Insertion of organometallic moieties into peptides and peptide nucleic acids using alternative “ click ” strategies †

The insertion of metal complexes in biologically active systems is of great interest in view of diagnostic and therapeutic applications as well as a precious tool to unveil biological mechanisms. Optimization of safe and biocompatible reactions is critical to achieve high functionalization e ﬃ ciency. Herein we present the application of two modi ﬁ ed versions of copper-catalyzed azide – alkyne cycloaddition ( click ) chemistry, namely a one-pot diazotransfer + azide – alkyne cycloaddition ( one-pot click ) and a copper-free photoactivated tetrazole – alkene cycloaddition ( photoclick ), for derivatization of peptides and peptide nucleic acids (PNAs) with ferrocene and cymantrene moieties. These metal fragments were chosen for their possible exploitation as redox and IR probes. We could demonstrate that one-pot click enables for e ﬃ cient functionalization of propargyl-glycine and an alkyne-containing peptide with an amino-contain-ing cymantrene precursor. In addition, we could show that photoclick allows for the insertion of male-imido-ferrocene into a peptide and a PNA sequence containing a tetrazole moiety.


Introduction
The functionalization of biologically active molecules is of great importance nowadays in medicinal chemistry. [1][2][3] In particular, their functionalization with metal-based compounds can confer novel properties to the formed bioconjugate compared to the biomolecule itself allowing for novel applications. 4 Metal complexes can be used as both diagnostic and therapeutic tools in the clinical treatment of patients. 5 The coupling of a targeting vector (i.e. peptide, antibody, etc.) to therapeutic/diagnostic metal-based drugs can, for example, increase their organ/cell selectivity. [6][7][8][9][10] In a similar vein, the insertion of metal fragments into biologically active molecules can yield biocompatible and extremely useful tools for biological investigations, such as luminescent probes. 11 The application of copper-catalyzed azide-alkyne cycloaddition (CuAAC, commonly referred to as click chemistry) for bioconjugation has been extensively exploited due to the high versatility, selectivity and functional group tolerance of this synthetic technique. 12 A great number of probes, including metal-containing probes, were "clicked" to an important variety of biologically active compounds such as peptides, antibodies, peptide nucleic acids (PNAs) or polymers. [13][14][15][16][17] However, this reaction presents two important drawbacks from both a chemical as well as a biological point of view. Since an azide is one of the protagonists of this transformation, this functional group

Gilles Gasser
Gilles Gasser completed a PhD thesis in supramolecular chemistry in the group of Prof. Helen Stoeckli-Evans at the University of Neuchâtel (2004). After postdoctoral stays at Monash University (Australia) with Prof. Leone Spiccia in bioinorganic chemistry and at the Ruhr-University Bochum (Germany) with Prof. Nils Metzler-Nolte in bioorganometallic chemistry, Gilles started his independent research career at the University of Zurich, first as Swiss National Science Foundation (SNSF) Ambizione fellow (2010) and then as a SNSF Assistant Professor (2011). Gilles is the recipient of several awards including the Werner Prize from the Swiss Chemical Society for outstanding independent chemical research (2015) and an ERC Consolidated Grant (2015).
has to be installed on one of the two molecules to be coupled. Safety concerns about the isolation of azides (especially with low molecular weight compounds) are still actual nowadays due to the potential instability of this class of compounds. 18 Therefore, from a chemical point of view, it would be more convenient to skip the azide isolation, for example, via an in situ insertion of this functionality. The second disadvantage of the CuAAC is its reliability on the presence of Cu I as catalyst for its mechanism to take place. This situation represents a problem for in cellulo applications of the technique due to the toxicity of the metal. 13,19 To address these limitations, a great effort has been made towards the optimization of new methodologies for clickmediated biocompatible conjugation. 13,20,21 Since the absence of a Cu catalyst makes the click reaction very inefficient, an alternative activation mechanisms had to be developed. Among the most relevant examples, Strain-Promoted Inverse Electron Demand Diels-Alder Cycloaddition (SPIEDAC) and Strain-Promoted Alkyne-Azide Cycloaddition (SPAAC) exploit the formation of an unstrained product as reaction driving force, circumventing the need of Cu catalysis. 19 Other approaches involve the use of light to form a reactive intermediate. 21 Herein, we present, to the best of our knowledge, the first examples of the application of two modified versions of the classical CuAAC to introduce organometallic fragments into biologically active molecules, namely a peptide and a PNA oligomer (see Fig. 2). In particular, we focused our attention on two modified click reactions, namely the copper-catalyzed onepot diazotransfer + azide-alkyne cycloaddition reaction (which will be called from now on in this article one-pot click) 22 and the metal-free photoactivated tetrazole-alkene cycloaddition (referred to from now on as photoclick). 23 The two organometallic components investigated in this study were a cymantrene derivative A and a ferrocene derivative B (Fig. 1), 24,25 chosen for their potential use as IR (A) and redox (B) probes in a biological environment. 26,27 Results and discussion

