The Critical Role of Aromatic-Aromatic Drug-Polymer Interactions to Provide Nanomedicines Containing Chloroquine: A Simple Proposal to Provide High Encapsulation, High Stability, and Prolonged Drug Release

Background: Chloroquine (CQ) is a drug commonly used to treat malaria. CQ has also gained interest for the treatment of other chronic diseases such as arthritis, lupus, cancer, diabetes, atherosclerosis, and dermatomyositis, among others. Since CQ is hydrophilic and low molecular-weight, attractive interactions with polymers in aqueous medium are weaker than with water, so that low encapsulation together with uncontrolled and fast release is observed. Importantly, a long-term administration of CQ is suggested, thus the development of formulations with controlled and prolonged release is desirable. Results: Here we propose the use of aromatic interactions between the cationic CQ and the FDA approved polymer poly(sodium 4-styrenesulfonate) (PSS) for the formation and stabilization of nanoparticles (NPs). The strategy consists on the simple mixture of two aqueous solutions containing the oppositelly charged molecules. UV-vis and NMR spectroscopy evidence intimate aromatic-aromatic interactions between CQ and PSS. CQ/PSS molar ratios from 0.2 to 0.5 allow obtaining NPs with spherical shape, size in the range of 170-410 nm, zeta potential from -18 to -45 mV, and particles number in the range of 0.9 -6.6 x 1010 NPs/mL. Selected NPs (CQ/PSS molar ratio 0.4) are stable to wide variations in ionic strength (0-200 mM), pH (2-10) and temperature (20-50 °C). In addition, CQ/polymer 0.4 was also tested but with the absence of the aromatic group in the polymer, and providing smaller (70 nm), lower-concentrated (6.1 x 109 NPs/mL), and unstable particles, conrming the key role of the aromatic group. Furthermore, CQ/PSS NPs are stable during months and can be converted to a reconstitutable powder. Importantly, the selected NPs (CQ/PSS 0.4) show full drug association eciency (100 %), very high drug loading (49 %), very high yield (89 %), and evidencing a drug entrapment/release governed by kinetic associations ( ≈ 99 %). Finally, release studies evidence a controlled and prolonged delivery. Conclusions: Considering the potential uses of CQ for chronic diseases, and the simplicity and eciency of our proposal, it could be considered as a valuable alternative to developed nanomedicines. In addition, this strategy could be used for other drugs and polymers showing similar characteristics to CQ and PSS.


