BACTERIOPHAGE T4 ENDONUCLEASES I1 AND IV, OPPOSITELY AFFECTED BY dCMP HYDROXYMETHYLASE ACTIVITY, HAVE DIFFERENT ROLES IN THE DEGRADATION AND IN THE RNA POLYMERASE- DEPENDENT REPLICATION OF T4 CYTOSINE- CONTAINING DNA

Bacteriophage T 4 mutants defective in gene 56 (dCTPase) synthesize DNA where cytosine (Cyt) partially or completely replaces hydroxymethylcytosine (HmCyt). This Cyt-DNA is degraded in vivo by T 4 endonucleases I1 and IV, and by the exonuclease coded or controlled by genes 46 and 47.-Our results demonstrate that T 4 endonuclease I1 is the principal enzyme initiating degradation of T 4 Cyt-DNA. T h e activity of endonuclease IV, but not that of endonuclease 11, was stimulated in the presence of a wild-type dCMP hydroxymethylase, also when no HmCyt was incorporated into phage DNA, suggesting the possibility of direct endonuclease IV-dCMP hydroxymethylase interactions. Endonuclease I1 activity, on the other hand, was almost completely inhibited in the presence of very small amounts of HmCyt (3-9% of total Cyt + HmCyt) in the DNA. Possible mechanisms for this inhibition are discussed.-The E. coli RNA polymerase modified by the products of T 4 genes 33 and 55 was capable of initiating DNA synthesis on a Cyt-DNA template, although it probably cannot do so on an HmCyt template. In the presence of an active endonuclease IV, Cyt-DNA synthesis was arrested 10-30 min after infection, probably due to damage to the template. Cyt-DNA synthesis dependent on the unmodified (33-55-) RNA polymerase was less sensitive to endonuclease IV action. PON infection of Escherichia coli with bacteriophage T4, host DNA is U broken down into nucleotides which are utilized for phage DNA synthesis. Two phage-coded endonucleases (I1 and IV, coded by genes denA and denB, respectively, and henceforth called endo I1 and endo IV) are involved in this degradation. Endo I1 plays a major role, whereas the action of endo IV is minor and not essential (WARNER et al. 1970; HERCULFS et al. 19'71; SOUTHER, BRUNER and ELLIOTT 1972). The resulting fragments are further degraded by the action of the exonuclease coded or controlled by genes 46 ' Present address: Department of Microbiology, Biomedical Center, University of Uppsala, S-751 23 Uppsala, Sweden. All reprint requests should be sent to this address. Genetics 114: 669-685 November, 1986. 670 K. CARLSON AND A. 0VERVATN and 47 (WIBERG 1966; KUTTER and WIBERG 1968) (henceforth called the 46/ 47 exonuclease or exo 46/47) and utilized for phage DNA synthesis. The endonucleases are not essential for phage development in common laboratory hosts. The 46/47 exonuclease is required, however, since 46 mutants arrest DNA synthesis, show aberrant late gene expression and do not recombine (EPSTEIN et al. 1963; SHALITIN and KAHANA 1970; HOSODA, MATHEWS and JANSEN 1971; SHAH and BERGER 1971; SHALITIN and NAOT 1971; PRASHAD and HOSODA 1972; HOSODA 1976). Bacteriophage T 4 mutants defective in gene 56 (dCTPase) incorporate dCMP to varying extents into their DNA, instead of hydroxymethyl-dCMP (Hm dCMP). A defect also in gene 42 (dCMP hydroxymethylase) and/or in gene 1 (deoxynucleotide kinase, which phosphorylates Hm dCMP to Hm dCDP) leads to complete replacement of glucosylated 5-hydroxymethylcytosine (GlcHmCyt) by cytosine (Cyt) in new phage DNA. This phage DNA is attacked by the same nucleases active in host DNA degradation: endo 11, endo IV and exo 46/47. BRUNER, SOUTHER and SUGGS (1972) and KUTTER et al. (1975) suggested a major role for endo IV in this degradation, since denB-56strains, which lack endo IV, make stable Cyt-DNA in vivo, whereas 56or denA-56-, which both have an active endo IV, did not. More recently, CARLSON and WIBERG (1983) examined the fate of phage Cyt-DNA made by 42-46-56mutants by