Genetic and Physicochemical Characterization of Escherichia coli Strains

Transcription

1 Proc. Nat. Acad. Sci. USA Vol. 69, No. 6, pp , June 1972 Genetic and Physicochemical Characterization of Escherichia coli Strains Carrying Fused F' Elements Derived from KLF1 and F57 (sucrose gradient centrifugation/extrachromosomal DNA) NEIL WILLETTS* AND FERNANDO BASTARRACHEA Departamento de Gen6tica y Biologia Celular, Centro de Investigaci6n y de Estudios Avanzados del I. P. N. M6xico 14, D. F. Mexico Communicated by Donald A. Glaser, February 9, 1972 ABSTRACT Although the two F' elements, F57 (carrying his+) and KLF1 (carrying leu+), cannot normally coexist in the same cell, crosses between RecA- strains of Escherichia coli carrying these two elements gave rise, at a low frequency, to progeny carrying both the his + and leu + markers in an extrachromosomal state. Genetic studies showed that the his+ and leu+ markers were now linked, and centrifugation studies of F' DNA isolated from these strains confirmed the presence of a new species of DNA molecule of altered size. It is proposed that the his+ and leu + genes in these strains are covalently linked on a single "fused" F' element. Normally, two F' elements cannot coexist in the same cell (1-3), and similarly, replication of an F' element is inhibited in an Hfr strain (1, 4, 5). This property is called incompatibility and may stem from the same control mechanism that limits the number of copies of an F' element to 1-2 per chromosome (6, 7). In all cases so far described in which two F or F' elements have been found in the cell, both are probably integrated into the bacterial chromosome (sometimes in an unstable manner) and replicated as a part of it; these cases include the double male (8) and F' derivatives of an Hfr strain (4, 9-11). Integration of an F' factor into the chromosome can occur at a low frequency even in a RecA- strain (a strain deficient in recombination) (11, 12). We have studied the rare His+ Leu+ progeny derived from F57 (an Fhis+) and KLF1 (an Fleu+) in a RecA- strain. These two F' elements were chosen since it was found that they could be easily separated by sucrose gradient centrifugation even when both were labeled with tritium. This was important since our intention was to investigate whether His+ Leu+ cells, apparently carrying both F57 and KLF1, contained two plasmid DNA molecules, one or both of which might have suffered a deletion, or a single plasmid DNA molecule. The genetic properties of the His+ Leu+ strains obtained indicated that the his+ and leu+ markers were now carried by a single F' DNA molecule. In agreement with this, the extrachromosomal DNA molecules carried by these strains were different in size from either F57 or KLF1. A model for their formation is proposed. While this work was in progress, Press et al. (13) reported the isolation of similar fused F' episomes. Abbreviation: CCC, covalently-closed circular form of DNA. * Present and permanent address: MRC Molecular Genetics Unit, Department of Molecular Biology, University of Edinburgh, Edinburgh, Scotland MATERIALS AND METHODS Bacterial Strains. The strains used are listed in Table 1. JC1553 (14) was used as host for the F' episomes. F57 is an Fhis isolated from Hfr AB311 by.t. Takano, and KLF1 was isolated from the HfrH strain AB259 by Low (15). Phage Strains and Techniques. Lysates of the male-specific phages MS2 and M12 were made by a confluent plate lysis technique. Sensitivity to these phages was determined as described (16). Bacteriophage X, was obtained from Dr. M. Monk. Media and Buffers. The bacterial growth media have been described (17). Cultures of strains carrying F' episomes to be used for DNA extractions were grown in a medium based on M9 medium (18) with the addition of 0.4% glucose, 0.01 M MgSO4, 0.2 ug/ml of thiamin, and 20 ug/ml of L-amino acids. EM9 medium contained 0.25% casamino acids instead of individual amino-acid supplements. TES is described by Bazaral and Helinski (19). TES-Cl is similar, but contains 0.3 M NaCl. ['4C]Thymidine [2-'4C]dT; 55 A&Ci/ml; 60.5 Ci/mol) and [8H]thymidine [3H]dT-6-T(n); 1 mci/ml; 5 Ci/mmol) were obtained from the Radiochemical Centre, Amersham, England. Segregation. The frequency of segregation or loss of an F' factor from its host strain was determined by inoculation of an L broth culture from a single colony grown on the appropriate minimal medium. After overnight growth, dilutions were plated to obtain single colonies on L plates. These were patched on L plates and, after an 8 hr incubation, they