Introduction
Chloroquine (CQ) is a cheap medicine that has been used worldwide for more than 70 years. It is part of the World Health Organization (WHO) model list of essential medicines and is primarily used to prevent and treat malaria [1][2][3]. Like most antimalarial agents, CQ is currently widely used in autoimmune diseases such as severe rheumatoid arthritis [4][5][6] and systemic lupus erythematosus [7][8][9]. In addition, several investigations indicates its potential administration in the treatment of another chronic diseases (i.e. cancer [10,11], diabetes [12,13], atherosclerosis [14], dermatomyositis [15], and Sjögren's syndrome [16], among others). Several factors have in uenced on the attention of CQ, highlighting their well-known immunomodulatory capacity [17].
on June 15th the Food and Drug Administration (FDA) revoked the emergency use authorization for CQ (and their analogous hydroxychloroquine) to treat COVID-19 in United States of America [18]. This decision was supported on the evidence of several side effects that includes heart rhythm problems and others safety issues. Importantly, FDA mention that this decision does not affect the use of CQ for malaria and other autoimmune diseases. In this respect, CQ is well tolerated by most people, even in longterm therapies [19]. This is supported due to their high capacity to bound plasmatic proteins and being eventually deposited into metabolically active tissues. Side effects as retinopathy, cardiomyopathy, myopathy, and neuromyopathy may could appear on prolonged treatments and high-level doses [20].
Considering the potential of this drug to treat several chronic diseases (malaria, arthritis, lupus, cancer, diabetes, atherosclerosis, dermatomyositis and SARS-COV-2, among others) and the appearance of side effects during prolonged treatments at high-level doses (retinopathy, cardiomyopathy, myopathy, and neuromyopathy) the development of formulations with controlled and prolonged drug release could represent a valuable alternative to optimize the e cacy, decrease side effects, and even to explore other uses.
Bearing in mind the physicochemical characteristics of CQ (cationic, hydrophilic, aromatic and low molecular-weight drug, Fig. 1, its entrapment in drug delivery systems such as nanoformulations to provide controlled and prolonged release, represents a main challenge [21][22][23]. The above statement is based on the very low capacity of this kind of molecules to undergo interactions with excipients that surpass their a nity with water, producing low drug association and, consequently, very fast drug release upon exposing the medicines to biological media [21][22][23]. In the state-of-the-art we could nd some attempts to encapsulate CQ in nanocarriers. Among them, PLA [24], dextran [25], PEG [26], solid lipids [27], and PLGA [28] are described as encapsulant agents. In most cases, a drug burst release (≈ 40%) was observed, and the complete release of CQ occurs within 1 to 3 days, thus not providing with controlled/prolonged drug release. We previously reported a green and highly e cient methodology to non-covalently attach three tricyclic drugs (imipramine, cyclobenzaprine, and amitriptyline), all of them hydrophilic and aromatic low molecular-weight molecules, to the hydrophilic and aromatic poly(sodium 4styrensulfonate) (PSS, Fig. 1) polymer. PSS is an approved excipient by the Food and Drug Administration (FDA) [29][30][31][32][33]. In that work we demonstrate that the formation of the NPs is based on ionic a nity and critically dependent on the occurrence of aromatic-aromatic interactions between the drug and the polymer.
The aim of this work is to provide a strategy based on aromatic-aromatic CQ-PSS interactions to create stable nanoformulations with high elaboration e ciencies (association e ciency, drug loading and yield) in addition to controlled and prolonged drug release. We also aim to evidence the critical role of the aromatic group in the selected polymer to form and stabilize de NPs at comparing formulations using a similar but non-aromatic polymer [poly (sodium vinylsulfonate)] (Fig. 1). Importantly, this study could be the basis to elaborate safer and more effective formulations to treat chronic diseases that require longterm therapies with CQ (or another hydrophilic, aromatic and low molecular-weight drug), such as malaria, rheumatoid arthritis, lupus erythematosus, diabetes mellitus, and SARS-COV-2, among others. Based on previous investigations where our group demonstrated the synthesis of NPs comprising imipramine, amitriptyline, or cyclobenzaprine with the hydrophilic and aromatic polymer PSS [21], polymer already approved by the FDA for therapeutic applications [29,30,32,34,35], we postulate in this work a simple strategy to create NPs loaded with CQ.

