Further insights into the mechanism of the reactionof activated bleomycin with DNAMarina S. Chow, Lei V. Liu, and Edward I. Solomon†

Department of Chemistry, Stanford University, Stanford, CA 94305-5080

Contributed by Edward I. Solomon, July 2, 2008 (sent for review April 6, 2008)

Bleomycin (BLM) is a glycopeptide anticancer drug that effectivelycarries out single- and double-stranded DNA cleavage. Activated BLM(ABLM), a low-spin ferric-hydroperoxide, BLM–FeIII–OOH, is the lastintermediate detected before DNA cleavage. We have previouslyshown through experiments and DFT calculations that both ABLMdecay and reaction with H atom donors proceed via direct H atomabstraction. However, the rate of ABLM decay had been previouslyfound, based on indirect methods, to be independent of the presenceof DNA. In this study, we use a circular dichroism (CD) feature uniqueto ABLM to directly monitor the kinetics of ABLM reaction with a DNAoligonucleotide. Our results show that the ABLM � DNA reaction isappreciably faster, has a different kinetic isotope effect, and has alower Arrhenius activation energy than does ABLM decay. In theABLM reaction with DNA, the small normal kH/kD ratio is attributed toa secondary solvent effect through DFT vibrational analysis of reac-tant and transition state (TS) frequencies, and the lower Ea is attrib-uted to the weaker bond involved in the abstraction reaction (C–H forDNA and N–H for the decay in the absence of DNA). The DNAdependence of the ABLM reaction indicates that DNA is involved inthe TS for ABLM decay and thus reacts directly with BLM–FeIII–OOHinstead of its decay product.

non-heme iron � kinetics � isotope effects � reactivity

Bleomycins (BLMs) (Fig. 1) are a group of glycopeptide antibi-otics that are used clinically to treat Hodgkin’s lymphoma, head

and neck tumors, and testicular cancer (1–4). Activated BLM(ABLM), a low-spin ferric hydroperoxide complex, BLM–FeIII–OOH, (5–7) is the last intermediate detected before DNA strandcleavage (8, 9). ABLM can be formed from the reaction ofFeIIBLM with O2 and one additional electron (8, 10). Throughdeuterium and tritium labeling studies, it was established that DNAcleavage involves H atom abstraction from the C4� of the deoxyri-bose sugar (11, 12). O18 isotope studies indicated that O–O bondcleavage is in the rate-determining step (13). We have recentlyprovided strong experimental and computational evidence thatABLM decay and reactions with H atom donors proceed via Hatom abstraction by the low-spin ferric hydroperoxide (14). How-ever, an alternative mechanism that parallels heme chemistry wasrecently reemphasized, where BLM–FeIII–OOH is protonated anddecays to produce the H atom abstracting species, BLM–FeVAO,via heterolytic cleavage of the O–O bond (15). This heterolyticcleavage mechanism would not show a dependence on H atomavailability and would appear to be ruled out by the fact that thedecay of ABLM is accelerated by H atom donors. However, the rateof ABLM reaction with DNA as the substrate had been experi-mentally observed to be very similar to that in the absence of DNA(8). In those experiments, the rate of the ABLM reaction wasstudied by using primarily the indirect method of quantitativeanalysis of DNA product. We have defined a method to directlymonitor the decay of ABLM in real time by using a circulardichroism (CD) feature specific to ABLM at 450 nm (Fig. 2 Inset)(14). In this study, we apply this method to evaluate the kinetics ofthe reaction of ABLM with a 10-bp DNA oligonucleotide.

