lable at ScienceDirect

Journal of Organometallic Chemistry 760 (2014) 89e100

Contents lists avai

Journal of Organometallic Chemistry

journal homepage: www.elsevier .com/locate/ jorganchem

The effect of specific modifications of the amine ligands on thesolubility, stability, CO release to myoglobin and whole blood, celltoxicity and haemolytic index of [Mo(CO)4(NR3)2] complexes

Lukas Kromer a, Ana Catarina Coelho a, Isabel Bento a, Ana Rita Marques b,Carlos C. Romão a,b,*

a Instituto de Tecnologia Química e Biológica, Universidade Nova de Lisboa, Av. da República, EAN, 2780-157 Oeiras, PortugalbAlfama Lda., Instituto de Biologia Experimental e Tecnológica, Av. da República, EAN, 2780-157 Oeiras, Portugal

a r t i c l e i n f o

Article history:Received 30 September 2013Received in revised form3 December 2013Accepted 5 December 2013

Keywords:CORMCO releasing moleculesMolybdenum carbonyl

* Corresponding author. Instituto de Tecnologiaversidade Nova de Lisboa, Av. da República, EAN, 278

E-mail addresses: carlosromao@mac.com, ccr@itqb

0022-328X/$ e see front matter � 2013 Elsevier B.V.http://dx.doi.org/10.1016/j.jorganchem.2013.12.009

a b s t r a c t

A series of cis-[Mo(CO)4(amine)2] complexes (NR3 ¼ morpholine 1; 4-Meepiperazine, 2;H2NCH2CH2NH2, 3; H2NCH2CH2emorpholine (4) R2NCH2CH2epiperazinee4-Me (R ¼ H, 5); R ¼ Me, 6);Me2NCH2CH2NMe2, 7) was prepared in good yields, in a one-step microwave-assisted synthesis. The X-ray diffraction structures of the complexes 4, 5 and 6 are reported. The stability of the complexes 1e7 inaqueous, aerobic media was studied by UVeVis spectrophotometry, RP-HPLC and gas chromatography atseveral pH values. Stability beyond 1 h requires bidentate ligands with at least one tertiary amine ligandand increases in the order 4 < 5 < 6. Stability is approximately the same at pH 7.5 and pH 3.9 for 5 and 6in solutions acidified with HCl. Acidification with CF3COOH induces decomposition. The order of COtransfer rate to deoxy-Mb and haemoglobin in whole blood is 1 > 2 > 3 > 4 > 5 > 6 >> 7, but it is muchfaster to whole blood. The haemolytic index of some compounds increases in a similar order:1 < 2 < 5 < 6; with the exception of 1, the complexes are not toxic to RAW264.7 cells up to a con-centration of 100 mM.

� 2013 Elsevier B.V. All rights reserved.

1. Introduction

Carbon Monoxide (CO) is an essential signalling molecule pro-duced in the body by the action of the enzyme Heme Oxygenase(HO), which catabolises toxic heme released by dying cells. It wasthe second reported member of the group of simple mediators nowcalled gaseotransmitters: NO, CO and H2S [1]. Its endogenousproduction plays a very important role in innate immune defenceand cytoprotection [2,3]. Accordingly, exposure of cells, tissues orrodents to exogenous CO produces important therapeutic results ina very broad range of medical indications, some of which do nothave alternative treatments [4e7]. Although this finding alreadyled to human clinical trials [8], the application of CO gas as atherapy, namely in humans, suffers from a number of shortcom-ings, as discussed elsewhere [9]. This fact was recognised soon afterthe first report on the therapeutic action of CO gas and led to thesearch for pro-drugs capable of delivering CO to biological systems

Química e Biológica, Uni-0-157 Oeiras, Portugal..unl.pt (C.C. Romão).

All rights reserved.

and animals in safer and more easily controllable manners,compatible with its ambulatory use. Such pro-drugs were namedCORMs or CO-Releasing Molecules, by Motterlini and Mann whoreported the first examples based on transition metal carbonyls[10]. These CORMs, the dimethyl sulfoxide (DMSO) soluble[Ru(CO)3Cl2]2 (CORM-2) and the water soluble [RuCl(CO)3(glyci-nate)] (CORM-3) [11,12], have demonstrated their therapeuticproperties in a broad variety of in vivo preclinical, animal modelstudies namely in cardiovascular disease, rejection of transplantedorgans, acute lung, kidney and liver failure, cancer, sepsis and shockas recently reviewed [13,14]. They also recapitulated the action ofCO gas in the same models with the important advantage ofgenerating much lower values of carboxyhaemoglobin (COHb) insystemic circulation. Our early studies used air-sensitive zerovalentMo carbonyls, like fac-[Mo(CO)3(histidinate)]Na (ALF186) and[Mo(CO)5Br][NEt4] (ALF062) in several animal models of diseasewith evident and extensive beneficial effects although producinghigher COHb values than the CORM-2 and CORM-3 congeners [15e17]. These four molecules remained the only transition metalCORMs reported to be used in vivo until 2012. In spite of theirencouraging results in proof-of-concept studies they lack a series ofphysical and chemical properties that are necessary to make their

L. Kromer et al. / Journal of Organometallic Chemistry 760 (2014) 89e10090

pharmaceutical use acceptable. As discussed elsewhere [9], therational implementation of these favourable properties on a tran-sition metal carbonyl complex implies building a suitable coordi-nation sphere around the metal ion of choice. This constructioncomprises two layers: an inner coordination sphere that de-termines the chemistry around the central metal, and an outercoordination sphere that creates an acceptable ADME profile, andtargets the diseased tissues. This methodology has led us to prepare[Mo(CO)3(CNCMe2COOH)3], an advanced drug-like CORM for thetreatment of acute liver failure [18]. Other authors have also pro-duced CORMs with many drug-like properties and rather sophis-ticated coordination spheres, like the vitamin B12 conjugate[Re(CO)2Br2(H2O)(cyanocobalamine)] [19], and the esterase orphosphatase triggered CORMs [Fe(CO)3(h4-C6H7OX)] (X ¼ C(O)Me;P(O)(OMe)2) [20e22] but they were never reported to be testedin vivo. Since CORMs are pro-drugs, the double sphere constructionprocess must have in mind that at some point, ideally at the site ofdisease, CORMs must breakdown to release CO. In other words,there must be a balance between stability in circulation andinstability in the tissues. The latter can only be tuned throughin vivo studies since it depends strongly on targeting and phar-macokinetics, that is, the outer sphere. On the contrary, makingcomplexes that are stable in circulation depends on the inner co-ordination sphere, relies on in vitro studies, and is the starting pointin CORM design. Accordingly, CORM candidates must be stable toair, water and plasma. Moreover, they must present sufficient sol-ubility in water and be devoid of haemolytic and cytotoxic prop-erties. However, the still limited knowledge of the chemistry ofmetal carbonyl complexes in aqueous, aerobic and protein con-taining media makes the assembling of inner coordination sphereswith these characteristics a rather difficult task with many unpre-dictable hurdles and assumptions. In fact, besides the kineticallyinert d6 tricarbonyl complexes of Tc(I) and Re(I), which have beenintensively studied for radiopharmaceutical use [23], very fewmetal carbonyl complexes have been studied in aqueous biologicalmedia.

Penta-, tetra- and tricarbonyl complexes of group 6 metals withligands like pyrrolidine, piperazine and morpholine [24], cyclo-hexylamine, pyridine and ethylenediamine [25], and ethylenedi-amine derivatives [26], have been known for a long time. Althoughpiperazine, morpholine and polyamine ligands offer excellentpossibilities for enhancing the water solubility of their hydrophobicmetal carbonyl complexes, due to either protonation or formationof hydrogen bonds, such complexes were never tested in aqueousaerobic conditions. There is only one example of a study of potentialCORMs of this type, namely that of the amino-acid complexes of thetype [Mo(CO)5(H2NCRR0CO2R00)] [27]. The traditional synthesis ofthese amine complexes was done by refluxing solutions of Mo(CO)6with the corresponding ligand in benzene or toluene. Dependingon the reaction time the corresponding mono- and disubstitutedcomplex was obtained usually with contamination by each other.Access to these species became much easier when Hogarth andArdon introduced a modified conventional microwave oven tosynthesise molybdenum carbonyl complexes. This technique al-lows the synthesis of molybdenum tetracarbonyl complexes[Mo(CO)4L2] with amine ligands in a much shorter time and withhigher purity than the traditional refluxmethod [28e30]. The over-heating of high boiling point solvents in open microwave systems(e.g. diglyme from 162 to 175 �C) leads directly to reaction accel-eration and therefore less time for air decomposition of unstableintermediates [31].

