J. Mol. Biol.

(1978) 122, 145-162

Sequence-specific Crosslinking Agents for Nucleic Acids Use of 6-Bromo-5,5-dimethoxyhexanohydrazide for Crosslinking Cytidine to Guanosine and Crossiinking RNA to Complementary Sequences of DNA JAMES SUMMERTOX~ Division. of Molecular and Cellular Biology National Jewkh Hospital and Research Center 3800 East Co!fax, Denver, Cal. 80206, U.S.A. A?r'D PAUL A. BARTLETT Department University

Berkeley,

of Chemistry of California

Calif. 94720, li.S.d.

(Received 6 Decem,ber 1977) A class of compounds of the form: NH2NHC0(CH,),C(0R),CHR’X has been designed to allow selective blocking of specific genetic sequences of DNA and RNA (Summerton, unpublished data). This paper describes the synthesis and use of 6-bromo-5,5-dimethoxyhexanohydrazide for such site-specific inactivation. 111 model reactions it is shown that this compound can be attached to the C-4 position of cytidine and that, after activation, the cytidine-bound agent crosslinks to the N-7 position of guanosine. This reaction sequence has been applied to the crosslinking of bacteriophage T7 RNA to its complementary DNA in a highly specific fashion. The RNA is derivatized with the hydrazide reagent, activated, and incubated under annealing conditions with the complementary DNA, resulting in crosslinks between the t,wo strands that are stable to denaturing conditions, dependent on the presence of the crosslinking agent, and specific for the complementary DNA sequence. These studies show that the title compound is a promising sequence-specific blocking agent for nucleic acids. The capability of introducing site-specific blocks in DNA and RNA in this way may have a wide variety of applications in the study of genet’ic processes. In particular, the combination of this compound with appropriate restriction fragments may enable systematic mapping and characterization of viral genomes.

1. Introduction Our objective is the development of a general met’hod for blocking specific genetic sequences, because a way of introducing sequence-specific lesions in DNA and RNA would have a wide variety of applications in the study of genetic processes. It may he useful for the study in vitro of mechanisms of replication, transcription and repair, t Present address: Department Corvallis, Ore. 97341, U.S.A.

of Biochemistry

anti Biophysics,

Oregon

State University,

14:; 0022- 28:36/7&J/1222%4562 $0:.00/O

% 1978 Academc

Press Tnc. (London)

Ltd.

I46

J. S~‘JlMEH’I’ON

.A S I) t’. A,

HAX.‘YLE’Y’l

as well as provide a, means for the *>,sterna.tic mapping of viral gcnomes. Should such an approach prove feasible irr rGr:o. it w~~dtl h;rw important applicat,ions in t ht. st,udy of gene expression and in thtx pot~~ntjial control of vira I dis(~lst5 l)y sel&ivC~ inactivation of gene tjranscription. Our approach (Summerton, unpublished dat,a) involves the attachmentJ of a bifunctional compound at multiple sites on a single-stranded polynucleotide (Carrie1 strand). This bifunctiona,l compound is designed to form a covalent crosslink when the derivatized carrier strand is annealed with a complementary polynucleotide (target strand). Such crosslinks between amcarrier and its target, should irrevcrsihly block that specitic genet’ic sequence with respect to all informat,ion transfrr proccsscxs (replication. transcription and t’ranslation). Many syshems have been developed rccent,ly for the purpose of crosslinking oligonucleotides t,o complementary structures. For the most part, these experiments have focused on the crosslinking of dcrivatized messenger RNA or transfer RNA species to ribosomal proteins and ribosomal RNA for mapping the topography of the ribosome (Bispink & Matthaei, 1973: Rreitmcbyer & Noller, 1976; Budker e:t u,Z.. 1974; Eilat et al., 1974, and references t.herein; Pongs & Lanka, 1975; Schoenmaker & Schimmel, 1974; Schwa,rtz & Ofengand. 1974: Sonnenberg et al., 1975; Wagner & Gassen. 1975. among others). These techniques are not generally adaptable to crosslinking a given oligonucleotidc sequencr to its complementary polynucleotide beca,usc they rcquirr unusual functional groups on t$hc carrier &and t.hat are specific for each system. c’urthermore, these techniques allow the introduction of only a single crosslinking agent per carrier strand. In a general approach to crosslinking tetra- penta- and hexaribonucleotides to RNA and DNA, Grineva and coworkers (Grineva & Karpova, 1973,1974; Grineva e6 al., 1972,1974.1975,1976) have investigated t’he use of nitrogen mustards of general structure i (see below). Attachment of these molecules by an acetal linkage to the 2’, 3’-hydroxyl groups at the 3’ terminus of t.he RNA oligonucleotide (ii, below) converts it to an effective alkylating agent for both RNA and DNA. However, no

il

CHO

O-P-O I -0

BOX O H 0

0

Y

(il)

DNA

AND

RNA

CROSSLINKING

r\GENTS:

TESTTNG

147

attempt has been made to apply this to an oligonucleotide long enough for discrimination between complementary and non-complementary RNA or DNA to be seen. To increase t,he crosslinking efliciency for polynucleotides. and because of the labilit’y of most’ crosslinked struct’ures and the presence of cellular repair mechanisms. we beliwe thnt multiple crosslinks will be necessary to ensure irreversible blockage of genetic sequences in complex biological systems. I$rlr,, have ta,krn advantage of the fact tha,t acylhydrazides can be substituted at the C-4 position of cytosine residues b? incorporating t’his moiety into a bifunctional crosslinking agent. of general st’ructure XH,XHCO(CH,),C(OR),CHR’S. which can be attached at multiple sites on both RNA and DNA. This paper provides the experimenta. details of the synthesis of 6-bromo-5.5dimethoxyhexanohydrazide (NH,NHCOCH,CH,CH,C(OCH,),CH,P,r) and describes model st’udies verifying the intended sequence of events in the polynucleotide crosslinking process. The compound binds covalently to the C-4 posit,ion of cytidine; the ketal moiety of the cytidinr-bound agent is cleaved by mild acid treatmt,nt, giving an activated alkylating agent : and the activated cytidine-agent complex alkylates bhe K-7 position of guanosine t,o give the adduct’ : cytidine-agent’guanosinr. as illustrated in Figure 1. That t,hr rcwtion wquenw in Figure I takes

0

CH,O OCH, $t:c;;en+*

NV”tiBr (I)

d

2!!?+re,“,4,0n

N-d Rib&e 0

Cytldme

(2) pH 2.5 I Activation

deactivation

(4) pti 13

i 0

0

HNL \ NH

‘-W (5) R-00, R’= H (61 R= H, R’=CHO

Fro. 1. Chemical scheme for t,hr crosslinking climot,hoxyhcxanohydrszide.

