Fluorescence Studies of Nucleotides and Polynucleotides

Fluorescence properties of nucleotides, polynucleotides and phosphorylated derivatives as function of temperature, pH and ionic strength

With RNA polymerase deazanebularin triphosphate acts as an ambiguous substrate in that it functions, with almost equal efficiency, as an analogue of either ATP or GTP.
The relevance of this observation to the mechanism of base selection by RNA polymerase is discussed.
The synthesis, characterization, and properties of polymers containing deazanebularin are also described. Some of these polymers form unique ordered structures with base pairs possessing only a single hydrogen bond; these permit an assessment of the contribution of hydrogen bonding to helix stability.
These polymers also exhibit some unusual fluorescence properties, including excited state energy transfer.
In a preceding communication (I) TX have described the fluorescence properties of three purine nucleoside analogues, and of their nucleotide and their polynuclcotide derivatives. These compounds were &own to be fluorescent at physiological conditions of temperature, solvent, and pH, and some applicatiolns of t,he fluorescence propelties to the study of conformational trxnsitioll;; in polynucleoticles were presented.
In this paper xve outline t,he characteristics of a fourth fluorescent, purine ribollucleosidc analogue, 7-deazauebularin (DW) June 7, 19 '1) ( Fig. 1). Like formycin, Z-aminopurine, and 2,6-diamino--. purine, DN 1s fluorescent at ambient temperature and in physiological solutions at neutral pH; the spectral and fluorescence characteristics of DN are therefore of interest and these are presented below.
However, there are several additional reasons \vhy UN is noteworthy.
The first concerns the ability of DN to engage in aberrant base pairing with complementary residues in polynucleotides during replication reactions. 1iq)ection of the DN structure shows that this analogue should be capable of forming at most a single hydrogen bond with thymidine or urscil in Dhe course of ordinary base pairing of the Watson-Crick type (see Fig. 2) ; no hydrogen bonds should be formed Tvith cytosine residues. We have observed that DN is ambiguous during transcription in the sense that the corresponding 5'.ribonucleoside triphosphate functions as an analogue of &her llTP or GTP.
We have also isolated pal>-mers containing various proportions of DN, including a homopolymer and an alternating copolymer of DN and uridine.
Some of these polymers form ordered structures which arc unique in that t,hcy yosscss bnse pairs containing only a si?zgZe hydrogen bond. These findings are also given belolv.
Finally, DN is of interest because it is very cytotosic to mammalian cells in culture and its metabolism in cultured cells shall-,5 several unusual properties.
The description of these characteristics of DN 1s the subject of the accompanying report. MATERIALS AND METHODS
The 5'-monophosphates of deazanebularin and the pyrimidine nucleosides were prepared by the POCI~ method previously described (3).
[@P]Nucleotides were prepared by the same procedure with [32P]phosphorous oxychloride obtained from New England Nuclear Corp. The corresponding di-and triphosphates were chemically synthesized by the methods of Michelson (4) and Smith and Khorana (5). Commercially available nonradioactive nucleotides were purchased from P-L Laboratories, Wlwaukee, Wise. Radioactive di-and triphosphates were obtained from Schwarz BioResearch Inc., Orangeburg, N. Y.
Escherichia coli B RNA polymerase was prepared and assayed according to the method of Chamberlin and Berg (6). The enzyme fraction eluted from DEAE-cellulose was used throughout; the specific activity of different preparations was in the range 2000 to 6000 units per mg of protein.
Polynucleotide phosphorylase was isolated from E. coli B as previously described (3). Bovine pancreatic RNase, bacterial alkaline phosphatase, venom phosphodiesterase, spleen phosphodiesterase, micrococcal nuclease, and pancreatic DNase were purchased from Worthington Biochemicals, Freehold, N. J. Ribonuclease Tz was the generous gift of Dr. Herbert Sober.
Routine uncorrected fluorescence analyses were done at a temperature of 20 f 0.5" on a Hitachi MPF-2A recording spectrofluorimeter.
Corrected excitation spectra were obtained on the spectrofluorimeter described by Stryer (7). We thank Dr. L. Stryer for the use of this instrument.
Optical rotatory dispersion spectra were obtained on a Cary 60 spectropolarimeter in a thermostated, water-jacketed cell with a 1 -cm light path. The observed rotations, after correction for solvent blanks, are expressed in terms of the molar rotation (~1') as described by T'so et al. (8).
The conditions used for synthesizing the various alternating ribocopolymers were as follows. Each l-ml reaction contained  ([""PI or [3H]) at specific activities of lo6 to 10' cpm per pmole; and one of the following nucleoside triphosphates at 0.4 mM: UTP, BUTP, rTTP, or *TP; RNA polymerase, 5 to 15 units.
The reactions were incubated at 37" and t,he extent of synthesis determined by measuring the acid-precipitable radioactivity in 50.~1 aliquots on Millipore filters (9). For the large scale synthesis of poly[r(DN-U)] the reaction conditions were modified as follows.
