1H-, 13C-, 31P-NMR studies and conformational analysis of NADP+, NADPH coenzymes and of dimers from electrochemical reduction of NADP+

Biochimica et Biaphysica Acta, 1076 (1991) 49-60 © 1991 Ekerier Science Publishers B.V. (Biomedical Division) 0167-a838/01/$03.5¢ ADONIS 016748389100064G

49

BBAPRO33798

~H-, ~C-, 3~P-NMR studies and conformational analysis of NADP ÷, NADPH eoenzymes and of dimers from electrochemical reduction of NADP ÷ Enzio Ragg 1, Leonardo Seaglioni t, R. Mondelli ~, I. Carelli 2, A Casini 3 and S. Tortorella 3 t Dipartimentodi Scienze Malecotan'Agraalimentan, Universit A di Mdang Milano (Italy), ~ Dipartimento di Chimica~ LC.M., Uaiversitd de L'AclUila, L'Aquila fltaly) and J Dipartimento di Studi di Chimica e Teotologia dells Soslanza Biologicameme Auive, Uni~tsit~ di Roma ' La Sapienz a ; Rama ( haly )

(Receiwd 9 August t990)

Key words: NMR; Conformation analysis; Electrochemical reductiom Nic~linamide adenine nuclemide

All H,H, H,P and several C,P coupling constants, including those between C-4' and the vicinal phosphorus atom, have been determined for NADP ÷, NADPH coensymes and for a 4,4-dimer obtained from one-deetron electrodamical reduction of NADP +. From these data the preferred conformation of the ribese, that of the 1,4-dihydtonieolinamide rings, and the conformation about bonds C(4')-C(5') and C(5')-O(5') were deduced. The preferred [arm of the 1,4and l~-dihydrolffridiue rings and the conformation about the ring-ring junetlon were also obtained far all the other 4,,L and 4,6-directs fonued in the same reduction. All Ihe dimers show a puckered stng'hn~, i.e., a boat form for the 1,4fiand a twist-beat for the 1,6-dihydronlcotinamide ring; both protons at the rlng-ring junctions are equatorial and have preferred gauche orientation. On the contrary, the reduced coenzyme NADPH displays a planar or highly flexible conformation, rapidly flipping between two limiting boat structures. The conformation of the ribose rings, already suggested for the NADP ¢uenzymes to be an equilibrium mixture of C(2")-endo (S.~pe) anal C(3')-ende (N.~pe) puckering modes, has been reexamined by using the Mtona procedure and the relative proportien of the two modes has been obtained. The S and N fmnflles of conformers have almost equal population for the adeniue.rlbnse, whereas for the nkollnamide.rlbose rings the S-type reaches the 90%, The rotatlon about the ester bond C(5')-O(5') and about C(4')-C(5'), defmad by torsion angles fl and ¥ respectively, displays a constant high preference for the trans conformer ~ (7S-80%), whereas Ihe retainers ¥ are spread out in a range of different populations. The values are distributed between the gauche ¥ + (48-69%) and the truns yt forms (28-73%). The 7 + conformer reaches a 90% value in the ease o| NA1DP+ and NMN +. The conformations of the mononuelcotides 5'.AMP, NMN + and NMNH were also calculated from the experimental coupling constant values of the literature.

lnlrnduelion In the preceding paper [1] we have reported the 1H-NMR analysis and the structure determination of the 4,4- and 4,6-1inked dimers (Scheme i) obtained from orb,e-electron electrochemical reduefioa of NADP* We report here on a 13C and 3,p study of the a,4-dimar (XI in the accompanying paper [!]) isolated as a 90f[ pure compound. In the course of this study, we needed to perform parallel NMR analyses, includkng ~H, on coeazymes NADPH and NADP +, as chemical

Correspondence: It, Mondelli,Dipanimento di ScienzeMolecolmi Agtoalimentari,Via C.eloria2, 20133,Milano,Italy.

shifts and coupling constants for the nuclei of file phosphodbose chain were not available. Actually, although a few errors have been corrected [2,4], there is great confusion in the literature [5] about these NMR parameters, which were generally derived through incomplete and often approximate approaches [2,3,6-9]. Therefore all H,H, P,H and the most important of the P,C coupling constants were obtained and are here reported together with those of dimer XI. The conformation of nicotinamid¢ coenzymes has been the subject of many papers [3,4,7-M], hat the preferred conformations in solution have not yet been unambiguously determined. Nevertheless. much of the discussion, for instance, on the stereoselectivity of enzymatic transfer of hydrogen from NADH [15l, has

50 been based on the assumption that the 1,4-dihydropyrldine ring exists in a boat form. Furthermore, it is usually accepted [5], with a few exceptions [10,12,13,16], that NADH occurs in folded forms, with adenine and nicotinamide rings stacked in parallel. The existence of folded structures was based on chemical shifts arguments [5] and again on the puckering of the dihydronicotinamide ring ]8]. The conclusion is that this model, after 30 years, still lacks sound experimental foundation, as recently pointed out by Bonnet and Duni!z [17]. Actually the determination of the crystal structure for N-substituted dihydronicotinamides showed that the ring is essentially planar [17] and NAD +, as free-acid and as Li salt, appears to be in the "extended" conformation in solid phase [16,18]. In addition X-ray studies of NADPH bound to a dihydrnfolate reductase, and of a number of dehydrogenases hound to NAD + demonstrated that generally the coenzyme exists in an 'ex~ tended' form [19]. We report here on the results of a conformational analysis of NADP +, NADPH and of 4,4- and 4,6-dimars [1] in aqueous solutions; specifically, on the prefel'red conformation of the ribose and dihydronicotinamide rings; on the conformation around bonds C(4')-C(5') and C(5')-O(5"), for dimer XI and for the coenzymes; on the conformation around the ring-ring junction, and on the preferred forms of ff~e 1,4- and 1,6-dihydropyridine tings for the dimers. Materials and Methods #-NADP + monosodium salt, tetrahydrate, M, 837.5 from U.S.B. (U.S.A.) and NADPH tetrasodium salt, M~ 833.4, 98% pure from Boehringer (F.R.G.) were used for NMR studies. The 4,4.dimer Xl was prepared by oneelectron electrochemical reduction of fl-NADP+, free acid. as described in the accompanying paper [l]. NMR spectra were ~ceordcd with Bruker CXP-300 and AM-500 spectrometers. Chemical shifts are in ppm (8) values from external 3-(trimethylsilyl)-propane-1mlphonic acid sodium salt hydrate (DSS) for lH and ~*C, and from external 85~o H~PO4 for 3tp. Estimated accuracy for tH and tZc 4_.0.005 ppm and +0.01 for ~tP. Coupling constants are in Hz, accuracy __+.0.1 Hz, unless specified in the tables. The spectra were measured in D~O (20 rag, ml -~) at pH 9.3 _+0.l (NADPH and Xl) and at pH 7.2 + 0.1 (NADP ÷); the solutions were adjusted to the desired pH with NH 3 and kept under nitrogen; NADPH and dimcr XI were also measured in 0.15 M sodium tetraborate buffer (pH 9.3). The variation in ~H and ~3C chemical shift with concentration in the range 20-50 mg-ml -I are within 0.1 ppm. All the compounds are stable for more than 24 h in the experimental conditions used. UC assignments were performed by heteronuclear shift-correlated two-dimensional experiments, with concentrated solutions;

