|Home | About | Journals | Submit | Contact Us | Français|
We report a detailed restriction map of the bacteriophage T4 genome and the alignment of this map with the genetic map. The sites cut by the enzymes BglII, XhoI, KpnI, SalI, PstI, EcoRI and HindIII have been localized. Several novel approaches including two-dimensional (double restriction) electrophoretic separations were used.
Detailed restriction maps of a number of viral genomes (Roberts 1976) have played an important role in the recent analyses of these systems. However, a detailed physical map has not yet been determined for one of the most intensively investigated bacterial viruses, bacteriophage T4. Two features of the T4 genome have hampered attempts to define the restriction map. First, because the DNA contains glucosylated 5-hydroxymethyl cytosine (HMC) rather than cytosine, it is resistant to restriction-enzyme cleavage. Second, the genome is very large (about 166 kb, Wood and Revel 1976). We have circumvented the first problem by using a multiple mutant of T4 that incorporates cytosine rather than HMC into its DNA (Snyder et al. 1975; Wilson et al. 1977). To cope with the large genome size we have used a new two-dimensional method for separation of the numerous restriction fragments (O’Farrell, in preparation), and have applied some novel approaches to the detection and ordering of these restriction fragments.
The positions of 231 restriction enzyme cutting sites, including all the sites cut by the restriction enzymes BglII, XhoI, KpnI, SalI, and PstI, and most of the sites cut by EcoRI and HindIII are reported here. The restriction map has been aligned with the genetic map by localizing, on the restriction map, the positions of a number of cloned fragments of T4 DNA. The positions of these cloned fragments on the genetic map had previously been determined by marker rescue procedures (Mattson et al. 1977). An additional reference point is a unique BamHI restriction site (Takahashi et al. 1979), whose location on the T4 genetic map has been determined (Wilson et al. 1980).
Two independent published efforts have identified positions of SalI, SmaI and KpnI restriction sites and some of the BglII sites but did not include alignment to the genetic map (Kiko et al. 1979; Rüger et al. 1979; Carlson and Nicolaisen 1979). Additionally, extensive analysis of DNA clones localized a number of EcoRI sites (Wilson et al. 1977). As will be discussed below, our results substantially agree with these reports, as well as with other recent data.
A T4 phage carrying four mutations, [g56amE51, g42amC87, ΔdenB NB5060, alcW7] was propagated on a permissive bacterial host QD Su3. Growth of this bacteriophage, designated alcW7, on a non-suppressing host ED8689 (sup°, trpR−,hsdR− derivative of W3350) gave a high titer of phage containing cytosine. Unless specified otherwise, T4 DNA refers to the cytosine-containing DNA isolated from this phage. E. coli 5K(R1) was used to check the approximate extent of substitution of C for HMC (Wilson et al. 1977). The C containing phage had a plating efficiency of 10−5 on this strain and appears to have no residual HMC. Bacterial strains and T4 alcW7 where obtained from G.G. Wilson (Wilson et al. 1977).
Endonuclease EcoRI was generously provided by Pat Bedinger and Janet Natzle. All other restriction endonucleases were obtained from New England Biolabs. The T4 DNA polymerase was a kind gift from Ursula Hibner. The polymerase was prepared according to Morris et al. (1979), had a specific activity of about 18,000 units per mg and was free of measurable nuclease contamination. Most notably no 5′ exonuclease was detected in this polymerase preparation. Agarose was Seakem ME grade, α 32P dATP was purchased from New England Nuclear.
All the restriction enzymes we have examined work well in a single buffer (designated TA buffer) having the following composition: 33 mM Tris acetate pH 7.9, 66 mM potassium acetate, 10 mM magnesium acetate, 0.5 mM dithiothreitol and 100 µg/ml bovine serum albumin (nuclease free). No EcoRI* activity or other “star” activities were observed even after extensive over-digestions of ØX174 DNA in TA buffer. However, one site in T4 DNA was cut very slowly by BglII although it may represent a “star site” (a sequence related to that preferred by the restriction enzyme), there is no evidence that the specificity of the enzyme is decreased in the buffer used here. Direct tests showed that EcoRI was more stable and more active in TA buffer than in one of the frequently used EcoRI buffers (100 mM Tris C1 pH 7.2, 100 mM NaC1 5 mM MgCl2). Other enzymes gave levels of activity roughly equal to the nominal activities quoted by the supplier.
