Correlation between in vitro stability and pharmacokinetics of poly ( ε -caprolactone)-based micelles loaded with a photosensitizer

Polymeric micelles are extensively investigated as drug delivery systems for hydrophobic drugs including pho-tosensitizers (PSs). In order to benefit from micelles as targeted delivery systems for PS, rather than only sol-ubilizers, the stability and cargo retention of the (PS-loaded) micelles should be properly assessed in biologically relevant media to get insight into the essential parameters predicting their in vivo performance ( i.e., pharmacokinetics). In the present study, asymmetric flow field-flow fractionation (AF4) was used to investigate the in vitro stability in human plasma of empty and meta-tetra(hydroxyphenyl)chlorin (mTHPC)-loaded dithiolane-crosslinked micelles based on poly( ε -caprolactone)- co -poly(1,2-dithiolane ‑ carbonate)- b -poly(ethylene gly co l) (p(CL-co -DTC)-PEG) and non (covalently


Introduction
Photosensitizers (PSs) are compounds that are activated by absorbed light to the excited state, following by release of its energy to dissolved oxygen to yield singlet oxygen species, which in turn cause cell death [1,2].This process is exploited in so-called photodynamic therapy (PDT), a modality that has been clinically approved for treatment of different types of cancer [3,4].However, some pharmaceutical properties of PSs need optimization to fully exploit their anti-cancer efficacy by PDT.For instance, m-tetra(hydroxyphenyl)chlorin (mTHPC) is a clinically approved second generation PS [5][6][7].However, its very hydrophobic character (log P of ~9 [8]) encounters obstacles similar to that of many other PSs and (extremely) hydrophobic chemotherapeutic drugs, such as extremely low-aqueous solubility, aggregation in aqueous media and limited tumor specificity [4,9,10].
Polymeric micelles have the capacity to accommodate hydrophobic PS to yield nanomedicines facilitating its administration and increase accumulation at target tissues via passive targeting (i.e.enhanced permeability and retention (EPR) effect) and/or direct targeting towards e.g.specific receptors or epitopes preferentially (over)-expressed on target cells [4,[11][12][13][14][15]. Indeed, as demonstrated in our previous publications [16,17], mTHPC can be solubilized in poly(ε-caprolactone)-bpoly(ethylene glycol) (pCL-PEG) based polymeric micelles.In order to exploit the potential of polymeric micelles as targeted delivery systems for PS, these drug carrier systems must have sufficient stability and simultaneously good cargo retention in the circulation [18][19][20][21].However, rapid release of their cargo in the circulation after intravenous (i.v.) administration of polymeric micelles loaded with mTHPC or other drugs was observed in previous studies [16,22,23].This can be likely ascribed to interactions of the micelle-forming block copolymer unimers and/or loaded PS with blood constituents among which plasma proteins [24][25][26][27][28][29].Therefore, advanced in vitro methods are required to get mechanistic insights into the stability of drug-loaded nanocarriers in biologically relevant media simultaneously avoiding the use of in vivo animal models.Indeed, there are currently some research methods available to explore the stability of drug-loaded polymeric micelles in well-defined and simple media, for example, dialysis [30,31] and particle size analysis [32,33].However, these methods cannot give a clear indication of the stability of nanomedicines in complex biological media such as blood.An attractive technology, and hardly exploited for this purpose, is asymmetrical flow field-flow fractionation (AF4), which allows high resolution separation of particles and plasma proteins based on size differences under physiologically relevant conditions [34].
In the present study, the in vitro and in vivo stability of mTHPC-loaded dithiolane-crosslinked micelles based on poly(ε-caprolactone)-co-poly (1,2-dithiolane-carbonate)-b-poly(ethylene glycol) (p(CL-co-DTC)-PEG) was investigated.In our previous study, it was shown that p(CL-co-DTC)-PEG micelles crosslink spontaneously by disulfide-exchange between dithiolanes in the core of the micelles [35].Here, the in vitro release of mTHPC from these covalently crosslinked p(CL-co-DTC)-PEG micelles and the in vitro stability of these micelles as well as of non (covalently)crosslinked pCL-PEG based micelles in human plasma was evaluated by fluorescence spectroscopy and AF4.To validate the in vitro findings, the in vivo pharmacokinetics including circulation kinetics and tumor accumulation of the crosslinked micelles as well as the loaded mTHPC were studied in A431 tumor-bearing mice and compared with that of free mTHPC and non (covalently)-crosslinked pCL-PEG based micelles loaded with the same PS.

Synthesis and characterization of Cy7-labeled p(CL-co-DTC)-PEG
Cy7-labeled p(CL 18 -DTC 7.5 )-PEG block copolymer was synthesized as previously reported [17] (Scheme S2, supporting information).The successful coupling of Cy7 to the polymer and complete removal of unreacted Cy7 was demonstrated by GPC analysis, with which the amount of Cy7 in the copolymer was quantified using UV/Vis detection at 755.5 nm, showing 17% coupling efficiency as reported previously [17].On average, one polymer chain carried ~0.2 Cy7 molecule.

