Influenza A virus mimetic nanoparticles triggerselective cell uptakeSara Maslanka Figueroaa, Anika Vesera, Kathrin Abstiensa, Daniel Fleischmanna, Sebastian Becka,and Achim Goepfericha,1

aDepartment of Pharmaceutical Technology, University of Regensburg, 93053 Regensburg, Germany

Edited by Robert Langer, Massachusetts Institute of Technology, Cambridge, MA, and approved April 2, 2019 (received for review February 12, 2019)

Poor target cell specificity is currently a major shortcoming ofnanoparticles (NPs) used for biomedical applications. It causessignificant material loss to off-target sites and poor availability atthe intended delivery site. To overcome this limitation, we designedparticles that identify cells in a virus-like manner. As a blueprint, wechose a mechanism typical of influenza A virus particles in whichectoenzymatic hemagglutinin activation by target cells is a manda-tory prerequisite for binding to a secondary target structure thatfinally confirms cell identity and allows for uptake of the virus. Wedeveloped NPs that probe mesangial cells for the presence ofangiotensin-converting enzyme on their surface using angiotensinI (Ang-I) as a proligand. This initial interaction enzymatically trans-forms Ang-I to a secondary ligand angiotensin II (Ang-II) that has thepotential to bind in a second stage to Ang-II type-1 receptor (AT1R).The presence of the receptor confirms the target cell identity andtriggers NP uptake via endocytosis. Our virus-mimetic NPs showedoutstanding target-cell affinity with picomolar avidities and wereable to selectively identify these cells in the presence of 90% off-target cells that carried only the AT1R. Our results demonstrate thatthe design of virus-mimetic cell interactive NPs is a valuable strategyto enhance NP specificity for therapeutic and diagnostic applications.Our set of primary and secondary targets is particularly suited for theidentification of mesangial cells that play a pivotal role in diabeticnephropathy, one of the leading causes of renal failure, for whichcurrently no treatment exists.

virus-mimetic nanoparticles | influenza A | heteromultivalent | targetspecific | enzyme responsive

Nanomaterials are valuable tools in the field of drug delivery asthey can target cells with high specificity, avoiding the side

effects of conventional drug administration (1). Nanoparticles(NPs) with the ability to bind to distinct target structures on cellsurfaces are particularly valuable (2). However, various publica-tions have raised doubts about the efficacy of cell targeting, in-cluding the metaanalysis published by Wilhelm et al. whichscreened 10 y of preclinical in vivo cancer therapy data (3). Theiranalysis revealed that the presence of a target cell-specific ligandon a NP surface gave only a modest increase in the percentage ofthe dose that entered solid tumors with a value of 0.9% comparedwith 0.7% for NPs without any cell recognition mechanism. Eventhough the authors identified a number of factors that could beresponsible for the low uptake values, they concluded that poor cellrecognition was particularly decisive and that the development ofbetter strategies for target-cell identification is a significant need.In fact, a plethora of strategies have been explored to improve

the ability of NPs to recognize a target cell with sufficient speci-ficity. Progress has been made especially in the understanding ofmultivalent ligand–receptor interactions and their contribution tothe avidity of a particular nanomaterial for cells that carry therespective receptor (4). Thus, we know that a NP equipped withhighly specific ligands tethered to its surface can achieve nano-molar avidity for the target cell (5). In this case, the multivalentbinding (6) compensates for the affinity loss stemming from themassive structural changes to the ligands when they are attachedto linkers or tethering molecules such as poly(ethylene glycol)

(PEG) (7). The mechanism has been well documented in the lit-erature for several nanomaterials (8–10) binding to G protein-coupled receptors (GPCRs) (5, 11, 12) or integrins (13). How-ever, high avidity is a necessary but not sufficient prerequisite foridentifying target cells. A fundamental problem that cannot beovercome with “simple” multivalent particles is the lack of speci-ficity. Since there may be more cell types in the organism that carrythe receptor chosen for targeting, this can increase the odds ofparticles binding to off-target sites (14).In attempts to overcome poor specificity, heteromultivalency

(15) has frequently been explored. This strategy takes advantage ofthe fact that targeting multiple receptors decreases the chance thatoff-target cells carry the same set of receptors. However, when NPscarried several different types of ligands rather than only one formore precise target-cell identification, the increases in specificitywere modest. Because the ligands are presented simultaneously bythe nanomaterials, any cell expressing either of the respective re-ceptors will bind the particles and interfere with them.To overcome this limitation, we developed NPs that interact

with cells in a virus-like manner (16). In contrast to simple heter-omultivalently binding NPs, viral cell attachment and entry is acomplex sequential process of two or more stages (17). Adenovi-ruses, for example, reveal their ligand for integrin binding thattriggers cell uptake only after first binding to the coxsackie ade-novirus receptor (18). Our approach to increase target-cell speci-ficity by using more than one target site for cell identification wasinspired by influenza A virus which requires enzymatic activation of

Significance

The identification and targeting of specific cell types are majorshortcomings of nanoparticles (NPs) for theragnostic applica-tions. Even in simple cell culture systems, today’s NPs cannotunequivocally distinguish target from off-target cells. To over-come this restraint, we designed NPs that identify cells in aninfluenza A virus-like manner. Our NPs probe mesangial cells forangiotensin-converting enzyme using angiotensin I (Ang-I) assubstrate. The resulting enzymatic reaction transforms Ang-Iinto Ang-II. Binding of the latter to the Ang-II type-1 receptor(AT1R) confirms the cells’ identity and triggers NP uptake. Theparticles identified their target cells even with off-target cellspresent. With cell avidities similar to those of antibodies, theparticles are promising transporters for therapeutic and di-agnostic applications with pinpoint accuracy.

Author contributions: A.G. proposed research; S.M.F., A.V., K.A., D.F., S.B., and A.G. de-signed research; S.M.F. performed research; S.M.F. analyzed data; S.M.F., A.V., K.A., D.F.,S.B., and A.G. discussed results and commented on the manuscript; and S.M.F. and A.G.wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Published under the PNAS license.1To whom correspondence should be addressed. Email: achim.goepferich@ur.de.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1902563116/-/DCSupplemental.

