The kalilo senescence plasmid of Neurospora intermedia has covalently-linked 5? terminal proteins

Curr

Current Genetics

Genet (1990)17:195-201

9 Springer-Vertag 1990

The kalilo senescence plasmid of Neurospora intermedia has covalently-linked 5' terminal proteins P. John Vierula 1, Chun K. Cheng 1, Deborah A. Court 1, Rick W. Humphrey 1, David Y. Thomas 2, and Helmut Bertrand 1 i Department of Microbiology, University of Guelph, Guelph, Ontario, Canada N1G 2W1 2 Biotechnology Research Institute, National Research Council, Montreal, Quebec, Canada Received September 7, 1989

Summary. Kalilo is a linear plasmid associated with senescence in Neurospora. The terminal Eco R1 restriction fragments of this element are linked to a protein component which remains bound despite denaturation with high concentrations of SDS. Following digestion with proteinase K, the 5' termini of the plasmid remain resistant to lambda exonuclease whereas the 3' termini are sensitive to exonuclease III, suggesting that the terminal protein is covalently linked. From an analysis of iodinated proteins released by nuclease digestion, the size of the terminal protein was estimated to be 120 kDa. The covalent linkage between D N A and protein was shown to be alkali-labile suggesting that it is a phosphodiester bond. Electron micrographs of the intact plasmid demonstrate that the associated proteins are terminal, and may be involved in replication. Key words: Neurospora - Plasmid - Terminal proteins Mitochondria

Introduction Several senescent strains ofNeurospora intermedia, isolated from the Hawaiian island of Kauai, contain a linear double-stranded DNA plasmid (Bertrand et al. 1986). This 8.6kb element, named kalilo, has no significant homology to either nuclear or mitochondrial (mt) DNA (Bertrand et al. 1985). The onset of senescence in vegetative cultures of N. intermedia has been correlated with the insertion of kalilo into any one of several possible sites in the mitochondrial genome. The gradual deterioration of vegetative cultures, which includes stop-start growth, formation of inviable conidia and disappearance of cytochromes aa3 and b, appears to result from the suppressive accumulation of defective m t D N A containing

Offprint requests to: H. Bertrand

an integrated plasmid (Griffiths and Bertrand 1984; Bertrand et al. 1985, 1986). Linear, double-stranded DNA plasmids were first identified in Zea mays (Pring et al. 1977). They have since been found in a number of different plants (Pring and Lonsdale 1985), bacteria (Hirochika and Sakaguehi 1982) and fungi (Garber et al. 1986). Many of these elements have been reported to be covalently linked to putative terminal proteins at their 5' ends (Kemble and Thompson 1982; Erickson et al. 1985; Kistler and Leong 1986; Meinhardt et al. 1986; Giasson and Lalonde 1987; Duvell et al. 1988) which are presumed to play a role in their replication. While the kalilo plasmid appears to have a structure similar to that of other linear mitochondrial plasmids, it is unusual in that almost full-length elements integrate into the mitochondrial chromosome by a mechanism that creates giant inverted repeats of mtDNA at the flanks of the resulting insertion sequence (Chan et al., submitted). It is believed that the ends of the plasmid function as "actve sites" for the integration, as well as replication, of the element (Chan et al., submitted). In this report we demonstrate that the kalilo plasmid has covalently linked proteins at its 5' termini. Evidence is also presented which suggests that these terminal proteins are linked to the DNA via a phosphodiester bond, and that they may be involved in the initiation of replication of the linear plasmid.

Materials and methods Neurospora strains and growth conditions. The senescent strain, P561, and the plasmid-free strain, P605, of Neurospora intermedia have already been described (Rieck et al. 1982; Griffiths and Bertrand 1984). A single ascospore derivative from a cross between the maternal P561 parent and P605, denoted as strain 9142, was selected for this study due to its high content of free plasmid and advanced senescencephenotype. Silica gel stocks from the third subculture of strain 9142 provided the inoculum for conida flasks prepared as described by Davis and deSerres (1979). A dense conidial suspension from these flasks was then used for subsequent inoculation of liquid

196 2-4h at 37~ At the highest concentration employed, the enzyme: terminal protein ratio was estimated to be 1: 2.

Iodination of the terminalprotein. Proteins associated with 250 lag aliquots of mitochondrial DNA were iodianted using the BoltonHunter Reagent (> 2,000 Ci/mmol; ICN Biomedicals, Inc.) according to the manufacturer's recommendations. The iodinated DNAprotein complex was recovered by ethanol precipitation, and the precipitate was rinsed three times with 70% ethanol.