(101

of cytidine

to guanosine

with

6.bromo-5,5-

148 place

,I. YtiMMEH’I’0N at the polynucleotidc

attached

level

to single-stranded

and Found

irrrvcrsibly

RNA

XSI)

I’. A. H.\ltTl,~;‘l”I

is suggested

by experiments

and activated,

t,o conlplemt,nt;bI’?

and thr

in which

dcrivatizcd

the agent

ENA

is

ih annc~ai~~cl

DNA.

2. Materials and Methods (a) Synthesis of li-bromo-5,[email protected]; preparation of structure (1) (Pig. 1) Two points should be considered in t,he following synthesis. First, the quantity of diazomethane used presents a serious explosion hazard and thus should be kept behind a safety shield at all times. Second, the bromoketone intermediate shonld be handled with caution since it, passes easily through rubber gloves and mg quantities produce severe skin lesions. A 0.06-mol portion of methyl chloroformylbutyrate (Aldrich Chemical Co.) was added with stirring to 0.14 mol diazomethane in ether at room temperature. After 2 11, gaseous HBr was bubbled through the solution until t,he yellow color of tile diazoketone was dissipated. The solution was neutralized M.itll sodium bicarbonate. waslled 3 times with water, concentrated at reduced pressure, and finally distilled (b.p. II l”C/O.O5 mm Hp) to give 12 g of the bromoketone. The product was contaminated with approx. 8% of the irl tile fortnat,ion of t.hc diazokrtone). chloroket.one analog (produced as a side-product A mixture of 0.05 mol of the bromoketone, 20 ml trimethylorthoformate, 30 ml methanol, and 0.1 g p-t.oluenesulfonio acid was kept, overnight at 50°C. The sohltion was neutralized with 10 ml of saturat.etl ammonium carbonate, concent.rated at. reduced pressure, and partitioned between ether and water. After Leashing the organic phase twice with water, the ether was removed at reduced pressure atrd t,he residue was mixed wit11 2 equivalents of hydrazine and sufficient metllanol to give a homopeueous solutiott. Aftrar 18 11 at 37”C, the mixture was concentrated in ?~acuo arId triturat,ed twice wit11 &her. leaving an insoluble residue which was discarded. After evaporation of tile etller, t,hr crude product was crystallized by dissolving in warm toluene and cooling slowly. By this procedure a 50% yield of 6-bromo-5,5-dirnrtll~)xyltexar~(~~l~~~r~ide (struct,ure (1) ; m.p. 58 to 6O”C), contaminated with 8:~ of the chloro analog, was obtained. Spect,ral analysis of the compound gave the following: IR(CHC1,) cm- 1 : 3350, 3470, 3640 (C - H) ; 1670, 1630 (C = 0); n.m.r.t (CDC13) p.p.m.: 1.7 (m. 4H). 2.1 (t, 2H), 3.2 (s, 6H), 3.8 (br, 2H), 7.4 (br, 1H). The bromomethyl resonance orcurs as a singlet. at 3.4 p.p.m., while tile corresponding peak of the contaminating cllloro analog occurs at 3.5 p.p.m. The mass spectrlltn does not show a molecular ion, but the fragmentation pattern is entirely consistent wit,11 the assigned structure. Elemental analysis for C8H17 BrN,O, (269.15) ; contnminatntl \vith 8% of the chloro analog: Calc.: Vound: (b) Linking

cytidine

C, 36.26;

H, 6.47; Br. 28.57;

C, 36.27;

H, 6.50; Br, 28.59; N, 10.54.

to guanosine;

preparation

of structures

N, 10.57.

(2),

(3), and (4)

(Fig.

1)

A mixture of 1.0 g 6-bromo-5,5dimethoxyhexanollydrazide, 0.2 g cytidine and 0.4 ml water was adjusted to pH 4.15 with 90 ~1 Yly/ formic acid and incubated for 6 h at 37°C. This mixture was added dropwise wit11 stirring to 75 ml of et)her. The ether-insoluble material was fractionated by descending paprr chromatography (Whatman 3MM, 16: 5: 2 (by vol.) (solvent 1)). The product) developed with n-butanol/water/ethanol, (structure (2)), visualized with 254 nm light (RF 0.55), was eluted with water. One ml of the elutant (AZoC value of 275 at pH 13) was dried and then activated by resuspending in 0.1 ml of 0.1 M-HCl. After 90 min at 25”C, the activated material (structure (3)) was mixed with 50 ,~l citrate buffer (0.2 &l-citric acid to pH 5.5 with NaOH) and 0.85 ml formamide followed by 113 mg guanosinc. The slurry was st,irred well and incubated for 6 days at 37°C. Paper electrophoresis of the reaction mixture in 0.05 M-ammonium acetate adjusted to pH 5.5 with acetic acid revealed a new component whicll moves wit,h an estimated +1 t Abbreviations

used: n.m.r., nuclear magnetic

resonance’; p.p.m., parts prr million.

DNA

AND

RNA

CROSSLINKING

AGENTS:

TESTING

149

cIIa.rge (relative to picric acid at - 1) and exhibits a deep blue fluorescence under 254 nm light. This material was eluted and further purified by descending chromatography developed with isopropanol/l9& aqueous ammonium formate (2: 1 (v/v) (solvent 2)). The material (structure (4)) chromatographed a,s a single component with an RF of 0.23. The yield was estimated to be 7.6%, based on an c310 of 18,800 for 4-deamino-4-pentanohydrazidocytidino at pH 13 (S ummerton, unpublished data). (c) Preparation