Elaboration and characterization of CQ/polymer formulations
First, we studied the binding between the only two components proposed to create the NPs (cationic CQ and the anionic polymers PSS or PVS (PVS is used as control due to the absence of the aromatic group).
The UV-vis spectra of CQ (Fig. 2) exhibit strong absorptions attributed to n-π* transitions in the absence and presence of the selected polymers. In particular, a bathochromic effect is evident in the CQ spectrum in presence of PSS (Fig. 2C), showing a clear red shift that is typical for intimate aromatic -aromatic interactions (this effect is not observed for CQ in the presence of PVS) [36,37]. In order to corroborate this observation, the 1 H-RMN technique was selected considering the strong evidence demonstrated to identify changes in the chemical environment of protons by the presence of intimate aromatic-aromatic interactions [38][39][40]. In Fig. 3 can be observed the behavior of the aromatic protons of CQ in the absence (Fig. 3A) and the presence (Fig. 3C) of PSS. As evidenced, the signals of the aromatic protons (between 6.0 and 8.0 ppm) of CQ are strongly affected by PSS and are up eld shifted together with a broadening of the signals (Fig. 3C). This ndings are characteristic for the presence of secondary site-speci c aromatic-aromatic interactions since the hydration shell of the components changes, and the magnetic elds produced by electronic currents of aromatic groups strongly affect the neighbor structures [21,41,42]. A decrease on the molecular mobility is responsible for the broadening of the bands and the low signal-to-noise values.
Further, the formation of CQ/polymer NPs was evaluated. The strategy involves the simply mixing of two aqueous solutions (one containing the cationic drug CQ and the other containing the anionic aromatic or non-aromatic polymers, PSS and PVS, respectively) at room temperature, varying the drug concentration In order to con rm our previous interpretation, the formulations were analyzed by dynamic light scattering (DLS) and laser doppler anemometry (LDA). As evidenced in Fig Other attempts for nanoencapsulation of CQ (using PLA, dextran, PEG, solid lipids or PLGA) reported hydrodynamic diameters around 50-500 nm with PdI in the range of 0.07-0.35 [24][25][26][27][28]. Moreover, nanoparticle tracking analysis (NTA) evidenced nanoparticle concentrations in the range of 0.9 × 10 10 to 6.6 × 10 10 NPs/mL for our formulations. Other studies analyzing this parameter for amitriptyline-PSS NPs, doxorubicin loaded in carbomer NPs, and doxorubicin loaded in skim milk derived nanovesicles, reported values in the range of 1.0 × 10 6 − 4.26 × 10 11 NPs/mL [21,43,44]. Interestingly, when comparing formulations with PSS versus PVS, despite CQ/PVS also formed NPs at molar ratio 0.4 (presumably due to ionic bonds between the CQ and PVS), less e cient interactions are evidenced. This statement is re ected in the signi cantly smaller size (≈ 70 nm for CQ/PVS versus ≈ 170 nm for CQ/PSS) together with the signi cantly lower concentrated particles (more than one order of magnitude less particles per mL) when comparing with CQ/PSS 0.4, as evidenced in Figure S1 (supplementary information section).
The above results con rm the importance of the presence of aromatic groups in the polymer to e ciently obtain NPs, and presumably also in their stabilization (see further studies in this paper). Considering the lower polydispersity (PDI 0.25) and higher nanoparticle concentration (6.6 × 10 10 NPs/mL) obtained for CQ/PSS 0.4, this formulation was selected to additional studies. The morphological structure of this formulation was characterized by scanning transmission electron microscope (STEM), as can be seen in Fig. 4D. The results showed uniform spherical structures with smooth surface. Interestingly, the size evidenced by STEM for this formulation (CQ/PSS 0.4) is smaller than that shown by DLS. These results could re ect the hydrated and partially swollen state of NPs during DLS studies; for STEM analysis, they shrink in the drying process during the sample preparation.
In order to investigate the stability of the obtained NPs (as a function of size and zeta potential), the formulations were subjected to different biological conditions and beyond, in addition to challenging storage conditions. Importantly  highlights the critical role of the aromatic group in the polymer to obtain stable NPs. Furthermore, freezedrying process was used to evaluate the transformation of the CQ/PSS nanoparticle suspension into a reconstitutable dry powder, thus facilitating the transport and storage while preventing biological contamination [45]. In Fig. 6, we can observe the achievement of optimal resuspension of the dry product, without signi cantly altering the size and zeta potential of the original formulations (fresh nanosuspension). Considering both the stability under physiological and storage conditions, and the possibility to convert the nanosuspension into a reconstitutable powder, the obtained CQ/PSS NPs are promising to postulate the formulation as a therapeutic alternative. Despite that in the literature we could nd some investigations focused on identifying the stability of nanoformulations with CQ, our study demonstrates higher simplicity and stability (in suspension and as a powder), promoting this strategy, based on aromatic -aromatic interactions, as an attractive alternative to the state-of-the-art procedures.

