Stretching the Bisalkyne Raman Spectral Palette Reveals a New Electrophilic Covalent Motif

Abstract Small heteroaryl‐diyne (Het‐DY) tags with distinct vibrational frequencies, and physiologically relevant cLog P were designed for multiplexed bioorthogonal Raman imaging. Pd−Cu catalyzed coupling, combined with the use of Lei ligand, was shown to improve overall yields of the desired heterocoupled Het‐DY tags, minimizing the production of homocoupled side‐products. Spectral data were in agreement with the trends predicted by DFT calculations and systematic introduction of electron‐ rich/poor rings stretched the frequency limit of aryl‐capped diynes (2209–2243 cm−1). The improved Log P of these Het‐DY tags was evident from their diffuse distribution in cellular uptake studies and functionalizing tags with organelle markers allowed the acquisition of location‐specific biological images. LC–MS‐ and NMR‐based assays showed that some heteroaryl‐capped internal alkynes are potential nucleophile traps with structure‐dependent reactivity. These biocompatible Het‐DY tags, equipped with covalent reactivity, open up new avenues for Raman bioorthogonal imaging.


Introduction
Raman microscopy is a powerful optical analytical method that measures the vibrational composition of a biological sample, allowing both non-destructive and non-invasive imaging of biomolecules. [1,2]Alkynes have a large Raman scattering crosssection in the cell-silent region and are seldom found in Nature, rendering them an important spectroscopically bioorthogonal handle. [3,4]In contrast to most fluorophores, their unidirectional vibrational mode gives very narrow Raman vibrational bands (FWHM ~20 cm À 1 ) [5] allowing them to be readily adapted to multi-channel imaging.[26][27][28] Compared to spontaneous Raman, SRS offers improved sensitivity, spectral resolution and imaging speeds.SRS, used in tandem with either polyyne-based Raman probes, or triple-bond-conjugated near-infrared dyes coupled with isotopic substitutions, has allowed up to 24 resolvable 'colours' to be imaged. [16,29]However, notwithstanding the many advantages, the large size, rigidity, planarity, low solubility and high lipophilicity of many polyyne probes challenge their wider application.When conjugated to important biomolecules or small molecular therapeutic agents, potential problems include, but are not limited to, perturbation of important biological functions including uptake, localization and target binding. [30,31]urthermore, the vibrational intensity variation with increasing alkyne units varies by multiple orders of magnitude, requiring multiple instrumental adjustments during the image acquisition process.
While alkyne extension and 13 C doping (Figure 1a) have been extensively exploited for frequency tuning of alkyne tags, [16] an approach that has been largely overlooked is the tuning of π-electron delocalization via modulation of end-cap electronics.Electron-withdrawing and electron-donating ring substitutions have been shown to cause minor frequency shifts in aryl end-capped alkynes, however in themselves these are not sufficient to facilitate multiplex imaging. [16,32]In the current study (Figure 1b), we sought to demonstrate that electronic modulation of the end-cap phenyl rings of BADY through heteroaryl substitution offers an alternative to the current strategies for multiplexed imaging.