Preparation and characterization of empty and mTHPC-loaded micelles
Non (covalently)-crosslinked micelles consisting of 10 mg/mL pCL-PEG without and with 0.5 and 5 wt% mTHPC loadings were prepared using a film hydration method, as described previously [16,17].Covalently crosslinked micelles based on different dithiolane containing p (CL-co-DTC)-PEG block copolymers (4 mg/mL polymer) were prepared by a nanoprecipitation method, as previously reported [17,22].Briefly, for empty micelles, 4 mg copolymer was dissolved in 100 μL of DMF (40 mg/mL), while for mTHPC loaded crosslinked p(CL-co-DTC)-PEG micelles (different loadings), a certain volume of mTHPC solution in DMF (5 mg/mL, volume depending on the aimed wt% loading), was added to the weighted polymer, followed by addition of a certain volume of DMF to obtain a final polymer concentration of 40 mg/mL.Subsequently, the resulting copolymer/mTHPC solution was added dropwise to PBS, at 1/ 9 volume ratio.Homogenous micellar dispersions were formed after gentle shaking and subsequently dialyzing the obtained dispersion using a dialysis tubing (MWCO =1 kDa) against PBS at room temperature for 12 h.
The Z-average hydrodynamic diameter (Z ave ) and polydispersity index (PDI) of the formed micelles were determined by dynamic light scattering (DLS) using a ZetaSizer Nano S (Malvern).The loading capacity (LC) and loading efficiency (LE) of mTHPC were determined by UV-Vis analysis and calculated using following equations as previously reported [16].The in vitro release of mTHPC from (de)crosslinked p(CL-co-DTC)-PEG micelles (10 wt% mTHPC loading, prepared in PBS as described in section 2.3) was studied in human plasma (from Seralab, UK) at 37 • C by monitoring the change of fluorescence intensity of mTHPC, as previously reported [16].Free mTHPC (i.e., mTHPC in a solution of ethanol/ propylene glycol (40/60, w/w)) was used as reference.In short, 960 μL of mTHPC-loaded crosslinked micelles in PBS (4 mg/mL polymer) were pre-incubated with 80 μL of 20 mg/mL DTT solution in PBS (final DTT concentration was 10 mM) to reduce disulfide bonds in the core of the micelles, or with 80 μL PBS at 37 • C for 12 h as a control.Subsequently, the pretreated micelles as well as free mTHPC were added to human plasma or PBS (as control) at a volume ratio of 1/9.During incubation with plasma at 37 • C, samples were taken at different time points (5 min, and 0.5, 1, 1.5, 2, 3, 5, 8 h) and placed in a 384-well plate to record the fluorescence intensity using a Jasco FP8300 spectrofluorometer (Japan) at 655 nm after excitation at 420 nm.

Stability of empty and mTHPC-loaded micelles as studied by AF4
Asymmetric flow field-flow fractionation (AF4) was used to investigate the in vitro stability of empty micelles in human plasma as well as the release of mTHPC from micelles in a solution of human serum albumin (HSA) or human plasma (both from Sigma Aldrich, Darmstadt, Germany).For this purpose, covalently crosslinked p(CL-co-DTC)-PEG micelles with random monomer order in the hydrophobic block (4 mg/mL polymer) and non (covalently)-crosslinked pCL-PEG micelles (10 mg/mL polymer) without and with mTHPC loadings (5 and 0.5 wt%) were prepared in PBS as described in section 2.3.In detail, empty micelles were mixed with human plasma or PBS at a volume ratio of 7/3.To study release, mTHPCloaded micelles were incubated either with HSA solution in PBS (final HSA concentration was 45 mg/mL) or human plasma at a volume ratio of 7/3, while free mTHPC (i.e., mTHPC in a solution of ethanol/propylene glycol (40/60, w/w)) at mTHPC concentrations of 0.5 and 0.05 mg/mL (corresponding to the amounts that were present in 5 and 0.5 wt% mTHPC loadings in pCL-PEG micelles) was employed as reference.During incubation at 37 • C, samples were taken at predetermined time points ranging from 0 to 24 h and analyzed by AF4.
AF4 measurements were performed using an AF2000 separation system (Postnova Analytics, Landsberg, Germany) equipped with PN1130 isocratic pumps, degasser and different detectors, namely a refractive index (RI, PN3150) detector, a fluorescence (PN3412) detector and a DLS (Zetasizer Nano ZS, Malvern Instruments, Herrenberg, Germany) detector, respectively.The separation channel consisted of a spacer with 350 μm thickness, deltoid shaped channel profile and 27 cm channel length, and a regenerated cellulose membrane with a cut-off of 10 kDa from Postnova Analytics (Landsberg, Germany).PBS was used as a mobile phase.
A 50 μL of sample was injected into the system during the focusing step with an injection flow rate of 0.2 mL/min, and a focus flow rate of 3 mL/min over 4 min.Next, in the fractionation step, after a transition time of 1 min, the crossflow was kept constant at 2.7 mL/min for 7 min and then decreased exponentially in 20 min to 0.1 mL/min.The crossflow was kept constant at 0.1 mL/min for 15 min and at 0.00 mL/min for another 4 min to ensure complete sample elution.Samples were eluted with a flow rate of 0.5 mL/min.Detailed flow rates in each step and mechanism of AF4 are presented in Table S3, Section S2 and Scheme S3 (supporting information), respectively.The detection of the eluted and fractionated protein components and (mTHPC loaded) micelles was performed sequentially by RI, fluorescence at λ em 650 nm with λ ex 420 nm (for mTHPC) and DLS.The AF2000 control unit software was used for data acquisition and processing.

