Optimization and modeling of the remote loading of luciferin into liposomes

Accepted Manuscript Title: Optimization and modeling of the remote loading of luciferin into liposomes Author: Anders Højgaard Hansen Michael A. Lomholt Per Lyngs Hansen Ole G. Mouritsen Ahmad Arouri PII: DOI: Reference:

S0378-5173(16)30338-6 http://dx.doi.org/doi:10.1016/j.ijpharm.2016.04.055 IJP 15719

To appear in:

International Journal of Pharmaceutics

Received date: Revised date: Accepted date:

5-2-2016 15-4-2016 19-4-2016

Please cite this article as: Hansen, Anders Hojgaard, Lomholt, Michael A., Hansen, Per Lyngs, Mouritsen, Ole G., Arouri, Ahmad, Optimization and modeling of the remote loading of luciferin into liposomes.International Journal of Pharmaceutics http://dx.doi.org/10.1016/j.ijpharm.2016.04.055 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Optimization and modeling of the remote loading of luciferin into liposomes Anders Højgaard Hansena,b, Michael A. Lomholta, Per Lyngs Hansena, Ole G. Mouritsena,b, Ahmad Arouria,b*

5

a

MEMPHYS-Center for Biomembrane Physics, Department of Physics, Chemistry, and

Pharmacy, University of Southern Denmark, Odense, Denmark. b

The Lundbeck Foundation Nanomedicine Research Center for Cancer Stem Cell Targeting

10

Therapeutics (NanoCAN), University of Southern Denmark, Odense, Denmark.

*Address for Correspondence: MEMPHYS-Center for Biomembrane Physics, Department of Physics, Chemistry, and Pharmacy, University of Southern Denmark, Campusvej 55, DK-5230 Odense, Denmark. Tel.: +45 6550 3506. Fax: +45 6550 4048. E-mail addresses: 15

[email protected] (Ahmad Arouri), [email protected] (Anders H. Hansen), Michael A. Lomholt

([email protected]),

[email protected]

(Ole

G.

Per

Lyngs

Mouritsen).

Hansen

([email protected]),

URL:

http://www.memphys.dk;

http://www.nanocan.org.

1

20

Keyword: Liposome; Luciferin remote loading; Carboxylate gradient; Drug encapsulation; Phospholipase A2 enzyme; Modeling; Diffusion.

2

Graphical abstract

25 Abstract We carried out a mechanistic study to characterize and optimize the remote loading of luciferin into preformed liposomes of 1,2-dipalmitoyl-sn-glycero-3-phosphocholine/1,2-dipalmitoyl-snglycero-3-phosphoglycerol (DPPC/DPPG) 7:3 mixtures. The influence of the loading agent 30

(acetate, propionate, butyrate), the metal counterion (Na+, K+, Ca+2, Mg+2), and the initial extraliposomal amount of luciferin (

) on the luciferin Loading Efficiency (LE%) and luciferin-to-

lipid weight ratio, i.e., Loading Capacity (LC), in the final formulation was determined. In addition, the effect of the loading process on the colloidal stability and phase behavior of the liposomes was monitored. Based on our experimental results, a theoretical model was developed 35

to describe the course of luciferin remote loading. It was found that the highest luciferin loading was obtained with magnesium acetate. The use of longer aliphatic carboxylates or inorganic proton donors pronouncedly reduced luciferin loading, whereas the effect of the counterion was modest. The remote-loading process barely affected the colloidal stability and drug retention of the liposomes, albeit with moderate luciferin-induced membrane perturbations. The correlation

40

between luciferin loading, expressed as LE% and LC, and conditions the maximum LC was attained using an

was established, and under our

of around 2.6 µmol. Higher amounts of

luciferin tend to pronouncedly perturb the liposome stability and luciferin retention. Our theoretical model furnishes a fair quantitative description of the correlation between

and

luciferin loading, and a membrane permeability coefficient for uncharged luciferin of 1×10-8 45

cm/s could be determined. We believe that our study will prove very useful to optimize the remote-loading strategies of moderately polar carboxylic acid drugs in general.

3

1. Introduction Liposomes are artificial lipid aggregates that comprise a spherical bilayer arrangement of amphiphiles encasing an aqueous inner core. The potential use of liposomes in systemic drug 50

delivery was highlighted back in the seventies (Gregoriadis et al., 1974), owing to their small colloidal size, controllable surface properties, large cargo capacity, biodegradability, and biocompatibility (Čeh et al., 1997). Liposomes can modulate the pharmacokinetic profile of the encapsulated drug as well as reduce the drug toxicity (Scheinberg et al., 2010). In addition, liposomal platforms can be further developed for active and selective drug delivery to the target

55

site (Arouri et al., 2013; Park, 2014). Because of the limited amount of lipids that can be administered systemically and in order to deliver the drug in clinical efficacious concentrations, it is essential that the liposomes are efficiently packed with the active drug compounds (Arouri et al., 2013). Whereas passive loading of water-soluble molecules often results in low encapsulation efficiencies, remote loading can

60

provide a highly efficient alterative for drug encapsulation (Gubernator, 2011; Sur et al., 2014; Zucker et al., 2009). The remote loading process is based on the transport of molecules against their concentration gradient as the molecules are transported from the bulk solution into preformed liposomes (Cullis et al., 1991). The procedure can be used to remote load amphipathic weak acids with carboxylic acid functional groups or weak bases with amine functional groups

65

(Nichols and Deamer, 1976; Zucker et al., 2009). Since most known drug compounds either contain acidic (20%) or basic (75%) functional groups (Barenholz, 2001; Joguparthi and Anderson, 2008; Kerns and Di, 2008), remote loading can prove very useful to achieve high drug-to-lipid ratios, high loading efficiencies, and improved retention of many available amphiphilic molecules.