Results and DiscussionA past study showed only small differences in the rates of ABLMdecay and its decay when bound to DNA (t1/2 � 120 and 105 s,

respectively), (8) leading to the proposal that decay of ABLMproduces the species involved in H atom abstraction from DNA. Incontrast, by using the CD feature at 450 nm to monitor the time courseof the reaction, we find that the ABLM reaction with DNA in factproceeded appreciably faster (2.5 times) in the presence of DNA (Fig.2 and Table 1). Additionally, our experiments show that theABLM � DNA reaction exhibits a relatively small, normal kineticisotope effect (KIE) of 1.7 � 0.2 compared to the KIE 3.6 � 0.9observed for ABLM decay (14). Furthermore, temperature-dependentkinetic experiments give an activation energy (Ea) of 4.7 � 0.9 kcal/molfor the ABLM � DNA reaction compared with 9.3 � 0.9 kcal/mol forthe ABLM decay (14). Therefore, our results show that DNA isinvolved in the rate-determining step of ABLM decay.

A series of control experiments were conducted to ensure thatthe reaction studied kinetically did indeed involve ABLM bound toDNA. First, complete binding of FeIIBLM to DNA oligonucleotidewas confirmed by near-IR magnetic CD (MCD), which showsdifferent spectra in the presence and absence of DNA (Fig. 3A).Second, the formation of ABLM upon addition of oxygenatedbuffer to the FeIIBLM–DNA complex was confirmed by using EPR(Fig. 3B).§ Third, colorimetric 2-thiobarbituric assay (�532 � 1.58 �105 mM�1�cm�1) of the product mixture was positive for basepropenals produced during DNA cleavage (Fig. 3C). The BLM-

Author contributions: M.S.C. and E.I.S. designed research; M.S.C., L.V.L., and E.I.S. performedresearch; M.S.C., L.V.L., and E.I.S. analyzed data; and M.S.C. and E.I.S. wrote the paper.

The authors declare no conflict of interest.

†To whom correspondence should be addressed. E-mail: edward.solomon@stanford.edu.

§Addition of O2 to BLM–FeII produces a FeIII–superoxide complex, which disproportionatesto give BLM–FeIII and BLM–FeIII–OOH.

This article contains supporting information online at www.pnas.org/cgi/content/full/0806378105/DCSupplemental.

© 2008 by The National Academy of Sciences of the USA

N H

OH

CH 3

NHC H3

O

HO

NH CH 3

O

NH

O

O

OH

HO

N N

NH 2

CH 3

H 2N

HO

O

N

N

S

NH2

O O NH 2HN

O

O

O

H O O

NH 2

OH

O H

NH

O NS

HN O

RH

S +CH 3

H 3CR

HN

H3C H

H

H

H

H

HH

H

H

D NA bin ding do main

bleom y cin A 2 (B L M )

c arb ohy dra te d om ain

S+C H3

H 3C NH

NH 2+

b leo m y c in B 2 (B LM )

=

peplom ycin (PEP )

m e ta l b in ding d o ma in

Fig. 1. Structure of BLM ligand. Atoms underlined and in bold type are likelyligands to the metal center.

www.pnas.org�cgi�doi�10.1073�pnas.0806378105 PNAS � September 9, 2008 � vol. 105 � no. 36 � 13241–13245

CHEM

ISTR

Y

based product was also identified by EPR and UV–visible CD aslow-spin FeIIIBLM (Fig. 3 B and D). The fact that the rate, KIE, andactivation energy of the ABLM � DNA reaction are all differentfrom those for ABLM decay demonstrates that DNA is a substratefor BLM–FeIII–OOH rather than for a subsequent decay product(i.e., BLM–FeVAO).

The reaction of ABLM with the oligonucleotide was measuredin H2O and 2H2O buffer to obtain the KIE (kH/kD � 1.7 � 0.2).Because the C�4–H is not labile to solvent-exchange, the isotopeeffect would appear to be attributed to the H/D exchange of theH on the Odis of the hydroperoxide unit, as shown in Fig. 4.

