Design and synthesis of theranostic antibioticnanodrugs that display enhanced antibacterialactivity and luminescenceSheng Xiea, Sesha Manuguria, Giampiero Proiettia, Joakim Romsona, Ying Fub, A. Ken Ingec, Bin Wud, Yang Zhanga,Daniel Hälla, Olof Ramströma,1, and Mingdi Yana,d,1

aDepartment of Chemistry, KTH Royal Institute of Technology, Teknikringen 30, S-10044, Stockholm, Sweden; bDepartment of Applied Physics, Science forLife Laboratory, KTH Royal Institute of Technology, SE-171 21, Solna, Sweden; cDepartment of Materials and Environmental Chemistry, StockholmUniversity, S-10691, Stockholm, Sweden; and dDepartment of Chemistry, University of Massachusetts, Lowell, MA 01854

Edited by Song Jin, University of Wisconsin-Madison, Madison, WI, and accepted by Editorial Board Member Richard Eisenberg June 15, 2017 (received forreview June 1, 2017)

We report the modular formulation of ciprofloxacin-based puretheranostic nanodrugs that display enhanced antibacterial activi-ties, as well as aggregation-induced emission (AIE) enhancementthat was successfully used to image bacteria. The drug derivatives,consisting of ciprofloxacin, a perfluoroaryl ring, and a phenyl ringlinked by an amidine bond, were efficiently synthesized by astraightforward protocol from a perfluoroaryl azide, ciprofloxacin,and an aldehyde in acetone at room temperature. These com-pounds are propeller-shaped, and upon precipitation into water,readily assembled into stable nanoaggregates that transformedciprofloxacin derivatives into AIE-active luminogens. The nano-aggregates displayed increased luminescence and were success-fully used to image bacteria. In addition, these nanodrugs showedenhanced antibacterial activities, lowering the minimum inhibitoryconcentration (MIC) by more than one order of magnitude againstboth sensitive and resistant Escherichia coli. The study represents astrategy in the design and development of pure theranostic nano-drugs for combating drug-resistant bacterial infections.

nanodrugs | aggregation-induced emission | fluoroquinolones

The need for new antimicrobial therapeutics is becoming in-creasingly urgent due to the rapid emergence and spread of

antibiotic-resistant pathogens (1, 2). A promising strategy againstdrug-resistant bacteria is to revert drug resistance mechanisms, withthe potential to rejuvenate large classes of previously powerful an-tibiotics (3). Nanotherapeutics have in this context emerged as a newplatform in drug development (4). Compared with small-moleculedrugs, which often show poor pharmacokinetic profiles and broadsecondary effects, nanotherapeutics have been shown to achievehigher therapeutic efficacy and lower off-target toxicity by alteringthe biodistribution of drugs. The increased uptake into target cellsleads to higher intracellular drug concentrations, higher therapeuticefficacy, and lower systemic toxicity. In addition, nanotherapeuticscan overcome multidrug resistance by disarming the bacteria’s effluxpumps which are unable to clear the nanoparticles (5).A relatively new form of nanotherapeutics is pure nanodrugs

(PNDs), i.e., nanoparticles that consist of pure drug molecules (6).A key difference between PND and other nanotherapeutics is thedrug-loading efficiency. Whereas PNDs are composed entirely ofthe drug molecule, the drug-loading efficiency for nanotherapeuticsusing the traditional drug carriers is significantly lower because thedrug carrier is usually much larger than the small drug molecules(7). The carrier may furthermore be more sensitive to the envi-ronment, and its subsequent degradation products can pose po-tential negative impacts (8). Because PNDs are made entirely ofdrug molecules, they are carrier-free and can be released as the freeform of the drug in a manner that results in increased activities.Examples of PNDs, primarily prepared from cancer drugs, havedemonstrated enhanced antitumor efficacy due to their improvedpermeability and retention, enhanced bioavailability, and improved

stability against enzymatic degradation (9–11). Furthermore, be-cause lower doses are used, side effects are decreased.PNDs can be straightforwardly prepared by reprecipitation (12).

Although the method is in principle applicable to many drug sub-stances, especially those having limited water solubility, some criticalissues exist. Reprecipitation may not give nanoaggregates, but ratherlong fibers, flakes, or even large crystals. Precise control of thenanoaggregate size to below 100 nm is difficult, and suffers frombatch-to-batch variations. In addition, the stability of nanoaggregatesis usually poor without stabilization. These challenges are associatedwith the very complicated assembly process of drug–solvent anddrug–drug interactions which are difficult to control (13). Therefore,only a few empirical protocols have been reported so far, for ex-ample, covalent dimerization (9), installation of bulky hydrophobictails (10), and use of polymeric backbones (14).These challenges can potentially be addressed by applying strate-

gies that promote stable nanoaggregate formation. One such tech-nique is aggregation-induced emission (AIE). During the last decade,AIE has been applied to various organic compounds for the devel-opment of new luminescent materials for optoelectronic devices,sensors, biomedical imaging, and smart materials (15). Whereas theindividual AIE-active luminogens fluoresce poorly in solution, theemission increases markedly upon their aggregation in response torestricted molecular motions. Such nanosized aggregates are fre-quently obtained using propeller-shaped molecules (16), in contrastto planar entities that are prone to form crystalline seeds and furthergrow into bulk crystals (17, 18). Furthermore, the restriction of

