fbound � Ffree

� �� Ka � ½Carrier�

1þ Ka � ½Carrier� ð1Þ

where Ffree and Fbound are fluorescence emission intensities, respec-tively, of the free and of the bound drug, [Carrier] the concentrationof the drug carrier and Ka the affinity constant defined by the fol-lowing relationship:

Ka ¼½Drug—Carrier�½Drug� ½Carrier� ð2Þ

where [Drug] and [Drug–Carrier] are the respective concentrationsof the free drug and of the drug–carrier complex. This method relieson the proportionality of Ffree and Fbound to the respective concentra-tions of these two forms, [Drug] and [Drug–Carrier]. However, in theparticular case of TPP derivatives, this is not the case. Indeed, freedrug is not an homogeneous form and covers in fact two differentforms: an aggregated one (poorly fluorescent) and a solubilizedone (moderately fluorescent). Then, fluorescence intensity of thefree drug is no more directly proportional to its concentration, be-cause it will depend on its aggregation rate, which is probably in-versely related with its concentration.

To overcome limitations of monowavelength monitoring in thisparticular case, multivariate curve resolution – alternating leastsquares (MCR-ALS) has been applied on fluorescence emissionspectra recorded with different carrier concentrations [25]. MCR-ALS consists in the decomposition of this data matrix (D) into theproduct of two matrices: (1) a C matrix containing concentrationprofiles of the different species, (2) a S matrix with their fluores-cence spectra.

D ¼ C � ST þ E ð3Þ

E matrix represents difference between experimental values anddata predicted by the model, that is residuals. Data analysis methodproposed by Diewok for MatLab [26] has been adapted here to Rsoftware [27]. Optimization is based on als algorithm contained inthe ALS package [28]. High aggregation of certain TPP derivativescombined with a strong affinity for some of the studied plasma car-riers reduces contribution of the solubilized drug. In as far as fluo-rescence emission spectra of this particular species are the samewhatever the carrier considered, a column-wise extended approachhas been used to improve results. D matrix is constituted by spectrarecorded on one TPP derivative with the four carriers studied: HSA,HSAlip, LDL, HDL. C and S matrices, respectively, contain concentra-tion and spectra profiles of five species: the free solubilized drug

and the four complexes formed by the TPP with each of the four car-riers studied. Because of its poor fluorescence, the aggregated freedrug is not included directly. Its presence is taken into account byapplying no concentration closure constraint (sums of concentra-tions of the other species at each carrier concentration are notforced to be equal to one). For each carrier, concentration profileof the bound drug is adjusted to follow relationship (2), before sub-sequent spectra optimization. When further optimizations no morereduce residues’ amount, the four binding constants are determinedby nonlinear regression of the concentration profile with Eq. (2).

2.5. Molecular docking simulations

Blind docking of TPP derivatives into human serum albumin(PDB code 1AO6) was performed with AutoDock Vina 1.0 (exhaus-tiveness value of 100 and maximum output of 20 structures) [29].Unsubstituted TPP crystal structure has been downloaded from theCambridge Structural Database (MOLFEZ). After substituents’ addi-tion with UCSF Chimera [30], ligands were prepared for dockingusing AutoDock Tools to calculate Gasteiger charges and set activetorsions (the four bonds between porphyrin core and phenyls, allrotatable bonds between the phenyl and the sugar residue). UCSFChimera was used to visualize dockings, calculate contact surfaces,and monitor hydrogen bonds. The selection of the main bindingconformation depended on the frequence of the different sitesamong the twenty output structures.

3. Results

3.1. Hydrophobicity of TPPs

As expected, glycoconjugation induces a decrease in hydropho-bicity relative to the hydroxylated parent compound (Table 1).Moreover, hydrophobicity is further reduced with increasing num-ber of sugar residues. If these conclusions apply both to para andmeta series, it is to note that para-derivatives are less hydrophobicthan their meta isomers. Thus, CHI of TPP(pObGluOH)4 (28.3) islower than that of the TPP(mObGluOH)4 (39.3). This also holds truefor hydroxylated compounds, when comparing TPP(pOH)4

(CHI = 100.2) and TPP(mOH)4 (117.2). Because of minor differencesof hydrophobicity between mannose and galactose residues, thelarge CHI increase between TPP(pObGalOH)3 (CHI = 40.8) andTPP(pODEGOaManOH)3 (CHI = 62.4) should be attributed to thepresence of a spacer between the sugar and the phenyl ring. Thepara-derivative with the spacer is even more hydrophobic thanthe meta-triglycoconjugated derivative, TPP(mObGluOH)3

(CHI = 55.7).

3.2. Distribution in human plasma

For eight of the ten studied compounds, more than 75% of thesensitizer is found in lipoproteic fraction (Table 1). Exceptions tothis rule are constituted by the two para-tetraglycoconjugatedderivatives, TPP(pObGalOH)4 and TPP(pObGluOH)4, lone com-pounds to be mainly bound – about 80% – to the proteic fraction.This behavior is particular striking when compared with the quiteexclusive lipoproteic transport of the meta-tetraglycoconjugatedderivative. Among compounds majoritary bound to lipoproteins,the para-triglycoconjugated TPP(pObGalOH)3 presents a signifi-cantly higher protein-bound fraction than other compounds,including TPP(pODEGOaManOH)3. Drug binding to proteic fractionconcerns one quarter of the former but is negligible in the case ofthe latter (less than 6%). This comparison shows that inclusion of aspacer between the sugar and the phenyl ring has a dramatic effecton plasma distribution.

