Aryl diazonium intermediates enable mild DNA-compatible C–C bond formation for medicinally relevant combinatorial library synthesis

Forging carbon–carbon (C–C) linkage in DNA-encoded combinatorial library synthesis represents a fundamental task for drug discovery, especially with broad substrate scope and exquisite functional group tolerance. Here we reported the palladium-catalyzed Suzuki–Miyaura, Heck and Hiyama type cross-coupling via DNA-conjugated aryl diazonium intermediates for DNA-encoded chemical library (DEL) synthesis. Starting from commodity arylamines, this synthetic route facilely delivers vast chemical diversity at a mild temperature and pH, thus circumventing damage to fragile functional groups. Given its orthogonality with traditional aryl halide-based cross-coupling, the aryl diazonium-centered strategy expands the compatible synthesis of complex C–C bond-connected scaffolds. In addition, DNA-tethered pharmaceutical compounds (e.g., HDAC inhibitor) are constructed without decomposition of susceptible bioactive warheads (e.g., hydroxamic acid), emphasizing the superiority of the aryl diazonium-based approach. Together with the convenient transformation into an aryl azide photo-crosslinker, aryl diazonium's DNA-compatible diversification synergistically demonstrated its competence to create medicinally relevant combinatorial libraries and investigate protein–ligand interactions in pharmaceutical research.


Introduction
Carbon-carbon (C-C) bond formation is a fundamental process in all facets of organic and bioorganic chemistry, in which palladium (Pd)-catalyzed cross-coupling reactions are integral. [1][2][3][4] These palladium-catalyzed cross-coupling reactions mainly focus on building C(sp 2 )-C(sp 2 ) connections that feature structural rigidity and increase drug-likeness. The importance of C(sp 2 )-C(sp 2 ) bonds, together with the broad substrate of Pdcatalyzed cross-coupling, render Suzuki-Miyaura, Heck and Hiyama reactions the most widely adopted synthetic approaches in medicinal chemistry. 5,6 Meanwhile, metalcatalyzed C-C bond formation has emerged as an important topic in chemical biology, as evidenced by the adaptation of Pdcatalyzed cross-coupling for chemoselective modication of biomacromolecules. For example, iodo/bromo-functionalized unnatural bases allowed uorogenic labeling of DNA/RNA molecules; 7-10 proteins bearing a site-specically incorporated iodophenyl handle could undergo bioorthogonal labeling with boronic acid-conjugated probes. 11,12 Consequently, expanding the scope of Pd-catalyzed C-C coupling reactions, especially in a biocompatible manner, is of great signicance to medicinal chemistry and chemical biology.
From the chemistry-biology interface, a drug interrogation technique named DNA-encoded chemical library (DEL) has emerged. Since its sprouting concept was proposed by Lerner and Brenner in 1992, three decades of efforts have proved the DEL to be an economical and efficient platform to expedite bioactive hit discovery. [13][14][15][16][17] Compared with the resourceintensive high-throughput screening (HTS) that contains separately synthesized and deposited library members, a DEL allows the miniature-scale synthesis, maintenance, and selection of all library members simultaneously in a single pool. Typically, each diverse building block (BB) is conjugated to the starting DNA fragment, accompanied by the enzymatic ligation of a corresponding DNA barcode. In this fashion, the molecular identity of each chemical subunit incorporated is recorded by using the tethered DNA sequence. By integrating the power of combinatorial synthesis with genetic barcoding in iterative "split-andpool" cycles, libraries comprising an exceptional magnitude (10 6 -10 12 ) of members could be achieved. When the library is a Chongqing Key Laboratory of Natural Product Synthesis and Drug Research, School of Pharmaceutical Sciences, Chongqing University, China. E-mail: yizhouli@cqu.edu.cn; gongzhang@cqu.edu.cn selected against the target protein, the molecular identity of positive binders is then interpreted via PCR and sequencing. Attracted by the DEL's cost and time efficiency, both academia and pharmaceutical companies have advanced DEL in terms of synthetic methodology and selection strategy. [18][19][20][21][22][23][24][25][26][27][28][29][30][31] Currently, although very few drug candidates in late-stage clinical trials stemmed from DEL, such as the soluble epoxide hydrolase (sEH) inhibitor GSK2256294 and the receptor-interacting protein (RIP1) kinase inhibitor GSK2982772, [32][33][34][35] DELs have shown their potential in high-throughput discovery of preliminary bioactive hits for dened targets, which might lead to the development of bioactive compounds or probes with structural optimization.