Significance

The work described represents a strategy in the design and devel-opment of theranostic pure nanodrugs for combating drug-resistantbacteria. The significance of this work includes: (i) a modular syn-thetic strategy for the preparation of ciprofloxacin derivatives inhigh yields in one step at room temperature without any catalysts;(ii) stable nanoaggregates, prepared following the principle ofaggregation-induced emission, and successfully used as fluorescentorganic dots to image bacteria; and (iii) formulation of nanodrugsthat showed aggregation-enhanced antibacterial activities.

Author contributions: O.R. and M.Y. designed research; S.X., S.M., G.P., J.R., B.W., Y.Z.,and D.H. performed research; Y.F. and A.K.I. contributed new reagents/analytic tools;S.X., G.P., J.R., Y.F., A.K.I., O.R., and M.Y. analyzed data; and S.X. and M.Y. wrotethe paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission. S.J. is a guest editor invited by the Editorial Board.

Data deposition: The atomic coordinates and structure factors have been deposited in theCambridge Structural Database, Cambridge Crystallographic Data Centre (CCDC) (acces-sion codes CCDC 1426170, CCDC 1553081, and CCDC 1553082).1To whom correspondence may be addressed. Email: ramstrom@kth.se or mingdi@kth.se.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1708556114/-/DCSupplemental.

8464–8469 | PNAS | August 8, 2017 | vol. 114 | no. 32 www.pnas.org/cgi/doi/10.1073/pnas.1708556114

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 5,

202

0

intramolecular motions in AIE often involves supramolecular inter-actions, which result in fairly stable aggregates (19–22). Indeed,aqueous suspensions of AIE aggregates can be stable over severalmonths in the absence of surfactants (23).Encouraged by these observations, we designed a type of PNDs

based on AIE-active fluoroquinolones. Fluoroquinolones such asciprofloxacin are among the most prescribed antibiotics for thetreatment of various infections (24). Their broad-spectrum efficacy isassociated with the mechanism of action, which is the inhibition ofDNA replication and repair (25). The emergence of fluoroquinolone-resistant microorganisms has long been observed and is partly at-tributed to the overuse of quinolone antibiotics (26). Resistance tofluoroquinolones generally arises from mutations of target enzymesthat reduce the susceptibility to the drug, or by a reduction of theintracellular drug concentration (27). The latter is achieved throughlimiting drug influx, e.g., by expressing fewer porins, and/or throughdrug extrusion by expressing multidrug efflux pumps.Most fluoroquinolones are electron donor–acceptor com-

pounds with the piperazinyl group serving as the electron donorand 4-oxoquinoline-3-carboxylic acid as the electron acceptor inthe neutral and zwitterion forms. In organic solvents like aceto-nitrile, fluoroquinolones are weakly fluorescent, owing to the highnonradiative rate constants which deactivate the singlet excitedstate through intramolecular charge transfer (28). We hypothesizedthat by forming nanoaggregates, the intramolecular motion, a pro-cess that promotes nonradiative relaxation pathway, would be re-stricted (15). This would open up the radiative pathway, leading tothe enhanced fluorescence emission allowing the nanoaggregatesto be used as imaging agents. In this proof-of concept study, wedesigned and synthesized propeller-shaped structures consisting of afluoroquinolone such as ciprofloxacin, a perfluoroaryl group, anda phenyl group through a straightforward protocol. We show thatthese compounds can be prepared into antibiotic nanodrugs thatserved as both imaging agents and therapeutic drugs.

Results and DiscussionThe ciprofloxacin derivatives were synthesized from readily ac-cessible perfluoroaryl azides, phenylacetaldehyde, and cipro-floxacin by stirring them in acetone at room temperature (Fig. 1).

Phenylacetaldehyde and ciprofloxacin initially formed the enaminein situ, which then reacted with perfluoroaryl azides to give theamidine products. This unique reactivity of perfluoroaryl azides isdue to their highly electron-deficient nature, which allows theseelectrophilically activated azides to undergo cycloaddition withenamines at room temperature without any catalysts (29, 30),and also with other dipolarophiles and nucleophiles (31–33).Ciprofloxacin is sparingly soluble in common organic solvents.However, the ciprofloxacin conjugates showed much improvedsolubility in organic solvents and enabled efficient separation byflash column chromatography or recrystallization. This reaction isefficient and modular, allowing the introduction of the perfluoroaryland phenyl moieties to ciprofloxacin by a straightforward protocol.The derivatization of ciprofloxacin showed minimal impact on its