quaternary carbons ~ere assigned on the basis of chemical shift, relaxation time and long-range coupling information. For COSY spectra and HOD signal suppres, stun, see the accompanying paper [1]. Spectra simulation was performed by using the PANIC program. Calculations of the populations relative to the different conformers were performed by solving the appropriate linear equation:

where Job~ is the experimental coupling constant, x~ denotes the molar fraction of the i-th species and J~ is the coupling constant relative to the pure i, th conformation. The equations are solved as sets of linear equations, with the additional constrain that V,x, = 1. The relative populations of S-type conformers are calculated as arithmetic means of two values, obtained independently from the experimental values of J(HI',H2') and J(H3',H4"). The estimated error on the populations is within _+5%. Molecular mechanics calculations were performed by using the CVFF force field [38], with the partial charges calculated with the MOPAC program [39].

Results and Discussion ~H-NMR spectra of N,4DPH and NADP + Both forms of the coenzyme were analysed following the procedure used for dimer XI [1] and the same experimental conditions, except for NADP +, which is not stable at pH higher than 7.5. The results are given in Tables 1 and 11; and partial proton spectra arc reported in Figs. 1, 2 and 3. The spectral pattern of some rihose protons, for the coenzyme and dimer X], are strongly dependent on th~ relative chemical shift difference between the two phosphorus nuclei of the pyrophosphate chain. Whenever this difference is very small, as is the case with NADPH (Fi~, 1) and with dimer XI in D 2 0 / N H ~ solution (Table I|l), the patterns of the adenine-rlbos¢ protons at C-4' and C-5' an: deceptively simple due to second order effects, while the corresponding protons of the nicotinamidc fragment lie together at 4.05 6, A firstorder analysis of A4' * and A5',5" leads to incorrect results, whereas the simulation of the spectrum is good if both phosphorus nuclei are included in the LAOCOON type calculation with a 0.01 ppm shift difference. When the cuenzyme is oxidized, the ~tp shift

* NumberedA's and N's indicatethe atoms(hydrogensand carbons) or the adenine nucleotide and those o1' nicotinamide necleolide unit. Tespectivety.Unspecifiednumbers indicate atoms of beth units.

51

TABLE 1

TABLE II

tH ¢l,¢mlcal shift values for NADP

+,

NADPH and direct Xi

H,H coupling constant values/or NADP ÷, NADPII and direct X i

Measured in ppm (,~) relative to external DSS: estimated aeeurac~ :k0.005 ppm. unless specified. NADP +

NADPH

(a)

(b)

(c)

XI

(v)

A2 A8 N2

8_130 8.=~(I 9.280 8.809

N5 N6 AI' A2" A3' A4' AS' AS" N]' N2' N3' N4' N5" N5"

8.180 9.105 6.093 4.055 4.594 4.363 4.283 4.187 6.023 4.456 4.403 4A93 4.311 4.204

8.230 8,453 6.919 Z853 d 2.719 ~ 4.79 r 5.893 6,194 4.933 4.577 4.375 4.286 4.194 4.?6"] 4.256 4.~8 4.028 4.091 4.015

8.198 8.427 7,091

N4

8.214 8.452 6.913 2.815 d 2,728 ~ 4.80 r 5.943 6.183 4.915 4.~65 4.354 4.277 4.172 4.77 r 4.146 4.197 4.045 4.020 4.0t?

CONH2

H

NADP+ a

j

2.903

4.4.56 6.099 6.167 4.920 4_586 4.372 2.272 4.209 4.85 4.302 4.300 4.044 4.125 4.042

DsO solution (pH 7.2)~ D~O/NH s solution (pH 9.3). c 0.I5 M sodium tetraborate buffer in 020 (pH 9.3)e.© The 8eminat protons at C.-4 are diastereotopic; (d) N4a, (e) N4b. f Par6ally overlapped by HOD signal: accuracy 4-0.01 ppra.

H

J vaktes are measured in Hz and are I~ivco without tlif~: estimalfd accuracy 4-0.1 Hz, unless Slz:cified.

H

N2,N4 N2,N6 N2,N5 N4,N4 N4,N5 N4.N6 NS,N6 NI',N2 ~ N2',N3' NY,N4' N4",N5' N4*,NS" NS",N5" AI',A2' A2',A3' A3',A4' A4',AS" A4',AS" AS',AS"

1-7 1.5

8.1 1.3 6,3 5,5 5.0 2.7 2.5 2.3 12.0 5.0 5.3 4.6 %6 4.6 11.8

NADPH b

X[ b

0.7 ¢ 0.5 c.d 1.7 0.3 d 18.2 f 3,3 ~ 3.7 e 1.9 = 1.7~ 8.2 7.5 h 5.4 h 2,1 h 3.5 5.0 11.8 4.7 4.9 5.1 2.7 4A 1[.8

0.6 IA 0.5 = 2.3 * 4.8 0.7 8.0 7.0 h $,5 h 2.3 h 3.7 5.0 12.0 5.0 5.0 4.8 3.0 4.6 11.8

= DzO solution (pH 7.2). b 0.t5 M sodium tetraborat¢ buffer iu I~O (pH 9.3). The atmlysis in DzO/NH J solution (pH 9.3) gave the same results; for XI, see Table II of Ref. l; for NADPH, erdy J(A5',A$") changes to 11.6 Hz. ¢ Coupling constants with the geminal protons, N4a (left) and N4b (tight), respectively (see Table I) d E~timated from the linevddth. = Not mcmsttred.,but obtained f,om th~ calculated ~p~-ttum. f Geminal coupling constant. a Coupling coeslant between protons at the ring-ring junction. h Accuracy *0.2 Hz.

