capillary-wall collagen as a biophysical marker of ......k. yokoi and m. kojic share ﬁrst...

Integrated Systems and Technologies

Capillary-Wall Collagen as a Biophysical Marker ofNanotherapeutic Permeability into the TumorMicroenvironment

Kenji Yokoi1,2, Milos Kojic1,3, Miljan Milosevic3, Tomonori Tanei1, Mauro Ferrari1, and Arturas Ziemys1

AbstractThe capillary wall is the chief barrier to tissue entry of therapeutic nanoparticles, thereby dictating their efficacy.

Collagen fibers are an important component of capillary walls, affecting leakiness in healthy or tumor vasculature.Using a computationalmodel alongwith in vivo systems, we compared how collagen structure affects the diffusionflux of a 1-nm chemotherapeutic molecule [doxorubicin (DOX)] and an 80-nm chemotherapy-loaded pegylatedliposome (DOX-PLD) in tumor vasculature. We found a direct correlation between the collagen content arounda tumor vessel to the permeability of that vessel permeability to DOX-PLD, indicating that collagen content mayoffer a biophysical marker of extravasation potential of liposomal drug formulations. Our results also suggestedthat while pharmacokinetics determined the delivery of DOX and DOX-PLD to the same tumor phenotype,collagen content determined the extravasation of DOX-PLD to different tumor phenotypes. Transport physicsmayprovide adeeper view intohownanotherapeutics cross biological barriers, possibly helping explain thebalancebetween biological and physical aspects of drug delivery. Cancer Res; 74(16); 4239–46. �2014 AACR.

IntroductionInterest in biomedical nanoparticle and microparticle

applications is growing, especially for imaging and drugdelivery (1–3). Therapeutic particles bring several advan-tages to drug delivery, including the enhanced permeabilityand retention (EPR) effect, which is important for tumortreatment (4). It is assumed that EPR is related to the factthat particles of a certain size preferentially accumulate inthe tumor environment (5). An increase in the number ofparticles within tumors allows the delivery of more thera-peutic payload, for example, chemotherapeutic drugs. Drugcarriers, like liposomes or other particulates, are distributedthroughout tissues by convective transport within the vas-culature tree. Outside of the vessel walls, concentrationgradients frequently drive the diffusive transport of a ther-apeutic payload released passively into surrounding tissues,such as the tumor microenvironment (6). The importance oftransport physics goes beyond drug delivery: The physical

laws and principles that define the behavior of matter areessential for understanding the initiation and progression ofcancer at all size scales (7). The complex nature of biologycreates many transport barriers at different scales, demand-ing multiscale approaches to solve the riddles of oncophy-sical transport (8).

Capillarywalls and the surrounding tissues formadense andcrowded medium, impede the diffusion of therapeutics, andare among the major physical barriers to drug delivery. Diffu-sion can be tissue specific, and as in the case of tumors,diffusion also depends on drug properties (9). Therefore,pharmacokinetic aspects—especially profiles of drug concen-tration in plasma—have direct relation to drug extravasation,because concentration in plasma controls drug gradientsacross vessel wall. Also, the endothelial cells that tile thevascular wall and separate the blood flow from the tissuescontain transporter proteins that function as molecularpumps, fluxing out drug molecules (10, 11). However, endo-thelial cells may engulf and endocytose particles carrying alarge amount of drug molecules inside (12), or they may eventranscytose particulates, actively transporting them across theendothelium (13). Sometimes, capillaries develop fenestra-tions: openings through the capillary walls that lack endothe-lial cells and are covered by collagenous diaphragms (14).Studies show that the diaphragms have a sieving function,allowing the mass exchange of small molecules such as wateror proteins (15, 16).

Drugs and particles that penetrate intact capillary walls orfenestrations encounter a basal membrane, where the majorconstituent is type IV collagen (17). Physical aspects of trans-port have an important place in the oncological context,including the role of collagen in the transport of therapeutics

Authors' Affiliations: 1The Houston Methodist Research Institute; 2TheUniversity of Texas MD Anderson Cancer Center, Houston Texas; and3Belgrade Metropolitan University, Research and Development Center forBioengineering, Kragujevac, Serbia

Note: Supplementary data for this article are available at Cancer ResearchOnline (http://cancerres.aacrjournals.org/).

K. Yokoi and M. Kojic share first authorship of this article.

Corresponding Author: Arturas Ziemys, Houston Methodist ResearchInstitute, 6670 Bertner Avenue, R7-116, Houston, TX 77030. Phone:713-441-7320; Fax: 713-441-7438; E-mail: [email protected]

doi: 10.1158/0008-5472.CAN-13-3494

�2014 American Association for Cancer Research.

CancerResearch

www.aacrjournals.org 4239

on July 10, 2021. © 2014 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst May 22, 2014; DOI: 10.1158/0008-5472.CAN-13-3494

http://cancerres.aacrjournals.org/

(18). The ability of collagen to modulate vessel permeability,and drug permeability within tumors in general, was noticedpreviously (19–21) and even related to serum biomarkers (22).Different theoretical analyses and computational techniqueswere applied to model diffusion across capillary walls (23–25).How the particle size and collagen properties may modulatediffusion flux has not been explored.

