Hydrophilic CO-Releasing Material of PEGlyated Ruthenium Carbonyl Complex

The poor water-solubility and instability of Ru(II) carbonyl complex hamper the therapeutic application as CO releasing materials (CO-RMs). To enhance the hydrophilicity and bio-utility of CO, a robust Ru(I) carbonyl sawhorse skeleton was grafted with water-soluble PEGylated sidearm. In this case, 12 PEGylated sawhorse Ru2(CO)4 complexes were prepared with satisfactory yields and characterized by IR and 1H- and 13C- NMR. X-ray diffraction analysis of CO-RM 8, 13 and 14 revealed the featured diruthenium sawhorse skeleton and PEGylated axial ligands. The flask-shaking method measures the water-solubility of CO-RMs, indicating that both bridging carboxylate ligands and PEGlyated axial ligands regulate the hydrophilicity of these CO-RMs. Under photolysis conditions, CO-RM 4–13 sustainable released therapeutic amounts of CO in the myoglobin assay. The correlation of the CO release kinetics and hydrophilicity of CO-RMs demonstrated that the more hydrophilic CO-RM released CO faster. The biological test found that the low cytotoxic CO-RM 4 showed a specific anticancer activity toward HT-29 tumour cells.


Introduction
Recently, biological experiments using transition metal carbonyl complexes as COreleasing molecules (CO-RM) [1,2] revealed the therapeutic effects of endogenous CO. These effects include anti-inflammatory function, vasodilatation, anti-apoptotic, anti-proliferative, and anti-hypoxia [3]. The organ protection of CO is desirable and attractive because controlling a low concentration of CO indeed protects donor tissues from ischaemiareperfusion injury [4]. The ruthenium carbonyl complex, [Ru(CO) 3 Cl 2 ] 2 , CO-RM-02, mainly were fascinating because it attenuated acute hepatic ischemia-reperfusion injury in rats by reducing serum AST/ALT levels and improved the liver histology score [5,6]. However, the poor solubility of CO-RM-02 and its unregulated CO releasing property in aqueous systems hinder its therapeutic application under physiological conditions [7,8]. To solve the watersolubility issue, CO-RM-02 was solubilized in DMSO, which readily reacts with the CO-RM dimer to generate DMSO-ligated monomeric ruthenium carbonyl species [9]. Motterlini and Mann found that glycinate ligands chelate to the Ru (II) carbonyl moiety, and the corresponding Ru(II) complex (CO-RM-03) was water-soluble. Unfortunately, CO-RM-03 degrades rapidly in human plasma, and the half-life of CO release is only 3.6 min [10]. In fact, due to the intrinsic hydrophobic nature of CO, most metal carbonyl complexes have minimal solubility in an aqueous solution. Although a few ionic transition metal carbonyl complexes are water-soluble, the CO release tests indicated that the M-CO bonds of simple ionic CO-RM degraded quickly under complicated physiological conditions and produced unpredictable side effects, such as blocking blood vessels and causing cytotoxic effects. Increasing the water-solubility and finely controlling the CO release kinetics of transition metal complexes are challenges in therapeutic CO-RMs [11][12][13][14][15].