d i f f e r e n c e is 0,35 pp~a, a n d t h e p r o t o n s p e c t r u m (Fig, 2) is nearly f i r s t - o r d e r ;rod can b e u n a m b i g u o u s l y analysed. For both N A D ? H a n d dimor XI, this degeneracy w a s r e m o v e d b y t~sing s o d i u m t e t r a b o r a t e b u f f e r iR D 2 0 a n d l e a v i n g t h e p H v a l u e o f 9.3 u n c h a n g e d , I n this solution t h e c h e m i c a l shift d i f f e r e n c e b e t w e e n t h e p h o s p h o r u s nuclei c f t h e c h a i n i n c r e a s e s to 0 . 2 - 0 . 3 F p m a n d

4,4 d i m e r s

TABLE Ill

CONH2

H

H

3/

P chemical '.~i~ifrvalues for NADP

+

, NA DPH

and dimer Xi

In ppm ft~m eatemal 85% H~PO,t, estimated accuracy :t:0.01 ppm. Negative numbers denote upfie[d shifts.

R

B

'-

6 --

3

CONH=

lk P-2'A P-5'A P-5'N 4,6 d l m e r s

Scheme 1.4,4.- and 4,6 -lirff~l NADP dimers [ll. R = adenosine (2% phosph at e)-.diphosphat =-ribo.~.

NADP +

NADPH

(~)

(b)

3.477 --10.$44 - 10.886

X| (c)

4.44)9

- ]0.474

4.434 --10.392 -10.170

(c)

(b) 4.26" - 10,505

" D:O solution (pH 7.2). b DzO/NH 3 solution {p~ 9.3). 0.]5 M sodium tctraboratc buffer in D~O (pH 9.3).

4.288 -- 10.534 - 10.308

52

N2"

N 3'

N4"* NS' hiS"

As*'

AS" A4'

I

I

4.4

4.3

'

t

I

4,2

I

I

4.1

4.0

ppm

Fig. 1. tH-NMRspectrum(300 MHz, D20/NH 3 {pH 9,2))of NADPH.regionfrom40 to 4.4 6.

leads to a small de.shielding of 0.05-0.1 ppm for protons N2", N3" and NS'. Although the dbose protons of adenine [ragments are not significantly affected (0.01 ppm), the shift difference belweea 3~p nuclei induces

N2"

variations in the patterns of the coupled protons A4' and AS' (Fig, 3); the analysis is thus easier, since the P,H coupling constant values can be cross-checked with those obtained from the 31p spectrum. In addilion, the

N3"

4I 1

4.5

I

4.4

I

I

4.3

4.2

Fig.2. 'H-NMR spectrum (500 MHz. D 2 0 / N H ~ (pH 7.2))of N A D P ~'.re,on from 4.1 to 4.5 &

~ ppm

(b)

N2' N4" 5

1

4.40

A4'

A5" .

A

4 30

4 20

Jl

,,

N5

t/ N5

410

400

/i

ppm

Fig. 3. (a), IH-NMRSlXa:tram(300 MHz.DzO/~etr~horate(pH 9.3))of NADPH,regionfrom3.9 to 4.4 6: (b),cal¢ulaledspectrum.

slightly Iowfield shift of NS' decreases the overlapping with NS", allowing a complete analysis also for the nicotinamide fragment. All H,H aad P,H coupling constants were obtained. ~lp a#d ~SC ~pectra of NADPH, NADP + and of lhe 4, 4-dimer XI The phosphorus at C-2' is always a doublet of ~bout 7 Hz for all compounds, while the two phosphate signals of the chain vary. NADP + shows a well-resolved patteru (Fig. 4), whereas both NADPH and dimer XI give a degenerate singlct in D20/NH3, even at 121.4 MHz. This is in agreement with the results of Feeney et aL [20,21[ who found the same degeneracy in the 31p

spectrum of the reduced coenzyme at pH 7.9 (KCI, Tris buffer, EDTA). With sodium tetraborate buffer in D20, P-5'N moves to lowfield, 0.3 ppm for NADPH and 0.2 ppm for X! (Fig, 5). The shift difference between the phosphates of the chain allowed to obtain all P,H and P,P coupling constartts (Table IV), thus confirming the results from the proton spectra, The assignment of adenosine-5'-phosphate, P-5'A, and nicotinamide riboside-5'-phosphat¢, P~5'N, can now be made with certainty (Table I11): for NADP + at pH 7,2, P-5'A is at low field and P-f'N at high field, while, for both NADPH and direct Xl at pH 9,3, P-5'N is deshielded with respect to the partner. The relative attribution of the two phosphates remains the same when NADPH is

';4

P-S'A

i

I

I

I

- 10.4

P-5'N

I

-10.6

i

I

I

-10.8

-11.0

ppm

Fig. 4. :qP-NMR spec~Jum (121.4 MHz, D 2 0 / N H 3 (pH 7.2)) of the pyrophosphate fragment in NADP +.

bound to the enzyme dih3,drofolate reductase [20,21], but for NADP + the shift appears to be reversed with respect to the enzyme-NADP*-folate complex [21].

]JC spectra were performed in the same conditions used for proton and phosphorus spectra. The signal assignments (Tab]e V), obtained by heterormclear two-

P-5"N

P - S'A

I

ppm

- 10.1

- 10.2 31

- 10.3

- 10.4

Fig. 5. P-NMR spectrum (121.4 MHz, D~O/tetraborate (pH 9.3}1 of the py~phosphate ffagmel~t in NADPH.