Published online April 29, 2019.

www.pnas.org/cgi/doi/10.1073/pnas.1902563116 PNAS | May 14, 2019 | vol. 116 | no. 20 | 9831–9836

APP

LIED

BIOLO

GICAL

SCIENCE

S

Dow

nloa

ded

by g

uest

on

Dec

embe

r 6,

202

0

hemagglutinin, a proligand glycoprotein in the viral envelope. Onceactivated, the ligand binds to cell surface sialic acid, triggering celluptake (19). We aimed to use this principle to engineer NPs withhigh specificity for mesangial cells, which play a pivotal role in thedevelopment of diabetic nephropathy (20). We designed NPs witha surface-immobilized substrate specific to an enzyme located onthe target-cell membrane, specifically the proligand Ang-I whichwe expected to be “visible” to cells bearing the primary target,angiotensin-converting enzyme (ACE) (Fig. 1A). The subsequentenzymatic reaction cleaves the dipeptide His-Leu from Ang-Iconverting it to Ang-II, the secondary active ligand. Ang-II canthen bind to the Ang-II type-1 receptor (AT1R) present on thetarget cell. As an agonist, it triggers receptor-mediated endocytosis(21). If NPs come into contact with off-target cells expressing onlyone of the two target receptors, ACE or AT1R, the NPs will eitherbe activated but not internalized or they will not be able to bind thecell at all (Fig. 1B)

ResultsFor the development of our virus-mimetic NPs, we relied on well-known polymers recognized for their biocompatibility and highreproducibility of particle manufacture. For NP synthesis, PEG-poly(lactic acid) (PEG-PLA) (23) block copolymers were stabi-lized against disassembly in cell culture medium with hydrophobicpoly(lactic-coglycolic acid) (PLGA) cores (24). The combinationof both polymers offers the possibility of integrating favorabletraits regarding drug encapsulation, size, and controlled drug re-lease (25, 26). Concomitantly it circumvents the size increasePEG-PLA NPs would undergo when stabilization was achieved byincreasing the molecular weight of the PLA block (27). We cou-pled lysine N-modified Ang-I (Lys–Ang-I) and Lys–Ang-II to NPsbearing carboxylic acid-modified PEG chains on their surface(NP-COOH) such that between 15% and 25% of PEG chains perparticle were modified (SI Appendix, Fig. S1 A and C). Using co-polymers with different PEG chain lengths allowed for preparation

of NPs with different sizes (SI Appendix, Fig. S1B). This is usefulbecause size has been found to be an important parameter affectingNP–cell interactions (28, 29). The angiotensin-modified NPs weprepared were 50 nm (NP210Ang-I/II) and 80 nm (NP510Ang-I/II) insize with narrow particle size distributions [polydispersity index(PDI) 0.1–0.2] (SI Appendix, Fig. S1).First, we investigated whether Ang-II–decorated NPs were

able to bind to the AT1R and meet this fundamental pre-requisite for the concept of enzymatic ligand activation. Sincethe AT1R is a Gq-coupled receptor (30), its response to agonistbinding can be followed via cytosolic calcium measurements. Tothis end, we incubated AT1R-positive rat mesangial cells (rMCs)(SI Appendix, Figs. S2 and S3) with Ang-II–positive NPs. Theresulting intracellular calcium increase with rising particle con-centration is in line with the ability of Ang-II to mediate calciummobilization (SI Appendix, Fig. S2B) and confirms that the sec-ondary ligand bound to its target receptor (Fig. 2). NP210Ang-IIand NP510Ang-II yielded similar results, indicating that NP sizewas of minor significance. NPs decorated with the proligandAng-I, in contrast, were not able to bind the AT1R. Likewise,ligand-free control methoxy-ended particles (NPMeO) did notcause any calcium influx. Overall, the results confirmed that NPscarrying the secondary ligand Ang-II in their particle corona areable to bind to target cells via the AT1R, the designated sec-ondary target in this two-stage binding concept, with high aviditythat yielded EC50 values in the picomolar range.After confirming that Ang-II–modified particles bind to the

secondary target receptor, we were intrigued to know if the NP-bound Ang-I could bind to ACE as the primary target structureand if it would be converted to the secondary ligand Ang-II. Inthese experiments, soluble enzyme served as a surrogate for en-zyme immobilized on the target-cell surface. We incubated Ang-I–modified NPs with 0.1 μM rabbit lung ACE and tested again forAT1R activation in rMCs. Fig. 3 shows that enzyme-incubatedNPAng-I yielded a high intracellular calcium signal from whichwe concluded that the NPs must have bound to the ACE and thatAng-I in the particle corona must have been converted to Ang-II.The affinity of the interaction between NP-bound Ang-I and ACEwas of the same order of magnitude as that for the free substrate(Km of 8.1 ± 2.4 μM and 1.4 ± 0.3 μM for NP210Ang-I andNP510Ang-I, respectively) (SI Appendix, Fig. S5). In control ex-periments, we showed that receptor activation can be completelysuppressed by blocking it with valsartan, a highly affine AT1Rantagonist (Kd = 1.44 nM) (31). This proves that the signal musthave been NP triggered, confirming that the NPs had bound to thereceptor. When the enzyme was inhibited by captopril (pKi =9.40) (32) the signal dropped back to background levels, indicatinga failed enzymatic reaction. Due to the velocity of the enzymaticconversion, we also blocked the cell membrane-bound enzyme onthe cells during the measurements to avoid conversion by the cellsthemselves (SI Appendix, Fig. S4).After NP binding to the primary and secondary target mole-

cules was confirmed, we tested the interaction of the two-stagevirus-mimetic NPs with cells. AT1R- and ACE-expressing rMCsand human kidney 2 (HK-2) cells (SI Appendix, Fig. S2) wereincubated with NPs carrying the proligand (Ang-I) and in-vestigated for particle uptake. Internalization of NP510Ang-I andNP210Ang-I after proligand activation by cell membrane-boundACE was observed in both rMCs and HK-2 cells (Fig. 4 A–D). Itwas shown that both ACE activation and AT1R binding werenecessary by adding either captopril, an ACE inhibitor of theprimary target, or valsartan, an antagonist of the secondary tar-get. The fluorescence intensities measured in these control ex-periments were comparable to the ones obtained for the ligand-free NPMeO, which shows that the NPAng-I first recognizes andbinds ACE on the cell membrane, activating a number of pro-ligand Ang-I molecules and transforming them to the secondaryligand Ang-II. The newly created NPAng-II then binds to the