In vivo studies of free mTHPC and Cy7 labeled micelles loaded with mTHPC in A431 tumor-bearing mice
For the in vivo studies, mTHPC (0.6 wt% loading) in dithiolanecrosslinked p(CL 18 -DTC 7.5 )-PEG micelles with random monomer order in the hydrophobic block and non (covalently)-crosslinked pCL 23 -PEG micelles were prepared in PBS as described in section 2.3, except that the micelles were labeled by mixing the corresponding non-labeled polymers with Cy7-labeled p(CL 18 -DTC 7.5 )-PEG (synthesis is described in section 2.2) (at a ratio of 98.5 to 1.5% w/w) before preparation of the micelles.mTHPC injection solution was prepared by 1:1 dilution of a 120 μg/mL mTHPC stock solution in solvent (i.e., ethanol/propylene glycol, 40/60 w/w) with PBS (final mTHPC concentration was 60 μg/ mL, equal to the concentration of injected micellar samples with 0.6 wt % mTHPC loading).
The animal experiments were approved by the local Utrecht animal ethics welfare committee and national regulatory authorities.Female Balb/c nude mice, weighing 20-28 g were purchased from Envigo (Horst, the Netherlands).Mice were housed in ventilated cages at 25 • C and 55% humidity under natural light/dark conditions.Food and water were provided ad libitum during the entire study.Mice were inoculated with 1 × 10 6 A431 cells suspended in 100 μL PBS subcutaneously into the right flank.When the tumors reached a size of 100-300 mm 3 , mice were included in the studies.Tumors were measured using a digital caliper.The tumor volume V (in mm 3 ) was calculated using the eq.V = (π/6)LS 2 where L is the largest and S is the smallest superficial diameter [19].
Blood samples were collected in tubes with EDTA-anticoagulant via submandibular puncture (~60 μL) at 1 min (100% injection control), and at 1 and 2 h, and via cardiac puncture (~200 μL) after 4 and 24 h, post injection.For the latter, mice were killed through cervical dislocation while under deep isoflurane anesthesia.The collected blood samples were centrifuged at 1000 ×g for 15 min at 4 • C. The plasma supernatant was collected, extracted using acetonitrile/DMSO (4/1, v/ v) and analyzed by high-performance liquid chromatography (HPLC) and a LI-COR Odyssey imaging system to quantify the amount of mTHPC and Cy7 labeled polymer, respectively, as described in previous papers [17,22].Pharmacokinetic parameters were determined by noncompartmental analysis with the PKSolver add-in for Microsoft Excel [36].

Biodistribution in tumor-bearing mice
Mice were sacrificed 4 and 24 h (3-6 animals per group) after i.v.administration of the formulations including free mTHPC and mTHPCloaded Cy7-labeled covalently crosslinked p(CL 18 -DTC 7.5 )-PEG and non (covalently)-crosslinked pCL 23 -PEG micelles.Tumors were excised and then stored at − 80 • C until further processing for quantification.Tumors from three untreated animals were used as controls.
To quantify the content of mTHPC and Cy7 labeled micelles in the tumors, the excised tumor samples were treated [22].In short, to weighted tumor samples, an equal volume of RIPA lysis buffer (v/w) was added.The mixture was homogenized at a speed of 6000/s for 60 s and the homogenate was subsequently aliquoted.
To determine the mTHPC concentration in the samples, an aliquot of the homogenate was vortex-mixed with acetonitrile/DMSO (4/1 v/v) at a volume ratio of 1 to 4 for 1 min.After centrifugation at 15,000 ×g for Y. Liu et al. 10 min, 50 μL of the obtained supernatant was injected into the HPLC system consisting of a Waters X Select CSH C18 3.5 μm 4.6 × 150 mm column coupled with a fluorescence detector set at λ ex 420 nm, λ em 650 nm to analyze the mTHPC concentrations in the different samples [22].To determine the concentration of Cy7 labeled micelles, another aliquot of the homogenate was vortex-mixed with RIPA lysis buffer (1/2, v/v) for 1 min.The fluorescence of Cy7 in the mixture (20 μL) was detected at the 800 nm channel (i.e.λ ex 785 nm and λ em 820 nm), using a LI-COR Odyssey scanner imaging system [22].