Introducing the hydrophilic functional groups into the coordination sphere of transition metal complexes is an efficient way to enhance the water-solubility of the leading CO-RM structure (Scheme 1). The hydrophilic auxiliary ligands bearing carboxylic acid groups improve the hydrophilicity of (Mo(CO) 3 (CNCR R CO 2 R ) 3 [16] and [Mn(CO) 4 {S 2 CNMe(CH 2 CO 2 H)}] [17], respectively. The classic water-soluble phosphine ligand, P(CH 2 OH) 3 coordinates to Ru(I) center and significantly enhances the solubility of Photo-CO-RM, 2, 2 -bipyridine tricarbonyl rhenium(I) in PBS solution [18]. Conjugation of transition metal carbonyl moiety with the hydrophilic biomacromolecules also improves the water solubility and biocompatibility of CO-RMs. The micellar bearing [RuCl(glycinate)(CO) 3 ] [19], peptide conjugates of [Mn(CO) 3 (tmp)] + [20], and a photo-CO-RM based on vitaminB12, namely, B12-ReCO-RM2 [21] were fabricated to deliver CO in the aqueous system. However, the synthesis and purification of these hydrophilic CO-RMs are sophisticated. More importantly, the CO release kinetics of most of the water-soluble CO-RMs are unpredictable and thus cannot satisfy the basic requirements of ADME proprieties for CO pre-drugs [22]. Polyethylene glycol (PEG) is a non-ionic polymer that improves water solubility and selective drug absorption. Inspired by the idea of the PEG-polymer linking drug system proposed by H. Ringsdorf [23], and the subsequent precedents of PEGlyated therapeutic agents by A. Abuchowski [24]. Herein, CO-RMs, which are PEGlylated Sawhorse Ru carbonyl complexes, were synthesized by incorporating a robust CO-RM leading structure [Ru 2 (CO) 4 (COOR) 2 ] with functionalized PEG chains. X-ray single-crystal analysis revealed that the designed PEG esters of amino acids coordinate to Ru(I) of the sawhorse CO-RM lead structure. From the logP measurements and myoglobin assay experiments, it was found that these hydrophilic CO-RMs show a well-controlled CO release property with broad kinetics under physiologic conditions. Materials 2022, 15, x FOR PEER REVIEW 2 of 8 CO release kinetics of transition metal complexes are challenges in therapeutic CO-RMs [11][12][13][14][15]. Introducing the hydrophilic functional groups into the coordination sphere of transition metal complexes is an efficient way to enhance the water-solubility of the leading CO-RM structure (Scheme 1). The hydrophilic auxiliary ligands bearing carboxylic acid groups improve the hydrophilicity of (Mo(CO)3(CNCR′R″CO2R‴)3 [16] and [Mn(CO)4{S2CNMe(CH2CO2H)}] [17], respectively. The classic water-soluble phosphine ligand, P(CH2OH)3 coordinates to Ru(I) center and significantly enhances the solubility of Photo-CO-RM, 2, 2′-bipyridine tricarbonyl rhenium(І) in PBS solution [18]. Conjugation of transition metal carbonyl moiety with the hydrophilic biomacromolecules also improves the water solubility and biocompatibility of CO-RMs. The micellar bearing [RuCl(glycinate)(CO)3] [19], peptide conjugates of [Mn(CO)3(tmp)] + [20], and a photo-CO-RM based on vitaminB12, namely, B12-ReCO-RM2 [21] were fabricated to deliver CO in the aqueous system. However, the synthesis and purification of these hydrophilic CO-RMs are sophisticated. More importantly, the CO release kinetics of most of the watersoluble CO-RMs are unpredictable and thus cannot satisfy the basic requirements of ADME proprieties for CO pre-drugs [22]. Polyethylene glycol (PEG) is a non-ionic polymer that improves water solubility and selective drug absorption. Inspired by the idea of the PEG-polymer linking drug system proposed by H. Ringsdorf [23], and the subsequent precedents of PEGlyated therapeutic agents by A. Abuchowski [24]. Herein, CO-RMs, which are PEGlylated Sawhorse Ru carbonyl complexes, were synthesized by incorporating a robust CO-RM leading structure [Ru2(CO)4(COOR)2] with functionalized PEG chains. X-ray single-crystal analysis revealed that the designed PEG esters of amino acids coordinate to Ru(I) of the sawhorse CO-RM lead structure. From the logP measurements and myoglobin assay experiments, it was found that these hydrophilic CO-RMs show a well-controlled CO release property with broad kinetics under physiologic conditions.