Polyacrylamidegel electrophoresis. Iodinated proteins were resolved on a 7.5% SDS polyacrylamide gel essentially according to the method of Laemmli (1979). Protein molecular weight standards were obtained from BRL.

Electron microscopy. Nickel grids were prepared by coating them

Fig. 1. Electrophoretic analysis of total mitochondrial DNA before and after treatment with proteinase K. Undigested mitochondrial DNA from strain 9142 treated as follows before loading: no SDS added (lane 1), DNA heated to 60~ for 5 rain with 1% SDS (lane2), DNA treated with proteinase K for 2 h (lane3). M and K indicate the positions of mitochondrial DNA and proteinase K-treated kaBlo plasmid, respectively

with Formvar and a thin layer of carbon, glow discharging the carbon films and immediately treating them with poly-L-lysine (Sigma) essentially as described by Williams (1977). Final preparation and spreading of DNA samples was also as described by Williams (1977). Nicked, circular bacteriophage c)X174 (New England Biolabs) was used as a size standard. Plasmids were initially photographed at 28,000-fold magnifications using a Philips model 400T electron microscope.

Results cultures. Conidia of the P605 wild type strain were prepared either from silica gel stocks or subculture slants. Liquid cultures for the isolation of mitochondria were grown in Vogel's minimal medium (Vogel 1947) at room temperature. The wild type strain was grown for 12-15 h and the senescent strain for 40-48 h

Isolation of mitochondrial DNA. Mitochondria were isolated using the flotation gradient procedure of Lizardi and Luck (1971) as modified by Lambowitz (1979). Mitochondrial DNA was isolated as described previously (Bertrand et al. 1985) except that the CsC1 gradient step was replaced by an additional extraction with phenol:chloroform:isoamyl alcohol (50:50:1; P:C:I) followed by extraction with chloroform:isoamyl alcohol (24:1). DNA was recovered by ethanol precipitation. Proteinase K-treated kalilo DNA was recovered from agarose gel slices by electroelution essentially as described by Maniatis et al. (1982).

Manipulation of DNA. Digestions with restriction enzymes were performed according to manufacturer's instructions. Digestions of iodinated mitochondriai DNA with DNase I (Pharmacia) were carried out at 37~ for 1-2 with 1.2 units of enzyme per lag DNA in 40 mM Tris-C1, pH 7.5, 6 mM MgClz, followed by incubation with five units of exonuclease III (Pharmacia) for 30rain in the same buffer. Digestions with lambda exonuclease and exonuclease III were performed according to the supplier's recommendations (Pharmacia). Approximately 10lag of mitoehondrial DNA were alkali-treated in a volume of 0.1 ml containing 500 mM NaC1, 1 mM EDTA and either 50mMNaOH or 0.5 M piperidine. Following incubations at 55~ or 37~ respectively, for various times, the reactions were neutralized by adding 10 gl of 3 M sodium acetate, pH 5.2, and the mixtures were diluted with 3 volumes of 100 mM Tris-C1, pH 7.6, 1 mM EDTA. Control reactions were neutralized immediately after addition of the alkali. The DNA was recovered by ethanol precipitation in the presence of 2 gg of yeast tRNA as carrier. Following resuspension, the denatured mitochondrial DNA was reannealed in 10 mM Tris-Cl, pH 7.6, 1 mM EDTA, 100 mM NaC1 by heating to 65~ for 5 rain and then incubating at 57~ for 60 rain.

Proteolytic digest. Digestions of 5-10 lag of mitochondrial DNA, with proteinase K (0.05 gg/ml), were carried out at 37~ for 2-4 h in 10raM Tris-Ct, pH7.6, 1 mM EDTA, 1% SDS. Digestions with TPCK (L-1-Chloro-3-[4-tosylamido]-4-phenyl-2-butanone)-treated trypsin were carried out in 10 mM Tris-C1, pH 7.6, 1 mM EDTA for