Extending our previous work on zerovalent molybdenumcarbonyl based CORMs, we now present our findings on a series of[Mo(CO)4(NR3)2] complexes bearing saturated amine ligands, andprepared by microwave-assisted synthesis. In this study we

investigate the impact that specific ligand modifications made tothe inner sphere of [Mo(CO)4(NR3)2] complexes have on the sta-bility, solubility, CO release, haemolytic and cytotoxicity profiles ofthe respective complexes. The objective is to identify trends andguidelines for the future design of CORM scaffolds based on amineligands for therapeutic use in vivo.

2. Results and discussion

2.1. Synthesis and characterisation

A series of cis-[Mo(CO)4(amine)2] complexes (1e7; Scheme 1)was prepared in a one-step synthesis directly from [Mo(CO)6] andseveral saturated mono- and bidentate amines, by usingmicrowave-assisted heating with an open reflux system. All com-plexes were obtained in less than 16 min reaction time, using apower of 700 W, as yellow crystalline solids in high purity andyields between 30 and 82%. Compounds 1 [24], 3 [25,30] and 7 [32]are well described and characterised in the literature. However,their preparation by this less conventional method led to excellentresults, reducing both time consumption and the formation ofsecondary products, when compared to traditional refluxtechniques.

The compounds 1 and 2 bearing the monodentate secondaryamines morpholine and 1-methylpiperazine, respectively, provedslightly sensitive to air and moisture changing colour from yellowto brownish when kept in closed vials under air. On the other hand,compounds 3e7 are air stable for periods of days and can bemanipulated on the bench without any inert atmosphere pre-cautions. This first observation indicates that the use of bidentatesaturated amines (NeN) improves the stability of [Mo(CO)4(NeN)]complexes to atmospheric conditions when compared to theirmonodentate analogues Mo(CO)4(NR3)2.

The FT-IR spectra of 1e7 show a pattern of four distinct COstretching frequencies in the 2021e1775 cm�1 range consistentwith a cis-tetracarbonyl geometry. The NeH stretching frequencieswere also observed for complexes 1e5, as sharp bands, between3240 and 3370 cm�1. The lowering of the NeH stretching fre-quency, when compared to free 1-methylpiperazine (3266 cm�1)indicates that in compound 2 the ligand binds via the secondaryamine, as drawn in Scheme 1, and not via the tertiary amine group.

Single crystals suitable for X-ray diffraction studies were ob-tained for complexes 4, 5 and 6 by layering the reaction solution(diglyme/THF) with diethyl ether or n-hexane. The structures arerepresented in Fig. 1. Each complex has the metallic centre coor-dinated to four carbonyl groups and to a saturated bidentate ni-trogen amine ligand via the amine of the side chain and through thetertiary amine of the morpholine or the piperazine ring, giving aslightly distorted octahedral geometry. Comparable crystal struc-tures of the type [Mo(CO)4(NR3)2] where (NR3) is a non-cyclamamine are limited to a few structures, namely [Mo(CO)4(piper-idine)2] (pip) [33], [Mo(CO)4(diamino-monosaccharide)] (saccha-ride) [34], [Mo(CO)4((PhCH2)HN(CH2)2NMe2)] (Bn-en-Me2) [26]and [Mo(CO)4(TMEDA)] (TMEDA) (7) [32]. Selected bond lengthsand bite angles are listed and compared in Table 1.

The MoeN bond lengths in complexes 4 and 5 represent twoextreme situations. On the one hand the MoeN1 distances of2.282(3) and 2.291(1) �A, respectively, are quite short for a MoeNbond but comparable with the other Mo-(primary amine) bonds ofthe saccharide complex (2.284(3) �A). On the other hand, MoeN2distances are extraordinarily long. The values of 2.410(3) and2.404(1) �A are the longest observed for Mo0eamine bonds. Thisdiscrepancy between the twoMoeN distances was also observed inthe corresponding octahedral [Ni(NeN)2X2] complexes and inter-preted as a consequence of steric constraints introduced by the

Scheme 1. Microwave synthesis of [Mo(CO)4(amine)2] complexes 1e7 used in this work: Mo(CO)6 and 5 equiv of the corresponding ligand in diglyme/THF are heated to reflux in amodified microwave oven for 2e16 min.

L. Kromer et al. / Journal of Organometallic Chemistry 760 (2014) 89e100 91

bulky morpholine or piperazine group [35e37]. Another conse-quence of these steric constraints is that in spite of the longMoeN2bonds, the N1eMoeN2 bite angles of 76.4(1) and 76.52(5)�,respectively, are remarkably small and only about 1.7� larger thanthe corresponding bite angle of 74.83(9)� found in the saccharidestructure, which is imposed by the rigid carbohydrate backbone. Inthe case of complex 6, the MoeN bonds are not so different fromeach other because both N donors are tertiary amines. In fact, thesebond lengths are close to those found in the TMEDA complex 7.Given the formation of such strong bonds to primary amines, it isrelevant to notice that the complexes 4 and 5 preferred a bidentatecoordination of only one diamine ligand to the coordination of twosuch ligands via the terminal NH2 group.

Fig. 1. ORTEP plot of complexes 4, 5 and 6 (Ellipsoids are drawn with 50% probability,

2.2. Solubility and stability studies

All the complexes prepared are soluble in commonpolar organicsolvents such as dichloromethane, methanol, acetone, acetonitrileand DMSO, and they are insoluble in non-polar solvents like n-hexane. In aqueous or physiological media (pH z 7) the com-pounds are poorly soluble and totally insoluble in the cases of 2 and7. However, the novel piperazine containing compounds 5 and 6can be easily solubilised in acidic conditions. Addition of 1 equiv ofHCl to their aqueous suspensions increases their solubility due tothe protonation of the distal, non-coordinated secondary N atom ofthe piperazine ring that generates a cationic complex. Interestingly,addition of more HCl to produce pH z 2 markedly increases the

solvent molecules and hydrogen atoms of carbon groups are omitted for clarity).

Table 1Selected bond length (�A) and angles (�) of complexes 4, 5, 6 and reference complexestaken from literature sources.

Complex MoeN1 MoeN2 N1eMoeN2 Ref.

4 2.282(3) 2.410(3) 76.4(1) This work5 2.291(1) 2.404(1) 76.52(5) This work6 2.339(2) 2.389(2) 78.32(7) This workTMEDA (7) 2.334(2) 2.345(2) 78.0(1) 35pip 2.345(2) 85.97(10) 36Bn-en-Me2 2.342(3) 2.317(3) 77.0(1) 29Saccharide 2.284(3) 2.320(3) 74.83(9) 37

Fig. 2. UVeVis spectra of solutions of 2 taken immediately after preparation in theaqueous media indicated (see colour in Fig. S7).

L. Kromer et al. / Journal of Organometallic Chemistry 760 (2014) 89e10092

solubility. For CO release and cell toxicity studies where such lowvalues of pH are not acceptable, the compounds were solubilised inwater with 10% methanol or DMSO. Even with this method, com-plex 7 couldn’t be solubilised. The approximate maximum solubi-lity obtained for all compounds is given in Table 2. Thedetermination of accurate solubility values would be very difficultand hampered by the instability of the compounds. In fact,decomposition of some of them was not unexpected and could beobserved rapidly after dissolution. Such decomposition was eval-uated for complexes 1e6, by means of UVeVis spectroscopy, RP-HPLC and GC chromatography.

The UVeVis spectrum of each compound was obtained inacetonitrile and was taken as a reference for the next studies sincethey are soluble and stable in this solvent. The stability of thecompounds in aqueous mediawas studied through the evolution oftheir UVeVis spectra, which were compared to the reference inacetonitrile. To this end, compounds were dissolved in water at pH10.2, distilled water (pH 6.2), water with 0.1% of TFA (pH 2.0)(TFA ¼ trifluoroacetic acid), methanol/water with 0.1% TFA(pH z 2.0) and aqueous HCl solution (pH z 2.0). The spectra weretaken as rapidly as possible after solution preparation. When thebis-morpholino complex 1 is dissolved in distilled water, thespectrum taken immediately after dissolution already reveals thatthemaxima at 221 and 260 nm have disappeared and a new band isrising at 250 nm (Fig. S6). Dissolving 1 in basic water (pH 10.2)shows the same type of changes. However, the spectra taken afterdissolution in acidic media (pH 2.0) indicate much more extensivedecomposition with a marked rise of the band at 250 nm. In thecase of the bis-N-methylpiperazine complex 2 this type ofdecomposition in acidic solutions is even more pronounced asshown in Fig. 2: the four maxima observed in the acetonitrilespectrum (l (nm) ¼ 221; 258; 295; 379) are replaced by newmaxima (l (nm) ¼ 249; 288; 400) indicating rapid decompositionupon solubilisation and the formation of other Mo complexesduring decomposition.