Results and Discussion
Since 1969 when J. Lewis first reported that Ru2(CO)4(O(C=O)R)2L2, features a sawhorse structure [25], a considerable number of the ruthenium(I) carbonyl complexes have been prepared and characterized. However, the inherited poor water-solubility and bio-Scheme 1. Selected Water-soluble CO-RMs.

Results and Discussion
Since 1969 when J. Lewis first reported that Ru 2 (CO) 4 (O(C=O)R) 2 L 2 , features a sawhorse structure [25], a considerable number of the ruthenium(I) carbonyl complexes have been prepared and characterized. However, the inherited poor water-solubility and biological incompatibility of sawhorse ruthenium complexes are obstacles to their application in in the biological system [26]. To increase the hydrophilicity of these ruthenium carbonyl complexes, the water-soluble amino acid glycol and PEG esters were tethered at the axial position of the sawhorse structure, respectively (Scheme 2). The thermolysis of Ru 3 (CO) 12 1 in acetic acid affords polymeric sawhorse ruthenium carbonyl 2, and the sequential ligand substitution reaction using glycine methyl esters flourishes the PEGylated sawhorse complexes 4-7 with 61-74% yield. The light-yellow products 4-7 are moisture and air-stable easily handled without proof from oxygen or light. The thermolysis of Ru 3 (CO) 12 1 and aromatic carboxylic acid at 120 • C in toluene generates hexacarbonyl Ru(I) intermediates 2 and with the characteristic carbonyl bands at 2103 w, 2079 vs, 2035 vs, 2004 vs and 1938 w cm −1 . The axial carbonyl ligands of 2 are labile and can readily be substituted by glycine esters of ethylene glycol monomethyl ester. Thermolysis using various bridged aromatic carboxylic acids afforded 8-15, and FT-IR, 1 H-NMR, mass spectrometry and elemental analysis were used to fully characterize complexes 4-15. The IR spectra of the Ru(I) complexes identified the four characteristic carbonyl bands at 2028-1938 cm −1 of sawhorse Ru 2 (CO) 4 complexes whilst the bridged-carboxylate ligand showed C=O band at 1743 cm −1 . The NMR spectra revealed more detailed information about the molecular structures of these ruthenium complexes with more details. In the 1 H-NMR spectrum of 4, the bridged-acetate was observed as a singlet at δ 1.94 ppm, and the protons of NH 2 appeared as a singlet at δ 2.90 ppm. CH 2 and OCH 3 of axial ethylene glycol monomethyl ester glycine ester appeared as triplets at δ 4.35, 3.75, 3.62 ppm, and a singlet at 3.39 ppm, respectively. The 13 C{ 1 H} NMR spectrum of 4 shows three types of CO resonances, and the Ru bonded carbonyl groups are at δ 204 ppm, the bridging carboxylate is at δ 172 ppm, and the ester group is at δ 184 ppm, respectively. 4-15 showed similar resonances for both bridged and axial ligands. Notably, the amino protons of µ 2 -acetato complex 4 appear at δ 2.90 ppm, lower than the corresponding chemical shift of µ 2 -arylcarboxylato complexes 8-15. The shielded amino proton of 4 reflects less electron donation to the sawhorse unite, indicating the corresponding axial ligand may be more labile.