absorption, but caused large changes in the fluorescence emission. Inthe dissolved state, strong emission from derivatives 5–6 (Φf = 19–26%, SI Appendix, Table S1) was observed in dry THF anddichloromethane (DCM) compared with ciprofloxacin (Φf = 2–9%).These are among the strongest quinolone fluorophores reported (34,35). The electron-withdrawing amidine decreases the electron-donating capability of the outer piperazinyl nitrogen, similar to theeffect of acylation (28). This results in blocking of the photoinducedelectron-transfer process from the outer piperazinyl nitrogen to thefluoroquinolone ring, which is the most efficient energy-dissipationpathway of ciprofloxacin. In polar solvents, such as acetonitrile,methanol, and water, derivatives 5–7 emitted faintly. As the polarityof the solvent increased from toluene to acetonitrile, the emissionpeak gradually red-shifted from 400 to 460 nm. The red shift anddecrease in emission with the increase of solvent polarity (SI Ap-pendix, Figs. S1 and S2) is indicative of a twisted intramolecularcharge-transfer process upon photoexcitation from the donor(piperazinyl ring) to the acceptor (fluoroquinolone ring) (34, 35), aprocess that is stabilized and promoted by polar solvents, however,susceptible to nonradiative quenching (36–38).Compounds 5-6, 8–9, and 11 displayed behavior as AIE-active

luminogens (SI Appendix, Fig. S3). For example, compound 5showed weak fluorescence (Φf ∼ 1.0%) in acetonitrile with anemission maximum at 465 nm (Fig. 2A). Addition of water showedpolarity-induced quenching, and decreased the absorption at

Fig. 1. Modular synthesis of ciprofloxacin derivatives having perfluoroaryl and aryl moieties. (A) ciprofloxacin (1, 1.0 mmol), azide (2, 1.3–2.0 mmol), al-dehyde (3, 2.0 mmol), acetone (80–60 mL), room temperature, 1–3 d; (B) seven days, purified by recrystallization; (C) norfloxacin instead of ciprofloxacin wasused; (D) piperidine instead of ciprofloxacin was used.

Xie et al. PNAS | August 8, 2017 | vol. 114 | no. 32 | 8465

CHEM

ISTR

Y

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 5,

202

0

λmax = 280 nm (Fig. 2C). However, when the water content in-creased to 80 vol %, the emission started to increase together witha blue shift to 457 nm. The maximal fluorescence intensity wasobserved at 90 vol % (Φf = 12%), which was 12- and 30-foldstronger than its solution in pure acetonitrile or 10 vol % water/MeCN, respectively (Fig. 2B). In addition, broad tails appeared

and the absorption intensity increased with water due to Miescattering, indicating the formation of nanoaggregates which wasalso confirmed by scanning electron microscopy (SEM, Fig. 2D).These observations are consistent with many other AIE-activeluminogens having donor–acceptor structures (21, 22, 37, 39–41), where the aggregation hampers the intramolecular motion

Fig. 2. Aggregation-induced emission enhancementof ciprofloxacin derivative 5. (A) Fluorescence spectrain water/MeCN mixtures, excitation 334 nm. (B) Plotof relative emission intensity against water content inwater/MeCN mixtures. I0: emission intensity at 0 vol %water. (C) Absorption spectra in water/MeCN mix-tures. (D) SEM image of 5 prepared from 90 vol%water/MeCN. Concentration: 10 μM.

Fig. 3. Characterization of the nanoaggregates prepared by reprecipitation. TEM images of nanoaggregates prepared from compound 5 (A), 6 (B), 7 (C), and8 (D) in 99 vol % water/MeCN, and nanoaggregates from compound 5 (E and F) in 99.9 vol % water/MeCN. (A, Inset) Diffraction pattern. Concentration:10 μM. See SI Appendix for TEM images of other samples. (G) Storage stability of nanoaggregates 5 formed from 99 vol % water/MeCN in Millipore water at4 °C. PDI (●) and particle diameter (○) were measured by DLS.

8466 | www.pnas.org/cgi/doi/10.1073/pnas.1708556114 Xie et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 5,

202

0

between the donor and the acceptor, hence leading to enhancedfluorescent emission as well as a blue shift in the emission peak(15). Spherical nanoaggregates were observed at high water con-tent (≥99 vol %) (Fig. 3). These nanoaggregates were amorphousas confirmed by diffraction (Fig. 3A, Inset). The nanoaggregatescould be prepared reproducibly and displayed relatively homo-geneous morphology and emission properties. The higher thewater content, the smaller the nanoaggregates (SI Appendix, TableS2) (17). Screening of organic solvents showed that the size of thenanoaggregates decreased with increasing dielectric constant (SIAppendix, Fig. S4) (42). This provides a convenient way to controlthe size of the ciprofloxacin nanoaggregates. In the case of com-pound 5, it precipitated into 40–60-nm nanoaggregates in DMF,MeCN, acetone, and DMSO, which were much smaller than the90-nm nanoaggregates from THF [measured by dynamic lightscattering (DLS); SI Appendix, Table S2]. The nanosized nano-aggregates also exhibited high storage stability. For example,compound 5 (d ∼ 60 nm) showed no significant change in size over2 mo at 4 °C (Fig. 3G). Other derivatives were also evaluated (SIAppendix, Table S2), and spherical amorphous nanoaggregateswere similarly observed (Fig. 3 and SI Appendix, Fig. S5). These