were replicated onto minimal plates to determine whether the His+ or Leu+ phenotype had been lost. Acridine Orange Treatment. Dilutions of 10-4 into L broth at ph 7.8 containing 10 gg/ml acridine orange were made from overnight cultures in minimal medium. These cultures were shaken at 370 for 20 hr, and the proportion of the population that had lost the F' factor was determined as described under Segregation. Matings. Matings in liquid medium were done in L broth at 370, with a donor: recipient ratio of 1:10 and a mating period of 30 min. Exponential cultures of the F' donor strains were obtained by diluting overnight cultures grown in M9 minimal medium 25-fold in L broth and shaking at 370 until these reached about 2 X 108 cells/ml. When an F' strain was

2 1482 Genetics: Willetts and Bastarrachea Proc. Nat. Acad. Sci. USA 69 (1972) TABLE 1. Bacterial strain8 Strain Episome Arg His Leu Met Try Pro Str Spc rec MX173 F R S recal MX176 KLF R S recal MX177 KLF S S reca66 MX178 EDFH R S recal ED332 F S R reca66 JC R S recal JC R R recal JC S R reca46 MX173, MX176, and MX178 are derivatives of JC1553. JC4121 is a SpcR derivative of JC1553. JC5466 is a reca56 derivative of JC5455 (16). to be used as recipient, cultures were phenocopied by shaking them for 24 hr in M9 minimal medium. These cultures were diluted 10-fold into L broth to obtain densities of 1-2 X 108 cells/ml and were used immediately. Donor populations were always tested to determine what proportion of the cells still carried the appropriate F' factor(s), and the results were corrected accordingly. Preparation of X [14C]DNA. AB1157 (17) was grown in 25 ml EM9 medium to about 4 X 108 cells/ml. Bacteriophage Xv was then added at a multiplicity of infection of 5. After 10 min, deoxyadenosine (1.25 ml of a 5 mg/ml solution), thymidine (0.01 ml of a 1 mg/ml solution), [14C]thymidine (0.1 ml of a 55 MCi/ml solution), casamino acids (0.5 ml of a 10% solution), and glucose (0.06 ml of a 40% solution) were added. After incubation for 2 hr at 370 the cells lysed. The lysate was cleared by centrifugation twice at low speed, and the X phage (5 X 1010 plaque forming units/ml) was sedimented by centrifugation at 20,0 rpm for 30 min in the International SB110 rotor. The phage were resuspended in about 0.5 ml buffer by gently shaking overnight at 4. The DNA was extracted from the phage by a procedure essentially similar to that described by Thomas and Abelson (20). Phenol remaining in the DNA solution was removed by 1 a. 80 loo - b FIG. 1. Sucrose gradient centrifugation of F57 and KLF1. [3H]KLF1 together with either [14C]F57 (a) or [3H]F57 and X [14C] DNA (b) was layered on top of the gradient and centrifuged for 25 min. 0-O, 8H; * 0,t 14C. dialysis against several changes of TES-Cl at 4. The DNA solution was stored at 40 with a drop of chloroform. This solution contained 350 ug/ml DNA and 250,0 cpm/ ml. For use in sucrose gradients, it was diluted 1: 10 in TES-Cl and heated at 65 for 1 min, followed by rapid cooling, to ensure linearity of the DNA molecules. 25 ul, which contains 6 cpm and <1 ug DNA [as recommended by Burgi and Hershey (21) ], were layered onto gradients. Preliminary tests showed a single peak in such gradients. Preparation of F' DNA. Cultures were grown in minimal medium lacking histidine or leucine or both, as appropriate, to ensure that most cells harvested still carried the F' episome. Labeling, with 10 MCi/ml ['H]thymidine or 1 MACi/ml [14C] thymidine, was done during exponential growth between about 108 and 4 X 108 cells/ml, in the presence of 250 ug/ml deoxyadenosine. DNA was then extracted from the cells by the procedure described by Bazaral and Helinski (19) and Nisioka, Mitani, and Clowes (22). 0.8 ml of a cell lysate from about 5 X 109 cells was mixed with 5.2 g CsCl, 3 ml H20, and 1.6 ml ethidium bromide solution (7 Mg/ml in 0.1 M phosphate buffer, ph 7.0), and the mixture was centrifuged at 44,0 rpm in a Spinco 65 rotor for 36 hr. The tubes were then punctured, and 8-drop (0.15-ml) fractions were collected. Samples of 25 Ml from these fractions were applied to squares of Whatman 3 MM filter paper (2 X 2 cm) and dropped into 5% C13CCOOH at room temperature. After the papers had been washed twice with 5%O C13CCOOH, and once with each of 80% ethanol, pure ethanol, and diethyl ether, they were dried and immersed in 5 ml of scintillation fluid [3.5 liters toluene, 14 g 2.5-diphenyloxazole (PPO), and 0.35 g [1,4 - bis (4 - methyl phenoxazolyl) - benzene (POPOP)]. Sucrose Gradient Centrifugation. The two or three fractions from