One-pot click reaction
The one-pot click reaction developed by Wittmann et al. allows in a subtle manner to circumvent the isolation of the azide intermediate. 22 More specifically, the one-pot click proceeds via two subsequent steps. First, an amine derivative is turned into an azide via a Cu II catalyzed diazotransfer. This intermediate can then be clicked to an alkyne by the classical CuAAC reaction. The Cu I catalyst for the click reaction is formed in situ by addition of a reducing agent to the mixture. In the present study, we applied this synthetic technique to the two aminoorganometallic derivatives A and B. These molecules were clicked to an unnatural amino acid carrying a propargyl group (Fmoc-L-Pra-OH) and a short peptide containing this amino acid (NLS-Pra, Fig. 2). This small peptide is known for its nuclear targeting properties and has been used in the past by our group to deliver different metal complexes to the cell nucleus. 6,28,29 NLS-Pra was synthesized by standard solid phase synthesis and the alkynyl containing unnatural amino acid was inserted at the N-terminus. 28 The original one-pot click procedure reported by Wittmann et al. was followed in this study with the exception of minor modifications: the amounts of copper salt, tris(benzyltriazolylmethyl)amine (TBTA) and sodium ascorbate were doubled and the reaction duration was extended to 1 h of microwave assisted refluxing, instead of 30 minutes. When A was used as the amine precursor, the reaction was found to be effective for labelling both the amino acid and the peptide (Fig. 3).
More specifically, product 2A was isolated in 50% yield and its structure confirmed by several analytical techniques. The UPLC-MS chromatogram showed one peak, whose integrated mass of m/z 694.2 is attributable to [M + H] + . The nature of 2A was further confirmed by HR ESI-MS, in which the peak of [M − H] − is visible. In addition, the IR spectrum clearly showed the characteristic vibrations of the CuO groups at 1921 and 2015 cm −1 . However, due to the presence of traces amounts of a paramagnetic impurity, we were unable to obtain a good 1 H-NMR spectrum. In the 13 C-NMR, however, resonances at 55.73 and 29.22 ppm can be attributed to the CH 2 and CH carbons between the newly formed triazole and the COOH moiety. The peptide derivative 3A was isolated via preparative RP-HPLC purification. UPLC-MS analysis showed