Het-DY tag design
A significant challenge to the use of BADY or polyyne structures for bio-imaging is their strong tendency to form π-π stacked aggregates.Using ESI-MS analysis and a series of all-carbon BADY analogues, we recently showed that reduced compound aggregation correlates favorably with an improved cLog P. [33] This earlier study underlines the pivotal need for improved physicochemical properties of conjugated alkyne Raman tags if they are to enter more widespread use.We predicted that electronic modulation via systematic replacement of the phenyl rings in BADY with heteroaromatic rings would not only induce π-electron delocalization to generate new Het-DY tags with frequencies tuned for multiplexed Raman imaging, but would also improve their log P while simultaneously maintaining their small molecular size (Figure 2a).
A library of Het-DY tags was designed, with the phenyl endcaps in BADY replaced either by electron-rich rings (5membered heteroaromatic rings in place of the phenyl ring), or electron-poor rings (nitrogen atom incorporation at the ortho-, meta-and/or para-positions in the phenyl ring relative to alkyne substitution) to tune their Raman vibrational frequencies.Library members A1-A94 (Supporting Information Table S1) were modeled using density functional theory (DFT) as their amide derivatives to best represent linkages between the tags and a target of interest in future applications and to enhance signal intensity. [3]Het-DY tags 3 a-n (Figure 2b and c, Table 1) were selected from this library for experimental evaluation.Ring modifications cause red and blue shifts relative to BADY (black); 13 C isotopes or alkyne extension cause red shifts.b) Systematic replacement of phenyl rings in BADY (black) with electron rich (red) or electron poor (cyan and blue) heteroaromatic rings widens the diyne spectral palette (2209 cm À 1 -2243 cm À 1 ).

Het-DY tag synthesis
Terminal alkynes 1 and activated haloalkynes 2 are the key intermediates required for the PdÀ Cu co-catalyzed synthesis of the selected 1,3-diynes 3 a-n; non-commercial coupling partners were synthesized as shown in Scheme 1.Where possible, terminal alkynes 1 were accessed directly by Seyferth Gilbert homologation of commercial aldehydes 4. Alternatively, Sonogashira coupling of commercial aromatic halides 5 and TMS acetylene under microwave conditions afforded the silyl protected alkynes 6. TMS-deprotection of these mono-substituted arylalkyne intermediates under basic conditions gave terminal alkynes 1. Activated haloalkynes 2 were obtained either by direct halogenation of commercial terminal alkynes (Supporting Information general procedure C), or by AgNO 3catalysed halodesilylation of ester-substituted TMS-alkyne intermediates 6 with NIS/NBS (Scheme 1).
The key step in the synthesis of heterodimeric 1,3-diynes is a C(sp)-C(sp) cross-coupling, which is often achieved via Cucatalysis in the presence of an excess (3-5 equiv.) of one of the alkyne coupling partners. [34]PdÀ Cu co-catalysed coupling of terminal alkynes with activated bromo/iodoalkynes affords greater selectivity for synthesis of the unsymmetrical 1,3-diyne without requiring the large stoichiometric excess. [34,35]In the current study, we employed the PdÀ Cu coupling conditions reported by Lei et al. for the synthesis of 3 b-n. [35][36][37] Excess alkyne coupling partners (1, 2) were avoided, homocoupled side-products and the difficulties associated with chromatographic separation of these side-products from heterocoupled products 3 were also minimized.However, the yield of cross-coupling reactions was lower in heteroaromatic substrates and steadily reduced upon each additional end-cap ring nitrogen incorporated ortho to the alkyne (3 e, 3 i and 3 j) (Supporting Information Table S3). [38]The synthesis of compound 3 k in which all four ortho-positions are occupied by nitrogen, failed.Replacing Lei ligand with the bulky (t-Bu) 3 P ligand, which also facilitates faster reductive elimination, enabled its isolation albeit in a low yield.The susceptibility of the electron rich rings of the pyrrole or imidazole alkynes to halogenation under the NXS-AgNO 3 halodesilylation conditions required an alternative synthesis of 3 a, which was achieved using a Cu/DMAP-catalyzed Glaser coupling (Supporting Information general procedure D1) [36] and an excess of the N-methyl imidazole capped alkyne (5 equiv.) to afford the tag 3 a (50 %).