nanoaggregates have negative zeta potential from −25 mV to−40 mV, likely due to the carboxylate groups on the quinolonering being exposed at the particle–solution interface (10).Crystal structures were collected to investigate the intermolecular

interactions in these compounds. Compound 5 adopts a more tightlypacked antiparallel dimer configuration (Fig. 4A). The penta-fluoroaryl and fluoroquinolone groups are partially in the face-to-face orientation, with the closest distance of 3.08 Å. The phenylgroups take a coplanar arrangement, with no additional interactionsidentified. The crystal structure of compound 5 prepared from THFadopts a similar dimer configuration as from acetonitrile, but showsa different amidine torsion angle in the amidine moiety (SI Appendix,Fig. S6). When replacing F at the p position on the perfluorophenylring in 5 with a methoxy (OMe) group, the resulting compound 6exhibits a more extended configuration (Fig. 4B), showing antipar-allel face-to-face stacking between the two pentafluoroaryl groups(∼3.4 Å) and the two fluoroquinolone groups (∼3.5 Å), andweaker interactions between the two phenyl rings (∼4.2 Å) (43).The intermolecular interactions efficiently restrict the intra-molecular motions in the nanoaggregates, contributing to the

Fig. 4. Crystal structures of compound 5 (A, CCDC 1553082) and compound 6 (B, CCDC 1426170). Crystals were grown by slow evaporation of an acetonitrilesolution. (Gray: C, red: O, yellow-green: F, blue: N. Hydrogen atoms are omitted for clarity.)

Fig. 5. Fluorescence microscopy images of E. coliORN 178 incubated with nanoaggregates of com-pound 5. (A) Merged micrograph of bright-fieldimage and fluorescence image (excitation: 405 nm,power intensity: 0.5%). A 410–463-nm filter (see theshadow band II in G and J) was used to excludeautofluorescence from the bacteria. (B) Closer viewshowing some unattached nanoparticles and bac-teria. (C and D) Enlarged views showing nano-particles attached to the bacterial cells. (E) Bright-field image of one E. coli bacterium with a nano-aggregate attached. (F) Fluorescence lambda-modemicrograph of the bacterium in E acquired 2 minlater (405 nm, power intensity: 0.5%). (G) Fluores-cence spectra of two ROIs on the bacterium,ROI1 and ROI2 shown in F. The fluorescence signalsfrom ROI1 are dominated by the nanoaggregateemission. The autofluorescence from the bacteria(480–550 nm) existed in both ROI1 and ROI2,showing as weak shoulders marked by dashed redlines in G. (H) Bright-field micrograph of E. coli ORN178. (I) Fluorescence lambda-mode image under theexcitation of laser (405 nm, power intensity 2.8%)showing autofluorescence from bacteria. (J) Nor-malized average autofluorescence intensity of all E.coli cells in the observation view. Note that the bluecolor in A–D is pseudo based on F, whereas the colorin F and I is the true color.

Xie et al. PNAS | August 8, 2017 | vol. 114 | no. 32 | 8467

CHEM

ISTR

Y

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 5,

202

0

enhanced emission in the nanoaggregates by slowing down theintramolecular charge transfer.Among all propeller-type ciprofloxacin derivatives, compound 5,

having a pentafluorophenyl group, yielded spherical nanoaggregateswith the smallest size, high stability, and strong AIE enhancement (SIAppendix, Table S2). Compound 11, a norfloxacin derivative, alsoyielded nanoaggregates [d ∼ 64 nm, polydispersity index (PDI) =0.21] comparable to compound 5 (d ∼ 60 nm, PDI = 0.24). Com-pound 12, lacking the quinolone structure, could form nano-aggregates with stability up to 4 d, albeit displaying a much larger size(d = 232 nm, PDI = 0.11). In contrast, compound 10, in which thephenyl ring was replaced by H, did not yield measurable nano-aggregates, supporting the hypothesis that a propeller shape is es-sential for nanoaggregate formation. Because many drugs haveamine functionality, this straightforward protocol of introducing bothperfluoroaryl and phenyl groups constitutes a general strategy for theformulation of PNDs.These nanoaggregates showed fluorescent properties with up to