Successful hit discovery lies in the DEL's vast chemical space and abundant structural diversity, which in turn depends on the DEL synthetic reactions employed. First, DEL-compatible reactions should be performed in aqueous solutions and circumvent inherent restrictions posed by DNA's physicochemical properties. Second, DEL synthesis should either create meaningful drug-like structures such as (hetero)aromatic cycles and natural products, [36][37][38][39] or link pharmacophore-containing building blocks with chemically stable and structurally rigid scaffolds, rather than merely build loose and exible connections between DNA-tagged chemical subunits. Notably, C(sp 2 )-C(sp 2 ) bond formation is a major approach to assembling and decorating aromatic BBs with structurally conned and less rotatable linkage. Therefore, forging C(sp 2 )-C(sp 2 ) linkage in DEL synthesis is a meaningful task, especially under mild conditions, and having broad substrate scope, and remarkable functional group tolerance.
Compared with aryl halides, aryl diazonium salts serve as versatile reagents for transition metal-catalyzed crosscoupling. [62][63][64] The higher reactivity of aryl diazonium permits the Suzuki-Miyaura or Heck cross-coupling reaction to occur at a lower temperature, under milder pH conditions, free of additional ligands, and in an environment-friendly fashion. 65,66 In addition, aryl diazonium salts can be readily prepared from abundant and inexpensive arylamines preferred in combinatorial chemistry. 17 As a consequence, we envision that the on-DNA aryl diazonium-based cross-coupling offers a complementary or even superior approach to forging C-C linkage (Fig. 1a, paths C and D), which could facilitate DELs to expedite bioactive compound discovery. Typical examples containing C-C bonds include urbiprofen, belinostat, and lacidipine, which contain hydrolyzable esters or hydroxamic acids (Fig. 1b).

Results and discussion
We set out to establish a standard condition for DNAcompatible diazotization and cross-coupling starting from the aniline-conjugated DNA headpiece (HP) 1a. Although aryl diazonium could be readily generated in situ by treating arylamine with nitrous acid, 67 nitrous acid (HNO 2 ) and nitrite (NO 2 − ) at high concentrations are notoriously known as carcinogens. While they are transformed into N-nitrosamine derivatives in the liver as a strong DNA-alkylating species, they can also directly cause hydrolytic deamination of nucleobases through diazonium intermediates in vitro, leading to mutations of adenine, guanine and cytosine. 68,69 Therefore, the diazotization reagent's preferred reactivity to the DNA-tethered aniline group rather than to native nucleobases is obligatory. Due to the instability of aryl diazonium salts in LC-MS detection, diazotization was monitored aer further nucleophilic substitution to afford phenyl azide 1a ′ . We compared two diazotization reagents, nitrous acid (NaNO 2 /HCl) and tert-butyl nitrite (tBuONO). While prolonged NaNO 2 /HCl treatment caused partial nucleobase deamination, good to excellent conversion was observed without detectable DNA damage by tBuONO treatment (nal conc. 5 mM, 25°C), indicating that tBuONO was a DNA-compatible diazotization reagent (ESI Table 1 and Fig. 1-3 †). Following the aryl azide transformation of several DNA-conjugated arylamines, subsequent copper-catalyzed azide-alkyne cycloaddition (CuAAC) reactions were performed with phenylacetylene to form 1,2,3-triazoles (ESI Fig. 4 †). This diversication demonstrated that arylamine building blocks could serve as masked precursors to generate active aryl diazonium intermediates, and further showed potential to synthesize DELs with click chemistry from diverse aryl azides facilely generated from arylamines.