The small magnitude (1.7) is consistent with a secondary KIE,which results when bond cleavage occurs adjacent to the isoto-pically substituted position, due to changes in force constant tothe isotope in going from reactants to the transition state (TS).The magnitude of 2° KIEs associated with the mechanism in Fig.4 can be estimated through vibrational analysis of isotopicallysubstituted bonds by using the Streitweiser approximation (16).

kD

kH� e��

hc2kT �(��‡

D � ��‡H) � �(��D � ��H)�, [1]

where h is Planck’s constant, c is the speed of light, k isBoltzmann’s constant, T is temperature, �� is the TS frequency,and � is the ground state vibration frequency. Vibrationalfrequencies of the reactant [BLM–FeIII–OOH/D, sugar] and theTS [BLM–FeIII–OOH/D � sugar] were obtained from DFTcalculations on a small BLM model, where the pyrimidine,deprotonated amide, and imidazole moieties of the metal-binding domain are retained, and the 1° and 2° amines aremodeled as NH3 (14). The calculations show that, in addition tothe O–H/D stretching mode, the OO–H/D in-plane (ip) andOO–H/D out-of-plane (op) bending modes are also significantlyaffected by H/D isotope substitution (Fig. 5). The frequencies ofbending modes have been shown to contribute significantly tothe normal secondary isotope effect observed for SN2 reactionsin alkyl and ethyl halides (17–20). In our computational model,the O–H stretch and OH ip bend are clean modes that are easilymapped in the H/D-substituted reactant and TS complexes.

0 60 120 180 240 300 360

CD

Inte

nsi

ty (

m°)

Time (s)

1200016000200002400028000

CD

Inte

nsi

ty (

m°)

Energy (cm-1)

450 nm ABLM

FeIIIBLM

CDCD450 nm

Fig. 2. ABLM reaction kinetics studied by CD spectroscopy. The black tracerepresents the decay of ABLM and the red trace represents reaction of ABLMwith DNA (5�-d(GGAAGCTTCC)2-3�). (Inset) Interval scans showing the changein CD spectral features during ABLM decay to FeIIIBLM.

2500 3000 3500 4000

EP

R In

ten

sity

Gauss

34s

104s

404s

6000 8000 10000 12000 14000 16000

∆ε (

mM

-1 c

m-1

)

Energy (cm-1)

-10

0

10

20

1200016000200002400028000

CD

Inte

nsi

ty (

m°)

Energy (cm-1)400 500 600 700 800

Ab

sorb

ance

Wavelength (nm)

453 nm

532 nm

A

C

B

D

2.262.17 1.94

2.172.45

1.89

Fig. 3. Spectral characterization of the ABLM � DNA reaction. (A) Near–IR MCD of FeIIBLM � DNA (filled red trace) vs. FeIIBLM (dotted black trace). Perturbationof the spectrum indicates DNA binding. (B) EPR spectra of FeIIBLM � DNA � O2. ABLM features (g � 2.26, 2.17, and 1.94) grow in over time and eventually decayto FeIIIBLM (g � 2.45, 2.18, and 1.89). (C) Colorimetric assay for base propenals in the ABLM � DNA reaction products. The purple spectrum shows the absorbanceat 532 nm due to the chromophore formed from 2-thiobarbituric acid and base propenals. The gray spectrum is of a FeIII/EDTA/2-thiobarbituric acid assay complex,which accounts for the absorbance observed at 453 nm. (D) CD spectra of the product of ABLM � DNA reaction (filled orange trace) compared to that of FeIII–BLM(dotted black trace).

Table 1. Kinetic Parameters of ABLM decay � DNA

Reaction ABLM � DNA* ABLM decay†

Rate,‡ s�1 0.044 � 0.02 0.018 � 0.003kH/kD

§ 1.7 � 0.2 3.6 � 0.9Ea, kcal/mol¶ 4.7 � 0.9 9.3 � 0.9

*Reaction mixtures contained 0.3 mM FeII, 0.34 mM BLM, 0.42 mMd(GGAAGCTTCC)2 oligonucleotide, and were kept anaerobic until reactionwas initiated by addition of oxygenated buffer.