Statistical analysis
Statistical analysis was done by GraphPad Prism 8.3.0 software.Statistical significance of biodistribution among different mTHPC formulations was determined by two-way analysis of variance (ANOVA).A value of p < 0.05 was considered significant.Statistical significance is depicted as * p ˂ 0.05, ** p ˂ 0.01, *** p ˂ 0.001.

Preparation and characterization of polymeric micelles
Spontaneously crosslinked micelles based on dithiolane (p(CL-co-DTC)-PEG) at a polymer concentration of 4 mg/mL were prepared by a previously reported nanoprecipitation method [17,22].Non (covalently)-crosslinked micelles consisting of 10 mg/mL pCL-PEG were prepared by a film hydration method [16,17].Different block copolymers were chosen for the preparation of crosslinked micelles, which differed in the sequence of CL and DTC (i.e., blocky or random) and/or molecular weight of the hydrophobic block (Table S1, supporting information).As reported previously and summarized in Fig. S4 (supporting information) [35], the covalently crosslinked p(CL-co-DTC)-PEG micelles had hydrodynamic diameters that slightly increased from 17 to 22 nm with increasing chain length of the hydrophobic block from ~12 to ~25 monomeric units, independent of the monomer sequence (i.e., blocky or random distribution of CL and DTC in the polymer chain).In contrast, for non (covalently)-crosslinked pCL-PEG micelles, the increase in sizes was more significant (from 17 to 43 nm), with increasing CL chain length from 9 to 23 units (Fig. S4, supporting information) [17] which can probably be explained by the more condensed p(CL-co-DTC)-PEG micellar core caused by crosslinking compared to the non (covalently)-crosslinked pCL-PEG micelles.At least, the sizes measured by DLS of non (covalently)-crosslinked pCL-PEG micelles were well in agreement with that obtained by TEM measurements [37].The different micelles were able to efficiently load mTHPC (~80%) in the micellar core at mTHPC feed concentration of ~0.5 mg/mL, which is in line with our previously published data [16,17].

In vitro stability studies 3.2.1. In vitro release of mTHPC from micelles in human plasma as studied by fluorescence spectroscopy
The in vitro release of mTHPC from the crosslinked p(CL-co-DTC)-PEG micelles was investigated in human plasma over time at 37 • C (Fig. 1), by making use of the quenched state when mTHPC is aggregating in the core of the micelles and fluorescence dequenching that occurs after its release from the micelles [16].To explain, when mTHPC releases from the micelles, fluorescence intensity increases resulting from a combination of fluorescent mTHPC when it is released and simultaneous less fluorescence quenching of retained PS in the core of the micelles.Similar as observed in our previous study [16], when free mTHPC (dissolved in a mixture of ethanol and propylene glycol) was added to PBS, severe precipitation occurred and hardly any fluorescence was detected (Fig. S5A, black line, supporting information).Upon 10× dilution of micelles in PBS, the fluorescence of mTHPC when loaded in the micelles was quite low and constant over 8 h at 37 • C due to the afore-mentioned fluorescence quenching (Fig. S5A, supporting information).However, upon 10× dilution in plasma, free mTHPC showed fluorescence intensity of ~2100 a.u.(Fig. 1, black lines), indicating the capability of plasma proteins to solubilize mTHPC.For mTHPC loaded in crosslinked micelles consisting of p(CL-co-DTC)-PEG with relatively short chain lengths of the hydrophobic blocks (9 units of CL and 4 to 7 units of DTC) (Fig. 1A, red, green and cyan lines), substantial increase of fluorescence was observed during the first 1 h incubation with plasma.This figure also shows that the fluorescence intensities levelled off at ~2100 a.u., identical to fluorescence intensity observed from free mTHPC (black line), suggesting complete and quite rapid release of mTHPC from these micelles in plasma.This was observed regardless of the monomer sequence and the DTC content in the polymer chains, and was consistent with our previous studies on non (covalently)-crosslinked pCL 9 -PEG and pCL 17.6 -PEG based micelles having similar short hydrophobic pCL blocks [16,22].However, for the crosslinked p(CL 18 -DTC 7.5 )-PEG micelles containing longer hydrophobic block (Fig. 1A, pink line, ~25 units of CL and DTC in total), quantitative release of mTHPC from the micelles took longer (~4 h), suggesting better mTHPC retention in these micelles.In our previous study, we also demonstrated that increasing the pCL hydrophobic block from 9 to 23 units resulted in better retention of mTHPC (0.5 and 8 h, respectively) [17].Therefore, as compared to micelles with shorter hydrophobic blocks, the slower release of mTHPC from p(CL 18 -DTC 7.5 )-PEG micelles can be explained by the chemical crosslinking and/or the increased hydrophobic interactions, both responsible for stabilization of micelles [38,39].Interestingly, despite more or less similar chain lengths of the core-forming hydrophobic blocks, the slower release of mTHPC from the non (covalently)-crosslinked pCL 23 -PEG micelles as compared to the covalently crosslinked ones (8 vs 4 h) might be ascribed to the physical state of the  S2 in supporting information).In other words, the semi-crystalline state with crystallites acting as physical crosslinks in the core-forming hydrophobic pCL block in the pCL 23 -PEG micelles may confer greater PS retention by decreasing its rate of diffusion [39][40][41] as compared to the liquid state of the core of p(CL 18 -DTC 7.5 )-PEG micelles at 37 • C.
After reduction by DTT of the disulfide bonds present in the core of the p(CL-co-DTC)-PEG based micelles (i.e., decrosslinking) and subsequent incubation in plasma, faster increase in the fluorescence intensity of mTHPC was observed (Fig. 1B) compared to the micelles before decrosslinking (Fig. 1A), i.e., the plateau fluorescence reached within 1 h.This observation demonstrates that a faster release of mTHPC from micelles can be triggered by reversible reduction of disulfide bonds present in the micellar core resulting in decrosslinking of the micelles.