2 is insoluble in distilled water hence the lack of absorbance inthe UVeVis spectra. In acetonitrile all the bidentate complexes 3e6present very similar UVeVis spectra with three maxima at ca. 260,300 and 390 nm. Complex 3 decomposes totally upon dissolution inall aqueous media, as can be seen by the complete disappearance of

Table 2Approximate solubility values (mg/mL) in aqueous media for compounds 1e7.

Compound Solubilitylimit (mg/mL)

Solvent medium

1 5 10%DMSO (or MeOH)/H2O2 1 10%DMSO (or MeOH)/H2O3 nd e

4 5 10%DMSO (or MeOH)/H2O5 1 H2O þ 1 equiv HCl or H2O at pH z 26 5 H2O þ 2 equiv HCl or H2O at pH z 27 Insoluble e

the bands at 260, 304 and 396 nm upon dissolution (Fig. S8). Incontrast, the absorption bands of complexes 4e6 are preserved inaqueous solution after immediate dissolution. Notwithstanding,they decay along the time with different rates, which also aremedium dependent. When dissolved in MeOH/H2O(0.1% TFA) 1:9complex 4 decayed by 92% after 1 h and a new specie was formed insolution with a band at 217 nm (Fig. S9). The complex is alreadymore stable than the three previous examples, but still less stablethan the N-methylpiperazine derivatives 5 and 6 where nodecomposition is seen upon dissolution in several media (Fig. S10,S11), and the ensuing decays are much less pronounced. In the caseof 5 and 6 at the same initial concentration (100 mM) we observeddifferent amounts of decay in three different aqueous media, asgraphically represented in Fig. 3. In MeOH/H2O the percentage ofdecomposition of both complexes is modest: after 1 h complex 5and 6 only lose 28 and 11% of the initial concentration, respectively.The pH of this solution (measured in the UVeVis cuvette) was 7.5.When these complexes are dissolved inwater with excess (2 equiv)of HCl, the pH of the solution in the UVeVis cuvette is 3.9 but thedecay after 1 h is not very different: 30% for 5; 21% for 6. Quiteremarkably, the decay of both complexes 5 and 6 is dramaticallyincreased in analogous solutions containing TFA at pH z 3.7: 86%and 67% for 5 and 6, respectively. This unexpected outcomecertainly indicates that pH is not the crucial factor that determinesthe decomposition process. In fact, the pH values of the solutions inH2O/HCl andMeOH/H2O (0.1% TFA) 1:9 are very close and the decayof both compounds after 1 h is 2.5 times higher in the latter case. Onthe other hand the decomposition rate in the neutral MeOH/H2O1:9 solution is very close to that measured in the acidic H2O/HClsolution. Somehow, trifluoroacetic acid seems to be a catalyst forthis decomposition. In any case, 6 is more stable than 5.

In an attempt to improve the understanding of the decompo-sition of 5 and 6 we performed several kinetic experiments usingUVeVis spectroscopy. One of our concerns was the possible influ-ence of the spectrophotometer irradiation on decomposition vialight activated CO loss. Such influence is absent in the case of thedecomposition of 6 since assays performed by acquiring UVeVisspectra at intervals of 2 min, 15 min or 1 h for 2e12 h alwaysidentified a first order decay with a k and half-life (t1/2) values of0.18 h�1 and 3.6 h, respectively, meaning that the light does notinfluence the decomposition of complex 6 which is very clean asdepicted in Fig. 4 (see Fig. S12 for more data).

In contrast, complex 5 has a far more complicated decomposi-tion process and we were unable to obtain reproducible data. The

Fig. 3. % of decay of compounds 5 and 6 after 1 h in different aqueous media (a)MeOH/H2O(0.1% TFA) (pH 3.7) 1:9; (b) MeOH/H2O 1:9 (pH 7.5); (c) H2O þ HCl (pH 3.9).Initial concentration ¼ 100 mM.

L. Kromer et al. / Journal of Organometallic Chemistry 760 (2014) 89e100 93

observed reaction order sometimes fitted first order plots andsometimes second order plots giving half-life (t1/2) values between1.4 and 4.2 h. It is obvious that more than one process is competingin the decomposition and no isosbestic points were detected as inthe case of 6 (Fig. 4). Anyhow, the decomposition of 5 does notdepend on spectrophotometric irradiation since in a given set ofparallel experiments acquiring spectra every 2 min and every60 min gave similar half-life values, t1/2 z 1.4 h.

The stability of complexes 1e6 in aqueous acidic media was alsoexamined using HPLC chromatography. To this end we used areverse phase C-18 column and an eluent gradient (see Table S7 inSI) with the same initial composition of 10% methanol and 90%water (with 0.1% TFA) used in some UVeVis stability studiesdescribed above. From those UVeVis studies we know that thepresence of TFA accelerates the decomposition of the complexes.However, TFA is commonly used as an additive in the mobile phasefor reversed phase HPLC since it helps achieve optimum resolution

Fig. 4. UVeVis spectra of 6 dissolved in water with HCl (pH 3.9). Spectra taken every1 h, for 12 h.

and stabilization of the column during the chromatographic run.Two experiments carried out with 6 with the same gradient butwithout 0.1% TFA severely damaged the HPLC column. Thus, solu-tions were prepared for all complexes in 10% MeOH/90% H2O (0.1%TFA). A first aliquot from each solution was injected as rapidly aspossible after the preparation and eluted. Further 4 aliquots werethen analysed at 20 min intervals. The chromatographic traces arereproduced in the SI (Fig. S1eS5). The chromatograms of the so-lutions of complexes 2 and 3 only showed the peak attributable toMo(CO)6 with a retention time tr ¼ 10.3 min. The absence of thepeak due to the respective parent complex in each chromatogram isconsistent with the UVeVis analysis performed above whichshowed these complexes decompose immediately after dissolution.No other intermediate was observed under these conditions. In thechromatographic traces of 3 as in several other ones, the Mo(CO)6peak starts to decrease after a certain time due to precipitationwhich is also visible in the vial. In fact, all chromatograms of all sixcompounds showed the formation of Mo(CO)6, which was identi-fied by inspection of the FT-IR spectrum of the colourless precipi-tate isolated from the sample vial and also by comparison of the UVabsorption spectrum of the respective chromatographic peak withthat of authentic Mo(CO)6. In the case of complexes 1, 4 and 5 peakscorresponding to other species were observed. These may be in-termediates in the formation of Mo(CO)6, which was also observedas in all cases. In the chromatogram of complex 4, but not in that of5, it was also possible to identify the peak corresponding to theintact complex. This is somewhat unexpected because according toUVeVis spectrophotometry 5 is more stable than 4.

The decomposition of complex 6 was followed under the sameHPLC conditions after the initial dissolution of the complex ineither the mixture MeOH/H2O (0.1% TFA) or in water with HCl as inthe UVeVis experiments. The chromatographic traces registeredalong the analysis of the five aliquots had the same profile in bothsolutions. In each case the peak corresponding to the complex 6appears with tr ¼ 8.06 min decreases with time while the peak ofMo(CO)6 increases, as graphically represented in Fig. 5 by thevariation of the area under the curve (AUC) for each peak along the5 traces taken at 20 min intervals.

In principle, decomposition of the complexes should producefree CO and the rate of release of this gas to the headspace of thevials should give another measure of the decomposition of thecomplexes. We, therefore, followed the decomposition of 1e6 by

Fig. 5. Decay of 6 and rise of Mo(CO)6 measured by the AUC in the HPLC chromato-grams of the aliquots chromatographed every 20 min.

Fig. 6. Spectral change over 2 h of a Mb solution incubated with complex 5; conver-sion of deoxy-Mb to CO-Mb.

L. Kromer et al. / Journal of Organometallic Chemistry 760 (2014) 89e10094

GC-RCP chromatographic analysis of the headspace of solutionsin MeOH/H2O(0.1% TFA) 1:9 and MeOH/H2O or, in the cases of thecomplexes 5 and 6 also in H2O/HCl. The concentrations of thecomplexes in these experiments were 25 mM, that is, much lowerthan those used in the HPLC and UVeVis experiments, but closerto those having biological meaning. None of the compounds 1e6releases H2, which could eventually arise from Hþ reduction byzerovalent Mo complexes or by the wateregas shift reaction.Much more surprising, however, was to observe the exceedinglysmall quantity of CO that is released in these conditions(Table S8eS13). This value consistently ranged between 0.1 and0.2% of the total load of CO in the compounds. This means thatfrom the 4 equiv of CO of these complexes only ca.0.005 equiv are released! This non-anticipated observation is,however, compatible with the HPLC stability studies that revealthe abundant formation of Mo(CO)6.