The reaction mixtures (20 ml) contained only 0.2 rnM triphosphates and they were incubated at 10". (An alternative procedure for the synthesis of poly[r( DN-U)] was to incubate the reactions at 15" in the presence of 2 to 4 x lo-* M spermine.) Aliquots were removed at approximately X-hour intervals until the rate of polymerization started to plateau (48 to 72 hours).
Purified pancreatic DNase (lo), 25 pg, was then added and the incubation continued for a further hour. The reaction mixture was deproteinized by shaking with an equal volume of redistilled phenol.
The aqueous layer was concentrated by vacuum dialysis to approximately 3 ml and the high molecular weight polymer fraction isolated by gel filtration on a column (50 x 2 cm) of Seyhadex G-100 with 0.05 I\I Tris-HCl, pH 7.9, as the eluting buffer. The material which emerged at the void volume was used for subsequent experiments.
The homopolymers of deazanebularin, uridine, adenosine, and 5.bromouridine were synthesized with E. coli polynucleotide phosphorylase using the following reaction conditions: Tris-HCl buffer, pH 8.3, 0.1 M; MgCIZ, 0.03 X; EDTA, pH 8.3, 3 rnM; the appropriate nucleoside diphosphate, 0.05 l"r; and approximately 1 to 2 units of enzyme per ml. The synthesis of nonradioactive polymers was followed by measuring the release of inorganic phosphate calorimetrically (11). Radioactive polymer synthesis was monitored by acid precipitation on Millipore filters. Radioactive random copolymers of deazanebularin and guanosine or deazanebularin and adenosine were prepared as above. Although the nucleoside diphosphate ratios were varied the total nucleotide concentration was maintained at 0.05 &I. The base composition of these copolymers was determined directly by double labeling techniques with t3H]DNDP and [W]ADP or [14C]GDP (specific activities of 2 to 5 PCi per pmole were used routinely).
Enzymatic degradation of the synthetic polymers was monitored at 25" with one or more of the following procedures: (a) measuring the decrease in acid precipitability of radioactive polymers; (b) determining the increase in fluorescence intensity and changes in energy transfer efficiency upon digestion; and (c) measuring the increase in ultraviolet absorpt'ion.
The reactions (0.25 to 1.0 ml) were incubated (with and without 0.02 M MgC&) in the following buffers: for RNase T1 and pancreatic RNase, 0.1 M sodium acetate buffer, pH 5.5; for spleen phosphodiesterase, 0.1 M Tris-HCl buffer, pH 7.9; for venom phosphodiesterase, 0.1 M Tris-HCl buffer, pH 7.0. The enzyme and polymer concentrations used in any given experiment are shown under "Results." spectra arc practically identical; comparable superposition of these +l)cctra was found for the corresponding 5'.mono-, di-, and tripho&ates.
The spectral identity and the homogeneous chromatographie behavior of these compounds suggest that they are free from fluorescent impurities.
Although the main absorption profile of l)N, lvith X ,,,aX at 270 nm, is similar to that of normal nucleic acid constituents, its spectra are noteworthy in t\vo respects.
First, the value of its molar extinction coefficient is extremely low, 3800;2 in addition there is an unusually broad tailing of the absorption spectrum toward longer wave lengths in acidic solution (Fig. 4). The wave length of maximum absorpt'ion of DN (270 nm) is closer to that of the normal 2 The extinction coelficient of deazanebularin reported in the paper of Gerster et ~1. (2) was misprinted. Fro. 5. The relative fluorescence mtensity of 7-deazanebularin as a function of pH. The nucleoside concentration was 1.3 X 10-L M. The excitation wave length was 290 nm; fluorescence emission was monitored at 400 nm. The buffers used were described previously (1). nucleotides than was found for the other fluorescent purine analogues we have studied.
However, selective excitation of the fluorophore in the presence of a large excess of normal bases can still be achieved by exciting at longer xvave lengths (up to 300 nm) ; in this case, owing to the lower extinction coefficient, the fluorescence intensity is appreciably reduced.
In contrast, the emission maximum occurs at a longer wave length (400 nm) than observed with the other analogues.
The effect of pH on the fluorescence intensity of deazanebularin is shown in Fig. 5. This fluorimet.ric pH titration yields excited state pK values of 4.2 and 11.8. The pK for protonation in the ground state, obtained by spectrophotometric titration is 4.3.
Quantum Yield Values-The quantum yields of DN and of its 5'.phosphorylated derivatives were determined according to the method of Parker and Rees (12) and are given in Table I 6. The effect of NaCl (0) and MnC12 (0) on the fluorescence intensity of 7-deazanebularin.
The solvent is 0.01 M Tris-HCl buffer, pH 7.9. Nucleoside concentration, excitat,ion, and emission wave lengths were as stated for 7. The relative fluorescence intensity of 7-deazanebularin in water as a function of temperature.