Table 1Plasma distribution of meso-tetraphenylporphyrin derivatives. CHI values have been determined by HPLC according to the method described by Valko et al. [22]. Plasmadistributions have been determined by fluorescence measurements after plasma ultracentrifugations (see details in Section 2.3).

Compound CHI Lipoproteins Proteins

Total HDL LDL

TPP(mOH)3 – 94.7 ± 1.3 74.2 ± 5.2 17.3 ± 4.8 5.3 ± 1.3TPP(mOH)4 117.2 ± 0.1 97.6 ± 0.4 71.3 ± 1.0 20.0 ± 3.0 2.4 ± 0.4TPP(mObGluOH)3 55.7 ± 0.5 97.8 ± 1.0 78.0 ± 4.9 14.1 ± 3.4 2.2 ± 1.0TPP(mObGluOH)4 39.3 ± 0.1 95.6 ± 1.2 60.8 ± 13.0 22.1 ± 5.4 4.4 ± 1.2TPP(pOH)3 – 95.0 ± 1.2 77.6 ± 4.7 13.4 ± 3.0 5.0 ± 1.2TPP(pOH)4 100.2 ± 0.2 96.4 ± 1.3 86.7 ± 5.4 7.7 ± 4.0 3.6 ± 1.3TPP(pObGalOH)3 40.8 ± 0.1 77.3 ± 1.6 67.7 ± 2.1 7.1 ± 1.1 22.7 ± 1.6TPP(pObGalOH)4 26.5 ± 0.1 10.4 ± 1.4 8.7 ± 1.6 1.4 ± 0.5 89.6 ± 1.4TPP(pObGluOH)4 28.3 ± 0.1 13.7 ± 4.2 11.3 ± 3.6 1.8 ± 0.4 86.3 ± 4.2TPP(pODEGOaManOH)3 62.4 ± 0.1 95.4 ± 1.3 85.8 ± 3.0 8.6 ± 4.0 4.6 ± 1.3

Table 2Binding affinities of meso-tetraphenylporphyrin derivatives (expressed as logKa).

Compound CHI Albumin Lipoproteins

HSA HSAlip LDL HDL

TPP(mOH) – 5.07 5.50 8.30 8.11

248 B. Chauvin et al. / European Journal of Pharmaceutics and Biopharmaceutics 83 (2013) 244–252

HDL are main lipoproteic carriers of photosensitizers. Indeed,with the exception of TPP(pObGalOH)4 and TPP(pObGluOH)4, thosestructures bind more than half of sensitizer present in plasma.Binding to LDL is always minoritary, the highest proportion beingreached with the TPP(mObGluOH)4.

3

TPP(pOH)3 – 5.60 5.77 8.32 7.11TPP(mOH)4 117.2 ± 0.1 5.77 5.99 8.21 7.65TPP(pOH)4 100.2 ± 0.2 6.32 6.17 8.77 7.35TPP(pODEGOaManOH)3 62.4 ± 0.1 4.90 5.19 7.78 7.01TPP(mObGluOH)3 55.7 ± 0.5 5.66 5.73 7.64 7.33TPP(pObGalOH)3 40.8 ± 0.1 5.80 6.17 7.89 7.33TPP(mObGluOH)4 39.3 ± 0.1 5.05 5.03 7.58 6.95TPP(pObGluOH)4 28.3 ± 0.1 5.57 5.83 6.87 6.51TPP(pObGalOH)4 26.5 ± 0.1 5.29 5.27 6.80 6.33

3.3. Binding constants toward plasma proteins and lipoproteins

Binding of TPPs to plasma carriers induces spectral modifications,accounting for the disruption of TPPs aggregates upon formationof a complex between the TPP and the carrier. Those equilibriacan be followed by absorption or fluorescence spectroscopies. Inthe absence of plasma carrier, absorption spectrum of TPP(pObGalOH)3 presents a large Soret band at 417 nm, with a distinctshoulder at 437 nm, the latter resulting from the formation ofJ-aggregates (Fig. 2). HSA addition leads to the disappearance ofthe 437-nm shoulder characteristic of aggregates and to theappearance of a new intense band at 422 nm, which attests forthe formation of the complex. Concerning fluorescence spectros-copy, binding of TPP to HSA induces a slight modification of spec-tral shape but a significant increase in fluorescence intensity.

