An enzyme can be considered to be made up of a number of peptide segments that together define the structural property of that enzyme and account for its high catalytic efficiency and specificity. A peptide, depending on its amino acid sequence, may also possess dynamic secondary and/or tertiary structure that could provide multiple cooperative intra- and inter-molecular interactions, which could be exploited for catalysis, just like enzymes do.^[1] Since the concept of peptide-based catalysis emerged for asymmetric synthesis a half century ago,^[2] enormous advances have been made in terms of reaction scope and mechanism diversity as well as catalyst design.^[3] In this regard, introduction of transition metal complexes to peptide-based ligands is an efficient approach for catalyst development,^[4] because it combines the richness of transition metal chemistry with the above-mentioned merits of peptidic catalysis together with its capability for facile sequence modulation. Along with this direction, palladium,^[5] rhodium,^[6] copper,^[7] and iron^[8] have been integrated with peptides to achieve asymmetric catalytic allylation, hydrogenation, oxidation, carbenoid insertion and cross-coupling reactions. However, we think the potential of this strategy is far from being fully revealed. For example, the competent cobalt,^[9] an essential trace element that is employed by nature in a number of metalloenzymes as the catalytic centre in Vitamin B12 cofactors (Figure 1a),^[10] is almost thoroughly ignored in this regard. We therefore launched a project aiming to fill this gap through preparation of peptide-cobalt hybrids that could be used as catalyst in organic synthesis.

Given their ease of preparation and broad applications, Cobalt-Salen complexes were elected for this purpose (Figure 1, b).^[11] Modification could occur by installation of a short peptide on the core of the Salen ligand, e.g. the phenyl moiety, to give peptide-modified Salen-Co(II) complexes (Figure 1, c). To this end, three salicylaldehydes SA1~SA3 were designed and made each bearing a free carboxyl acid group for peptide ligation (Scheme 1). 5- Carboxysalicylaldehyde SA1 was obtained readily through mono-formylation of 4-hydroxybenzoic acid 1 with hexamethylenetetramine (HMTA). Selective alkylation of commercially available 5-hydroxysalicylaldehyde (2) with tert-butyl bromoacetate (3) afforded compound 4 in 35% yield, which was converted into salicylaldehyde SA2 quantitatively via treatment with trifluoroacetic acid (TFA). 6-Hydroxysalicylic acid (5) was transformed into 6-hydro- xysalicylaldehyde (6) through a two-step process, and the same protocol was effective to afford SA3 readily.

Fmoc solid-phase peptide synthesis (SPPS) was used to make oligopeptides (Scheme 2). Wang resin was loaded with a first Fmoc amino acid through condensation using the N,N'-diisopropylcarbodiimide (DIC)-1-hydroxyben- zotriazole (HOBt) combination in N,N-dimethylformamide (DMF). Subsequent end-capping acetylation was achieved with excess Ac₂O/Py to give resin supported Fmoc amino acid 8 ready for elongation. Then repetitive Fmoc cleavage, coupling and washing were carried out with selected amino acids. Then at the end, cleavage from the resin with simultaneous global deprotection was accomplished by treatment with TFA. Following this protocol, short peptides P1~P4 were prepared in good yields.