The above results indicate that the stability of the[Mo(CO)4(NR3)2] complexes can be tuned by the judicious choice ofthe amine ligands. Replacement of monodentate ligands bybidentate ligands improves the air stability of the respective com-plexes. 1 and 2 are much less stable in air than all other ones.Increased alkylation of the N donors results in increased stabilityboth in air and in solution, but may lead to lower solubility inaqueous media as exemplified by 7 which is highly air stable butvery insoluble inwater containing solvents. However, the solubilityproblem can be circumvented appending charged (or chargeable)functions at distal positions of the ancillary ligands. The compari-son between 5 and 6 illustrates all these points. The N donors of 6are fully alkylated, hence stability is increased relative to 5. Both areessentially insoluble in water at neutral or basic pH yet sufficientlysoluble after being protonated. The increased stability obtainedupon alkylation of the N donors atoms is not fully understood yet.In principle, it could be due to the elimination of the pathway thatleads to the conjugate base dissociative substitution mechanism,which accelerates dissociative substitution at the position cis toNR2H ligands in octahedral complexes [38,39]. Deprotonation of aNR2H ligand forms the strong p-donor NR2

� ligand (amine conju-gate base), which stabilizes the empty coordination position leftupon dissociation of a ligand at an adjacent (cis) position. Thismechanism is operative in the substitution of CO ligands in[W(CO)5(k2-O2CCH2NRR0)]� (R, R0 ¼ H, Me) only when R or R0 areprotons [38]. However, in our case, the expected enhancement ofthis effect at higher pH is not evident in Fig. 3 where decompositionat pH 7.5 is shown to be slower than in any of the other acidicconditions. In any case, 5, which has a NH2 ligand decomposes al-ways faster than 6, which has one NMe2 ligand in the corre-sponding coordination position. Another tempting, yet speculativeexplanation is that increased hydrophobicity around the central Moatom protects it from the reactionwith H2O and protons improvingthe stability of the coordination sphere.

Another important observation is that, in this case, decompo-sition of the metal carbonyl complexes is not directly reflected inthe amount of CO that is set free to the headspace. Regardless of thedetails of the decomposition mechanism it is evident that unsatu-rated fragments resulting from the dissociation of amine ligandsare powerful and efficient scavengers of CO that ultimately lead tothe formation of the insoluble and inert Mo(CO)6. Although theformation of Mo(CO)6 in the decomposition of several otherMo(CO)xL6�x has been reported [40,41], it is not always observed atleast with this efficacy, as proven by the extensive liberation of COto the headspace of solutions of [Mo(CO)3(L3)]Na (L3 ¼ histidinate,nitrilotriacetate) [15]. Notwithstanding, for biological purposes it isimportant to note that the CO scavenging activity of these species iseasily overridden by that of heme proteins as described in the nextsubsection.

2.3. CO release studies

The amount of CO released in the decomposition of the com-plexes 1e6 was measured by the myoglobin assay [10], and by thequantification of the amount of carboxyhaemoglobin (COHb)formed by addition of the compound to whole sheep blood inAlsever’s solution. The less soluble complex 7 did not allow the Mbassay to be carried out with accuracy, since its solutions in 10%DMSO/H2O presented some turbidity, a well-identified cause oferror in these tests [42].

Before running the Mb assay, we ascertained that dithionitedoes not induce the liberation of CO from these complexes in thesame medium and stoichiometry used for the Mb assay. Therefore,the Mb test is fully reliable for the determination of these COrelease values [43]. As an example, Fig. 6 shows the conversion ofdeoxy-myoglobin (deoxy-Mb) to carboxymyoglobin (COMb) whenMb was incubated with complex 5. As mentioned in the UVeVisspectrophotometric decay studies above, the decomposition of 5was found irreproducible. In contrast, Fig. 6 shows that it isperfectly well behaved in the context of the Mb assay. Similarbehaviour was observed for the other complexes studied in Table 3.This suggests that side reactions perceived through UVeVis spec-troscopy do not interferewith the release of CO, which is scavengedby deoxy-Mb. Fig. 7 shows the equivalents of CO released over timefor compounds 1e6. The results show the high instability of com-plexes 1 and 2, which instantly transferz0.6 andz1.5 equiv of COto myoglobin, respectively, reaching a plateau after 5 min of incu-bation. It is possible that such plateau results from the formation ofa stable side product due to the interaction of the carbonyl specieswith side chains of the myoglobin or to the formation of Mo(CO)6which does not release any more CO. The half-lives for CO transferof the complexes are given in Table 3.

Complexes 3e6 show a slower transfer with half-lives between15 min (3, 4) and 89 min (6). The results also show that complexeswith bidentate amine ligands have a more controllable CO releaseprofile. It shouldn’t be forgotten that these results are obtained inthe essentially anoxic conditions of the reduced deoxy-Mb solution.Therefore, Mo0 is not being oxidised and the transfer of CO resultsfrom deoxy-Mb scavenging the CO dissociated from the Mo0

complex either through displacement by the solvent or reactionwith the protein. Such interactions between CORM fragments andproteins have been already identified for [Ru(CO)3]2þ based CORMs

Table 3The half-lives (t1/2 [min]) of CO transfer to deoxy Mb and whole blood Hb forcomplexes 1e7. t1/2 in Mb assay: time to transfer 0.5 equiv of CO to myoglobin inPBS. t1/2 inwhole blood assay: time to transfer of 0.5 equiv of CO to haemoglobin inwhole sheep blood.

Complex COMb COHb

1 <1 <12 <1 <13 16 <14 15 45 20 36 89 157 nd 110

L. Kromer et al. / Journal of Organometallic Chemistry 760 (2014) 89e100 95

like CORM-2 and CORM-3, but not yet for Mo based CORMs [44e47,15].

These conditions are obviously very different from those met bya CORM when it reaches the blood stream following i.p. or i.v.administration. In these clinically relevant conditions O2, plasmaproteins and red blood cells may interfere with the release of CO,which will be scavenged by Hb in circulation. This can result in anundesired raise of the %COHb level in systemic circulation, whichshould be avoided due to toxicity reasons. The resistance of metalcarbonyl complexes to blood can be practically evaluated by bloodoximetry where %COHb, %O2Hb, %metHb and total Hb are readily

Fig. 7. Equivalents of CO per molecule transferred from complexes 1 (-) and 2 (C)(top), and 3 (:), 4 (�), 5 (A) and 6 (✴) (bottom) to deoxy-Mb, in PBS at 25 �C.

measured after incubating the complexes with whole sheep bloodin Alsever’s solution at 37 �C [15].

The water soluble complexes 5 and 6were dissolved in saline byprotonation with 1 equiv. of HCl and added to the blood in a dosecorresponding to the administration of 10 mg/kg to a 20 g mouse(0.125 mg/mL of blood) therefore not saturating all the haemo-globin with CO and working in therapeutically relevant concen-trations. From the oximetric readout of that blood the equivalentsof CO, which were transferred per molecule, were calculated andplotted vs time (Fig. 8). The complexes 1e4 and 7were dissolved inDMSO as a concentrated solution and measured in the same way.Both vehicles (saline/HCl and DMSO) were measured separatelyand did not show a relevant change in the composition of hae-moglobin as read in the oximeter.

In essence, the results of the blood assay are an amplifiedversion of those obtained in the myoglobin assay. Such amplifica-tion in rate and extension of CO release can be due not only to thehigher temperature of the test, but also to the higher affinity of Hbfor CO and to the presence of oxygen. The effect of the plasmaproteins was not ascertained. Complexes 1 and 2 release ca.1 equiv of CO very rapidly (within the time of mixing), but thisvalue grows very little over the time very much like in the case oftheir reaction with deoxy-Mb but with an inverted reactionextension: 1 releases more CO to the blood than 2, whereas thereverse is true in the Mb assay. The ranking of the rate and

Fig. 8. Equivalents of CO per molecule transferred from complexes 1 (-) and 2 (C)(top) and 3 (:), 4 (�), 5 (A), 6 (✴) and 7 (6) (bottom) to haemoglobin of whole sheepblood in Alsever’s solution at 37 �C.