4

70

So far, the two most extensively investigated remote loading methods rely on establishing either a transmembrane pH- or a transmembrane concentration-gradient. Other strategies also exist, like utilizing transmembrane valinomycin-dependent K+ diffusion potentials (Mayer et al., 1985). In the pH-gradient method, a low intra-liposomal pH and a high extra-liposomal pH are established to trap basic compounds containing amine functional groups (Cullis et al., 1991;

75

Madden et al., 1990; Mayer et al., 1986), such as propranolol and doxorubicin. As the uncharged drug diffuses through the liposomal bilayer into the intra-liposomal interior it becomes protonated (charged) and therefor cannot diffuse back into the bulk. By an equivalent mechanism, carboxylic acids can be remotely loaded using a high intra-liposomal pH and a low extra-liposomal pH (Kheirolomoom et al., 2010).

80

In the concentration-gradient method, a high intra-liposomal concentration of an appropriate organic salt can drive the accumulation of the drug inside the liposomes, which is coupled to the countertransport of the membrane-permeant state of the encapsulated ion pair. For instance, basic drugs, like bupivacaine, quinidine, and doxorubicin, can be remotely loaded using an ammonium sulfate gradient (Chen et al., 2010; Lasic et al., 1992; Zucker et al., 2009). At alkaline pH,

85

ammonium sulfate will dissociate into sulfate anions and ammonium cations that will further produce ammonia and protons. Ammonia can freely exit the liposomes down its concentration gradient (membrane permeability coefficient, P = 0.13 cm/s), whereas the sulfate counter anion has a much lower membrane permeability (P < 10-12 cm/s) and therefore will remain inside the liposomes (Clerc and Barenholz, 1995). The process will decrease the intra-liposomal pH,

90

thereby allowing for the trapping of basic drug compounds inside the liposomes in a fashion similar to the pH-gradient method (Clerc and Barenholz, 1995). Acidic drugs, like diclofenac, glucocorticoid prodrugs, and methylprednisolone succinate, can also be remotely loaded using an

5

acetate gradient of different salts (Avnir et al., 2007; Hwang et al., 1999; Zucker et al., 2009), since acetic acid (AH) can diffuse freely through the liposomal membrane (P = 6.6x10-3 cm/s 95

(Walter and Gutknecht, 1984)), which will increase the intra-liposomal pH allowing for trapping of carboxylic acid compounds. The remote loading of weak acids and bases will maintain electroneutrality in both intra-liposomal and extra-liposomal compartments (Clerc and Barenholz, 1995). The concentration-gradient method appears to be more efficient, especially for drugs with limited water solubility (Hwang et al., 1999).

100

The drug surrogate

and

molecular

reporter

(S)-2-(6-hydroxybenzo[d]thiazol-2-yl)-4,5-

dihydrothiazole-4-carboxylic acid (luciferin) is a readily water soluble compound. As illustrated in Fig. 1, luciferin (L) can be remotely loaded into liposomes using a carboxylic acid gradient, which will drive the transport of the protonated (uncharged) form of luciferin (LH2) through the lipid membrane leading to an accumulation of luciferin inside the liposomes (Armarego and 105

Chai, 2013; Cern et al., 2012; Zucker et al., 2009). Unlike doxorubicin, the salts of which can form low-solubility gel-like or fibrous-like bundles inside the liposome (e.g., sulfate and citrate salts) (Lasic et al., 1992; Li et al., 1998; Madden et al., 1990), luciferin molecules do not easily form precipitates. Despite the advantages of drug remote-loading methods, only few attempts have been made to

110

understand the effect of the loading conditions, which could allow for process optimization as well as prediction of the content and properties of the final formulation (Zucker et al., 2009). In the present mechanistic study, we investigated the process of luciferin remote-loading into preformed liposomes as well as the effects of the loading agent (acetate, propionate, butyrate), the metal counterion, and the initial extra-liposomal luciferin concentration on the luciferin

115

loading efficiency (LE% = (encapsulated luciferin / total luciferin)×100%) and the luciferin-to-

6

lipid weight ratio in the final formulation. The colloidal stability and the phase behavior of the luciferin-loaded liposomes were also checked using dynamic light scattering (DLS) and differential scanning calorimetry (DSC), respectively. In order to enable the prediction of the outcome of the luciferin remote-loading process under 120

various loading conditions, developing a model based on our experimental results will prove very useful in this regard. Theoretical frameworks modeling the final equilibrium state for molecule loading into liposomes with semipermeable membranes already exist (Čeh and Lasic, 1995, 1997). We have here extended available models by including the kinetics of the loading process taking into account the effect of a finite loading time. A theory has also been developed