Nucleoside concentration, excitation, and emission wave lengths were as in Fig. 5. The fluorescence intensity at 20" was arbitrarily taken as 100%. changes in quantum yield. These changes are qualitative13 similar to those which were observed with the formgcin, 2-aminopurine, and 2,6-diaminopurine nucleotides (I). The fluorescence intensity of deazanebularin compounds decreases as a function of increasing ionic strength (Fig. 6). Salts such as sodium phosphate, Tris, IiCl, l\lgCIZ, and sodium citrate quench to the same degree as NaCl.
The effect of ionic st'rength on the fluorescence of previously studied nucleosides (1) was less pronounced.
Excited Rate Lifetime-The value for the excited state lifetime of DN in water was determined by Dr. L. Stryer and was found to be 6.0 nsec.
Eflect oj Temperalure-The fluorescence intensity of DN in water decreases with increasing temperature (Fig. 7), but there is no change in the absorption or emission spectra. The absorbance is minimally affected by heating, showing that the decrease in fluorescence reflects changes in quantum yield. These effects of changing temperature are fully reversible, and apply also to the 5'-phosphorylated derivatives of DN. Eflect of Solvent-Examination of solvent effects on the fluoresence of DN were restricted by the limited solubility of the nucleoside and its derivatives in nonaqueous media. Studies of DN fluorescence in alcohol-water mixtures gave the following results (Table II).
Increasing concentration of ethanol (or methanol) and propylene glycol cause progressive and small increases in quantum yield, without detectable changes in the excitation spectrum.
In absolute ethanol, or methanol, there was a decrease in fluorescence intensity, whereas in propylene glycol there was a 4-fold increase. Under these anhydrous conditions the emission spectrum was shifted by approximately 10 nm toward the blue.
In summary, the fluorescence properties of DN are qualitatively very similar to those of the formycin, 2-aminopurine riboside, and 2,6-diaminopurine riboside.
The quantum yield of fluorescence of DN (0.08) and the excited state lifetime (6 nsec) are in a range which is convenient for several types of fluorescence studies, including polarization.
In its response to changes in solvent, temperature, pH, and ionic strength, the fluorescence of DN recalls that of the purine ribonucleoside analogues previously described. Il. c'. Wad and E. Reich 709 several tcmplb3. DNMP is incorporated into RK;i with both natural and synthetic DT\TA templates, and wit,h native as well as denatured DNA (Tables Ill to VI).
The data in Table  III show that DKTP can substitute for either XTP or GTP (but not for CTP or GTP) in reactions directed by RN,\ polymerase. It is of interest that DNTP replaces llTP and GTP (but not both) IJ-ith almost equal effect.iveness in reactions requiring four nncleotides (Table III). However, the total amount of lt?jA product which can be formed in t,his way is only a small fraction of that produced \\-ith the normal substrates (1 to 47;) although the reaction is linear for up to 45 min. The polymerizat'ions WCJX performed with both radioactive and nonradioactive DNTP, and t'he results of these and of the appropriate control experiments show clearly that DNTP is incorporated into RNI\ by RXX polymerase and that its coding behavior with respect to the DNh template is ambiguous.
In reactiolzs directed by natural DE&, the requirement for tem- plate, the need for three additional nuclcoside triphosphates, and the sensitivity to actinomycin all indicate that the incorporation of DNMP into RNA is mediated by the normal catalytic function of RKX polymerase, and not by some contnmillating enzymatic activity.
The results of a more quantitative test of the ability of DXTP to substitute for either -iTP or GTP are illustrated in Fig. S. In these experiments native calf thymus DNA was the template, and DNTP replaced, respectively, ATP (Curves B and B) or GTP (Curves C and U). The progress of RNA synthesis \yas monitored by following the incorporation of two radioactive nucleotides, namely [w~~P]DNTP and [3H]CTP. When DNTP is substituting for .Yl'P, somewhat more DNMP is incorporated into product than CMP.
In fact, the ratio DNMP:CilIP in the newly formed RN:1 is 1.27, which is exactly equal to the ratio dTRlI':dGlLIP in the template. The RNA product therefore faithfully reflects the complementary base composition of the template, just as it doe? IT-hen all four normal triphosphate subskates are used, and DAMI' is being polymerized at precisely  the same relative frequency a:: would be expected for AMP. Conversely, when DNTP is replacing GTE', the ratio DNMP: CXP = 1.0 in the product coincides exactly with that required both by the relative content of dG;\II' and dCIllP in the tem-plate, and by the ratio expected in RNA if I)N?\'lP were really being incorporated at the sites which would ordinarily be occupied by GPVIP. We conclude that DNTI' does indeed show coding ambiguities during transcription irk vitro and can serve as a purine analogue of either ATP or GTP in the RNA polymerase reaction.
Despite this ambiguity, the relative utilization of DNTP corresponds exactly with that expected for the nucleotide it is replacing.