If all TPP are likely to bind to LDL, HDL, and HSA, affinities dra-matically vary according to carrier and substitution of the TPP core(Table 2). However, it is remarkable to observe that, whatever theTPP considered, affinities toward the different plasma carriers de-crease when passing from LDL to HDL and finally to HSA (whether

Fig. 2. Spectral modifications of TPP(pObGalOH)3 upon binding to HSA. (Forinterpretation of the references to color in this figure legend, the reader is referredto the web version of this article.)

fatty acid free or not). Even compounds mainly bound to proteinsin plasma (TPP(pObGalOH)4 and TPP(pObGluOH)4) present a higheraffinity for LDL than for other studied plasma components. Thosepara-derivatives present higher affinity constants toward HSAand HSAlip than their meta-homologous, an observation that ap-plies whatever the substitution considered.

Another noteworthy result is the large difference in bindingaffinities for compounds with similar plasma distribution. That isthe case of TPP(pOH)4 and TPP(pODEGOaManOH)3, two com-pounds bound at �85% to HDL. Binding affinity to LDL and HSAis 10-fold higher for the former than for the latter. When comparedwith TPP(pObGalOH)3, TPP(pODEGOaManOH)3 presents the sameorder of magnitude in their binding constants toward LDL andHDL. Spacer mainly affects binding to HSA, decreasing 10-foldbinding affinities, which could account for the lower protein bind-ing of this compound when compared with TPP(pObGalOH)3.

3.4. Molecular docking simulations

Depending on their substitution, TPPs interact at different loca-tions on the HSA molecule. The most noticeable result is theimpossibility for these bulky structures to insert into the twohydrophobic pockets that constitute Sudlow binding sites commonto most drugs. It is difficult to privilegiate one binding site for non-glycoconjugated TPPs. These structures are spread at differentlocations depending on their substitution. On the opposite, glyco-conjugated porphyrins present preferential clusters.

If considering glycoconjugated porphyrins (Fig. 3), the mostnoticeable result is the drastic effect of sugar position. Sugar natureand number do not seem to affect binding location. The two meta-derivatives, TPP(mObGluOH)3 and TPP(mObGluOH)4, bind on thesame location in the inter-domain crevice, whereas the three para-derivatives without spacer, TPP(pObGalOH)3, TPP(pObGalOH)4,and TPP(pObGluOH)4, share the same binding site. For the latter

B. Chauvin et al. / European Journal of Pharmaceutics and Biopharmaceutics 83 (2013) 244–252 249

three compounds, the tetrapyrrole is located between residues Q104and K466, with two phenyl rings of both sides of residue K106.

TPP(mObGluOH)3 binds between subdomains Ib and IIIa, withthe TPP core located below residue R114. The three sugar residuesinsert into three polar pockets: (i) the first formed by residuesR114, R117, R186, and K519, (ii) the second constituted by residuesN109, S419, T422, K466, and T467, and (iii) the third composed byamino acids D108, H146, K190, R197, and Q459. In the case of thetetraglycoconjugated TPP(mObGluOH)4, three sugars insert in thesame pockets, the fourth interacting with K524.

Of particular interest is the modulation of distribution patterninduced by the presence of the spacer. If this particularity doesnot prevent TPP(pODEGOaManOH)3 from interacting at the samelocation than TPP(pObGalOH)3, it favors binding on a site next tothat of TPP(mObGluOH)3, on a site inaccessible to the tri-paragly-coconjugated derivative without spacer (TPP(pObGalOH)3). In thisparticular conformation, the tetrapyrrole is close to residue P421,and one sugar is located between residues Q33 and E86, one otherbetween residues K419 and K500. The last mannose residue insertsinto the third polar pocket described for TPP(mObGluOH)3.

The fact that sugar residues are susceptible to insert into polarpockets in the case of TPP(pODEGOaManOH)3 or meta-derivativesresults in an higher contribution of the substituent in the interac-tion surface for those derivatives (Table 3). For those particularstructures, once bound to HSA, sensitizer is less accessible to sol-vent than in the case of para-derivatives without spacer. This latterfact is confirmed by the percentage of the TPP derivative involvedin the interaction (Table 3). Interaction surfaces increase withincreasing surfaces of the TPP derivatives. The main exception tothis rule is represented by para-tetraglycoconjugated derivatives,their interface surfaces being lower than that of TPP(pObGalOH)3.This fact probably results from the rigidity of para-conformation,which induces a reduced possibility to insert into favorable pocketsupon increasing molecular volume. Indeed, flexibility of meta-derivatives confers to those derivatives the ability to form higherinterface surfaces with the protein than para-derivatives. Analysisof interaction modalities shows that TPPs interact with HSA mainlythrough hydrophobic interactions but also through hydrogenbonds. The latter, which are stronger interactions, mainly concernglycoconjugated compounds, due to their increased number of hy-droxyl groups.

Fig. 3. Binding sites of glycoconjugated TPPs according to blind docking results. Binding sgreen), TPP(pObGalOH)4 (in dark green), TPP(pObGluOH)4 (in sea green) and TPP(pODElegend, the reader is referred to the web version of this article.)

4. Discussion

4.1. Plasma distribution of photosensitizers

Plasma distributions of glycoconjugated TPPs are consistentwith common considerations on the relationships between plasmadistribution and hydrophobicity. Differences in hydrophobicitymainly result from differences in exposure of the TPP ring due tothe presence of polar substituents. This principle accounts for theeffect of substituent’s nature and number but also position. Indeed,para-substitution confers to the molecule a planar conformationdifferent from the globular conformation resulting from meta-sub-stitution. The latter allows an easier access to the hydrophobic TPPcore (Fig. 4).