the ethidium bromide-cscl density gradient that contained the F' satellite DNA (usually separated by about eight fractions from the chromosomal DNA peak) were pooled, and dialyzed at 40 for 6 hr against four changes of a 1-volume excess of TES-Cl to remove the ethidium bromide and CsCl (22). Samples of 50 Ml were then layered onto 5 ml of 5-20% linear sucrose gradients made up in TES-Cl in International 14.5 X 96 mm polyallomer tubes. After the tubes had been filled with paraffin oil (Nujol), they were centrifuged at 40,0 rpm at 20 in an International SB283 rotor. The tubes were then punctured, and 5-drop (0.15-ml) fractions were collected directly into 10 ml of scintillation

3 Proc. Nat. Aca. Sci. USA 69 (1972) fluid (250 ml toluene, 750 ml ethyl cellosolve, 17.5 g PPO, and 140 g naphthalene). RESULTS Properties of the Original F' Episomes, F57 and KLF1. DNA molecules of F57 and KLF1 could easily be distinguished from one another in the same sucrose gradient even if both were labeled with tritium (Fig. 1). The fast-moving peaks containing most of the DNA presumably represent the covalently-closed circular (CCC) forms, since the F' DNA had been isolated on this basis as a satellite peak after CsClethidium bromide density gradient centrifugation. Further evidence for this was obtained by storing the DNA preparations at 40 for several days and then repeating the sucrose gradient centrifugations (Fig. 2). In both cases, the fastmoving peaks were converted to slower sedimenting forms presumably representing nicked circles (23, 24). The third, slowest-moving peak seen in gradients containing the large KLF1 episome may result from further degradation to linear, and perhaps fragmented, molecules: a similar slow-moving peak was formed in a progressive fashion on treatment of nicked-circular F57 DNA (Fig. 2c) with low concentrations of DNase. The ratios of the distances sedimented by the peaks containing nicked circles and those containing presumptive linear molecules were 1.7 (KLF1) and 1.5 (F57 + DNase), different from the ratio of 1.14 found by Bode and Kaiser (25) for similar forms of the smaller X DNA molecule. Loss of the CCC forms of F57 and KLF1 was almost complete after about 10 days of storage. This is very rapid compared to the period of months necessary for loss of the CCC form of ColEl DNA (23), but the large sizes of the F57 and KLF1 DNA molecules may accelerate this process. Assuming that the velocity of sedimentation in a linear sucrose gradient is proportional to the sedimentation coefficient and that the sedimentation coefficient of linear X DNA is 34 S (21), the sedimentation coefficients of the CCC forms of F57 and KLF1 were estimated to be 97 S and 132 S, respectively. Fused F' Episomes in E. coli 1483 C. 60- E 40 A~ 2, ; 20 1 C. F57 (day 9) f. KLF I (day 9) so A FIG. 2. Degradation of F' DNA on storage. [3H]F57(a-c) and [3H]KLF1 DNA (d-f) were stored at 40 for the number of days indicated. Samples were then layered on sucrose-gradients together with X [14C] DNA and centrifuged for 25 min. O-O, 3H; the positions of the X [14C]DNA peaks are indicated by arrows. The numbers indicate the percent of 3H label in the peaks. JC1553 was used as host strain in these experiments, and the F57 derivative of this, MX173, proved to be a surprisingly poor donor of his+ (Table 2). This may be due to a defect carried by F57 and expressed only in JC1553, since when F57 was transferred from MX173 to JC5466 it regained its usual high transfer ability. MX173 was also correspondingly less sensitive to male-specific phage. The donor properties of KLF1 (in MX176) were not affected. TABLE 2. Characteristics of the His+ Leu+ strains % Segregationt % Curingt MS2t Progeny per 1 donor cells and Strain* Hist Leut His Leu His Leu M12 His+ % Leu+ Leu+ % His+ MX S 8 X MX R <5 X MX S 7.4 MX MX S MX S MX S MX S X 10-5 MX R <5 X 10-5 <5 X 10-5 MX R <5 X <5 X 10-5 * All strains are derivatives of the RecA strain JC1553; all were UV'. MX185, MX186, and MX187 were derived from crosses between ED332 (donor) and MX176 (recipient). MX188 and MX189 were derived from crosses between MX177 (donor) and MX173 (recipient). MX191 and MX193 were derived from crosses between MX177 (donor) and MX173 (recipient). t These characteristics indicate episomal markers. t The techniques used are described in Methods. Matings with JC4121 were performed as described in Methods, contraselecting [SpcR]. 1 progeny from each cross were tested for coinheritance of the other marker by a replica-plating method. The area of lysis remained turbid.