Here, by combining a diffusionmodel and in vivo studies, weanalyzed the impact of the structure of the collagen sleeve onthe diffusive mass transport of the small molecule doxorubicin(DOX) and the 80-nm pegylated liposome (PLD), which togeth-er represent Doxil loaded with DOX inside.

Materials and MethodsCells

The 4T1,murine breast cancer, and 3LL,murine lung cancer,cells were kindly provided by Dr. I.J. Fidler (University of TexasMD Anderson Cancer Center, Houston, TX). The cells weremaintained in minimal essential medium supplemented with10% fetal bovine serum and supplements, as described previ-ously (26).

MiceFemale Balb/c and C57/BL6 mice were maintained in ani-

mal facilities at Houston Methodist Research Instituteapproved by the American Association for Accreditation ofLaboratory Animal Care and in accordance with currentregulations and standards of the United States Departmentof Agriculture, Department of Health andHuman Services, andNational Institutes of Health. Houston, TX.

TracersThe area of tumor tissue perfused by blood was evaluated

by imaging of a lysine-fixable 70-kDa fluorescein dextrantracer (Molecular Probes, Inc.) 1 hour after intravenousinjection.

Establishment of experimental tumorsTo produce tumors, 4T1 and 3LL cells were harvested from

subconfluent cultures. A tumor cell suspension (1 � 105 cells/100 mL) was injected subcutaneously into Balb/c or C57/BL6mice, respectively.

Therapy and necropsyStarting 10 days after the inoculation with the tumor cells,

the mice were intravenously injected with 6 mg/kg of PLD(Doxoves-Liposomal Doxorubicin HCl, FormuMax ScientificInc.) and the mice were sacrificed 24 hours after the injection.Then, the tumors were excised, frozen, and stocked to evaluateendothelial cells, basementmembranes, and the accumulationof PLD in the tumors.

Immunofluorescent staining of endothelial cells (CD31),basement membranes (type IV collagen), proliferatingcells (Ki67), and imaging of PLD in the tumors

The frozen sections were fixed in cold acetone for 15minutes. After a protein block, immunofluorescent staining

of the endothelial cells or basement membrane was performedusing antibodies to CD31 (BD Biosciences), type IV collagen(Abcam), or Ki67 (Abcam), respectively (26). The sections werethen incubated with the corresponding fluorochrome-labeledsecondary antibodies (Jackson Immunoresearch). The imageswere captured using a confocal laser-scanning microscope(Carl Zeiss MicroImaging Inc.) and analyzed using the built-in image-analysis software (26). The ruby-red fluorescence ofanthracyclines in the tumor tissue was also imaged usingconfocal microscopy at the excitation and emission wave-lengths of 488 and 590 nm, respectively (27, 28). The imagedfluorescence of DOX originates mostly from free DOX, becausethe solid state of DOX inside the PLD strongly quenches thefluorescence of liposomal DOX (29).

Imaging analysisThe ImageJ program was used to analyze the tissue sam-

ples (30), using the adequate RGB channels as 8-bit images,which were treated with background subtraction and auto-matic thresholding using the triangle method. The resultspresented in this article were calculated from five replicatemeasurements. The whole images were used in the analysis.The numbers of collagen and vessel entities after automatictriangle thresholding were calculated using a standard par-ticle analysis procedure. To avoid misinterpreting monocytesas vessels in the CD31 staining, the minimal area was set to250 mm2. The colocalization between collagen and vessels wascalculated by using ImajeJ/Fiji program (coloc2 plug-in;ref. 31), where M1 coefficient denotes the overlap of vesselsby collagen and M2—the overlap of collagen by vessels. Theextravasation of the dextran tracer andDOXwas calculated asa function of the distance in terms of pixels (after automaticthresholding). In the case of dextran, the distance was cal-culated to the nearest vessel, whereas in the case of DOX, thedistance was calculated to the centers of red-cell nuclei. Thecount of cells (DOX-affected and proliferating cells stainedwith Ki67) and their statistics were calculated by usingstained nuclei centers.

Transport modelWe applied a recently developed, hierarchical, multiscale

diffusion modeling technique (32–35), which has an advantageover other approaches because it accounts for the materialmicrostructure and the physico-chemical characteristics of thediffusing molecules, including the size of the molecules orparticles. See Supplementary Material for details, and Supple-mentary Figs. S1 to S5 and Table S1.

ResultsTo examine the role of collagen in vascular permeability, we

used a mouse model to analyze the type IV collagen contentand its colocalization with blood vessels and the effects ofcollagen on extravasation and tumor-cell proliferation. The invivo study was followed by a computational analysis of thepermeability of the collagen network to DOX and PLD parti-cles, to interpret the role of collagen in modulating transportbetween the vessel and tumor microenvironments.

Yokoi et al.