With carboxyl acid functionalized salicylaldehydes SAz and peptides Px at hand, we headed on to synthesize peptide-modified salen-Co(II) complexes. Activation of SAz with O-(7-azabenzotriazol-1-yl)-N,N,N',N'-tetramethyluro- nium hexafluorophosphate (HATU)/diisopropylethylami- ne (DIPEA) in DMF followed by addition of oligopeptide Px afforded peptide modified salicylaldehyde Px^zSA, which were purified via reverse phase column chromatography. Condensation of these Px^zSA with ethylenediamine (EDA, 0.5 equiv.) in DMF under nitrogen atmosphere at room temperature gave corresponding Salen ligands in 6 h. Upon completion of the reaction, as indicated by thin layer chromatography (TLC), a suspension of Co(OAc)₂ in DMF was introduced. The reaction mixture was then heated to 80 °C for 12 h, affording a deep red solution. Removal of the solvent and washing with deionized water provided a brown solid. ESI-MS confirmed the successful preparation of the desired peptidic metal complexes Px^zSalenCo. The UV-Vis spectra of these complexes have been recorded and analyzed to further confirm the metal centre to be Co(II) rather than Co(III).^[12] The yields for peptide installation and the metal complex assembly are tabulated in Scheme 3. Meanwhile, the same protocol was also applied to prepare Co(II) complexes ¹SalenCo, ²SalenCo and ³SalenCo directly from corresponding salicylaldehydes SA1~SA3.

Recently, great advances have been made in hydrofunctionalization of alkenes and alkynes via first row transition metal catalysis.^[13] In this context, the catalytic activity of our peptidic Salen cobalt complexes was evaluated with the model reaction of cobalt-catalyzed hydrohydrazination of cinnamic alcohol 9 with azodicarboxylate 10.^[14] We envisioned that the hydroxyl group in olefin 9 could engage in hydrogen-bonding interactions with the peptide on the cobalt Salen complex. The reaction was carried out in MeOH under N₂ at 30 ℃ with excess tetramethyldisiloxane (TMDS) as the hydride source. The results are shown in Table 1. Although 1 mol% P1¹SalenCo was ineffective, the same amount of complexes P1²SalenCo and P1³SalenCo exhibited excellent catalytic activity, resulting in clean conversion (in 3 h) of allylic alcohol 9 to 1,3-hydroxyl- alkylhydrazine 11 in 81% and 90% isolated yields, respectively (Entry 1 vs Entries 2, 3). With P2 modified Salen- Co(II) complexes, a similar trend was still observed, even lower yields were obtained (Entry 4 vs Entries 5, 6). P3²SalenCo, carrying highly hydrophobic pentapeptide P3, was effective to promote the reaction as well to give 11 almost quantitatively (Entry 7). Both heptapeptide P4 modified cobalt complexes P4²SalenCo and P4³SalenCo could catalyze this hydrohydrazination at 30 ℃, affording 11 in excellent yields (Entries 8, 9), even though the former needed more time to achieve full conversion. At 15 ℃, the catalytic efficiency for all the peptidic cobalt complexes derived from framework SA2 decreased remarkably (Entries 10~13). When quenched at 3 h, an 80% conversion was realized with tetrapeptide P1 modified P1²SalenCo as the catalyst. Similarly, with complexes P2²SalenCo, P3²SalenCo and P4²SalenCo, 60%, 40% and 50% conversion were obtained respectively in the same reaction time, which roughly correlated proportionally with the reaction rates. The rate differences related to the decorating peptides are recognizable and should not be ignored. In order to make further comparison, non-peptide Salen-Co complexes ¹SalenCo~³SalenCo were also tested with the same reaction. Surprisingly, ¹SalenCo displayed much higher catalytic activity than P1¹SalenCo and P2¹SalenCo, and the reaction could proceed to completion in 3 h even at 15 ℃ (Entry 14 vs Entries 1, 4). In contrast, both ²SalenCo and ³SalenCo were able to achieve only partial conversion with the same reaction time and temperature, showing slightly decreased catalytic efficiency than their peptidic congeners (Entries 15, 16 vs Entries 10~13). These preliminary data clearly manifest that not only the electronic nature of the Salen core is strongly correlated to the catalytic activity of the metal centre, but also the identity of the peptide can impose a significant influence, a feature reminiscent of metal cofactors in metalloenzymes. Although the exact reason remains elusive at this stage, the temperature effects on these catalytic reactions may reflect the common principle that peptides have lower conformational flexibility and hydrogen bonding strength at lower temperatures.