Association parameters, nature of the entrapment and release studies of CQ/PSS formulations
In order to validate a selected nanoformulation for drug delivery purposes, the drug encapsulation parameters are critical. The association e ciency (AE), drug loading capacity and yield of CQ/PSS formulations were evaluated for all the range of tested ratios (CQ/PSS ratio between 0.1 and 1.2), and the results are shown in Fig. 7. Interestingly, it could be appreciated AE values in the range of 80-100% and drug loading values in the range of 15-49% for all tested formulations. Additionally, production yield values between 55-89% were obtained. Importantly, for the obtained CQ/PSS 0.4 formulation, an AE of 100%, a drug loading of 49% and a yield of 89% was obtained. The very high values for these encapsulation parameters could be attributed to the high a nity between the CQ and the polymer PSS.
These results are very promising since previous works for CQ loaded NPs comprising PLA, dextran, PEG, solid lipids or PLGA-tocopherol polyethylene glycol succinate reported AE values in the range of 64.1-99.9% [24][25][26][27][28], and drug loading in the range of 12.9-27.8% for PEG or solid lipid NPs [26,27]. The success of our approach stands on the advantage of making intimate interactions (aromatic-aromatic interactions that presumably complement the ionic drug-polymer interactions) to bind the drug to the carrier. Thus, in this proposal, the drug is a structural component for the NPs formation, allowing the use of no more than one excipient to build the NPs, and requiring high load of the drug for this purpose.
Dia ltration studies were carried out to determine the nature of the drug entrapment/release, since the technique allows distinguishing kinetically bound molecules to molecules subjected to binding equilibrium (thermodynamically bound). The former are thought to place at the inner part of the NPs, stabilized in hydrophobic domains, and the latter are thought to bind into hydrated segments at the surface of the NPs. As previously reported, formulations with a higher kinetically bound fraction (% KB) are associated with a controlled and prolonged release pro le [21]. The dia ltration parameters for all tested CQ/PSS formulations (CQ/PSS between 0.1 and 1.2) are shown in Table S1 (see supplementary   information). Importantly, the formulation CQ/PSS 0.4 presented very high values of kinetically bound molecules (99.3 ± 0.3%), thus re ecting a high content of CQ stabilized in the inner structure of the NPs. In addition, a low dissociation constant (K diss ≈ 0.7) is found, re ecting the high tendency of this drug to bind the polymer, thus con rming the potential of the developed formulations and suggesting a controlled and prolonged release of CQ from the NPs.
To determine the release pro le of the NPs CQ/PSS 0.4, conventional dialysis, and continuous dissolution studies in a USP 4 dissolution apparatus were selected as methodologies. Firstly, a sustained and prolonged release pro le of CQ from the NPs in Milli-Q water was obtained. The cumulative drug release by dialysis as a function of time evidenced a maximum release of 3% in 30 days (Fig. 8A-a). In this condition, the binding between the aromatic drug CQ and the aromatic polymer PSS remains mainly inalterable and the release process is conditioned by the fraction of CQ molecules located on the NPs surface (thermodynamically bound) and consequently available to be released. Interestingly, under simulated biological conditions (pH 7.4, 0.13 M NaCl and 37 °C) a biphasic release pro le was observed with an initial release phase (20% in 6 hours), followed by a controlled and prolonged release phase until 30 days (Fig. 8A-a'). The observed controlled and prolonged release pattern represents a big improvement over other nanocarriers encapsulating CQ; dextran and PEG NPs showed a release pro le that reaches 45-93% during 3-8 h (analyzed by conventional dialysis) [25,26], and PLA NPs show 40% of release during 10 h (analyzed by static Franz vertical diffusion cell) [24]. Furthermore, with the aim to study a more dynamic process, the continuous ow cell apparatus USP 4 was performed (Fig. 8B). As could be expected, a greater cumulative release of 16.2% in the rst 6 hours was obtained, evidencing that the exposition of these NPs to a continuous ux of medium accelerates the release of CQ.
Finally, the release mechanism obtained by the t of the data to different mathematical models (i.e., zero order, rst order, Higuchi and Korsmeyer-Peppas) was evaluated. The Korsmeyer-Peppas tted better the experimental data (Table S2, see supplementary information). Considering the characteristics of the components (hydrophilic molecules), and that CQ is a structural component of the NPs, the drug release process should involve the release of CQ molecules from the surface of the NPs, due to equilibrium displacement, and followed by hydration and rearrangement of the boundary between the inner and the surface of the NPs. The detachment of the drug from the NP enhances the hydration of the system, so that the subjacent con ned drug molecules achieve a hydrated environment and establishing equilibrium with the bulk. This process proceeds in a cyclic manner. This complex process is consistent with deviations of the linear or rst order behavior of the release as a function of time.

