Article pubs.acs.org/molecularpharmaceutics
An In-Depth View into Human Intestinal Fluid Colloids: Intersubject Variability in Relation to Composition Danny Riethorst,† Peter Baatsen,‡ Caroline Remijn,§ Amitava Mitra,∥ Jan Tack,⊥ Joachim Brouwers,† and Patrick Augustijns*,† †
Drug Delivery and Disposition, Department of Pharmaceutical and Pharmacological Sciences, KU Leuven, 3000 Leuven, Belgium Center for the Biology of Disease, KU Leuven, and VIB Bio Imaging Core, VIB-KULeuven, 3000 Leuven, Belgium § Nutrition & Health, Unilever R&D Vlaardingen, 3133 AT Vlaardingen, The Netherlands ∥ Biopharmaceutics, Pharmaceutical Sciences and Clinical Supply, Merck & Co., West Point, Pennsylvania 19486, United States ⊥ Department of Gastroenterology, University Hospitals Leuven, 3000 Leuven, Belgium ‡
ABSTRACT: Intestinal fluids dictate the intraluminal environment, and therefore, they substantially affect the absorption of orally taken drugs. The characterization of human intestinal fluids (HIF) and the design of simulated intestinal fluids (SIF) mainly focus on composition, not necessarily taking into account the ultrastructure of HIF. Colloidal structures in HIF and SIF can enhance the solubilizing capacity for lipophilic drugs while decreasing the bioaccessible fraction. As such, colloids present in HIF play a crucial role and require an in-depth characterization. Therefore, the present study pursued a comprehensive characterization of the ultrastructure of fasted and fed state HIF, focusing on (i) intersubject variability in relation to composition and (ii) differences between the ultrastructure of HIF and SIF. Individual as well as pooled HIF were collected from human volunteers near the ligament of Treitz and compositionally characterized previously. A HIF population pool (20 healthy volunteers) for both fasted (FaHIF) and fed state (FeHIF) was compared to current SIF, as well as selected HIF from different individuals. The selected individual HIF represented the full spectrum of compositional characteristics. Three complementary electron microscopy techniques, cryo-TEM (transmission electron microscopy), negative stain TEM, and cryo-SEM (scanning electron microscopy), were employed to provide a comprehensive view of the colloidal structures in HIF and SIF. The use of complementary EM techniques provided a unique insight into the ultrastructure of HIF, including their native conformation. These characterizations showed that FaHIF and FaSSIF (fasted state simulated intestinal fluids) only consist of (mixed)-micelles with minimal intersubject variability. Ultrastructures in FeSSIF (fed state simulated intestinal fluids) and FeSSIF-v2 are not representative of the colloids in FeHIF since SIF lack (multi)-lamellar vesicles and lipid droplets. Furthermore, the images demonstrated significant intersubject variability in the ultrastructure of FeHIF, which may contribute to variable absorption of lipophilic drugs. KEYWORDS: human intestinal fluids, colloidal structures, cryo-SEM, cryo-TEM, micelles, vesicles
■
INTRODUCTION Intestinal fluids dictate the intraluminal environment, and therefore, they substantially affect the absorption of orally taken drugs. In the past, simple aqueous buffers were used for in vitro absorption profiling, thereby only simulating the pH effect on absorption. Since additional components in human intestinal fluids (HIF), including bile salts, phospholipids, cholesterol, and lipid degradation products, affect drug absorption as well, more biorelevant simulated intestinal fluids (SIF) that mimic HIF composition have been developed and are still evolving.1−5 The characterization of HIF and the design of SIF mainly focus on composition,6−9 not necessarily taking into account the ultrastructure of HIF, i.e., the colloidal structures formed by the interaction between HIF components: (mixed-)micelles, (multi)laminar vesicles, and lipid droplets.10 © 2016 American Chemical Society
Colloidal structures in HIF have already been described in the early nineties.11,12 More recently, these structures have been gaining interest for their influence on the oral bioavailability of lipophilic drugs with limited aqueous solubility, belonging to classes II and IV of the Biopharmaceutics Classification System (BCS). 13 Exemplary is the utilization of lipid based formulations to enhance the intestinal absorption of poorly water-soluble drugs.14 Regardless of the formulation, orally taken drugs come into contact with endogenous colloidal structures, especially in fed state conditions. Indeed, postReceived: Revised: Accepted: Published: 3484
June 3, 2016 July 20, 2016 August 31, 2016 August 31, 2016 DOI: 10.1021/acs.molpharmaceut.6b00496 Mol. Pharmaceutics 2016, 13, 3484−3493
Article
Molecular Pharmaceutics
Figure 1. Complementary EM imaging techniques. Low magnification images of cryo-TEM (a), negative stain TEM (b), and cryo-SEM (c) are shown. Below every technique, schematic legends are presented (d−f) to facilitate identification of different colloidal structures in the figures below.