Materials 2022, 15, x FOR PEER REVIEW 3 of 8 logical incompatibility of sawhorse ruthenium complexes are obstacles to their application in in the biological system [26]. To increase the hydrophilicity of these ruthenium carbonyl complexes, the water-soluble amino acid glycol and PEG esters were tethered at the axial position of the sawhorse structure, respectively (Scheme 2). The thermolysis of Ru3(CO)12 1 in acetic acid affords polymeric sawhorse ruthenium carbonyl 2, and the sequential ligand substitution reaction using glycine methyl esters flourishes the PEGylated sawhorse complexes 4-7 with 61-74% yield. The light-yellow products 4-7 are moisture and air-stable easily handled without proof from oxygen or light. Yellow crystals of three diruthenium (I) complexes 8 (Figure 1), 13 and 14 were obtained via the diffusion of petroleum ether to CH2Cl2 solution of the complexes at 0 °C. Single crystal X-ray diffraction was used to analysis confirmed the molecular structures of these complexes, and the selected bond distances and angles are shown in Table 1. The molecular structures feature a typical sawhorse structure that consists of a diruthenium tetracarbonyl core surrounded by two ethylene glycol monomethyl ester glycine esters as axial ligands and two arylcarboxylate ligands at equatorial positions. Three crystals belong to the monoclinic system with the C2/c space group. The Ru-Ru bond distances in these sawhorse skeletons are 2.6694(10) Å (8), 2.6634(7) Å (13) and 2.6727(5) Å (14), respectively; these values are with a metal-metal single bond [27]. The Ru-CO bond length of each terminal carbonyl is slightly different. For instance, the average Ru-CO bond length of these complexes is about 1.83 Å, which is shorter than Ru-CO (1.943(3) Å and 1.903(3) Å) of CO-RM-3 [10], but longer than Ru-CO(1.76 Å) in those of axial triphenylphosphine analogues [27]. In complex 8, Ru(1)-N(1) 2.239(5) Å is slightly more than Ru (2) Yellow crystals of three diruthenium (I) complexes 8 (Figure 1), 13 and 14 were obtained via the diffusion of petroleum ether to CH 2 Cl 2 solution of the complexes at 0 • C. Single crystal X-ray diffraction was used to analysis confirmed the molecular structures of these complexes, and the selected bond distances and angles are shown in Table 1. The molecular structures feature a typical sawhorse structure that consists of a diruthenium tetracarbonyl core surrounded by two ethylene glycol monomethyl ester glycine esters as axial ligands and two arylcarboxylate ligands at equatorial positions. Three crystals belong to the monoclinic system with the C2/c space group. The Ru-Ru bond distances in these sawhorse skeletons are 2.6694(10) Å (8), 2.6634(7) Å (13) and 2.6727(5) Å (14), respectively; these values are with a metal-metal single bond [27]. The Ru-CO bond length of each terminal carbonyl is slightly different. For instance, the average Ru-CO bond length of these complexes is about 1.83 Å, which is shorter than Ru-CO (1.943(3) Å and 1.903(3) Å) of CO-RM-3 [10], but longer than Ru-CO(1.76 Å) in those of axial triphenylphosphine analogues [27]. In complex 8, Ru(1)-N(1) 2.239(5) Å is slightly more than Ru(2)-N(2) 2.210(5) Å, and the average Ru-Ru-N angle is 158 • , indicating the former axial ligand might be more labile and readily dissociate during the CO releasing process. Interesting, in complex 13, Ru(1)-N(1) and Ru (2) 2.210(5) Å, and the average Ru-Ru-N angle is 158°, indicating the former axial ligand might be more labile and readily dissociate during the CO releasing process. Interesting, in complex 13, Ru(1)-N(1) and Ru(2)-N(2) with same distance at 2.249(4) Å, but Ru(2)-Ru(1)-N(1) = 160.28(11)° bigger than Ru(1)-Ru(2)-N(2) = 159.09(11)°.   (5) 157.86 (15) 158.27 (15)  13 2.6634 (7) 2.249 (4) 2.249 (4) 159.09 (11) 160.28 (11)  14 2.6727 (5) 2.233 (4) 2.240 (4) 158.27 (11) 159.84 (10) The CO release activity of each CO-RM in vivo was measured with the "golden standard" of myoglobin assay. Firstly, sodium dithionite was added to reduce myoglobin to deoxy-myoglobin (deoxyMb) in PBS (pH = 7.4) at 37.8 °C. A stock solution of CO-RM was added and then activated by LED-UV radiation at 365 nm, releasing CO in vivo. The change of deoxyMb to carbonmonoxy-myoglobin (MbCO) was monitored by UV-vis spectroscopy. A typical series of electronic absorption spectra (Figure 2) showed the conversion of deoxyMb to MbCO in the presence of CO released from 4. The half-lives of CO release rate of 4 at 60 μM is 166 s, 40 μM is 172 s, and 20 μM is 267 s. Four isosbestic points demonstrated the biocompatibility of this CO-RM. The dark-controlled CO release experiment showed that all CO-RMs are stable and do not spontaneously degrade under physiological conditions. Tuning the time and density of UV radiation also control the CO releasing kinetics from CO-RM whilst the molecular structures of CO-RMs determine their photo-sensitivity and CO release activity.    (11) 159.84 (10) The CO release activity of each CO-RM in vivo was measured with the "golden standard" of myoglobin assay. Firstly, sodium dithionite was added to reduce myoglobin to deoxy-myoglobin (deoxyMb) in PBS (pH = 7.4) at 37.8 • C. A stock solution of CO-RM was added and then activated by LED-UV radiation at 365 nm, releasing CO in vivo. The change of deoxyMb to carbonmonoxy-myoglobin (MbCO) was monitored by UVvis spectroscopy. A typical series of electronic absorption spectra (Figure 2) showed the conversion of deoxyMb to MbCO in the presence of CO released from 4. The half-lives of CO release rate of 4 at 60 µM is 166 s, 40 µM is 172 s, and 20 µM is 267 s. Four isosbestic points demonstrated the biocompatibility of this CO-RM. The dark-controlled CO release experiment showed that all CO-RMs are stable and do not spontaneously degrade under physiological conditions. Tuning the time and density of UV radiation also control the CO releasing kinetics from CO-RM whilst the molecular structures of CO-RMs determine their photo-sensitivity and CO release activity.

Ru(1)-Ru(2) Ru(1)-N(1) Ru(2)-N(2) Ru(1)-Ru(2)-N(2) Ru(2)-Ru(1)-N(1)
2.210(5) Å, and the average Ru-Ru-N angle is 158°, indicating the former axial ligand might be more labile and readily dissociate during the CO releasing process. Interesting, in complex 13, Ru(1)-N(1) and Ru(2)-N(2) with same distance at 2.249(4) Å, but Ru(2)-Ru(1)-N(1) = 160.28(11)° bigger than Ru(1)-Ru(2)-N(2) = 159.09(11)°.    (4) 158.27 (11) 159.84 (10) The CO release activity of each CO-RM in vivo was measured with the "golden standard" of myoglobin assay. Firstly, sodium dithionite was added to reduce myoglobin to deoxy-myoglobin (deoxyMb) in PBS (pH = 7.4) at 37.8 °C. A stock solution of CO-RM was added and then activated by LED-UV radiation at 365 nm, releasing CO in vivo. The change of deoxyMb to carbonmonoxy-myoglobin (MbCO) was monitored by UV-vis spectroscopy. A typical series of electronic absorption spectra (Figure 2) showed the conversion of deoxyMb to MbCO in the presence of CO released from 4. The half-lives of CO release rate of 4 at 60 μM is 166 s, 40 μM is 172 s, and 20 μM is 267 s. Four isosbestic points demonstrated the biocompatibility of this CO-RM. The dark-controlled CO release experiment showed that all CO-RMs are stable and do not spontaneously degrade under physiological conditions. Tuning the time and density of UV radiation also control the CO releasing kinetics from CO-RM whilst the molecular structures of CO-RMs determine their photo-sensitivity and CO release activity. To identify the structural features of CO-RMs that govern the CO releasing behaviour, the CO release kinetics of 4-15 were correlated to the corresponding M-CO band and lipophilicity in Table 2. Firstly, the oil-water partition coefficient logP values of complexes were measured by the "flask-shaking" method with n-octanol and water, respectively [28]. The logP value of 4-13 ranges from 0.39 to 1.78. The water-solubility of these sawhorse ruthenium complexes mainly depend on both the axial glycol amino esters and bridging carboxylate, respectively. As shown in Table 2, the more hydrophilic CO-RMs release carbon monoxide faster. 