Cancer Res; 74(16) August 15, 2014 Cancer Research4240




Collagen content and localizationThe content of type IV collagen (hereafter referred to as

collagen) in 3LL and 4T1 tumors was quantified in in vivo.The representative images of stained collagen show differ-ent structures, with 3LL tumors possessing brighter colla-gen nodules as compared with 4T1 tumors (Fig. 1A and B)and indicating larger collagen content. The majority of thecollagen nodules visually overlapped with the red-stainedcapillaries in 3LL tumors, but not in 4T1 tumors (Fig. 1C andD). Using imaging analysis, we outlined each vessel and zoneof high collagen content in the 3LL and 4T1 tumors, reveal-ing that the stained vessels (red) and collagen (green)overlapped substantially in the 3LL tumors (yellow zones;Fig. 1E and F). In contrast to the 3LL tumors, the 4T1 tumorsdid not show the same overlap. The main reason for thedifferent results could be the automatic segmentation anal-ysis of the tissue images. The lack of intensity of the stainedcollagen did not allow collagen segments to be identified.The 4T1 tumors showed more vessels and less vessel wallcollagen content than the 3LL tumors (Fig. 1G), which mayexplain the large amount of vessel and collagen colocaliza-tion observed in the images of the 3LL tumors but not inthose of the 4T1 tumors (Fig. 1H). Therefore, the vessels ofthe 4T1 tumors lacked collagen compared with the 3LLtumors.

Extravasation in the microenvironmentFITC-tagged dextran was used as a tracer to evaluate the

extravasation of small molecules into the tumor microen-vironment. Our imaging data showed that the tracer extrav-asated in both tumor types (Fig. 2). The extravasation oftracer in the 3LL tumors was localized around vessels andshowed a decaying intensity with increasing distance fromthe vessels (Fig. 2A); whereas in the 4T1 tumors, the extrav-asation was more scattered (Fig. 2B). Interestingly, many ofthe vessels in the 3LL tumors did not display visual extra-vasations of the tracer, meaning that the leakiness of the 3LLvasculature was heterogeneous, a characteristic feature oftumor vasculature (36). The normalized tracer extravasationfrom the vessel wall was calculated from the imaging data(Fig. 2C). Within 1 hour and within the detection limit, thetracer extravasated up to 40 and 60 mm from the vessel in the3LL and 4T1 tumors, respectively. Based on the half-width ofthe stained areas (Fig. 2C), the characteristic distance ofextravasation was approximately 20 and 30 mm for the 3LLand 4T1 tumors, respectively.Although the 3LL and 4T1 tumors showed different, but

comparable, extravasation patterns of dextran tracers, theextravasation of PLD was very different. The 3LL tumors didnot show any substantial concentration field around thevessels (Fig. 2D), whereas the 4T1 tumors showed a verydistinct concentration of red fluorescence (DOX) around thevessels (Fig. 2E). The density of DOX-containing nuclei wasquantified and showed that the 4T1 tumors had substantiallymore DOX-stained nuclei around the vessels (Fig. 2F). Takinginto account that only free DOX can be imaged (29), we canconclude that our results indicate free DOX delivered andreleased from PLD.

Cell proliferationProliferating cell nuclei were imaged and quantified by

immunofluorescence staining of the tumor sections with Ki67antibody to estimate the effects of DOX extravasation andcollagen content around the vessels. Figure 3A shows prolif-erating 3LL cells, which are positively stained with Ki67antibody (purple) with minor red nuclei containing DOX.Proliferation was dominant throughout all the imaged regions.

Figure 1. In vivo collagen features in 3LL and 4T1 tumors. The 3LL tumorscontained thick type IV collagen (green) structures (A), which weremissing in 4T1 tumors (B). The visual colocalization of vessels (red) andcollagen was stronger in the 3LL tumors (C) than in the 4T1 tumors (D).The image analysis revealed substantially more structural collagen/vessel overlap (yellow) in the 3LL tumors (E) than in the 4T1 tumors (F).Quantitative analysis showed more vessels in the 4T1 tumors than in the3LL tumors, but the collagen content in the 4T1 tumors was dramaticallylower (G). The colocalization of vessels and collagen was expressed inboth tumors, but the 3LL tumors have substantiallymore vessels coveredby collagen; M1 and M2 are Mender coefficients, indicating vesseloverlap with collagen (M1) and collagen overlap by vessel (H).

Vessel Collagen as a Biophysical Marker for Drug Delivery

www.aacrjournals.org Cancer Res; 74(16) August 15, 2014 4241




The 4T1 tumors showed zones of proliferation that do notcontain DOX (red nuclei; Fig. 3B). The average number ofproliferating cells decreased substantially in the 4T1 tumorscompared with controls without treatment; but in the 3LLtumors, the average number of proliferating cells actuallyincreased slightly compared with the controls (Fig. 3C). Also,the correlation analysis shows a decreased proliferating celldensity around vessels in the 4T1 tumors (Fig. 3D).

Calculated permeability to PLD and DOXTo investigate how the collagen sleeve alone may modulate

mass transport through the capillary wall, we developed acollagen-sleeve model that mimics the collagen surroundingcapillaries (17, 37–39). We studied the diffusion transport ofPLD and DOX, which differ in size and physico-chemicalproperties, in response to different structures of collagen-fibermesh.