Photoclick reaction
As a parallel approach to the one-pot click reaction, we investigated the applicability of the photoclick reaction to our metal precursors, to pursue a more biocompatible method for organometallic labelling. This procedure exploits the lightmediated formation of a reactive acyl-nitrene intermediate from a tetrazole group. This activated intermediate is then prone to 1,3-dipolar cycloaddition with a substituted alkene. 23 This reaction belongs to the class of photo-mediated orthogonal reactions, which have been extensively exploited for in vivo applications due to their high selectivity and the lack of the toxic copper catalyst. 20,21,[33][34][35] The diphenyltetrazole group 7, which was synthesized as previously reported, 36 was installed on the PNA backbone 8, 37 to obtain a tetrazole-containing PNA monomer as a t-Bu ester 9 (Fig. 4). Removal of the t-Bu yielded compound 10, which can be inserted into a PNA or a peptide sequence by solid phase synthesis. The initial design of the system envisaged the use of a phenyl-allyloxy derivative as the alkene-reactive component. However, this molecule was found to be poorly reactive and only afforded the products in low yields. For these reasons, we decided to employ a maleimide group as alkene functionality. Maleimide derivatives of cymantrene and ferrocene (4A and 4B, Fig. 5) were therefore prepared and "photoclicked" with the protected PNA monomer 9 bearing the tetrazole reactive moiety. The Fc-containing compound 5B was isolated in 63% yield and fully characterized. The UV chromatogram of UPLC analysis showed one peak and the HR ESI-MS showed a peak at m/z 912.3041, corresponding to [M + H] + (calculated: 912.3054). On the other hand, the Mn derivative 5A was isolated from the crude as a minor component (8.5%), whereas the main product of the reaction (40%) was still a product of photoclick lacking, however, the Mn(CO) 3 moiety (m/z 891.5 [M − Mn(CO) 3 + H] + ). Comparison of the IR spectra confirmed the absence of the CuO vibrations in the spectrum of the product isolated in higher amount, signals which are well visible in the IR spectrum of 5A (1927 and 2019 cm −1 ). This outcome is probably due to the known ability of cymantrenyl compounds to release CO upon light irradiation. 38 Therefore, it is likely that the product of the photoclick decomposes during the reaction. Motivated by the successful formation of 5B, we attempted the insertion of ferrocene-maleimide in a NLS peptide carrying   the modified PNA monomer (NLS-tetra, Fig. 2), synthesized by solid phase synthesis. 28 In a procedure reporting on the modification of a DNA strand with a maleimide functionalized dye via photoclick, this reaction was performed in PBS affording yields in the range of 8-34% depending on the conditions used, whilst the use of DMSO as solvent did not afford the desired product. 39 Surprisingly, when we performed the reaction in water or ACN : PBS (5 : 1), after just 30 min irradiation NLS-tetra was found to have entirely turned into the hydrolysis byproduct of the acyl-nitrene intermediate (Fig. 5

PNA functionalization with photoclick
In order to assess the full potential of the photoclick synthetic procedure, we applied it to a PNA sequence. PNAs are DNA/ RNA mimics which were first reported by Nielsen et al. 40 They have gained significant interest owing to their ability to hybridize with DNA/RNA complementary strands with very high affinity, their proven stability to nuclease and protease degradation and their lack of unspecific interactions with proteins. Thanks to these characteristics, PNAs are very promising for several applications ranging from nanotechnology 41,42 to in vivo antisense therapy 43 and tumour pre-targeting. 8 However, there is room for improvement of the potential of PNAs in biological applications, for example by increasing their low solubility in physiological environment or by equipping them with a probe for cellular localization. It is therefore important to develop techniques for PNA functionalization. In particular, substitution with metal complexes have been of great interest to exploit additional properties of the new PNA-metal complex conjugate. 8,15,16,37,[44][45][46][47][48][49][50][51] In this work, we applied photoclick to achieve the insertion of a ferrocene moiety into a PNA sequence that was previously applied as an antisense agent on hepatitis B virus. 52 This sequence, which was synthesized via solid phase synthesis, 53 was further modified with the insertion of the PNA monomer bearing the modified tetrazole and a cysteine for conjugation via Michael addition (PNA, Fig. 2). The photoclick reaction was performed several times to optimize the conditions and the structures of the isolated products are reported in Fig. 6. Having the PNA sequence as substrate, the reaction showed a similar behaviour to that reported previously with the DNA strand. 39 When DMSO was used as solvent, the PNA starting material was converted to the hydrolyzed form of the acyl-nitrene intermediate 14. Therefore, in the second trial, the reaction was performed in water with 1 eq. of 4B. After 1 h irradiation, we could isolate all the products reported in Fig. 6. The photoclick product 11, containing a single Fc moiety, was isolated in 6.5% yield. The doubly Fc labelled PNA 12 was also formed in 8% yield, due to the occurrence of both photoclick and Michael addition. Furthermore 13, a PNA containing the hydrolyzed tetrazole and the ferrocene conjugated at the cys group via Michael addition, was isolated in 14% yield. Due to the presence of water, the main component of the mixture was found to be the unlabelled hydrolyzed product 14 (30%). Next to these components, two compounds with m/z 5380 and 5566, respectively were isolated but their structures were not elucidated. In the attempt to improve the labelling efficiency, increasing amounts of 4B were used (5 eq.). The yield of formation of the photoclick product 11 could be increased to 10%. Additionally, the double Fc containing 12 was isolated in 2.6% yield, the hydrolyzed intermediate 14 in 17% yield and 13 in 36% yield. The component with m/z 5566 was also formed. As became evident from the last reaction conditions used, it is possible to increase the formation of the photoclick product by increasing the amount of the Fc-maleimide starting material. The kinetic of the competing Michael reaction seems however to be faster than the one of the photoclick, therefore, higher yields could be obtained if the cysteine would not be present.

