Fluorescent and Water Dispersible Single‐Chain Nanoparticles: Core–Shell Structured Compartmentation

Abstract Single‐chain nanoparticles (SCNPs) are highly versatile structures resembling proteins, able to function as catalysts or biomedical delivery systems. Based on their synthesis by single‐chain collapse into nanoparticular systems, their internal structure is complex, resulting in nanosized domains preformed during the crosslinking process. In this study we present proof of such nanocompartments within SCNPs via a combination of electron paramagnetic resonance (EPR) and fluorescence spectroscopy. A novel strategy to encapsulate labels within these water dispersible SCNPs with hydrodynamic radii of ≈5 nm is presented, based on amphiphilic polymers with additional covalently bound labels, attached via the copper catalyzed azide/alkyne “click” reaction (CuAAC). A detailed profile of the interior of the SCNPs and the labels’ microenvironment was obtained via electron paramagnetic resonance (EPR) experiments, followed by an assessment of their photophysical properties.


Introduction
Among many nanosized carriers single chain nanoparticles (SCNPs) are the most versatile,both in molecular design, structural diversity and embedded functionalities. [1] Derived from single polymer chains by covalent or noncovalent crosslinking they directly link the vast synthetic space of controlled polymer synthesis with their function, embedding chemical functionalities into the final SCNPs to allow applications of SCNPs in the areas of catalysis, [2] drug delivery [3] or imaging technologies [3b,4] such as photoacoustic imaging. [5] Crucial thereto is the placement of specific chemical functionalities within the SCNP,f urther stimulated by its nanosized dimension, excellent dispersion and internal compartmentation. Thef ormation of nanosized compartments within such SCNPs has been intensely discussed, [1a,6] proposing their use as functional units to generate microenvironments for catalysis, [7] photophysics, [8] or subsequent embedding of chiral elements. [9] As the formation of SCNPs is based on the collapse of single polymer chains at low concentration via covalent or non-covalent intramolecular bonding interactions, [10] it "freezes-in" the conformational state of the polymer chain into the final, crosslinked SCNP. [11] However,d ynamics of the polymer chains often is at least partially preserved depending on the degree of crosslinking and solvent effects during the crosslinking process. [6] Thus recently as trong solvent dependencyo nthe final size of single-chain nanoparticles has been observed using combshaped polymers,bearing for example,polyisobutylene (PIB) sidechains with ad irect correlation between solvent quality and the final SCNP size. [8b] Strong photophysical effects as for example an increasing rate of photoinduced dimerization has been observed within pre-folded SCNPs via an increase of quantum yields by the confinement-effects within the SCNP. [8b] Structural dynamics can lead to am odulation of the band structure of SCNPs and the excited states of bound dye molecules on the sub-ns timescale which determines the achieved fluorescence quantum yield and both, the absorption and fluorescence spectra of the dyes. [12] Although the formation of nanosized compartments within the SCNPs has been often proposed and observed indirectly,adirect proof has not been accomplished as many of the used methodologies provide an only indirect measure of compartmentation. Thus monitoring the molecular dynamics of water molecules around SCNPs via Overhauser dynamic nuclear polarization (ODNP) hinted to supramolecular organized hydrophobic benzene-1,3,5-tricarboxamides (BTA) into chiral folds within the SCNP,r esembling their self-assembly in solution. [13] Also the solution behavior of polymers within SCNPs has been shown to deviate from those of the respective uncrosslinked polymers due to confinement within nanosized compartments,l eading to significant changes in their lower critical solution (LCST) behavior. [14] We here probe the formation of compartments within synthetic SCNPs based on amphiphilic polymers,e quipped with an active spin label, via continuous wave electron paramagnetic wave resonance (CW EPR), allowing to probe the internal compartments of the formed SCNPs.A dditionally,f luorescent dyes,a ppropriate for use in pump-probe photoacoustic imaging [15] are embedded into small (r h < 10 nm) SCNPs to allow sufficient permeation for cellincorporation. [16] Thus the SCNPs act as ab arrier against defense mechanisms of the targeted tissues and stabilize the dye against external photobleaching. Important in our endeavor was the quest to position the dye within the hydrophobic core of the SCNP,a ssuch its accessibility is reduced and thus ap roposed protection effect of the SCNP could be expected to be most effective.
With the here reported single-chain nanoparticles one can therefore generate as caffold for encoding the required different functionalities stimuli-response,f or example,t hermo-or pH-responsivity, [3c, 17] and biocompatibility, [18] by using three types of monomers to adjust both, the required amphiphilic balance for the single-chain collapse,a nd the dye attachment.