Binding to the proteic fraction of para-tetraglycoconjugatedderivatives can be explained by the more pronounced hydrophiliccharacter of those compounds. TPP(pObGalOH)3 presents an inter-mediate CHI and an intermediate behavior between hydrophilicprotein-bound derivatives and more hydrophobic compoundsquite exclusively bound to lipoproteins. The latter compoundspresent the typical behavior of amphiphilic compounds, mainlybound to HDL. Binding to LDL concerns always a minoritary pro-portion of TPPs on the studied series. The effect of para-glycocon-jugation appears similar to that of para-sulfonation as describedby Kongshaug et al. [14]: only the tetrasubstituted compoundbinds mainly to proteins, other derivatives (whether mono-, di-or tri-sulfonated) bind mainly to lipoproteins, majoritarily HDL.

Binding to LDL is commonly associated with the hydrophobiccharacter of TPPs. However, in our series, there is no correlation be-tween proportion bound to LDL and CHI. This finding is similar tothat described in the case of the sulfonated TPPs: a disulfonatedTPP presents a higher proportion bound to LDL than the morehydrophobic monosulfonated derivative [14]. Moreover, in our ser-ies, similar hydrophobicities do not imply similar distribution pat-terns, as can be evinced by comparing TPP(pObGalOH)3 andTPP(mObGluOH)4.

4.2. From plasma distribution to binding constants

The most striking conclusion of the comparison between plas-ma distribution and binding constants is that even compounds pre-dominantly bound to proteins in plasma have a higher affinity

ites of TPP(mObGluOH)3 (in red), TPP(mObGluOH)4 (in yellow), TPP(pObGalOH)3 (inGOaManOH)3 (in blue). (For interpretation of the references to color in this figure

Table 3Properties of interface surfaces between HSA and the different TPP derivatives.

Interface surface Percentage of the TPP surfaceinvolved in the interaction (%)

Contribution of the substituentin the interactiona (%)

Polar Apolar Total

TPP 129.6 315.1 444.7 35.1 0.0TPP(mObGluOH)3 296.3 412.4 708.7 37.7 64.4TPP(mObGluOH)4 391.1 530.3 921.4 42.9 62.6TPP(mOH)3 121.9 297.1 419.0 34.4 11.5TPP(mOH)4 134.7 271.4 406.1 32.9 23.4TPP(pObGalOH)3 200.6 404.7 605.3 29.0 55.5TPP(pObGalOH)4 276.2 305.2 581.4 28.0 52.7TPP(pObGluOH)4 260.3 304.7 564.9 24.5 54.3TPP(pOH)3 97.4 234.2 331.5 27.9 11.4TPP(pOH)4 94.7 216.4 311.1 25.9 11.5TPP(pODEGOaManOH)3 352.9 464.5 817.4 30.6 69.5

a The ratio between the surface of the substituent in contact with the protein and the total surface of the TPP derivative interacting with the protein.

250 B. Chauvin et al. / European Journal of Pharmaceutics and Biopharmaceutics 83 (2013) 244–252

toward lipoproteins, especially LDL. This striking result recalls thatrelative affinities toward separated plasma carriers is just a part ofits plasma distribution, the latter being also the result of relativeconcentrations of plasma carriers. Involvement of plasma proteinand lipoprotein concentrations has been underlined by Kongshaugin the case of hematoporphyrin [31]. This compound presents amajoritary binding to HDL in plasma, despite a higher affinity to-ward LDL than toward HDL. Thus, plasma distributions ofTPP(pObGalOH)4 and TPP(pObGluOH)4 are not the consequenceof a particular affinity toward albumin, but the result of a ratio ofaffinities toward lipoproteins and albumin not high enough toovercome the difference in the concentrations of those carriers. In-deed, albumin is the most abundant plasma protein (�0.5–0.8 mM), whereas lipoprotein concentration is much lower(�1 lM for LDL and 13 lM for HDL).

Despite the presumed protein-affinity of hydrophilic com-pounds, there is no clear relationship between affinity towardHSA and CHI. Hydrophilic compounds, such as TPP(mObGluOH)4,present low binding constants, but it is also the case of mosthydrophobic structures such as TPP(mOH)3. Highest binding con-stants are characteristic of compounds (TPP(pOH)4, TPP(mOH)4 orTPP(pObGalOH)3) with intermediate hydrophobicities. On thecontrary, TPPs’ affinity toward lipoproteins can be globally ac-counted for by their hydrophobicity. Affinity increase with CHIapplies both to HDL and LDL but is more pronounced in the caseof the latter. This observation can be linked to the classical idea ofa preferential binding of more hydrophobic structures to LDL.However, this rule knows exceptions and in the studied series,despite the tight link between LDL-affinity and CHI, proportionof LDL-binding is not correlated with hydrophobicity. The latterfact is the consequence of the absence of link between affinity to-ward HSA and CHI.