This shows that DNTP is not incorporated at significant levels in the presence of one of its normal counterparts and cannot compete effectively with them. That is, when DNTP is substituted for GTP, and hTP is present in the reaction mixture, DNTP is functioning and being utilized only as a ~-GTP analogue; DNMP is not being incorporated detectably at both AMP and GMP sites. The failure of I)NTP to compete efficiently with normal triphosphates is seen also in the reactions directed by synthetic templates, t'he characteristics of which are presented below. DNTP is not utilized by RN.A polymerase in the absence of a normal substrate.3 This is seen in the homopolymer reactions directed by denatured DNA (Table IV), poly(dT) ( Table V), poly(dC) ( Table Vl), poly(rC) and poly(rU).
In all of these cases, no detectable acid-prccipitable product is formed in the presence of DNTP alone; no experiments were performed to test for the formation of short oligonucleotides in these reactions. However, in poly(dT)-and poly (dC (Fig. 9). With this template, relatively substantial and prolonged synthesis occurs when DNTP serves as the only "pm-me" nucleotide substrate.
The K, value found for DNTP (1.0 x 1 O-4 M) does not differ markedly from that of ATP (6.6 x 1O-5 M) xvith the same template.
No K, value has been dctermincd for DNTI' when it is acting as an analogue of GTP.
The synthesis with poly[d(;l-T)] template of alternating copolymers containing 1)X and a pyrimidine was studied in some detail.
Analysis of the RKX product using radioactive double labeling techniques reveal that DN and uridine are incorporated at equal rates as is required by the base composition of the template. In addition, alkaline hydrolysis of the product synthesized with s-321-'-labeled 1)NTP (and iiornadioactive UTP) as substrate show quantitative transfer of [321'] t)o U>lP (Table  VII).
The reciprocal experiment using [cr-32P]UTP shows quantitative label transfer to DNi\Il'.
These facts show that the copolymer contains a perfectly alternating sequence of DK and uridine residues. In this case, DN is replacing adenosine during the synthesis and in the structure of the product.
Similar results were obtained from the analysis of poly[r(DN-\k)] (Table  VII).
With poly[d(iZ-T)] template the incorporation of DNMP in the presence of XT1 is very poor; even with an input ratio DNTP:ATP = 10 the incorporation ratio is at best I>NMP: AMP = 0.05. Thus the DN content of alternating copolymers containing both DN and adenosine has been confined to the range of 0.1 to 5$;. Slthough the RNA product fait'hfully reflects the template compositiolr aild sequence t,he kinetics of synthesis are unusual in several respects.
I. At 37", the polymerization of UMI' and 1)NMP proceeds at a reasonable rate for about 1 hour. of the reaction the rate is al)prosimatcly one-tenth that achieved with AMP and UMP mider identical conditions. The reaction then stops abruptly although only a small fractioli of the inlmt nucleotide has been polymerized.
In contrast, the synthesis of poly[r(A-U)] continues at a rapid rate for extended periods (up to 12 hours) until more than 5091 of the substrate has bec~l utilized.
Holvever, with iI!TP as the pyrimidine substrate polymerization proceeded at the initial rate for prolonged periods (Pig. 10).
3. The arrest in synthesis at, 37" which occurred typically  after npl)rosimatcly 60 min could be eliminated by the addit,ion of >permil~c (Fig. 11). In the presence of the polyamine the kinetics of synthesis with all pyrimidincs resembled that which n-as fount1 with WY'. This effect of spermine was seen only in a very narrow range of concentrations IT-hich was somewhat variable in tliffercnt eq)eriments; the effective concentration of spermine could be established only by performing a titration curve durilrg each set of experiments.
In spite of systematic analysis of the cspcrimental variable-; no esplanat~ion was found for the incollstnlley in these obscurations. The effect of sper-mine could not be duplicated by addition of KC1 or NaCl (0.05 to 0.30 M) or b,-JIgCl, and i\InCls (0.002 to 0.1 M), although a wide range of concentrations was tested.

A prolonged, but slow synthesis of poly[r(DN-U)]
could also be achieved in the absence of spermine by lowering the temperature of incubation to 10". With this low rate of polymerization, the reaction typically proceeded for as long as 48 or 72 hours, by which time up to 107; of the nucleotide had been incorporated into polymer. These low temperature incubations were routinely used for the isolation of poly[r(DN-U)] (see "Materials and Methods"). 5. For reasons which we do not understand, great difficulties were always encountered lx-hen syntheses were attempted on a large scale, even when all the reagents and reaction conditions had been proportionately adjusted. One possible interpretation of these unusual kinetic properties is presented below (see "Discussion").
Polynucleotide Phosphorylase-As shown in Table  VIII, DNDP is an effective substrate for E. coli polynucleotide phosphorylase.
Since DNDP is also efficiently lltilized by the enzyme when the normal nucleoside diphosphates are being polymerized, mixed polymers can be prepared containing any desired proportion of DN residues, in addition to the homopolymer, poly(DN) (see "LR/Iaterials and Methods").

Alternating Copolymer of Deazanebularin and Uricline: poly-[r(DN-U)]-Poly[r(Dr-G)]
TX-as isolated from a large scale incubation as described under "Materials and hlethods." The minimum molecular weight, judged by gel filtration and end group analysis, was in excess of 75,000. As isolated (in 0.05 M Tris-HCl buffer, pII 7.9) poly[r(DN-U)] possesses none of the spectral characteristics of multistrandedness.