of bacteriophage

T7 DNA

The procedures for preparation of T7 DNA were adapted from Summers & Szybalski (1968). Sterile tryptose broth (80 ml) (Difco Co.) was inoculated with Escherichia coli B and incubated overnight at 37°C with aeration. The following day the culture was added to 3.5 1 sterile tryptose broth and incubation was continued at 37°C with aeration, to an was continued. At 40 min A 550 value of 0.6. T7 phage (2 x lOI’) were added and incubation after infection, 80 g NaCl was added. After dissolution of the salt, 630 ml of 30% polyethylene glycol 6000 and 70 ml of 10% dextran sulfate 500 were added and the lysate was stored at 4°C for 18 h. The top phase was discarded and the interphase and lower phase were centrifuged for 10 min at 4’ C and 10,000 g. The interphase cake was resuspended in 14 ml of water and 2.1 ml of 3 ~-Kc1 was added. This preparation was centrifuged at 10,000 g for 10 min at 4°C. The supernatant was decanted and its density was adjusted to 1.5 g/cm3 wit11 CsCl. The T7 were banded by centrifugat,ion for 18 h at 4°C in a Beckman 5OTi angle-head rot,or at 35,000 rc,\.s/min. The T7 band was removed and dialyzed against 0.01 >I-EDTA (pH 7). (d) Preparation

of DNA

Jilters

‘I’7 phage were diluted into 15 ml of 0.1 >I-NaOH and the solution was adjusted to p = 1.72 g/cm3 with CsCl. Centrifugation was carried out at 30,000 revs/min at 18°C for 70 h in a Becklnan 50.2 Ti rotor. The DNA band was removed and diluted with 0.1 MNaOH to a final concn with an Azso value of 1.0. Calf thymus DNA (Sigma Chemical Co.) was dissolved in 0.1 M-N&H to a final concn wit11 an A,,, va,lue of 1.0. The 2 DNA preparations were chilled to 4°C and then adjusted to pH 6 with acetic acid. Portions (40 ml) were mixed with 10 ml of cold 25 x saline phosphate (1 x saline phosphat,r: 0.2 M-Nacl, 0.01 M-NaH,PO,. NaOH to pH 7.3) and 0.05 ml of 0.1 M-MgCl,. Both DNA preparations were divided into 2 portions and eacll portion was passed slowly through a 47.mm nit)rocellulose filter (Millipore, type GS, 0.22 pm). After passage of t,lle DNA solutions, the filters were washed with 5 x salinr? phosphate, dried, and baked under vacuum for 2 11 at 80°C. Blank filters were treated the same way, with the exception of the loading sol&on, which was DNA-free. Finally, 7.men diamet’er circles were punched out of the large filters and stored at -20°C. T7 filt,crs ront,ainetl 18 pg DNA/filter and calf thymus filters contained 22 rg DNA/filter. (~1) Attachment

of acylhydrazide-containing

compounds

to RX-4

Yeast RNA (Sigma Chemical Co.) was dissolved in 10m3 M-EDTA (pH 7) to give a solution of 2 mg/ml. One-ml portions of this preparation were mixed with 2.5 g of the crosslinking agent and with 1.075 g pentanohydrazide (an analog of the crosslinking agent containing only the attachment, moiety; Summerton, unpublished data), and each was acidified with 62.5 ~1 formic acid. The pentanohydrazide preparation w&s incubated for 1 11 and the agent, preparation for 2 h at 55°C. The RNA was precipitated with ethanol, resuspended in 0.1 II-ammonium acetate, and reprecipitated wit11 ethanol. These pellets and a sample of untreated RNA were digested for 16 h in 0.2 ml of 0.2 M-NaOH. Each sol&on was neutralized with a piece of solid CO,, treated overnight with alkaline phosphatase to convert t,he nucleotides to nucleosides, and fractionated by descending paper cllrornatopraplty using solvent I. Compounds (2) and (9) of Fig. 1 were chromatographed in parallrl with the RNA hydrol>,satcs. The pent,alrohydrazide-treated RNA exhibited a single new product with an RF and ultraviolet spectrnm identical to that of compound (9). and thr agent-treated RNA gave a new product with an R, and u.v. spectrum identical to t.tlat, of compound (2). Trettt,ment of these new products at pH 2.5 followed by alkali con\,erted that, from tire agent-treated RNA to n csompountl with an RF and spectrum

Tile procedures for preparat,iorr of T7 RNA \\-ere adapt,r:tl from Hummers &, Szyhalski (1968) and Minkley (1974). E. coli B wcrc incubated at, 35 ‘C wit11 aoration iti 150 11~1of tryptose brot,h to an A,,, value of 0.6. Tt le culturcx was tllen irradiated at 25-C tltltlcr 254 nm light in a dark room (approx. 15,000 ergs/mmz). After irradiation thr cells \vore incubated a further 10 min in the dark, chilled in an icp-bath, centrifuged, and the pellet was resuspended in 3 ml of cold 0.1 nl-ammonium acet,ato containing 1.5 x 1012 ‘I’7 phage. After 10 min at 4”C, the phage-infected cells were added to I5 ml of 37°C tryptosc hrotlt containing 5 mCi H, “PO, (carrier-free). After incuhatioll at 37°C for IO mill, the cult>urci was rapidly chilled in an ice-bath and centrifuged at 4°C:. Ttie pellet, was resuspended irl 3 ml of cold 0.1 M-ammonium acetate and quickly added to 3 ml of lysis rnixturo (0.1 >IEDTA, 1.576 sodium dodecyl sulfate, NaOH to pH 7) at 100°C. After 2 min in a boiling water bath, the lysate was chilled and centrifuged at 10,000 g for 10 mirr at. 0°C. The supernatant was decanted and adjusted to a final volume of 14..i ml wit11 water. and 6 g C&r, 15 g CsCl and 0.6 g LiCl were dissolved in t,he mixtnrc, which was centrifuged in thickwalled polycarbonate bottles at 30,000 revs/min at, 18°C for 70 h in a Beckman 50.2 ‘l’i rotor. The RNA band (p = 1.81 g/cm”) was removed and precipitated wit,11 2 vol. ettranol. The pellet was washed 3 times by suspension irl 5 ml 0. t RI-ammotlium acetate and reprecipitat,ed with 3 vol. ethanol. The final RNA pellet, was resuspended in lo-” nl-EDTA contained 1270 cts/min (pH 7) at a concn with an Azso value of 20.8. A typical preparation anrrr~aled per rg of RNA (CerBnkov counting). 499b of the 32P label iu this RNA preparatiotl to an excess of T7 DNA bound to nitrocellulose filters. whereas less ttuul 0.1 ok of tllo 32P label adhered to blank filters during a control annealing step. (g) Attachment

