Quantum Defects as a Toolbox for the Covalent Functionalization of Carbon Nanotubes with Peptides and Proteins

Abstract Single‐walled carbon nanotubes (SWCNTs) are a 1D nanomaterial that shows fluorescence in the near‐infrared (NIR, >800 nm). In the past, covalent chemistry was less explored to functionalize SWCNTs as it impairs NIR emission. However, certain sp3 defects (quantum defects) in the carbon lattice have emerged that preserve NIR fluorescence and even introduce a new, red‐shifted emission peak. Here, we report on quantum defects, introduced using light‐driven diazonium chemistry, that serve as anchor points for peptides and proteins. We show that maleimide anchors allow conjugation of cysteine‐containing proteins such as a GFP‐binding nanobody. In addition, an Fmoc‐protected phenylalanine defect serves as a starting point for conjugation of visible fluorophores to create multicolor SWCNTs and in situ peptide synthesis directly on the nanotube. Therefore, these quantum defects are a versatile platform to tailor both the nanotube's photophysical properties as well as their surface chemistry.


Introduction
Since their discovery [1] single-walled carbon nanotubes (SWCNTs) attracted al ot of attention not only because of their unique chemical structure,b ut also because of their outstanding photophysical properties such as non-bleaching/ blinking near-infrared (NIR) fluorescence. [2][3][4] This NIR fluorescence is beneficial especially for bioimaging as the emission wavelength of SWCNTs falls into the so-called tissue-transparencyw indow where the absorption of water, hemoglobin, and lipids reaches ac ombined minimum and scattering is reduced compared to that of visible light. [5] Consequently,SWCNTs already found application in diverse settings ranging from in vivo NIR imaging [6][7][8] over drug delivery vehicles [9,10] to NIR optical sensors. [11][12][13][14][15][16][17][18][19][20] In order to carry out their desired function in these important applica-tions,t he pure carbon tube has to be modified with, for example,t he cargo to be carried or ar ecognition unit imparting specificity for the analyte to be detected. Furthermore,c arbon nanotubes are not water-soluble,p reventing applications in aqueous systems unless the hydrophobic surface is coated with an amphiphilic surfactant. [12] In the last 20 years,b oth covalent and noncovalent functionalization have been employed for the decoration of SWCNTs with functional units.N oncovalent functionalization such as adsorption of DNAi sb yf ar the most frequently applied approach owing to its ease-of-use and mild conditions. [21][22][23] Furthermore,c onformational changes of the coating molecule can directly translate into changes of the NIR fluorescence,which is interesting for sensing. [16,24] On the other hand, covalent functionalization leads to more stable conjugates, but destroys the SWCNTsextended p-network and thus also the NIR fluorescence. [2,25] In 2017, Setaro et al. reported preserved fluorescence by using a[ 2 + +1]-cycloaddition with electron-poor aromatic nitrenes,which they also used for the attachment of gold nanoparticles and spiropyranes. [26] Very recently,G odin et al. used spyropyran-switchable SWCNTs for NIR superresolution microscopy, [27] while Chio et al. used the same nitrene [2+ +1]-cycloaddition for the attachment of small molecules such as biotin. [28] However,s p 3 defects can not only diminish the NIR fluorescence,b ut were also found to modulate it depending on the nature and density of the defects.In2010, Ghosh et al. reported on aN IR emission peak red-shifted by approximately 130 nm (termed E 11 * )u pon introduction of oxygen defects. [29] Later,Piao et al. observed asimilar peak shift and enhanced fluorescence quantum yield upon introduction of aryl defects using diazonium salts (quantum defects). [30] This technique also enabled tuning of the defect fluorescence both in terms of intensity and emission wavelength via different substituents on the aryl/alkyl defect. [31][32][33][34] In the last years,an umber of different oxygen and aryl defects were reported that are now very promising tools for the generation of brighter/modified SWCNTs, [31,35,36] pH [37] and saccharide [38] sensors,s hort fluorescent SWCNTs for superresolution microscopy, [39] and single-photon sources for quantum computing. [40,41] Furthermore,t his new defect-induced fluorescence feature moves the emission even further into the biotransparencywindow,leading to even better tissue penetration properties. [42,43]

Results and Discussion
In this work, we expand the use of quantum defects from modulation of the SWCNTs photophysical properties to-wards using them as modular anchors for the attachment of peptides and proteins ( Figure 1). To this end, we employed an (N-maleimido)phenyldiazonium salt (MalPh-Dz, 2), whichafter defect introduction-can be targeted by thiol-containing molecules such as proteins,while at the same time generating af urther red-shifted E 11 * emission feature (defect-carrying SWCNTs are thus referred to as SWCNT* in the following).