4 exhibited the lowest logP value of 0.39, which release CO fastest and convert 30 µm Mb to MbCO for just 163 s using 60 µm CO-RM. Since the benzyl substitutes of axial ligand significantly reduce the hydrophilicity of 5, The logP value of 5 increased to 1.41 and the CO release half-life of 5 t 1/2 , 60 µm decreased to 276 s. With the increase in the degree of polymerization of PEG chain, it is conducive to improving the water solubility of CO-RM 5-7, and the CO release rate increases with t 1/2 , 60 µm from 276 s to 189 s. The substitutes on the arene are related to the hydrophilicity of aromatic carboxylate bridging CO-RM. 8, 9 and 12 showed higher logP values of 1.71, 1.67 and 1.78, respectively, which released CO much slower with t 1/2 ,60 µm around 1000 s. Interestingly, The parasubstituted methyl-, chloro-10, 11 and meta methoxy-13 groups increase the hydrophilicity and the corresponding CO rate of each aromatic CO-RM. 14 and 15 are too hydrophobic to be evaluated via the "flask-shaking" method, in which t 1/2 , 60 µm is over 2000s. Compared with compounds 4, 8-15 with benzyl H atom and polymerization degree n = 1, the structure of acetic acid bridging is better than that of aryl acid bridging. The methyl and methoxy substituents on the benzene ring can improve the water solubility to a certain extent, but the increase in the chain length of bridged alkyl carboxylic acid will significantly reduce the water solubility. Therefore, acetic acid as bridging ligand and hydrophilic PEGlyated ester are essential to improving the water solubility of CO-RM and accelerating the release of CO.

Entry Ru(1)-Ru(2) Ru(1)-N(1) Ru(2)-N(2) Ru(1)-Ru(2)-N(2) Ru(2)-Ru(1)-N(1)
Each component of CO-RM, such as Ru, carboxylic acid and amino acid esters, is considered non-toxic than most of the other CO-RMs. The cytotoxicity of CO-RM is still unknown. To evaluate the bioactivity of PEG-CORM with different functional groups, we investigated the cytotoxicity of 4 and 8 with the murine macrophage cell line, RAW 264.7 and the human colon adenocarcinoma cell line, HT29. Generally, IC50 of 4 over two cell lines showed less cytotoxicity in the dark, in constant with the cellular protection effect of endogenous CO. The MTT experiments showed that 100 µM to 500 µM of 4 significantly impacted the cell viability of RAW264.7 (Figure 3a), IC50 value is 253.3 µM. 100 µM of 4 started to reduce the cell survival rate. As the concentration of CO-RM increased, the survival rate of the cell dropped sharply. In the presence of 400 µM and 500 µM of 4, RAW264.7's survival rate was 22.3% and 10%, respectively. Due to its limited watersolubility, the MTT experiments of 8 were carried out at 10-50 µM (Figures S17 and S18). impacted the cell viability of RAW264.7 (Figure 3a), IC50 value is 253.3 μM. 100 μM of 4 started to reduce the cell survival rate. As the concentration of CO-RM increased, the survival rate of the cell dropped sharply. In the presence of 400 μM and 500 μM of 4, RAW264.7's survival rate was 22.3% and 10%, respectively. Due to its limited water-solubility, the MTT experiments of 8 were carried out at 10-50 μM (Figures S17 and S18). 8 showed little effect on RAW264.7 cell viability at a concentration of 10 μM, 99.6%. The survival rate of RAW 264.7 cells decreased slightly with the increase in concentration; 50 μM of 8 decreased to 94.6%, similar to the toxicity of 4.