The kalilo-free plasmid has an associated terminal protein Total mitochondrial D N A was initially prepared f r o m the senescent strain 9142 by P: C : I extraction without prior treatment with proteinase K. W h e n this D N A was electrophoresed on an agarose gel, without first adding SDS to the sample buffer, no plasmid D N A was a p p a r e n t below the mitochondrial D N A b a n d (Fig. 1, lane 1). Heating this D N A to 60~ for 5 min in the presence of 1% SDS, before loading, facilitated entry o f the free plasmid into the gel (lane 2). However, the plasmid D N A did not f o r m a sharp band, presumably as a result of aggregation mediated by the terminal proteins. Pretreatment of total mitochondrial D N A f r o m strain 9142 with proteinase K resulted in an increased electrophoretic mobility o f the plasmid and its resolution f r o m the m t D N A (lane 3). A n analysis o f mitochondrial D N A digested with various restriction enzymes is shown in Fig. 2B. F r o m the restriction m a p of the integrated f o r m of kalilo (Fig. 2A), it was k n o w n that EcoRI cuts within the inverted long terminal repeats (LTRs) but HindIII and BglII do not (Bertrand et al. 1986). Two additional restriction fragments of the expected lengths for the terminal fragments were resolved by agarose gel electrophoresis when either HindIII- or BglII-restricted mitochondrial D N A (lanes 2 and 5) was treated with proteinase K (lanes 3 and 6). Only one additional fragment was resolved following proteinase K treatment of the EcoRI-digested D N A (lane 9) relative to the untreated D N A (lane 8). This demonstrated that the protein is associated with two identical size Eco R 1 fragments which contain the LTRs. In the absence o f a proteinase K treatment, the kalilo element does not b a n d in a CsCI gradient. N o r is the D N A dissociable f r o m its protein c o m p o n e n t by extraction with 50% formamide, 8 M urea or 6 M guanidine isothiocyanate (data not shown). Therefore, either the protein

197

Fig. 3. Electrophoretic analysis of EcoRI-digested total mitochondrial DNA, from strain 9142, treated with the following amounts of trypsin: one unit (lane 4), 0,1 units (lane 5), 0.01 units (lane 6), 0.001 units (lane 7). Lane l, EcoRI digests of P605 mitochondrial DNA. Lanes 2 and 3, 9142 mitochondrial DNA before and after treatment with proteinase K. Lane 8, EcoRI-digested 9142 mitochondrial DNA treated with one unit of trypsin in the presence of 1 mM PMSF

Fig. 2. A partial restriction map of the kalilo mitochondrial plasmid. Restriction sites for EcoRI (E), HindIII (H) and BglII (17) are indicated. Cross-hatching delineates the inverted LTRs. B release of terminal restriction fragments by proteinase K treatment of kalilo DNA. Total mitochondrial DNA from plasmid-free strain P605 (lanes 1, 4 and 7), from strain 9142, not treated with proteinase K (lanes 2, 5 and 8) or from strain 9142 treated with proteinase K (lanes 3, 6 and 9) were digested with the indicated restriction enzymes. The terminal fragments released proteinase K treatment are indicated by the arrowheads

component of this plasmid binds tenaciously to the DNA, or it is covalently linked, as has been conclusively demonstrated for the genome-linked viral proteins (VPg) of adenovirus (Tamanoi and Stillman 1982) and bacteriophage 029 (Hermoso and Salas 1980). In the presence of SDS, proteinase K attacks peptide bonds with little specificity. This should result in the thorough hydrolysis of any protein associated with the plasmid, regardless of the nature of its interaction with the DNA. To help discriminate between covalent and noncovalent interactions between the DNA and protein, we

used varying concentrations of trypsin, under non-denaturing conditions, to cleave accessible lysine and arginine residues in preparations o f E c o R I - d i g e s t e d mitochondrial DNA from the senescent strain. Evidence from this experiment, which favours the hypothesis that protein is covalently linked to the kalilo plasmid, is provided by the results in Fig. 3. At low trypsin concentrations, the terminal EcoR1 fragment was visible as a diffuse, slower migrating, band. With higher enzyme concentrations, the terminal Eco RI restriction fragment formed a sharp band with higher mobility but which still migrated more slowly than the terminal fragment that was obtained after treatment with proteinase K. Since the residual peptide was still capable of retarding the electrophoretic mobility of the terminal E c o R I restriction fragment, a covalent linkage was deemed likely. Because the substrate was not denatured prior to trypsin digestion, additional arginine or lysine residues may still have been present within the tryptic peptide but were less accessible to the enzyme. The presence of a single, discrete, terminal DNA fragment (with attached peptide) in the tryptic digests (Fig. 3 lanes 4 and 5) also suggests that the same protein is found at both ends of the plasmid.