L. Kromer et al. / Journal of Organometallic Chemistry 760 (2014) 89e10096

extension (amount) of CO release within each assay is the same:3 > 4 > 5 > 6 > 7. However, the absolute value of CO released inboth tests for each compound is quite different: the maximumvalue of CO released for this group of compounds in the Mb assay isca. 1 equiv. for compounds 3, 4 and 5 after 1 h. In the case of theblood assay the values for CO released by compounds 3, 4 and 5after 1 h are 4, 3 and 2 equiv. respectively. Complex 3 shows thehighest variation between both assays because all 4 equiv areactually released to blood Hb within 10 min. In general termscomplexes with primary amine ligands are less stable and releasemore CO: 3, 4 and 5 always release more and faster than 6 and 7.Although the difference between 5 and 6 can be seen but is small,the difference between 3 and 7 is always much larger.

The results of this test confirm that the structural variationsintroduced in the ligands (Scheme 1) on going from 1 to 7 areactually reflected in the CO release profile of these compounds inwhole blood. The order of complex stability that was observed inthe UVeVis and HPLC experiments in aqueousmedia, and in theMbassay turned out to be essentially the same depicted in Fig. 8although the latter is the most important one from the point ofview of the evaluation of the biological applicability of thesemolecules.

The present data strongly hints to the fact that none of thecomplexes is still a useful CORM for in vivo applications.

In any case, it is clear that it is possible to tune the coordinationsphere of a given family of metal carbonyls in order to improve theprofiles of stability, solubility and CO release. From a given initialmetal carbonyl scaffold the margin for improvement and tuningcan be unexpectedly large.

2.4. Cell toxicity

The toxicity of compounds 1, 2, 5 and 6 was evaluated in themacrophage cell line RAW264.7. Compound 7 could not be evalu-ated in this test due to insolubility in the 10% DMSO/buffer solu-tions. Cells were incubated with the corresponding compound atdifferent concentrations up to 100 mM for 24 h. The viability of thecells was evaluated by the MTT assay and the results are presentedgraphically in Fig. 9. Complexes 2, 5 and 6 didn’t show significanttoxicity to RAW264.7 cells at concentrations up to 100 mM. Complex1 was toxic to RAW264.7 cells at concentrations above 50 mM,reducing cell survival by approximately 60% at a concentration of100 mM.

Regardless of any other information these tests exclude 1 fromany biological use. There is no obvious chemical reason that we can

Fig. 9. Cytotoxicity of complexes 1 (-), 2 (C), 5 (A) and 6 (✴) to RAW264.7 cells.Complexes were tested at 10, 50 or 100 mM concentration. The cytotoxicity of thecomplexes was evaluated by MTT assay after 24 h incubation with the cells.

identify from all the preceding tests that hints for a justification ofthis toxicity profile for 1. However, the reassuring thing is that allother compounds proved essentially non-toxic which means thatamine ligands and Mo are a likely association for the search ofpharmaceutically acceptable, non-toxic CORMs. It is also worthnoting that these standard tests cytotoxicity are measured after24 h incubation. Since at this time it can be considered that allcompounds are already decomposed, one may conclude that thedecomposition products are also non-toxic.

2.5. Haemolysis

Administration of haemolytic compounds in vivo has severedeleterious consequences [47]. Therefore, in the development ofsafe CORMs it is important to guarantee that they do not causehaemolysis of red blood cells (RBC). The factors that lead to RBChaemolysis by CORMs have never been studied in depth and veryfew have been reported but it is known that many metal speciesand chemicals are haemolytic. Following a protocol that we havedetailed elsewhere we measured the haemolytic index (HI) ofseveral complexes [15]. A haemolytic index above 10% indicateshaemolysis. Considering the very rapid interaction of 1 and 2 withblood (Fig. 8, top), we decided to compare their haemolytic profilewith that of the more stable complexes 5 and 6. The results aredepicted in Fig. 10. The variation of HI with concentration shows aclear influence of the inner coordination sphere on the HI. In fact, 1is haemolytic above 0.125 mg/mL. Replacing morpholine by Me-piperazine, approximately doubles this value for complex 2. Thevalues obtained for 5 and 6, which are haemolytic above ca. 0.5 mg/mL, further confirm the effect of the coordination sphere on HI. Asfar as we are concerned, the limited existing data does not allow atthe moment an explanation for the observed differences. Thehaemolytic effect of zero-valent Mo carbonyls is not solely depen-dent on the rate of decomposition of the complexes because[Mo(CO)3(his)]Na (ALF186), also an air-sensitive, zero-valent Mocomplex that releases all its CO to blood within the time of mixing,is not haemolytic at a concentration of 1 mg/mL [18]. The sparinglysoluble, air-sensitive [Mo(CO)5Br][NEt4] (ALF062) is haemolyticalso at 0.5 mg/mL as 5 or 6, but the water soluble, air-stable[Mo(CO)5(CN)]Na is not haemolytic at 1 mg/mL [48]. Interest-ingly, the complexes found to be non-haemolytic at 1 mg/mL arenot toxic to RAW264.7 macrophages below 100 mM. This suggests adirect correlation between cytotoxicity and haemolysis but thisremains to be proven. In any case, the haemolytic profile of the

Fig. 10. Variation of the haemolytic index with concentration of complexes: of com-plexes 1 (-), 2 (C), 5 (A) and 6 (✴).

L. Kromer et al. / Journal of Organometallic Chemistry 760 (2014) 89e100 97

complexes in this family still seems too high for therapeutic ap-plications. We believe this is the first report on the variation of HIalong a series of related metal carbonyl complexes.

3. Concluding remarks

This work describes one strategy to improve the physical andchemical properties (stability, solubility), safety and bloodcompatibility of metal carbonyl complexes in order to improvetheir pharmacological profile as potential CO Releasing Molecules.To this end, we performed a number of systematic variations in theancillary ligand substituents of saturated amine complexes[Mo(CO)4(NR3)2], previously untested in aerobic aqueous condi-tions, and generally recognized chemically unstable substances. Itwas found that stability to water, air and blood, requires bidentate,tertiary amine ligands. Primary, and secondary amine ligands areunacceptable. The increase in hydrophobicity generated by thetertiary amines has to be counteracted by the presence of chargedgroups appended to the amine alkyl substituents. Accordingly, thecomplexes with morpholine ligands proved inferior to their Me-piperazine congeners, which can be protonated to improve solu-bility in water. At the end of this limited exercise, the qualitativeorder of increasing “drug-likeliness”, as defined elsewhere for aCORM [9], is 1 < 2 < 3 < 4 < 5 < 6. The first five complexes pre-sented are clearly inadequate CORM candidates due to lack of sta-bility, too high haemolytic index or too fast CO loss in systemiccirculation. Complex 6, with a half-life in blood and a haemolyticindex close to becoming acceptable and 7, which needs to be madesoluble, suggest architectures for pharmacologically compliantCORMs. Further confidence on this direction comes from realizingthat, with the exception of the least stable complex 1, all the othertested complexes show negligible cytotoxicity at 100 mM concen-trations, and that the haemolytic index decreases with increasingstability and solubility of the complexes.

A final note concerns the importance of performing a compre-hensive set of different stability tests before labelling metalcarbonyl complexes as useful CORM candidates for in vivo appli-cations. Air and water stability, ascertained by HPLC or spectro-scopic means are crucial requirements to enable reliable cellculture and other in vitro studies. Stability to blood ensures that thecomplex survives systemic circulation keeping its CO load intact. Asatisfactory sequential response to these tests means that the coreof the scaffold is chemically fit to enter biological testing. Whenapplicable, further tuning of the drug sphere will equip it with thesafety, ADME and tissue specificity characteristics that a real CORMprodrug should possess. We believe that the systematic and com-bined use of all these tests will accelerate the discovery of appro-priate metal carbonyl scaffolds that may be further transformedinto disease specific CORMs.

4. Experimental

4.1. General remarks

Microanalysis (CHN) and ESI-MS data were obtained by Con-ceição Almeida at the Analytical Services Unit from Instituto deTecnologia Química e Biológica. FT-IR spectra were measured on aMattson 7000 FTIR spectrometer at room temperature, using KBrpellets. NMR spectra were recorded on a Bruker Avance II 400 MHzspectrometer. Chemicals shifts are quoted in parts per million.Diglyme, THF, n-hexane, Mo(CO)6, morpholine, 1-methylpiperazine, ethylenediamine, 4-(2-aminoethyl)morpholine,1-(2-aminoethyl)piperazine, 1-[2-(dimethylamino)ethyl]pipera-zine and N,N,N0,N0-tetramethylethylenediamine were obtained

from commercial sources and used as received. Complexes 1, 2 [24],3 [25] and 7 [32] were compared to literature data.