No sharp or reversible optical transitions are observed on heating to 70'; however, a variable and substantial number of chain breaks are introduced during the heating cycle. These are manifest as slight irreversible increments in fluorescence and absorption, and as corresponding decreases in acid precipitability of the polymer.

Since poly[r(Dn'-U)] is produced under the direction of a poly[d(A-T)]
template, it appeared likely that some sort of base pairing should occur betlveen D?r' and a complementary pyrimidine. The fluorescence and ultraviolet absorption of the polymer were used to monitor possible physical transitions which might point to the formation of base pairs and structural order. The presence of II&+, even in low concentrations (2 X 10W4 M) gives rise to all the spectral characteristics associated with base pairing and double strandedness.
Thus, addition of MgClt instantly produces a significant (25%) and reproducible degree of hypochromism at 260 nm. This call be seen in the spectra which are presented in Fig. 12. When the polymer is denatured (45") the ultraviolet absorption profile is the same as that lvhich is obtained at lower temperatures in the absence of stabilizing amounts of Mg++.
Heating the polymer solution in the presence of ;\Ig++ produces a characteristic melting curve (Fig. 13). This optical transition is sharp, cooperative, and reversible. In addition, it occurs over a narrow range of temperatures (5-S"), just as in the case of the double stranded poly[r(h-U)].
The midpoint of the thermal t,ransition (T,) is a function of the hlg++ concentration (Fig. 14). Similar T, profiles can be observed with the addition of NaCl or KCI, but the concentra- at Xmx (205 nm  The spectra were taken on a Hitachi MPF-2A spectrophotofluorimeter.
above and below the region of thermal transition.
Above the T, the excitation spectrum corresponds to that of the DNMP monomer, whereas below the T,, the excitation spectrum is quite different and provides evidence for energy transfer from uracil to the DN residues (Fig. 15). 2. The kinetics of nuclease digestion of poly[r(DN-U)], seen in Fig. 16  ( Fig. 17) show that the addition of Mg ++ ions results in an increased molar rotation which is indicative of an increased structural organizatiolr. The spectral changes induced by MgCl:! can be completely reve~~~cl by heating; the variation in rotational strength as a funct'ion of temperature exhibits a profile typical of polynucleotide dcnaturation. Additional considerations relating to the structure of poly[r(DT\;-U)] arc presented under "Discussion." Poly(rDN) and Random Copolymers Containing DN--l-'oly-(rDN) was prepared with polynucleotide phosphorylase under the reaction conditions described under "Materials and Methods." The minimum molecular weight, as judged by Sephadex gel filtration was 40,000. End group analysis of [3H]p~ly(rDN) gave an average chain length of 160, which corresponds to an average molecular weight of 56,000. The ultraviolet absorption spectrum of the polymer at neutral pH is almost identical with that of the corresponding monomer spectrum (see Fig. 4).

Poly(rDE)
is also virtually devoid of hypochromicity relative to the nucleoside; the estinctioll coefficient (per mole of phosphate) at A,,,,, being 3600 compared with 3800 for l)Y, illdicating little, if any, structural order for single stranded poly(rl)X).
Attempts were made to prepare homopolymer pair complexes 7 of poly(rl)rU) with either poly(rIY) or poly(rl3rG).
Xo convincing evidence was obtained for the formation of such complexes at room. temperature by either absorption or fluorescence mcasurements although equimolar amounts of polymers were mixed in solutions of high ionic strength (up to 3.0 nr KC1 and SaCl or 0.02 BI RlgC12; MgCIZ concentrations higher than 0.02 M cause precipitation of poly(rDN)). The effect of DIX on the stability of double stranded RNA hclices was therefore studied by using random copolymers containing DN and either adenosine, guanosine, or cytidine in various base ratios.
A detailed spectral and fluorimetric study of the polymers will be presented elselvhcre,4 however, two points deserve mention here. As shown in Fig.  18, the T, values of poly[r(DN, A)] + poly(lT) complexes are considerably lower than the T,, of poly(r12 .rU), although a relntively small amount of DN is present in these polymers.
The degree of hyperchromicity observed on denaturation and the cooperativeness of the melting profiles are also considerably rcduced. The fact that denaturation curves obtained fluorimetritally yield the same T,, and occur over the same temperature range as the absorption measurements suggests that there is no selective early denaturation of the DN-containing regions. The second point of interest is that deazallebularin-adcnosine copolymers exhibit pronounced energy transfer (14, 15) from $15 adenosine to deaznncbularin residues (Fig. 19). Xo such energy transfer is seen in tleazalrehularirr-guatlosille or deazancbularincytidine copolymers.
Some of the fluorescence properties of thrse polymers are presented below.