Different collagen sleeve thicknesses were modeled bychanging the number of collagen-fiber layers (from 1 to 6;

each layer was 10-nm thick; ref. 37), covering the physiologicalthickness of collagen in a capillary wall (39). Increasing thenumber of fiber layers reduced the PLD flux much faster thanthe DOX flux (Fig. 4A). Therefore, PLDmet stronger resistancethan DOX diffusing through the fiber mesh, because of thedifference in particle size (PLD and DOX have sizes of 80 and 1nm, respectively). Furthermore, the influence of the collagenfiber density was calculated for a 3-layer sleeve with meshopenings ranging from 50 to 800 nm in size (Fig. 4B). Transportof DOX molecules was not affected substantially by increasingthe collagen-fiber density, mostly because the DOX molecularsize is substantially smaller than any opening in the mesh, andconsequently there was a lack of resistance from the collagensleeve. However, the calculated PLDfluxwas sharply decreasedby shrinking the mesh opening below 200 nm, and the flux wasdiminished below an opening size of 100 nm—as the size of theopenings approached the size of PLD. The ratios of the PLDparticle size to the mesh opening size were 1:2.5, 1.25, and 0.62for mesh sizes of 200, 100, and 50 nm, respectively.

Each PLD contains a large amount of DOX, similar to Doxil.Based on our volumetric estimates, a Doxil liposome couldcontain DOX in a ratio of approximately 1:4,000 ormore, whichis the same order of magnitude as the 1:10,000 ratio reportedpreviously (40). Using different DOX/PLD ratios, we calculatedthe DOX flux for different mesh sizes and compared the resultsto those calculated for free DOX flux. Figure 5 shows that for avery tight collagen mesh with a 50 to 100 nm openings, thepacking of DOX did not affect the DOX flux, so free DOX wouldbe more efficient. However, the DOX packing inside liposomesbecame advantageous for a collagen mesh with an opening of200 nm or larger. The strongest effect was found for the

Figure 3. Cell proliferation in the 3LL and 4T1 tumors was a function ofDOX extravasation. The 3LL sample (A) shows that proliferating cells(purple) dominate the imaging area, while proliferation is absent whereDOX extravasated (red) in the 4T1 tumor (B). The count of proliferatingcells in the 3LL and 4T1 tumors (C) and the proliferating cell densityaround the vessels as a function of distance (D).

Figure 2. The extravasation differences of the tracer (left) and the PLD(right) in the 3LL and 4T1 tumorswithin 1 and 24 hours after the tracer andPLD administration, respectively. The tracer (tagged dextran; green)extravasated from capillaries (red) in the 3LL (A) and 4T1 (B) tumors. Thetracer concentration field appeared slightly farther from the vessels inthe 4T1 tumors than in 3LL tumors (C). The PLD (red) extravasation fromthe vessels (green) was negligible in the 3LL tumors (D), but substantial inthe 4T1 tumors (E). The 4T1 tumors contained a larger number ofnuclei affected by DOX than the 3LL tumors (F).

Yokoi et al.





liposomes with the highest DOX packing density, suggestingthat it could potentially determine the therapeutic propertiesof such particles.We created a diffusionmodel containing a source (capillary),

vessel wall (represented by the collagen sleeve), and a cellularmicroenvironment surrounding the capillary. But it is impor-tant to note that this model is focused to the collagen roleand does not seek to explain the actual extravasation in thepresence of endothelium or other biological factors. Figure 6Ashows a DOX concentration field developed around a capillaryin a 4T1 tumor within 24 hours after treatment with PLD,whereDOX intensity decreases with increasing distance from avessel (Supplementary Fig. S5). In view of collagen as a physicalbarrier in drug transport, the models suggest that DOX canpenetrate the collagen sleeve, or any other dense environment,more efficiently than PLD. Therefore, we calculated PLD andDOX extravasation in a phenomenological fashion using theirconcentration profiles in plasma, similar to those found in aother study (41), where the same DOX concentration wasinjected in two different formulations (pure DOX and PLD),and found that concentration of DOX released from PLD washigher than that of free DOX over 24 hours (Fig. 6B). The

Figure 5. The effect of DOX loading into PLD on DOX flux through acollagen mesh with the same DOX and PLD particle concentrations.

Figure 4. The calculated flux of DOX and PLD particles through thecollagen sleeve. The PLD diffusive flux was reduced faster by increasingthe collagen thickness with a collagen-mesh size of 400 nm (A), whereasmesh size had limited impact on DOX (B). The model revealed a smalleffect of collagen mesh size on DOX flux, whereas PLD flux wasdiminished abruptly when the mesh size approached the size of PLD(below 200 nm). Flux is presented on a logarithmic scale.