fasted state simulated intestinal fluid (FaSSIF), containing taurocholate (3 mM) and lecithin (0.75 mM), 2.24 mg of SIF powder was dissolved per milliliter of FaSSIF buffer (pH 6.5), which contained NaOH (10.5 mM), Na2HPO4 (28 mM), and NaCl (106 mM). To prepare fed state simulated fluids FeSSIF(pH6.5) and FeSSIF-v2(pH6.5), 11.2 mg of SIF powder and 9.76 mg of FeSSIF-v2 powder, respectively, were dissolved in FaSSIF buffer (pH 6.5). Collection and Characterization of HIF. Human intestinal fluids were collected and characterized as part of a previously published study.6 Briefly, human intestinal fluids were collected from the duodenum (near the ligament of Treitz) of healthy volunteers (HV) every 10 min for a period of 90 min in both fasted and fed state. For every time point a maximum of 10 mL was collected. After an overnight fast of at least 12 h (no food and only water), volunteers were given 250 mL of water before initiating the sampling of fasted state intestinal fluids. Following fasted state sampling, 400 mL of Ensure Plus was ingested to simulate a standard meal; this condition is referred to as the fed state. Sampling was initiated after drinking 250 mL of water, 20 min after the intake of the liquid meal. For each volunteer, the collected fractions were pooled to obtain fasted and fed state volunteer pools. In addition, a “population” pool of 20 volunteers was created for both fasted human intestinal fluids (FaHIF) and fed state human intestinal fluids (FeHIF) by taking an equal volume from each individual volunteer pool. All samples and pools were characterized with respect to bile salts, phospholipids, cholesterol, and lipid degradation products. For a detailed description of the collection procedure and the methods used for characterization, we refer to Riethorst et al.6 (Cryo)-TEM. Samples were vitrified with an ethane plungefreezer (EMBO, Precision Engineering, Heidelberg, Germany). To this end, 2.5 μL of sample was applied to a glow-discharged carbon coated 300 mesh lacy grid held by the plunger-forceps, incubated for 30 s, blotted for 5−10 s, plunged into liquid
prandial HIF contains similar colloidal structures15 as observed after in vitro digestion of lipid based formulations.16 While colloidal structures in HIF and SIF enhance the solubilizing capacity17−19 for lipophilic compounds, strong entrapment20 may result in a decreased absorptive flux, as demonstrated in vitro,2,19 ex vivo,21 and in situ.22 Thus, endogenous colloidal structures are a determining factor in the solubility− permeability interplay underlying the bioavailability of lipophilic drugs. To improve the predictive capacity of SIF, endogenous colloidal structures in HIF should be further studied and implemented in future versions of SIF. Since the composition of HIF is highly variable with respect to surfactants (e.g., phospholipid and bile salts) and lipid degradation products,6 it can be expected that the formation of micelles, vesicles, and lipid droplets is prone to a similar variability. Therefore, the present study pursues a comprehensive characterization of the ultrastructure of fasted and fed state HIF, focusing on (i) intersubject variability in relation to composition and (ii) differences in ultrastructure between HIF and SIF. The presence of lipid droplets, (multi)-lamellar vesicles, and micelles in HIF has already been visualized with cryogenic transmission electron microscopy (cryo-TEM).4,15,19,23,24 CryoTEM imaging provides the best possible differentiation between solid and membrane colloidal structures in a nearnative state.15 Since cryo-TEM has certain limitations caused by a narrow field of view, however, it is less suitable to compare HIF from different subjects or to correlate HIF ultrastructure with composition (as demonstrated in the current study). In the present study, we have therefore explored negative stainTEM2 and cryo-scanning electron microscopy (SEM) as complementary techniques alongside cryo-TEM.