Results and Discussion
Ther equired poly(poly(ethylene glycol) methacrylate) (PEGMA)-copolymer (polymer I)was synthesized by RAFT copolymerization of poly(ethylene glycol) methacrylate (M n = 300 Da, n = 4.5), azidopropyl methacrylate,a nd 3-(trimethylsilyl)propargyl methacrylate in DMF,u sing AIBN as initiator and cyanoisopropyl dithiobenzoate (CPDB) as chain transfer agent (CTA), followed by the removal of the CTAu sing an excess of AIBN. [19] TheC TA, [20] them ono-mers, [21] and the dye labels [22] were synthesized according to literature with small adaptions (see Supporting Information). Fort he polymerization the molar fractions of the three monomers were set as 0.8, 0.12, and 0.08, respectively,t o achieve an appropriate number of crosslinking groups [23] and to retain residual attachment sites for modification with the labels after the single-chain collapse (polymer I: M n = 36.1 kDa, PDI = 1.7, DP = 129). Furthermore,t his composition allows to induce thermoresponsivity together with sufficient dispersibility and hydrophobic packaging of the labels.F inal composition and structural integrity of the resulting polymer was proven by NMR-spectroscopy,a llowing to adjust the desired functionalities along the polymer chain in relation to the initially used amounts of the three monomers (see Supporting Information, including the final copolymer-compositions). Subsequently,single-chain collapse of polymer I and labelling (see Scheme 1) was accomplished in at wo-step/one-pot reaction, where the alkyne group was first deprotected by tetra butyl ammonium fluoride in water, followed by slow addition of the resulting polymer solution to an aqueous CuSO 4 /NaAsc solution via as yringe pump to adjust low concentrations (c polymer < 10 À6 M) for the required single-chain folding.T he amphiphilic polymer thus is adopting apreorganized nanoparticle by forming an intramolecular core-shell structure with the hydrophilic PEG sidechains in the shell and the reactive azide-groups in the core,resulting in sole intrachain crosslinking of the reactive groups to form SCNP II via CuAAC-reaction (see Scheme 1). [24] Forf urther labelling of SCNP II,t he catalyst was reactivated by adding additional sodium ascorbate and al abel-alkyne solution in water (with the help of Kolliphor EL as detergent for the Scheme 1. Synthesis of the precursor polymerIby RAFT polymerization followed by the formation of the unlabeled SCNP II and labelling to obtain the labelled SCNP IIIa-c. 2,2,6,6-tetramethylpiperidine oxide (TEMPO)-label and aza-BODIPY (aBOD)). This allows areaction of the still present residual terminal azido moieties in the hydrophobic core to attach the two fluorescent dyes Rhodamine B( RhoB) and aBOD as aN IR-fluorescent dye.E specially aBOD is attractive as it displays at umor targeting functionality via deprotonation, resulting in intramolecular charge transfer, and thereby suppression of fluorescence. [22d, 25] TheEPR-label 2,2,6,6-tetramethylpiperidine oxide (TEMPO) was attached for subsequent EPR investigations of the SCNPsi nner core (see Scheme 1).
Progress of the crosslinking was followed by attenuated total reflection infrared (ATR-IR) spectroscopy (see Figure S1a and b) indicating af ull disappearance of the alkyne band at 2178 cm À1 in the SCNP II,and apartial and complete removal of the azide band at 2100 cm À1 for SCNP II and the SCNPs III,r espectively.T he new bands of the triazole ring are overlapped by the bands of the polymer backbone and the PEG sidechains and therefore not visible in the spectrum. 1 H-NMR spectroscopy proves acomplete reaction of the reactive groups under formation of triazole rings (see Figure S1d, removal of the TMS-protecting group at 0.2 ppm, absence of the alkyne proton at 2.6 ppm). Ac rosslinking density of % 9 crosslinks per SCNP was roughly calculated from the stoichiometric ratios,l eaving % 4r emaining attachment sites per SCNP for the labels.T he so obtained SCNPs are water dispersible and can be redispersed after freeze-drying.