Use of T4 DNA polymerase, which is also active in TA buffer, provides an extremely convenient method for labeling restriction fragments. In the absence of deoxyribonucleoside triphosphate, the 3′ exonuclease activity of the polymerase removes nucleotides from 3′ ends of the DNA fragments and these are replaced with labeled nucleotides in a polymerization reaction initated by addition of radioactive substrates (Englund 1972). In a typical reaction 100 ng of T4 DNA in 10 µl was cut with a restriction enzyme; 100 ng of T4 DNA polymerase (in 1 µl of 0.2 M KPO4, pH 7.5, 50% glycerol, 1% mercaptoethanol) was then added and the reaction incubated for about 2 min at 37 °. Finally, α32P dATP was added (10 µM) along with the other three unlabeled deoxyribonucleotide triphosphates (100 µM) and incubation was continued for about 4 min to replace about 100 terminal nucleotides. The labeling reaction was terminated by heating to 70 ° for 5 min. The detailed characteristics of this labeling reaction will be reported separately (O’Farrell, in preparation). It is important to note that the number of nucleotides replaced by labeled nucleoitdes can be regulated. Here, about 100 nucleotides at the termini of each restriction fragment have been replaced and, unless a subsequent restriction enzyme cut is made within this short labeled region, the label behaves as an end-specific label.
Restriction fragments labeled by the above procedure can be redigested with a second restriction enzyme immediately after the heat step. In this case, only the termini generated by the first digestion will be labeled. Such a digest with, for example, XhoI and PstI will be designated XhoI (32P)/ PstI, indicating that the termini generated by XhoI cleavage are labeled while the termini generated by PstI are not labeled.
To obtain partial digests, T4 DNA was treated with different levels of restriction enzyme and, after each digest was end labeled, it was analyzed on one dimensional gels. Selected levels of digestion were then analyzed by a two-dimensional scheme (described in Results and Discussion). Generally, on the basis of the one-dimensional separations, several digests were mixed to produce a fairly equal distribution among higher order partials and the smaller partials. Even though some enzymes, such as PstI, produced digests in which the starting material and/or the limit restriction fragments were vastly more abundant than the intermediates, the high sensitivity of autoradiography allowed adequate detection of the partials.
Restriction fragments were separated electrophoretically on flat bed gels of a suitable agarose concentration in TAE buffer (40 mM Tris acetate pH 8.1, 20 mM sodium acetate, 2 mM EDTA), at a voltage gradient of 2 V/cm (Sharp et al. 1973; Helling et al. 1974). Gels were stained with ethidium bromide and the fluorescence observed under ultraviolet illumination. For autoradiography, gels were dried thoroughly under vacuum and exposed directly to X-Omat (Kodak) X-ray film, with or without an intensifying screen (Kodak Lighting Plus).
Two-dimensional restriction analysis was performed using a new method which will be described in detail in a separate publication (O’Farrell, in preparation) but is outlined here. As described by Potter et al. (1977), restriction fragments separated according to size by agarose gel electrophoresis can be recovered, redigested with a second restriction enzyme and each fraction again separated according to size by agarose gel electrophoresis to generate a two-dimensional separation. To simplify this procedure, some investigators, after separating restriction fragments in the first dimension, have digested the DNA fragments with a second restriction enzyme within the agarose gel (Rosenvold and Honigman 1977; Kovacic and Wang 1979). The new two-dimensional method used here incorporates an electrophoresis step which brings together the second restriction enzyme and the separated fragments in a narrow zone stacked at the edge of the gel and thereby greatly improves both the in situ digestion and the subsequent separation. This method produces sharp spots, and allows the use of numerous restriction enzyme combinations which were previously impractical. When combined with end labeling and analysis of partial digests (Smith and Birnstiel 1976; Kovacic and Wang 1979), it greatly extends our capacity to determine restriction maps of complex genomes. Its application is illustrated more thoroughly under “Results” below.
Chemically activated paper (diazophenylthioether paper) was prepared according to a method described by Brian Seed (personal communication). T4 DNA was either cut with a single enzyme (BglII, XhoI, EcoRI, HindIII or PstI), with various pairwise combinations of these enzymes, or with a combination of all five enzymes. Each digest was separated on both a 0.5% agarose gel and a 1.5% agarose gel and the separated DNA fragments transferred to the activated paper according to the methods of Alwin et al. (1977), as subsequently modified (Wahl et al. 1979). As described by the latter authors, the same filters were hybridized in sequence with a series of labeled probes. Total T4 DNA and 6 fragments of T4 DNA cloned in pBR322 were labeled and used as probes.