We started our investigations by optimizing the narrow window of reaction conditions [44] between sodium dodecylbenzenesulfonate (SDBS)-dispersed SWCNTs 1 and 4-(Nmaleimido)phenyldiazonium tetrafluoroborate (MalPh-Dz, 2), allowing for the NIR fluorescence to be altered but not diminished (Figure 2a). Forafast and efficient screening of conditions (reaction time and reactant concentration, see Figure S1), we made use of a9 6-well green LED array for SWCNT excitation driving the radical arylation reaction. The best results were achieved when a1 0nm SWCNT (c carbon % 530 mm)s olution (dispersed in 1% SDBS/H 2 O) was mixed with 2 (100 mm)and irradiated for 15 minutes.Higher MalPh-Dz concentrations would lead to too many defects and diminish NIR fluorescence. [30] Thesuccess of the defect reaction was monitored by NIR fluorescence spectroscopy (Figure 2b)a nd the SWCNTs structural integrity and colloidal stability was controlled using Vis/NIR absorbance spectroscopy ( Figure S2). Thef luorescence spectra and the excitation/emission map (see Figure 2c) clearly show ag rowing E 11 * peak at approximately 1135 nm with increasing concentration of 2 (control 2D spectrum with an onmodified SWCNT/SDBS sample can be found in Figure S3). Thei ncreased E 11 * /E 11 ratio of the 2D versus the 1D NIR fluorescence spectrum could be attributed to prolonged exposure to MalPh-Dz 2 before spin filtration and measurement of the 2D spectrum or due to preferential redispersion of MalPh-SWCNT*.
After successful introduction of (N-maleimido)phenyl quantum defects (MalPh defects), we tested whether biomol-ecules can be conjugated to this anchor.Asafirst example,we chose an anobody against green fluorescent protein (GFP). Nanobodies are the isolated antigen-binding region of heavychain antibodies found, for example,i nCamelidae and are only 10 %o ft he size of conventional antibodies. [45,46] This renders them very useful as binders for diverse applications such as superresolution microscopy, [47][48][49] live-cell immunostaining after modification with cell-penetrating peptides, [50] and isotopic labeling of biological samples for secondary-ion mass spectrometry (SIMS) imaging. [51] Nanobodies binding GFP (GFP-binding protein, GBP) in particular can be used as ap latform technology due to the widespread availability of GFP-fusion proteins or even whole genetically modified organisms expressing GFP-fusion proteins,g iving the possibility to target awhole variety of proteins with just one single conjugate.Similar to our previous (noncovalent) work, [13] we used aG BP with as ingle ectopic C-terminal cysteine for oriented conjugation to the MalPh-SWCNT* 3 leaving the antigen-binding region pointing away from the SWCNT* surface.
With the fast hydrolysis kinetics of N-aryl maleimides [52] in mind, we evaluated both the sequential defect introduction followed by nanobody conjugation as well as ao ne-step approach combining all three reaction partners at once ( Figure 3a). Forthe sequential reaction, the excess diazonium salt 2 was removed using 300 kDa molecular weight cutoff (MWCO) spin filters followed by resuspension of the now naked MalPh-SWCNT* 3 in 1x phosphate-buffered saline (PBS,p H7.4) and reaction with 500 equiv.(% 25 equiv.w ith respect to introduced maleimides) of the nanobody 4 (16 hat room temperature). In the one-step approach, the same excess of GBP was added directly during the defect introduction (30 minutes instead of 15 minutes for the sequential reaction) and left to react for 16 ha tr oom temperature as well. After defect-introduction/bioconjugation, the excess nanobody was removed using 300 kDa-MWCO-spin-filtration and the SWCNT*-GBP conjugate 5 resuspended in 1xPBS using 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-5000] (ammonium salt, PL-PEG5000). As shown in Figure S4, the redispersion step was efficient only for the conjugate 5 synthesized via the one-step approach, yielding ah ighly concentrated solution (OD = 1.8). Consequently,weproceeded with the resuspended conjugate 5 resulting from the one-step procedure in the following. Figure 3b shows the NIR fluorescence spectra of the conjugate 5 as well as the negative control, where no diazonium salt 2 was added, both resuspended using PL-PEG5000. Ac omparison of both spectra shows successful introduction of sp 3 quantum defects and E 11 * emission. To further evaluate whether these defects also contain the covalently attached nanobody,w ee mployed atomic force microscopy (AFM). Here,t he one-step-synthesized sample clearly shows nanobodies attached to the SWCNT* (see Figure 3c,control without 2 in Figure S5) with the additional height introduced by the GBP fitting both the value measured by AFM (d = 4.3 AE 0.9 nm) as well as the diameter obtained from the crystal structure (PDB:3 G9A, d = 3.4-4.3 nm, see Figure S6). In af ew cases,a lso larger heights of SWCNT*- GBP conjugates (approximately 7nm) were measured, which could be explained by the possible side reaction of the diazonium salt 2 with the GBP'saromatic residues leading to dimer formation. Taken together,t hese results indicate successful conjugation of the nanobody to maleimide-bearing quantum defects,which in turn are still able to modulate the SWCNT*sN IR fluorescence,y ielding emission at 1143 nm.