of crovslinking

agent

to labeled

RXA

A 120-~1 portion of the RNA solution described in section (f) was added to 300 mg of 6.bromo-5,5-dimethoxyhexanohydrazide. After rnixing welt, 7.5 ~1 of 91y0 formic acid were added and the preparation was incubated at 55°C for 2.5 h. The derivatized RNA was precipitated with 1 ml ethanol (2 h at -2(W), centrifuged, and t)lle pellet was resuspentled in 0.25 ml 0.2 M-ammonium acetate. Afttlr adding I.2 ml ethanol. the preparation was chilled and centrifuged. The RNA pellet was drained well and resuspended in lo-:] IUEDTA (pH 7) to give a solution of 226 pg RNA/ml. After vigorous mixing, the preparation was centrifuged briefly to pellet insoluble debris. By this procedure approx. 25% of thr cytosine residues were derivatized. The methods for estimating the percelitage of cytosinr residues derivatized are described in section (c) of Results and Discussion. For control studies, 80 ~1 of the RNA solution (section (f)) was mixed with 86 mg pentanohydrazide, 5 ~1 91% formic acid was added, and the mixture was incubated for 34 min at 55°C. Precipitation, washing, and resuspension were as described above. Approx. 17% of the cytosine residues were derivatized in this preparation. Finally, 80 ~1 of t’he RNA solution (section (f)) was precipitated, washed, and resuspended as above. The II.\-. spectra of non-derivatized (C) agent-derivatized (A), pentanohydrazidr-derivatieed (Bj RNA are shown in Fig. 2. (11) I’veactivation

of R,IA

Portions (87 ~1) of the 3 RNA solutions (section derivatized and (B) pentanohydrazide-derivatizod) activation buffer (0.05 M-chloroacetic acid to pH 40 min at 25°C. (i) Annealing

RNA

(g)) ((C) non-derivatized, (A) agentwere mixed with 87.~1 portions of 2.5 wi-ith NaOH) and incubated for

with filter-bound

DNA

Preactivated agent-derivatized RNA was prepared for annealing by adding in order: 230 ~1 formamide, 100 ~1 25x saline phosphate, 5 ~1 0.1 M-EDTA (pH 7), 10 ~1 I.0 Macetic acid, and 170 ~1 of agent-derivatized RNA (section (h)) (preparation A). Two

DNA

AND

RNA

CROSSLINKING I

AGENTS:

TESTING

151

I

(b)

260

310

260 Wovelenqth

310 (nm)

Fro. 2. 11.“. spectra of derivatized and non-derivatized RNA. RNA preparations were diluted int,o 0.1 M-NaOH and their spectra were recorded from 240 nm t,o 340 nm (-----). The solutions were then adjusted to pH 6 with acetic acid, and theiT apectra again recorded (---). (a) Non-derivatized RNA (solution C); (b) agent-derivatized RNA (solution A); (c) pentanohydrazide-derivatized RNA (solution B). Shown for comparison are the alkaline (-----) and neukal (-.-----) spectra of the corresponding nucleosides. (cl) Cytidine; (e) agent,-derivat,ized cvtidine (structure (2) of Fig. 1); (f) pentanohyclrazidr,.tlcrivatized rytidine (structure (9) of Fig. 1).

contjrol preparat#ions differing only in the RNA component were prepared; one cont,aining 170 ~1 of preactivated prntanohydrazide-derivatized RNA (preparation B), and the ot,her containing 170 ~1 of preactivated norI-deri\,atizetl RNA (prepara,tion C). Thr nrmealing mixtures had a final pH of 6.9. Non-activated agent-derivat,ized RNA was prepartxci for annealing by adding in order: 10 ~1 0.1 M-EDTA (pH 7), 20 ~1 I.0 M460 ~1 formamide, 200 ~1 25 x saline phosphat’e, acetic acid, 170 ~1 of activation buffer, and 170 ~1 of the non-activat,ed agent-derivatized R,NA solution (section (g)) (preparation D). Two control preparations differing only in the RNA component were prepared: one containing non-activated pentanohydrazidederivatized RNA (preparation E), and one containing non-activated non-derivatized RNA (preparation F). Portions (40 ~1) of preparation A were added to caclr of 4 TT DKA filters, 4 calf thymus DNA filters, and 4 blank filters. Preparations B and c‘ wcrr addt,d in a. like manner to T7 DNA filters, calf thymus DNA filters, and blank filters. PO&ions (40 ~1) of preparation D were added to each of 8 T7 DNA filters, 8 calf thymus DNA filters, and 8 blank filters. Preparations E and F were added in a like manner to Ti DNA filters, calf thymus DNA filters, and blank filters. In all of the experiments, the ratio of target DNA to derivatized RNA was of the order of 12 to 15. All IO8 filters were incubated wit,h their respective RNA solutjions for 6 h at 37°C’. rinsed in 2 changes of saline phosphate (30 min each). dried, and counted in 1 ml of saline phosphate in a scintill&ion counter (Cer&nkov counting).

12

J. SlJMIblE:RTOS (j ) Postactiz~ation

.\NI)