The DNA fragments were labeled using T4 DNA polymerase, rather than the more common nick-translation method described by Rigby et al. (1977). For these reactions, DNAs to be labeled were first restricted, then, using the T4 polymerase reaction, about 800 bases at the termini were replaced with α32P deoxynucleotide triphosphates.
The substitution of cytosine by dHMC in T4 DNA is achieved by two phage encoded enzymes: The product of gene 42, a hydroxymethylase (HMase), generates dHMC from dCMP, while the product of gene 56, dCTPase, reduces the level of the more usual substrate for DNA synthesis. Bacteriophage T4 encoded nucleases endoII and endoIV are involved in host DNA breakdown, but will degrade T4 DNA containing cytosine (Kutter and Wiberg 1968). A variety of multiply mutant T4 strains capable of incorporating cytosine into viable phage have been constructed (Snyder et al. 1976; Morton et al. 1978; Wilson et al. 1977) and are referred to as C-T4 strains. The combination of mutations most effective in eliminating HMC are deficient in dCTPase, endoIV and HMase (Morton etal. 1978); an additional mutation, alc, is required to overcome a block affecting growth of cytosine containing T4 (Kutter et al. 1975; Snyder et al. 1976).
Because of the impact on restriction enzyme analysis, the choice of alternative alleles used for C-T4 strain construction must be considered. In particular, mutations in the denB locus, which encodes endoIV, confer no distinctive phenotype. Usually, to simplify identification, deletions which remove not only denB but a nearby selectable marker are used. For the map construction reported here, we chose a strain described by Wilson et al. (1977) [am E51 (dCTPase), am C87 (HMase), deletion denB rlI NB5060, alc W7]. This C-T4 strain grows unusually well, and its DNA can be cleaved to completion with restriction enzymes. As shown in Fig. 1, the deletion in this strain, rIIB NB5060, begins approximately 0.3 kb in from the N terminus of the rIIB gene and extends through the entire D region into ac--a total of approximately 3.2kb (Bautz and Bautz 1967; Depew etal. 1975; D. Pribnow, personal communication). Other endoIV deletions commonly used in C-T4 strains are also illustrated in Fig. 1.
Figure 2 and Tables 1–5 present a detailed physical map of the T4 genome. To facilitate comparison we have set the origin of the physical map at the origin of the genetic map. Methods used in map construction include: (1) one- and two-dimensional agarose gel analysis of multiple restriction-enzyme digests; (2) analysis of partial digestion products using a new two-dimensional technique; and (3) hybridization of identified cloned fragments to blots of one-dimensional gel analyses of various multiple enzyme digests. The following description is designed to illustrate the capacity, limitations and accuracy of the methods; more precise procedural details will be presented in a subsequent paper (O’Farrell, in preperation).
The number and size of fragments produced by various restriction enzyme digestions of T4 DNA were first determined by one-dimensional agarose electrophoresis using digests of the bacteriophage DNA’s of lambda and ØX174 as molecular-weight standards. Generally, the fragments were end-labeled to improve detection of small fragments and simplify analysis of multiple bands. This particularly aided in analysis of multiple digests when only the ends created by cleavage with one of the enzymes were labeled (see Materials and Methods). Figure 3 shows samples of these separations. As indicated on the figure, bands are numbered consecutively from high to low molecular weight. When two or more restriction fragments are found to comigrate they are distinguished by lower-case-letter subscripts. For those restriction enzymes making relatively few cuts, the fragments have also been given a second name – a letter designation. Here, fragments are ordered alphabetically according to their map position. For example, XhoI 8-C is the eighth band on a gel and the third on the map. Fragments can still be uniquely identified by either of the abbreviated names, (ie. XhoI 8 or XhoI C). This naming system greatly facilitates reference to both gel data and the map. Our band-numbering system coincides with that used by Rüger et al. (1977), but not with that used by Carlson et al. (1979) or Takahashi et al. (1978).
The one additional complication in terms of nomenclature results from the fact that most C-T4 strains carry a deletion of some sort near the origin (Fig. 1). This generally will not affect our letter designations. However, these deletions affect the sizes of fragments, and thus their number designation. Fragments have been numbered here according to our own analyses. The expected sizes of the fragments in the absence of any deletions are also indicated, using the data of Takahashi et al. (1978 and personal communication) to determine the approximate location of additional restriction enzyme sites.