Primary Target(ACE)

Secondary Target(AT1R)

Enzymaticprocessing

Receptorbinding

Primary Secondarybinding recognition

Recognition failure

AT1Rmediated

Endocytosis

ACE

AT1R

Ang-I

Ang-IICell-uptakefailure

B

Proligand(Ang-I)

Secondary Ligand(Ang-II)

A

Target CellOff-target Cell I Off-target Cell II

Fig. 1. Illustration of the concept of heteromultivalent interactive target-cell identification by virus-mimetic NPs. (A) Target cell-specific recognitionand internalization make up a sequential two-stage process. During the firststage, Ang-I binds to ACE, the primary target on the cell surface with highaffinity (Km = 19 μM) (22). The subsequent cleavage reaction produces Ang-II,the secondary ligand. During the secondary recognition stage, Ang-II bindsto the secondary target AT1R. As a receptor agonist, it triggers endocytosisof the NP upon receptor binding (21). (B) NPs fail to recognize off-target cellsthat carry only the primary (off-target cell I) or secondary (off-target cell II)target receptor.

9832 | www.pnas.org/cgi/doi/10.1073/pnas.1902563116 Maslanka Figueroa et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 6,

202

0

secondary target AT1R, triggering receptor-mediated internali-zation. In rMCs (Fig. 4C) and HK-2 cells (Fig. 4D), the smallerNP210Ang-I were taken up in higher amounts than their largercounterparts, NP510Ang-I. This demonstrates a size-dependentinteraction between NPs and their target molecules and opensinteresting new possibilities for NP optimization.To confirm these results, the interaction of the cells with the

NPs was investigated by confocal laser scanning microscopy(CLSM) (Fig. 4 E–H). NP-associated fluorescence was observedin rMCs and HK-2 cells incubated with NP210Ang-I andNP510Ang-I (Fig. 4 E and F and Fig. 4 G and H, respectively).Cells incubated with ligand-free NPMeO show considerably lessNP-associated fluorescence. Again, NP uptake could be com-pletely suppressed by adding captopril and/or valsartan.To assess the target-cell specificity of the virus-mimetic NPs, we

investigated their interaction with cell lines that have differentexpression patterns of the target molecules (Fig. 5). In contrast torMCs and HK-2 cells, NCI-H295R cells lack the primary recog-nition molecule, ACE (SI Appendix, Figs. S2 and S3) and carryonly the secondary target receptor AT1R. Our virus-mimetic NPsshowed the highest uptake rates in rMCs and HK-2 cells and onlymoderate uptake in NCI-H295R cells. This moderate uptake wascomparable to that of ligand-free NPMeO control particles and isdue to the base level of nonspecific cellular uptake of NPs (33).These results confirm the high specificity of our interactive NPs, astheir uptake in NCI-H295R cells, which carry only one of twomandatory recognition structures, is as low as for nontargetedNPMeO. Furthermore, there is a strong correlation between theenzymatic activity of the different cell types and the NP-targetingability. rMCs, which had the highest ACE activity (SI Appendix,Figs. S2 and S3) also had the highest particle uptake. This dem-onstrates that even better targeting specificity can be achievedwhen ACE expression is increased, making these NPs a promisingtool for the treatment of diseases that present with overexpressedACE levels, as in the case of mesangial cells in diabetic ne-phropathy (34, 35) or Alzheimer’s disease (36).In a final set of experiments, we investigated whether our

virus-mimetic NPs would be able to distinguish between targetand off-target cells when presented with both of them simulta-neously. To this end, we cocultured rMCs together with AT1R-positive but ACE-negative HeLa or NCI-295R cells. To betterdistinguish between cell types, we marked rMCs fluorescently withCellTracker Green (rMCs-CTG). For CLSM images, off-target

cells were marked with a different fluorescent label, and all cellnuclei were stained for better visualization. After incubating thecocultures with virus-mimetic NP210Ang-I, we examined NP uptakeusing CLSM (Fig. 6A) and flow cytometry (Fig. 6 B and C). In bothexperiments, we observed remarkable selectivity. CLSM imagesrevealed that NP-associated red fluorescence was strongly colo-calized with the green fluorescence of target rMCs, but not with off-target HeLa or NCI-H295R cells. Flow cytometry confirmed themicroscopy results. We found that NPs were not only taken up insignificantly higher amounts by rMCs, but this specificity wasachieved in an environment with 10-fold more off-target cells thantarget cells. Comparable results were obtained for two different off-target cell types. While both carry only the secondary target mole-cule responsible for NP uptake, NCI-H295R cells display over 100-fold higher receptor expression than HeLa cells, as shown previouslyby our group (37), proving that AT1R expression on off-target cellsis not of concern for the exceptional specificity of virus-mimetic NPs.