125

for the case of coupled transport involving diffusion of uncharged components (like AH and LH2 in our case), namely, a mechanism by which the influx of luciferin (LH2) into the liposome is driven by the efflux of AH. The lipid formulations used in this study are based on earlier work in our lab on secretory phospholipase A2 (sPLA2)-sensitive liposomes (Hansen et al., 2015). sPLA2 is a lipolytic enzyme

130

that is overexpressed in several cancer types and can efficiently degrade liposomes releasing their payload (Andresen et al., 2005; Arouri et al., 2013; Arouri et al., 2015). In the present study,

luciferin

was

remote

loaded

into

preformed

liposomal

systems

of

phosphatidylcholine/phosphatidylglycerol (PC/PG) 7:3 mixtures. The presence of anionic lipids enhances the enzymatic activity of sPLA2 as well as aids colloidal stabilization of the liposomes. 135

In this context, luciferin will serve both as a drug surrogate and a reporter molecule, which is able to undergo a luminescent redox reaction in the presence of the enzyme luciferase (Marques and Esteves da Silva, 2009), allowing for the detection of luciferin release (Ignowski and Schaffer, 2004; Kheirolomoom et al., 2010). The acetate loading agent, irrespective of the metal

7

counterion, and moderate luciferin concentrations were found to provide the optimum loading 140

efficiency. In general, the liposome stability and drug retention appear not to be strongly influenced by the remote-loading process applied in this study.

2. Materials and Methods 2.1. Materials 145

1,2-Dipalmitoyl-sn-glycero-3-phosphocholine

(DPPC),

1,2-dipalmitoyl-sn-glycero-3-

phosphoglycerol (DPPG), and 1,2-dimyristoyl-sn-glycero-3-phosphoglycerol (DMPG) were purchased from Corden Pharma LLC (Switzerland). Luciferin (sodium salt) was purchased from Regis Technologies (USA). All other chemicals and solvents were purchased from SigmaAldrich Chemicals Co. (Germany). All substances were used as received without any further 150

purification or modification.

2.2. Preparation of liposomes The lipid mixtures were prepared in a chloroform/methanol 5:1 mixture. The lipid solutions were heated under nitrogen flow to evaporate the organic solvents until a lipid film was formed, and 155

they were placed thereafter in vacuo for 24 h. The lipid film was hydrated with the remote loading solution (120 mM, pH 6.0) of acetate salts (sodium, potassium, calcium, and magnesium), sodium propionate, or sodium butyrate. If necessary, pH of the lipid-containing solution was re-adjusted to 6.0 using hydrogen chloride or sodium hydroxide. Ionic strength and osmolarity of the remote loading solutions were adjusted to match those of the potassium sulfate

160

buffer (see below) using sodium chloride and D-glucose, respectively. Each lipid sample (30 – 40 mg/mL, 2 mL) was heated above the lipid main phase transition temperature (Tm) and transferred

8

to a cryo-vial. The lipid suspension was freeze-thawed 10 times before being extruded 15 – 18 times through two 100 nm polycarbonate filters using an Avanti Mini Extruder (Avanti Polar Lipids, Alabaster, AL, USA) at a temperature higher than the phase transition temperature of the 165

lipid or lipid mixture.

2.3. Remote loading of luciferin into preformed binary liposomes The freshly prepared liposomal suspension was passed through a size-exclusion column from Bio-Rad, containing Sephacryl S-200 from GE Healthcare. This step was performed in order to 170

exchange the bulk remote loading solution for a 120 mM potassium sulfate solution at pH 6.0. The remote loading of luciferin into preformed liposomes was realized via the addition of luciferin, dissolved in 120 mM potassium sulfate (pH 6.0), to the liposomes in the potassium sulfate solution. The samples were covered in aluminum foil to protect luciferin from photodegradation. Subsequently, free luciferin was removed using size-exclusion column

175

chromatography as described above. A FLUOstar Omega Microplate Reader (BMG LABTECH) in a 96-well microplate mode was used to determine the amount of encapsulated luciferin by measuring the absorbance of luciferin in 0.5 M carbonate buffer pH 11.5 at 385 nm (liposomes were lysed using 1 vol.% Triton X-100). Luciferin concentration was calculated using an extinction coefficient of 18,200 M-1cm-1 determined in 0.5 M sodium carbonate buffer

180

(pH 11.5). Triton X-100 showed no effect on luciferin absorbance (unpublished observations). The phospholipid concentration was determined using a procedure adapted from the Bartlett’s phosphate assay (Bartlett, 1959).

2.4. Differential scanning calorimetry (DSC)

9

185

DSC measurements were carried out using a Nano-DSC (Calorimetry Science, Provo, Utah, USA). The liposomes were freshly prepared, and a total lipid concentration of 1 – 2 mM was used for each experiment. The samples were degassed for 10 min before being loaded into the DSC cell, and the reference cell was filled with buffer. A scanning rate of 0.5°C/min was applied.