Instruments and methods
All commercial chemicals were used without further purification. 1

Syntheses
A, B, 7, 8 were synthesized as previously described. All analytical data match with that reported in the literature. 24,25,36,37 Fmoc-PNA-tetra-OtBu 9. The protected PNA backbone 8 as the chloride salt (517 mg, 1.195 mmol, 1.0 eq.) was dissolved in DCM (100 mL) and washed with NaHCO 3 (2 × 40 mL). The aqueous solution was back-extracted with DCM (2 × 30 mL). The organic phases were reunited and evaporated using a rotary evaporator to yield a yellow oil. 7 (350 mg, 1.31 mmol, 1.1 eq.) was dissolved in 10 mL of DMF. HATU (500 mg, 1.31 mmol, 1.1 eq.), HOBt (275 mg, 2.03 mmol, 1.7 eq.) and DIPEA (340 μL, 2.03 mmol, 1.7 eq.) were then added to this solution and the mixture stirred for 20 min. The yellow oil containing 8 in 5 mL DMF was added to the mixture containing 7. The mixture was stirred overnight. Afterwards, the solvent was removed at the rotary evaporator. The crude product was purified by column chromatography on silica gel with EtOAc : Hex (1 : 1) as the eluent to isolate the product (r. f. 0.65, 630  NLS-Pra, NLS-tetra and PNA. The peptides and the PNA sequences were synthesized by solid phase synthesis, as previously reported by our group on a TENTA Gel-S-RAM resin. 28  Mn-maleimide 4A. A as trifluoroacetate salt (30 mg, 0.067 mmol, 1.0 eq.) and maleic anhydride (66 mg, 0.67 mmol, 10.0 eq.) were dissolved in glacial acetic acid (1 mL) and then refluxed for 8 h. The solvent was then evaporated using a high vacuum pump. The crude product was then purified by column chromatography on silica gel with DCM : MeOH (100 : 2.5) as the eluent. The product was isolated as the first band (r.f. 0.67). The solvent was removed using a rotary evaporator to yield 16.6 mg of the expected product as a light brown oil (60%). 1  Fc-maleimide 4B. B (30 mg, 0.140 mmol, 1.0 eq.) and maleic anhydride (137 mg, 1.40 mmol, 10.0 eq.) were dissolved in glacial acetic acid (1 mL) and then refluxed for 6 h. The solvent was then evaporated using a high vacuum pump. The crude product was then purified by column chromatography on silica gel with DCM as the eluent. The product was isolated as the first orange band (r.f. 0.75). The solvent was removed using a rotary evaporator to yield 1.8 mg of product (4%).  2B. B (20 mg, 0.093 mmol, 1 eq.) in 158 μL of MeOH, CuSO 4 (0.297 mg, 0.00186 mmol, 0.02 eq.) in 30 μL of H 2 O and NaHCO 3 (15.7 mg, 0.186 mmol, 2 eq.) in 424 μL of H 2 O were mixed in a 10 mL round bottom flask. TfN 3 (0.279 mmol, 3 eq.), freshly prepared as previously reported, 22 in 258 μL of DCM was added to the mixture. 2.5 mL of MeOH were added to the solution to make it homogeneous. The mixture was stirred for 1 h, until when both TLC (Hex : EtOAc 1 : 1) and UPLC (r.t. 1.60 min, m/z = 240.9) confirmed the full conversion of the amine group to azide. Then the mixture was transferred in a microwave reactor and the following reagents were added to it: Fmoc-L-Pra-OH (31.2 mg, 0.093 mmol, 1 eq.), TBTA 5A. 4A (4.48 mg, 0.011 mmol, 1 eq.) and 9 (7 mg, 0.011 mmol, 1 eq.) were dissolved in 3 mL CH 3 CN in a fluorescence quartz cuvette and irradiated at 302 nm for 1 h. The mixture turned yellow during irradiation. The solvent was removed at rotatory evaporator and the crude was purified by preparative r. 5B. 4B (1.2 mg, 0.004 mmol, 1 eq.) and 9 (2.63 mg, 0.004 mmol, 1 eq.) were dissolved in 0.6 mL CH 3 CN in a fluorescence quartz cuvette and irradiated at 302 nm for 1 h. The mixture turned yellow during irradiation. The solvent was dried at rotatory evaporator and the crude was purified by column chromatography on silica gel with eluent DCM : MeOH (10 : 1). The yellow spot was isolated (r.f. 0.3) to yield the product and a yellow/green solid (2.3 mg, 63%). 1 H-NMR not meaningful due to traces of paramagnetic impurities. 13  Fc-PNA 11. 4B (0.54 mg, 0.00184 mmol, 5 eq.) and PNA as TFA salt (2.5 mg, 0.00037 mmol, 1 eq.) were dissolved in 1 mL of H 2 O in a quartz luminescence cuvette and irradiated for 1 h. The components of the mixture were isolated by preparative r.p. HPLC (linear gradient of A (double distilled water containing 0.1% v/v trifluoroacetic acid) and B (acetonitrile): t = 0-1.