■
MATERIALS AND METHODS Chemicals. Simulated intestinal fluid (SIF) powders were purchased from Biorelevant.com (Croydon, UK). To prepare 3485
DOI: 10.1021/acs.molpharmaceut.6b00496 Mol. Pharmaceutics 2016, 13, 3484−3493
Article
Molecular Pharmaceutics
Figure 2. Example of heterogeneous cryo-TEM imaging. Images originate from a single sample on a single grid. (a) Large colloids with a lipid droplet surrounded by >300 nm multicompartmental vesicles. (b) Small (mixed)-micelles 200 nm. Nevertheless, a wide range of structures (10 nm to 10 μm) were observed (Figure 6a−c). From the cryo-TEM images, structures above 200 nm appear particularly as (multicompartmental) vesicles. Below 50 nm, only micelles can be seen. Fiber-like structures are regularly observed in FeHIF, including HV 5. These structures were confirmed with both negative stain and cryoTEM imaging and are most likely proteins. HV 6 FeHIF stands out because of the high bile secretions in this sample (Table 1). This results in a high quantity of (mixed-)micelles (10−100 nm). The negative stain image at low magnification (Figure 6d) depicts large dark clusters alongside a lonely lipid droplet in what further seems an empty sample. Larger magnifications (Figure 6e,f) show that these clusters consist of colloids between 50 and 100 nm; additionally, the presumed empty space appears dense in structures ranging from 10 to 100 nm. Further investigation with cryo-TEM (Figure 7d−f) clearly demonstrates that the structures between 10 and 50 nm are micelles. Considering the low lipid content, structures between 50 and 100 nm are considered to be mixed-micelles. HV 16 FeHIF has high lipid content, accompanied by the highest phospholipid and cholesterol concentrations (Table 1). In negative stain TEM, this sample appears like a web of structures that range between 20 and 300 nm, with only a limited amount of large structures (>1−2 μm) (Figure 6g−i). All open spaces were predominantly filled with colloids 400 nm comprise greaseball-like lipid droplets and large vesicles. Large vesicles, in high lipid containing samples, occurred alongside high cholesterol concentrations. HV 3 FeHIF is a peculiar example as it only contains colloids 500 nm) structures (white dashed arrow) in the clusters. The highest magnification shows a clear image of the smaller structures (e.g., micelles): 10 nm (black arrow), 20 nm (dashed black arrow), and up to 40−50 nm (white arrow). The cryo-TEM images of fed state HIF (f−h) show the different types of colloidal structures observed: micelles (solid white arrow), large (mixed)-micelles/small lipid droplets (solid black arrow), vesicles from 50 to 500 nm (black dashed arrow), and a tubular membrane structure (black asterisk).
Interindividual Variability in HIF. The ultrastructures of five individual FaHIF showed little variability, as only micelles were present (data not shown). Again, none of the individual FaHIF samples showed vesicle formation, contradictory to a previous report in literature.25 For FeHIF samples, ten individual volunteer pools (pooled from time-dependent samples per volunteer, as explained in Materials and Methods) were selected for imaging, covering a wide range of compositions. The composition and ultrastructure of all ten volunteer pools are summarized in Table 1. FeHIF obtained 3488
DOI: 10.1021/acs.molpharmaceut.6b00496 Mol. Pharmaceutics 2016, 13, 3484−3493
Article
Molecular Pharmaceutics Table 1. Composition and Colloids Formed in Pooled and Individual HIFa
a
The individual FeHIF were grouped according to their overall ultrastructure. Additionally, both the FaHIF and FeHIF pools were included as they are a reference for average composition. The HIF samples and the corresponding numbering of healthy volunteers (HV) originate from a previous study6.
sizes, enclosing the sample solutions as a very thin vitrified film. Large vesicles may be forced into certain shapes, following the borders of the lacy carbon film. Examples of this are clearly seen in Figure 7h,i. By visualizing full droplets, cryo-SEM images are not influenced by the above. Therefore, it provides complementary native insights into colloid distribution and 3D conformation. To gain insight into the droplet, it has to be split, creating a broken surface as seen in Figure 8a. After water sublimation, salts and sugars create maze-like walls in which smaller colloidal structures are nestled. These wall-like structures are straighter and much less dense in FaHIF (Figure 8b). In all images, colloids appear to be spherical or sometimes elliptical. One exception is the grain-like structures between 200 and 400 nm in Figure 8c, which are most likely salt artifacts and could only be observed in FaHIF. Structures between 20 and 200 nm (micelles and vesicles) were mixed similar to the observations using negative stain TEM. Larger structures stand out of the salt/sugar “walls” and were all spherical as well. Figure 8f clearly shows the solid inside of a broken lipid droplet.