LVzlcZease S'?tsceptibiZit~ of DN I-'olllmers-PolS[r(nN-IT) ] is susceptible to degradation b!-all the enzymes tested (Table IX, Fig. lci), although the rate of degradation is, in every case, slower in the presence of added i\IgCIZ. The decreased nuclease susceptibility in MgCls is most likely a consequence of an increase in the double stranded character of poly[r(DN-U)]; the enzymes used preferentially degrade single stranded polymers.
Complete tlegradation of poly[r(DN-U)] by RSasc T? results in a 5.fold increase in fluorescence intensity and yields the expected mononucleotides on electrophoresis at 1113 3.5. Limit digests with pancreatic RiYase or with either of the phosphodiesterases renders the polymer totally acid soluble but gives only a 3.fold increase in fluorescence.
The full j-fold increase in fluorescence ih obtained only after addition of RXase TZ to the digests. Since the major product of these enzyme reactions, prior to RNase T2 addition, has the Same clectrophoretic mobility at pH 3.5 as ,4pIip, it is assumed that the end products in these limit digests arc dinucleotides.
-1lthough D& pTTp is the expected product of pancreatic R-l~ase digestion, t'he results suggest that the phosphocliesterases cleave certain 17)h -containing di~~ucleotidcs slowly, if at' all.
The four nuclenscs &scribed above were also tested for their ability to tlegradc I)oly(rDX).
The polymer was resistant to pancreatic Ri\'ase and susceptible to RNase Ta as anticipated from the known substrate specificity of these enzymes. Poly-(rDN) was, however, unexpectedly resistant to digestion by both qleen xnd venom phosphodiesterases.
High enzyme concentrations (50 pg per ml) and extended incubation times (up to 24 hours) were required to obtain partial h-glrolysis of the 1)olgmer. 111 contrast, ai1 identical couccntratio1r of poly(r,l) was totally depolgmerized in 10 min when incubated with 10 g per ml of these ~)hosl)~loclicstcr~~scs.
Fluoresce?lce Properties of DN Polgl,lers-The fluorescence properties of poly(rI)N) resemblr qualitatively those of other fluorescent polynucleotides.
The excitation and emission spectra are the same as those of the monomer (see Fig. 3), but the fluorescence intensity is decreased to one-fifth that of DNMP. This degree of quenching is much less than that observed with other polynucleotide fluorophorcxs and can be correlated with the low hypochromicity of the polymer. Both of these observat,ions suggest that poly(rDN) has \-cry little structural order in the single stranded state.
The fluorescence of copolymers of 1)X and adenosine differs from that of poly(rDX).
The fluoresccncc emission spectrum is virtually identical with that of the monomer; however, the escitation spectrum is modified in two ways. Compared with poly(rDN) and l)N111', the lvavc length of maximum excitation is shifted by up to 10 nm toward the blur.
The copolymers t,hat contain high adPuosinc to Dh ratios eshibit the largest spectral changes (Fig. 19). Since the X,,,, of adenosine and of DN also differ by 10 nm (260 nm for adenosiue, 270 nm for DN) it is likely that energy transfer is taking Illace between the AMP and D~MI r&dues in the polymer. A second change in the excitation spectrum is the appearance of a prominent shoulder at npproximatcly 300 am, which has no counterpart in the corre-;;ponding absorption spectrum.
It is likely, therefore, that this shoulder represents a transition involving only residues which

716
Fluorescet~ce hb&'es of Nucleoticles and Polyrmcleotides. II Vol. 247,so. 3 are in the excited state; this could represent a charge transfer complex of adenosine and DN in the excited state. Treatment of these copolymers with RXase Tz abolishes both of these spectral characteristics; the resultant excitation spectrum is identical with that of Di\r monomers.
The addition of poly(U) to poly[r(DX, A)] under conditions which lead to the formation of a 1:l ordered homopolymer pair structure also rapidly converts the excitation spectrum to one resembling that of poly (rDN) and reduces the fluorescence intensity to approximately one-fifth that of the single stranded copolymer.
When the newly formed double stranded structure is heated, a fluorescent Z', is obtained and the excitation spectrum reverts to that of single stranded poly[r(DN, A)]. However, the maximal fluorescence intensity after melting is only one-half the r&e found for the copolymer in the absence of an equal concentration of poly(U). The fluorescent T,,, which decreases with increasing 1)N content of the copolymer is identical with that obtained by monitoring the ultraviolet absorption (Fig. 18).
The fluorescence emission and excitation spectra of copolymers containing DN and guanosine are qualitatively the same as those of poly (DN) and DNMP. There appears to be no detectable energy transfer between these two nucleosides.
The fluorescence properties of the alternating copolymer poly-[r(DN-U)] differ from those of the other polymers containing DN.
In t,he single stranded state (i.e. in the presence of Mg++ at temperatures above Z',), the excitation and emission spectra are like those of DNMP, although the fluorescence intensity is quenched to one-fourth that of the monomer.