Stability of empty micelles in human plasma. AF4 as a fraction-
ation technique provides opportunities to separate particles and proteins in complex mixtures based on their hydrodynamic sizes [34].Therefore, this technique was exploited to assess whether micellar disassembly resulting from protein-micelle interactions plays a role in the mTHPC release profiles shown in section 3.2.1.To this end, empty covalently crosslinked p(CL 9 -DTC 3.9 )-PEG and p(CL 18 -DTC 7.5 )-PEG micelles and non (covalently)-crosslinked pCL-PEG micelles that showed different retention characteristics of mTHPC in the presence of plasma (as discussed in section 3.2.1)were incubated with plasma at 37 • C and subsequently fractionated and analyzed using AF4.
Fractograms obtained by refractive index (RI) detection of the empty micelles in PBS at 37 • C did not change upon 24 h incubation, regardless of being covalently crosslinked or not and of their initial sizes (Fig. S6, supporting information), demonstrating excellent stability of the Fig. 2. AF4 fractograms of micelles using refractive index (RI) detection.The micelles were incubated for 24 h with human plasma (HP) at 37 • C: non (covalently)crosslinked micelles composed of pCL 9 -PEG (A), pCL 15 -PEG (B) and pCL 23 -PEG (C) and dithiolane crosslinked micelles consisting of p(CL 9 -DTC 3.9 )-PEG (D) and p (CL 18 -DTC 7.5 )-PEG (E) were incubated with plasma at a volume ratio of 7/3 for 24 h.After being mixed with plasma, the final polymer concentrations of non (covalently)-crosslinked pCL-PEG and crosslinked p(CL-co-DTC)-PEG based micelles were 7.0 and 2.8 mg/mL, respectively.A sample of human plasma (donated as 30% HP in (A)) was used to assign the peaks of plasma components.The signals were normalized to the highest signal.micelles in PBS.In line with this, the size distribution of these micelles as measured by in-line DLS did not change over 24 h in PBS (Fig. S7, supporting information).When the different micelles were incubated with plasma, two well-separated main peaks were observed (Fig. 2), namely one at retention time of ~10 min, representing human serum albumin (as verified by overlaying the fractograms with that of human plasma, e.g., Fig. 2A, blue vs black line) and another at retention times ranging from approximately 20 to 30 min, corresponding to micelles (as evidenced by the fractograms of micelles in PBS, Fig. S6, supporting information).In-line DLS measurements of the micelles in plasma (Fig. S8, supporting information) showed an increase in particle size with increasing retention time, as expected, which did not significantly change over the course of the 24 h incubation time.However, it is clear (Fig. 2A and B) that the peak corresponding to both non-(covalently)crosslinked pCL 9 -PEG and pCL 15 -PEG micelles with relatively small sizes (<25 nm) shifted to higher retention times and thus increased in size with increasing incubation time, suggesting a certain degree of instability of these micelles.Probably, the observed release of mTHPC from the pCL 9 -PEG and pCL 17.6 -PEG micelles in plasma as described in our previous publications [16,22] is related to the instability of these micelles as shown by the AF4 data.Interestingly, shifts of the micellar peaks were not observed for the covalently crosslinked p(CL 9 -DTC 3.9 )-PEG and p(CL 18 -DTC 7.5 )-PEG micelles with similar relatively small sizes (<25 nm) (Fig. 2D and E), as well as for the physically crosslinked pCL 23 -PEG micelles (Fig. 2C) with a larger size (43 nm), suggesting that these micelles were stable in this medium.
From above data, we can draw several conclusions.First, it is noted that covalently crosslinked p(CL 9 -DTC 3.9 )-PEG micelles showed superior stability in plasma, as compared to the non (covalently)-crosslinked pCL 9 -PEG micelles of similar size (18 vs 17 nm) and pCL 15 -PEG micelles that have comparable chain length of the hydrophobic segment (~14 monomer units) but slightly differed in size (~18 vs ~ 24 nm).This indicates that indeed covalent-crosslinking improves the stability of small micelles in plasma.Likewise, covalently crosslinked p(CL 18 -DTC 7.5 )-PEG micelles had better stability as compared to non (covalently)-crosslinked pCL 15 -PEG micelles having comparable sizes (22 vs 24 nm).On the other hand, non (covalently)-crosslinked pCL 23 -PEG micelles displayed similar high stability when compared to p(CL 18 -DTC 7.5 )-PEG micelles, which have similar chain lengths (~24 monomer units) of the core-forming block but different sizes (43 vs 22 nm).As discussed in section 3.2.1, this is probably attributed to physical crosslinks resulting from the crystallization of pCL blocks in the core of the  (C, D).After being mixed with plasma, the final polymer concentrations were 7 (A, C) and 4 (B, D) mg/mL, respectively.As a control, free mTHPC at a final concentration of 0.035 mg/mL was added to plasma (black line), corresponding with the concentrations obtained from micelles with 0.5 wt% mTHPC loadings.The signals were normalized to the highest signals.
pCL 23 -PEG micelles (so these micelles are referred to physically crosslinked pCL 23 -PEG micelles from here).In fact, according to the obtained AF4 results for the non (covalently)-crosslinked pCL-PEG micelles, the stability of these micelles improved by increasing the chain length of hydrophobic pCL block from 9 to 23 units (see Figs. 2A-C).