†Data taken from ref. 14.‡Rates measured at 20°C, 100 mM Hepes buffer, pH 7.4.§From the ratio of reaction rates measured in H2O and D2O buffers.¶Arrhenius activation energy obtained from temperature-dependent exper-iments performed at 5, 10, 20, and 30°C.

13242 � www.pnas.org�cgi�doi�10.1073�pnas.0806378105 Chow et al.

However, the O–H op bend mixes with other vibrational modelsof the BLM and the sugar, and cannot be directly compared.Hence, we calculated the frequencies of the localized O–H opbend by manually distorting the O–Odis–H angle around theoptimized equilibrium point (details in Materials and Methods).The frequencies of these three modes and their calculated 2°KIEs are listed in Table 2.

The frequency of the O–H stretch does not significantlychange in going from reactants to TS and thus only has a small,normal contribution to the overall isotope effect. The fre-quency of the OO–H ip bend, however, decreases substantiallyfrom 1,310 to 640 cm�1 due to the elongation of the O–O bond(from 1.5 to 2.7 Å) in the TS (14), and gives a significant normalsecondary isotope effect of 1.55. The OO–H op bend increasesslightly from 420 cm�1 in reactant to 520 cm�1 in TS and contributesa small inverse isotope effect of 0.93. The product gives a totalsecondary KIE of 1.47 at 293 K, which is in reasonable agreementwith the experimental value (1.7 � 0.2).

It should be noted that in our TS model, the Odis–H of theFe-hydroperoxide is H-bonded to the O3 of the sugar ribose(OdisOO3 � 2.8 Å) (Fig. 6 Right), whereas, in the reactantsmodel where the sugar is optimized at van der Waals distancefrom the ABLM, this H bond is longer and weaker (OdisOO3 �3.45 Å) (Fig. 6 Left). The presence of this H bond influences thefrequencies of the OO–H bends and thus the 2° KIEs [supportinginformation (SI) Table S1]. In our DFT calculations, removal ofthe OdisOH���O3 H bond (by omitting the sugar molecule) fromboth reactants and TS gives a 2° KIE of 1.7; removal from TS onlyresults in 2° KIE � 1.8; and removal from the reactants only gives2° KIE � 1.3. Our computational model likely underestimatesthe strength of the OdisOH���O3 H bond in the reactants—theNMR structure of BLM–CoIII–OOH shows a shorter OdisOO3

bond (2.9 Å) (21)—and, hence, underestimates the 2° KIE(calculated � 1.47 vs. experimental � 1.7 � 0.2).

This DFT vibrational analysis was also applied to theABLM � DNA reaction where the C4� position on the ribosesugar was H/D labeled. In this case, we considered only the C–Hstretch and estimated the primary KIE for C–H cleavage to be

2.61,¶ within the broad range observed for the cleavage of2H-labeled DNA (kH/kD of 2–7) (11). The combined calculated1° and 2° KIE is 3.8, which is consistent with that observed forthe decay of ABLM (Table 1).�

Based on observed KIEs, we had previously proposed that inthe absence of substrate, ABLM abstracts a H atom from anamide N–H on the peptide linker region of the BLM tail. TheArrhenius activation energy (Ea) of this process is 9.3 kcal/mol,which is �5 kcal/mol higher than the Ea for the DNA reaction(Table 1). One apparent difference between these two reactionsis the strength of the X–H bond cleaved during H atom abstrac-tion by ABLM. The C–H bond (BDE � 92 kcal/mol) cleaved inABLM � DNA is weaker than the N–H bond (BDE �105kcal/mol) (22), which would be cleaved in this model for ABLMdecay, leading to a lower reaction energy (�Erxn) for theABLM � DNA reaction. The activation barrier for a reactioncan be expressed in terms of the intrinsic barrier (when thereaction has no driving force) and a contribution arising from theenergy differences between reactants and products (23). Thisrelationship between activation barrier (Ea), intrinsic barrier(�E0