that contains high bile salt secretions, which result in “micellar type” FeHIF. Group 2 (HV 1, 3, 5, 20) consists of FeHIF that predominantly has micelles 2 μm) are indicated by a white asterisk. Images b and e contain fiber-like structures up to several micrometers in length. An example of such a fiber is highlighted by a white frame in image b. 3490
DOI: 10.1021/acs.molpharmaceut.6b00496 Mol. Pharmaceutics 2016, 13, 3484−3493
Article
Molecular Pharmaceutics
Figure 7. Interindividual variability in fed state HIF visualized with cryo-TEM. Four fed state HIF from different healthy volunteers (HV) are depicted with varying magnifications: HV 5 (a−c), HV 6 (d−f), HV 16 (g−i), and HV 3 (j−l). The HIF and numbering of HV originate from a previous study.6 Below each HIF, its relative composition is shown in a log2 scale, compared to the mean from 20 individual HIF (1.0 represents equal to the mean). HIF from HV 5 (a−c): fiber-like structures (solid black arrow), large multicompartmental vesicles (black dashed arrow), and micelles (solid white arrow). HIF from HV 6 (d−f): in image d, low magnification, only micelles are present. Images e and f contain micelles of 20− 50 nm (black arrow) and larger mixed-micelles 50−100 nm. HIF from HV 16 (g−i): in image g, a large number of lipid droplets (solid white arrow) are visible; additionally, laminar (dashed white arrow) and bilaminar (dashed black arrow) vesicles were observed. HIF from HV 3 (j−l): micelles (solid white arrow), small vesicles 50−100 nm (dashed white arrow), and large vesicles 200−800 nm (solid black arrow).
species of colloids as their size ranged from 10 nm to 5 μm. Besides the majority of laminar vesicles, multicompartmental 3491
DOI: 10.1021/acs.molpharmaceut.6b00496 Mol. Pharmaceutics 2016, 13, 3484−3493
Article
Molecular Pharmaceutics
Figure 8. Visualization of colloidal structures in HIF using cryo-SEM. Image a depicts the total surface of the broken droplet that is imaged with cryo-SEM. The white box indicates the area that is fractured correctly for imaging. The FaHIF pool is depicted in b and c. FeHIF examples are shown in d (HV 5), e,f (FeHIF pool), and g (HV 6). In between the salt/sugar aggregated “walls” (solid black arrows), several colloidal structures are indicated: micelles (dashed black arrow), micelles and vesicles between 50 and 500 nm (solid white arrow), and large lipid droplets (white asterisk). Salt grain-like structures (dashed white arrow) were observed in FaHIF.
already shown by Fatouros et al.26 However, while these advanced simulated media contain vesicles, they still do not fully resemble the currently observed ultrastructure of FeHIF. Presumably, intraluminal processes, like gastrointestinal transfer and enzymatic degradation of food components, affect the formation of colloids in the GI tract. As such, implementation of these dynamic processes might be considered when preparing fed state simulated media. In addition, a FeHIF population pool does not capture the substantial variability in colloidal structures observed in HIF from individual subjects.