190 2.5. Dynamic light scattering (DLS) Size distributions were measured in a DLS apparatus from Brookhaven-BI-2000SM systems equipped with a photodiode from Avalanche. The light source was a 35 mW 633 nm (red) HeNe laser, and the scattered light was collected at an angle of 90°. The samples were measured for at 195

least 2 min at 25°C, and the CONTIN algorithm was used to fit the acquired data.

2.6. Modeling of loading kinetics To describe the loading kinetics, we have formulated a simple pH-based remote loading model with slow diffusion of neutral luciferin (LH2) and acetic acid (AH) moieties across the liposomal 200

membranes followed by rapid local acid-base equilibration. In this model, the liposomal membrane separates two different types of compartments with total volumes of

and

,

representing the volume inside the liposomes and the bulk volume outside the liposomes, respectively. In the present study, this model is used to determine the membrane permeability coefficient for luciferin by fitting the model to the experimental data. 205

Acid-base equilibria for LH , LH , LH , L , AH and A are described by standard bulk massaction equations (as illustrated in Fig. 1): [LH ] [H ] = [L ] [H ] =

[LH ] , [L ] [H ] =

[LH ] , [LH ] [H ] =

[LH ] , [A ] [H ] =

[LH ] ,

[AH] , [OH ] [H ] =

10

and similar equations for the outside compartment, e.g., for [LH ] , with the equilibrium constants given in Table 1. It should be noted that self-consistency requires the constraint =

210

.

Conservation of the total amount of luciferin [L] = [LH ] + [LH ] + [LH ] + [L ] and acetate [A] = [AH] + [A ] , with similar equations for the outside compartment, can be described by and 215

[L] +

[L] =

[A] +

and

[A] =

, respectively. The symbols

denote the number of moles of luciferin and acetate ions, respectively.

The transport of neutral species across the membrane is governed by [ ]

=

(1)

→

where →

=

([

] −[

])

[A]

→

(2)

and =

(3)

where →

with 220

=

being the total surface area of the liposomes,

acetic acid, and

and

(4)

([AH] − [AH] ) →

the influx of luciferin,

→

the influx of

the permeabilities of neutral luciferin and acetic acid, respectively.

Equations 1 to 4 express that the only way for luciferin and acetic acid to exchange between the inside and the outside of the liposomes is by diffusive transport (Fick’s law) of neutral species across the liposome membranes.

11

Finally, the concentrations [H ] and [H ] 225

of protons are determined through charge

conservation, i.e., [

] −[

] −[

] −[

] −[

] − [

] =

(5)

Here the constant is determined by the initial concentrations, in particular the initial pH inside the liposomes. A similar equation is also valid for the outside compartment. Given well-defined initial [L] , [A] and pH values inside and outside the liposomes, and assuming that [L] and [A] are zero initially, one may proceed to the time dependence of all 230

species by solving the above equations. In practice, we implemented this procedure by differentiating Equation 5 and its outside partner with respect to time, which together with Equations 1 and 3 yield four coupled differential equations that can be solved numerically for the four unknown quantities [L] , [A] , [H ] and [H ] as functions of time. All the other concentrations entering the differential equations are eliminated through the constraints and

235

mass-action equations. Using the parameter values from Table 1, the differential equations for the durations of our experiments could be solved.

3. Results and Discussion 3.1. Optimization of the remote-loading process 240

In order to optimize the remote loading of luciferin into liposomes, different carboxylate salts serving as loading agents were screened. The effects of the chain length of the aliphatic carboxylic acids (C2 – C4) and the metal counterion (sodium, potassium, calcium, magnesium) on the luciferin loading efficiency (LE%) into DPPC/DPPG 7:3 liposomes are shown in Fig. 2. Whereas the use of potassium sulfate was associated with a low LE% of 3%, the use of aliphatic

245

sodium carboxylates resulted in a much higher LE% that was inversely proportional to the length

12

of the aliphatic chain ranging from 34% for C2 to 16% for C4. Since the pKa values for the studied loading agents are comparable (Lide, 2009), the chain-length dependence can be attributed to the membrane partitioning coefficient (K) of the carboxylates (see Fig. 2A) (Walter and Gutknecht, 1984); a higher K will enhance the accumulation of the carboxylates inside the 250

lipid bilayer and consequently decrease their effective molecular concentration. The accumulation of the carboxylates inside the lipid bilayer can also lead to membrane perturbations that reduce luciferin retention. The low LE% obtained with potassium sulfate is linked to its extremely low permeability coefficient (P < 10-12 cm/s) (Clerc and Barenholz, 1995), and therefore an LE% of 3% can represent the LE% of passive loading.

255

As acetate was the most effective loading agent, the role of the acetate counterion was investigated. As shown in Fig. 2B, the acetate counterion (sodium, potassium, calcium, and magnesium) barely influenced the LE% and the luciferin-to-lipid weight ratio. The highest luciferin-to-lipid weight ratio (61 µg L/mg lipid) was obtained with magnesium acetate, which therefore was used throughout the rest of the study.

260

The impact of the remote-loading procedure on the colloidal stability of the liposomes was assessed using dynamic light scattering (DLS). As shown in Fig. 3A, the use of magnesium acetate buffer (pH 6.0), the exchange of the bulk magnesium acetate buffer for potassium sulfate buffer (pH 6.0), and the encapsulation of luciferin only modestly altered the average size of the liposomes. The luciferin-loaded liposomes were relatively stable for 55 days (see Fig. 3B).