‡), and reaction energy (�Erxn) is given by the Marcusequation (24):

Ea � �E0‡ �

12�Erxn �

�E rxn2

16�E0‡ . [2]

The �E0‡ values for the ABLM � DNA and ABLM decay

reactions were calculated by using the experimentally deter-mined Ea values, and the 13 kcal/mol energy difference of theC–H and N–H bond in the reaction energies in Eq. 2. Theseresults are given in Table 3 for an ABLM � DNA reactionenergy of �7 kcal/mol. The calculated �E0

‡ for ABLM � DNAand ABLM decay is very similar (8 vs. 6 kcal/mol), indicating thatthe difference in the experimentally determined Ea values likelyreflects the difference in reaction energies associated with theBDEs of the C–H vs. N–H bond cleaved. This is the case for valuesof �Erxn below �3 kcal/mol for the ABLM � DNA reaction,although it should be noted that, as �Erxn becomes more exother-mic, the �E0

‡ increases. For the ABLM � DNA reaction, whereBLM–FeIII–OOH � H–C–(DNA)3BLM–FeIVAO � •C–DNA � H2O, our DFT calculations [B3LYP/6–311G* for optimi-zation and B3LYP/6–311�G** and polarized continuum model(PCM) � � 4.0 for a single point calculation] give �Erxn of �2kcal/mol††, suggesting that the ABLM � DNA reaction likely has alow driving force and intrinsic barrier.

¶DFT frequencies for sugar and TS(ABLM � sugar), where C4�–H/D:TS C–H � 1,758 cm�1 andC–D � 1,338 cm�1; difference � �420 cm�1. For the reactant sugar C–H � 3,904 cm�1 andC–D � 2,278 cm�1; difference � �816 cm�1.

�Only the zero point energy term is presented above. The mass–moment of inertia andvibrational excitation terms have values of unity when calculated by using the frequenciesof the OH stretch and OO–H bends.

F e III

OOH/D

B LM

H C 4'-s ug ar

Fe III

O

H /D

B LM

H C 4'-s ug arO

F eIV B LM

O

C 4'-s ug ar

O

H /D

H

T S:

Fig. 4. The proposed mechanism for the ABLM � DNA reaction illustratingthe exchangeable H on Odis of the hydroperoxo moiety of ABLM as thepotential origin of the observed KIE.

Fig. 5. The three O–H vibrational modes in the ABLM small model that aresensitive to H/D substitution. The sugar molecule is omitted for clarity.

Table 2. DFT frequencies (in cm�1) and resulting calculated 2°KIEs for Odis–H normal modes

Mode R* �(D–H)R TS† �(D–H)� kH/kD

O–H stretch‡ 3620 (2640) �980 3570 (2600) �970 1.02OO–H ip bend‡ 1310 (975) �335 640 (485) �155 1.55OO–H op bend§ 420 (310) �110 520 (380) �140 0.93

Total KIE 1.47

*Reactants � (BLM–FeIII–OOH, sugar).†TS � (BLM–FeIII–OOH � sugar).‡Frequencies from DFT calculations.§Frequencies from manual distortion of O–H along normal bending mode.Frequencies of O–D modes are given in parentheses.

Chow et al. PNAS � September 9, 2008 � vol. 105 � no. 36 � 13243

CHEM

ISTR

Y

Whereas the Ea values for the two reactions in Table 3 arequite different, their rates only differ by a factor of 2.5 (Table1). Thus, the �G� are similar (19.5 vs. 19.0 kcal/mol, Table S2)and there is apparently a difference in entropy between the tworeactions (27, 28). Indeed, Eyring plots of the temperature-dependent kinetic data yield significant T�S� terms of �10 and�14 kcal/mol at 293 K for ABLM decay and ABLM � DNAreaction respectively (Fig. S1 and Table S2), ref lecting adifferent decrease in entropy in going from reactant to TS. TheT�S� for the ABLM � DNA reaction is more negative andindicates that the ABLM � DNA TS is more ordered than thatfor ABLM decay. The lower activation barrier observed for theABLM � DNA reaction (4.7 kcal/mol vs. 9.3 kcal/mol forABLM decay) is consistent with an earlier, tighter TS (29–32).