Figure 6. The concentration fields of DOX andPLD around the capillary. A,DOX concentration (red) around capillaries (green) in a 4T1 tumor; someproliferating nuclei staining in purple with Ki67. By using experimentalconcentration profiles in plasma (B), the concentration fields for DOX (C)and PLD (D) were calculated, with the vessel being a sourcesurrounded by a 100-mm-thick microenvironment medium. E, thecalculated profiles of average DOX concentration in tissue domain overtime; the adjacent numbers show the AUC in mmol � hours.






concentrations of both PLD and DOX inside the tumor micro-environment increased and then diminished over 24 hours inourmodels (Fig. 6C andD). The concentration of the larger andmore slowly diffusing PLD particle changed slower and per-sisted longer in the microenvironment, whereas the concen-tration gradient of DOX developed quickly and was rapidlycleared. The overall concentration behavior was very similar tothat in a previous study (42), comparing the DOX concentra-tions after treating a tumorwith pureDOX and liposomal DOX.The concentration of the pure DOX in the tumor was muchlower than that of the liposomal formulation. The longerpersistence of PLD in the plasma resulted in a more sustainedDOX concentration in tumor microenvironments that cansynergize with DOX loading in PLD particles and is reflectedin the calculated average concentration of DOX (Fig. 6E).

DiscussionWe have attempted to establish a physical linkage between

vessel wall collagen and drug extravasation. To understandsome fundamental properties of collagen type IV in basementmembrane, we have developed computational models ofcollagen sleeve that surround capillaries in tumor microen-vironment. The models did not use any parameters fitting,and rely merely on laws of physical and chemistry, and knownphysiological properties, such as the thickness of collagenfibers, the thickness of collagen sleeve around capillaries,dimensions of therapeutic substances, plasma concentrationprofiles of PLD or DOX, and their diffusivities. The resultsfrom collagen models revealed that a collagen sleeve alonedoes not function as a substantial barrier to DOX, which alsocould be extrapolated to other small molecules. However,vessel wall collagen may be a barrier to PLD carrying DOX,especially if PLD has to extravasate through dense and thickcollagen layers. Those characteristics are crucial in furtherdiscussion of our results.

Our quantified in vivo results indicated that 3LL and 4T1tumors are different in the contents of collagen type IV, whichis the major constituent is type IV collagen (17). Moreover, thecolocalization of collagen and capillaries was observed byfinding yellow zones (Fig. 1), which indicates the colocalizationof red and green fluorescence (43, 44). McDonald and collea-gues showed similar colocalization of vessels and collagen (44).Their study found that tumor vessels—in contrast to healthyvessels—had separated layers of collagen, endothelial cells, andpericytes. The results of the same study suggested that collagencontent might be different in different vascular segments,leading to differences in how tightly the collagen in basementmembrane and endothelial cells adhere to one another. There-fore, the imaging results illustrated in Fig. 1 could indicatedifferent phonotypic states of vessels and associated collagenin the 3LL and 4T1 tumors. In our experiment, no noticeablePLD extravasation was observed in healthy organs (Supple-mentary Fig. S6). This finding supports lower systemic toxicityfor liposomal formulation of DOX (40) and suggests thatcollagen type IV could be very dynamic property of tumormicroenvironment and deserves deeper research spanningbeyond the scope of this study.

In view of collagen transport properties obtained from phys-icalmodels and collagen content from in vivo studies, the logicalcorrelates were found between collagen content in basementmembranes of capillaries and drug extravasations. The extrav-asation of the tracer in the 3LL and 4T1 tumors correlated withthe respective collagen contents, suggesting that vascular col-lagen could modulate the transport of small molecules tolimited amount. However, the DOX extravasation using PLDhad a strong negative correlation with the collagen content inthe 3LL and 4T1 tumors. The extravasation occurred fromvessels that were not tightly covered by collagen type IV(Supplementary Fig. S7). Other studies have shown that theextravasation of liposomes is limited and heterogeneous, form-ing perivascular clusters that do not move significantly andcould be observed for up to 1 week (45). Collagen, as the majorcomponent of the extracellular matrix, substantially limits thediffusion of PLD because of the liposome size (18). Moreover,tumor cells canhave a very limited role in internalizingPLD (46).Altogether, PLD displays impeded extravasation into the tumormicroenvironment beyond thepremises of vessels, as calculatedby our models and supported by our and other experimentalresults. As a consequence of an absence of thick collagenbundles around the vessels, the decrease in the proliferationof the 4T1 cells may be related to the large amount of extrav-asated DOX loaded inside PLD. In the case of the 3LL tumors,the presence of a thick collagen coating around the vessels wascorrelated with extremely limited or absent extravasation ofDOX loaded inside PLD, which did not affect cell proliferation.

The collagen type IV content around vessels by itself couldpotentially be a biophysical marker predicting drug extrava-sation and therapeutic efficacy. Because passive transport, thatis, diffusion, obeys the laws of physics—whether in vitro orwithin biological tissues—the models help to decouple theroles of physical and biologic transport mechanisms and aresupported by in vivo results. The use of PLD prolonged thecirculation time of DOX, facilitated DOX extravasation, anddecreased the tumor-cell proliferation rate in the 4T1 tumors.Despite the 3LL tumors have more collagen, the tracer per-meability was similar in the 3LL and 4T1 tumors (althoughslightly more extravasation of the tracer was found in the 4T1tumors). This is in agreement with the model results andsuggests that collagen may have a small effect on the extrav-asation of small molecules, because small molecules shoulddiffuse relatively easily through capillary collagen.