vesicles could be observed regularly as well. Finally, cryo-SEM showed that the majority of the above colloids are indeed spherical or elliptic in shape. Comparison of simulated and human intestinal fluids demonstrated that the ultrastructure in FaHIF is represented by FaSSIF. In contrast, both FeSSIF and FeSSIF-v2 do not adequately simulate the complex ultrastructure of FeHIF. A likely cause is the lack of lipids and their degradation products in FeSSIF compared to FeHIF, which could be improved in future FeSSIF. The added value of lipolysis products was 3492
DOI: 10.1021/acs.molpharmaceut.6b00496 Mol. Pharmaceutics 2016, 13, 3484−3493
Article
Molecular Pharmaceutics
digestion and absorption. 1. Phase behavior and aggregation states of model lipid systems patterned after aqueous duodenal contents of healthy adult human beings. Biochemistry 1990, 29 (8), 2028−40. (13) Porter, C. J.; Trevaskis, N. L.; Charman, W. N. Lipids and lipidbased formulations: optimizing the oral delivery of lipophilic drugs. Nat. Rev. Drug Discovery 2007, 6 (3), 231−48. (14) Williams, H. D.; Trevaskis, N. L.; Yeap, Y. Y.; Anby, M. U.; Pouton, C. W.; Porter, C. J. Lipid-based formulations and drug supersaturation: harnessing the unique benefits of the lipid digestion/ absorption pathway. Pharm. Res. 2013, 30 (12), 2976−92. (15) Mullertz, A.; Fatouros, D. G.; Smith, J. R.; Vertzoni, M.; Reppas, C. Insights into intermediate phases of human intestinal fluids visualized by atomic force microscopy and cryo-transmission electron microscopy ex vivo. Mol. Pharmaceutics 2012, 9 (2), 237−47. (16) Fatouros, D. G.; Bergenstahl, B.; Mullertz, A. Morphological observations on a lipid-based drug delivery system during in vitro digestion. Eur. J. Pharm. Sci. 2007, 31 (2), 85−94. (17) Augustijns, P.; Wuyts, B.; Hens, B.; Annaert, P.; Butler, J.; Brouwers, J. A review of drug solubility in human intestinal fluids: implications for the prediction of oral absorption. Eur. J. Pharm. Sci. 2014, 57, 322−32. (18) Clarysse, S.; Psachoulias, D.; Brouwers, J.; Tack, J.; Annaert, P.; Duchateau, G.; Reppas, C.; Augustijns, P. Postprandial changes in solubilizing capacity of human intestinal fluids for BCS class II drugs. Pharm. Res. 2009, 26 (6), 1456−66. (19) Vertzoni, M.; Markopoulos, C.; Symillides, M.; Goumas, C.; Imanidis, G.; Reppas, C. Luminal lipid phases after administration of a triglyceride solution of danazol in the fed state and their contribution to the flux of danazol across Caco-2 cell monolayers. Mol. Pharmaceutics 2012, 9 (5), 1189−98. (20) Holmstock, N.; De Bruyn, T.; Bevernage, J.; Annaert, P.; Mols, R.; Tack, J.; Augustijns, P. Exploring food effects on indinavir absorption with human intestinal fluids in the mouse intestine. Eur. J. Pharm. Sci. 2013, 49 (1), 27−32. (21) Wuyts, B.; Riethorst, D.; Brouwers, J.; Tack, J.; Annaert, P.; Augustijns, P. Evaluation of fasted and fed state simulated and human intestinal fluids as solvent system in the Ussing chambers model to explore food effects on intestinal permeability. Int. J. Pharm. 2014, 478 (2), 736−744. (22) Stappaerts, J.; Wuyts, B.; Tack, J.; Annaert, P.; Augustijns, P. Human and simulated intestinal fluids as solvent systems to explore food effects on intestinal solubility and permeability. Eur. J. Pharm. Sci. 2014, 63, 178−86. (23) Mullertz, A.; Fatouros, D. G.; Vertzoni, M.; Reppas, C. Unravelling the ultrastructure of ascending colon fluids from patients with ulcerative colitis by cryogenic transmission electron microscopy. J. Pharm. Pharmacol. 2013, 65 (10), 1482−7. (24) Kleberg, K.; Jacobsen, J.; Mullertz, A. Characterising the behaviour of poorly water soluble drugs in the intestine: application of biorelevant media for solubility, dissolution and transport studies. J. Pharm. Pharmacol. 2010, 62 (11), 1656−68. (25) Mullertz, A.; Reppas, C.; Psachoulias, D.; Vertzoni, M.; Fatouros, D. G. Structural features of colloidal species in the human fasted upper small intestine. J. Pharm. Pharmacol. 2015, 67 (4), 486− 92. (26) Fatouros, D. G.; Walrand, I.; Bergenstahl, B.; Mullertz, A. Colloidal structures in media simulating intestinal fed state conditions with and without lipolysis products. Pharm. Res. 2009, 26 (2), 361−74. (27) Armand, M. Lipases and lipolysis in the human digestive tract: where do we stand? Curr. Opin. Clin. Nutr. Metab. Care 2007, 10 (2), 156−64.