~ti rlericatizecl

1’. .I. RS.3

li:1RTl,~:‘I’T annealetl

with I).V.-1

\vitli preparat,ioiis I), E, alit1 II-XaOH at 37”C), as expected since both (9) and simple x-halo ketals are unaffected under these conditions (Summerton, unpublished data). At pH 2.5. t’he ketal of (2) is hydrolyzed to afford the bromoketone adduct (3). This compound ha,s spectral properties at pH 1 and 5.5 identical to those of (2), but, it is now quite sensitive t’o alkali; therefore, to prevent degradation of (3) during chromatography, the developing solvents were adjusted to pH 3.5 with formic acid. The lability in alkali of (3) is expect.ed on the basis of a. comparable lability in alka.li for simple x-halo ket)ones (Summerton, unpublished data). When structure (3) is incubated for 12 hours at neutral pH or for a few minutes at pH 11. it is converted into two compounds of unknown structure; one (structure (7)) displays a light blue fluorescence under 254 nm light. t.he other (8) is non-fluorescent and displavs spectra at pH 1. 5.5 and 13 essentia,lly identical to those of (2) and (3). The acyl hydrazide reagent, was shown to bind covalently at the C-4 position of the cytidine residue in RNA as well. Alkaline cleavage of the derivatized polynucleotide chain and dephosphorylat’ion afforded a mixture of nucleosides which contained material wit,h the same chromatographic and spectral properties as the adduct (2). Similarly, compound (9) was isolated from hvdrolysis of pentanoh,vdrazide-treated RNA. When activated adduct (3), 14C-labeled in the cvtosine moiety, is incubated with guanosine, the product (4) contains 14C and carriei a positive charge at pH 5.5, and has the spect,ral properties at pH 1 and 5.5 expected for a compound containing both the 4deamino-4-alkanohydrazidocytidine (9) and 7-oxoa,lkylguanosine (10) moieties, as illustrated in Figure 3(a) and (b). When the crosslinked material (4) is treated with 0.1 rw-NaOH at 37°C for as little as five minutes or as much as 16 hours, the resulting products, (5) and (6), still carry 14C but no longer move in an electric field at pH 5.5. Furthermore, as illustrated in Figure 3(c) and (d) and Table 1, their spectral properties are now those expected for compounds containing both the cytidine adduct (9) and an alkali-treated 7-oxoalkylguanosine. The structures (5) and (6) are proposed in analogy to bhe ring-opened products resulting from alkali treatment of 7-methylguanosine (Haines et al., 1962). The properties of the compounds in Figure 1 are listed in Table 1. The fact that (5) and (6) carry both the cytidine adduct and t’he modified guanosine after oxt’ensive alkali treat)ment suggests that the crosslinks formed by this agent will remain intact during such procedures as sediment’ation t,hrough alkaline gradient)s and alkaline hydrolysis of RNA. (c) Estimation

of the fraction of acylhydrazide-derivatized residues in RNA

cytoske

Two methods were developed for estimating what percentage of the cytosine residues were derivatized in RNA treated with pmtanohpdrazide or the crosslinking

154

d)

j0

330

340 Wavelength

(m-n)

FIG. 3. WV. spectra of structure (4) compared with WV. spectra of a mixtjure of (9) and (10). U.V. spectrum of (4) (--------) at a concn with an AzB5 value of 0.438 at pH 1. u.v. spectrum of a mixture of (9) and (10) (------) at concns with an A2s7 value of 0.216 and AZ58 value of 0.172, respectively (at pH 1). (a) pH 1.0 (0.1 M-HCI); (b) pH 5.5 (0.2 nl-citric acid t,o pH 5.5 with NaOH); (c) pH 1.0 (0.1 ix-NaOH for 5 min followed by cont. HCl to pH 1.0); (d) pH 13.0 (0.1 ix-NaOH).

One method involves alkaline hydrolysis of the derivatized RN$, dephosphorylation of the resulting nucleotides, and quantitation of the nucleosides after chromatographic separation (see section (e), Materials and Methods). After treatment of RNA with pentanohydrazide and subsequent fractionation. spectral analysis of the cytidine fraction indicates that it is pure. However, under the same experimental conditions, agent-treatsed RNA yields a cytidine fraction (RF == 0.18 in solvent 1) which is contaminated with small amounbs of other u.v.-absorbing substances (probably compounds (7) and (8), arising from hydrolysis and subsequent degradation of RNA-bound agent during the attachment’ and purification steps). In contrast to the cytidine fractions: the guanosine fractions (RF = 0.13 in solvent 1) from all preparations have the spectral characteristics of pure guanosine. Since guanosine is not modified by acylhydrazides (Hayatsu & Ukita, 1964; Gal-Or et al., 1967; Kikugawa et al., 1969), it is used as a reference for comparison with the amount of modified cytidine in the derivatized material. This chromatographic method measures only that fraction of derivatized cytosines which contain t’he agent in the protected form (as in structure (2)); it’ does not include those which contain the agent in other forms (i.e. activated, degraded, or crosslinked to intrastrandguanine residues). Table 28 presents the result from a representative experiment, employing yeast RNA derivatized as described in Materials and Methods, section (e). agent.

0.18 0.64 0.20 0~30

(8)

‘:

0.58

0.2 I

(7)

t I)escending f 0.2 M-citric 8 Solvent was 11Summerton 7 (10) - OH-

(9) I’ 10) j( 10) -~ OH-,

0.24

0.0”

(6)

2

Appearance

Deep

Light

Deep

paper

on

papw;

blue fluorescence 1)ark Dark blue fluorescence Dark

Dark

Dark

Dark Dark blue fluorescence

I)ark

chromatography

paper chromatography using Whatman 3MM acid adjusted to pH 5.5 with NaOH. adjusted to pH 3.5 with formic acid. (unpublished data). represents alkali-treated 7.acetonylguanosine.

0.5”

0.6“ 0.78

0.20

0.02

0.28

0.78

Solvent

0.95

RF?

0.55 0.408 0.04

1

0.90

Solrent

(5)

(4)

(1) (9 (3)

structure

Propertie.

solvents

tl

described

0

0 0 0

0

0

0 0 0 t-1

Electrophoretic mobility, pH 5.5 1

3 10 "87 287 "58

263 285

287 287

pH

in Materials

of mTodi$ed nucleosides

TABLE 1

and

-

5.5:

13

310 276 175 25x

311 198 199 264 264

300

"65

300

265

"65

300

259

299

pH

A,“XX (nm)

276

975 275

Methods.

pH

pH

3 10 287 287 171 271

273

273

274

387

-

1, aftor alkaline treatment

Guanosine Cytidine Derivatized-cytidine

( pmol) t (pmol)t (pmol)t

y0 Cytosines modified, Derivatized-cytidine/cytidine Derivatized-cytidine/guanosine

B. By speetrophotometric

based

on : ratio3 ratio3

1~000 0.671 0

1~000 0.542 0~1:10

14~00 0.499 0~12.5

0 0

19 19

20 19

determimtion

Untreated A 260 of RNA, A 310 of RNA, A 310 of RNA,

pH pH pH

13 13 6

‘//, Cytosines modified, based Eqn (Z), using RI1 Eqn (2), using R’II

RNA preparations Pentanohydrazidederivatized

Agent

derivatized

0.990 0.006 0.007

1 .OOO 0~102 0.020

0.984 0.107 0.019

1.7 0

28 23

30 25

on :

t (vol. eMant) . (il,,.)/(~n~,,). $ Per cent derivatization = (mol modified cytidine)/(mol modified cytidine $- unmodified cytidine). 5 Per cent derivatization = (mol modified cytidine)/(mol guanosine) (X), where X -: cytidine/ guanosine ratio in non-derivatized RNA preparation. 11R and R’ as defined in section (c) of Results and Discussion. The base rat,io for this yeast RNA preparat)ion was: A, 0.29; C, 0.21; G, 0.31; U, 0.18.