Diffusion-ordered NMR spectroscopy (DOSY-NMR) and dynamic light scattering measurements of polymer I and SCNP II (see Figure 1a nd Table 1) independently show ar eduction of the hydrodynamic radius after the CuAAC from 5.2 nm to 4.2 nm. This proves the single-chain collapse in good agreement with literature,w here ar eduction of the hydrodynamic radius is an indication of as ingle-chain collapse. [26] Upon further labelling of the SCNP II yielding SCNPs IIIa-c,b oth methods show the trend of an increased hydrodynamic radius of % 1.4 nm (DOSY-NMR, see Table 1) and of % 0.5 nm (DLS) indicating the successful attachment of the respective labels,w hich lead to an expansion of the measured volume,a sthe dye-labels are comparably large in dimension. Atomic force microscopy (AFM) further proved the formation of SCNPs with an average height of 7.5 AE 1.5 nm (see Figure 1c,d and Figure S2). This underscores the formation of distorted particles due to the surface/SCNPinteraction between the polar mica-surface and the hydrophilic PEG-shell, also indicative of the high responsivities of the SCNPs during adsorption, inline with previous observations. [27] Size exclusion chromatography (SEC) measurements in water further prove the covalent attachment of the labels to the SCNPs via an overlap of the R.I. and UV/Vis track for the dye-labelled SCNPs IIIb and IIIc (see Figure S4 da nd e). During single-chain collapse,the hydrophobic reactive groups (azide/alkyne) are proposed to be located in the inner core of the nanoparticles,hidden from the aqueous surroundings.W e therefore observe ap olarity shift in SEC,i nline with an increased hydrophobic collapse (T cp )a fter the single-chain collapse and with further modification moving from the native polymer I (T cp = 48 8 8C) to SCNP II (T cp = 54 8 8C) to SCNP IIIa-c (T cp = 65 8 8C, 69 8 8Ca nd 75 8 8C, respectively) (see Table 1and Figure S3). It should be noted that this is the first observation of such aconsequent increase of T cp ,initially only observed for noncrosslinked, dynamic random copolymers. [14] Additionally,this behavior explains the decrease of retention time in SEC moving from polymer I to SCNP II,w ith concomitantly increasing T cp .Labelling SCNP II with the dyes to yield SCNPs IIIa-c leads to af urther decrease of the retention time and afurther increase of T cp .
To achieve ad eeper understanding of the microstructure and the distribution and location of the individual polymer and label segments within the SCNPs,C WE PR measurements were conducted. Forthe protective effect in SCNPs to be effective,t he exact location (spatially and in chemical environments) of the labels within the SNCPs and the desired location within the hydrophobic core is crucial. Moreover, arough quantification of the number of labels in one SCNP is needed. With the CW EPR spectra of the attached TEMPO label it is possible to specify the chemical environment of the   Figure S3 b).
nanostructure surrounding the probe within ar adius of % 1nm. [28] Forc omparison of the TEMPO-labelled SCNP IIIa with an equally labelled linear polymer,t he TEMPOlabelled polymer I' 'a was synthesized analogously to polymer I,b ut without the alkyne moieties (see Supporting Information). Upon covalently binding TEMPO as al abel (Figure 2b) to polymer I' ' yielding polymer I' 'a,asignificant broadening of the EPR peaks is observed, caused by the hindered rotation of the label attached to the large polymer.T his effect is even stronger in the TEMPO-labelled SCNP IIIa,i ndicating ar eduction of the mobility of the label upon attachment to the SCNP.F igure 2a in comparison shows the CW EPR spectra of polymer I and SCNP II in water in pure physical mixtures with the free TEMPO probe.T he three identical normalized spectra indicate free TEMPO in water, proving the absence of specific,m ostly hydrophobic interactions of the amphiphilic TEMPO probe with the polymer and the SCNP,respectively,asweusually and regularly observe it with amphiphilic (LCST-type) polymers. [28a] Simulated spectra of the samples quantify the mobility reduction and the low polarity of the environment as expected for water-depleted, polymer-rich regions (see Figure 2c). For SCNP IIIa,but not for the respective polymers an exchange coupling of 6MHz had to be included to the calculation, indicating high local concentrations and direct contact of two or more TEMPO labels with distances < 1nm. Even though acertain amount of TEMPO was probably deactivated by reduction during the labelling process because of sodium ascorbate present in the reaction mixture,the degree of labelling was approximated to be 1.3 active labels per particle (see Figure S10). This, together with our spectra, leads to the conclusion that the SCNPs bear one to two chemically attached active spin labels and even more EPR-inactive labels.
Concentration dependent measurements of SCNP IIIa in Figure 3a,b indicate that the label is hidden from the  environment outside the nanoparticle as expected for the specific topology and compartmentation in the SCNPs.A ll normalized spectra show the same shape,a part from small differences in the signal-to-noise ratio.E ven at high concentrations the surroundings of the labels remain the same.Thus, the label seems to be completely unaffected by external influences outside of the particle,e ven though collision or agglomerates with other particles at high concentrations might occur (see Figure 3c). Summing up the EPR results,the up to two active labels per nanoparticle are covalently bound in SCNP IIIa in aconfined space surrounded by the non-polar polymer backbone,w hich is reflected in the lower hyperfine splitting of labels on SCNPs,i nt he nanoparticles core, altogether as as trong proof for the formation of such ac ompartment. External influences do not affect the label, indicating its embedding in the SCNPscore.Considering that TEMPO is more hydrophilic than the hydrophobic dyes embedded for optical applications,t he SCNPs are therefore able to protect ap hotoactive label from photooxidation or other degradational processes.