4 1484 Genetics: Willetts and Bastarrachea Construction and Properties of His+ Leu+ Derivatives. These strains were constructed either by crossing ED332 (carrying F57) and MX176 (the KLF1 derivative of JC1553), or MX177 (carrying KLFl) and MX173 (the F57 derivative of JC- 1553). Donors were used in exponential phase and recipients (the JC1553 derivatives) in stationary phase. His+ Leu+ [StrR] progeny were selected, and these clones, which were obtained at a low frequency (<10-4 per donor cell), were patched on the same selective medium. Only a proportion of these patches grew, perhaps indicating that a secondary event was necessary for the formation of a stable His+ Leu+ clone. These were purified and characterized. Three clones, MX186, MX187, and MX188, had similar properties. The His+ Leu+ phenotype was stable during growth in broth, and on treatment with acridine orange, both phenotypes were lost together in all cases (Table 2). In matings, his+ and leu+ were both transferred at about the same frequency (that characteristic of KLF1), and all His+ progeny were Leu+ and vice versa (Table 2). Genetically, then, all three strains seemed to contain a single F' episome carrying both his+ and leu+. MX186, carrying an F' episome tentatively designated MXF1, was selected for study of its F' DNA. A further clone, MX185, was similar to these three clones except that it lost both markers together at a high frequency (Table 2). The F' DNA from this strain, tentatively designated MXF2, was also examined. Other clones, such as MX189, apparently still carried an autonomous F57 episome, but the leu+ marker, perhaps together with some part of F from KLF1, seemed to be integrated into the chromosome (Table 2). These strains were not further investigated. Characterization of MXF1 and MXF2. F' ['H ]DNA was obtained by CsCl-ethidium bromide density gradient centrifugation as described in Methods; it was used for mixedlabel sucrose gradients with X (not shown), F57, or KLF1 [14C]DNA (Fig. 3, a-d). Both MXF1 and MXF2 gave a three-peak pattern similar to that observed with KLF1 (Fig. 2). Upon storage of the F' DNA at 40 for 5 and 11 days, and recentrifugation with X [14C]DNA, the fastest-sedimenting peaks decreased in size at a rate similar to the F57 and KLF1 Proc. Nat. Acad. Sci. USA 69 (1972) CCC DNA peaks: they therefore contain CCC DNA. No new peaks appeared, indicating that the two slower-moving peaks represent the degradation products of the CCC form, i.e., nicked circular and linear (or fragmented) molecules, respectively. Why these slower-moving peaks should contain so much of the F' DNA after only 1 day's storage is not clear. The CCC forms of MXF1 and MXF2 sedimented at different velocities than the CCC forms of either of the parental F' episomes. MXF1 sedimented faster than either F57 or KLF1, and MXF2 sedimented at an intermediate rate (Fig. 3, a-d). The sedimentation coefficients were estimated by comparison with X [14C]DNA to be 153 S (for MXF1) and 108 S (for MXF2). These results show that a new species of CCC F' DNA of altered size was present in cells carrying the presumptive fused Fhis+ leu+ episomes, and in the two cases tested, these new species differed in size from each other. Transfer-Deficient His+ Leu+ Strains. If the His+ Leu+ strains arise by recombination between F57 and KLF1, then a part of one or both F factors is probably deleted, perhaps as a requirement for stability of the resultant fused F' episome. Such a deletion can be deduced in the case of MXF2 from its size, which was intermediate in comparison with the parental F' episomes. An attempt was made to demonstrate deletion by genetic means, with a transfer-deficient mutant of F57 carrying the recessive