-10.5

55 TABLE IV

Conformation of ribose rings

P,H P,C and P,P coupling eonstunl values for NADP +, NADPH wrd dimer XI

An inspection of Tables i - V shows that coupling constants and chemical shifts of dimer X! are equal to those of NADPH, except for the nuclei involved in the ring-ring junction; in addition, the coupling constant values are similar Io those of the corresponding mononuelcotides 5'-AMP and NMN [3,7,23,24]. The ribose rings in these mononucleotides have been reported to exist as an equilibrium mixture of C(2" )-undo (S-type) and C(Y).endo IN-type) conformers; the same conformations have also been suggested for the coenzymes in solution I3,23,24,25]. We have calculated, following tLe method of Altona et al. [26-281, the relative population of the S-type vs. N-type conformers, for dimer XI, NADP +, NADPH and, by using the coupling constant values of the literature [23,24], for the mononuclcotides 5'-AMP, NMN +, NMNH (Table VI). A large number of X-ray data give averaged values of the pseudorotational angles PN = 9 ° and Ps -~ 162% Calculated J values from model geometries of PN = 9* and Ps ~ 153" are in agreement with our experimental values of J ( H I ' , H 2 ' ) + J ( H 3 ' , H 4 " ) (9.6-9.8 Hi0. This sum for NMN* and for the nicotinamide fragment of the oxidized coenzyme is 7,8 and 8,2 Hz, respectively, because the substituent effect of the positively charged nitrogen atom directly attached to C-I' decreases the

J values are measured in Hz from IH, 3lp uod I~C spectra and are given without sign:, estimated accuracy ±0.1 Hz unlegs specified, NADP+ a J(P.HI P2',A2' PS",A4"

NADPH b

XI h

7.1 2.0

6.8 2.0

6.8 2,0

Ps",As' PS",AS" PS'.N4'

5.0 4.9 3.0

S.2 4.8

1.7

5.0 5.0 1.7

PS',N5'

4.2

5.0

5.0

PS',N~"

5.0

s.6

5.2

5.5 5.0 4.5 8.5 5.0 8.4 5.2

8.0 4,2 2.0 9.0 5.0 8.2 6-0

8.5 ~ 4.2 ¢

20.5

~r0.0

/(p.c) P2'.hl" P2'.A2' P2'.A3' PS",A4' PS',A5" PS',Na' PS',N5 ~ J(P,P}

8.5 c 8.5 ~ 20.0

a DzO solution (pH 7.2}. 0.15 M sodium tctraborate buffer in D~O (pH 9.31. The coupling coaslanls for adeaine-nbose fragment and J(P5'N4"; ~',¢,,. also measured in D 2 0 / N H ~ solution fpH 9.3) and show similar values. These cv.plings were obtained onty [rum D~O/NH~ solution.

TABLE V

~'C chemical shift values for NADPH and dimee XI

dimensional correlation, are in agreement with those reported [4,22] for the coenzymes; only NS' vs. AS' are reversed for NADPH. When tetraborate solution is used, NADPH and dimer XI display dcshielding effects on NI', N2' and NY of 3.5-6.5 ppm, which are significantly larger than those found for the corresponding protons and for phosphorus. At present we do not yet have a definite interpretation of the effect being studied, but from preliminary results it appears that it is due to the borate ion mid not to the counter-ions Na + or NH~, An investigation on the effect of these and other counter-ions is in progress. Several carbon-phosphorus coupling constants for ribose nuclei were also measured (Table IV). Of partieular interest are the three-bond interactions of A4' and N4' with the vieinal phosphorus atom of the diphosphate chain never determined before for the coenzymes. The coupling constant values were calculated from the experimental a3C spectra, since there are second order effects. For instance A4" and N4' patterns of NADPH and dimer XI are triplets of 4.0 Hz. The same signals for NADP + arc doublets of 9.0 Hz. The strong degeneracy in the former cases is again a consequence of the coincident shifts of P-5'N and P-5"A. The three cornpounds show similar values for these couplings (8.2-9.0 Hz).

In ppm (8) from exlemal DSS~ accuracy _+0.01 ppm The assignments were performed by heterouuduar shift-correlated twoMimen~ioual experiments. NADPH

co A6 A2 A4 A8 hi2 N6 AS N~ N3 NI' AI' A4"

N4' A2' N2' N3" A3' NS' AS' N4

XI

(a)

(h)

(a)

(h)

175.33 158,16 155,35 151,62 142.75 140.79 126.61 121.33 107.83 102.73 97.58 89.49 85.48 84.93 78,99 73,36 73.10 72.65 68.33 67.88 24,48

175.55 158.29 155.39 151.59 142.82 141.13 127.13 121.47 107.80 102.1.'1 101.07 89.67 85.63 85.57 79.31 80.22 78.69 72,76 68.33 67.94 24.58

175.61 157.98 15524 15t.54 142.61 13"].47 131.55 121,18 105.71 t05.61 97.14 89.16 85.52 84.75 78.98 74,64 72.95 72.76 68.38 68~03 41~63

174.67 158. I ', 155.37 151.63 142.78 137.98 132.13 121.(13 105.96 105.60 100.61 8%28 85.62 85.62 79.18 81÷12 78.52 72.86 69.02 68.07 41.1~4

' DzO/NH J solulion (pH 9.3), h 0,15 M sodium tgtrgbQrate buffer in D20 (pH 9.3).

56 TABLE V1

P~putaOuns(%) olcon/ormeesfor NADP ", NADPH~dimer XI andfor Ihe correspondingmonunurleot~des The populations were ohtained from the experimental coupling constants of Table il and Ill; for the monucleotides 5'-AMP, NMN + and NMNH, the experimental values of the literature [23.24] were used. The calculations were performed by solving the 0ppropriate equation (see Materials and Methods). Estimated actor :1:5~ unless specified. Compound