265

The influence of the different buffer systems and the luciferin loading on the phase behavior of the liposomes was characterized using differential scanning calorimetry (DSC). Since DPPC and DPPG have a similar main phase transition temperature (Tm around 42°C), DPPG (C16) was replaced with DMPG (C14) that is shorter by two carbons and has a lower Tm (Tm around 23 °C)

13

to enable the detection of possible lipid segregation. This should not influence our results since 270

luciferin and the buffer systems interact mainly with the head-group region and do not insert deeply into the lipid bilayer. In addition, DPPC and DMPG lipids are completely miscible and form homogeneous bilayers in both gel and liquid–crystalline phases (Arouri et al., 2009; Garidel et al., 1997). As displayed in Fig. 4, the presence of magnesium acetate inside and outside the liposomes caused phase segregation resulting in two peaks at 38.5°C and 40.3°C.

275

DPPC/DMPG liposomes undergo the main transition at around 33°C (Garidel et al., 1997). The peak at 38.5°C probably corresponds to DMPG-enriched domains stabilized by Mg2+ ions, whereas the peak at 40.3°C belongs to DPPC-enriched domains (DPPC has Tm around 42°C). When the magnesium acetate outside the liposomes was exchanged for potassium sulfate, only one DSC peak was observed at 38.9°C. This Tm value is still higher than that of the pure system

280

(around 33°C), which indicates the pronounced stabilization of the lipid bilayer by the magnesium ions, however without any detectable lipid demixing. The equilibrated DSC thermograms of luciferin-loaded DPPC/DMPG 7:3 liposomes demonstrate that the lipid bilayer was moderately perturbed by luciferin molecules inside the liposomes, which resulted in a reduction in the transition enthalpy. As our encapsulation and DLS data suggest, the luciferin-

285

perturbing effects do not seem to pronouncedly alter the stability and drug retention of the liposomes.

3.2. Efficiency of the remote-loading process The most commonly used quantities to evaluate the efficiency of the drug-loading process into 290

nanoparticles are (A) Loading Efficiency (LE%), which is equal to the percentage of the drug added to the system that has successfully been encapsulated, and (B) Loading Capacity (LC),

14

which represents the drug content of the nanoparticle, which can be expressed as weight of the encapsulated drug per weight of the carrier. Although the two quantities are interrelated, the relationship between LE% and LC is not well understood and it highly depends on the drug295

carrier system. The ultimate goal of a drug loading process is to maximize both quantities, i.e., LE% and LC. Fig. 5 illustrates the dependence of luciferin LE% and LC, expressed as weight of luciferin in microgram per weight of lipid in milligram, on the amount of luciferin initially added to the system (

300

, in µmol). At low

, namely from 0 to 1.2 µmol, LC increased almost linearly

before a maximum of 54 µg L/mg lipid was reached at 2.6 µmol

. Increasing

to 3.1

µmol reduced LC pronouncedly. This observation could be attributed to the membraneperturbing effect of luciferin at this high luciferin concentration, which would reduce the permeability barrier of the liposomes, i.e., reduce luciferin retention. A similar “overloading” effect has been reported before in literature (Zucker et al., 2009). In contrast to LC, the LE% 305

parameter increased rapidly with which the LE% steadily declined with

reaching a maximum of 39% at 0.5 µmol luciferin, after approaching 6% at 3.1 µmol of added luciferin.

For low-cost molecules, a low LE% would be acceptable to achieve a high LC, whereas for highcost molecules a compromise should be made to achieve a reasonable LC without wasting much of the starting material. It appears from our data that there is an optimum initial drug 310

concentration window for achieving the maximum LC with acceptable LE%, since a higher initial drug concentration can adversely affect the liposome stability and drug retention. For luciferin and under our experimental conditions, the optimum window sets around 2.6 µmol luciferin.

15

315

3.3. Modeling of the loading kinetics We developed a simple pH-based remote loading model describing the slow diffusion of neutral luciferin (LH2) and acetic acid (AH) moieties across the liposomal membranes as depicted in Fig. 1 (see section 2.6). The model was used to calculate the membrane permeability coefficient for luciferin by fitting the model to the experimental data.

320

The total luciferin concentration inside the liposomes [L]i at a certain time t can be determined by solving equation (1) (

[ ]

=

→

). In this context, the termination time

refers to the

length of the remote loading procedure. In our experiments, the average termination time was three days. Subsequently, and after translating the experimentally obtained luciferin concentrations inside the liposomes [L]i into amounts of encapsulated luciferin, i.e., using 325

=

[L] , a plot showing the relationship between the amount of added luciferin

amount of encapsulated luciferin model (

at time

was constructed (see Fig. 6). By fitting the

(model fit) onto the experimental data set (

permeability coefficient

and the

(experimental)), a membrane

for neutral luciferin of 1×10-8 cm/s could be determined. Our

permeability coefficient value is roughly three times larger than a value of 3.6×10-9 cm/s 330

obtained by Ferrara and co-authors (Kheirolomoom et al., 2010). The difference is likely caused by the different lipid systems and encapsulation conditions used in the two studies. In addition, the permeability coefficients for lipid systems usually fall within a wide range (Tamm and Kalb, 1993). To illustrate the impact of the termination time, theoretical model curves have been implemented

335

using the same remote loading conditions as in our experiment but at longer

of 1, 2, and 3

weeks and at infinity (see Fig. 6). The curves clearly demonstrate the dependence of the loading efficiency on the duration of the experiment, which appears to follow a hyperbolic-type pattern.