For enzymes making many cuts, such as EcoRI and HindIII, we have simply designated the fragments according to the map position (in kb) of their distal end and their approximate size. It should be noted that the method of determining the location of these sites, discussed below, does not involve the identification of specific fragments with specific bands in the total EcoRI or HindIII digests. Where such identification is known from cloning data (cf. Wilson et al. 1977), the band number is also indicated in Table 3, using the assigments of Wilson et al. (1977).
After digestion of the T4 DNA to completion with one restriction enzyme, the resulting fragments were transferred to a slab gel for in situ digestion with the second enzyme and separation in a second dimension, as described in “Materials and Methods”. Different enzyme combinations were used in this way; thus, the BglII(32p)- XhoI separation shown in Fig. 4 represents a separation of end-labeled BglII fragments in the first dimension followed by XhoI cleavage and separation in the second dimension. In an identical fashion, we have carried out XhoI(32P)-BglII, BglII(32P)- EcoRI, EcoRI(32P)-BglII, XhoI(32P)-EcoRI, and EcoRI(32P)-XhoI separations to derive the initial map of the BglII and XhoI sites.
In principle, the deduction of map order from such a set of gels is straightforward. For example, the sizes of the terminal fragments produced by XhoI cleavage of each BglII fragment can be determined (Fig. 4). Likewise the size of each terminal fragment produced by BglII cleavage of XhoI fragments can be determined from the reciprocal gel (XhoI(32P)-BglII). In theory, it should then be possible to identify a unique set of overlaps of the XhoI and BglII fragments and determine a map order. In practice, the analysis is more complex due to several factors: (1) Even with only 13 BglII fragments and 17 XhoI fragments, certain fragments comigrate in the first dimension (Fig. 3); as a result, some of the end fragments detected in the second dimension could not be unambiguously assigned to individual fragments. For example (Fig. 4), the second band in the BglII digest contains two fragments, which generate four end-labeled fragments in the second dimension (1.6 kb, 2.2 kb, 2.7 kb and 6.9 kb). (2) Not all end fragments are resolved in the second dimension (e.g., the two ends of BglII 3 in Fig. 4); this resolution problem was further complicated by difficulties in precisely comparing mobilities on opposite sides of the gel. (3) Because the method of end labeling actually labels a stretch of DNA near the ends of the fragments (see Materials and Methods), if the second digestion cleaves off a very small end fragment (< 200 base pairs), the fragment immediately internal to the end fragment may be labeled. Furthermore, such very small end fragments may be lost during electrophoresis in the second dimension. (4) Some of the fragments produced by the first enzyme digestion will not be cut by the second enzyme and will migrate in the second dimension with the same molecular weight as in the first dimension; if two or more consecutive fragments are not cut by the second enzyme, their order cannot be deduced from this analysis. (Note that if the same percentage of agarose is used for the first and second dimensions all such altered fragments should form a straight diagonal across the gel. However, a higher percentage gel is usually used in the second dimension, and the non-ideal electrophoretic behavior of high-molecular-weight DNA causes the unaltered DNA fragments – and any products of incomplete cleavage – to form an arc (see Fig. 4).)
We circumvented the above difficulties primarily by repeating the analysis with additional combinations of enzymes, in particular by performing two-dimensional analyses with XhoI and BglII combined with EcoRI. This gave an unambiguous order to the fragments produced by XhoI and BglII digestion, as well as determining the position of the EcoRI sites immediately adjacent to these sites. The accuracy of the positioning of these sites should be quite high (probably less than 5% error) since calibrated one-dimensional gels were used to determine the sizes of all the XhoI, BglII and XhoI/BglII double-digest fragments, as well as most of the XhoI(32P)/EcoR1 and BglII(32P)/EcoRI fragments. The calibration of the two-dimensional gels was generally accurate enough to allow assignment of particular spots to bands in the one-dimensional double digest for size determination.
At this stage in the analysis, we received a preprint of Carlson’s (1979) map ordering the T4 KpnI and SalI sites. These were aligned and oriented with our map using one-dimensional analysis of digests with combinations of BglII, XhoI, KpnI and SalI. After accounting for the differences in the deletions, agreement was excellent. However, we saw one extra KpnI fragment, which Carlson has since confirmed (personal communication).