mutation traj90. This mutant episome was made by recombination between F57 and Flac traj90 (16, 26), and is designated EDFH1 in accordance with the nomenclature recommended in (16). His+ Leu+ strains derived from EDFH1 and KLF1 should be transfer-deficient if the traj+ allele is deleted from KLF1 in formation of the fused F' episome. His+ Leu+ strains were isolated from crosses between MX177 (carrying KLF1) and MX178 (carrying EDFH1) as described above. One of these, MX193, was similar to the His+ Leu+ strains described above in that both the His+ and Leu+ phenotypes were lost together on treatment with acridine orange (Table 2). MX193, however, was transfer-deficient, showing that the traj+ allele of KLF1 had been deleted E oa.a.C-F57+3H-MXF2 IC C.'C-F57+3H-MXFI e'cefih-x3 6C0C 610 4CD0 w w Il l E -sot , '0 :2 n l b. '4C-KLFI+3H-MXF2. d. 'C-KLFI+3H-MXFI f4@c-klfi+3h-mxf3 0 6C oo 3 IC DO I.& -_ I I iu FIG. 3. Mixed gradients of fused and parental F' episomes. The mixtures of F' DNA preparations indicated in the figure were layered on sucrose gradients and centrifuged for 24 min (MXF2), 15 min (MXF1), or 18 min (MXF3). O-O, 8H; 14C.

5 Proc. Nat. Acad. Sci. USA 69 (1972) in formation of the His+ Leu+ strain. Since the donorability of wild-type F57 is reduced in JC1553, the F' episome in MX193 was transferred to JC5466, using the techniques described in (26); it was found to remain transfer-deficient. Other His+ Leu+ strains such as MX191 seemed similar to strains such as MX189 described above, probably carrying an autonomous EDFH1 episome and a chromosomal leu+ marker (Table 2). F' DNA isolated from [3H]MX191 was, in fact, indistinguishable from [14C]EDFH1 DNA in a sucrose gradient. The F' DNA from MX193, tentatively designated _MXF3, gave three peaks in a sucrose gradient with X [14C]DNA, similar to the patterns obtained with the other fused F' episomes. Again, the fastest-moving peak was gradually lost on storage of the DNA, showing that it represented the CCC form, the radioactivity appearing in the slower-moving peaks. The sedimentation coefficient of the CCC form was calculated to be 147 S, and mixed-label sucrose gradients with 3H-fused F' DNA and either ['4C]EDFH1 or [14C]KLF1 DNA confirmed that it sedimented faster than the CCC forms of either parental F' DNA (Fig. 3e, andf). DISCUSSION Trhe results show that autonomous F57 and KLF1 episomes cannot coexist in the same cell but that stable His+ Leu+ clones could be obtained by formation of a single "fused" Fhis+ leu+ episome. The resultant genetic linkage of the his+ and leu+ markers was correlated with the appearance in the cell of a new species of F' DNA molecule. The F' DNA molecules isolated from three such strains showed a three-peak pattern similar to KLF1 when subjected to sucrose gradient centrifugation, the fastest-moving peak representing the CCC form. The molecular weights of the fused F' episomes in MX185 (MXF2), MX186 (MXF1), and MX193 (MXF3) were estimated from the sedimentation coefficients of the CCC forms by application of the formula of Bazaral and Helinski (23); they were 155 X 106, 330 X 106, and 3 X 106, respectively. The molecular weights of F57 and KLF1, similarly determined, were 115 X 10' and 240 X 106, respectively. However, the estimates for KLF1, MXF1, and MXF3 are questionable since the formula may. not hold for such high molecular weights. Although the ratios of the sedimentation coefficients of the CCC forms and the nicked circular forms predicted theoretically by Fukatsu and Kurata (27) were confirmed experimentally by Bazaral and Helinski for DNA molecules of molecular weights up to to 108 (23), the ratios for KLF1, MXF1, and MXF3 (1.9, 2.3, and 2.5, respectively) were much