16

4. Conclusions Although liposomal systems are among the most widely investigated platforms for drug delivery 340

owing to their versatile and promising characteristics, the number of liposome-based formulations that have made it beyond the pre-clinical stage is still limited. The main hurdle is technical, namely preparing stable liposomes efficiently packed with drug molecules that can reliably deliver the drug precisely to the target site in clinically efficacious concentrations. This has been addressed by exploring more advanced strategies including the addition of targeting-

345

modules and the design of bio-responsive liposomes, all of which added to the complexity of the liposomal formulations (Arouri et al., 2013; Park, 2014). Notwithstanding all efforts, scaling up the production of advanced liposomal formulations from laboratory scale to pilot scale production and providing the quantities and qualities required for clinical testing are still very challenging.

350

In the present paper, we carried out a mechanistic study to understand the luciferin remoteloading process into preformed liposomes. In this context, the use of luciferin as a model drug will allow for monitoring the drug (i.e., luciferin) release process by adding the luciferase enzyme to the bulk and monitoring the generated bioluminescence. In our case, the drug release can be triggered by sPLA2 enzymes by using sPLA2-liable lipid formulations. The efficiency of

355

the luciferin remote-loading process applied in the present study and the feasibility of exploiting sPLA2 to activate the release of luciferin from liposomes have been demonstrated before in model as well as cell-based systems (Arouri et al., 2015; Hansen et al., 2015). As our results show, the highest luciferin loading was obtained with the aliphatic carboxylate salt magnesium acetate. The use of longer aliphatic carboxylates or inorganic proton donors

360

pronouncedly reduced luciferin loading, whereas the effect of the counterion was much milder.

17

The luciferin remote-loading process barely affected the colloidal stability and drug retention of the liposomes, albeit the moderate membrane perturbations observed with DSC. The established dependence of the luciferin loading efficiency (LE%) and loading capacity (LC) on the initial amount of added luciferin suggests that, under our conditions, the maximum LC with acceptable 365

LE% can be attained using around 2.6 µmol luciferin. Higher amounts of luciferin tend to pronouncedly perturb the liposome stability and luciferin retention. Our model, though relatively simple, is capable of describing the correlation between the initial amount of added luciferin and luciferin loading capacity. We believe that our observations and approach will prove very useful to optimize the remote-loading strategies of moderately polar carboxylic acid drugs in general.

370 Acknowledgments This work was supported by The Lundbeck Foundation Center of Excellence NanoCAN (Nanomedicine Research Center for Cancer Stem Cell Targeting Therapeutics), the Danish Council for Independent Research-Technology and Production Sciences (DFF-FTP), and a 375

scholarship from Novo Nordisk Foundation to AHH.

18

References Andresen, T.L., Jensen, S.S., Jørgensen, K., 2005. Advanced strategies in liposomal cancer therapy: Problems and prospects of active and tumor specific drug release. Prog. Lipid Res. 44, 380

68-97. Armarego, W.L.F., Chai, C., 2013. Purification of Laboratory Chemicals. Elsevier, Waltham, MA, USA. Arouri, A., Dathe, M., Blume, A., 2009. Peptide induced demixing in PG/PE lipid mixtures: A mechanism for the specificity of antimicrobial peptides towards bacterial membranes? Biochim.

385

Biophys. Acta 1788, 650-659. Arouri, A., Hansen, A.H., Rasmussen, T.E., Mouritsen, O.G., 2013. Lipases, liposomes and lipid-prodrugs. Curr. Opin. Colloid Interface Sci. 18, 419-431. Arouri, A., Trojnar, J., Schmidt, S., Hansen, A.H., Mollenhauer, J., Mouritsen, O.G., 2015. Development of a cell-based bioassay for phospholipase A2-triggered liposomal drug release.

390

PLoS ONE 10, e0125508. Avnir, Y., Ulmansky, R., Wasserman, V., Even-Chen, S., Broyer, M., Barenholz, Y., Naparstek, Y., 2007. Amphipathic weak acid glucocorticoid prodrugs remote-loaded into sterically stabilized nanoliposomes evaluated in arthritic rats and in a Beagle dog: A novel approach to treating autoimmune arthritis. Arthritis Rheum. 58, 119-129.

395

Barenholz, Y., 2001. Liposome application: problems and prospects. Curr. Opin. Colloid Interface Sci. 6, 66-77. Bartlett, G.R., 1959. Phosphorus assay in column chromatography. J. Biol. Chem. 234, 466-468. Čeh, B., Lasic, D.D., 1995. A rigorous theory of remote loading of drugs into liposomes. Langmuir 11, 3356-3368.