Release of mTHPC from micelles in buffer with human serum albumin and human plasma as studied by AF4.
Besides protein-polymer interactions, favorable interactions of the loaded cargo with plasma proteins can also lead to its premature release from the micelles, even when the nanoparticles are stable in plasma [42][43][44].Previous studies showed high binding affinity of mTHPC for plasma (lipo)proteins leading to mTHPC redistribution from intact liposomes to lipoproteins [26][27][28][29].Indeed, we also observed that even though the empty covalently crosslinked p(CL 9 -DTC 3.9 )-PEG and p(CL 18 -DTC 7.5 )-PEG micelles are stable in plasma for 24 h according to the AF4 data shown above, mTHPC was quantitatively released from these micelles in 1 and 4 h, respectively (see Fig. 1A, red and pink lines).To investigate this further, mTHPC loaded in stable micelles (i.e., covalently crosslinked p(CL 9 -DTC 3.9 )-PEG and p(CL 18 -DTC 7.5 )-PEG micelles or physically crosslinked pCL 23 -PEG micelles) with different mTHPC loadings, were incubated with human serum albumin (HSA) or human plasma for 16 h at 37 • C. RI detection was used to record fractograms of plasma components and micelles, while mTHPC was also detected making use of its intrinsic fluorescence using a fluorescence detector coupled to the AF4.
By comparing the RI fractograms of HSA (Fig. 3A, red line) and plasma (Fig. 3B, green line), it can be concluded that the peak of plasma at approximate 10 min represents the most abundant component, i.e., albumin, whereas the broad peak at around 20 and 30 min corresponds to low-and high-density lipoproteins, respectively.Importantly, it is obvious from Fig. 3B (purple lines) that free mTHPC preferably binds to lipoproteins, particularly low-density lipoproteins which is in agreement with previous studies [29,45], while in the absence of lipoproteins it just slightly binds to albumin (Fig. 3A, blue lines).
For mTHPC-loaded pCL 23 -PEG micelles, fractograms obtained by RI detection show that the peak of the micelles (retention time of ~30 min.)did not change upon incubation regardless of the medium (HSA or plasma) (0.5 wt% loading, Fig. 4A, B and 5 wt% loading, Fig. 5A, B), which is in line with the stability results of the corresponding empty micelles (Fig. 2C).The fractograms obtained by recording the intrinsic fluorescence of mTHPC (Fig. 4C and D) after 7 h incubation with either 30 or 60% plasma show no detectable release of mTHPC from pCL 23 -PEG micelles with 0.5 wt% loading.This suggests good mTHPC retention in pCL 23 -PEG micelles, which is in good agreement with the release profile based on the quenching of mTHPC fluorescence reported in our previous study [17].On the other hand, upon incubation with 30% plasma of pCL 23 -PEG micelles with 5 wt% mTHPC loading, a small fluorescence peak at approximately 15 min in the fractogram (Fig. 5C) was observed after 3 h and did not change till 16 h, suggesting that initially a small fraction (< 5%) of mTHPC was released and subsequently bound to lipoproteins probably as a result of overloading of the core of the micelles with mTHPC.In contrast, the release of mTHPC in HSA was negligible (Fig. 5D), which demonstrates that the release of this PS from pCL 23 -PEG micelles is medium dependent and emphasizes the high affinity of the PS for lipoproteins and not for albumin.The release of mTHPC from covalently crosslinked p(CL-co-DTC)-PEG micelles could not be evaluated by AF4 because released mTHPC that bound to lipoproteins has the same retention time as the micelles (~17 min).
Overall, the results of AF4 analysis indicate that this technique is a powerful analytical tool to evaluate the stability of PS loaded micelles in biological media, e.g., blood plasma.It can separate the micelles and different plasma proteins, which is necessary for analysis of possible interactions between proteins and particles but challenging using conventional separation methods, e.g., size exclusion chromatography or ultracentrifugation.In addition, PS (re)distribution to specific plasma proteins is observed, giving strong indications for the (in)stability of PS loaded micelles in plasma.This information is important to explain or even predict micelles' behavior in vivo and importantly, provides a valuable reference to take effective measures to prevent possible and unwanted premature cargo release in the circulation.Fig. 5. Fractograms of mTHPC-loaded (5 wt%) pCL 23 -PEG micelles incubated at 37 • C for 16 h with plasma (7/3, v/v) (A, C) or 45 g/L HSA (B, D), obtained by refractive index (RI) (A, B) and fluorescence intensity detection (FI) (λ ex 420 nm, λ em 650 nm) (C, D).After being mixed with plasma or HSA (solution in PBS), the final polymer concentrations of the different micelles were 7.0 mg/ mL.As a control, free mTHPC at a final concentration of 0.035 mg/mL was added to plasma and HSA, corresponding to the same concentration of mTHPC in 0.5 %wt loading in the micelles.The signals were normalized to the highest signals (RI or FI).