4.2. General synthetic procedure for [Mo(CO)4(amine)2] complexes

A glass vessel was loaded with diglyme (15 mL) or a mixture ofdiglyme/THF (5:1), Mo(CO)6 (10 mmol) and the respective amine(50 mmol). The mixture was heated to reflux in an open modifiedcommercial microwave oven (700W) for 2e16min. After cooling toroom temperature and storing the reaction mixture at �30 �Covernight, or by adding n-hexane (5e10 mL), yellow crystallinesolids precipitated. They were isolated by filtration and dried invacuum.

[Mo(CO)4(morpholine)2] (1): Reaction time: 6 min. Yield: 32%.Anal. Calc. for C12H18MoN2O6: C, 37.71; H, 4.75; N, 7.33. Found: C,37.62; H, 4.80; N, 7.24. Selected IR (KBr, cm�1): 3260 m (NH), 2011 s(CO), 1877 vs (CO), 1840 vs (CO), 1778 vs (CO). ESI-MS (MeOH) [m/z,98Mo]: 385 [M þ H]þ, 88 [L þ H]þ.

[Mo(CO)4(1-methylpiperazine)2] (2): Reaction time: 16 min.Yield 52%. Anal. Calc. for C14H24MoN4O4: C, 41.18; H, 5.92; N, 13.72.Found: C, 41.10; H, 5.41; N, 13.78. Selected IR (KBr, cm�1): 3259 m,3246 m (NH), 2013 s (CO), 1870 vs (CO), 1852 vs (CO), 1775 vs (CO).ESI-MS (MeOH) [m/z, 98Mo]: 411 [M þ H]þ, 383 [M � CO þ H]þ, 101[L þ H]þ.

[Mo(CO)4(ethylenediamine)] (3): Reaction time: 2 min. Yield78%. Anal. Calc. for C6H8MoN2O4: C, 26.88; H, 3.01; N, 10.45. Found:C, 26.20; H, 3.07; N, 10.48. Selected IR (KBr, cm�1): 3368 m, 3310 m(NH), 2021 s (CO), 1883 vs (CO), 1836 vs (CO), 1779 vs (CO). ESI-MS(MeOH) [m/z, 98Mo]: no ionisation observed.

[Mo(CO)4(4-(2-aminoethyl)morpholine)] (4): Reaction time:6 min. Yield 82%. Anal. Calc. for C10H14MoN2O5: C, 35.52; H, 4.17; N,8.28; Found: C, 35.00; H, 3.87; N, 8.30. Selected IR (KBr, cm�1): 3356m, 3297 m (NH), 2016 s (CO), 1888 vs (CO), 1866 vs (CO), 1855 vs(CO), 1793 vs (CO). ESI-MS (MeOH) [m/z, 98Mo]: 131 [L þ H]þ. 1HNMR (400MHz, 25 �C, (CD3)2CO): d 4.24 (t, 2H, morph), 3.88 (dt, 2H,morph), 3.76 (br, 2H, NH2), 3.10 (d, 4H, morph þ CH2NH2), 2.75 (t,2H, NCH2CH2), 2.56e250 (m, 2H, morph). 13C NMR (100MHz, 25 �C,(CD3)2CO): d 220 (CO), 210 (CO), 207 (CO), 66 (Cmorph), 63 (CH2N),62 (Cmorph), 40 (H2NCH2).

[Mo(CO)4(1-(2-aminoethyl)piperazine)] (5): Reaction time:6min. Yield: 77%. Anal. Calc. for C10H15MoN3O4: C, 35.62; H, 4.48; N,12.46. Found: C, 35.50; H, 5.22; N, 12.91. Selected IR (KBr, cm�1):3337 m (NH), 2009 s (CO), 1864 vs (CO), 1843 vs (CO), 1807 vs (CO),1785 vs (CO). ESI-MS (MeOH) [m/z, 98Mo]: 340 [M þ H]þ, 312[M � CO þ H]þ, 284 [M � 2CO þ H]þ, 130 [L þ H]þ. 1H NMR(400 MHz, 25 �C, (CD3)2CO): d 3.80 (br, 2H, H2N), 3.42 (t, 2H, pip),3.14e3.06 (m, 4H, pip þ CH2NH2), 2.96 (d, 2H, NCH2CH2), 2.74 (br,1H, NH), 2.67 (t, 2H, pip), 2.38 (t, 2H, pip). 13C NMR (100MHz, 25 �C,(CD3)2CO): d 223 (CO), 220 (CO), 207 (CO), 65, 64, 63, 45, 40.

[Mo(CO)4((1-(2-dimethylamino)ethyl)piperazine)] (6): Reactiontime: 6 min. Yield: 70%. Anal. Calc. for C12H19MoN3O4: C, 39.46; H,5.24; N, 11.50. Found: C, 39.40; H, 5.44; N, 11.57. Selected IR (KBr,cm�1): 3362 m (NH), 2009 s (CO), 1889 vs (CO), 1856 vs (CO), 1800vs (CO). ESI-MS (MeOH) [m/z, 98Mo]: 368 [M þ H]þ, 340[M � CO þ H]þ, 312 [M � 2CO þ H]þ, 158 [L þ H]þ. 1H NMR(400 MHz, 25 �C, (CD3)2CO): d 3.34 (t, 2H, pip), 3.25 (d, 2H, pip),2.99e2.89 (m, 6H, pip þ CH2NH2 þ NCH2CH2), 2.83 (s, 6H, CH3),2.50e2.44 (m, 2H, pip), 2.21 (s, 1H, NH). 13C NMR (100 MHz, 25 �C,(CD3)2CO): d 222 (CO), 207.0 (CO), 64, 62, 59, 58, 56, 47, 45.

[Mo(CO)4(N,N,N0,N0-Tetramethylethylenediamine)] (7): Reactiontime: 6 min. Yield: 30%. Anal. Calc. for C10H16MoN2O4: C, 37.05; H,4.97; N, 8.64. Found: C, 37.40; H, 4.78; N, 8.51. Selected IR (KBr,cm�1): 2009 s (CO), 1858 vs (CO), 1806 vs (CO); ESI-MS (MeOH) [m/z, 98Mo]: 117 [L þ H]þ.

L. Kromer et al. / Journal of Organometallic Chemistry 760 (2014) 89e10098

4.3. X-ray data collection and refinement statistics

X-ray data collection was performed at 110 K with mono-chromated Mo Ka radiation (l ¼ 0.71073 �A), using a CCD BrukerSAMRT APEX II diffractometer with an Oxford cryosystem. All datasets were processed and scaled using the SAINT and SADABS(Bruker (2005), APEX2, SAINT and SADABS, Bruker AXS Inc., Mad-ison, Wisconsin, USA) programs, the structure was solved by directmethods and refinement proceeded using the program SHELX [49].Hydrogen atoms were included in calculated positions and all non-hydrogen atoms were refined with anisotropic thermal displace-ment parameters. Data and refinement statistics are listed inTable 4.

4.4. Stability studies by UVeVis spectroscopy

The spectra were recorded at 25 �C in a UV-1800 Shimadzu UVspectrophotometer. Quarts 1 cm path length cuvettes were used.Freshly solutions of each complex were prepared in acetonitrile,distilledwater (pH 6.2), aqueous basicmedia (pH 10.2) and aqueousacidic media (pH 2.0): a mixture of water with 0.1% of TFA, meth-anol with water with 0.1% of TFA and water with HCl solution.

4.5. Stability studies by HPLC

A sample of the compound (1 mg) was dissolved in 1 mL of theappropriate solvent: 1 (1% acetone/water); 2 (MeOH); 3, 4, 5, 6, 7(10%MeOH/H2O þ 0.1% TFA) and immediately injected (10 mL) intothe Varian LC940 HPLC system operating at room temperature. Theseparation was performed with a reversed phase column OmniS-pher from Varian C18RP (5 mm particle size; 250 � 4.6 mm) at theflow rate 1 mL/min under the linear gradient program (MeOHeH2O) given in the Supplementary information (SI; Table S7). PDAdetector operated at 236 and 261 nm wavelengths.

4.6. Myoglobin assay

10 mL of a freshly prepared 5 mM stock solution of eachcomplex in DMSO was added to 990 mL of the horse skeletalmyoglobin (Mb) in PBS buffer 7.4 previously reduced to deoxy-

Table 4Data collection, and Refinement statistics.