Similar considerations should explain an exception to the clas-sical rule reported by Hasan et al. [14]. Protoporphyrin and hema-toporphyrin bind in the same proportions to plasma proteinsdespite the higher hydrophobicity of the former. This result mustbe viewed as the consequence of the difference in substitution,which confers a much higher affinity toward albumin for protopor-phyrin (280 � 106 M�1) than for hematoporphyrin (1.4 � 106 M�1).This albumin-affinity increase counterbalances the probablehydrophobicity-induced increase in affinity toward lipoproteins,resulting in a similar plasma distribution.

4.3. Interactions with human serum albumin

Contrary to HDL- and LDL-affinities, an increase in hydropho-bicity does not result in an increased affinity toward albumin. In-deed, this absence of correlation between overall hydrophobicity

and albumin affinities leads to consider a possible effect of otherfactors, such as the size of the TPP derivative (which could limitits ability to access to favorable binding pockets of albumin mole-cule) or distribution of polar substituents on the hydrophobic TPPcore (which could favor or restrict interactions with amino acids).Confronted with similar observations, some authors have under-lined the importance of the amphiphilic character of the photosen-sitizer in its interactions with proteins [32]. Those conclusionsstrengthen the interest of docking simulations to better under-stand phenomena governing interactions between TPPs and HSA.Docking results have shown that substitution affects location ofthe TPP derivative on the protein. Moreover, they have led to ex-clude interactions at classical drug binding sites I and II, unlikewhat has been described for some sensitizers: chlorin p6, purpurin18 [33], or bacteriochlorin derivatives [34]. This difference proba-bly results from steric difference between the tetrapyrroles notbearing phenyls at meso positions and the bulky TPP core. Resultsobtained with other tetra-para-substituted TPPs conclude to abinding at the surface of the albumin molecule, a result consistentwith our findings. Fluorescence lifetime studies performed on aseries of sulfonated phthalocyanines have shown that degree ofsulfonation influences insertion in hydrophobic pockets. Tetrasulf-onated derivatives bind at the surface of the protein whereas lowersulfonation degree allows insertion into hydrophobic cavities [35].However, effect of substituent is only partly steric. It also plays arole in interactions modalities between sensitizer and HSA. Sulfonegroups could form ionic interactions with basic amino acids (histi-dine and lysine), a hypothesis strengthened by sensitivity of inter-actions to ionic strength [36].

The double acting effect of the substituent, likely to form directinteractions with HSA but also to induce steric limitations, also ap-plies to our series of hydroxylated and glycoconjugated porphy-rins. Glycoconjugated derivatives form more hydrogen bondsthan hydroxylated ones, and meta-derivatives more than para-derivatives. However, even when glycoconjugated, TPP derivativesinteract with the protein mainly through hydrophobic interactions.The direct involvement of the substituent in the binding distin-guishes TPP interactions with proteins from their interactions withthe C18 surface in the HPLC experiments. Indeed, CHI values arehighly correlated with ratios of TPP nucleus surface to the totalTPP derivatives surface (r2 = 0.94 when excluding the highly flexi-ble TPP(pODEGOaManOH)3), which illustrates the probable lack ofdirect interactions between the substituent and apolar surfaces. Inthe case of interactions with albumin, susbtituents interact directlywith the protein, especially if the TPP derivative possesses someflexibility (case of meta-derivatives and TPP(pODEGOaManOH)3).Rigidity of planar para-derivatives prevents them to form specificinteractions with albumin, which could explain the absence of dif-

Fig. 4. Conformations of meta- and para-derivatives. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

B. Chauvin et al. / European Journal of Pharmaceutics and Biopharmaceutics 83 (2013) 244–252 251

ference in distribution pattern between TPP(pObGalOH)4 andTPP(pObGluOH)4 despite the modification of the nature of sugarresidue. This observation also applies to the respective affinitiesof those particular derivatives.

When compared with the more widespread distribution patternof para-derivatives, meta-derivatives seem to present stronger andmore specific interactions. This result, conflicting at the first sightwith affinity constants (higher in the para series), should be prob-ably considered differently: globular conformation of meta-deriva-tives prevents them from interacting at the surface of albuminmolecule, thus restraining their possible binding sites. In this per-spective, higher overall binding constants measured on para-deriv-atives could result from a higher number of sites of almostequivalent affinities.

4.4. Considerations about the particular affinity for LDL

Photosensitizers are likely to interact with lipoproteins accord-ing to two modes, whether with the proteic portion and/or withthe lipidic one [37]. Existence of high affinity sites on apoproteincoexisting with secondary solubilization in lipidic portion has beensupposed in the case of interactions of chlorin e6 with LDL [38]. Ifglobal binding constant is of the same order of magnitude as theone obtained for glycoconjugated TPPs, a preferential binding toapoprotein is unlikely for the latters. Tight link between affinity to-ward lipoproteins and hydrophobicity tends to privilegiate the ideaof an interaction with the lipidic portion. It seems probable thatinteractions of TPPs with the hydrophobic stationary phase in HPLCare quite similar to their interactions with the hydrophobic lipidicportion. Moreover, lower binding affinity toward lipoproteins ofglycoconjugated derivatives – likely to interact strongly with pro-teic portion through hydrogen bonding – reinforces the hypothesisof an interaction with the lipidic portion. At last, this hypothesis isconfirmed by comparison with affinities of TPPs toward liposomes[39]. Ranking of binding affinities toward those phospholipidic ves-icules is close to that obtained with HDL.