As anext step,wev erified that the nanobody is still able to bind GFP even after covalent conjugation to the SWCNT*. Therefore,w ei mmobilized GFP (patterned using polydimethylsiloxane (PDMS)-based microcontact printing) on ap oly-l-lysine (PLL)-coated glass surface,f ollowed by blocking (with bovine serum albumin [BSA])/washing steps and incubation with the conjugate 5.
Theo bserved colocalization of the GFP and the NIR channel indicates retained function of the GBP even after covalent conjugation to the SWCNT* (Figure 3e,c ontrol without the MalPh defect on the right and without GBP in Figure S7). This is,t ot he best of our knowledge,t he first covalent conjugation of af unctional (immuno)protein to aS WCNT under preserved/enhanced NIR fluorescence.
After having successfully established MalPh quantum defects as an anchor for the attachment of (immuno)proteins, we wanted to challenge this defect-based approach even  further with the aim of synthesizing peptide chains directly on the SWCNT sidewall. While there are afew reports on the use of coiled-coil or cyclic peptides for SWCNT dispersion [53][54][55] and the noncovalent immobilization of RGD motifs, [56] covalent immobilization of peptides is less explored. Pantarotto et al. and Bianco et al. utilized the 1,3-dipolar cycloaddition of azomethine ylides for the covalent modification of SWCNTs and for subsequent attachment of previously synthesized, short peptides. [57,58] However,t his approach destroys the SWCNTso ptical properties and rules out NIR fluorescence imaging applications.
To use quantum defects as as tarting point for peptide growth, we synthesized adiazonium salt containing af luorenylmethoxycarbonyl (Fmoc)-protected l-phenylalanine (Fmoc-Phe-Dz, 6)i naone-step procedure.N ext, we again optimized the conditions for defect introduction (Fmoc-Phe defects) and evaluated the success using 1D/2D NIR fluorescence spectroscopy (Figure 4b,c, control 2D spectrum in Figure S8). We carried out the Fmoc deprotection using 20 % piperidine/DMF in a1mL syringe reactor equipped with astandard 20 mmpore-size frit. After washing, we coupled the fluorophore 5(6)-carboxyfluorescein (CF) to assess addressability of the unprotected amine.F igure 4d shows colocalization of the NIR and the CF channels after immobilization on glass and washing steps using 1xPBS with 0.1 %T riton-X-100 as opposed to the negative controls [without 6 ( Figure 4d)or without Fmoc deprotection ( Figure S9)].T his result shows that the unprotected amine is still addressable and the conjugation of carboxyfluorescein led to the generation of covalently linked multicolor SWCNTs*.
Encouraged by these promising results,wewanted to test next whether it is also possible to synthesize awhole peptide sequence on the Fmoc-Phe-SWCNT* 7.H ere,w ec hose apositively charged hexaarginine peptide to also evaluate its impact on SWCNT* solubility in aqueous environments (Figure 5a). To evaluate the success of the SWCNT*-peptide synthesis,n either the Kaiser test for free amines nor UV measurements after Fmoc cleavage could be used due to their insufficient sensitivities on the small scale of these experiments (n SWCNT = 100 pmol in 1mLsolution). Thus,wedecided to couple 5(6)-CF to the N-terminus before global deprotection of the argininess ide chains using ad eprotection cocktail [75 %t rifluoroacetic acid (TFA), 15 %d ichloromethane (DCM), 5% ddH 2 O, 5% triisopropylsilane (TIS)]. This was followed by tip-sonication in 1xPBS (3 min, 30 % amp,4 8 8C) and centrifugation (16 100 g, 30 min) to remove insoluble SWCNTs.F igure 5b shows colocalization of the NIR and the CF channels on the single-nanotube level, indicating successful synthesis of SWCNT*-F-R 6 -CF in contrast to the control (without 6,s ee Figure S10) and retained optoelectronic properties after TFAdeprotection. In fact, the negative control did not contain any SWCNTs,i ndicating increased solubility in aqueous environments by covalent peptide functionalization. However,due to the small number of defects (approximately one defect per 20 nm tube [30] )t he SWCNT*-F-R 6 -CF (8)did not display high solubility in water and therefore additional wrapping was used to increase the concentration and colloidal stability.I nf uture studies,t his aspect could be further evaluated with higher defect densities and/or longer peptide sequences.