Fragmenl

Ribose a

Around C(5')-O(5') h

NADP + NADPH

A N A

S .58 85 52

#' 80 79 87

N

94

76

Dimer XI

A

56 87 73 ~ 93 90

~ 80

N

5'-AMP e NMIq * f NMMH t

P' 75 76 75

Around C(4') C(5') ~

H(4')-C(4')~C(5') ~O-(5")-P d w~ 54 81 54

#" 12 15 12

#13 9 13

C 62 88 63

v' 36 It 34

v2 l 3

72

15

13

75 ")4 78 ~ 8t 78

13 14 It l0 1l

13 13 11 9 II

50

37

13

46

59 48 70 ~ 91 K 57 s

34 36 28 7 24 h

7 15 2 2 l~ ~

54 46

P'-Y" 47 67 47 36 44 36

' Population calculated followingthe proceduxeof Altona el aL [26-28l and assuming pseadorotatioaal angles Ps 153° and P~ 9 ° and puckering angle ~= 37 0. b The ~8t population in the first column was calculated from J(C4'-PS') hy using Eqa. 2 and J values [31] of 10.0 Hz and 2.5 Hz for angles of 180° and 60". respectively.The populations of the.6' conformers in the other colursns were calcolal~ by using Eqn, 3 and J values [31] of 23.0 Ha and 2.4 Hz for angl~s of 180 ° and 60*~ resp~:tivdy. The populations of the T conformers were ¢-alcuiutodby using Eqn. 3 and the "non~:lassicar model angles [29.301 y+ 53°, ¥' IS2* and Y- 292°. d Percemage or w conformation [33,34] for this fragment. Obtained from 'J(H4'-IPS"), assuming a maximum coupling og 3.7 Hz [30] and zero coupling for non-W geometry. r Calculated from the ¢xperaneatal coupling constants at pH 8,0-8.3 123.24]. In agreement with the reported values [23.24l: 67, 85, 65. 93 and 505. respectively. h These values derive from estimated coupling constants [23].

value o,*"J ( H I ' , H 2 ' ) . For these compounds, the calculations were thus performed omitting this coupling. The puckering angle ¢,~, assumed to be equal for N- a n d S-type conformation, was estimated from J ( H 2 ' , H 3 ' ) . The high values (4.9-5.5 Hz) of this coupling indicate a low puckering amplitude of about 37 o, which is in agreement with the majority of the X-ray d a t a (¢bm = 36-38 ° ) [26]. Thus, with the values Pr~ = 9 ° , Ps = 153° and em = 37¢, we can calculate the molar fraction of the S-type puckering mode. Actually with a Ps value of 153 ° the S conformations are intermediate between the C(2')-eado =E and the twist ~T. The results are reported in Table VI The eonformational equilibrium for the adenineribose ring results a 50% mixture of the two puckering modes (except for 5 ' - A M P with a 75% of S), whereas for the ribose adjacent to the pyridine moiety the population of the S conformer increases to 85-95%.

mational flexibility a b o u t these bonds makes the study of the pyrophosphat¢ chain very difficult in solution. In the s~lid phase, the conformations of the backbone chain for N A D + [16,18] and for several coenzym¢ protein complexes [19] have been determined. The orientation about C ( 5 ' ) - O ( 5 ' ) and C ( 4 ' ) - C ( 5 ' ) lies in the trans (,/] t) and gauche-gauche (7 + ) range, respectively, for both A and N unils, conformation which is usually considered as preferred for 5'-nuelcotidas [19]. A similar

"ON~ :(bridge)

o

Conformation about bonds C{5')-0(5') and C{4')-C(5') The proton-proton and proton-phosphorus coupling constants invohdng H-4', H-5', H-5" a n d P atoms have never been measured for these coenzymes [5,29-31]. The P,H three-bond couplings for dinucleotides were generally estimated from the line width of phosphorus signals [7,20,2I], which are always very close to each other, and oflen have short Tz relaxation times, leading to a considerable broadening [20]. Actually, the eonfor-

-~-- C$'

Os' I

_

base

SchemeIL Mononucleofidefragment,

57 conformation has also been suggested for the coenzymes in solution 17]. The complete set of couplings for the nuclei of the pyrophosphate chain is given in Tables II and lIl. From the values of J(P, HS'), J(P, HS") and J(P,C4') we derived the relative populations of the three main conformers around C(5'}-O(5') (j8 conformers), and from J(H4',H5'), 3(H4",HS") and aJ(P.H4") those aronnd C(4')-C(5') (7 conformers). The assignment of H-5' at low field with respect to H-5" was based on the results of stereospecifie partial deuteration of adenosine and its derivatives [32]. With this assignment, we have calculated from the experimental coupling constants that conformer 7 is always very little populated, for both A and N fragments. This result is in agreement with the finding, from a large number of crystal structures, that this rotamer is rare [19]. For the relationship between angles and P,H vie'real couplings we have used the Karplus/Altona equation reported in [29]. The relative populations of the three conformers Bt(trans), ~+(gauche) and #-(gauche) were then calculated following usual procedures (see Materials and Methods) and by means of the limiting coupling constants values Jl~ = 23.0 l-k. and J~ = 2.4 Hz, which are supported by m~ny experimental data 1311. The populations of the minor conformers (#+ and /~-) could be obtained, as the individual coupling constants of phosphorus with H-5' and H-5" were both available. The population of the preferred conformation, ,8', was independently deduced from an other parameter: the carbon-phosphorus coupling J(P, C4'), which is particularly significant, since it is dose (8.2-9.0 Hz), for

H~Hs.

I C~' [$t ( I r , m n * ) l a p l

--~

HS' H4' 7 + (gll)[*==l

H~C4

+

Hs' i-[o=uche)(-=© I

r-----~

Hs" Os' k14' Hs' ,ft {mr)lap]

C'

H~'

Hs" p* {g.uuc=he}l',~.c=l

/--- --'3

Hs" 1~14, Os' ¥-(qu}l-*©l

Scheme IIL Newman projeclionsalong the backbone angles ,0= C(4")-C..(5")-O(5')-P,~ad ',t=CA3')-C(4")-CO')=O(5'). Atomic nernberingconforms with the IUPAC-[UBconvention,as ~[erenced byAltona[30],