(a) (b) The human colon adenocarcinoma cell line HT29 received particular interest in studies focused on food digestion and bioavailability due to its ability to express characteristics of mature intestinal cells. To evaluate the potential of PEG-CO-RMs for CO therapy as anticancer agents, we measured the cytotoxicity of 4 and 8 over HT-29. Experiments with HT29 cells used the concentration of 4 in the range from 50 μM to 700 μM (Figure 3b), and 8 in the range of 10-50 μM (Figure S19). At a concentration of 50 μM 4 and 8, 12.5% and 6.65% of the cells lose activity in light stimulation, respectively. Compound 4 showed a better inhibitory effect on cancer cells, indicating that alkyl bridged carboxylic acid ligands with higher toxicity than aryl carboxylic acid ligands. Moreover, when the concentration increased to 700 μM, HT29 cells survived just 24.7%. Interestingly, 4 showed similar anticancer activity in the dark, indicating the anticancer activity of 4 might result from PEG-CO-RM as a whole rather than its' released CO.

Materials and Methods
All manipulations were accomplished with standard Schlenk techniques. Decacarbonyl-ruthenium (Ru3(CO)12) and mPEG amino acid esters were prepared according to literature procedures [27]. CO releasing test were performed using myogblin assay. The cytotoxicity and anticancer activity were measured with RAW264.7 and HTC-29 cells, respectively. The details of experiments were listed in ESI.

Conclusions
In conclusion, the robust sawhorse skeletons of the diruthenium carbonyl complex were devised with PEGylated ligands to tune the CO releasing and bioavailability of CO- The human colon adenocarcinoma cell line HT29 received particular interest in studies focused on food digestion and bioavailability due to its ability to express characteristics of mature intestinal cells. To evaluate the potential of PEG-CO-RMs for CO therapy as anticancer agents, we measured the cytotoxicity of 4 and 8 over HT-29. Experiments with HT29 cells used the concentration of 4 in the range from 50 µM to 700 µM (Figure 3b), and 8 in the range of 10-50 µM (Figure S19). At a concentration of 50 µM 4 and 8, 12.5% and 6.65% of the cells lose activity in light stimulation, respectively. Compound 4 showed a better inhibitory effect on cancer cells, indicating that alkyl bridged carboxylic acid ligands with higher toxicity than aryl carboxylic acid ligands. Moreover, when the concentration increased to 700 µM, HT29 cells survived just 24.7%. Interestingly, 4 showed similar anticancer activity in the dark, indicating the anticancer activity of 4 might result from PEG-CO-RM as a whole rather than its' released CO.

Materials and Methods
All manipulations were accomplished with standard Schlenk techniques. Decacarbonylruthenium (Ru 3 (CO) 12 ) and mPEG amino acid esters were prepared according to literature procedures [27]. CO releasing test were performed using myogblin assay. The cytotoxicity and anticancer activity were measured with RAW264.7 and HTC-29 cells, respectively. The details of experiments were listed in ESI.

Conclusions
In conclusion, the robust sawhorse skeletons of the diruthenium carbonyl complex were devised with PEGylated ligands to tune the CO releasing and bioavailability of CO-RMs. The myoglobin assay test on the CO releasing rate showed well-controlled release kinetics of CO-RMs 4-13 with t 1/2 , 60 µM from 166 s to 2699 s. The logP values of CO-RMs were correlated with CO release rates, revealing the intrinsic relationship between CO-RMs' water-solubility and CO releasing activity. The CO-RMs with smaller logP released CO faster, which might prove the concept of enhancing water-solubility to improve the CO release properties. The hydrophilicity of CO-RM was finely tuned via selecting carboxylate bridging ligands and glycol amino acid esters. MTT assay confirmed that CO-RM 4 consisted of acetate and glycol glycine ester less cytotoxic to RAW264.7, but toxic to HT29 cancer cells. These CO releasing and bioactivity experiments demonstrated the PEGylated Sawhorse ruthenium carbonyl complex's drug-like properties and the promising therapeutic potential.