These variations in FeHIF ultrastructure may contribute to variable absorption of lipophilic drugs, which should be further investigated.
■
AUTHOR INFORMATION
Corresponding Author
*Address: Drug Delivery and Disposition, Gasthuisberg O&N 2, Herestraat 49, box 921, KU Leuven, 3000 Leuven, Belgium. Tel: +32 16 330301. Fax: +32 16 330305. E-mail: patrick.
[email protected]. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS The authors acknowledge ARIADME, a European FP7 ITN Community’s Seventh Framework Programme under Grant Agreement No. 607517.
■
REFERENCES
(1) Khoshakhlagh, P.; Johnson, R.; Langguth, P.; Nawroth, T.; Schmueser, L.; Hellmann, N.; Decker, H.; Szekely, N. K. Fasted-State Simulated Intestinal Fluid ″FaSSIF-C″, a Cholesterol Containing Intestinal Model Medium for In Vitro Drug Delivery Development. J. Pharm. Sci. 2015, 104, 2213. (2) Markopoulos, C.; Thoenen, F.; Preisig, D.; Symillides, M.; Vertzoni, M.; Parrott, N.; Reppas, C.; Imanidis, G. Biorelevant media for transport experiments in the Caco-2 model to evaluate drug absorption in the fasted and the fed state and their usefulness. Eur. J. Pharm. Biopharm. 2014, 86 (3), 438−48. (3) Galia, E.; Nicolaides, E.; Horter, D.; Lobenberg, R.; Reppas, C.; Dressman, J. B. Evaluation of various dissolution media for predicting in vivo performance of class I and II drugs. Pharm. Res. 1998, 15 (5), 698−705. (4) Kleberg, K.; Jacobsen, F.; Fatouros, D. G.; Mullertz, A. Biorelevant media simulating fed state intestinal fluids: colloid phase characterization and impact on solubilization capacity. J. Pharm. Sci. 2010, 99 (8), 3522−32. (5) Jantratid, E.; Janssen, N.; Reppas, C.; Dressman, J. B. Dissolution media simulating conditions in the proximal human gastrointestinal tract: an update. Pharm. Res. 2008, 25 (7), 1663−76. (6) Riethorst, D.; Mols, R.; Duchateau, G.; Tack, J.; Brouwers, J.; Augustijns, P. Characterization of Human Duodenal Fluids in Fasted and Fed State Conditions. J. Pharm. Sci. 2015, 105, 673. (7) Clarysse, S.; Tack, J.; Lammert, F.; Duchateau, G.; Reppas, C.; Augustijns, P. Postprandial evolution in composition and characteristics of human duodenal fluids in different nutritional states. J. Pharm. Sci. 2009, 98 (3), 1177−92. (8) Bergstrom, C. A.; Holm, R.; Jorgensen, S. A.; Andersson, S. B.; Artursson, P.; Beato, S.; Borde, A.; Box, K.; Brewster, M.; Dressman, J.; Feng, K. I.; Halbert, G.; Kostewicz, E.; McAllister, M.; Muenster, U.; Thinnes, J.; Taylor, R.; Mullertz, A. Early pharmaceutical profiling to predict oral drug absorption: current status and unmet needs. Eur. J. Pharm. Sci. 2014, 57, 173−99. (9) Fuchs, A.; Dressman, J. B. Composition and physicochemical properties of fasted-state human duodenal and jejunal fluid: a critical evaluation of the available data. J. Pharm. Sci. 2014, 103 (11), 3398− 411. (10) Fagerberg, J. H.; Bergstrom, C. A. Intestinal solubility and absorption of poorly water soluble compounds: predictions, challenges and solutions. Ther. Delivery 2015, 6 (8), 935−59. (11) Hernell, O.; Staggers, J. E.; Carey, M. C. Physical-chemical behavior of dietary and biliary lipids during intestinal digestion and absorption. 2. Phase analysis and aggregation states of luminal lipids during duodenal fat digestion in healthy adult human beings. Biochemistry 1990, 29 (8), 2041−56. (12) Staggers, J. E.; Hernell, O.; Stafford, R. J.; Carey, M. C. Physicalchemical behavior of dietary and biliary lipids during intestinal 3493
DOI: 10.1021/acs.molpharmaceut.6b00496 Mol. Pharmaceutics 2016, 13, 3484−3493