both coenzyrnes tad dimer Xi, to the maximum value (10 Hz) observ+;c for truss interactions i . . u d ~ i d ¢ ~ [311. The population P(/]~) was calculated by using the limiting coupling constant values J,.... --10 Hz and J , . . c ~ ~ 2.5 Hz [311. As shown in Table V], both adenine and nicotinamide fragments display a preference of 75-80% for conformer ~t. and the results from proton and carbon data are in agreement, The minor ,.onformers/]÷ and . ~ are equally populated, except for the N fragment of NADP +, where ,8+ is twice/~-. The conformation about C(4')-C(5') bonds was studied by using the 'non classical' model for the three main conformers, as suggested by Altona et al, [29,30], i.e., 7 +_ 53 °, -i,t = 182 ° and 7 - = 292 °. The relative population of the thcee conformers were thus obtained (Table VI) and compared with those similarly calculated from the experimental data [2324] of mononucleotides 5'-AMP and NMN. The adenine fragment showz a constant behaviour, with a preference for conformer ~,+ (59-63%) and -rt (34-36%}. Conformer "t- is pratically absent or shows a very low population, in agreement with the expectation [191. The nicotinamide fragment instead varies from a high preference for isomer 7* in NADP + (88%) to lower values (about 48-505) in NADPH and dimer XI, with a consequent larger participation of the minor conformers 7 ~ and y-. The values of mononueleotides are very similar: y+ 70% (5'-AMPL 90% (NMN +) and 57% (NMNH). These results are also confirmed by the values of the long-range couplings (J(P, H4'). The maximum value of this interaction, corresponding to a W geometry for the fragment H(4')-C(4')-C(5')-P [33,34], has been taken equal to 3.7 Hz, with an average of experimental co-piing constants of 3.3 Hz [30]. Among the compounds studied, NADP ÷ shows the largest values, with a 3.0 Hz coupling for the nicotinarnide fragment, in agreement with the population found for ¥+ and .B~ conlormers. The lowest value (1.7 Hz} occurs for the same fragment of NADPH. Assuming n maximum coupling of 3.7 Hz and a zero coupling for all "non-W' geometries, we have calculated the 'percentage of W conformation" (Table VI), which are in line with the product p t ~+ obtained from the vicinal couplings.

Conformation of the dihydronicotinamide ring in NADPH and in the 4,4- and 4,6-dimers A conformational analysis of the tetrahydrobipyddine system of the 4,4- and 4,6-dimers must consider the puckering of the dihydronicotinamide ring and the rotation about the C(4)-C(4) and C(4)-C(6) bond, respectively, at the ring-ring junction. In the case of 4,4-dimers, the proton-proton internelions J(N4,NS) and J(N4,N6), are diagnostic for the ring puckering, whereas the rotational process affects

58 ',he coupling at the junction site, J(N4,N4). It is interesting to note that "~J(N4,Nr) differs greatly from dimerle to monomeric ~pecies, being 1.7 and 1.9 Hz for NADPH and 0.5-0.7 Hz for XI and the other two diastercoisomers [1]. As a transoid aUylic coupling shows its maximum value (3.0-3.5 Hz) when the dihedral angle ~,= H(4)-C(5)-C(6)-H(6) is about 90 °, and a very small magnitude for angles near 0 ° [33], we can deduce that the N4 protons of the dimers are predominantly in pseudo-equatorial orientation, with z, angles of not more than 200-30 °. This corresponds lo the boat Bi.4 and lAB conformations for the 4R and 4S configuration, respectively. For axially oriented N4 protons, • angles measure 100", which should lead 1o a coupling of about 3 Hz. A planar structure of the 1,4-dihydropyridine rings, or flexible forms inlerconverling between two limiting boat conformations, are e~pected to give allylic coupling of 1.5-2.0 Hz. Consequently, the values of 1.7 and 1.9 Hz found for the geminal N4 protons of NADPH are in agreement with a neatly planar ring, or with a flexible structure (averaged couplings) rapidly flipping between two boat forms. The energy difference between planar and boat conformation has been reported to amount to only 2.7 kJ. m J [35]. Thus, the co~,clusions drawn by Kaplan etal. [8] that the dihydropyridine ring of the reduced coenzyme NADH is locked in a puckered conformation, are not consistent with our results. These authors utilized a 0.:5-0.8 Hz difference in coupling cone:ants, which in addition were obtained from not enough resolved spectra of deuterated materials. An examination of the vicinal coupling J(N4,NS) confirms, for 4,4-rimers, the equatorial preference of N4 protons, since values of 5.0-5.5 Hz exclude ~.ngles of 90-1(10 °. The geminal N4 protons of NADPH show coupling with N5 of 3,1 and 3.7 Hz: the small difference again cannot be taken as evidence for a preferred ring puckered conformation [8]. The non-equivalence of N4 protons is a consequence of the presence of chiral centers, which makes the two nuclei diastereotopic. The c,lher ;we |oug-range couplings are not very informative; J(N2,N6) shows for the coenzymes and 4,4-directs similar values, as expected, which are in agreement with a W geometry [33]. J(N2,N4) is not a pure allylic interactio,l, because the conjugative effect of CONH2 group increases t h e , contribution to the coupling mechanism with respect to the ~r contribution. The vlnylogous amide delocalization, as shown in the structure of Scheme 4, is considered to be dominant, with respect to the amide delocalizalion, in NADH [36] and in the 1,4-dihydronicotinamides [371. In addition, the effect of the electron-withdrawing substituent, at the central atom of the coupling pathway, is strongly dependent on the orientation of the substituent. The different values found for the three dimers [1], ranging I¥om 0.5 to 0.9 Hz, clearly retlect different steric situa-

NH2

[

C

[ R Scheme IV. Coplanar tr,,nsoid eonformaticmof lA-dihydronicolinamides, with vinylogc~Jsamide delocalization, lions, which however are difficult to evaluate. Actually the problem of the preferred conformation of the amide group for NADPH in aqueous solution is still an open question [36,371. A coplanar t r a n ~ i d conformation as shown in Scheme IV has been established for N-substituted 1,4-dihydronicolinamides in solid phase [17]. The same conformation has also been suggested [37] for NADH and NMNH in solution, on the basis of ultraviolet absorption and fluorescence data, and of the chemical shift difference between N2 and N6 protons in model compounds. The deshielding (about 1.0 ppm) of N2 vs. N6 for NAPDH, as well as for the three 4,4-dimers, might indicate that the oxygen atom of the amide is preferenlially oriented towards N2. Molecular mechanics calculations showed that the rotation around bond N4,N4 is easy, when the dihydropyridine rings of 4,4-rimers assume the boat conformation with H-4 equatorially oriented becanse the two rings remain approximately parallel. On the other hand the rotation becomes hindered in the ease of axial orientation. Preliminary results of these calculations also showed that the conformations of the rings are preferentially boat, instead of planar, and the H-4 protons are equatorial, in agreement with experimental data. The small values of the vicinal coupling at the ring-ring junction, J(N4,N4), are in favour, for the three isomers, of gauche conformations, the dihedral angles • = H(4)-C(4)-C(4)-H(4) having values of about 60 °. Significant populations of other forms with u angles between 120 ° and 180" can be excluded; the eclipsed forms with o = 120 o although compatible with NMR data, are energetically disfavoured. Thus a dynamic mixture of gauche conformations, or a single gauche, with boat forms of the rings, are the most preferred. This may lead to the stacking of the two dihydropyridine systems, enough to, allow possible z, orbital interactions between them. The relative population of the two gauche forms (Scheme V) can vary in the three 4,4-directs, and this accounts for the larger difference in chemical shifts found [1] for N4 than for the other protons.