Conclusions
In conclusion, our results offer an e cient strategy to nanoencapsulate chloroquine (CQ) in nanocarriers. The strategy involves taking advantage of aromatic-aromatic interactions between CQ and PSS (as a unique excipient) for the formation and stabilization of the NPs. CQ/PSS NPs are stable to variations in ionic strength, pH, and temperature. Conversely, a similar formulation but with the absence of the aromatic ring in the polymer (testing ionic-ionic drug-polymer interactions for NPs formation and stabilization), provided smaller, low-concentrate, and unstable particles, con rming the key role of the aromatic group in the polymer. In addition, CQ/PSS NPs are stable at storage conditions and can be easily freeze-dried and reconstituted in water. Furthermore, the selected formulation of NPs shows a full drug association e ciency (≈ 100%), very high drug loading (≈ 49%), and very high yield (≈ 89%). In addition, it was shown that the drug entrapment is governed mainly by kinetic interactions (≈ 99%), and being the drug mostly con ned at the inner of the NPs. Finally, using different release methodologies, we have demonstrated that CQ/PSS NPs offer a controlled and prolonged drug release pro le. Considering the wide therapeutic possibilities for CQ in the treatment of a variety of chronic diseases, and the advantages that our technology offers in comparison with other approaches (being simple, cheap, e cient, and low-pollutant), the proposed strategy could represent the basis for a medicine with controlled/prolonged release to treat different pathologies where this release pattern of CQ is required (malaria, rheumatoid arthritis, lupus erythematosus, and diabetes, among others). In addition, this strategy could be used for other drugs and polymers showing similar characteristics to CQ and PSS.

Materials
Chloroquine diphosphate (515.9 g/mol) was purchased from Sigma Aldrich (USA) and was used as

Preparation of CQ/polymer formulations
The CQ/polymer formulations were synthetized according to the method previously reported by our group [21]. Brie y, 5.0 mL of an aqueous solution containing CQ was added to 5.0 mL of an aqueous solution containing the anionic polymers (PSS and PVS), at pH 7, and exposed to continuous stirring (5 min). The nal apparent concentration was de ned in order to obtain different molar ratios de ned as

Physicochemical characterization of the nanoparticles
The hydrodynamic diameter and zeta potential of the formulations were determined by dynamic light scattering (DLS) and laser Doppler anemometry (LDA) using a ZetaSizer NanoZS (Malvern Instruments, UK).
The determination of nanoparticle concentration was performed in a NanoSight NS300 (Malvern Instruments, UK). The samples were diluted up to 100 times with Milli-Q water to achieve an optimum concentration range of 10 7 -10 9 particles/mL. A minimum of ve videos (one minute each one) of the particles moving under Brownian motion were captured by the NanoSight. The videos were then analyzed for size distribution and particle concentration using the built-in NTA v 3.0 software (Malvern, UK).
The morphological characterization was carried out in a scanning transmission electron microscope (STEM), model Inspect F-50 (FEI, Holland). STEM images were obtained by sticking a droplet (20 µL) of the formulation to a copper grid (200 mesh, covered with Formvar) for 2 min, then removing the droplet with lter paper avoiding the paper touching the grid, then washing the grid twice with a droplet of Milli-Q water for 1 min and removing the droplet with lter paper. Subsequently, the sample was stained with a solution of 1% (w/v) phosphotungstic acid by adding a droplet of this solution to the grid for 2 min and then removing with lter paper. Finally, the grid was dried at room temperature for at least 1 h before being analyzed.