DiscussionIn recent years, the investigation into the concept of viral-mimetic, enzyme-responsive NPs has been limited. In most cases,extracellular enzymes such as matrix metalloproteinases (MMPs)(38–41) or proteases (42) were targeted to unveil active NPs thatcould then interact with tumor cells. Surprisingly, virus-mimeticectoenzymatic activation following receptor-mediated endocy-tosis has never been explored. Our work shows that it is possibleto design NPs that interact with target cells in a manner similarto the influenza A virus, using a sequential, interactive two-stageprocess. Ang-I–decorated copolymer NPs made the initial target-cell contact by binding to ACE via the proligand Ang-I (Fig. 3).As a result of this primary binding process, the NPs were enzy-matically activated to unveil Ang-II, the secondary ligand. Ang-IIprompts the second stage of binding in which the AT1R is thetarget receptor (Fig. 2). NP binding triggers cell uptake byreceptor-mediated endocytosis (Fig. 4). Our study of the in-teraction of such NPs with cells carrying only one of the twotarget structures showed that the presence of both receptors is asine qua non for cell uptake that significantly increases target-cell specificity. Moreover, we found a correlation between pri-mary target (ACE) expression and cell uptake. rMCs, whichexpress higher levels of ACE than HK-2 cells, took up signifi-cantly more NPs (Fig. 5). This makes such materials a promisingtool for the treatment of diseases in which an enzyme is over-expressed, as in the case of mesangial cells during diabetic ne-phropathy (35). In this instance, it is beneficial that NPs have

Fig. 3. Enzymatic activation of NPs by soluble ACE. Incubation of NP210Ang-Ior NP510Ang-I with 0.1 μM rabbit lung ACE transformed the NPs to AT1R ac-tivating NPAng-II, as shown by the increase in intracellular calcium concen-tration (yellow bar). Ang-I served as a positive control. In the absence of ACE,neither Ang-I decorated NPs nor Ang-I could activate the AT1R (gray bar)similarly to the ligand-free NPMeO. The addition of captopril (Cap) inhibitedthe Ang-II formation (green bar), and the AT1R antagonist valsartan (Val)blocked the receptor directly (orange bar). Results represent mean ± SD (n = 3measurements, levels of statistical significance are indicated as ***P ≤ 0.001and ****P ≤ 0.0001 compared with ACE-treated samples). n.s., non-significant.

A B

Fig. 2. AT1R interaction with angiotensin-labeled NPs. (A) Interaction ofsecondary ligand-modified NPs (Ang-II) with the secondary target (AT1R) onrMCs as measured by intracellular calcium levels. A total of 50 nm NP210Ang-IIand 80 nm NP510Ang-II particles bind to the AT1R of target cells withsimilar avidities of 53.7 ± 6.4 pM and 30.7 ± 3.8 pM, respectively. The ligand-free control particles (NPMeO) and Ang-I–modified particles (NP510Ang-I) didnot interact with the receptor. (B) AT1R stimulation by angiotensin peptidesas measured by calcium mobilization assay in rMCs. Increasing concentra-tions of Ang-II produce increasing intracellular calcium concentrations. Onlyat the highest Ang-I concentrations applied (10 and 30 μM) could low levelsof intracellular calcium be measured, which is a result of Ang-I activation byACE on the cell membrane during the experiments (SI Appendix, Fig. S4).Results represent mean ± SD (n = 3).

Maslanka Figueroa et al. PNAS | May 14, 2019 | vol. 116 | no. 20 | 9833

APP

LIED

BIOLO

GICAL

SCIENCE

S

Dow

nloa

ded

by g

uest

on

Dec

embe

r 6,

202

0

been reported by multiple groups to enter the mesangium, whichis a prerequisite for a nanotherapeutic intervention (43).In cell cocultures, particles were able to distinguish target cells

that carried both the primary and secondary targets (rMCs ofHK-2) from off-target cells that carried only the secondary target(HeLa and NCI-H295R cells) (Fig. 6). Our particles exclusivelybound to target cells. This sheds light on the mechanism of theNP–cell interaction. Particles that had been enzymatically acti-vated by ACE and, therefore carried Ang-II in their corona, didnot bind to neighboring off-target cells despite their 10-foldhigher prevalence. NPs seemed to remain bound to the cellsthat they initially made contact with via their primary target li-gand Ang-I. The low Km value for the enzymatic conversion ofAng-I by ACE, reflecting the high affinity of the NP-boundsubstrate/enzyme interaction, can provide some explanation forthis phenomenon (SI Appendix, Fig. S5). It is reasonable to as-sume that the particles are subject to multivalent interactionssuch that several Ang-I molecules on one NP interact simulta-neously with several cell membrane-bound ACE molecules. Thekinetics of these parallel interactions significantly hinder thedissociation of NPs from the cell surface while continuously in-creasing the number of Ang-II molecules in the particle corona.This increases the density of Ang-II in the cleft between particleand cell surface rapidly enough to bind neighboring AT1Rs.Alternatively, the particles, since they are completely substratecovered, could undergo a rotation on the cell surface until AT1Rbinding. In both cases, the continuous creation of secondary li-gand increasing the avidity for the secondary receptor to EC50values in the 30–50 pM range (for purely Ang-II covered NPs)(Fig. 2) could explain why particles cannot move freely and bindto neighboring off-target cells. Overall, our data show that de-spite the clean-cut design principles that we used for translating

the natural mechanism of influenza A virus–cell interaction tosynthetic, therapeutic NPs, the detailed mechanism of NP–cellinteraction will need to be further elucidated in future studies.A critical, forward-looking question is whether our particles are

limited to the model enzyme receptor combination we chose. Alook at the literature reveals that there are a plethora of well-known ectoenzymes (44, 45). These could be used to cleavepeptide sequences on NPs masking not only natural peptidic butalso nonpeptidic synthetic receptor ligands, which have enhancedstability in biological environments and concomitantly, a high

Fig. 5. Target-cell specificity of virus-mimetic NPs investigated in differentcell lines using flow cytometry. Virus-mimetic NPs were internalized in ACE-and AT1R-positive rMCs and HK-2 cells serving as target cells. Only moderateuptake comparable to the negative control particles (NPMeO) was detectedin NCI-H295R cells, which served as off-target cells and carried only thesecondary receptor. There was a correlation between the cellular ACE ac-tivity and cell binding/uptake, as rMCs have the highest enzymatic activity.Results represent mean ± SD (n = 3, levels of statistical significance are in-dicated as ***P ≤ 0.001 compared with untargeted NPMeO and #P ≤ 0.01compared with untreated cells). AFU, arbitrary fluorescence units.