In summary we have shown through differences in rate,isotope effect, activation energies, and T�S, that theABLM � DNA reaction is different from that of ABLM decay.These results strongly support a mechanism in which BLM–FeIII–OOH directly, rather than through a species produced byits decay, reacts with DNA through H atom abstraction anddemonstrates that non-heme Fe can be very different from hemein reactivity.

Materials and MethodsExperimental Details. Preparation of FeIIBLM–DNA. Ferrous BLM was prepared aspreviously described (14, 33–35). Oligonucleotides [10 bp; 5�-d(GGAAGCT-TCC)2-3�] from Operon DNA were dissolved in Hepes buffer (100 mM, pH 7.4)and heated in a 90°C sand bath for 5 min to anneal the single strands. Thecooled DNA solution was then added to FeIIBLM to form the FeIIBLM–DNAcomplex. Stock solutions contained 5 mM FeII, 5.6 mM BLM, and 6–7 mMoligonucleotide.Kinetics. A small amount of anaerobic FeIIBLM–DNA stock solution was injectedinto a Teflon-capped cell and brought to the CD instrument (J810 spectropo-larimeter; Jasco). A larger volume of aerobic buffer was injected into thecuvette to initiate the reaction. The reaction was performed in H2O and 2H2O.Constant reaction temperatures were maintained by using a circulating waterbath. All reactions were run in triplicate.MCD spectroscopy of FeIIBLM–DNA. Deuterated glycerol was added anaerobi-cally to the FeIIBLM–DNA to give a final 60% (vol/vol) mixture, which wasinjected into an MCD cell and quickly frozen in liquid N2. Near-IR MCD data(2,000 – 600 nm) was collected at 5K and 7T on a J200 CD spectropolarimeter(Jasco) fitted with an SM4000 –7T superconducting magnet (Oxford).X-band EPR. Spectra were obtained on a EMS spectrometer (Bruker), ER 041XGmicrowave bridge, and ER 4102ST cavity. Samples were run at 9.4 GHz at 77 K.FeIIBLM � DNA was frozen anaerobically then thawed aerobically and al-lowed to react with atmospheric O2. The samples were refrozen and EPRspectra were collected at consecutive time intervals of 34, 104, and 404 s.DNA product analysis. Base propenals produced during the ABLM � DNA reac-tion were quantified by using the 2-thiobarbituric acid assay according topublished procedures (8).

Computational Details. DFT calculations were performed by using the Gaussian03 package (36). Frequency calculations (B3LYP and Lanl2DZ) were performedon previously optimized small models of reactant [ABLM and sugar] and TS[ABLM � sugar] from ref. 14. To estimate the localized Odis–H ip and opbending frequencies, the H was manually moved in �5° increments along theip or op normal modes, and single point calculations were performed. Theenergies were fit to a parabola to obtain the force constant and frequency.Reaction energies were calculated by B3LYP/6–311G*//B3LYP/6–311�G**scheme. Solvation effects were accounted for by using the PCM with dielectricconstant � � 4.0).

††DFT calculated reaction energy depends on the functional used (25) and can vary by �10kcal/mol (26).

ACKNOWLEDGMENTS. We thank Prof. John I. Brauman, Dr. Stephen R. Lynch,and Dr. Richard M. Burger for helpful discussions. This work was supported bya National Institutes of Health Grant GM40392.