agarose gel electrophoresis, and they found that it is subjected to in vivo degradation where endo I1 plays a major role, whereas the presence or absence of endo IV only causes minor changes in the fragmentation pattern. This degradation generates discrete, genetically unique double-stranded fragments of foreign DNA (since normal GlcHmCyt-containing T 4 DNA is immune to these endonucleases) and thus is a restriction by the common definition of this term. The present investigation was undertaken in order to clarify the roles of endonucleases I1 and IV in the metabolism of T 4 Cyt-DNA. We show that endo I1 is the principal enzyme initiating degradation also of phage Cyt-DNA. Intriguingly, the activity of endo I1 was virtually completely inhibited by very small amounts of 5-hydroxymethylcytosine (HmCyt) in the DNA. Endo IV activity, on the other hand, was stimulated by the presence of a wild-type (wt) gene 42 product (gp42, dCMP hydroxymethylase), also when no HmCyt was incorporated into the DNA. These results account for the stable DNA synthesis by denB-56strains, where 3-9% of the Cyt residues are replaced by HmCyt. We further demonstrate that the gp33-gp55-modified E. coli RNA polymerase is capable of initiating replication of T 4 Cyt-DNA, but that this replication differs from replication initiated by the unmodified RNA polymerase in being very sensitive to endo IV. A preliminary report of some of these results was presented at the Evergreen International T 4 Meeting in 1985. MATERIALS AND METHODS Most strains of bacteria and bacteriophage, methods of construction of phage multiple mutants, growth media, in vivo I4C-labeling of DNA, DNA extraction, gel electrophoT4 ENDONUCLEASES I1 AND IV 67 1 resis and autoradiography have been described (CARLSON and WIBERG 1983). E. coli AJ1 (BOYD and ZILLIG 1974) is recA1, argG6, metB1, thi-1, lacZ53, gal-6, malA1, xyl-7, mtl-2, rpsLIO4, tsx-1, XR, Rif'; strain AJ7 is isogenic to AJ1, except for rpoB70 and, thus, Rip. Both AJ1 and AJ7 were obtained from B. BACHMANN. T 4 1 amB24X5 was from J. WIBERG. Quantitation of in Vivo DNA synthesis: DNA synthesis was estimated essentially according to KUTTER and WIBERG (1968): Bacteria were grown in glycerol-casamino acids medium (GCA) to 3 X 10' cells/ml at 37" and were infected with a multiplicity of infection (MOI) of 8-10 plaque-forming units per milliliter. Less than 5% of the infected cells remained viable 5 min after infection. Four minutes after infection a mixture of ['HIdThd (Amersham TRK 418) to yield 10 fiCi/ml and 6 X lo-' g/ml (2.4 X IO-' M), and dAdo to yield 250 fig/ml, was added. Samples of 30 f i I were withdrawn at different times and were pipetted onto glass-fiber filter discs (Whatman GF/A or GF/C), which were immediately submerged in about a liter of ice-cold 10% trichloroacetic acid (TCA). After 30 min on ice, the TCA was exchanged with the same volume 5% TCA. This was repeated after another 30-min incubation, and finally, the filters were washed with ice-cold acetone, dried and counted, using a nontoxic scintillation liquid (Safe-FluorTM from Lumac Systems, no. 3057). Reutilization of host DNA: A mixture yielding 20 fiCi/ml of ['HIdThd and 100 pg/ ml of dAdo was added to E. coli B grown to 7 X lo' cells/ml in GCA. After two generations of further growth in the radioactive medium, the cells were centrifuged out, washed once and resuspended in an equal volume of prewarmed GCA. Aliquots were infected at MO1 8 with different T 4 mutants. The acid-precipitable radioactivity remained constant throughout infection with all the mutants tested. Thirty minutes after infection, DNA was isolated from the infected cells (CARLSON and WIBERG 1983) and was hybridized to nitrocellulose membraned filters (Millipore HAWP) charged