As an ext step we scaled up the SWCNT*-based peptide synthesis and synthesized multiple SWCNT*-peptide conjugates at the same time in a9 6-well format (equipped with 0.2 mmp ore size filters,F igure 5c). Again, the synthesis followed the same protocol as above,yet with smaller reaction volumes.U sing this technique,w es ynthesized twelve different peptide sequences directly on NIR fluorescent carbon nanotubes.A fter side-chain deprotection using the same deprotection cocktail as for SWCNT*-F-R 6 -CF,t he carbon nanotubes were redispersed in an aqueous 1% DOC solution via tip-sonication. While DOC leads to slightly red-shifted emission compared to SDBS-dispersed SWCNT*-F-Fmoc/ SDBS (7), [59,60] both Figure 5d and es how the impact of peptide sequence on the NIR fluorescence.F or some sequences the original E 11 peak was almost twice as intense as the E 11 * peak-other sequences showed exactly the opposite behavior with the E 11 * signal being 2.5-fold stronger than the signal arising from the E 11 transition (Vis/NIR absorbance and NIR fluorescence spectra in Figure S11). A closer evaluation of the sequence dependence of this fluorescence modulation shows that the E 11 * /E 11 ratio decreases with an increasing number of hydrophobic residues in the peptide sequence attached to the defect responsible for exciton trapping (Figure 5f). As imilar effect was already observed by Kwon et al.,w ho found changing E 11 * emission wavelengths and E 11 * /E 11 ratios for differently substituted (fluoro)alkyl/aryl sp 3 defects. [32] This interesting impact on the SWCNT photophysics could be attributed to the peptides folding differently on the SWCNT and changing the charge landscape through which the exciton diffuses or where it gets trapped, thus leading to enhanced E 11 * fluorescence for less hydrophobic sequences.D ifferent folding of peptides is known from noncovalent SWCNT/peptide hybrids. [61] Furthermore,acomparison of the sequences 9-12 consisting of identical amino acids shows that not only the nature of the attached amino acids,but also their sequential arrangement is of high importance for the SWCNT* NIR fluorescence properties.T hese results demonstrate the possibilities of employing Fmoc-protected phenylalanine defects for the growth of peptidic chains directly on the nanotubessidewall and indicate that this method can not only be used for modulation of the SWCNTs fluorescence,b ut also to tailor their surface properties.T his in turn could enable SWCNTs with enhanced cellular uptake/retention, [62,63] tailored molecular recognition motifs,a nd novel and more stable optical sensors operating in the NIR. Furthermore,t he coupling of asecond optically active molecule (fluorophore) via apeptide sequence to aSWCNT could serve as general design principle for molecular recognition and signal transduction. Similar to Fçrster resonance energy transfer (FRET), conformational changes upon binding to at arget structure could affect the SWCNTs NIR fluorescence and enable novel fluorescent probes and labels.

Conclusion
In summary,weintroduced two new sp 3 quantum defects in SWCNTs,w hich serve as anchors for the attachment of biomolecules.T he versatility of this new functionalization platform was demonstrated by conjugation of aGFP-binding nanobody as an example for ap rotein and the synthesis of peptides directly on the carbon nanotube surface.T his new technique for covalent decoration of SWCNTs with biomolecules opens up great possibilities for applications in (bio)photonics,b iosensing,and biomedicine. Figure 5. In situ peptide synthesis and modulation of E 11 *peak intensities. a) Strategy for the generation of covalent and fluorescentSWCNT*-F-R 6 -CF conjugates based on Fmoc/O t Bu-solid-phase peptide synthesis (SPPS) followed by N-terminal CF coupling and Pbf deprotectiono fthe arginine side chains. b) SWCNT*-R 6 -CF spin-coated on aglass coverslip showing colocalization of the NIR and the CF channels, indicating successful peptide synthesis and N-terminal CF coupling directly on the SWCNT sidewall (scale bars = 10 mm). c) 96-Well peptide synthesis for the generation of aS WCNT-peptide pool following the same Fmoc/O t Bu-SPPS protocol as shown in (a), yet here in a96-well plate with filters (0.2 mm pore size). d) Normalized NIR-PL spectra before and after synthesis of two selected peptide sequences showing the modulation of the defectinduced fluorescence. The red-shift from the protected to the unprotected sample could be attributed to the different surfactants (SWCNT*-F-Fmoc:S DBS, SWCNT*-Peptide:D OC). e) The SWCNT fluorescence properties (in particular the E 11 */E 11 ratio) depends on peptide sequence on the sidewall (mean AE SD, n = 3). f) E 11 */E 11 ratio increases with the number of hydrophobic residues (mean and individualv alues, n = 6, 2, 4).

Conflict of interest
F. Opazo is shareholder of NanoTag Biotechnologies GmbH; all other authors declare no conflict of interest.