Difference in binding affinities toward HDL and LDL leads toconsider a possible role of certain lipids in the preferential bindingof TPPs to LDL than HDL. Interactions of hypericin with biologicalmembranes have shown that this structure presents a particularaffinity for cholesterol [40], a fact that could account for its locationin LDL, between hydrophobic core and phospholipid shell [41].Involving cholesterol is unlikely for our compounds, more amphi-philic than hypericin, and thus less able to insert deeply in the lipo-protein core. This hypothesis is supported by studies of inclusion ofdendrimeric porphyrins in biological membranes that show no im-pact of cholesterol proportion [42], contrary to what could have

been described for others photosensitizers, such as deuteroporphy-rin [43]. Preferential affinity for LDL than for HDL could result fromdifferences in surface properties: LDL surface is less hydrophobicand its outer layer is more fluid [44]. More hydrophobic characterof HDL surface results from the presence of more triglycerides andcholesterol esters in the outer layer [45]. Combined together,amphiphilic structures could better interact with LDL, insertionof hydrophobic pole being easier and interaction of hydrophilicpart with the surface being favored.

5. Conclusion

Our results underline the complementarity between determina-tion of affinity constants and measurement of sensitizer effectivedistribution between plasma components. Taken together, thesedata give a new insight into the role of the hydrophobicity of thesensitizer in its plasma distribution, the importance but also itslimits. Increasing hydrophobicity should orient distribution towardLDL, whereas lowering this parameter results in a majoritary pro-tein binding. Exceptions to this rule should result from specificinteractions between a photosensitizer and a carrier, interactionsnot directly related to its hydrophobicity. Our study also showsthat measuring the fraction bound to LDL is not sufficient to under-stand the behavior of TPPs in plasma. Binding constant determina-tions are essential. If it is commonly admitted that plasmadistribution plays a decisive role in orientating biodistribution,binding affinities are likely to affect photosensitizer’s ability topass from the carrier to its final target, a fact that should not beunderestimated when reconsidering the link between plasmabehavior and tumor localization.

Acknowledgements

B. Chauvin has benefited from a ‘‘Postes d’accueil CNRS – CEA –APHP’’ grant. The authors thank technicians of HEGP Biochemistryservice for their precious contribution to plasma distributionstudies.

References

[1] T.J. Dougherty, C.J. Gomer, B.W. Henderson, G. Jori, D. Kessel, M. Korbelik, et al.,Photodynamic therapy, J. Natl. Cancer Inst. 90 (1998) 889–905.

[2] F. Doz, H. Brisse, D. Stoppa-Lyonnet, X. Sastre, J. Zucker, L. Desjardins,Retinoblastoma, in: Paediatric Oncology, Pinkerton, R Plowman, N Pieters, R,London, 2004.

[3] J.B. Winther, The effect of photodynamic therapy on a retinoblastoma-liketumour. An experimental in vitro and in vivo study on the potential use ofphotodynamic therapy in the treatment of retinoblastoma, Acta Ophthalmol.Suppl. (1990) 1–37.

252 B. Chauvin et al. / European Journal of Pharmaceutics and Biopharmaceutics 83 (2013) 244–252

[4] M.-C. Desroches, A. Bautista-Sanchez, C. Lamotte, B. Labeque, D. Auchère, R.Farinotti, et al., Pharmacokinetics of a tri-glucoconjugated 5,10,15-(meta)-trihydroxyphenyl-20-phenyl porphyrin photosensitizer for PDT. A single dosestudy in the rat, J. Photochem. Photobiol. B, Biol. 85 (2006) 56–64.

[5] P. Maillard, B. Loock, D. Grierson, I. Laville, J. Blais, F. Doz, et al., In vitrophototoxicity of glycoconjugated porphyrins and chlorins in colorectaladenocarcinoma (HT29) and retinoblastoma (Y79) cell lines, Photodiagn.Photodyn. Ther. 4 (2007) 261–268.

[6] M. Lupu, C.D. Thomas, P. Maillard, B. Loock, B. Chauvin, I. Aerts, et al., 23Na MRIlongitudinal follow-up of PDT in a xenograft model of human retinoblastoma,Photodiagn. Photodyn. Ther. 6 (2009) 214–220.

[7] B. Chen, B.W. Pogue, P.J. Hoopes, T. Hasan, Combining vascular and cellulartargeting regimens enhances the efficacy of photodynamic therapy, Int. J.Radiat. Oncol. Biol. Phys. 61 (2005) 1216–1226.

[8] M.R. Horsman, J. Winther, Vascular effects of photodynamic therapy in anintraocular retinoblastoma-like tumour, Acta Oncol. 28 (1989) 693–697.