Theoptical properties of dyes can be constructively tuned by binding to SCNPs. [12,29] Figure 4a nd Table 2s how the absorption and fluorescence spectra and wavelengths of the RhoB and aBOD labelled SCNPs in water and aqueous phosphate buffer (pH 6.0), respectively,s pectra of the free dyes can be seen in Figure S5. In comparison to free RhoBalkyne,t he RhoB-labelled SCNP IIIb shows ar ed shifted absorption at 559 nm and ab lue shifted fluorescence at 582 nm (stokes shift reduced to 23 nm). In the aBOD-labelled SCNP IIIc,resonances are blue shifted by 6nmincomparison to the free aBOD dye in aqueous phosphate buffer,w ith astokes shift of 27 nm for the free aBOD dye and SCNP IIIc, respectively.Inline with the results from EPR measurements, those shifts occur by ad ifferent solvation of the dyes in the particlesh ydrophobic core,w ell separated from the sur-rounding bulk water phase,again indicative for the compartmented nanostructure of the SCNP.
Then ear neighborhood of two or more bound dye molecules leads to interactions and, dependent on their distance and orientation, possible excitonic coupling between the dyes which changes the fluorescence quantum yield, the wavelength maximum and FWHM of the fluorescence emission. [12] In addition internal relaxation processes on the sub-ns time scale and the interaction of the dye with the polymer at different sites and possible distributions of the molecular structure modulate the fluorescence emission and determine the fluorescence lifetime of excited states. [30] Time resolved fluorescence spectroscopy enables the analysis of sub-band structures,dynamics of interaction between excited states and the environment and possible heterogeneous decay channels resulting from excitonic coupling and/or compartmentation that contribute to the integral fluorescence spectrum. Just like in the EPR measurements we synthesized alinear aBOD-labelled polymer (polymer I' 'c see Supporting Information) to compare it to SCNP IIIc in view of their excitational behavior and heterogeneity. Figure 5r eveals how an excited state heterogeneity that might either result from molecular coupling and/or as tructural heterogeneity,p ossibly with subsequent relaxation dynamics after light absorption in the bound dyes,l eads to as plit of the formerly homogeneous excited singlet states of different molecules as indicated in Figure 5d.W hile the free aBOD quickly decays with atime constant of 110 ps in water due to quenching of the surrounding aqueous medium the fluorescence lifetime and fluorescence quantum yield significantly rises after binding the dye to ap olymer or SCNP.F or polymer I' 'c one can see as trong heterogeneity in the decay associated spectra (DAS,F igure 5b)w ith as pectral separation of two states with distinct different lifetimes of 1.2 AE 0.2 ns (red curve,e nergetically lower level) and 2.7 AE 0.2 ns (blue curve,e nergetically higher level) additionally to af ast decay component that is similar to the free dye (black curve, 220 AE 50 ps) with contributions from molecular interaction (see below). This is caused by aw eak and dynamic compartmentation, as depicted in Figure 5e,w ith partially folded regions in which the dye molecules can have direct contact to each other and free unfolded regions without direct contact between dye molecules that can be quenched by the aqueous surrounding.T his heterogeneity between the bound dye molecules also explains the strong inhomogeneous broadening of the absorption band in Figure S6. Thea bsorption is therefore even broader as compared to SCNP IIIc because the elongated polymer has al arger degree of freedoms as compared to the SCNP.A st he fluorescence maxima of all components in the DASa re similar the observed heterogeneity refers to structural differences rather than excitonic interaction of the different dye molecules.
For SCNP IIIc as tronger compartmentation can be observed from the DAS ( Figure 5c). Thed ecay components redistribute with distinct different maxima, indicating that absorbed light energy potentially is transferred between strongly coupled dye molecules.S light spectral separation of the two states which were also observed in polymer I' 'c can be seen. The1 .2 AE 0.2 ns component (red curve) seems to be  emitted from the energetically lower level while the 2.9 AE 0.2 ns component (blue curve) results from the energetically higher level. Thef astest component slows down further as compared to polymer I' 'c and was measured 300 AE 50 ps with strong spectral asymmetry and negative value at 730 nm which is typical for an excitonic relaxation from strongly blue shifted states with an emission around 700 nm to the observed fluorescence maximum around 720-730 nm.