59 H-

['14

..-Ca~. c;'~cs

:

i

g'°

c,'~-'-c~

4S,4R dimers g,

H4

r . . . ' C ~ H4 cZY'C~ ! 4S,4S dirnere H4

N.~'"

H4

"~"~N--

48,6H dimeri SchemeV. Newmanprojection~f the gaucheformsfor somediraers, alongthejunctionbond betv,ee.'thetwodihydtop~fidineziags~

The 4,6-rimers, described in the preceding paper [1], contain both the 1,4-dihydro and tl3e 1,6-dihydropyridine systems. The couplitlg constants for the former are very similar to those found (or Xl and its diastereoisomers, thus indicating a boa[ conformation with H-4 equatorially oriented. The J values are symmetrically comparable to those of the partner 1,6-dihydropyridine ring (see Table V of Ref. 1); in particular, the nearly zero values of the ailyLic interaction 4j(4B,6B), the low values (2,3--3.8 Hz) of the coupling constant at the ring-ring junction, and the 5.4-5.6 Hz of the threebond coupling (J(SB,6B). This shows that the conformation of 1,6-dihydropyridine rings is a twist-boat, with N6 protons equatorially oriented, and that the rotamers about N(4)-N(6) bond at the ring-ring junction are preferentially gauche (Scheme V). Coaclnsiens The preferred conformation of the dihydronicotinamide rings in solution, for the dimeri¢ species studied, (NADP)2, is a puckered structure; i.e., a boat form for the 1,4-dhhydro- and a twist-boat for the 1,6-dihydro-rings, with equatorial orientation of both protons at the ring-rlng junction. The rotation about

the b~-'td at the junction site is not hindered, but gauche co~:ormations are preferred. This allows some stacking of the two dihydropyridine units. In the reduced coenzyme NADPH, the dihydronicotinamide ring is planar or rapidly flipping between two boat forms. Recently [1"/] the crystal structure of N-substituted dihydronicotinamides, models fat the ~e~zymes, was reported to be practically planar. Since the energy difference between planar and boat form amounts only to 2.7 kJ-mo1-1 [35], a highly flexible conformation appears the most favoured in solution. Although the NMR data cannot distinguish between planar and interconverting boat structures, the puckered conformation given for NADPH by Kaplan et at. [8] and generally accepted [5,15,17], can definitely be excluded. The reduced coenzyme NADPH and the 4,4-dimer examined~ atthough different in the form of the dihydronicoPmamide ring, are similar in the ribose struclure and in the conformation about the C(4')-C(5") and C(5')-O(~') bonds. The conformation of the ribose rings consists of an equilibrium mixture of C(2').endo(S-type) and C(3")eado (N-type) families of conformers, as suggested by Sarma et al. [3,23]. The two puckering modes have approximately equal populations for the adenine-ribose in NADP +, NADPH and in the 4,&dimer (S 52-585), Whereas the nicotinamide-ribose shows a significant increase (85-94~) of the S-type. The mononucleofides NMN ÷ and NMIqH behave in a manner analogous to the corresponding units in the coenzymes, whereas 5'AMP shows 75~ of S. The increase of C(2')-endo puckering does not appear to be correlated with the positive charge on the aicotinamide nitrogen, but rather it should be related to the geometry at the glycosidic bond and to the nature of the attached base. This correlation has mainly been postulated on the basis of X-ray data and energy calculations [19]; if this is true also in solution, we should expect a larger population of syn conformer for the pvridyl than for the adenosyl moietyl. Actually, the syn and anti conformers were found to be equally populated for the pyridyl fragment of HAD + in solution [12,13], while definite and unambiguous evidence of a presumed [24] anti preference for the adenosyl moiety, is still lackin& An attempt to delermine the X angles by NOESY technique was unsuccessful because the correlation time of dimers in water lies in the region where the NOE effect is close to the zero-crossing point between positive and negative NOE. As concerns the rotation about the C(5')-O(5') ester bond, all the compounds studied display the same high preference for copra .,~..,~/~+ (75-80%), in agreement with the X-ray analyses of HAD + [16,18] and of the majority of nucleotides [19|, where angle # is largely limited to the ap (trims) range. However, the minor gauche conformers ~+ and ~ - at: present in solution

60 and are equally populated, except for the nicotinamide fragmenl of N A D P +, where [/+ is about twice/~-. The rotamers about C 1 4 ' ) - C 1 5 ' ) are distributed between -r + (48-69%) and yt (28 37~), with a preference for the gauche conformation. The preference increases to about 90% for the nicolinamide fragment of N A D P + and for the correspondingmonomer N M N +. This is no1 related to the ribose puckering, since N A D P H and N M N H , with the same puckering mode. show ealy 50~ of conformer 7++ Actually, the 3' angle does not seem to be correlated with any other angle 119], contrary to what has been previously suggested [24t; rather, the type of base seems to play a role in stabilizing the ~,+ form, expecially when electron-wilhdlawing groups are present on the base [19]. These results confirm the above predictions: the posilive charge on 1he nicotinamide ring appears to be responsible of ~he stabilization of the 7 + conformer, possibly inducing electrostatic attractive interactions between the base and lhe phosphate group, also mediated by water molecules. That these interactions occur, has also been postulated [40] on the b a i l s of the atp npfield ~hift generally observed in the oxidized dinueleotides and mononucleotides, with respect to the reduced forms. N o w the assignment of ~lp signals es(ablishes that the most affected of the two is indeed the phosphate in the nicotinamide unit. However, it is known [311 that strong changes in 31p chemical shift are due to variation in the a torsion angles about the P-O{5' ) ester bonds, as welt as to variation in the bond angles at 1he phosphorus atoms; variations that might also be induced by electrostatic interactions. Several data are available from X-ray analyses [19], but in solution the only information on these angles comes from chemical shift values. In the pyrophosphale chain, an additional and important factor are the torsion angles at the P O(bridge); these angles have been considered to be the main source of conformational flexibility for these coenzymes, in which the two n u d c o t i d e units have always been seen as rigid entities 119]. This model should probably be revised as well as the old que,',tion of whether these coenzymes exist in a folded or extended form, a n d if so to what extent, After the papers of Ellis et al. [12331 this problem has no longer been tackled, and is still waiting for an unambiguous answer. Re|eRnces 1 Ragg E.. Sc~ag/ioni,L., Mondelli, R., Cardli, V.. Cardli, I., Casini, A.. Finazzi-Agr6, A., Liberator.'. F. and Touorella. S. (1991) Bioehim l]iophys. Aeta 1076, 37-48+ 2 Oppenheimer, N.J., Arnold, I-.k and Kaplan, N.O, 11971) Prec. Nat. Aead+Sci, USA 68. 32110-3205. 3 Sarma, R.H. and Mynott, 1LJ. (19731 J. Am, C h c m Soc., 74707480. 4 Birdsall, B. and Feeney, J+ 41972) J. Chem. Soc. Perkin [I, I6431649. S You, K. (198S) CRC Cril. Rev. giochem. 17, 313-365.