with 3to 5-fig of denatured bacterial or T 4 DNA as described by DENHARDT (1966). Hybridization was carried out at 42" in the presence of 50% formamide (CARLSON and WIBERG 1983); filters were washed extensively, dried and counted as described under DNA synthesis. HPLC chromatography: DNA from infected cells was isolated as previously described (CARLSON and WIBERG 1983), with the addition of RNase treatment (100 fig/ml pancreatic RNase for 30 min at 37 " ; RNase heated to 100 " for 15 min before use) followed by two ethanol precipitations. Samples were finally dissolved in DNA-buffer (10 mM Tris-HC1, pH 7.4, 10 mM NaCI). Reference DNA was extracted from CsCI-purified phage particles (CARLSON and NICOLAISEN 1979): T 4 Cyt-0 (CARLSON 1980) to yield Cyt-DNA, and T 4 orgt57 pgt l4 (from H. REVEL; GEORGOPOULOS and REVEL 1971) or T 4 wt to give HmCyt-DNA. Unglucosylated or glucosylated HmCyt-DNAs gave similar results. The hydrolysis procedure was patterned on that described by CARRIER (1 98 1) for formic acid hydrolysis, using trifluoroacetic acid (TFA) since both perchloric and formic acid hydrolysis gave irreproducible results. About 100 fig DNA (determined by 0.D.260) was precipitated with ethanol, and the resulting pellet was dried and dissolved in 0.2 ml of concentrated TFA. The sample was transferred to an ampule, frozen at -70° , incubated at 100" for 1 hr and again frozen at -70". The ampule was opened, the contents thawed and the liquid blown off with gaseous nitrogen. The bases were dissolved in 100 f i I of DNA-buffer containing 1 mM ethylene-diamino tetraacetic acid and 1 mM bmercaptoethanol. Free bases were purchased from Sigma and were dissolved in the same buffer to provide standards. Samples were analyzed in a PerkinElmer series 4 HPLC apparatus, using a Whatman Particil-IO-SCX cation exchange column (length 25 cm, inner diameter 4.60 mm, particle size 10 fim, void volume 2.58 ml, sample volume 10 f i l , flow rate 1.5 ml/min). The running buffer was 20 mM ammonium formate, pH 4.0 [prepared from ammonia, Merck supra pure and formic acid (BDH Chemicals)], and 4.8% methanol. Analysis of E. coli DNA or T 4 Cyt-0 DNA showed only Cyt and no HmCyt, whereas T 4 HmCyt-DNA analysis showed only HmCyt and no Cyt. An equimolar mixture of 672 K. CARLSON AND A. 0VERVATN adenine, thymine, HmCyt, Cyt and guanine gave identical patterns before and after being subjected to the hydrolysis treatment, and doubling the amount of HmCyt or Cyt in the sample lead to twofold increase in the area under the respective peaks. Separation of E. coli and T4 DNA on KI-bisbenzimid gradients: DNA was extracted from infected cells as described above for HPLC chromatography. The sample for centrifugation contained (FERRIS and V ~ C T 1982) 100-400 fig of DNA, 1 mM NaZSOs, 30 fig/ml Bisbenzimid H (Sigma, B2883) and KI (Merck) to obtain a refractive index (Na, 25") of 1.414. Samples were centrifuged in a Spinco VTi65 rotor at 240,000 X g for 18 hr. Fractions of 0.2 ml were collected from the bottom of the tube, and the refractive index and fluorescence (excitation 340 nm, emission 490 nm) were measured using a Perkin-Elmer MPF-3 Fluorescence Spectrophotometer. E. coli DNA and T4 HmCyt-DNA banded at refractive indices between 1.465 and 1.470, whereas T4 CytDNA banded at a refractive index of 1.414. Restriction analysis of DNA: Restriction enzymes were purchased from New England Biolabs, or were generous gifts from SVEIN LUND, University of Tromsd, and were used as recommended by New England Biolabs. When samples of intracellular DNA were to be treated with restriction enzymes in vitro before gel electrophoresis, they were first dialyzed for 30 min at room temperature against 500 ml of DNA-buffer on a Millipore dialysis membrane (VMWP 02500).