1. Burger RM (1998) Cleavage of nucleic acids by bleomycin. Chem Rev 98:1153–1169.2. Hecht SM (2000) Bleomycin: New perspectives on the mechanism of action. J Nat Prod

63:158–168.3. Petering DH, Mao Q, Li W, DeRose E, Antholine WE (1996) in Probing of Nucleic Acids

by Metal Ion Complexes of Small Molecules, eds Sigel A, Sigel H (Marcel Dekker, NewYork), Vol 33, pp 619–648.

4. Stubbe J, Kozarich JW, Wu W, Vanderwall DE (1996) Bleomycins: A structural model forspecificity, binding, and double-strand cleavage. Acc Chem Res 29:322–330.

5. Neese F, Zaleski JM, Loeb-Zaleski K, Solomon EI (2000) Electronic structure of activatedbleomycin: Oxygen intermediates in heme versus non-heme iron. J Am Chem Soc122:11703–11724.

6. Sam JW, Tang X-J, Peisach J (1994) Electrospray mass spectrometry of iron bleomycin:Demonstration that activated bleomycin is a ferric peroxide complex. J Am Chem Soc116:3250–5256.

7. Westre TE, et al. (1995) Determination of the geometric and electronic structure of activatedbleomycin by using X-ray absorption spectroscopy. J Am Chem Soc 117:1309–1313.

8. Burger RM, Peisach J, Horwitz SB (1981) Activated bleomycin: A transient complex ofdrug, iron, and oxygen that degrades DNA. J Biol Chem 256:11636–11644.

9. Hecht SM (1986) The chemistry of activated bleomycin. Acc Chem Res 19:383–391.

10. Kuramochi H, Takahashi K, Takita T, Umezawa H (1981) An active intermediateformed in the reaction of bleomycin–Fe(II) complex with oxygen. J Antibiot 34:576 –582.

11. Worth L, et al. (1993) Isotope effects on the cleavage of DNA by bleomycin: Mechanismand modulation. Biochemistry 32:2601–2609.

12. Wu JC, Kozarich JW, Stubbe J (1985) Mechanism of bleomycin: Evidence for a rate-determining 4�-hydrogen abstraction from poly(dA–dU) associated with the forma-tion of both free base and base propenal. Biochemistry 24:7562–7568.

Fig. 6. Models of reactant [ABLM and sugar] and TS [ABLM � sugar]showing H bonds between hydroperoxide and sugar ribose. At the TS, theO–O bond has elongated from 1.5 to 2.73 Å and the C–H bond from 1.10 to1.17 Å. The Odis–H distance is 1.45 Å. The TS is described in greater detail inref. 14.

Table 3. Estimated energies and energy barriers forABLM reactions

�Erxn0

,

kcal/molEa(expt),kcal/mol

�E�0,

kcal/mol

ABLM � DNA (C–H) �7 4.7 8 � 1ABLM decay (N–H) �6a 9.3 6 � 1

*�Erxn0 for ABLM decay was estimated to be 13 kcal/mol more endergonic than

the DNA reaction due to a stronger N–H bond.

13244 � www.pnas.org�cgi�doi�10.1073�pnas.0806378105 Chow et al.

13. Burger RM, Tian G, Drlica K (1995) Oxygen isotope effect on activated bleomycinstability. J Am Chem Soc 117:1167–1168.

14. Decker A, Chow MS, Kemsley JN, Lehnert N, Solomon EI (2006) Direct hydrogen atomabstraction by activated bleomycin: An experimental and computational study. J AmChem Soc 128:4719–4733.

15. Kumar D, Hirao H, Shaik S, Kozlowski PM (2006) Proton-shuffle mechanism of O–Oactivation for formation of a high-valent oxo-iron species of bleomycin. J Am Chem Soc128:16148–16158.

16. Streitweiser AJ, Jagow RH, Fahey RC, Suzuki S (1958) Kinetic isotope effects in theacetolysis of deuterated cyclopentyl tosylates. J Am Chem Soc 80:2326–2332.