198

Fig. 4. Exonuclease digestions of kalilo DNA pre-treated with proteinase K. pBR322 DNA, linearized with EcoRI, was included in each reaction as a control. Lanes 1 and 3, plasmids incubated in lambda exonuclease buffer and exonuclease III buffer respectively. Lane2, plasmid treated with ten units lambda exonuclease for 60 min. Lane 4, plasmid treated with 100 units of exonuclease III for 60 min.

Fig. 6. Electrophoresis ofEco RI-digested, mitochondrial DNA from strain 9142 following alkaline treatment and renaturation. Incubations in 50mMNaOH or 0.5M piperidine were performed as described in Materials and methods. Lanes A and B, the same untreated DNA before and after digestion with proteinase K respectively. The duration of alkaline treatment, in min, is indicated above each lane. The position of the terminal restriction fragment of kalilo is indicated by the arrow

(lane 4), but proved to be resistant to lambda exonuclease activity (lane 2). This suggests that the 5' termini of the plasmid are not susceptible to exonuclease digestion due to blockage by the covalently-linked residual peptide. Fig. 5. Autoradiograph of 125I-labeled proteins resolved by SDSPAGE. Iodinated proteins from preparations of total mitochondrial DNA from strain P605 and strain 9142 before (--), and after DNase I digestion (+) as indicated. The putative terminal protein is indicated by the arrow. The molecular weight standards were visualized by staining with Coomassie brilliant blue R250

Additional evidence in support of a covalent linkage between the terminal protein and kalilo is provided by the results shown in Fig. 4. Proteinase K-treated plasmid D N A was recovered from an agarose gel by electroelution and treated either with exonuclease III, a processive enzyme which initiates degradation at free 3' ends, or the 5' specific lambda exonuclease. Linearized pBR322 D N A was included in each reaction as a control. The kalilo plasmid was degraded by exonuclease I I I digestion

Visualization o f the terminal protein

F r o m the size of the plasmid it was inferred that any terminal protein would be present in low amounts relative to the mass of the DNA. Therefore, purified mitochondrial D N A from both the plasmid-free P605 and senescent 9142 strains were labeled in vitro with 125I using the Bolton-Hunter reagent. A number of proteins were visible on autoradiograms when the iodinated mitochondrial D N A preparations of both strains were electrophoresed through a denaturing polyacrylamide gel (Fig. 5). This was not surprising, since the D N A was purified by three extractions with P : C : I in order to limit losses of the plasmid into the organic phase and interface. The identity

199

Fig. 7A. Electronmicrographsofkalilo DNA. A gel-purifiedplasmidDNA pre-treatedwithproteinaseK. B, C and D plasmidDNA identified in preparations of total mitochondrial DNA. Terminalproteins are indicated by the small arrows. Possible replication intermediatesare indicated by the large open arrowheads. The bars correspond to 0.2gm of the more prominent labeled polypeptides has not been established, but they may have an affinity for DNA which may increase their partitioning into the aqueous phase during phenol extraction. When these iodinated preparations were treated with DNase 1, and subsequently with exonuclease III, a novel protein with an approximate molecular weight of 120,000 could be detected only in the 9142 mitochondrial DNA preparation. Since this protein was released by enzymatic digestion, but not by boiling in the standard SDS sample buffer (2% SDS, 2% 2-mercaptoethanol), it is likely to be covalently linked to the DNA. The presence of the terminal protein may prevent complete digestion of all of the plasmid DNA. Therefore, 120kDa could be an over-estimate of the molecular weight of the terminal protein. Although the 120 kDa protein was the only species visible, it is also possible that other polypeptides were liberated by the nuclease digestion but obscured by one of the other labelled proteins in the one-dimensional gel.

Alkaline hydrolysis of the linkage between the terminal protein and kalilo DNA The alkylphosphodiester bond joining the phage 029 and adenovirus genomes to their respective genome-linked viral proteins (VP's) are sensitive to alkaline hydrolysis (Hermoso and Salas 1980; Desiderio and Kelly 1981). Using similar reaction conditions, the terminal protein of kalilo was also released from the terminal EcoRI restriction fragment (Fig. 6). The lability of this linkage to 50 mM NaOH or 0.5 M piperidine suggests that the terminal protein is covalently linked via a serine or threonine residue to the plasmid DNA.