Therefore, differences in DOX extravasation might be relat-ed to endothelium as a barrier to PLD before even reachingcollagen itself. The role of collagen in angiogenesis has beenshown to the degree that the vessel collagen sleeve remainsafter antivascular therapy and may serve as a scaffold torevascularize tissues (47). Moreover, the disruption of contactbetween endothelial cells and the vessel basement membranecan lead to endothelial apoptosis (44). On the other hand,fenestrations are structures of the basementmembrane, wherecollagen plays a major role too. It is widely accepted thatfenestrationsmay help to deliver therapeutic payloads (48–50).Furthermore, the plasma concentration kinetics may play animportant role, favoring long-circulating PLD, as in the exam-ple presented in Fig. 6. The results show that even at small

Yokoi et al.





1:1000 DOX packing inside PLD, the calculated area under thecurve (AUC) is larger than using free DOX. The drug packing inPLD revealed strong effects to drug delivery in our transportmodels, even if PLD extravasation is limited comparing to freedrug. The PLD concentration in plasma may be lower byseveral orders of magnitude than that of the free therapeuticmolecule, but still that can possess a similar or higher ther-apeutic efficacy with respect to the free molecule formulation,because each liposome may carry many molecules of payload.However, according to our models, the advantages of PLD willdiminish and smaller molecules would penetrate the collagensleevemore efficiently, if collagenmesh size gets comparable tothe size of PLD or gets smaller.The correlation between the collagen content around tumor

vessels and the permeability of a liposomal formulation of DOXwas found, indicating that collagen content could be a bio-physical marker useful to predict drug extravasation potential.Experiments and model synergistically show that collagensleeve may modulate vessel permeability to therapeutic par-ticles by size exclusion. Our results show differences in DOXextravasation among tumor phenotypes when PLD formula-tion is used. This can be explained by differences in capillarycollagen content. Larger amount of collagen associated withvessels correlates with poorer DOX extravasation into tumormicroenvironment. The differences in pharmacokinetics ofDOX and PLD suggest the benefits of using PLD in the presenceof collagen as a barrier for drug delivery.To thebest of our knowledge, biophysicalmarkers that utilize

the EPR effect and assist in the prediction of the therapeuticefficacy have not been reported to date. The analysis ofnanotherapeutics transport from systemic circulation to tumormicroenvironment provides additional framework to improvecancer therapy. The differences in the collagen content and

structure—as the marker of therapeutics extravasation—couldbe taken into account for planning the therapeutic strategy forpatients. However, further studies focused on evaluating thepotential of biophysical markers are needed to help in person-alizing nanotherapeutic for cancer therapy.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design: K. Yokoi, M. Kojic, T. Tanei, A. ZiemysDevelopment of methodology: M. Kojic, M. Milosevic, A. ZiemysAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): K. YokoiAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): K. Yokoi, M. Kojic, M. Milosevic, A. ZiemysWriting, review, and or revision of the manuscript: K. Yokoi, M. Kojic,A. ZiemysAdministrative, technical, or material support (i.e., reporting or orga-nizing data, constructing databases): A. ZiemysStudy supervision: M. Kojic, M. Ferrari, A. Ziemys

AcknowledgmentsThe authors thank the Texas Advanced Computing Center (TACC) at The

University of Texas at Austin for providing HPC resources that contributed to theresults reported in this article.

Grant SupportThis work was supported by the grant OI 174028 of the Serbian Ministry of

Education and Science (M. Kojic), the NIH (U54CA143837 to M. Ferrari andK. Yokoi; U54CA151668 to M. Ferrari), the Ernest Cockrell Jr. DistinguishedEndowed Chair (M. Ferrari), and the U.S. Department of Defense (W81XWH-09-1-0212 to M. Ferrari).

The costs of publication of this article were defrayed in part by the payment ofpage charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received December 9, 2013; revised February 27, 2014; accepted March 22,2014; published OnlineFirst May 22, 2014.

References1. KooOM, Rubinstein I, Onyuksel H. Role of nanotechnology in targeted

drug delivery and imaging: a concise review. Nanomedicine 2005;1:193–212.

2. Ferrari M. Cancer nanotechnology: opportunities and challenges. NatRev Cancer 2005;5:161–71.

3. Peer D, Karp JM, Hong S, Farokhzad OC, Margalit R, Langer R.Nanocarriers as an emerging platform for cancer therapy. Nat Nano-technol 2007;2:751–60.

4. Maeda H, Wu J, Sawa T, Matsumura Y, Hori K. Tumor vascularpermeability and the EPR effect in macromolecular therapeutics: areview. J Control Release 2000;65:271–84.

5. Matsumura Y, Maeda H. A new concept for macromolecular thera-peutics in cancer chemotherapy: mechanism of tumoritropic accu-mulation of proteins and the antitumor agent smancs. Cancer Res1986;46:6387–92.