Circulation kinetics and biodistribution of free mTHPC and Cy7labeled micelles loaded with mTHPC
In our previous study, we showed and discussed the pharmacokinetics of mTHPC-loaded Cy7-labeled physically crosslinked pCL 23 -PEG micelles and i.v.injected free mTHPC as control [17].In the present study, the pharmacokinetic data of Cy7-labeled dithiolane-crosslinked p (CL 18 -DTC 7.5 )-PEG micelles loaded with mTHPC and injected in A431 tumor-bearing mice are reported and discussed, and compared with free mTHPC.Micelles based on p(CL 18 -DTC 7.5 )-PEG were selected because of the high in vitro stability of these empty micelles in plasma (see Fig. 2E) and relatively slow release of mTHPC (see Fig. 1A, pink line).It is noted that no (short term) side effects were observed in mice that received mTHPC-loaded crosslinked p(CL 18 -DTC 7.5 )-PEG micelles during or after their administration, similar to other pCL-PEG based micelles [17,22].
As indicated in Fig. 6A, the plasma concentration curves of mTHPC after administration of either free mTHPC or the micellar formulation (black and blue lines) were comparable, which is similar to the Cy7 labeled micelles (purple line).It is shown that ~45% of the injected dose (ID) of mTHPC and Cy7 was rapidly cleared in the initial α-elimination phase (first 1 h).Non-compartmental analysis of the mTHPC and Cy7 curves was used to determine pharmacokinetic parameters including terminal half-lives (t½-β), area under the curves (AUCs), distribution volumes and clearances (Table 1).Pharmacokinetic parameters calculated from mTHPC analysis (Table 1, top part) were tightly associated with those derived from Cy7 analysis (Table 1, bottom part).It is however not possible to conclude whether these parameters of mTHPC calculated for the micellar formulation reflect (released) free mTHPC or that remaining associated with the micelles because of coincidently similar data observed for free mTHPC and Cy7.However, combined with the in vitro release study (Fig. 1A, pink line), premature release of mTHPC from micelles most likely occurred in the circulation, which was observed in various liposomal mTHPC formulations and other mTHPCloaded micelles [16,17,46].
Interestingly, the circulation kinetics of pCL 23 -PEG micelles and mTHPC loaded in those physically crosslinked micelles as well as the corresponding pharmacokinetic parameters presented in [17], were significantly superior as compared to the covalently crosslinked p(CL 18 -DTC 7.5 )-PEG micelles (Table 1).For instance, we demonstrated that after i.v.injection, the PS loaded in pCL 23 -PEG micelles displayed markedly prolonged blood circulation time (t½-β:14 h), compared to free mTHPC, which is due to the long circulating pCL 23 -PEG micelles (t½-β:18 h) (Table 1).These results are very well in line with the in vitro stability presented in section 3.2 (Figs. 4 and 5) and our previous publication [17], which both showed sufficient mTHPC retention in pCL 23 -PEG micelles during 8 h in human plasma and again indicates that physical crosslinks resulting from the crystallization of pCL blocks in pCL 23 -PEG micelles confer greater PS retention compared to covalently crosslinks in p(CL 18 -DTC 7.5 )-PEG micelles.
Finally, the biodistribution of mTHPC loaded in covalently crosslinked p(CL 18 -DTC 7.5 )-PEG (that showed short circulation time and poor PS retention) and physically crosslinked pCL 23 -PEG micelles (that had  a Initial phase half-lives (t½-α) were estimated from plasma disappearance rates during the first hour after administration while half-lives of the terminal phase (t½β) were calculated from non-compartmental analysis.Distribution volumes and clearances were calculated assuming a standard plasma volume of 58.5 mL/kg [16].b Taken from reference [17].c Calculated from the plasma concentration curves reported in reference [17].
prolonged circulation time and good PS retention) was investigated and compared with free mTHPC in tumor-bearing mice.Tumors were excised from mice that were sacrificed at 4 and 24 h post i.v.injection of the formulations, and tumor accumulation of Cy7labeled micelles was quantified by fluorescence intensity measurements (Fig. 6B, open bars).It was shown that Cy7 levels from covalently crosslinked p(CL 18 -DTC 7.5 )-PEG micelles in tumors were quite low (< 1% ID/g) and remained constant at 24 h (Fig. 6B, open blue bars).Interestingly, tumor accumulation of physically crosslinked pCL 23 -PEG micelles was much higher than that of covalently crosslinked p(CL 18 -DTC 7.5 )-PEG micelles (5% vs < 1% ID/g) at 4 h and importantly, progressively increased to ~8% upon 24 h after administration (Fig. 6B, open red bars), which led to a significant difference with the crosslinked p(CL 18 -DTC 7.5 )-PEG micelles.This result demonstrates that long circulating micelles are indeed favorable for EPR-mediated tumor accumulation [18,20,47].
Fig. 6B (solid bars on the left side) displays the biodistribution in tumors of mTHPC that was injected in free form and as covalently crosslinked p(CL 18 -DTC 7.5 )-PEG or physically crosslinked pCL 23 -PEG formulations.For free mTHPC and mTHPC loaded in covalently crosslinked p(CL 18 -DTC 7.5 )-PEG micelles, the mTHPC accumulation in the tumors was similar for 4 and 24 h (~2-3% ID/g) (Fig. 6B, solid black and blue bars).These mTHPC levels were actually higher than Cy7 levels (Fig. 6B, open blue bars), suggesting that mTHPC was not retained within the micelles, i.e., mTHPC was rapidly released from the micelles, in line with the in vitro release study (Fig. 1A, pink line).However, when loaded in physically crosslinked pCL 23 -PEG micelles, the tumor accumulation of mTHPC increased progressively in time (1.8 at 4 h to 5.1% ID/g at 24 h) (Fig. 6B, solid red bars).Despite relatively lower mTHPC level as compared to Cy7 level at each timepoint, this upward trend was correlated well with the Cy7 accumulation in the tumor, demonstrating the ability of long circulating polymeric micelles to facilitate PS targeting to tumors via the EPR effect [18,20,47].