Thep ossible band structure in the material is described schematically in Figure 5d.T he decay associated spectra in Figure 5c indicate that excitation from the ground state (GS) is followed by radiative decay from the higher state with at ime constant of 2.9 ns while fluorescence from the lower state decays with 1.2 ns.L ight absorption and relaxation of the exciton induce ac omplex dynamic into the surrounding polymer and SCNP environment that might cause an apparent relaxation time of about 300 ps possibly accompanied by rearrangements of different compartments or molecular dynamics/ conformation changes of the polymer. [30] Thes pectroscopic properties are strongly determined by the exact localization of the states at the binding sites of the aBOD molecules in SCNP IIIc.Specific configuration for the bound molecules by specific molecular design of the binding sites in the SCNP would allow for the creation of highly individual optical properties of the coupled aBOD dyes and allow for the fine tuning of color and fluorescence yield lifetime of the bound molecules.
Thel ocation of the aBOD dye in the SCNPsc ore also effects its pK a and thereby its pH-responding functionality.As depicted in Table 2and Figure S7 the fluorescence of the label is strongly changed by the pH value of the solvent, with a pK a of 6.8 for the free aBOD and 7.6 for SCNP IIIc.Changing the pH of the SCNP IIIc dispersion from 6.0 to 8.6 strongly decreases its fluorescence,b y9 2%.T he pH-responsivity has no strong influence on the fluorescence lifetime (see Figure S8 and S9). Thelifetimes in both, the free aBOD dye and SCNP IIIc,s tay nearly unchanged over the whole pH range, prospective for ap roper tumor tissue targeting. These data are in compliance with the overall fluorescence yield shown in Figure S7 with stronger amplitude reduction observed for SCNP IIIc (see Figure S9, Figure S7d) as compared to aBOD in solution ( Figure S8, Figure S7b).

Conclusion
In summary we could observe the formation of internal compartments within single chain nanoparticles (SCNPs) via EPR-and fluorescence spectroscopic methods.T he SCNPs are synthesized by using ac lickable PEGMA based copolymer synthesized by RAFT polymerization, followed by an intramolecular CuAACwithin its hydrophobic core.DLS and DOSY-NMR measurements proved single-chain collapse with hydrodynamic radii of 4-5 nm. Thew ater dispersible SCNPs all displayed a T cp higher than that of the free chain of the native polymer I,i ndicative for the formation of coreshell structured nanoparticles with the hydrophilic PEG chains in the shell and the hydrophobic groups with the embedded labels located inside the core.T he inherent dynamics of the SCNPs allows for am edium-responsive nanostructure as indicated by their thermoresponsivities,and the formation of contacts between two or more labels in one SCNP.Definite proof of this compartmented model consisting of acore-shell structure was accomplished via EPR measurements,p roving the presence of at least two covalently attached labels within ac onfined space with distances below 1nm, surrounded by the non-polar polymer backbone and shielded from outer influences by the particless hell. Based on the quite unique combination of thermal, EPR and photophysical measurements we could prove the nanocompartmented structure of such SCNPs,c onsisting of the hydrophobic labels embedded within the corresponding hydrophobic compartments,w ith the short PEG-chains generating the outer shell. Coupling between dye molecules and the local environment strongly influences their optical properties giving rise to as trategy that allows for specific molecular design of the optical properties.D ue to the so enabled shielding of the fluorescent dyes inside the SCNPs, Figure 5. Decay associated spectra measured on a) aBOD in water (c = 25 mM), b) polymer I' 'c in water (c = 0.1 mg mL À1 ,dye concentration ca. 5-10 mM) and c) SCNP IIIc in water (c = 0.1 mg mL À1 ,d ye concentration ca. 5-10 mM) after Fit with 3e xponential functions (excitation wavelength 632 nm, see SI for further information).d )Excitonic interaction in SCNP IIIc and resulting split of the S 1 states and qualitative band structure due to excitonic splitting and or interaction with the environmentofthe first excited singlet state (S 1 ). The green arrow indicates the excitation from the ground state (GS) and the blue arrow fluorescence from the higher excitonic level with atime constant of 2.9 ns while fluorescence from the lower level decays with 1.2 ns. e) Schematic interpretation of the time resolved fluorescence spectroscopic data of polymer I' 'c (left side) and SCNP IIIc (right side). the fluorescent SCNPs IIIb and IIIc prospect their application as contrasting agents in photoacoustic imaging,s till maintaining their targeting properties uninfluenced by their encapsulation in the SCNPs.