Compound 4 Compound 5 Compound 6

Empirical formula C10N2O5H14Mo1 C10N3O4H15Mo1 C12N3O4H19Mo1Fw 338.20 337.20 365.26Crystal system Monoclinic Orthorhombic MonoclinicSpace group P21/c P212121 P21/nCell dim (�A, �) a ¼ 11.9960(9)

b ¼ 12.2420(9)c ¼ 15.1910(11)b ¼ 109.840(3)

a ¼ 7.5109(3)b ¼ 13.1151(5)c ¼ 13.6361(5)

a ¼ 9.3140(14)b ¼ 11.5760(10)c ¼ 13.9010(14)b ¼ 95.7587(34)

V (�A3) 2099.7(3) 1343.24(9) 1491.2(3)Z 7 4 4rcalc (g cm�3) 1.450 1.593 1.568m (mm�1) 0.665 0.984 0.893F(000) 920 620 692Rflx collected 27269 16777 28023Independent

reflections [I > 2s(I)]4279 4094 36933330 3906 3379

No. of variables 301 175 205GooF 0.920 0.965 1.027Rint 0.0624 0.0265 0.0367Final R

indices[F2 > 2s(F2)]aR1 ¼ 0.0330wR2 ¼ 0.0813

R1 ¼ 0.0250wR2 ¼ 0.0718

R1 ¼ 0.0342wR2 ¼ 0.0918

R indices (all data)a R1 ¼ 0.0513wR2 ¼ 0.1023

R1 ¼ 0.0272wR2 ¼ 0.0736

R1 ¼ 0.0377wR2 ¼ 0.0949

a R1 ¼ PjjFoj � j:Fcj:j=P jFoj; wR2 ¼ fP½wðF2o � F2c Þ2�=

P½wðF2o Þ2�g1=2:

Mb with 0.1% of sodium dithionite. The concentration of Mbwas established between 50 and 60 mM being determined inevery experiment from the absorbance of the deoxy-Mb solutionat 555 nm (ε ¼ 9.21 mM�1 cm�1), and the final concentration ofcomplex was 50 mM. Control spectra were acquired in order toconfirm that Mb had been properly reduced with sodiumdithionite. Two controls were done in duplicate, the negativecontrol (0% CO-Mb), a deoxy-Mb solution and the positive control(100% CO-Mb), obtained by bubbling pure CO gas into the deoxy-Mb solution for 5 min. The changes in the deoxy-Mb spectrawere recorded between 500 and 600 nm over time at 25 �C. Theexperimental spectra were fitted as a weighted sum of the deoxy-Mb and the CO-Mb spectra. Solver function in MS Excel� wasused to calculate the percentage of CO-Mb in each time point bydeconvolution of the obtained spectra [50]. The absorbancespectrum was converted into a percentage of CO-Mb and basedon the initial amount of complex in solution the amount of COliberated was calculated as equivalents of CO.

4.7. CO release to headspace by GC-RCP detection

Stock solutions of compounds 1e6 (1 mg/1 mL) were preparedin the first two of the following media (1e4) and in all three media(5 and 6): MeOH/H2O(0.1% TFA) 1:9, MeOH/H2O 1:9 and H2O with2 equiv of HCl. The solutions were diluted in the respective me-dium obtaining final solutions of 2 mL with concentrations of25 mM for each compound. The final solutions were prepared andimmediately closed in a 7.0 mL Roth� sample vial equipped with amagnetic stirrer inside and capped with a Silicone/PTFE septa andan aluminium cap. Silicone/PTFE and aluminium cap were ac-quired from VWR�. The solutions were stirred at room temper-ature and in air. Samples of 200 mL were taken from the headspaceover time (0, 30, 60, and 90 min), with a Gastight Hamilton� sy-ringe, and diluted in another 7.0 mL Roth� vial, capped with aSilicone/PTFE septa and an aluminium cap, and analyzed quanti-tatively on a Peak Performer 1 RCP gas chromatograph (GC), whichallows the CO in gas to be quantified to concentrations as low as to1e2 ppb. The reducing compound photometer (RCP) bed and thecolumn were set at constant temperatures, 265 and 105 �C,respectively.

The amount of CO was calculated using a calibration curvepreviously obtained in vials of the same volume (7.0 mL), using aLinde minican� of CO (30 ppm CO rest in synth. air) ref14960013.

4.8. CO release in blood

A solution of each complex (ca. 5e7 mM) was prepared in sa-line/HCl (for 5 and 6) or DMSO (for the others) and an aliquot of50 mL was added to 1 mL of sheep whole blood in Alsever’s solution(Innovative Research cat no. IR1-020N) and incubated at 37 �C.Samples were kept inside closed plastic culture tubes with closures(5 mL; 12 � 75 mm) and analyzed over time in the oximeter(Avoximeter 4000 from Avox Instruments Inc.; disposable cuvettesfor Avoximeter 4000 from Avox Instruments Inc.) to follow theincrease in COHb levels.

The amount of CO liberated was calculated based on the amountof compound initially added, total amount of hemoglobin and %COHb (both given by the oximeter). The compounds were tested ina concentration calculated in order to mimic relevant in vivo doses.Assuming a 20 g mouse with 8% blood volume (1.6 mL), 10 mg/kgcorrespond to 0.2 mg of compound per animal, therefore 0.2 mg/1.6 mL of blood or 0.125 mg/mL of blood. Control spectra of the freemedium with the corresponding vehicles were always recorded.

L. Kromer et al. / Journal of Organometallic Chemistry 760 (2014) 89e100 99

4.9. Cell culture experiments: cytotoxicity

Murine RAW264.7 monocyte macrophages were obtainedfrom the European Collection of Cell Cultures (ECACC 91062702;Salisbury, Wiltshire, U.K.) and cultured in Dulbecco’s modifiedEagle’s medium (cat. 41966029, Gibco, Invitrogen) supplementedwith 10% fetal bovine serum (cat. 10500064, Gibco, Invitrogen).Cultures were maintained at 37 �C in a 5% CO2 humidifiedatmosphere.

The RAW264.7 macrophages were incubated in the presence ofcomplexes 1, 2, 5 and 6 (at a concentration of 10, 50 or 100 mM) for24 h, at 37 �C in a 5% CO2 humidified atmosphere. Culture mediumwas replaced by fresh culture medium supplemented with 1 mg/mL of MTT (3-(4,5-Dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetra-zolium bromide; A2231, Applichem). Cells were incubated in thesame conditions for 1 h. The culture medium was discarded andthe formazan crystals were solubilized in dimethyl sulfoxide(DMSO). The absorbance of the solutions was measured at 550 nmin a BioRad Microplate Reader. The percentage of survival wascalculated considering the growth of the cells in buffer alone as100%.

4.10. Haemolysis determination

Red blood cells (RBC) obtained upon centrifugation of sheepwhole blood (in Alsever’s solution; Innovative Research Cat no. IR1-020N) were used to evaluate the potential of the [Mo(CO)4(NR3)2]complexes to induce RBC hemolysis.

A 2% RBC suspension in PBS (100 mL) was distributed in thewells of a 96-well plate. The effect of the complexes was evaluatedin concentrations between 0.0078 and 1 mg/mL. A 2 mg/mL solution of the complex was prepared followed by 12 serialdilutions in PBS. These solutions were added (100 mL) to the RBCsuspension. The complex-RBC suspension was then incubated for1 h at 37 �C. A 2% RBC solution in water was used as a positivecontrol (RBC lysis).

The plate was then centrifuged and the absorbance of thesupernatant was measured at 550 nm in a microplate reader(Bio-Rad). The haemolytic index (HI) was determined using theformula 1,

HIð%Þ ¼ODðcomplex sampleÞ�ODðcomplex referenceÞODðpositive controlÞ�ODðnegative controlÞ �100

Formula 1: determination of HI, where: OD (complexreference) ¼ OD of the corresponding complex solution (endoge-nous abs). OD (positive control) ¼ OD of the solution obtained bylysis of the RBC (1% RBC). OD (negative control) ¼ OD of the 1% RBCsuspension in PBS after centrifugation.

A haemolytic index above 10% indicates hemolysis.

Acknowledgements

The Swiss National Science Foundation (LK, Grant No. PBZHP2-125507) and Alfama Lda. are gratefully acknowledged for financialsupport. ACC thanks the Fundação para a Ciência e Tecnologia (FCT)for the grant SFRH/BPD/70163/2010. The authors thank Jessica M.Carvalho for valuable assistance in experimental work. The NMRspectrometers are part of the National NMR Network and werepurchased in the framework of the National Program for ScientificRe-equipment, Contract REDE/1517/RMN/2005, with funds fromPOCI 2010 (FEDER) and FCT. The work was partially financed byAlfama Lda, through FP7-BSG-SME, Grant Agreement #230629/CORM-RA and FCT project PTDC/QUI-BIQ/117799/2010.