[9] P. Agostinis, K. Berg, K.A. Cengel, T.H. Foster, A.W. Girotti, S.O. Gollnick, et al.,Photodynamic therapy of cancer: an update, CA – Cancer J. Clin. 61 (2011)250–281.

[10] F. Danhier, O. Feron, V. Préat, To exploit the tumor microenvironment: passiveand active tumor targeting of nanocarriers for anti-cancer drug delivery, J.Control. Release 148 (2010) 135–146.

[11] W.G. Roberts, T. Hasan, Role of neovasculature and vascular permeability onthe tumor retention of photodynamic agents, Cancer Res. 52 (1992) 924–930.

[12] G. Jori, L. Schindl, A. Schindl, L. Polo, Novel approaches towards a detailedcontrol of the mechanism and efficiency of photosensitized processes in vivo, J.Photochem. Photobiol. A, Chem. 102 (1996) 101–107.

[13] J.T.C. Wojtyk, R. Goyan, E. Gudgin-Dickson, R. Pottier, Exploiting tumourbiology to develop novel drug delivery strategies for PDT, Med. Laser Appl. 21(2006) 225–238.

[14] T. Hasan, B. Ortel, A.C. Moor, B.W. Pogue, Photodynamic Therapy of Cancer, in:Cancer Medicine 6, sixth ed., BC Decker, 2003 (Chapter 40).

[15] B.A. Allison, P.H. Pritchard, A.M. Richter, J.G. Levy, The plasma distribution ofbenzoporphyrin derivative and the effects of plasma lipoproteins on itsbiodistribution, Photochem. Photobiol. 52 (1990) 501–507.

[16] H.J. Hopkinson, D.I. Vernon, S.B. Brown, Identification and partialcharacterization of an unusual distribution of the photosensitizer meta-tetrahydroxyphenyl chlorin (temoporfin) in human plasma, Photochem.Photobiol. 69 (1999) 482–488.

[17] A.P. Castano, T.N. Demidova, M.R. Hamblin, Mechanisms in photodynamictherapy. Part 3 – Photosensitizer pharmacokinetics, biodistribution, tumorlocalization and modes of tumor destruction, Photodiagn. Photodyn. Ther. 2(2005) 91–106.

[18] I. Laville, T. Figueiredo, B. Loock, S. Pigaglio, P. Maillard, D.S. Grierson, et al.,Synthesis, cellular internalization and photodynamic activity ofglucoconjugated derivatives of tri and tetra(meta-hydroxyphenyl)chlorins,Bioorg. Med. Chem. 11 (2003) 1643–1652.

[19] I. Laville, S. Pigaglio, J.-C. Blais, B. Loock, P. Maillard, D.S. Grierson, et al., A studyof the stability of tri(glucosyloxyphenyl)chlorin, a sensitizer for photodynamictherapy, in human colon tumoural cells: a liquid chromatography and MALDI-TOF mass spectrometry analysis, Bioorg. Med. Chem. 12 (2004) 3673–3682.

[20] I. Laville, S. Pigaglio, J.-C. Blais, F. Doz, B. Loock, P. Maillard, et al.,Photodynamic efficiency of diethylene glycol-linked glycoconjugatedporphyrins in human retinoblastoma cells, J. Med. Chem. 49 (2006) 2558–2567.

[21] D. Oulmi, P. Maillard, J.-L. Guerquin-Kern, C. Huel, M. Momenteau,Glycoconjugated porphyrins. 3. Synthesis of flat amphiphilic mixed meso-(glycosylated aryl)arylporphyrins and mixed meso-(glycosylatedaryl)alkylporphyrins bearing some mono- and disaccharide groups, J. Org.Chem. 60 (1995) 1554–1564.

[22] K. Valko, C. Bevan, D. Reynolds, Chromatographic hydrophobicity index byfast-gradient RP-HPLC: a high-throughput alternative to logP/logD, Anal.Chem. 69 (1997) 2022–2029.

[23] Q. Wang, H.J. Altermatt, H.B. Ris, B.E. Reynolds, J.C. Stewart, R. Bonnett, et al.,Determination of 5,10,15,20-tetra-(m-hydroxyphenyl)chlorin in tissues byhigh performance liquid chromatography, Biomed. Chromatogr. 7 (1993) 155–157.

[24] T. Ohnishi, N.A.L. Mohamed, A. Shibukawa, Y. Kuroda, T. Nakagawa, S. ElGizawy, et al., Frontal analysis of drug–plasma lipoprotein binding usingcapillary electrophoresis, J. Pharm. Biomed. Anal. 27 (2002) 607–614.

[25] A. de Juan, R. Tauler, Multivariate Curve Resolution (MCR) from 2000: progressin concepts and applications, Cr. Rev. Anal. Chem. 36 (2006) (2000) 163–176.

[26] J. Diewok, A. de Juan, M. Maeder, R. Tauler, B. Lendl, Application of acombination of hard and soft modeling for equilibrium systems to thequantitative analysis of pH-modulated mixture samples, Anal. Chem. 75(2003) 641–647.