Following the consideration of diazotization conditions, we next investigated the subsequent Suzuki-Miyaura coupling from aryl diazonium intermediates. When palladium(II) acetate and phenylboronic acid 2a were sequentially added to the aryl diazonium intermediate, the biphenyl coupling product 3a was observed with 60% conversion ( Table 1, entry 1). To reduce deamination/hydrolysis of diazonium intermediates and increase the efficiency of the coupling reaction, the stability of aryl diazoniums and the reactivity of boronic acids required elegant balancing. In detail, in the rst step, a weak acidic environment enhanced the stability of the diazonium intermediate, consistent with the fact that diazonium was prone to hydrolysis under basic conditions. In the second step, we examined different bases to activate phenylboronic acid, and found that cesium carbonate (Cs 2 CO 3 ) yielded better conversion ( reconciled reactivity with stability by ne-tuning the pH and temperature, providing 73% conversion (Table 1, entry 10). To verify the product's absolute structure, we conjugated the off-DNA synthesized and characterized small molecule 4-phenylbenzoic acid S3a to HP-DNA as the standard. Identical retention times were observed in co-injection assays between the on-DNA generated product and off-DNA prepared standard (ESI Fig. 5 †). Altogether, our optimized approach allowed the on-DNA architecture of the C-C bond via aryl diazonium-based Suzuki-Miyaura coupling (Fig. 1b).
In this vein, we inspected the substrate scope of Suzuki-Miyaura coupling by testing a wide variety of boronic acids or boron-containing chemical subunits according to the established protocol (Fig. 2a). To our delight, this reaction exhibited good conversion (60-91%) towards a broad range of electronrich and electron-decient substituted phenylboronic acids (2b-2k), including multi-substituted substrates (2l and 2m). Notably, labile functional groups such as the ester (2k) were well-tolerated, together with reactive handles permitting further diversication such as carboxylic acid (2f). In addition,  heterocyclic and polycyclic boronic acids yielded cross-coupling products with good to excellent conversion (60-85%, 2n-2t), offering an opportunity to assemble pharmaceutically privileged aromatic scaffolds (e.g., benzofuran and thiophene) by usin C(sp 2 )-C(sp 2 ) linkage. Furthermore, the reactivity of boronic ester (2u) and potassium triuoroborate salt (2v) was demonstrated. This signicantly expanded the chemical space of the aryl diazonium-based Suzuki-Miyaura C-C coupling reaction by accommodating available suites of boroncontaining compounds in combinatorial synthesis. Altogether, 20 boronic-containing subunits provided good to excellent conversions (>70%), while 13 resulted in acceptable conversions (50-70%) among the 33 BBs tested (ESI Table 3 †).
Besides, we explored the substrate scope regarding the DNAconjugated arylamines (Fig. 2b). The ester bond-containing 2k was chosen as the standard substrate. As expected, a variety of substituted anilines, whether with electron-rich or electronwithdrawing substituents provided good conversion rates without ester hydrolysis (62-80%, 1a-1g).
In addition, other DNA-tethered arylamines containing naphthalene, biphenyl, or sulfonamide groups yielded satisfactory conversions (66-70%, 1h-1j). Supported by the abundant commercial availability of bifunctional amine BBs 17 and 22 representative substrates tested here (ESI Table 4 †), we could anticipate that a wide range of DNA-conjugated arylamines would increase the diversity of synthesized libraries.
Encouraged by the accomplishment of the aryl diazoniumbased Suzuki-Miyaura coupling reaction which proceeded under mild conditions, we accordingly managed to adapt analogous conditions for the Heck coupling reaction between aryl diazonium intermediates and olens. To our delight, in the presence of palladium(II) acetate and phosphate buffer (pH 5.5), the Heck coupling reaction with styrene 4a yielded stilbene product 5a with good efficiency (83%) at 25°C (Fig. 3a). Similarly, the on-DNA Heck coupling product's structure was validated by off-DNA synthesis and co-injection experiments (ESI Fig. 6 †). Therefore, we demonstrated the capability to forge C-C bonds starting from arylamine BBs and olens in DEL synthesis under exceptionally mild conditions (Fig. 1b).