Conclusions
The functionalization of biologically relevant molecules with metal complexes is a very intriguing topic nowadays. The insertion of a metal fragment can strongly improve the properties of the system, from a therapeutic or diagnostic point of view as well as for the investigation of biological mechanisms. The use of copper-catalyzed azide-alkyne cycloaddition is deeply exploited for the functionalization of biomolecules, but some drawbacks limit its application for example in vivo. Therefore, the optimization of modified click procedure was extensively studied in the recent years. In this paper we present the use of modified click strategies (namely one-pot click and photoclick) to insert organometallic fragments in peptides and PNA sequences. One-pot click allows for the in situ conversion of an amine to the azide intermediate, which is then directly clicked with the alkynyl counterpart. Therefore, the isolation of the azide containing compound is not required, relieving the safety concerns linked to the manipulation of azide compounds. We obtained high labelling efficiencies using one-pot click on an amine derivative of cymantrene for derivatization of both a model alkynyl amino acid and a NLS peptide carrying this amino acid. When a ferrocene derivative was used as an amine substrate, yields were found to be lower. The harsh conditions used for one-pot click (1 h of microwave assisted reflux at 80°C) might lead to the decomposition of the methylaminoferrocene precursor. We also investigated the application of metal-free photoclick to our metal-containing precursors, to establish a biocompatible strategy for organometallic derivatization of biomolecules. While the use of cymantrene as substrate for photoclick yielded mainly a by product of the light-mediated CO release from the metal core, the application of photoclick to a ferrocene-maleimide derivative allowed for fast and effective functionalization of both a tetrazole-derivatized PNA monomer and a NLS peptide containing this PNA residue. This technique was also used for Fc insertion into a 15-mer PNA sequence carrying the tetrazole group, with yields comparable to those of a previously reported photoclick application for dsDNA functionalization. 39 Overall, we envisage the application of these very versatile techniques for the functionalization of biomolecules with other metal complexes. As a result, a wide range of metal fragments could be introduced into bioactive systems to prepare novel diagnostic tools useful to study cellular mechanisms. One could also imagine that novel therapeutic entities could be designed, for example, with the introduction of ruthenium complexes into targeting peptides to obtain selective compounds for traditional chemotherapy or photodynamic therapy. Overall, this study will expand the pool of metal-based probe-containing systems, which can be profitably applied in chemical biology or in medicinal chemistry.