NPMeO NP Ang-I + Cap + Val + Cap/Val

NPMeO NP Ang-I + Cap + V

NPMeO NP Ang-I + Cap + Val + Cap/Val

al + Cap/Val

NPMeO NP Ang-I + Cap + Val + Cap/Val

A E

B

C

D

F

G

H

Fig. 4. Sequential two-stage interaction by virus-mimetic NPs with primary and secondary receptor-positive rMCs and HK-2 cells as shown by flowcytometry (A–D) and CLSM (E–H). (A) NP510Ang-I and(B) NP210Ang-I were internalized by both cell typescompared with the negative control (NPMeO). NPuptake was shown to be size dependent. Incubationof (C) rMCs and (D) HK-2 cells with either NP210Ang-Ior NP510Ang-I showed that the smaller NPs weretaken up in significantly higher amounts by both celllines. Results represent mean ± SD (n = 3, levels ofstatistical significance are indicated as **P ≤ 0.01,***P ≤ 0.001, and ****P ≤ 0.0001). CLSM imageswere taken after incubating (E and F) rMCs and (Gand H) HK-2 cells with (F and H) NP210Ang-I and (Eand G) NP510Ang-I. Fluorescence indicating NP up-take can be seen in cells incubated with NPAng-I,compared with the weak uptake in cells receivingnontargeted NPMeO. In both experiments, NP up-take could be inhibited by 1 mM captopril and/orvalsartan. (Scale bar: 10 μm.)

9834 | www.pnas.org/cgi/doi/10.1073/pnas.1902563116 Maslanka Figueroa et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 6,

202

0

affinity for a secondary cell surface target of choice. This wouldincrease the number of potential target cells, besides mesangialcells, for our model particles, expanding the applicability and ver-satility of our concept significantly.

Materials and MethodsMaterials. Lys–Ang-I and Lys–Ang-II (purity > 98%) were purchased fromGenscript. Ang-I and Ang-II were purchased from Bachem. Valsartan waspurchased from Santa Cruz Biotechnology. O-aminobenzoicacid-phenylalanyl-arginine-lysine (2,4-dinitrophenyl)-proline [Abz-FRK(Dnp)-P] was purchasedfrom Enzo Life Sciences. Carboxylic acid-terminated PEG was obtained fromJenKem Technology USA, Inc. All other chemicals were obtained from Sigma-Aldrich in analytical grade. Ultrapure water, which was obtained from a Milli-Qwater purification system (Millipore), was used for NP and buffer prepara-tion. The buffers used for NP modification were 0.1 M 2-(N-morpholino)ethanesulfonic acid (MES) buffer at pH 6, and Dulbecco’s PBS (DPBS) at pH 7.4,containing 1.5 mM KH2PO4, 2.7 mM KCl, 8 mM Na2HPO4, and 138 mM NaCl.The assay buffer used for ACE activity quantification was a 0.1 M Tris bufferwith 50 mM NaCl and 10 μM ZnCl2 at pH 7.

Cell Culture. rMCs were a kind gift from Armin Buschauer, University ofRegensburg, Regensburg, Germany. NCI-H295R and HeLa cells were pur-chased from ATCC (CRL-2128 and CCL-2, respectively). All three cell lines werecultured in RPMI 1640 medium containing 10% FBS (Biowest), supplementedwith insulin-transferrin-selenium (Life Technologies) and 100 nM hydrocor-tisone. HK-2 cells were purchased from ATCC (CRL-2190) and maintained inDMEM-F12 (1:1) medium supplemented with 10% FBS.

Polymer Synthesis and NP Preparation. To synthesize block copolymers withdifferent functional groups at the end of the PEG chain, we used commer-cially obtained carboxylic acid- ormethoxy-terminated PEG asmacroinitiatorsfor ring-opening polymerization of cyclic lactide using 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) as a catalyst after Qian et al. (23) with slight modifications,as previously described by our group (46). Resulting block copolymers were a10-kDa PLA with a carboxylic acid-terminated 2-kDa or 5-kDa PEG chain or amethoxy-terminated 5-kDa PEG chain. The respective 1H-NMR spectra areshown in SI Appendix, Fig. S6.

To detect NPs in vitro, the carboxylic acid-terminated PLGA (13.4 kDa) inthehydrophobic corewas labeledwith twodifferent fluorescent dyes (47), CF647-amine (for FACS) and tetramethylrhodamine (TAMRA)-amine (for CLSM) before

NP preparation using 4-(4,6-dimethoxy-1,3,5-triazin-2-yl)-4-methylmorpholiniumchloride (DMTMM) as a coupling agent, as previously reported by our group(29). Briefly, amine functionalized CF647 or TAMRA dyes were coupled to car-boxylic acid-terminated PLGA using DMTMM as a linker. PLGA was used in afivefold excess so that not all of the polymer was labeled. The fluorescent PLGAwas purified from noncoupled dye through dialysis over 24 h.

For NP preparation, PEG-PLA block copolymers were mixed with 13.4 kDaPLGA in acetonitrile at a set 70/30 mass ratio and a final concentration of10 mg/mL. Polymer solutions were used to manufacture NPs via bulk pre-cipitation by adding them dropwise under stirring water to a final concen-tration of 1 mg/mL. NP formulations were agitated for 2 h until the organicsolvent was completely evaporated and then concentrated by ultracentri-fugation using a 100-kDa molecular weight cutoff Microsep advance cen-trifugal device (Pall Corporation) at 959 × g for 15 min.