11% quantum yield (SI Appendix, Table S2). The ability of thesenanoaggregates in bioimaging was tested on bacteria. In the ex-ample shown in Fig. 5, a suspension of nanoaggregates from com-pound 5 was incubated with Escherichia coli (E. coli) ORN178. Tomaintain cell viability, the concentration of nanoaggregates (0.28 μM)was lower than the minimum inhibitory concentration (MIC)(Table 1), and the incubation time was much shorter (4 h vs. 16–18 h). Subsequently, the nanoaggregates were removed by centri-fugation and the bacteria were redispersed for imaging. Fluores-cence from the nanoaggregates of compound 5 was observed (Fig.5A), some at the locations where the bacteria were present (Fig. 5B–D). Fig. 5E shows one bacterium with a nanoaggregate attached.At the location where the nanoaggregate was present [region ofinterest (ROI1) Fig. 5F], the fluorescence signals were dominatedby the emission from the nanoaggregate (410–600 nm, Fig. 5G),which can be distinguished from the weak autofluorescence fromthe bacterium (480–550 nm, Fig. 5 I and J). Residual fluorescencefrom compound 5 was observed on the bacterium in regions with-out the nanoaggregate (ROI2), likely due to the disassembly ofthe nanoaggregate during incubation. The colocalization of nano-aggregates on the bacteria suggests the potential for an increase inthe concentration of drug molecules in the vicinity of the bacteria

compared with the soluble drugs, paving the way for enhancedantimicrobial activity of these nanoaggregates.To test whether the nanoaggregate formulation leads to en-

hanced antibacterial activities, these nanoaggregates and thecorresponding compounds in the soluble form were tested againstseveral strains of E. coli [American Type Culture Collection(ATCC) 25922, ORN 178, and ORN 208]. The compounds in thenanoaggregate form generally resulted in lower MICs than thesoluble forms (Table 1). For example, compound 5 has an MIC of233 μM in the soluble form against E. coli ORN 178, whereas thenanoaggregates of compound 5 (39–74 nm) gave an MIC of 1.25–8.7 μM, up to two orders of magnitude lower. Similar results wereobserved for other ciprofloxacin derivatives, for example com-pounds 6–8, showing lower MICs for the nanoaggregates incomparison with their soluble forms. Enhanced antibacterial ac-tivity was also observed for E. coli ORN 208 (SI Appendix, TableS3), a nalidixic acid (Nal)-resistant variant from E. coli K-12 (44).These data demonstrate that the PND strategy could indeed leadto enhanced antibacterial activity of the free drug.In summary, we report a modular formulation of theranostic

PNDs based on fluoroquinolones. The resulting fluoroquinolonederivatives formed nanoaggregates with enhanced fluorescentemission. The AIE effect was attributed to the packing of thepropeller-shaped structures in the aggregated form, which restrictedthe intramolecular motions. The nanoaggregates were successfullyused as luminescent organic dots to image bacteria. Furthermore,these nanoaggregates showed aggregation-enhanced antibacterialactivity. Thus, the PNDs lowered the MIC against E. coli by morethan one order of magnitude. The enhanced bacterial-killing effi-ciency of PNDs can be in part attributed to the colocalization of thenanoaggregates with the bacterium (Fig. 5) leading to a higher localdrug concentration, and enhanced uptake of drug molecules whendelivered in the nanosized particles (45). The present study repre-sents a strategy in the design and development of antimicrobialtherapies for combating drug-resistant bacterial infections. It can befurthermore used in combination with other strategies, for example,the prodrug approach to modulate drug solubility and activity, tofurther enhance the efficiency of the PNDs.

Materials and MethodsSynthesis of Ciprofloxacin Derivatives. The following describes the synthesis ofcompound 5. All other compounds were prepared following similar proce-dures. Pentafluorophenyl azide (414 mg, 2 mmol), phenylacetaldehyde(400 mg, 4 mmol), and ciprofloxacin (546 mg, 1.65 mmol) were mixed withacetone (100 mL), and stirred at room temperature until the solution becameclear. After acetone was removed, the crude solid was dissolved in DCM, andwashed with aqueous HCl (0.5 M). After evaporation of DCM, the crudeproduct was purified by flash column chromatography over silica gel (95:5DCM/methanol, Rf = 0.4) to yield compound 5 as a pale-yellow solid (860 mg,85%). 1H NMR (400 MHz, CDCl3): δH 1.19 (quartet, 2H, J = 7.1 Hz), 1.38(quartet, 2H, J = 7.1 Hz), 3.23 (br, 4H), 3.53 (septet, 1H, JHH = 3.8 Hz), 3.71(s, 2H), 3.82 (br, 4H), 7.15 (d, 2H, JHH = 7.2 Hz), 7.23 (t, 1H, JHH = 7.3 Hz), 7.28(d, 1H, JHH = 6.9 Hz), 7.32 (t, 2H, JHH = 7.3 Hz), 7.89 (d, 1H, J = 12.9 Hz), 8.67 (s, 1H),14.89 (s, 1H, COOH); 13C NMR (101 MHz, CDCl3): δ 8.4, 35.2, 35.4, 44.8 (br, 2C),49.4, 105.0 (d, 1C, J = 3.0 Hz), 108.2, 112.4, 112.6, 120.0, 120.1, 126.2 (td, 1C, J =15 Hz), 127.3, 127.8, 129.3, 134.5, 135.2 (m, 1C), 138.1 (dm, 2C, JCF = 249 Hz),139.1, 140.1 (dm, 2C, JCF = 230 Hz), 145.4, 145.5, 147.5, 152.6, 154.6, 161.1,166.9, 177.0 (d, 1C, J = 2.4 Hz); 19F NMR (376 MHz, CDCl3): δ-121.0 (s, 1F),−153.3 (m, 2F), −164.3 (m, 2F), −161.0 (m, 1F); electrospray ionization high-resolution mass spectrometry: calculated for C31H25F6N4O3 [M+H]+: 615.1831,found 615.1830. IR (attenuated total reflection): 3,357; 1,718; 1,592; 1,455;1,330; 1,253; 1,197; 1,030; 983, 955, 914, 892, 831, 807, 748, and 722 cm−1.