Het-DY tag Raman properties
Het-DY tag 3 a, end-capped with electron rich N-methyl imidazole and N-methyl pyrrole rings (Figure 2b), was predicted by DFT calculations to show a significant red shift in its vibrational frequency compared to BADY (Table 1).This red shift is systematically reduced (Table 1, Supporting Information Table S1) as the electron rich rings are replaced with relatively electron poor rings in 3 b-d (Figure 2b).Whilst the trend for the experimental spectral maxima of Het-DY tags 3 a, 3 b and 3 d  S1. was in agreement with DFT predictions (Table 1, Figure 3), the frequency range was somewhat reduced (3 a predicted 2198 cm À 1 , experimental 2209 cm À 1 ) and 4-thiazole capped Het-DY tag 3 c appeared out of sequence suggesting limitations to current DFT calculations.In contrast, end-capping with electron poor rings is predicted to cause a blue shift (Table 1, Supporting Information Table S1) compared to BADY.DFT calculations for the Het-DY tags 3 e-g (Figure 2b) indicated that a single nitrogen incorporation, depending on its regioisomeric position (ortho-and para-to the alkyne in 3 e and 3 g respectively) causes a modest blue shift in frequency of 5 cm À 1 compared to BADY (Table 1).Further 5 cm À 1 blue-shifts were predicted with additional ortho-N-incorporations in 3 i-k (Figure 2b).Experimental maxima in the Raman spectra for tags 3 e-k were in close agreement with the DFT predictions (Table 1, Figure 3) and the trend of step-wise increments in frequency going from one to four ortho-N-incorporations.Since DFT predictions showed that meta-N-incorporation did not substantially alter the Raman vibrational frequency, tags 3 l-n (Figure 2c) were chosen as frequency matched analogues of tags 3 h-j with improved physicochemical properties (cLog P < 2) (Table 1).

Intracellular Raman activity of Het-DY tags
The relative intensities of BADY and Het-DY tags in DMSO solution (10 mM) compared to the internal standard EdU (100 mM) indicate that the high signal intensities of aryl-aryl end-capped diynes are largely retained (Table 1, Supporting Information Figure S1).The cLog P values of the Het-DY tags vary by > 3 Log units compared to BADY and fall within the optimum range of 0-3 for biological uptake and distribution of small molecules (Table 1, Supporting Information Table S1). [39]In cells, BADY (cLog P 4) accumulates in lipid droplets, (Figure 4a).While the localization of 3 f (cLog P 2.7) appears unchanged compared to BADY (Figure 4b), the more diffuse distribution of tag 3 a (cLog P 0.5) was predominantly in the cytoplasm which may be explained by the greater reduction in cLog P (Fig- ure 4c).The localization of BADY can be altered by tagging organelle targeting motifs to disparate parts of the cell, for example plasma membrane, mitochondria, lipid droplets and lysosomes (Supporting Information Figure S2).Several of the Het-DY tags including 3 a (2209 cm À 1 ), 3 e (2225 cm À 1 ), 3 f (2219 cm À 1 ), 3 i (2230 cm À 1 ) and 3 k (2243 cm À 1 ) were selected for functionalization as organelle markers for multiplexed biorthogonal imaging.These markers were synthesized via ester hydrolysis of the Het-DY tags 3 to the corresponding acids 7, followed by amide coupling to afford the organelle specific markers 8 (Supporting Information Experimental).Despite the expected improvement in Log P due to heteroatom substitutions and concomitant improvement in intracellular distribution of the tags, imaging with some of the Het-DY tags proved to be surprisingly challenging.Notwithstanding, tags 3 a and 3 f functionalized as imaging probes of lysosomes (Lyso-Het-DY, 8 a) and lipid droplets (LD-Het-DY, 8 f) were enriched in lysosomal (Figure 5a) and lipid rich regions (Figure 5b