NP-COOH was modified with Lys–Ang-I or Lys–Ang-II using 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide (EDC) and N-hydroxysuccinimide (NHS)chemistry. NPs (0.3 μmol PEG) were activated with EDC (10 mM) in presenceof NHS (80 mM) in MES buffer (100 mM, pH 6). After 30 min of activation, 2-mercaptoethanol was added in excess (45 μmol) to quench the reaction overa period of 15 min. Subsequently, either Lys–Ang-I or Lys–Ang-II was addedto the activated NPs (0.45 μmol), and the pH was raised by the addition ofDPBS (pH 7.4). After 2 h of gentle stirring, angiotensin-modified NPs werepurified by dialysis against 4 L water using a 6- to 8-kDa molecular weightcutoff regenerated cellulose dialysis membrane (Spectrum Laboratories, Inc)over 18 h (with medium change after 0.5, 1, 2, 3, 4, and 6 h), followed byultracentrifugation using a 100-kDa molecular weight cutoff Microsep ad-vance centrifugal device (Pall corporation) at 959 × g for 15 min.

NP Characterization. NP size was measured using aMalvern ZetaSizer Nano ZS(Malvern Instruments, GmbH) with a 633 He-Ne laser at a 173° backscatterangle, using a microcuvette. The electrophoretic mobility (ξ-potential) of theNPs was measured with a Malvern ZetaSizer Nano ZS using a folded capillarycell. Both measurements were conducted at 25 °C in 10% DPBS.

The PEG concentration of the NP solutions was determined using a col-orimetric iodine complexing assay adapted from Childs (48), as previouslydescribed (29). The absorbance was measured using a FLUOstar Omegamicroplate reader (BMG Labtech). Angiotensin peptide concentration wasdetermined fluorometrically using an LS-5S fluorescence plate reader (Per-kinElmer) bymeasuring the arginine content using 9,10-phenanthrenequinone(29, 49).

Intracellular Calcium Mobilization Assay. The affinity of NP-bound Ang-II to-ward the AT1R was investigated using a ratiometric Fura-2 Ca2+ chelatormethod as previously described by our group (7). Samples used in the assaywere Ang-II dilutions to test the functionality of the AT1R on the tested cells(SI Appendix, Fig. S2B) and NPAng-II to investigate the affinity of NP-boundangiotensin-II toward the AT1R (Fig. 2). In addition, NPAng-I enzymaticallyactivated to NPAng-II by a soluble form of ACE was used to confirm theoccurrence of enzymatic processing (Fig. 3). A stock solution of NPAng-I witha concentration of 10 μM Ang-I was incubated with 0.1 μM rabbit lung ACE(Sigma-Aldrich) for 2 h at 37 °C. To inhibit the conversion of NPAng-I toNPAng-II by the soluble enzyme form, samples were also incubated in thepresence of 20 μM captopril. To also inhibit ACE on the membrane of rMCsduring the measurements, the cells were additionally preincubated with5 mM captopril in PBS for 15 min. As a control, unmodified NPMeO wasused. Inhibition of the intracellular calcium mobilization of activated NPswas performed by the addition of 20 μM valsartan after the incubationperiod. NPs were then used as agonist samples in an intracellular calciummobilization assay, as described above. A Student’s t test was performedusing GraphPad Prism to assess statistical significance.

Flow Cytometry. To investigate the sequential two-stage interaction of virus-mimetic NPs with target cells, rMCs and HK-2 cells were seeded into 24-wellplates at densities of 30,000 and 50,000 cells per well, respectively, and in-cubated for 48 h. NP solutionswere prepared at a concentration of 0.7mgNP/mLin Leibovitz medium (LM) supplemented with 0.1% BSA. After washingthe cells with DPBS, the prewarmed NPs were pipetted onto them and in-cubated for 45 min at 37 °C. To confirm uptake specificity, cells were in-cubated with 1 mM captopril and/or valsartan for 15 min before NP addition.To assess the target-cell specificity of the NPs, rMCs stained with 10 μM CTGdye (Thermo Fisher Scientific) for 30 min at 37 °C in serum-free RPMI 1640were seeded in 24-well plates in coculture together with NCI-H295R or HeLacells at densities of 10,000 and 75,000, respectively, and incubated for 48 h.NP solutions at 0.02 mg/mL in LM with 0.1% BSA were subsequently addedand left on the cells for 45 min at 37 °C. In both experiments, the NP solutions

Hoechst CTG CTDR NP Ang-I Merged

aLeH/

CMr

R592HI

CN/

CMr

B

A

C

Fig. 6. Target-cell specificity of virus-mimetic NPs when presented to co-cultures of target- and off-target cells. (A) CLSM images of green-stainedfluorescent rMCs (green) and deep red-stained off-target HeLa or NCI-H295Rcells (white). Cells were additionally stained with Hoechst 33258 for nucleivisualization. NP210Ang-I–associated red fluorescence could only be detectedin rMCs, demonstrating the high selectivity of virus-mimetic NPs. Cocultureflow cytometry analysis of (B) rMCs with HeLa cells or (C) rMCs with NCI-H295R in the presence or absence of virus-mimetic NPs corroborated theresults. Results represent mean ± SD (n = 3 measurements, levels of statisticalsignificance are indicated as ***P ≤ 0.001 and ****P ≤ 0.0001 comparedwith rMCs-CTG). AFU, arbitrary fluorescence units. (Scale bar: 20 μm.)

Maslanka Figueroa et al. PNAS | May 14, 2019 | vol. 116 | no. 20 | 9835

APP

LIED

BIOLO

GICAL

SCIENCE

S

Dow

nloa

ded

by g

uest

on

Dec

embe

r 6,

202

0

were discarded after the incubation time and the cells washed thoroughly withDPBS, trypsinized, and centrifuged (2×, 200 ×g, 5 min, 4 °C). Finally, the cells wereresuspended in DPBS, and their fluorescence was analyzed with a FACS Caliburflow cytometer (Becton Dickinson). NP fluorescence was excited at 633 nm andthe emission was recorded using a 661/16 nm bandpass filter. rMC fluorescencewas excited at 488 nm and recorded using a 530/30 bandpass filter. FACS resultswere analyzed using Flowing software 2.5.1 (Turku Centre for Biotechnology).The population of viable cells and, in the coculture experiments, the stained rMCsand nonfluorescent HeLa or NCI-H295R cells were gated. The geometric mean ofthe NP-associated fluorescence was analyzed. A Student’s t test was performedusing GraphPad Prism to assess statistical significance.