However, in the ordered form, at temperatures below T,, the excitation spectrum is altered so that it resembles qualitatively that of the single stranded random copolymer, poly[r(DN, A)]. That is, compared with the monomer or with the denatured polymer, the wave length of maximum excitation of double stranded poly- [r(m-U)] is shifted by about 10 nm toward the blue, and a prominent shoulder appears at approximately 300 nm (Fig. 15). Since the X,,, for ultraviolet absorption of uridine is at 260 nm, whereas that of DN is at 270 am the blue shift in the excitation spectrum suggests that energy transfer is occurring between uridine and DN.
As in the case of poly[r(DN, A)] the shoulder in the activation spectrum at 300 nm is not matched by any corresponding change in the absorption spectrum and therefore probably represents a property only of residues in the excited state. This change in the shape of the excitation spectrum which is a property of double strandedness, is associated also with a change in the emission spectrum (Fig. 15). All these facts suggest t,hat the native polymer contains an emitting species which is absent in DNXP, and in single stranded poly[r(DN-U)]. The position of the 300.am shoulder in the excitation spectrum could be consistent with the existence of a charge transfer complex in the excited state.
As mentioned earlier, poly[r(DN, A)] exhibits energy transfer in the single stranded state but not when complexed with poly(U).
Poly[r(DN,U)], on the other hand, exhibits energy transfer only in the double stranded form, These observations taken together imply that the energy transfer which is observed in the copolymer poly[r(DN-U)] (see below) is the product of the interaction between m and the adjacent uracil residues in the same strand. DISCUSSION Many of the observations described in this paper arc relevant for problems of polynucleotide synthesis and structure, and these require comment and encourage speculation.
Coding in Transcription and Interaction with RWA Polymernse -The data in Figs. 8 to 10 and Tables III to VI leave little doubt that DNTP is an ambiguous substrate for transcript,ion, that it functions as an analogue of either ATP or GTP, and with almost equal efficiency in the two cases. This finding is significant for theories concerning the mechanism of base selection by RNA polymerase.
It is important to note that aside from the related structure, nebularin,5 DN is the only base analogue with demonstrable coding ambiguity in transcription. Although the mutagenicity of 2-aminopurinc has often been considered to result from pairin, 01 ambiguities of the kind obser\Ted here (16, 17) no detectable evidence of such behavior has been obtained in vitro for 2-aminopurine n-ith either DNA polymerase or RNA polymerase.jr 6 Thus the obvious ambiguity of Dx is a phenomenon which merits some consideration.
The ability of DNTP to replace ATP is not altogether surprising for the following reasons. (a) DN functions as an analogue of adenine with several enzymes that do not require templatev; (b) DN possesses one of the characteristic hydrogenbonding positions of ndenine, namely, the equivalent of the unprotonated N-l of adenine. DN can easily be visualized as forming a complementary base pair with urncil or its derivatives (see Fig. 2). This base pair would possess only a single hydrogen bond but could otherwise conform to all the geometrical requirements of a base pair of Dhe Watson-Crick type. At first sight, the ability of DNTP to replace GTP appears difficult to understand on the basis of the proposal that RXd polymerase selects the incoming nucleotide through some "allosteric instruction" from the template. DN possesses none of the functional groups characteristic of guanine, and specifically lacks all of the specific hydrogen-bonding functions ofanosine; in addition the chemical and physical properties of DN bear no similarity at all to those of guanosine.
The function of DXTP as a GTP analogue during transcription would thus seem to require some other explanation.
The incorporation of DNRIP in place of GMP can best be rationalized in one of three ways. The first is to imagine that DN can form a base pair with the rare imino form of cytidine. Under these conditions the base pair should contain a single hydrogen bond, which would make it structurally similar to the DN-U base pair. We regard this as unlikely, because of the rarity of the required cytosine tautomer and because the tautomer in question should also be capable of forming a base pair containing two hydrogen bonds with adenine.
Thus ATP would be expected to replace GTP as effectively as DNTP does, and this is not the case. A second possibility is that DN is forming a base pair with the protonated form of cytosine; again this base pair would contain a single hydrogen bond. This is unlikely for at least two reasons: (a) the pK of cytidine in polymers (5.8 to 7.4) (18) is too low to permit adequate protonation at '71s of only a single hydrogen bond, namely, an increased tendency to rotation at the glycosyl bond. This can be anticipated because two coplanar hydrogen bonds per base pair would tend to fis a plane in a cooperative fashion, which is not possible with a single hydrogen bond.
Finally, the structure proposed for poly[r(DN-U)] is also consistent with the synthetic patterns observed in the RNA polymerase reaction.
In t,hc absence of spermine, the newly formed poly[r(DN-U)] would be single stranded. Bs shown by Krakow (25) single st,randed RN;\ quickly inhibits RNh polymerase, and the same would be expected of poly[r(DN-U)].
The addition of spermine would tend to maintain the product as a double stranded structure thereby pcrmitt,ing continued synthesis as occurs normally with poly[r(A-U)].
The fact that no spermine is needed for the synthesis of poly[r(DN-*)I, is also consistent with the known pattern of thermostability of base pairs which contain uracil and its analogues: base pairs containing pseudouridine are very thermostable (26), and presumably poly[r(DNq)] remains in the helical form even without sperminc.