17. Barnes JA, Williams IH (1993) Theoretical investigation of the origin of secondary�-deuterium kinetic isotope effects. J Chem Soc Chem Commun 1993:1286–1287.

18. Glad SS, Jensen F (1997) Transition state looseness and �-secondary kinetic isotopeeffects. J Am Chem Soc 119:227.

19. Morlok MM, Janak KE, Zhu G, Quarless DA, Parkin G (2005) Intramolecular N—H���Shydrogen bonding in the zinc thiolate complex [TmPh]ZnSCH2C(O)NHPh: A mechanisticinvestigation of thiolate alkylation as probed by kinetics studies and by kinetic isotopeeffects. J Am Chem Soc 127:14039–14050.

20. Poirier RA, Wang Y, Westaway KC (1994) A theoretical study of the relationshipbetween secondary �-deuterium kinetic isotope effects and the structure of the SN2transition state. J Am Chem Soc 116:2526.

21. Hoehn ST, Junker HD, Bunt RC, Turner CJ, Stubbe J (2001) Solution structure ofCo(III)–bleomycin–OOH bound to a phosphoglycolate lesion containing oligonucleo-tide: Implications for bleomycin-induced double-strand DNA cleavage. Biochemistry40:5894–5905.

22. Luo Y-R (2003) in Handbook of Bond Dissociation Energies in Organic Compounds (CRCPress, Boca Raton, FL), pp 233–269.

23. Murdoch JR (1972) Rate-equilibriums relations and proton-transfer reactions. J AmChem Soc 94:4410–4418.

24. Marcus RA (1968) Theoretical relations among rate constants, barriers, and Bronstedslopes of chemical reactions. J Phys Chem 72:891–899.

25. Szilagyi RK, Metz M, Solomon EI (2002) Spectroscopic calibration of modern densityfunctional methods by using [CuCl4]2�. J Phys Chem A 106:2994–3007.

26. Davis MI, et al. (2003) Spectroscopic and electronic structure studies of 2,3-dihydroxybiphenyl 1,2-dioxygenase: O2 reactivity of the non-heme ferrous site inextradiol dioxygenases. J Am Chem Soc 125:11214–11227.

27. Dunitz JD (1995) Win some, lose some: Enthalpy–entropy compensation in weakintermolecular interactions. Chem Biol 709–712.

28. Ford DM (2005) Enthalpy–entropy compensation is not a general feature of weakassociation. J Am Chem Soc 127:16167–16170.

29. Frost AA, Pearson RG (1953) in Kinetics and Mechanism (Wiley, New York), pp191–223.

30. Hammett LP (1935) Some relations between reaction rates and equilibrium constants.Chem Rev 17:125–136.

31. Hammett LP (1938) Linear free energy relationships in rate and equilibrium phenom-ena. Trans Faraday Soc 34:156–165.

32. Shaik SS (1988) How is transition state looseness related to the reaction barrier? J AmChem Soc 110:1127–1131.

33. Kemsley JN, et al. (2003) Spectroscopic studies of the interaction of ferrous bleomycinwith DNA. J Am Chem Soc 125:10810–10821.

34. Loeb KE, Zaleski JM, Hess CD, Hecht SM, Solomon EI (1998) Spectroscopic investigationof the metal ligation and reactivity of the ferrous active sites of bleomycin andbleomycin derivatives. J Am Chem Soc 120:1249–1259.

35. Loeb KE, et al. (1995) Spectroscopic definition of the geometric and electronic structureof the non-heme iron active site in iron(II) bleomycin: Correlation with oxygen reac-tivity. J Am Chem Soc 117:4545–4561.

36. Frisch MJ, et al . (2004) Gaussian 03, (Gaussian, Inc., Wallingford, CT).

Chow et al. PNAS � September 9, 2008 � vol. 105 � no. 36 � 13245

CHEM

ISTR

Y