Subsequently, we performed the DNA-conjugated arylamine scope study. We tested the conversion of a series of arylamines reacting with ester bond-containing styrene 4o and methyl acrylate 4r, respectively (Fig. 3b). Similarly with the arylamines' scope for Suzuki-Miyaura coupling, among the 34 DNA-tagged substrates tested, 29 substrates achieved good conversions (>70%) with styrene (ESI Table 6 †), while 31 afforded good conversions with methyl acrylate (ESI Table 7 †). As a complement to the Heck reaction above, we took advantage of the abundant arylamine BBs to reversely investigate their reactivity proles with the DNA-conjugated styrene 6a (Fig. 3c). According to our previous work, 27 a panel of arylamines (7a-7m) was treated with NaNO 2 /HCl to readily form aryl diazonium salts in situ. The mixture reacted smoothly with 6a to afford the expected products with high conversion without detectable damage (ESI Table 8 †), no matter with electron-rich (7b), electron-decient (7c), or reactive (7d, 7f) substituents. Altogether, Heck coupling between aryl diazoniums and olens was achieved regardless of which end was in conjugation with DNA. This offered various methods to generate cinnamic acid-derived and stilbene-like chemical scaffolds in DEL synthesis.
Besides the widely explored Suzuki-Miyaura and Heck reactions, we were interested in exploring more C-C bond coupling reaction types enabled by the aryl diazonium intermediates. Considering the possible Pd-catalyzed Hiyama reaction between aryl diazonium salts and organosilanes, 70,71 we tested and optimized the Hiyama reaction condition on DNAconjugated aryl diazonium intermediates. A panel of DNAconjugated arylamines reacted with phenyltrimethoxysilane to afford C(sp 2 )-C(sp 2 )-conjugated products in moderate conversion (51-67%, Fig. 3d, ESI Table 9 and Fig. 7 †). To our knowledge, this represented the rst-case exploration of Hiyama reaction on DNA, offering an alternative and supplementary approach to architect C(sp 2 )-C(sp 2 ) linkages in DELs using organosilicon compounds. We also tested the on-DNA reactions on a large scale exemplied by Suzuki-Miyaura and Heck reactions, demonstrating their applicability in industrialized DEL construction (ESI Fig. 8 and 9 †).  To demonstrate the wide applicability of the new methodology, diversication to multi-dimensional C-C-assembled DELs was performed. Considering the different conditions of aryl diazonium-based and aryl halide-based coupling approaches, we predicted them to be orthogonal. To illustrate this viewpoint, we started with a 3-bromo-5-aminobenzoic acidconjugated DNA headpiece 1f, a bifunctional starting DNA bearing both aryl bromide and arylamine handles. As expected, 1f underwent the rst round of the Heck coupling reaction on the arylamine handle (via aryl diazonium intermediate) to afford 8a with high efficiency, leaving aryl bromide intact. Next, the aryl bromide further underwent the second step Suzuki-Miyaura coupling reaction, generating branched C-Cconnected scaffold 8b (Fig. 4a). Likewise, a linear topological scaffold was constructed. Bromo-substituted styrene was tolerated to react with the DNA-linked aryl diazonium, followed by a second-round Suzuki-Miyaura coupling via aryl bromide (ESI Fig. 10 †). These data together proved the capability to construct various C-C-assembled scaffolds in a DEL via orthogonal Pdcatalyzed cross-coupling.
Furthermore, the compatibility of this suite of Pd-catalyzed reactions with DEL synthesis was carefully examined. First, enzymatic DNA ligation was assessed to rule out any possible impairment by metal ions or nitrous species. Starting from the uniformed DNA-conjugated aniline HP-P-1a, different reaction conditions reported above (including Suzuki-Miyaura and Heck cross-coupling reactions) were applied, and then the ligation assay was performed on the reaction product subsequently (ESI Fig. 11 †). To our delight, all sets of enzymatic DNA barcode ligation proceeded smoothly, as shown by the polyacrylamide gel electrophoresis (PAGE) analysis (ESI Fig. 12 †). In addition, we investigated the compatibility with the decoding process by preparing a longer aniline-conjugated DNA named PCR-1 resembling the length of a fully encoded DEL (ESI Fig. 13 and 14 †). PCR-1 was subjected to various reaction conditions (including azide transformation, Suzuki-Miyaura coupling, and the Heck coupling reaction), and the resulted products, as PCR templates, underwent PCR amplication and Sanger sequencing (ESI Fig. 15 †). If obvious DNA damage occurred, either the PCR process would be hampered, or the sequencing results would lose delity. Nevertheless, no detectable mutation or increased sequencing noise was observed (ESI Fig. 16 †), indicating that DNA was tolerable in these combinatorial synthetic procedures. With these data, we inferred that the construction of multi-dimensional DNA-encoded C-Cassembled combinatorial libraries via diazonium intermediates was conceivable.