Fluorescence of Polymers Containing DX-The general fluorescence properties of DN are qualitatively similar to those of other fluorescent nuclcotides previously described (I, 27, 28). However, the fluorescence characteristics of polynucleotides containing DNMP differ appreciably in several respects from the comparable polymers containing formycin, 2-aminopurine, or fluorescent guanine derivatives.
The degree of fluorescence quenching which occurs on polymerization of DSMI is much less than that seen with formycin and 2-aminopurine, although the relative differences in the fluorescence intensity of single and double stranded polymers is approximately the same for all. It might be reasonable to consider that the limited extent of fluorescence quenching of DK in polynucleotides is due to the relatively low stability of structural order which seems to be characteristic of them. However, this explanation cannot suffice since the thermostability of poly[r(DnT-U)] in the presence of RiIg+f is in the same temperature range as that of poly-[r(2AP-U)] or poly[r(F-U)] m solutions of sodium chloride and the difference in fluorescence quenching of the different polymers persists under these conditions (1, 26). and a single pyrimidine.
The copolymers thus consist of mixtures of two base pairs of differing thermostability.
We have observed that the Z', is a perfectly linear function of t,he fraction of the respective purines (or type of base pairs). For example, the T, values of poly[dCA-T)] and poly [d(nZu-T)] fall at the extremes of a straight line given by a plot of T, of a series of mixed copolymers of the type poly[d(A, n?Pu-T)] containing varying proportions of each purine.
This makes it possible to predict the Ym of either alternating copolymer contai&g only a-single base pair by extrapolttting the line obtained from the T, values of the mixed alternating copolymers.
The T, of poly[r(DN-U)] has been calculated in this manner by extrapolating the T, of a series of mixed copolymers of the type poly[r(DN,A-U)] to a value corresponding to an adeninc content of zero. The T, of poly[r(DX-U)] given by this calculation is approximately 55" below that of poly[r(A-U)] under identical ionic conditions. Several facts suggest that the low degree of fluorescence quenching of DN polymers is an intrinsic property of Dr\: residues in both single and double stranded polynucleotides.
Thus, it appears likely that the fluorescence of DN polymers is due t,o the emission from residues throughout the polymer, and not simply from tcrmiiial residues a? it al)pears to be in polymers containing 2-aminopurine or formyciii. The most l)crsuasire fact is the relatively high fluorescence intensity of the polymer compared with that of the monomer.
To account for the emission of polymers (either single or double stranded) entirely ill terms of the fluorescence of the ends would require a quantum yield in excess of unity for the terminal residues. The behavior of the copolymers poly[r(I)N, A)] is also especially revealing i n this respect. The relative magnitude of the two main conponents of the excitation spectrum directly reflect the resljective adenine and DN content of the polymers; this would be difficult to understand if the only fluorescent' residues were DN molecules at the termini of the polymer.
The most noteworthy aspect of DN fluorescence in polymers concerns the changes in the fluorescence spectrum observed ill copolymers, and the spectral modificat,ions which occur as a function of secondary structure.
The excitation spectrum of the alternating copolymer poly[r(DN-U)] indicates that energy transfer from uridine to nx is taking place only in the doublcstranded form. In polylr(DN, A)], on the other hand, energy transfer from adenosine to DN 1s observed only in the single stranded state and is eliminated on complexing with poly(L). Taken t,ogether, these observations show that the energy transfer from uridine to DN which occurs in poly[r(DN-U)] is a COIIXquence of the interaction between Dlv and adjacent uridine residues in the same strand, and not between the bases which constitute a base ljair. -4nother conclusion which can be drawn from these findings is that energy transfer in both cases appcnrs to depend on interactions between adjacent bases in the same strand, and that the geometrical requirements for those interactions cannot be satisfied in both single and double stranded polymers.
Our preferred explanation for these effects is to visualize slight differences ill pitch between single and double stranded polynucleotide helices. This would product corresponding differences in the physical overlap of adjacent residllcs in the two types of polymer structure, and thereby influence energy transfer between them. However, them are no good grounds as yet for excluding several other, equally reasonable, interpretations.
As noted above, the changes on the blue side of the fluorescence excitation spectrum and the slight alterations in the excitation spectrum which accompany the appearance of the energy tr:ulafer band suggest that in both poly[r(DN, A)] and in poly [r(DN-U)] charge transfer complexes or "excimers" (14) rnaJ-be formed in the excited state. Further work will be required to establish or to reject this possibility with more confidence.
Applications-Several l)otent,ially useful applications of DS should be mentioned.
From its ambiguity in base pairing i-1% should be an excellent mutagen for all systems which incorl)orate it into polynucleotide.
hs shown in the accompanying report (29), DN is effect.ively incorporated into the nucleic acids of animal cells and of both DPIJX and RNB viruses. ,Lnother series of potential applications derives from its fluorescence properties; in particular, its ready incorporation into nucleoside terminus of tRX;\ can bc of special value in studying structural