6. Yuan F. Transvascular drug delivery in solid tumors. Semin RadiatOncol 1998;8:164–175.

7. Michor F, Liphardt J, FerrariM,WidomJ.What doesphysics have to dowith cancer? Nat Rev Cancer 2011;11:657–70.

8. Ferrari M. Frontiers in cancer nanomedicine: directing mass transportthrough biological barriers. Trends Biotechnol 2010;28:181–188.

9. Yuan F, Dellian M, Fukumura D, Leunig M, Berk DA, Torchilin VP, et al.Vascular permeability in a human tumor xenograft: molecular sizedependence and cutoff size. Cancer Res 1995;55:3752–6.

10. Sharom FJ. Abc multidrug transporters: structure, function and role inchemoresistance. Pharmacogenomics 2008;9:105–27.

11. Gottesman MM. Mechanisms of cancer drug resistance. Annu RevMed 2002;53:615–27.

12. Sahay G, Alakhova DY, Kabanov AV. Endocytosis of nanomedicines.J Control Release 2010;145:182–95.

13. SimionescuM,PopovD,SimaA. Endothelial transcytosis in health anddisease. Cell Tissue Res 2009;335:27–40.

14. Schechter J. Ultrastructural changes in the capillary bed of humanpituitary tumors. Am J Pathol 1972;67:109.

15. Roberts WG, Palade GE. Increased microvascular permeability andendothelial fenestration induced by vascular endothelial growth factor.J Cell Sci 1995;108:2369–79.

16. Bearer EL, Orci L. Endothelial fenestral diaphragms: a quick-freeze,deep-etch study. J Cell Biol 1985;100:418–28.

17. Yurchenco PD, Ruben GC. Basement membrane structure in situ:evidence for lateral associations in the type IV collagen network. J CellBiol 1987;105:2559–68.

18. Ramanujan S, Pluen A,Mckee T, Brown E, Boucher Y, Jain R. Diffusionand convection in collagen gels: implications for transport in the tumorinterstitium. Biophys J 2002;83:1650–60.

19. Robert A, Godeau G, Miskulin M, Moati F. Mechanism of action ofcollagenase on the permeability of the blood-brain barrier. NeurochemRes 1977;2:449–55.

20. Utoguchi N, Mizuguchi H, Dantakean A, Makimoto H, Wakai Y, Tsut-sumi Y, et al. Effect of tumour cell-conditioned medium on endothelialmacromolecular permeability and its correlation with collagen. BrJ Cancer 1996;73:24.






21. Netti PA, Berk DA, Swartz MA, Grodzinsky AJ, Jain RK. Role ofextracellular matrix assembly in interstitial transport in solid tumors.Cancer Res 2000;60:2497–503.

22. Yokoi K, Tanei T, Godin B, Ven AVD, Hanibuchi M, Matsunoki A, et al.Serumbiomarkers for personalizationof nanotherapeutics-based ther-apy in different tumor and organmicroenvironments. Cancer Lett 2014Apr 1;345:48–55.

23. Jain RK. Transport of molecules across tumor vasculature. CancerMetastasis Rev 1987;6:559–93.

24. Bassingthwaighte JB, Chan ISJ, Wang C. Computationally efficientalgorithms for convection-permeation-diffusion models for blood-tis-sue exchange. Ann Biomed Eng 1992;20:687–725.

25. Fujimori K, Covell DG, Fletcher JE, Weinstein JN. Modeling analysis ofthe global and microscopic distribution of immunoglobulin g, f(ab0) 2,and fab in tumors. Cancer Res 1989;49:5656–63.

26. Yokoi K, Godin B,Oborn CJ, Alexander JF, Liu X, Fidler IJ, et al. Poroussilicon nanocarriers for dual targeting tumor associated endothelialcells and macrophages in stroma of orthotopic human pancreaticcancers. Cancer Lett 2013 Jul 1;334:319–27.

27. Vaage J, Donovan D, Uster P, Working P. Tumour uptake of doxoru-bicin in polyethylene glycol-coated liposomes and therapeutic effectagainst a xenografted humanpancreatic carcinoma. Br JCancer 1997;75:482.

28. Erikson A, Tufto I, Bjønnum AB, Bruland�S, Davies CL. The impact ofenzymatic degradation on the uptake of differently sized therapeuticmolecules. Anticancer Res 2008;28:3557–66.

29. Lee RJ, Low PS. Folate-mediated tumor cell targeting of liposome-entrapped doxorubicin in vitro. Biochim Biophys Acta 1995;1233:134–44.

30. Abr�amoff MD, Magalh~aes PJ, Ram SJ. Image processing with imagej.Biophotonics Int 2004;11:36–42.

31. Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M,Pietzsch T, et al. Fiji: an open-source platform for biological-imageanalysis. Nat Meth 2012;9:676–82.

32. Ziemys A, Kojic M, Milosevic M, Kojic N, Hussain F, Ferrari M, et al.Hierarchical modeling of diffusive transport through nanochannels bycoupling molecular dynamics with finite element method. J ComputPhys 2011;230:5722–31.