Electron microscopy of the kalilo plasmid To visualize both the DNA and terminal proteins ofkalilo, the poly-L-lysine spreading technique described by Williams (1977) was employed. The samples of kalilo, with intact terminal proteins, also contained total mitochon-

200 drial DNA. However, the plasmid was about 5 to 10-fold more abundant than the mitochondrial DNA in these preparations and was identifiable by size relative to a bacteriophage 9 174 standard. The kalilo plasmid, treated with proteinase K, was prepared by electroelution. The electron micrographs of intact plasmid (Figs. 7B, C and D) show linear molecules ending in structures presumed to be the terminal proteins (small arrowheads). In Fig. 7 B, the two protein molecules (large arrowhead) appear to be linked to short segments of DNA constituting a possible replication fork. In Fig. 7C, approximately 0.2 gm of one end of the plasmid DNA appears to be thicker than the remainder of the molecule (possibly three strands) and terminates in a protein doublet (large arrowhead); also suggestive of a replicating element. No structures were visible at the plasmid termini in samples treated with proteinase K (Fig. 7A). These molecules also were not seen in preparations of mitochondrial DNA from a plasmid-free strain.

Discussion

In this report we have demonstrated that the kalilo senescence plasmid of N. intermedia has a covalentlylinked protein at both 5' termini. The susceptibility of this linkage to alkaline hydrolysis suggests that it is a phosphodiester bond (Daubert and Bruening 1984). Such a linkage has already been established conclusively for adenovirus (Rekosh et al. 1977) and Bacillus subtilis q529 (Salas et al. 1978). The 55 kDa terminal protein of adenovirus, linked by a phosphodiester bond between a serine residue and the terminal deoxycytidine, can be released by incubation in 50 mMNaOH (Desiderio and Kelly Jr 1981). Similarly, the phosphodiester bond between the terminal deoxyadenosine and a serine residue of the 27 kDa bacteriophage q~29 terminal protein is labile to alkaline treatment (Hermoso and Salas 1980). The only terminal protein from a fungal plasmid for which a size has been reported is the 28 kDa protein of the killer plasmid pGKL1 from Kluyveromyces lactis (Stam et al. 1986). Otherwise, the presence of covalently-linked terminal proteins has been inferred on the basis of differing electrophoretic mobilities of terminal restriction fragments or whole plasmids before and after proteinase K treatment, and resistance of these plasmids to 5' specific exonucleases. Elements with presumptive terminal proteins include the maize mitochondrial plasmids S 1 and $2 (Kemble and Thompson 1982), a Brassica mitochondrial plasmid (Erickson et al. 1985) and plasmids from filamentous fungi (Kistler and Leong 1986; Meinhardt et al. 1986; Giasson and Lalonde 1987; Duvell et al. 1988). The kalilo plasmid thus joins a growing class of linear DNA elements which have, or are presumed to have, covalentlylinked terminal proteins. The kalilo terminal protein has an estimated molecular weight of 120,000. Such a large protein would require at least 3,000 bp of DNA coding capacity. An integrated form of the kalilo plasmid has now been entirely sequenced (Chan and Bertrand 1988). There are only two major open reading frames, encoding proteins with do-

mains homologous to an RNA and a DNA polymerase, respectively (Chan and Bertrand, submitted). Their respective sizes are 93 kDa and 103 kDa. Therefore, if the 120 kDa terminal protein is encoded by the plasmid itself, it must consist of all, or at least part of, one of these two proteins. An alternative hypothesis would require recruitment of some "host" protein by kalilo to serve as a terminal protein. Its origin is currently under investigation. The terminal proteins of adenovirus and q~29prime the replication of their respective genomes (Salas 1983; Stillman 1983). However, efficient replication of adenovirus and q529 depends on a number of ancillary factors in addition to the terminal protein and DNA polymerase. For q)29, three other viral-encoded proteins enhance elongation (Salas 1988). Four other proteins, in addition to the terminal protein and DNA polymerase, are required for adenovirus replication; one is viral-encoded while the remainder are cellular (Kelly et al. 1988). The replication of other linear elements, including kalilo, is probably primed in a similar manner. The electron micrographs presented here (Fig. 7) are suggestive of a protein-primed initiation of replication from the termini of the kalilo element. However, more substantive evidence for this concept is not yet available. In addition, the limited coding capacity of the linear mitochondrial plasraids suggests that their replication may also be dependent on host-encoded mitochondrial factors. The role of the terminal protein in replication can best be addressed with in vitro assays using purified components, or by studying mutants which have altered proteins. Neither of these options are yet available for the kalilo plasmid of Neurospora. A unique facet of the biology of kalilo is its ability to integrate into the mitochondrial genome of Neurospora and generate very large inverted repeats of mitochondrial DNA flanking the insert (Dasgupta et al. 1988; Chan et al., submitted). All of the kalilo insertion elements analyzed to date are nearly full length, with terminal deletions of less than 20bp (Dasgupta et al. 1988). The terminal proteins are also implicated in this process by their proximity to the terminal nucleotides which are presumed to constitute "active sites" for integration into the mitochondrial chromosome (Chart et al., submitted).