With the potential to construct multi-dimensional DELs, we further sought to strengthen the utility by demonstrating the efficient synthesis of medicinally relevant and biologically active compounds on DNA. Since the condition of the aryl diazonium-based Heck coupling reaction was extremely mild, we envisioned that various bioactive warheads would remain intact via the aryl diazonium-based synthesis.
Given the evidence that susceptible functional groups such as esters and oxazolidones were elegantly preserved in our synthetic strategy (see Fig. 2 and 3), we launched a comparative study to incorporate a variety of functional groups into DNA conjugates (Fig. 4b, ESI Fig. 17-20 †). Notably, the HDAC inhibitor bearing an easily hydrolysable hydroxamic acid warhead was on-DNA synthesized via our aryl diazonium-based approach in parallel with the traditional aryl halide-based approach. Starting from the arylamine-or aryl iodideconjugated DNA, a HDAC ligand unit containing a styrene moiety was cross-coupled in parallel (see the ESI † for detailed methods). As shown by the LC-MS data, the high temperature and strong base required for the aryl iodide-based reaction caused almost complete hydrolysis of the hydroxamic acid warhead. In contrast, the ambient temperature and mild pH conditions in the aryl diazonium group yielded the desired product, leaving the hydroxamic acid warhead intact (Fig. 4b).
These data emphasized our approach's excellent functional group tolerance.
In addition, aryl diazonium could be readily transformed into aryl azide (as shown in Table 1), a canonical photocrosslinking probe. To prove its utility, we chose the model system of sulfonamide binding to the carbonic anhydrase CA II to perform the DNA-programmed photoaffinity labeling (DPAL) assay 72,73 that permits facile identication of target-binding ligands (Fig. 4c). Starting from the versatile DNA-conjugated aniline, divergent functionalization was implemented. A binding probe (BP) bearing an aniline handle was conjugated with 4-vinyl benzenesulfonamide to form BP-VBS, while a capture probe (CP) bearing an aniline moiety was transformed into the phenyl azide-containing CP-N 3 with photo-crosslinking properties. Upon the formation of a ternary complex among BP-CBS, CP-N 3 , and CA II, proximity-enabled covalent crosslinking occurred upon UV irradiation, suggesting the robust binding of the sulfonamide ligand and efficient crosslinking of the phenyl azide warhead (Fig. 4c). Altogether, these experiments indicated that the on-DNA aryl diazonium-centered synthesis was a versatile platform to afford various pharmaceutically active compounds and valuable chemical probes.

Conclusion
In conclusion, we have implemented the DNA-compatible Suzuki-Miyaura, Heck and Hiyama coupling reactions that employ aryl diazonium salts as the intermediates, enabling the facile architecture of C-C linkage in medicinally relevant DEL synthesis under mild condition, and having extensive substrate scope and exceptional functional group tolerance. Starting from DNA-tagged arylamines, we examined the diazotization condition to realize chemo-selective modication of the arylamine in preference to native nucleotides. We ne-tuned the buffer to realize efficient Suzuki-Miyaura, Heck and Hiyama crosscoupling across a diverse suite of chemical subunits, and then investigated the substrate scope of arylamines, boronic acids/ esters, and olens. Moreover, the diazonium-based approach adopted mild conditions and proved orthogonal with traditional aryl halide groups, providing an avenue to assemble branched or linear multi-dimensional drug-like libraries with structurally conned C-C connections. Remarkably, the aryl diazonium-based Heck coupling reaction was carried out under ambient room temperature and mild pH conditions, allowing the on-DNA synthesis of pharmaceutical compounds bearing susceptible bioactive warheads that would otherwise decompose (e.g., HDAC inhibitor with the hydroxamic acid group). In addition, aryl diazonium could be readily transformed into aryl azide as a covalent biomacromolecule photo-crosslinker. These altogether demonstrated the versatility of aryl diazonium to create medicinally relevant combinatorial libraries and the potential of resulting DNA probes to investigate ligand-protein interaction under physiological conditions. Finally, the DNAcompatible aryl diazonium-based synthetic strategies developed here may allow a broader panel of diversity-oriented transformations, providing potential solutions to medicinal chemistry and chemical biology in the future.