CLSM. To investigate the sequential two-stage interaction of virus-mimeticNPs with target cells, rMCs and HK-2 cells were seeded into eight-wellμ-slides (Ibidi) at 10,000 cells per well and incubated for 24 h. Cells werewashed with DPBS and incubated for 45 min at 37 °C with prewarmed NPsolutions at concentrations of 0.7 mg/mL in LM supplemented with 0.1%BSA. For uptake inhibition, cells were incubated 15 min before NP additionwith 1 mM captopril and/or valsartan. For NP target-cell specificity experiments,

CTG-stained rMCs (10 μM, 30 min, 37 °C) were seeded in coculture togetherwith CellTracker deep red-stained HeLa or NCI-H295R cells (25 μM, 30min, 37 ºC)in eight-well μ-slides at densities of 2,000 and 10,000 cells per well, respectively,and incubated for 24 h. Cells were washed with DPBS and incubated for 45 minwith 0.2 mg/mL NPs in LM with 0.1% BSA. After the incubation period, the NPsolutions were discarded, and the cells were washed thoroughly with DPBS. LMwas added before viewing the cells under a microscope. Images of live cells wereacquired with a Zeiss Axiovert 200 microscope with a LSM 510 laser-scanningdevice using a 63× Plan-Apochromat (N.A. 1.4) objective, and AIM 4.2 software(Zeiss). NP-associated fluorescence was excited with a 543-nm He-Ne laser andrecorded with a 560–615 bandpass filter. rMC and HeLa or NCI-H295R fluores-cence was excited with a 488 argon or a 633 nm He/Ne laser and recorded witha 560–615 bandpass and a 650 longpass filter, respectively. The focal plane wasset at 1.1 μm.

ACKNOWLEDGMENTS.We thank Renate Liebl, Viktoria Messmann, and Uwede Vries for their excellent technical assistance and Prof. Dr. Ralph Witzgallfor enabling CLSM imaging. Financial support was provided by the GermanResearch Foundation (DFG), Grant GO 565/17-3.

1. Allen TM, Cullis PR (2004) Drug delivery systems: Entering the mainstream. Science303:1818–1822.

2. Mout R, Moyano DF, Rana S, Rotello VM (2012) Surface functionalization of nano-particles for nanomedicine. Chem Soc Rev 41:2539–2544.

3. Wilhelm S, et al. (2016) Analysis of nanoparticle delivery to tumours. Nat Rev Mater 1:16014.

4. Peer D, et al. (2007) Nanocarriers as an emerging platform for cancer therapy. NatNanotechnol 2:751–760.

5. Hild W, et al. (2010) G protein-coupled receptors function as logic gates for nano-particle binding and cell uptake. Proc Natl Acad Sci USA 107:10667–10672.

6. Mammen M, Choi SK, Whitesides GM (1998) Polyvalent interactions in biologicalsystems: Implications for design and use of multivalent ligands and inhibitors. AngewChem Int Ed Engl 37:2754–2794.

7. Hennig R, Pollinger K, Veser A, Breunig M, Goepferich A (2014) Nanoparticle multi-valency counterbalances the ligand affinity loss upon PEGylation. J Control Release194:20–27.

8. Fasting C, et al. (2012) Multivalency as a chemical organization and action principle.Angew Chem Int Ed Engl 51:10472–10498.

9. Silpe JE, et al. (2013) Avidity modulation of folate-targeted multivalent dendrimersfor evaluating biophysical models of cancer targeting nanoparticles. ACS Chem Biol 8:2063–2071.

10. Wang S, Dormidontova EE (2011) Nanoparticle targeting using multivalent ligands:Computer modeling. Soft Matter 7:4435.

11. Dreaden EC, et al. (2012) Antiandrogen gold nanoparticles dual-target and overcometreatment resistance in hormone-insensitive prostate cancer cells. Bioconjug Chem 23:1507–1512.

12. Hennig R, Veser A, Kirchhof S, Goepferich A (2015) Branched polymer-drug conju-gates for multivalent blockade of angiotensin II receptors. Mol Pharm 12:3292–3302.

13. Kluza E, et al. (2010) Synergistic targeting of alphavbeta3 integrin and galectin-1 withheteromultivalent paramagnetic liposomes for combined MR imaging and treatmentof angiogenesis. Nano Lett 10:52–58.

14. Cheng Z, Al Zaki A, Hui JZ, Muzykantov VR, Tsourkas A (2012) Multifunctionalnanoparticles: Cost versus benefit of adding targeting and imaging capabilities.Science 338:903–910.

15. Modery-Pawlowski CL, Gupta AS (2014) Heteromultivalent ligand-decoration for ac-tively targeted nanomedicine. Biomaterials 35:2568–2579.

16. Marsh M, Helenius A (2006) Virus entry: Open sesame. Cell 124:729–740.17. Boulant S, Stanifer M, Lozach P-Y (2015) Dynamics of virus-receptor interactions in

virus binding, signaling, and endocytosis. Viruses 7:2794–2815.18. Grove J, Marsh M (2011) The cell biology of receptor-mediated virus entry. J Cell Biol

195:1071–1082.19. Peitsch C, Klenk H-D, Garten W, Böttcher-Friebertshäuser E (2014) Activation of in-

fluenza A viruses by host proteases from swine airway epithelium. J Virol 88:282–291.20. Qian Y, Feldman E, Pennathur S, Kretzler M, Brosius FC, 3rd (2008) From fibrosis to

sclerosis: Mechanisms of glomerulosclerosis in diabetic nephropathy. Diabetes 57:1439–1445.