[27] R Development Core Team, R: A Language and Environment for StatisticalComputing, R Foundation for Statistical Computing, Vienna, Austria, 2009.

[28] I.H. van Stokkum, K.M. Mullen, V.V. Mihaleva, Global analysis of multiple gaschromatography–mass spectrometry (GC/MS) data sets: a method forresolution of co-eluting components with comparison to MCR-ALS,Chemometr. Intell. Lab. 95 (2009) 150–163.

[29] O. Trott, A.J. Olson, AutoDock Vina: improving the speed and accuracy ofdocking with a new scoring function, efficient optimization, andmultithreading, J. Comput. Chem. 31 (2010) 455–461.

[30] E.F. Pettersen, T.D. Goddard, C.C. Huang, G.S. Couch, D.M. Greenblatt, E.C.Meng, et al., UCSF Chimera – a visualization system for exploratory researchand analysis, J. Comput. Chem. 25 (2004) 1605–1612.

[31] M. Kongshaug, J. Moan, Separation of lipoproteins, albumin and y-globulin bysingle-step ultracentrifugation of human serum. Application I: Binding ofhematoporphyrin to human serum and to albumin, Int. J. Biochem. Cell Biol. 21(1995) 371–384.

[32] O. Rinco, J. Brenton, A. Douglas, A. Maxwell, M. Henderson, K. Indrelie, et al.,The effect of porphyrin structure on binding to human serum albumin byfluorescence spectroscopy, J. Photochem. Photobiol. A, Chem. 208 (2009) 91–96.

[33] S. Patel, A. Datta, Steady state and time-resolved fluorescence investigation ofthe specific binding of two chlorin derivatives with human serum albumin, J.Phys. Chem. B 111 (2007) 10557–10562.

[34] Y. Chen, R. Miclea, T. Srikrishnan, S. Balasubramanian, T.J. Dougherty, R.K.Pandey, Investigation of human serum albumin (HSA) binding specificity ofcertain photosensitizers related to pyropheophorbide-a andbacteriopurpurinimide by circular dichroism spectroscopy and its correlationwith in vivo photosensitizing efficacy, Bioorg. Med. Chem. Lett. 15 (2005)3189–3192.

[35] K. Lang, J. Mosinger, D.M. Wagnerova, Photophysical properties ofporphyrinoid sensitizers non-covalently bound to host molecules; modelsfor photodynamic therapy, Coordin. Chem. Rev. 248 (2004) 321–350.

[36] A. Filyasova, I. Kudelina, A. Feofanov, A spectroscopic study of the interactionof tetrasulfonated aluminum phthalocyanine with human serum albumin, J.Mol. Struct. 565–566 (2001) 173–176.

[37] S. Bonneau, C. Vever-Bizet, P. Morlière, J.-C. Mazière, D. Brault, Equilibrium andkinetic studies of the interactions of a porphyrin with low-densitylipoproteins, Biophys. J. 83 (2002) 3470–3481.

[38] H. Mojzisova, S. Bonneau, C. Vever-Bizet, D. Brault, The pH-dependentdistribution of the photosensitizer chlorin e6 among plasma proteins andmembranes: a physico-chemical approach, Biochim. Biophys. Acta 1768(2007) 366–374.

[39] H. Ibrahim, A. Kasselouri, C. You, P. Maillard, V. Rosilio, R. Pansu, et al., Meso-tetraphenyl porphyrin derivatives: the effect of structural modifications onbinding to DMPC liposomes and albumin, J. Photochem. Photobiol. A, Chem.217 (2011) 10–21.

[40] Y.-F. Ho, M.-H. Wu, B.-H. Cheng, Y.-W. Chen, M.-C. Shih, Lipid-mediatedpreferential localization of hypericin in lipid membranes, BBA –Biomembranes 1788 (2009) 1287–1295.

[41] G. Lajos, D. Jancura, P. Miskovsky, J. Garcia-Ramos, S. Sanchez-Cortes,Interaction of the photosensitizer hypericin with low-density lipoproteinsand phosphatidylcholine: a surface-enhanced Raman scattering and surface-enhanced fluorescence study, J. Phys. Chem. C 113 (2009) 7147–7154.

[42] A. Makky, J.P. Michel, S. Ballut, A. Kasselouri, P. Maillard, V. Rosilio, Effect ofcholesterol and sugar on the penetration of glycodendrimericphenylporphyrins into biomimetic models of retinoblastoma cellsmembranes, Langmuir 26 (2010) 11145–11156.

[43] K. Kuzelova, D. Brault, Interactions of dicarboxylic porphyrins with unilamellarlipidic vesicles: drastic effects of pH and cholesterol on kinetics, Biochemistry34 (1995) 11245–11255.

[44] J.B. Massey, H.J. Pownall, Surface properties of native human plasmalipoproteins and lipoprotein models, Biophys. J. 74 (1998) 869–878.

[45] L.S. Kumpula, J.M. Kumpula, M.-R. Taskinen, M. Jauhiainen, K. Kaski, M. Ala-Korpela, Reconsideration of hydrophobic lipid distributions in lipoproteinparticles, Chem. Phys. Lipids 155 (2008) 57–62.