4.2.4
Drug association e ciency, drug loading, and yield of the CQ/PSS formulations Drug association e ciency, drug loading, and yield were obtained as previously described [21,45]. Brie y, the association e ciency of CQ in the CQ/PSS formulations was determined by analyzing the ratio between the amount of drug associated in the formulation and the total initial drug (associated and nonassociated). The drug loading (% w/w) was calculated by dividing the amount of drug associated by the total weight of the formulations. The yield was calculated by dividing the total nal weight of each formulation by the total initial weight of the components (CQ + PSS). The drug content into the formulations was calculated indirectly by quantifying the free drug in the medium; the separation of NPs and free drug was done by using Vivaspin®6 tubes (MWCO 3 kDa, 5000 G x 40 min). The quanti cation of the CQ was done by measuring the absorbance at 343 nm (Agilent 8453 spectrophotometer, USA), respectively. The standard curve of CQ was linear (R 2 > 0.999) in the range of concentrations between 4 × 10 − 5 M and 3 × 10 − 6 M (molar extinction coe cient was 30449 M − 1 cm − 1 ). Finally, for the calculation of the total nal weight, 1 mL of each formulation was lyophilized in glass vials, which were weighed before adding the formulation and after freeze-drying to assess the total solid mass (glass vials + formulation). The lyophilization procedure was done in the freeze-dryer equipment FreeZone 1 (Labconco, USA) using a high vacuum pump (50 mTorr) for 24 h. 4.2.5 Kinetic and/or thermodynamic drug entrapment, and dissociation constants evaluated by dia ltration The dia ltration method was selected to investigate the kinetic and/or thermodynamic drug entrapment, and the dissociation constants of their equilibrium binding to excipients/formulations [36,37,[46][47][48][49][50][51][52][53].
The unit used for dia ltration analyses consisted on a dia ltration cell (10 mL, Amicon 8010), a regenerated cellulose membrane (cutoff of 5000 Dalton, Merck, Germany), a reservoir, a selector, and a pressure source (Merck-Millipore, Germany). The method consists on passing through the dia ltration cell containing the formulation of NPs in water, a continuous liquid supply from the donor chamber (reservoir) keeping a constant volume in the dia ltration cell. A two-compartment system model is considered for data treatment [49]. For the dia ltration experiments, aliquots of 10 mL of the formulations were added into the dia ltration cell and then ltered under 3 bars of pressure and magnetic stirring. Milli-Q water (pH 7) was used as solvent for dia ltration. A total of 8 samples (approx. 5 mL) were collected and the concentration of CQ in each sample was determined by spectrophotometry [21,53].
In this paper, dia ltration was performed to determine the fraction of CQ kinetically or thermodynamically bound to PSS. Details of dia ltration procedures, mathematical analysis and results are widely explained in previous studies [36,37,[46][47][48][49][50][51][52][53]. Brie y, the parameters v and u represent the initial fraction of CQ thermodynamically bound to the particles, thus in equilibrium, and the initial fraction of drug bound to the nanoparticles whose release is kinetically controlled, respectively. The parameter j is related to the strength of interaction corresponding to the reversibly bound drug fraction (v). The parameters u m and k m correspond to u and j values, respectively, obtained in blank experiments as controls, performed by dia ltration of the drug in the absence of other excipients or formulations.
Subsequently, the thermodynamically bound (TB) and kinetically bound (KB) fractions were determined as follows: In addition, the dissociation constants (K diss ≡ c free /c rev−bound ), where c rev−bound is the concentration of drug reversibly bound to the matrix, and c free is the concentration of drug free in the bulk, could be also determined by dia ltration following the Eq. (3).

In-vitro drug release studies
In-vitro drug release assays were carried out using two different methods: conventional dialysis and USP apparatus 4 (continuous ow-cell).
4.2.6.1 Dialysis: 5 mL of CQ/PSS formulations were added in a dialysis bag (MWCO 10 kDa, ThermoScienti c, USA). The dialysis system was immersed in 95 mL of Milli-Q water (pH 7.0) or simulating biological conditions (Milli-Q water, 0.13 M NaCl, pH 7.4), and kept at 37 °C and continuous agitation (C-MAG HS 7, IKA, Staufen, Germany). The experiments were carried out until 30 days, aliquots (500 µL) of the solution were withdrawn at certain time intervals and replaced with an equal volume of fresh release medium. The amount of released CQ was determined by measuring the absorbance of each aliquot by spectrophotometry (Agilent 8453 spectrophotometer, USA).

USP apparatus 4:
For this assay, the set-up of the continuous ow method is combined with a dialysis membrane to contain the nanoformulations into the cell [54]. In brief, 5 mL of CQ/PSS NPs were added in a dialysis bag (MWCO 10 kDa, ThermoScienti c, USA) and then immersed into the ow-cell (12.5 mL capacity).
Drug release studies were assessed using 250 mL of Milli-Q water at pH 7. The continuous ow-cell (Sotax CE 6, Sotax AG, Switzerland) is operated in close con guration at 37 °C and with a ow rate of 4 mL/min approximately. The experiment was carried out for 6 hours, aliquots (500 µL) of the solution were withdrawn every 15 min and replaced with an equal volume of fresh Milli-Q water. The amount of released CQ was determined by measuring the absorbance of each aliquot by spectrophotometry (Agilent 8453 spectrophotometer, USA).
To investigate the release mechanism (in both conventional dialysis and the USP apparatus 4 data), mathematical kinetics modelling was done using the program DDSolver [55]. The coe cient of determination (R 2 ), the Akaike information criteria (AIC), and the model selection criteria (MSC) parameters were considered for the model selection. Finally, the release data was tted to zero order, rst order, Higuchi and Korsmeyer-Peppas [56].

Availability of data and materials
All data generated or analysed during this study are included in this published article [and its supplementary information les].