Preparation of Drug Nanoaggregates. Typical procedure: To an 8 mL vialcontaining Millipore water (5.94 mL), a stock solution of compound 5 inMeCN or DMSO (60 μL, 1 mM) was injected using a micropipette whileshaking. After addition, the vial was hand-shaken rapidly for 20 s. Othernanoaggregates were prepared following the same protocol. DLS mea-surements were performed directly from the suspensions. For transmissionelectron microscopy (TEM) analysis, the nanoaggregate suspension was

Table 1. MICs of compounds 5–8 against E. coli strains ATCC25922 and ORN 178

Free drug Nanoaggregates

MICFree [μM] n MICAggregate, μM n Size†, nm MICfree/MICaggregate*

E. coli ATCC 259225 116–232 3 1.25–8.6 6‡,§ 39–74 366 29–116 3 5.0 3 49–127 127 19–78 3 2.5 3{ 38–51 238 75–150 3 5.0–8.9 5‡ 39–90 22

E. coli ORN 1785 233 2 1.25–8.7 8# 39–80 447 78 2 2.5–5.3 5‡ 37–90 25

n, number of assays, each performed in duplicates or triplicates. In thecase of nanoaggregates, the concentration refers to that of the moleculerather than the nanoaggregates.*Calculated from the averages of MICfree and MICaggregate.†The particle size was measured by DLS, obtained from the main peak byintensity. Samples prepared from different batches having the particle sizefalling in the range were tested.‡1 assay extrapolated (48).#4 assays extrapolated (48).§2 additional assays showing MIC greater than the highest concentration,thus the upper limit is slightly underestimated.{1 additional assay showing MIC greater than the highest concentration,thus the upper limit is slightly underestimated.

8468 | www.pnas.org/cgi/doi/10.1073/pnas.1708556114 Xie et al.

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 5,

202

0

dropped onto a copper grid and quickly dried. For SEM analysis, the sus-pension was dropped onto an Al substrate and quickly dried under vacuum.

Imaging Bacteria. E. coli ORN 178 were incubated overnight in Mueller–Hintonbroth at 37 °C with shaking at 250 rpm. The bacteria were pelleted (6,500 rpm,4,000 × g, 15 min) and redispersed in pH 7.4 PBS to an OD 620 of 0.070(∼108 cfu/mL). A 0.400 mL suspension of compound 5 aggregates preparedfrom 90 vol % water/acetonitrile (100 μM) was added to a 1.00-mL aliquot ofthe bacteria suspension, and the mixture was incubated at 37 °C for 4 h whileshaking at 500 rpm. The mixture was then centrifuged (6,500 rpm, 4,000 × g,15 min); the supernatant was discarded and the pellets redispersed in pH 7.4PBS buffer (1.00 mL). An aliquot of the suspension (0.100 mL) was spread on aglass-bottom dish. Confocal fluorescence microscopy images were collectedusing a Zeiss LSM 780 confocal microscope (Carl Zeiss) with a Plan Apochromat63×/1.4 oil DIC M27 objective and a 32-channel GaAsP spectral detector. Themicroscope was focused on the surface of the glass-bottom dish where livebacteria transiently sedimented before they migrated away. A 405-nm exci-tation source was used for all imaging experiments.

MIC Assays. MIC was determined by broth microdilution (46, 47), with slightmodifications. Liquid cultures of bacteria (E. coli ATCC 25922, E. coli ORN

178, E. coli ORN 208) were grown in log phase and diluted to 106 cfu/mL Theaqueous nanoaggregates suspension was diluted sequentially with water. Ali-quots of the aggregate suspension (50 or 100 μL) was mixed with the equalvolume of bacterium suspension in a 96-well plate (in duplicates or triplicates).For theMIC assay of the free drugs, the compounds were first dissolved inMeCN,and further diluted in PBS buffer (pH = 7.4) to give a maximum final concen-tration of 30 vol % organic solvent. The plates were incubated at 37 °C for 16–18 h. A resazurin solution (10 or 20 μL of 0.01 w/vol% towells with 100 or 200 μLdrug/bacteria mixture, respectively) was added to each sample, and the plateswere incubated for 2 more hours, whereafter the color change was noted. TheMICs obtained by color change corresponded to those from turbidity measure-ment (47), except where free drug precipitated. For extrapolation of partialgrowth inhibition by Gompertz function (48), the turbidity was recorded.