A second method for estimating the fraction of derivatized cytosine residues exploits the very high molar absorption in alkali of cytosine residues containing a,liphatic acylhydrazides (-NHNHCO(CH,),R, n :> 1) bound at the C-4 position data). (c310 = 18,800 at pH 13 and 800 at pH 6; Summerton, unpublished This rapid spectrophotometric method can be used for assessing the degree of derivatization in both RNA and DNA, and measures the t’otal amount of derivatized cytosine residues, regardless of whether they are protected, activated, degraded to crosslinked. This method requires knowledge of the base ratio of the polynucleotide, the ~~~~ and c310 values of the derivatized cytosine residues (abbreviated CMPA) in alkali, and the A,60 and A310 values of the derivatized polynucleotide in alkali. The four common ribonucleotides, AMP, CMP, GMP and UMP have negligible absorbance at 310 nm in either neutral or alkaline solution. For brevity, the following terms are used : a. fraction of AMP in RNA : 9, fraction of GMP in RNA; u, fraction of UMP in RNA; c, fract’ion of CMP in RNA before derivatization step; x, fraction of CMPA in RNA after derivatization; R, the ratio

DNA

AND

for derivatized approximated by equation &O/A26ll

RC?

(~2~” of AMP)(n)

RNS

CROSSLINKING

RNA (1).

in alkali.

AGENTS:

R for alkaline

TESTING

solutions

of RNA

157

can be

( cslo of CMPA) (2) + (~26~ of GMP)(g) -r ( cZ6,,of UMP)(u) 4 ( czco of CMP)(c ~~ .r) t ( czeo of CMPA)(cr)

in alkali are: 15,300 for AMP; 11,500 The cZGOvalues for the four 5’ ribonucleotides for GMP: 7300 for UMP; and 7400 for CMP (Dawson rt ul., 1962) ; the cZeOof CMPA is 10.000 (Summerton. unpublished data). Rearrangement of equation (1) and substitution of the appropriate values gives equation (2). z!?

K(15,3OOn + 11.5OOg + 7300% -I 74OOc) 18,800 -

(2)

26OOR

The AZ1, value of derivatized RNA in alkali is relatively low and thus it is important to avoid extraneous a,bsorbance at 310 nm if equation (2) is to remain valid. Such extraneous absorbance arises when the RNA is treated under attachment conditions for prolonged periods (presumably due to premature activation of the croxslinking moiety) or when the RNA is not wa’shed sufficiently following derivatization. Fortunately, the AsI value of the contaminating materia,l is approximately the same at pH 6 and pH 13. Since derivatized RNA absorbs significantly at 310 nm only in alkali, the difference &,,(pH

is taken as the correct value for iz,,, R’ will be defined as: &,,(pH

13) -

&a(pH

6)

due to CMPA in alkali. 13) - &,,(pH &&PH

13)

6) ’

By exchange of R’ for R in equation (2). an estimation of the fraction of cytosines derivatized in RNA preparations containing minor impurities can be made. To estimate the fraction of cytosines derivabized, a sample of the treated RNA is thoroughly washed and suspended in 0.1 M-NaOH at a concentration with an A,,, value of O-95. The A,,, and A,,, values are recorded. acetic acid is added to give a solution of pH ‘v 6, and the A,,, value of bhis neutralized solution is recorded. The base ratio data are measured experimentally, found in the literature, or estimated. As an example, consider an RNA with a base ratio of 0.25 AMP, 0.25 GMP, 0.25 UMP and O-25 CMP. Suppose that) after derivatization, an alkaline solution of t.he RNA has an A,,, value of 0.957 and an Aal value of 0.085, and after adjusting to pH 6, an A,,, value of 0.018. In this instance. R’ would have a value of 0.070, and z a value of 0.039. Thus, 3.9:,,, of the nucleotides would be CMPA. indicating that approximately ISo/;, of the cytosines had been derivatized. Table 2B presents the results of this spectrophotometric analysis of the derivatizcd yeast RNA. (d) Crodinking RNA to com,plQm~en.tar,yDNA Our RNA-DNA crosslinking assay is based on t,he rationale that agent-derivatized RNA that has annealed and subsequently crosslinked to filter-bound DNA should remain bound to the filters during a denaturing wash, whereas annealed but noncrosslinked RNA should be removed during this wash. Figure 4 presents the results of two experiment’s which demonstrate crosslinking that is specific for complement,ary DX.1.

(1)

Filters (a)

(b)

FIG. 4. Crosslinking with preactivated RKA. Yreactivated T7 RNA preparations A ( n , treated with crosslinking agent), B ( q , treated with pentanohydrazide), and C (0, treated with buffer) were annpaled with T7 DNA filters, calf DNA filters, and blank filters as dcscribcd in section (i) of Materials and Methocls. The mean cts/min of 3ZP-labeled RNA annealed to ‘~7 DNA filters is defined as the 100% value for each of the respective RNA preparations; the mean amount of RKA remaining bound to filters after the denaturing wash is plotted (vertical bars) as a percentage of this (100%) value. Vertical lines above the bars signify the standard deviations of the respective means. (a) RNA preparations with 250/b of the cytosine residues derivatized. (b) RNA,preparations with 10% of the cytosine residues derivatized.

In these experiments, we define the crosslinking efficiency as the percentage of agent-derivatized RNA initially annealed to complementary DNA which remains bound through the final denaturing wash. In determining the crosslinking efficiency, we assume that RNA derivatized with the monofunctional compound, pentanohydrazide, does not bind covalently to DNA, i.e. any pentanohydrazide-derivatized RNA remaining on a filter after the final wash is bound non-covalently. Thus, the percentage of pentanohydrazide-derivatized RNA bound to the T7 DNA after the final wash is subtracted from the percentage of agent-derivatized RNA bound to T7 DNA after the final wash. The assumption is that similar percentages of both RNA derivatives are bound in a non-covalent’ manner. This adjustment corrects fot binding to the filter disc as well. Tt should be noted that the values calculated in this manner are actually lower limits for the true crosslinking efficiencies. Using an RNA preparation which was heavily derivatized (approx. 250’b of the cytosine residues derivatized), a crosslinking efficiency of 200/:, was obtained by activating the RNA preparation before (preactivation) the annealing step (Fig. 4(a)). In a similar experiment (Fig. 4(b)), using an RNA preparation in which only 107:, of the cytosines were functionalized, a crosslinking efficiency of 70,/, was seen.