Appendix A. Supplementary material

Compound 4, CCDC 979190; Compound 5, CCDC 979191; Com-pound 6, CCDC 979189 contain the supplementary crystallographicdata for this paper. These data can be obtained free of charge fromThe Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.

Appendix B. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.jorganchem.2013.12.009.

References

[1] C. Szabó, Sci. Transl. Med. 2 (2010), 59ps54.[2] L.E. Otterbein, Respir. Care 54 (2009) 925e932.[3] L.E. Otterbein, A. May, B.Y. Chin, Cell. Mol. Biol. 51 (2005) 433e440.[4] L.E. Otterbein, F.H. Bach, J. Alam, M. Soares, H.T. Lu, M. Wysk, R.J. Davis,

R.A. Flavell, A.M.K. Choi, Nat. Med. 6 (2000) 422e428.[5] I. Bauer, B.H.J. Pannen, Crit. Care 13 (2009) 220.[6] R. Motterlini, L.E. Otterbein, Nat. Rev. Drug Discov. 9 (2010) 728e743.[7] B.E. Mann, R. Motterlini, Chem. Commun. (2007) 4197e4208.[8] http://clinicaltrials.gov/ct2/results?term¼carbonþmonoxide.[9] C.C. Romão, W.A. Blättler, J.D. Seixas, G.J.L. Bernardes, Chem. Soc. Rev. 41

(2012) 3571e3583.[10] R. Motterlini, J. Clark, R. Foresti, P. Sarathchandra, B. Mann, C. Green, Circ. Res.

90 (2002) E17eE24.[11] J. Clark, P. Naughton, S. Shurey, C. Green, T. Johnson, B. Mann, R. Foresti,

R. Motterlini, Circ. Res. 93 (2003) E2eE8.[12] T.R. Johnson, B.E. Mann, I.P. Teasdale, H. Adams, R. Foresti, C.J. Green,

R. Motterlini, Dalton Trans. (2007) 1500e1508.[13] B.E. Mann, Top. Organomet. Chem. 32 (2010) 247e285.[14] B.E. Mann, Organometallics 31 (2012) 5728e5735.[15] J.D. Seixas, A. Mukhopadhyay, T. Santos-Silva, L.E. Otterbein, D.J. Gallo,

S.S. Rodrigues, B.H. Guerreiro, A.M.L. Gonçalves, N. Penacho, A.R. Marques,A.C. Coelho, P.M. Reis, M.J. Romão, C.C. Romão, Dalton Trans. (2012), http://dx.doi.org/10.1039/c2dt32174b.

[16] S.S. Rodrigues, J.D. Seixas, N.M.P. Pereira, C.C. Romão, W.E. Haas, I.M.S. Gon-çalves, 2011, US 2011/0038955 A1.

[17] W.E. Haas, C.C. Romão, S.S. Rodrigues, J.D. Seixas, A.R.M. Pina, 2007, US2007/0207993 A1.

[18] A.R. Marques, L. Kromer, D.J. Gallo, N. Penacho, S.S. Rodrigues, J.D. Seixas,G.J.L. Bernardes, P.M. Reis, S.L. Otterbein, R.A. Ruggieri, A.S.G. Gonçalves,A.M.L. Gonçalves, M.N.D. Matos, I. Bento, L.E. Otterbein, W.A. Blattler,C.C. Romão, Organometallics 31 (2012) 5810e5822.

[19] F. Zobi, O. Blacque, R.A. Jacobs, M.C. Schaub, A.Y. Bogdanova, Dalton Trans. 41(2012) 370e378.

[20] S. Romanski, B. Kraus, U. Schatzschneider, J.-M. Neudörfl, S. Amslinger, H.-G. Schmalz, Angew. Chem. Int. Ed. 50 (2011) 2392e2396.

[21] S. Romanski, B. Kraus, M. Guttentag, W. Schlundt, H. Rücker, A. Adler, J.-M. Neudörfl, R. Alberto, S. Amslinger, H.-G. Schmalz, Dalton Trans. 41 (2012)13862e13875.

[22] S. Romanski, H. Rücker, E. Stamellou, M. Guttentag, J.-M. Neudörfl, R. Alberto,S. Amslinger, B. Yard, H.-G. Schmalz, Organometallics 31 (2012) 5800e5809.

[23] R. Alberto, R. Schibli, R. Waibel, U. Abram, A. Schubiger, Coord. Chem. Rev. 192(1999) 901e919.

[24] G.W.A. Fowles, D.K. Jenkins, Inorg. Chem. 3 (1964) 257e259.[25] C.S. Kraihanzel, F.A. Cotton, Inorg. Chem. 2 (2001) 533e540.[26] K.B. Shiu, S.L. Wang, F.L. Liao, J. Organomet. Chem. 420 (1991) 207e215.[27] W.-Q. Zhang, A.C. Whitwood, I.J.S. Fairlamb, J.M. Lynam, Inorg. Chem. 49

(2010) 8941e8952.[28] M. Ardon, P.D. Hayes, G. Hogarth, J. Chem. Educ. 79 (2002) 1249e1251.[29] M. Ardon, G. Hogarth, D. Oscroft, J. Organomet. Chem. 689 (2004) 2429e2435.[30] S.L. VanAtta, B.A. Duclos, D.B. Green, Organometallics 19 (2000) 2397e2399.[31] D.R. Baghurst, D.M.P. Mingos, J. Chem. Soc. Chem. Commun. (1992) 674e677.[32] M.A. Bakar, H.K. Fun, K. Chinnakali, S.G. Teoh, O.B. Shawkataly, F.M. Lopez,

Acta Crystallogr. C49 (1993) 582e584.[33] J. Grundy, M.P. Coles, P.B. Hitchcock, Dalton Trans. (2003) 2573e2577.[34] C. Fowelin, A. Matyja, M. Schmidt, J. Heck, Z. Anorg. Allg. Chem. 633 (2007)

2395e2399.[35] I.R. Laskar, T.K. Maji, D. Das, T.H. Lu, W.T. Wong, K. Okamoto, N.R. Chaudhuri,

Polyhedron 20 (2001) 2073e2082.[36] T. Chattopadhyay, M. Ghosh, A. Majee, M. Nethaji, D. Das, Polyhedron 24

(2005) 1677e1681.[37] D. Das, N.P. Nayak, A.K. Mukherjee, N.R. Chaudhuri, Polyhedron 16 (1997)

3305e3309.[38] D. Darensbourg, E. Atnip, K. Klausmeyer, J. Reibenspies, Inorg. Chem. 33

(1994) 5230e5237.[39] D. Darensbourg, J. Draper, J. Reibenspies, Inorg. Chem. 36 (1997) 3648e3656.

L. Kromer et al. / Journal of Organometallic Chemistry 760 (2014) 89e100100

[40] W.-Q. Zhang, A.J. Atkin, R.J. Thatcher, A.C. Whitwood, I.J.S. Fairlamb,J.M. Lynam, Dalton Trans. (2009) 4351e4358.

[41] A.J. Atkin, I.J.S. Fairlamb, J.S. Ward, J.M. Lynam, Organometallics 31 (2012)5894e5902.

[42] A.J. Atkin, J.M. Lynam, B.E. Moulton, P. Sawle, R. Motterlini, N.M. Boyle,M.T. Pryce, I.J.S. Fairlamb, Dalton Trans. 40 (2011) 5755e5761.

[43] S. McLean, B.E. Mann, R.K. Poole, Anal. Biochem. (2012) 1e21.[44] R. Motterlini, B.E. Mann, T.R. Johnson, J.E. Clark, R. Foresti, C.J. Green, Curr.

Pharm. Des. 9 (2003) 2525e2539.[45] T. Santos-Silva, A. Mukhopadhyay, J.D. Seixas, G.J.L. Bernardes, C.C. Romão,

M.J. Romão, J. Am. Chem. Soc. 133 (2011) 1192e1195.

[46] M.F.A. Santos, J.D. Seixas, A.C. Coelho, A. Mukhopadhyay, P.M. Reis,M.J. Romão, C.C. Romão, T. Santos-Silva, J. Inorg. Biochem. 117 (2012) 285e291.

[47] R.P. Rother, L. Bell, P. Hillmen, M.T. Gladwin, J. Am. Med. Assoc. 293 (2005)1653e1662.

[48] Alfama Lda, unpublished work.[49] G.M. Sheldrick, Acta Crystallogr. Sect. A Found. Crystallogr. 64 (2008) 112e

122.[50] D.C. Harris, J. Chem. Educ. 75 (1998) 119e121.