19

400

Čeh, B., Lasic, D.D., 1997. A rigorous theory of remote loading of drugs into liposomes: Transmembrane potential and induced pH-gradient loading and leakage of liposomes. J. Colloid Interface Sci. 185, 9-18. Čeh, B., Winterhalter, M., Frederik, P.M., Vallner, J.J., Lasic, D.D., 1997. Stealth® liposomes: from theory to product. Adv. Drug Del. Rev. 24, 165-177.

405

Cern, A., Golbraikh, A., Sedykh, A., Tropsha, A., Barenholz, Y., Goldblum, A., 2012. Quantitative structure-property relationship modeling of remote liposome loading of drugs. J. Controlled Release 160, 147-157. Chen, J., Lin, A., Chen, Z., Wang, W., Zhang, T., Cai, H., Cai, B., 2010. Ammonium sulfate gradient loading of brucine into liposomes: effect of phospholipid composition on entrapment

410

efficiency and physicochemical properties in vitro. Drug Dev. Ind. Pharm. 36, 245-253. Clerc, S., Barenholz, Y., 1995. Loading of amphipathic weak acids into liposomes in response to transmembrane calcium acetate gradients. Biochim. Biophys. Acta 1240, 257-265. Cullis, P.R., Bally, M.B., Madden, T.D., Mayer, L.D., Hope, M.J., 1991. pH gradients and membrane transport in liposomal systems. Trends Biotechnol. 9, 268-272.

415

Garidel, P., Johann, C., Mennicke, L., Blume, A., 1997. The mixing behavior of pseudobinary phosphatidylcholine phosphatidylglycerol mixtures as a function of pH and chain length. Eur. Biophys. J. 26, 447-459. Gregoriadis, G., Swain, C.P., Wills, E.J., Tavill, A.S., 1974. Drug-carrier potential of liposomes in cancer chemotherapy. The Lancet 202, 1313-1316.

420

Gubernator, J., 2011. Active methods of drug loading into liposomes: recent strategies for stable drug entrapment and increased in vivo activity. Expert Opin. Drug Delivery 8, 565-580.

20

Hansen, A.H., Mouritsen, O.G., Arouri, A., 2015. Enzymatic action of phospholipase A2 on liposomal drug delivery systems. Int. J. Pharm. 491, 49-57. Hiyama, M., Akiyama, H., Yamada, K., Koga, N., 2013. Theoretical study of firefly luciferin 425

pKa values-relative absorption intensity in aqueous solutions. Photochem. Photobiol. 89, 571578. Hwang, S.H., Maitani, Y., Qi, X.R., Takayama, K., Nagai, T., 1999. Remote loading of diclofenac, insulin and fluorescein isothiocyanate labeled insulin into liposomes by pH and acetate gradient methods. Int. J. Pharm. 179, 85-95.

430

Ignowski, J.M., Schaffer, D.V., 2004. Kinetic analysis and modeling of firefly luciferase as a quantitative reporter gene in live mammalian cells. Biotechnol. Bioeng. 86, 827-834. Joguparthi, V., Anderson, B.D., 2008. Liposomal delivery of hydrophobic weak acids: Enhancement of drug retention using a high intraliposomal pH. J. Pharm. Sci. 97, 433-454. Kerns, E.H., Di, L., 2008. Drug-like Properties: Concepts, Structure Design and Methods: From

435

ADME to Toxicity Optimization, 1st ed. Elsevier. Kheirolomoom, A., Kruse, D.E., Qin, S., Watson, K.E., Lai, C., Young, L.J.T., Cardiff, R.D., Ferrara, K.W., 2010. Enhanced in vivo bioluminescence imaging using liposomal luciferin delivery system. J. Controlled Release 141, 128-136. Lasic, D.D., Frederik, P.M., Stuart, M.C.A., Barenholz, Y., McIntosh, T.J., 1992. Gelation of

440

liposome interior. A novel method for drug encapsulation. FEBS Lett. 312, 255-258. Li, X., Hirsh, D., Cabral-Lilly, D., Zirkel, A., Gruner, S., Janoff, A., Perkins, W., 1998. Doxorubicin physical state in solution and inside liposomes loaded via a pH gradient. Biochim. Biophys. Acta 1415, 23-40. Lide, D.R., 2009. CRC handbook of chemistry and physics. CRC Press/Taylor and Francis.

21

445

Madden, T., Harrigan, P., Tai, L., Bally, M., Mayer, L., Redelmeier, T., Loughrey, H., Tilcock, C., Reinish, L., Cullis, P., 1990. The accumulation of drugs within large unilamellar vesicles exhibiting a proton gradient: a survey. Chem. Phys. Lipids 53, 37-46. Marques, S.M., Esteves da Silva, J.C.G., 2009. Firefly bioluminescence: A mechanistic approach of luciferase catalyzed reactions. IUBMB Life 61, 6-17.

450

Mayer, L.D., Bally, M.B., Cullis, P.R., 1986. Uptake of adriamycin into large unilamellar vesicles in response to a pH gradient. Biochim. Biophys. Acta 857, 123-126. Mayer, L.D., Bally, M.B., Hope, M.J., Cullis, P.R., 1985. Uptake of antineoplastic agents into large unilamellar vesicles in response to a membrane potential. Biochim. Biophys. Acta 816, 294 -302.