6 Sarma. RH. and Kaplan. N O 0970) Biochemistry 9, 539-54g and 557-564.

7 Sarma, R.H., Mynou. R.J.. Hruska, F.E. and Wood, DJ. 11973) Can. J. Chem. 51. 1843-L851. 8 Oppenheimer, N.J., Arnold. LJ., Jr. and Kaplan, N.O. 11978) Biocheraistry 17. 2613-2619. 9 LappL D.A.. Evans. F.E. and Kaplan. N.O. 11980) Biochemistry 19. 3841-3845. ]fl Jacobus. 3.11971) Biochemislry 10, 161-164+ ]1 Hamill, WJ., Jr. Pugmirc. R.I. and Grant, D.M, 0974) J. Am. C;~-,- ..q~c.9fi. 2895-2~$7. "12 Zeus, A.P., Williams, T-L Wisewaty, J.C., Fisher, R.R., Dunlap. ILB., Bryson, T+A. arid Ellis. P.D. 4197-5)J, Am. Chem. $oc. 9"L 2g~0-2857. 13 ZetlS, A.P., B ~ n , T,A+, Dunlap, R.B.+ Fisher, R.R. and Ellis, P.D. 11976)J. Am. Chem. Soc. 98, 7559-7564. 14 Sovago. I. and Martin. R.B, 11979) FEBS Len. 106, 132-134. 15 Namhiat, K.P. Stauffer. D.M., Kolodziej. P.A. and Benner, S.A. 41983) J_ Am. Chcrm Soc. lOS, 5886-5890. 16 Rcddy+ B.S., Saengcr, W., Miihtcgger and Weimaun, G. 41981) J. Am. Chem. Soc. 10'3,907-914. 17 Glasfdd+ A~ Zbinden, P., Dnb|er, M,, Bermer+ S,A. and Dunitz, J.D. 11988)J. Am. Chem. See. 110. 5152-5157. 18 Panhasarathy, R. and Fridey, S.M.. 11984)Science 226. 969-971. 19 Saenger, W_ (¢&) 41984l Principles of Nucleic Acid Structure Springer. New York. 20 Fccney, $,. Bi,d,'iall.B., RobeJrts,G-C.K- and Burgen, A.S.V. (1975) Nature 257. 564-566. 21 Hyde+ E.I,. Birdsalh IL Roberts, G.C,K., Feeney, J. and Burgen, A.S.V. (19801 Biochemistry 19. 3746-3754. 22 Williams, T.J., Zens, A.P., Wisowaty, J.C.. Fisher. R.R., Dunlap, R.B.. Brysan. T.A. and Ellis, P.D. 11976)Arch. Biochem. Biophys. 172. 490-50L 23 Sarma, ILH. and Mynolt, R.J. 11973) J. Am. Chem. Soc. 9.5. 1641-1649. 24 Satma, R.H., Lee, C-H., Evans, F.£.. Yathlndra. N. and Sundaralingam. M. 11974)J. Am. Chem. Sos:. 96. 7337 7348. 25 Birdsa]l, B.. Birdsall, NJ.M., Feeaey, I. and Thornton, J. (1975) J. Am. Chore. Six:. 97, 2845-2850. 26 De Leenw, A.A.M. and Altona, C. (1982) J. Chem. Soc. Perkin It, 375-384. 27 De Lecuw, F.A.A.M. and Ahona, C. (1983) J. Comp. Chem. 4, d25~137. 28 Rapp. J~ Van Boom+ J.H, Van Lieshout, H.C. and Hansnoot, C.A.G. (1988}J+ Am. Chem. Soc. 110. 2736-2743. 29 Mellema, J.-H., Reters, J.M.L+, Van der Marel, G., Van Boom, J.H., Hansnoot, C.G.A. and Altona. C. (t984) Eur. J. Biochem. 143, 2~15-301. 30 Allona, C. 0982) Recueil 101. 413-433_ 31 G0~enstein, D.G. led.) 11984) Phosphorns-31 NMR, Principles and Applications, AeademJnPress, Orlando. 32 gitchig, R.G.$. and Perlin, A.S. 119"/71Carbohydr. Rcs..55, 121128. 33 Barfield, M.+ Dean, A.M., Fallick, CJ., Spear, RJ., Stemhell, S. and Westerman, P.W. (1975) J. Am. Chem. Snc_ 97, 1482-1492, 34 Sarma, R,H,, MynoU. R.J.. Wood. D.J. and Hruska, F,E. t]973) J. Am+ Chem, Soc, 95, 6457-6459+ 3.5 Hofmann, HJ. and Cirimaglia. R. {198B)FEBS Lett 241, 38-0,0. 36 Tmpp, J, and Redfield, A.G. 0980) J. Am. Chem. Soc+ 102, 534-538. 37 Fischer, P, Fleekenstein, J. and Hbnes, J. 11988) Pholcchem. Photobiol. 47, 193-199. 38 H~gle~,A.'f,. Litson, S. and Da,bcr, P. 11979)L Am, Chem. Soc. lOl, 5122-5130. 39 Slewazt, J./.P. 11987)Quantum Chemistry Program Eachange No. 45.5. version4.0. dO Blumenstein, M. and Raftery, M.A. 11972) Biochemistry It, 1643.