Acknowledgements. We would like to thank Dr. E. Daub for advice on the poly-L-lysinespreading technique, C. Ackerly for rotary shadowingthe electronmicroscopegrids, and B. B. Searsfor critical assessment of the manuscript. This work was supported by grants from the Natural Sciences and EngineeringResearch Council and the National Research Councilof Canada to HB. References

Bertrand H, Chan BS-S, GriffithsAJF (1985) Cell 41:877-884 Bertrand H, Griffiths AJF, Court DA, Cheng CK (1986) Cell 47:829-837 Chan BS-S, Bertrand H (1988) Genome 30 (Suppl 1):318 Dasgupta J, Chan BS-S, Bertrand H (1988) Genome 30 (Suppt 1):318 Daubert SD, BrueningG (1984) MethodsVirol 8:347-379 Davis RH, deSerres FJ (1970) MethodsEnzymol17A:79-143

201 Desiderio SV, Kelly TJ Jr. (1981) J Mol Biol 145:319-337 Duvell A, Hessberg-Stutzke H, Oeser B, Rogmann-Backwinkel P, Tudzynski P (1988) Mol Gen Genet 214:128-134 Erickson L, Beversdorf WD, Pauls KP (1985) Curr Genet 9:679-682 Garber RC, Lin JJ, Yoder OC (1986) Mitochondrial plasmids in Cochliobolus heterostrophus. In: Wickner RB, Hinnebusch A, Lambowitz AM, Gunsalus IC, Hollaender A (eds) Extrachromosomal elements in lower eukaryotes. Plenum, New York London, pp 105-118 Giasson L, Lalonde M (1987) Curr Genet 11:331-334 Griffiths AJF, Bertrand H (1984) Curr Genet 8:387-398 Hermoso JM, Salas M (l 980) Proc Natl Acad Sci USA 77:6425-6428 Hirochika H, Sakaguchi K (1982) Plasmid 7:59-65 Kelly TJ, Wold MS, Li J (1988) Adv Virus Res 34:1-42 Kemble R J, Thompson RD (1982) Nucleic Acids Res 10:8181-8190 Kistler HC, Leong SA (1986) J Bacteriol 167:587-593 Laemmli UK (1970) Nature 227:680-685 Lambowitz AM (1979) Methods Enzymol 59:421-433 Lizardi PM, Luck DJL (1971) Nature New Biol 229:140-142 Maniatis T, Fritsch EF, Sambrook J (1982) Molecular cloning: a laboratory manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York

Meinhardt F, Kempken F, Esser (1986) Curr Genet 11:243-246 Pring DR, Levings CS III, Hu WWL, Timothy DH (1977) Proc Natl Acad Sci USA 74:2904-2908 Pring DR, Lonsdale DM (1985) Int Rev Cytol 97:1-46 Rekosh DMK, Russell WC, Bellet AJD (1977) Cell 11:283-295 Rieck A, Griffiths AJF, Bertrand H (1982) Can J Genet Cytol 24:741-759 Salas M, MeUado RP, Vifiuela E, Sogo JM (1978) J Mol Biol 119:269-291 Salas M (1983) Curr Topics and Microbiol and Immuno1109:89-106 Salas M (1988) Curr Topics and Microbiol Immunol 136:71-88 Stare JC, Kwakman J, Meijer M, Stuitje AR (1986) Nucleic Acids Res 14:6871-6884 Stillman BW (1983) Cell 35:7-9 Tamanoi F, Stillman BW (I982) Proc Natl Acad Sci USA 79:22212225 Vogel HJ (1947) Microbiol Genet Bull 13:42-43 Williams RC (1977) Proc Natl Acad Sci USA 74:2311-2315

C o m m u n i c a t e d b y C. S. Levings I I I