higher than the theoretically predicted values. The ratios for the smaller F57 and MXF2 molecules, on the other hand, were both 1.6, close to the theoretically predicted values. Although the molecular weight of MXF2 was clearly smaller than the sum of the molecular weights of F57 and KLF1, it is more difficult to assert this for MXF1 and MXF3. However, MXF3 was shown genetically to have lost some F DNA, including the traj+ allele of KLF1. It seems likely, then, that these F' episomes arise by two recombination events between the two parental F' DNA molecules. The first would give a DNA molecule with F57 and KLF1 in tandem. Such molecules were not found, so that as would be expected, this situation must be unstable. A second recombination event would then lead to deletion of the gene(s) causing this in- Fused F' Episomes in E. coli 1485 stability [the incompatibility gene(s) ], but retention of both the his+ and leu+ markers. The two recombination events might be between nonhomologous regions of DNA, and neither involved the reca + product since RecA- strains were used. They may be similar to other illegitimate recombination events (28). The size of the deletion would be expected to vary, depending upon the relative position of the two recombination events, so that this mechanism can account. for the formation of the differently-sized F his+ leu+ episomes observed. Other F genes could be deleted simultaneously, as found for the fused F' episome in MX193, which had lost the traj+ allele of KLF1. Deletion of incompatibility genes together with transfer genes has also been observed in the formation of F' derivatives of an HFr strain (29). We are grateful to Dr. Jaime Martuscelli for advice on the centrifugation procedures, and to Esther Tam for invaluable technical assistance. 1. Scaife, J. & Gross, J. D. (1962) Biochem. Biophys. Res. Commun. 7, De Haan, P. G. & Stouthamer, A. H. (1963) Genet. Res. 4, Echols, H. (1963) J. Bacteriol. 85, Maas, R. (1963) Proc. Nat. Acad. Sci. USA 50, Dubnau, E. & Maas, W. K. (1968) J. Bacteriol. 95, Jacob, F. & Wollman, E. L. (1961) in Sexuality and the Genetics of Bacteria (Academic Press, New York and London). 7. Revel, H. R. (1965)J. Mol. Biol. 11, Clark, A. J. (1963) Genetics 48, Cuzin, F. & Jacob, F. (1967) Ann. Inst. Pasteur, Paris 113, Bastarrachea, F. & Clark, A. J. (1968) Genetics 60, DeVries, J. K. & Maas, W. K. (1971) J. Bacteriol. 106, Broda, P. & Meacock, P. (1971) Mol. Gen. Genet., 113, Press, R., Glansdorff, N., Miner, P., DeVries, J., Kadner, R. & Maas, W. K. (1971) Proc. Nat. Acad. Sci. USA 68, Clark, A. J. & Margulies, A. D. (1965) Proc. Nat. Acad. Sci. USA 53, Low, B. (1968) Proc. Nat. Acad. Sci. USA 60, Achtman, M., Willetts, N. S. & Clark, A. J. (1971) J. Bacteriol. 106, Willetts, N. S., Clark, A. J. & Low, B. (1969) J. Bacteriol. 97, Adams, M. H. (1959) in Bacteriophages (Interscience Pubs. Inc., New York). 19. Bazaral, M. & Helinski, D. R. (1968) J. Mol. Biol. 36, Thomas, C. A. & Abelson, J., (1967) in Procedures in Nucleic Acid Research, eds. Cantoni, G. L. & Davies, D. R. (Harper and Row, Scranton, Pa.), pp Burgi, E. & Hershey, A. D. (1963) Biophys. J. 3, Nisioka, T., Mitani, M. & Clowes, R. C.(1970) J. Bacteriol. 103, Bazaral, M. & Helinski, D. R. (1968) Biochemistry 7, Cohen, S. N. & Miller, C. A. (1970) J. Mol. Biol. 50, Bode, V. C. & Kaiser, A. D. (1965) J. Mol. Biol. 14, Achtman, M., Willetts, N. S. & Clark, A. J. (1972) J. Bacteriol., in press. 27. Fukatsu; M. & Kurata, M. (1966) J. Chem. Phys. 44, Franklin, N. C. (1971) in The Bacteriophage X, eds. Hershey, A. D., Kaiser, A. D., Echols, H. & Campbell, A. (Cold Spring Harbor Laboratories), pp Mass, W. K. & Goldschmidt, A. D. (1969) Proc. Nat. Acad. Sci. USA