DXA

AND

REA

C‘ROSSLIKKING

AGENTS:

TESTING

150

In evaluating the specificity of the crosslinking react’ion. a correction for the amount’ of RNA which binds irreversibly to the filter disc i&elf must be applied because this becomes significant with heavily derivatized preparations. In both experiments which are summarized in Figure 4: almost all of the agent-derivat,ized RNA irreversibly bound to non-complementary calf DNA filt’er is accounted for in this way, indicating a very high specificity. This represents the first, example of a seyuence-specific crossbetween complex polynucleotide strands. linking agent which is able to discriminate In previous experiments (Grineva et al., 1974.1975.1976: Grineva & Karpova, 1973. 1974), a high degree of binding to all DNA and RNA was s:ccn with relatively short derivatized sequences. A problem with t’he use of preactivated RKX could conceivably arise if. following activation of agent-derivat#ized single-st’randed RNA. some of bhe activated crosslinking moieties formed intrastrand links t)o proximal guanosine residues, thereby becoming unavailable for crosslinking to subsequently annealed complementary DNA. non-activated derivat’izrd RX.1 was first annealed To circumvent this possibility. with filter-bound DNA and the act,ivation step \Vas performed on the derivatizedRNA-DNA duplex. A larger fraction of the cyt,osine-bound crosslinking agents should then be available for linking t’o base-paired deoxyguanosine residues, bhereby giving a commensurate increastb in crosslinking effic:iencirs. The result,s from two such

DCA

Calf DNA

Blank

Filters (a)

(b)

FIG. 5. Crosslinking with postactivatcd RNA, Ken-activated T7 RKA preparations D ( n , treated with crosslinking agent), E ( q , treated wit,h pentanohydrazide), and F (C. treated with buffer) were utilized as in the experiment of Fig. 4. (a) RNA preparations with 25yb of the cytosine residues derivatized. Postactivation wa$ carried out for 40 min at pH 3.14 and 37°C. (b) RNA preparations with 10% of the cytosine residues derivatized. Postactiration was carried out for 90 min at pH 3.50 and 37°C’.

160

.J. SUM

NE

K’I’O

S

.\ N I)

I’.

.\

1%.A K’I’ I, ET’1

postactivation experimentIs are given in Figurcx 5. In t)hcse expniments. thr crosslinking efficiencies were 25”; and 26”,,, which compares favorably wit)h t.he 20’!,, and ‘i’I;i, crosslinking efficiencies. respectively, for t’he analogous preactivat,ion ~lxperimerrts. Again, the RNA bound to t’he calf DNA is account)ed for by the blank filter. Control experiments were carried out to evaluate t,he necessity for specific activation of the crosslinking moiety in order for crosslinking to occur. The non-activat’ed RNA preparations were annealed with filter-bound DNA and the filters were subjected to denaturing conditions without an intervening activation st)ep. The results from these experiments with derivatized but non-activated RNA are shown in Figure 6. Although some highly specific crosslinking is seen. it is significantly less than occurs with either the preactivated (Fig. 4) or (9*50/, and l-70/,, respectively) postactivated (Fig. 5) preparations. The source of this crosslinking by non-activated preparations is not clear, although it is likely that some ketal hydrolysis is taking place in the attachment (pH 4.1) or annealing (pH 6.9) steps (Summerton, unpublished data). Alkylation of a deoxyguanosine residue at the N-7 position (the postulated site of the crosslinking reaction) labilizes the glycosidic bond to heat and acid (Haines et al.. 1962; Lawley $ Brookea, 1963). Cleavage of this labilized glycosidic bond (depurination) leads to opening of the deoxyribose ring, rendering the adjoining 3’-phosphoester sensitive to hydrolysis by heat or alkali (Tamm el al., 1953). In view of this sequence of reactions, it was conceivable that crosslinking RNA to filterbound DNA could lead to considerable strand cleavage of t,he bound DNA. This in

3(

Calf DNA

Blank

Blank

Filters (0)

(b)

FIG. 6. Crosslinking with non-activated RNA. Non-activated RNA preparations D, E and F (as in Fig. 5) were utilized Fig. 4. No activation step was carried out prior to the denaturing wash. (a) RNA preparations with 25% of the cytosine residues derivatized. with 10% of the cytosine residues derivatized.

as in the experiment (b) RNA

preparations

of

DNA

AND

RNA

CROSSLINKING TABLE

ISffect of RJA

crosslinking

AGENTS:

16 1

TESTLNG

3

on the amount of DNA

rermining

on jilters

RPirZ preparations “/b of RNA crosslinked

Postactivated

[3H]DNA remaining on filter (ctsjmin) % of RNA crosslinked

Non-activated

[3H]DKA remaining on filter (cts/min)

Agentderivatized

Enntanohydrazidrderivat,ized

26.1

0

1094.5.jb41.0

1060.1 : 148.2

1.7

0

1116.6:: 169.9

1091.7 +27o.s

Nonderivatized 0

1153.9173.7

0

1067.5&50.4

In experiments similar to those of Fig. 5 (postactivated) and Fig. 6 (non-activated), but using filter-bound T7 [3H]DNA, the crosslinking efficiencies were calculated as described in section (d) of Results and Discussion. The filters, which had been through the final, denaturing wash, were then assayed for the amount of DNA remaining. For each set of filters, the crosslinking efficiencies are tabulated with the corresponding amounts of DNA remaining filter-bound after the final wash. In these experiments, an estimated 6.6% of the cyt,osine residues were derivatized in the agenttreated RNA preparations, and 13.1% of the cytosine residues were derivatized in the pentanohydrazide-treated RNA preparations. The DNA/RNA ratio in the annealing mixture.+ was 6. Postactivation was carried out at, pH 3.5.