Conclusion
In the present study, we demonstrated that AF4 is a powerful analytical technique for the separation and characterization of (drug loaded) nanoparticles and its stability in complex biological media e.g., human plasma.PS (re)distribution to lipoproteins was observed, giving strong indications for the (in)stability of the drug-loaded micelles in plasma.With this tool, the (in)stability of drug loaded nanoparticles after i.v.administration could be predicted, which is favorable to screen promising delivery systems with reduced experimental time and costs and without excessive use of animals.

Fig. 1 .
Fig. 1.Fluorescence intensity (λ ex 420 nm, λ em 655 nm) of free mTHPC and mTHPC-loaded (10 wt%) (de)crosslinked p(CL-co-DTC)-PEG micelles as a function of time at 37 • C in human plasma; free mTHPC and micelles were 10× diluted with human plasma to obtain final polymer and mTHPC concentrations of 0.4 mg/mL and 32 μg/mL, respectively.The fluorescence intensities of the corresponding mTHPC-loaded micelles diluted with PBS were used as 0 h timepoint.Decrosslinked micelles were prepared by incubation of their crosslinked counterparts with DTT (10 mM in PBS, to reduce the disulfide bonds) for 12 h at 37 • C.

Fig. 3 .Fig. 4 .
Fig. 3. Fractograms of free mTHPC in PBS with 45 g/L HSA (A) and 100% human plasma (B) obtained by refractive index (RI) (A, red line and B, green line) and fluorescence intensity (FI) (λ ex 420 nm, λ em 650 nm) (A, blue lines and B, purple lines); free mTHPC was added at a final concentration of 0.035 and 0.35 mg/mL, respectively.The signals were normalized to the highest signals.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 1
Pharmacokinetic parameters of mTHPC in different formulations and the corresponding (Cy7 labeled) micelles.