33. Kojic M, Milosevic M, Kojic N, Ferrari M, Ziemys A. On diffusion innanospace. J Serb Soc Comput Mech 2011;5:104–18.

34. Kojic M, Ziemys A, Milosevic M, Isailovic V, Kojic N, Rosic M, et al.Transport in biological systems. J Serb Soc Comput Mech 2011;5:101–28.

35. Kojic M, Milosevic M, Kojic N, Kim K, Ferrari M, Ziemys A. A multiscalemd-fe model of diffusion in composite media with internal surfaceinteraction based on numerical homogenization procedure. ComputMethods Appl Mech Eng 2014;269:123–38.

36. EberhardA,Kahlert S,GoedeV,HemmerleinB,PlateKH,AugustinHG.Heterogeneity of angiogenesis and blood vessel maturation in humantumors: implications for antiangiogenic tumor therapies. Cancer Res2000;60:1388–93.

37. Timpl R, Wiedemann H, Delden V, Furthmayr H, Kuhn K. A networkmodel for the organization of type IV collagen molecules in basementmembranes. Eur J Biochem 1981;120:203–11.

38. Van Der Rest M, Garrone R. Collagen family of proteins. FASEB J1991;5:2814–23.

39. LaurieG, LeblondC,MartinG. Localization of type IV collagen, laminin,heparan sulfate proteoglycan, and fibronectin to the basal lamina ofbasement membranes. J Cell Biol 1982;95:340–4.

40. Barenholz YC. Doxil�—the first FDA-approved nano-drug: lessonslearned. J Control Release 2012;160:117–34.

41. Junping W, Maitani Y, Takayama K, Nagai T. In vivo evaluation ofdoxorubicin carried with long circulating and remote loading prolipo-some. Int J Pharm 2000;203:61–9.

42. Laginha KM, Verwoert S, Charrois GJ, Allen TM. Determination ofdoxorubicin levels in whole tumor and tumor nuclei in murine breastcancer tumors. Clin Cancer Res 2005;11:6944–9.

43. Cretu A, Roth JM, Caunt M, Akalu A, Policarpio D, Formenti S, et al.Disruption of endothelial cell interactions with the novel hu177 crypticcollagen epitope inhibits angiogenesis. Clin Cancer Res 2007;13:3068–78.

44. Baluk P, Morikawa S, Haskell A, Mancuso M, Mcdonald DM. Abnor-malities of basement membrane on blood vessels and endothelialsprouts in tumors. Am J Pathol 2003;163:1801–15.

45. Yuan F, Leunig M, Huang SK, Berk DA, Papahadjopoulos D, Jain RK.Mirovascular permeability and interstitial penetration of sterically sta-bilized (stealth) liposomes in a human tumor xenograft. Cancer Res1994;54:3352–6.

46. Ishida O, Maruyama K, Sasaki K, Iwatsuru M. Size-dependent extrav-asation and interstitial localization of polyethyleneglycol liposomes insolid tumor-bearing mice. Int J Pharm 1999;190:49–56.

47. Inai T, Mancuso M, Hashizume H, Baffert F, Haskell A, Baluk P, et al.Inhibition of vascular endothelial growth factor (VEGF) signaling incancer causes loss of endothelial fenestrations, regression of tumorvessels, and appearance of basement membrane ghosts. Am J Pathol2004;165:35–52.

48. GaumetM, Vargas A, Gurny R, Delie F. Nanoparticles for drug delivery:the need for precision in reporting particle size parameters. Eur JPharm Biopharm 2008;69:1–9.

49. Jain RK, Stylianopoulos T. Delivering nanomedicine to solid tumors.Nat Rev Clin Oncol 2010;7:653–664.

50. Davis ME. Nanoparticle therapeutics: an emerging treatment modalityfor cancer. Nat Rev Drug Discov 2008;7:771–82.


Yokoi et al.




2014;74:4239-4246. Published OnlineFirst May 22, 2014.Cancer Res Kenji Yokoi, Milos Kojic, Miljan Milosevic, et al. Permeability into the Tumor MicroenvironmentCapillary-Wall Collagen as a Biophysical Marker of Nanotherapeutic

Updated version

10.1158/0008-5472.CAN-13-3494doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerres.aacrjournals.org/content/suppl/2021/03/31/0008-5472.CAN-13-3494.DC1

Access the most recent supplemental material at:

Cited articles

http://cancerres.aacrjournals.org/content/74/16/4239.full#ref-list-1

This article cites 50 articles, 13 of which you can access for free at:

Citing articles

http://cancerres.aacrjournals.org/content/74/16/4239.full#related-urls

This article has been cited by 1 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/74/16/4239To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-13-3494

http://cancerres.aacrjournals.org/content/suppl/2021/03/31/0008-5472.CAN-13-3494.DC1

http://cancerres.aacrjournals.org/content/74/16/4239.full#ref-list-1

http://cancerres.aacrjournals.org/content/74/16/4239.full#related-urls

http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://cancerres.aacrjournals.org/content/74/16/4239


capillary-wall collagen as a biophysical marker of ......k. yokoi and m. kojic share ﬁrst...

Documents