455

Nichols, J., Deamer, D., 1976. Catecholamine uptake and concentration by liposomes maintaining pH gradients. Biochim. Biophys. Acta 455, 269-271. Park, K., 2014. Targeted vs. non-targeted delivery systems: Reduced toxicity over efficacy. J. Controlled Release 178, 126. Scheinberg, D.A., Villa, C.H., Escorcia, F.E., McDevitt, M.R., 2010. Conscripts of the infinite

460

armada: systemic cancer therapy using nanomaterials. Nat. Rev. Clin. Oncol. 7, 266-276. Sur, S., Fries, A.C., Kinzler, K.W., Zhou, S., Vogelstein, B., 2014. Remote loading of preencapsulated drugs into stealth liposomes. Proc. Natl. Acad. Sci. U. S. A. 111, 2283-2288. Tamm, L.K., Kalb, E., 1993. Microspectrofluorometry on supported planar membranes, in: Schulmann, S.G. (Ed.), Molecular Luminescence Spectroscopy, part 3. John Wiley and Sons,

465

New York, pp. 253-305. Walter, A., Gutknecht, J., 1984. Monocarboxylic acid permeation through lipid bilayer membranes. J. Membr. Biol. 77, 255-264.

22

Zucker, D., Marcus, D., Barenholz, Y., Goldblum, A., 2009. Liposome drugs' loading efficiency: A working model based on loading conditions and drug's physicochemical properties. J. 470

Controlled Release 139, 73-80.

23

Table 1. Parameters used for solving Equations 1 to 5. Parameter

Value

Value

−

a

2.8

[μL]

−

a

3.8

[μL]

−

a

7.7

6.0

−

a

8.7

6.0

=

[μL]d

150

4.8

[M]

1.9x10-2

[cm/s]

1.0x10-8

[μmol]

2.4

[cm/s]c

6.6x10-3

b

−

a,b

10.1

14

−

475

Parameter

[μm2]

+

−

4.5x1011

Acid dissociation constants for luciferin and acetic acid were obtained from (Hiyama et al.,

2013). cPermeability coefficient for acetic acid was obtained from (Walter and Gutknecht, 1984). d

The volume

volume

corresponds to the volume of a solution with liposomes that is added to a

of a solution with luciferin at concentration

at the beginning of the experiment.

24

Figure Captions 480

Fig. 1. Schematic illustration of the remote loading of luciferin (L) into liposomes using a carboxylate/carboxylic acid gradient. (A) Initially, the loading agent (e.g., acetic acid) is encapsulated inside the liposomes. The untrapped loading agent in the bulk is exchanged for potassium sulfate by size-exclusion column chromatography and L is added to the bulk solution. At this stage, the uncharged L (i.e., LH2) diffuses down its own concentration gradient into the

485

liposomes. (B) The stage at which the L concentration is equal inside and outside the liposomes. (C) L is being transported against its own concentration gradient driven by the efflux of the carboxylic acid loading agent (e.g., acetic acid) down its own concentration gradient.

Fig. 2. Effect of the loading agent (A) and metal counterion (B) on the loading efficiency (LE%) 490

of luciferin (L) into 100 nm DPPC/DPPG 7:3 liposomes at 25°C. The error bars represent the standard deviation of three different preparations (n=3). The sodium acetate column appears twice for clarity. The partitioning coefficients (K) were adopted from (Walter and Gutknecht, 1984).

495

Fig. 3. (A) N-weighted average diameter of DPPC/DPPG 7:3 liposomes after the different preparation steps of luciferin-loaded liposomes. Buffer conditions: 120 mM magnesium acetate (pH 6.0) (Mg-Ac.) and 120 mM potassium sulfate (K2SO4) (pH 6.0). (B) N-weighted average diameter of luciferin-loaded DPPC/DPPG 7:3 liposomes after 1, 4 and 55 days of their preparation. The error bars represent the standard deviation.

500 Fig. 4. DSC equilibrated thermograms (second heating scan) of DPPC/DMPG 7:3 liposomes subjected to the different buffers during the preparation of luciferin-loaded liposomes.

25

Fig. 5. Dependence of the luciferin loading efficiency (LE%) and loading capacity (LC, µg L/mg 505

lipid) on the amount of luciferin initially added to the system (

) in µmol. The error bars

represent the standard deviation.

Fig. 6. The experimental remote-loading event is depicted (blue squares) along with the model fit (white/black circles), showing the dependence of the amount of luciferin remote loaded ( 510

the amount of luciferin initially added to the system (

) on

). The model values were determined

using Mathematica as described in the modeling section. The last experimental point (

= 3.1

µmol) was not considered for the fit. The only parameter fitted is the permeability for neutral L in our system, which was estimated to be 1×10-8 cm/s. The effect of varying the termination time (

) is also shown. The average termination time for our experiments was three days.

515

Fig. 1

26

520

Fig. 2

27

Fig. 3

525

Fig. 4

28

Fig. 5 530

Fig. 6

29