21. Bkaily G, et al. (2003) Angiotensin II AT1 receptor internalization, translocation and denovo synthesis modulate cytosolic and nuclear calcium in human vascular smoothmuscle cells. Can J Physiol Pharmacol 81:274–287.

22. Rice GI, Thomas DA, Grant PJ, Turner AJ, Hooper NM (2004) Evaluation ofangiotensin-converting enzyme (ACE), its homologue ACE2 and neprilysin in angio-tensin peptide metabolism. Biochem J 383:45–51.

23. Qian H, Wohl AR, Crow JT, Macosko CW, Hoye TR (2011) A strategy for control of“random” copolymerization of lactide and glycolide: Application to synthesis of PEG-b-PLGA block polymers having narrow dispersity. Macromolecules 44:7132–7140.

24. Owen SC, Chan DPY, Shoichet MS (2012) Polymeric micelle stability. Nano Today 7:53–65.

25. Xiao RZ, et al. (2010) Recent advances in PEG-PLA block copolymer nanoparticles. Int JNanomedicine 5:1057–1065.

26. Makadia HK, Siegel SJ (2011) Poly lactic-co-glycolic acid (PLGA) as biodegradablecontrolled drug delivery carrier. Polymers (Basel) 3:1377–1397.

27. Yang ZL, Li XR, Yang KW, Liu Y (2008) Amphotericin B-loaded poly(ethylene glycol)-poly(lactide) micelles: Preparation, freeze-drying, and in vitro release. J Biomed MaterRes A 85:539–546.

28. Jiang W, Kim BYS, Rutka JT, Chan WCW (2008) Nanoparticle-mediated cellular re-sponse is size-dependent. Nat Nanotechnol 3:145–150.

29. Abstiens K, Gregoritza M, Goepferich AM (December 6, 2018) Ligand density andlinker length are critical factors for multivalent nanoparticle–Receptor interactions.ACS Appl Mater Interfaces, 10.1021/acsami.8b18843.

30. de Gasparo M, Catt KJ, Inagami T, Wright JW, Unger T (2000) International union ofpharmacology. XXIII. The angiotensin II receptors. Pharmacol Rev 52:415–472.

31. de Gasparo M, Whitebread S (1995) Binding of valsartan to mammalian angiotensinAT1 receptors. Regul Pept 59:303–311.

32. Bevilacqua M, et al. (1996) Affinity of angiotensin I-converting enzyme (ACE) inhib-itors for N- and C-binding sites of human ACE is different in heart, lung, arteries, andveins. J Cardiovasc Pharmacol 28:494–499.

33. Huang X, Teng X, Chen D, Tang F, He J (2010) The effect of the shape of mesoporoussilica nanoparticles on cellular uptake and cell function. Biomaterials 31:438–448.

34. Cristovam PC, et al. (2008) ACE-dependent and chymase-dependent angiotensin IIgeneration in normal and glucose-stimulated human mesangial cells. Exp Biol Med(Maywood) 233:1035–1043.

35. Vidotti DB, et al. (2004) High glucose concentration stimulates intracellular reninactivity and angiotensin II generation in rat mesangial cells. Am J Physiol Renal Physiol286:F1039–F1045, and erratum (2006) 290:F1277.

36. Ohrui T, et al. (2004) Effects of brain-penetrating ACE inhibitors on Alzheimer diseaseprogression. Neurology 63:1324–1325.

37. Hennig R, et al. (2015) Multivalent nanoparticles bind the retinal and choroidal vas-culature. J Control Release 220:265–274.

38. Callmann CE, et al. (2015) Therapeutic enzyme-responsive nanoparticles for targeteddelivery and accumulation in tumors. Adv Mater 27:4611–4615.

39. Nguyen MM, et al. (2015) Enzyme-responsive nanoparticles for targeted accumula-tion and prolonged retention in heart tissue after myocardial infarction. Adv Mater27:5547–5552.

40. Han H, et al. (2017) Dual enzymatic reaction-assisted gemcitabine delivery systems forprogrammed pancreatic cancer therapy. ACS Nano 11:1281–1291.

41. Guo F, et al. (2018) PEGylated self-assembled enzyme-responsive nanoparticles foreffective targeted therapy against lung tumors. J Nanobiotechnology 16:57.

42. Adamiak L, Touve MA, LeGuyader CLM, Gianneschi NC (2017) Peptide brush polymersand nanoparticles with enzyme-regulated structure and charge for inducing orevading macrophage cell uptake. ACS Nano 11:9877–9888.

43. Choi CH, Zuckerman JE, Webster P, Davis ME (2011) Targeting kidney mesangium bynanoparticles of defined size. Proc Natl Acad Sci USA 108:6656–6661.

44. Alexander SP, et al.; CGTP Collaborators (2015) The concise guide to PHARMACOL-OGY 2015/16: Enzymes. Br J Pharmacol 172:6024–6109.

45. Kreutzberg GW, Reddington M, Zimmermann H (1986) Cellular biology of ectoen-zymes. Proceedings of the International Erwin-Riesch-Symposium on EctoenzymesMay 1984 (Springer, Berlin).

46. Abstiens K, Maslanka Figueroa S, Gregoritza M, Goepferich AM (2019) Interaction offunctionalized nanoparticles with serum proteins and its impact on colloidal stabilityand cargo leaching. Soft Matter 15:709–720.

47. Valencia PM, et al. (2011) Effects of ligands with different water solubilities on self-assembly and properties of targeted nanoparticles. Biomaterials 32:6226–6233.

48. Childs CE (1975) The determination of polyethylene glycol in gamma globulin solu-tions. Microchem J 20:190–192.

49. Smith RE, MacQuarrie R (1978) A sensitive fluorometric method for the determinationof arginine using 9,10-phenanthrenequinone. Anal Biochem 90:246–255.

9836 | www.pnas.org/cgi/doi/10.1073/pnas.1902563116 Maslanka Figueroa et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 6,

202

0