Covalent reactivity of Het-DY tags
Terminal alkynes have been shown to have latent reactivity towards thiol nucleophiles. [40]Alkynyl heterocycles with terminal  and methyl-substituted acetylenes designed to mimic Michael acceptor-like systems show cysteine-selective reactivity. [41,42]hile reduced cell viability of some of the Het-DY tags provided an early indication, their electrophilic nature was first revealed in the process of base-catalyzed hydrolysis of 3 k.Aqueous hydrolysis of 3 k resulted in multiple degradation products.However, methanolic hydrolysis under mildly basic conditions provided clear NMR evidence of the formation of Michael addition products (Figure 6a and Supporting Information Experimental).LC-MS based glutathione (GSH) reactivity assays (Figure 6b and c) of BADY and Het-DY tags, with reaction progression recorded at 5 min, 2 h and 24 h to identify GS-Tag adduct formation, shed light on their (a) cysteine trapping reactivity and (b) relative reactivity.Perhaps surprisingly, given the widespread use of the aryl-capped diynes in ATRI bioimaging to date, all the tags, including BADY, formed GS-adducts (Supporting Information Figure S4), albeit at different rates.Tags 3 a and BADY were most stable, showing relatively low (1-2 %) conversion to the adduct over 24 h.Tag 3 d formed (~25 %) GS-adduct within 5 min of GSH addition and reacted completely over 24 h.Correlating with the trend in increasing electron deficiency from 3e to 3 k, 3 e formed ~25 % adduct over 24 h, and 3 j and 3 k reacted completely within 5 min of GSH addition.Finally, NMR based GSH-reactivity analysis of 3 a, BADY and 3 k reaffirmed their relative reactivity.Despite forming GS-adducts as visible on LC-MS, 3 a and BADY show no changes in their 1 H NMR 24 h post GSH addition, indicating their low reactivity.However, in accordance with results from the LC-MS assays, within 5 min of GSH addition, 1 H NMR of 3 k showed multiple new peaks in the region corresponding to the alkene CH(sp 2 ) protons of the Michael addition products (Supporting Information Figure S5).Together, the above results suggest that diynes can be tailored via careful end-cap modifications as biocompatible imaging agents for multiplexing, or as covalently reacting Michael acceptors via conjugation with electron-deficient heterocyclic end caps.

Conclusions
We have rationally designed and synthesized Het-DY tags 3 a-n for fine-tuned Raman frequencies.A frequency difference of 34 cm À 1 was achieved between the two diyne tags 3 a and 3 k, which is considerably larger than single 13 C incorporation (1 0 cm À 1 ) and marginally higher than dual 13 C incorporation (2 0 cm À 1 ).To fully exploit diyne ATRI for multiplexing, we recommend incorporations of electron rich end-caps in the design of new diyne Raman probes.The many advantages of the PdÀ Cu co-catalyzed synthesis of 1,3-diynes (i.e., clean, convenient, efficient, and versatile), validate this method for further screening in the synthesis of a broad range of unsymmetrical 1,3-diynes, specifically extending its scope to heteroaromatic substrates.We also demonstrate for the first time that electron deficient, aryl-capped diyne Raman tags are nucleophile traps.Covalent capture alkyne probes offer an excellent opportunity to develop new and unexplored avenues in ATRI including high resolution imaging due to cellular trapping, real time tracking aided by the shift in the Raman activity of the alkyne and the alkyne-nucleophile adducts, and the determination of covalent reaction kinetics in cellulo.

General procedure for Pd-catalyzed Cadiot-Chodkievicz (CC) cross-coupling
To an oven-dried 2-necked RBF with an oven-dried PTFE-coated magnetic stir-bar, Pd 2 (dba) 3 (4 mol %), Lei ligand (4 mol %), and CuI (2 mol %) were added.Anhydrous DMF (2 M) was added via a syringe and the mixture vacuum purged with nitrogen for three cycles.After stirring the mixture under nitrogen for 10 min, a vacuum purged and N 2 -filled solution of terminal acetylene 1 (1.2 equiv.), in anhydrous DMF (1 M) was added via a syringe, followed by TEA (2 equiv.).The reaction mixture was stirred for another 5 min, then a vacuum purged and N 2 -filled solution of haloacetylene 2 (1.0 equiv.) in anhydrous DMF (1 M) was added last via a syringe.The system was stirred at room temperature under N 2 for 4 h.Reaction progress was monitored by TLC, which showed loss of starting material and appearance of three new spots.Upon completion, a minimum amount of MeOH and Celite (3 × weight of the crude) were added and the solution was evaporated to afford a plug.The resulting plug was loaded on to an automatic flash column for purification.Fractions with desired R f (TLC) were pooled and evaporated to afford the heterocoupled products BADY, 3 b-n. 45][46][47][48][49][50][51][52][53][54][55][56][57][58][59][60]