turn could result in loss of a substantial portion of the DNA from its attachment sites on the filter. The final wash would then remove this cleaved DNA with its crosslinked RNA, leading to a serious underestimation of the crosslinking efficiency. To evaluate this possibility, experiments similar to those summarized in Figures 5 and 6 were repeated, using filter-bound 3H-labeled T7 DNA. For these experiments, the DNA/ RNA ratio in the annealing mixture was 6, compared to 12 to 15 for the experiments of Figures 5 and 6. After making the measurements necessary for calculating the crosxlinking efficiency, the filters were dried and counted in scintillation fluid to assess the amount of [3H]DNA remaining on the filters. The results are presented in Table 3 and suggest that RNA crosslinking causes no serious loss of DNA from the filters. Our aim at the outset was to develop an effective means of inactivating specific nucleotide sequences, and we investigated the use of 6-bromo-5,5-dimathoxyhexanohydrazide for this purpose. Under conditions where RNA and DNA are not degraded, bhis bifunctional crosslinking agent can be attached to the C-4 position of cytidine, converbed to an alkylating agent, and shown to crosslink with guanosine by alkylation at t,he N-7 position. These model reactions provide support for the postulated sequence of tbvents in the specific crosslinking of T7 RNA with its complementary DNA. T7 RNA can be derivatized with the crosslinking agent, activated, and incubated under annealing conditions with both complementary and non-complementary DNA. The derivatized RNA binds irreversibly and highly specifically to its complementary DNA in ij reaction which requires the bifunctional crosslinking agent. Because of the

162

.J. SC’JII\1ER1’0S

=\SI)

L’. ;I.

W.-\H.TI,l~:‘I”r

specificity of this binding. and the stab My of the linkage undw cwldit,ions \+.hic:lr dissociate normal RNA-DNd duplexes. \ve expect Urat this and related crosslinking agents will be useful sequence-specific blocking agents for nucleic acids. Our future goals include structural verification and quantita.tion of botjh RNA--DSA intorstjrand cro4inks and RKA-RNA intrast,rand crosslinks. as \r.rll as further refinement arltl applications of this class of crosslinking rwgent. Mucll of this project, was carried out in tile Department of Molecular Biology, Utri\-clrsitjof California, Berkeley, under the joint sponsorship of H. Fraenkol-Conrut and B. Singer. Their considerable contribut,ions of wlpport and advice are grat,efully acknowledged. An additional part of this project was done in the Division of Molecular and Celh&u Biology at National Jewis11 Hospital and Researcll Ceritrr, Denver, Colorado, under thtl sponsorship of Dr Kathleen Hercules; l,(,r support iI1 tllia project is appreciated. S\lbstantial help in the cllomical aspects of tllis project was provided by F. R. (:IYXYI 111. Department of Chemistry, University of (‘alifornia, Berkeley. Finally, the preparation of T7 DNA was greatly facilitated by advice and assistance frown Robert Fiscller and Warrcltl Maltzman, both of the Department) of MolCcIIlar Biology, Ul1ivcirsit-y of California. Brrkel(~y. Financial support for this work included National Institutes of Health Postdoctoral Fellowships CA-04039 and CA-05883, Reseercl~ grants CA - 12316 and CA-1 9057 from thcs National Cancer Institute, Kesearell grant PCM-75-07854 from tllc, National Sciarlcc Foundation. A portion of this work was done during tile t)enurc of an Established Jnvestigatorship of tllc America11 Heart Association II&J hy OII(~ of 11s (J. S.). REFERENCES Bispink, L. & Matthaei, H. (1973). FEBS Letters, 37, 291-294. Breitmeyer, J. & Nollcr, H. (1976). J. Mol. RX. 101, 297-- 306. Budker, V., Knorre, D., Kravchcrtko, V.. Lavrik, O., Ncvinsky, G. si Teplova, N. (1974). FEBS Letters, 49, 159-162. Dawson, R., Elliott, D., Elliott, J$‘. & Jones, K. (1962). Data for Biochemical Research, pp. 74--79, Oxford University Press, London. Eilat, D., Pellegrini, M., Oen, H., de Groat. N., Lapidot, >r. & Cantor, C. R. (1974). Nature (London), 250, 514--516. Gal-Or, L., Mellema, J., Moudrianakis, K. 8r. Beer, M. (1967). Biochemistry, 6, 1909-1915. Grineva, N. I. & Karpova, G. G. (1973). FEBS Letters, 32, 351-355. Grineva, N. I. & Karpova, G. G. (1974). lgol. Biol. (U.S.S.R.), 8, 832-844. Grineva, N. I., Karpova, G. Cr., Kutznetsova, L. M., Venkstern, T. V. & Bayer, A. A. (1972). Nucleic Acids Res. 4, 1609-1631. Grineva, N. T., Karpova, G. 0. & Shamovskii. G. G. (1974). ;Wol. Biol. (V.S.S.R.), 8, 358371. Grineva, N. I., Karpova, G. U., Myzina. S. D. & Chemesova, A. N. (1975). Bioorg. Rhim. (U.S.S.R.), 1, 1707~1715. Grineva, N. I., Karpova, G. G., Myzina, 8. D., Fodor, 1. & Baev, A. A. (1976). Dokl. Akad. NaukS.S.S.R., ser. Biochtm. 223, 1477.-1480. Haines, J., Reese, C. & Todd, L. (1962). J. Chem. Sot. 5281- 5288. Hayatsu, H. & Ukita, T. (1964). Biochem. Biophys. Kes. Commun. 14, 198-203. Kikugawa, K., Hayatsu, H. & Ukita, T. (1969). Chem. Biol. Interactions, 1, 247-256. Lawley, P. & Brookes, P. (1963). Biochem. J. 89, 127.-138. Minkley, E. (1974). J. Mol. BioZ. 83, 289. 304. Pongs, 0. & Lanka, E. (1975). Proc. IVat. Acad. Sci., IjT.S.A. 72, 1505-1509. Schoenmaker, H. J. P. & Schimmel, I’. R. (1974). J. iclol. Biol. 84, 503-513. Schwartz, I. & Ofengand, J. (1974). t’roc. IVat. Acad. Sci., U.S.A. 71, 3951. 3955. Sonnenberg, N., Wilchek, M. & Zamir, A. (1975). Proc. iVat. Acad. Sci., V.S.A. 72, 4332-4336. Summers, W. C. & Szybalski, W. (1968). I’irology, 34, 9-16. Tamm, C., Shapiro, H., Lipshitz, R. & Chargaff, E. (1953). J. Biol. Chem. 203, 673-688. Wagner, R. BE Gassen, H. (1975). Rio&em. Biophys. Res. Commun. 65, 519-529.