In addition, the ee value of the hydrazine product 11 for some reactions was measured by HPLC, but almost no enantiomeric excess could be detected. Despite being disappointed, it is not totally out of our expectation when taking into account the radical nature of the stereogenic C—N bond forming event and the fact that no asymmetric version of related cobalt catalyzed alkene hydrofunctionization has ever been documented till now.

In order to evaluate the potential effect of the free hydroxyl group as an excellent hydrogen bonding donor on the catalytic reaction, parallel experiments with trans-cinnamyl methyl ether 12 in place of 9 were carried out employing our peptide-modified Salen-Co(II) complexes as catalysts (Table 2). At 30 ℃, the hydrohydrazination of cinnamyl methyl ether 12 with 10 all proceeded to completion in 3 h except those catalyzed by P1¹SalenCo and P2¹SalenCo (Entries 1, 4). With the same catalyst, the yield of 13 from the reaction of 12 was lower than the yield of 11 from the correlated reaction of 9. This divergence in yields did demonstrate an observable beneficial impact of the hydroxyl group on the metal catalyzed hydrohydrazination, which might derive from the formation of hydrogen-bonding between the substrate and the peptide on the catalyst.

The catalytic reactivity of these peptide modified Salen- Cobalt complexes was further investigated with N-cinna- myl acetamide 14 as the substrate for this hydrohydrazination (Table 3). Under the same conditions, all complexes except those derived from scaffold SA1 demonstrated excellent activity to achieve clean reactions with full conversion of 14 and the hydrohydrazine product 15 was obtained in exceptional isolated yields. This restoration of clean reaction may be attributed to the presence of a second amide group which can readily participate in hydrogen-bonding interactions acting as either a donor or an acceptor or both.

Next, peptide modified cobalt complex P1²SalenCo was selected to explore the substrate scope. As shown in Table 4, secondary N-cinnamyl aromatic amides 16~19 all underwent this cobalt catalyzed hydrohydrazination reaction smoothly with 1 mol% catalyst loading, giving 24~27 in high yields (Entries 1~4). Surprisingly, under the same conditions, the reaction of (E)-propenylbenzene 20 was much slower than previous amides with a significant drop in yield (Entry 5). This phenomenon clearly demonstrates the critical role that the hydrogen bonding has played in current peptide-based SalenCo catalyzed reaction. Methyl cinna- mate 21 underwent this catalytic process readily to deliver 29 in 87% yield (Entry 6). Substrates 22 and 23, both carrying a coordinating N-heterocycle, namely pyridine and piperidine respectively, exhibited no reactivity under the standard conditions (Entries 7, 8). We reason that the catalytic activity of P1²SalenCo is inhibited completely by the substrates.

In summary, nine peptide modified Salen-Co(II) complexes based on three Salen ligand frameworks SA1~SA3 and four oligopeptides P1~P4 have been made successfully. These peptide-based metal complexes were submitted to catalyze the hydrohydrazination of cinnamic alcohol 9, cinnamyl methyl ether 12 and N-cinnamylacetamide 14 with azodicarboxylate 10 using TMDS as a hydrogen atom donor. Whereas Salen-Co(II) complexes derived from SA1 barely showed catalytic activity, those based on SA2 and SA3 were capable of accomplishing this transformation with only 1 mol% catalyst loading at 30 ℃ in 3 h. The catalytic activity is pretty sensitive to the reaction temperature as evidenced by the partial conversions observed in the reaction at 15 ℃. At lower temperature, the impact of peptide on the catalyst activity is also prominent. Compared to alcohols 9 and 14, ether 12 provided the hydrohydrazination product in lower yields. These findings indicate peptide modification as an effective means for Salen-Co(II) catalyst optimization and hydrogen-bonding may play a significant role. Although no chiral induction has been achieved with these catalysts, we still hold the faith in light of many enantioselective radical reactions enabled by metalloenzymes and continue to work on this project in our laboratory.