Figure 1 .
Figure 1.a) Raman probes from the Min and Graham groups.[16,32]Ring modifications cause red and blue shifts relative to BADY (black);13 C isotopes or alkyne extension cause red shifts.b) Systematic replacement of phenyl rings in BADY (black) with electron rich (red) or electron poor (cyan and blue) heteroaromatic rings widens the diyne spectral palette (2209 cm À 1 -2243 cm À 1 ).

Figure 2 .
Figure 2. a) Het-DY tag design workflow with green forward arrows representing developments highlighted in this paper and blue dotted arrows feedback points in Het-DY tag development; b) Structures of Het-DY tags 3 a-k, electron rich and electron poor end-cap rings cause a red-shift and blue-shift in the Raman vibrational frequency of alkynes; c) Structures 3 l-n were designed to improve the physicochemical properties of the tags (cLog P < 2) 3 h-j, while retaining the incremental changes in Raman shift.
[c] + + and + represent RIE ranges of 20-30 and 10-20, respectively.[d] The cLog P for each tag was calculated using ChemDraw 17.1.[e] Difficult to purify from homodimers and concentration of pure fractions not sufficient to acquire spectra with good intensity.[*] Tag 3 c designed as an electron rich end-capped Het-DY showed a blue-shifted Raman frequency compared to BADY.
) of the cells respectively as expected.Viability studies of Het-DY tags 3 a-n were performed in ES2 cells (EC 50 0.54 -> 100 μM, Supporting Information Figure S3).The EC 50 values following incubation with the Het-DY tags for 72 h show that only the alkyne tags with electron rich (3 a and 3 b) or neutral (BADY) and meta-N-incorporated endcaps (3 f) are tolerated well (EC 50 values � 100 μM).Viability reduced with an ortho-or para-nitrogen in the end cap relative to the alkyne: 3 e (EC 50 17.57μM) and 3 g (EC 50 24.10μM); or with increased number of ortho-nitrogens: 3 i (EC 50 1.63 μM), 3 j (EC 50 0.54 μM) and 3 k (EC 50 2.35 μM).

Figure 6 .
Figure 6.a) Methanolic hydrolysis of 3 k at room temperature resulted in Michael addition; only one of the possible Michael addition products is shown.b) Representative LC-MS spectrum with detection at 254 nm; formation of GS-Tag adduct in ACN:PBS buffer (1 : 1) at 37 °C.c) Relative reactivities of tags 3 a, 3 d, BADY, 3 e, 3 j and 3 k calculated at 5 min, 2 h and 24 h post addition of the diyne tags to GSH in ACN-PBS buffer at 37 °C.Tags 3 a and BADY formed 1-2 % adducts over 24 h whilst tag 3 e showed ~25 % conversion.Tags 3 d, 3 j and 3 k showed complete conversion to GS-adducts; 3 j and 3 k showed complete conversion within 5 min post addition.

Table 1 .
Calculated Raman shifts; experimental Raman shifts of 10 mM DMSO solutions; relative intensities of 10 mM DMSO sample solutions compared to internal standard EdU (100 mM); and cLog P values of Het

-DY tags 3 a-n.
[a] DFT calculations at B3YLP 6-31G(d,p) level with the 6-31G(d,p) doublezeta plus polarization basis set were performed on the amide linked tags to accommodate frequencies and intensities of biologically imaged functionalized tags.[b] For the full list of DFT calculated Raman frequencies and relative signal intensities see Supporting Information Table