ACKNOWLEDGMENTS.We thank Dr. Andreas Fischer (KTH) for help with theX-ray crystal structure of compound 6, Dr. Nanjing Hao for help with TEM,and Prof. Paul Orndorff (North Carolina State University) for E. coli strainsORN 178 and ORN 208. The study was supported in part by KTH and theNational Institutes of Health (R01GM080295 and R21AI109896 to M.Y.). S.X.and Y.Z. thank the China Scholarship Council for special scholarship awards.

1. Taubes G (2008) The bacteria fight back. Science 321:356–361.2. Berendonk TU, et al. (2015) Tackling antibiotic resistance: The environmental

framework. Nat Rev Microbiol 13:310–317.3. Payne DJ (2008) Microbiology. Desperately seeking new antibiotics. Science 321:1644–1645.4. Zhang L, et al. (2008) Nanoparticles in medicine: Therapeutic applications and de-

velopments. Clin Pharmacol Ther 83:761–769.5. Yu YJ, et al. (2011) Boosting brain uptake of a therapeutic antibody by reducing its

affinity for a transcytosis target. Sci Transl Med 3:84ra44.6. Merisko-Liversidge EM, Liversidge GG (2008) Drug nanoparticles: Formulating poorly

water-soluble compounds. Toxicol Pathol 36:43–48.7. Cao Z, Yu Q, Xue H, Cheng G, Jiang S (2010) Nanoparticles for drug delivery prepared

from amphiphilic PLGA zwitterionic block copolymers with sharp contrast in polaritybetween two blocks. Angew Chem Int Ed Engl 49:3771–3776.

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

9. Kasai H, et al. (2012) Creation of pure nanodrugs and their anticancer properties.Angew Chem Int Ed Engl 51:10315–10318.

10. Wang Y, et al. (2014) Disulfide bond bridge insertion turns hydrophobic anticancerprodrugs into self-assembled nanomedicines. Nano Lett 14:5577–5583.

11. Zhang J, et al. (2015) Preparation and size control of sub-100 nm pure nanodrugs.Nano Lett 15:313–318.

12. Chan HK, Kwok PCL (2011) Production methods for nanodrug particles using thebottom-up approach. Adv Drug Deliv Rev 63:406–416.

13. Wu L, Zhang J, Watanabe W (2011) Physical and chemical stability of drug nano-particles. Adv Drug Deliv Rev 63:456–469.

14. MacKay JA, et al. (2009) Self-assembling chimeric polypeptide-doxorubicin conjugatenanoparticles that abolish tumours after a single injection. Nat Mater 8:993–999.

15. Mei J, et al. (2014) Aggregation-induced emission: The whole is more brilliant thanthe parts. Adv Mater 26:5429–5479.

16. Mei J, Leung NL, Kwok RT, Lam JW, Tang BZ (2015) Aggregation-induced emission:Together we shine, united we soar! Chem Rev 115:11718–11940.

17. Kaeser A, Schenning AP (2010) Fluorescent nanoparticles based on self-assembled pi-conjugated systems. Adv Mater 22:2985–2997.

18. Bucar D-K, Elliott JA, Eddleston MD, Cockcroft JK, Jones W (2015) Sonocrystallizationyields monoclinic paracetamol with significantly improved compaction behavior.Angew Chem Int Ed Engl 54:249–253.

19. Xue X, et al. (2014) Spatiotemporal drug release visualized through a drug deliverysystem with tunable aggregation-induced emission. Adv Mater 26:712–717.

20. Okazawa Y, Kondo K, Akita M, Yoshizawa M (2015) Polyaromatic nanocapsules dis-playing aggregation-induced enhanced emissions in water. J Am Chem Soc 137:98–101.

21. Zhao E, et al. (2015) A luminogen with aggregation-induced emission characteristicsfor wash-free bacterial imaging, high-throughput antibiotics screening and bacterialsusceptibility evaluation. Adv Mater 27:4931–4937.

22. Feng G, et al. (2015) A light-up probe with aggregation-induced emission charac-teristics (AIE) for selective imaging, naked-eye detection and photodynamic killing ofGram-positive bacteria. Chem Commun (Camb) 51:12490–12493.

23. An B-K, Kwon S-K, Jung S-D, Park SY (2002) Enhanced emission and its switching influorescent organic nanoparticles. J Am Chem Soc 124:14410–14415.

24. Castro W, Navarro M, Biot C (2013) Medicinal potential of ciprofloxacin and its de-rivatives. Future Med Chem 5:81–96.

25. Hooper DC (1999) Mode of action of fluoroquinolones. Drugs 58:6–10.26. Thomson CJ (1999) The global epidemiology of resistance to ciprofloxacin and the changing

nature of antibiotic resistance: A 10 year perspective. J Antimicrob Chemother 43:31–40.

27. Hooper DC (1999) Mechanisms of fluoroquinolone resistance. Drug Resist Updat 2:38–55.28. Cuquerella MC, Miranda MA, Bosca F (2006) Role of excited state intramolecular

charge transfer in the photophysical properties of norfloxacin and its derivatives.J Phys Chem A 110:2607–2612.

29. Xie S, Lopez SA, Ramström O, Yan M, Houk KN (2015) 1,3-Dipolar cycloaddition re-activities of perfluorinated aryl azides with enamines and strained dipolarophiles.J Am Chem Soc 137:2958–2966.

30. Kong N, et al. (2016) Catalyst-free cycloaddition reaction for the synthesis of glyco-nanoparticles. ACS Appl Mater Interfaces 8:28136–28142.

31. Xie S, Zhang Y, Ramström O, Yan MD (2016) Base-catalyzed synthesis of aryl amidesfrom aryl azides and aldehydes. Chem Sci 7:713–718.

32. Xie S, Ramström O, Yan M (2015) N,N-diethylurea-catalyzed amidation betweenelectron-deficient aryl azides and phenylacetaldehydes. Org Lett 17:636–639.

33. Xie S, Fukumoto R, Ramström O, Yan M (2015) Anilide formation from thioacids andperfluoroaryl azides. J Org Chem 80:4392–4397.

34. Park H-R, Kim TH, Bark K-M (2002) Physicochemical properties of quinolone antibi-otics in various environments. Eur J Med Chem 37:443–460.

35. Albini A, Monti S (2003) Photophysics and photochemistry of fluoroquinolones. ChemSoc Rev 32:238–250.

36. Willets KA, Nishimura SY, Schuck PJ, Twieg RJ, Moerner WE (2005) Nonlinear optical chro-mophores as nanoscale emitters for single-molecule spectroscopy. Acc Chem Res 38:549–556.

37. Hu RR, et al. (2009) Twisted intramolecular charge transfer and aggregation-inducedemission of BODIPY derivatives. J Phys Chem C 113:15845–15853.

38. Zhang GF, Aldred MP, Gong WL, Li C, Zhu MQ (2012) Utilising tetraphenylethene as adual activator for intramolecular charge transfer and aggregation induced emission.Chem Commun (Camb) 48:7711–7713.

39. Gao BR, et al. (2010) Time-resolved fluorescence study of aggregation-induced emission en-hancement by restriction of intramolecular charge transfer state. J Phys Chem B 114:128–134.

40. Wang B, et al. (2011) Starburst triarylamine donor-acceptor-donor quadrupolar derivativesbased on cyano-substituted diphenylaminestyrylbenzene: Tunable aggregation-inducedemission colors and large two-photon absorption cross sections. Chem Eur J 17:2647–2655.

41. Mukherjee S, Thilagar P (2014) Insights into the AIEE of 1,8-naphthalimides (NPIs):inverse effects of intermolecular interactions in solution and aggregates. Chem Eur J20:8012–8023.

42. Chung HR, Kwon E, Kawa H, Kasai H, Nakanishi H (2006) Effect of solvent on organicnanocrystal growth using the reprecipitation method. J Cryst Growth 294:459–463.

43. Blanchard MD, Hughes RP, Concolino TE, Rheingold AL (2000) π-Stacking between pen-tafluorophenyl and phenyl groups as a controlling feature of intra- and intermolecularcrystal structure motifs in substituted ferrocenes. Observation of unexpected face-to-facestacking between pentafluorophenyl rings. Chem Mater 12:1604–1610.

44. Harris SL, et al. (2001) Characterization of Escherichia coli type 1 pilus mutants withaltered binding specificities. J Bacteriol 183:4099–4102.

45. Shimanovich U, et al. (2015) Tetracycline nanoparticles as antibacterial and gene-silencing agents. Adv Healthc Mater 4:723–728.

46. Hao N, Jayawardana KW, Chen X, Yan M (2015) One-step synthesis of amine-functionalized hollow mesoporous silica nanoparticles as efficient antibacterial andanticancer materials. ACS Appl Mater Interfaces 7:1040–1045.

47. Jayawardena HS, Jayawardana KW, Chen X, Yan M (2013) Maltoheptaose promotesnanoparticle internalization by Escherichia coli. Chem Commun (Camb) 49:3034–3036.

48. Lambert RJ, Pearson J (2000) Susceptibility testing: Accurate and reproducible mini-mum inhibitory concentration (MIC) and non-inhibitory concentration (NIC) values.J Appl Microbiol 88:784–790.

Xie et al. PNAS | August 8, 2017 | vol. 114 | no. 32 | 8469

CHEM

ISTR

Y

Dow

nloa

ded

by g

uest

on

Feb

ruar

y 5,

202

0