Information from several sources was combined to position the EcoRI sites. (1) Internally labeled DNA was analyzed on two-dimensional separations using XhoI-EcoRI ang BglII-EcoRI combinations; appropriate end-labeled fragments, at a higher concentration, were run to aid in analysis of the rather complicated pattern. (2) Fragments end-labeled at sites of KpnI cleavage or sites of SalI cleavage were further cut with BglII and then analyzed on two-dimensional gels with EcoRI cleavage prior the second dimension. These gels (KpnI(32P)/BglII-EcoRI and SalI(32P)/BglII-EcoRI) allowed us to position the EcoRI sites that are immediately adjacent to SalI and KpnI sites. (3) The alignment with the genetic map allowed us to take advantage of partial restriction maps obtained from cloned regions of the genome, the references for which are in Table 4. (4) Procudcts obtained by partial EcoRI digestion of SalI(32P) fragments were analyzed using methods which will now be described.
As described by Smith and Birnstiel (1976), a DNA fragment labeled at one end and partially digested with a restriction enzyme can be separated into a series of labeled bands whose sizes correspond to the distances between the labeled end and each restriction site. While this has proved to be a powerful tool for restriction mapping, its usefulness has been limited by the considerable effort required to purify starting restriction fragments specifically labeled at only one end. If a mixture of terminally labeled restriction fragments were partially digested with a second restriction enzyme and separated in one dimension, it would not be possible to determine which of the labeled ends generated a particular band. If, however, these separated partials are digested to completion with the second enzyme and electrophoresed in a second dimension, each labeled end generates a distinctive row of fragments in the resulting pattern. Thus, a labeled end fragment of particular size in the second dimension can be used to identify the positions of all partials extending from that end. (Fig. 5). This approach allows the simultaneous mapping of many sites on a number of restriction fragments. Kovacic and Wang (1979) independently described a similar use of two-dimensional analysis of partial digests.
In the map presented here, the HindIII and PstI sites, plus some of the EcoRI sites, were mapped by a series of such analyses of partials. In these experiments, C-T4 DNA was cut with an enzyme which makes relatively few cuts, such as SalI, and then end-labeled. The end-labeled SalI fragments were partially digested with a second enzyme (either PstI, EcoRI or HindIII) and the partials separated in the first dimension. Redigestion to completion with the second enzyme and separation in the second dimension then produced horizontal trails of fragments, each member of the trail having the same size in the second dimension. For example, Fig 5b shows the two end fragments generated from the SalI 7 fragment after HindIII digestion. The larger end fragment is denoted as le and the smaller as se. The HindIII partials of the SalI 7 fragment extending from end se or le to internal HindIII sites will migrate in the first dimension as fragments smaller than intact SalI 7. Each of these partials can be recognized by the fact that they generate fragment se or le after redigestion. Thus electrophoresis in the second dimension produces the observed trail of fragments (denoted P) of the same molecular weight. By calibrating the first dimension separation we can determine the molecular weights of all partials extending from the given end. In the case of SalI 7, each end generates three partials which are clearly resolved and a fourth partial which is partly resolved and a fifth which is nearly as large as intact SalI 7 and is not resolved from it. Reading the positions of the HindIII sites from both end le and end se produces redundant data confirming the existence of six HindIII sites within SalI 7; cleavage at one site produces the end fragment while cleavage at the remaining five sites produces the partials. Since HindIII should produce about 150 such partials, it is not surprising that they are not all resolved in a single analysis. We were able to map most of the HindIII sites by analyzing end-labeled BglII, XhoI and SalI fragments by this partial method. In each of these analyses, we were able to unambiguously identify only a fraction of the end-labeled fragments and their trails, but when combined these analyses gave an almost complete map of the HindIII sites. The realtive orientation of the HindIII maps of individual fragments was determined by overlaps of maps derived from the different enzyme combinations and by matching the sizes of fragments in the maps to cloned fragments of known map location and to HindIII fragments detected in our blot-hybridization experiments.
Certain features of the HindIII map should be noted. The HindIII sites closest to SalI, BglII and XhoI sites will be most accurately positioned because these positions are derived directly from the measured molecular weights of end labeled doubly cut fragments (SalI(32P)/HindIII, BglII(32P)/HindIII and XhoI(32P)/HindIII fragments). The positions of the internal HindIII sites are expected to be less precise because the calibration of the two-dimensional gels is less accurate. Additionally, the sizes derived from high-molecular-weight partials cannot be determined as accurately as those from lower molecular weight fragments, since we are often looking at relatively small differences between two fairly large numbers. A gap in the HindIII map between 49.3 and 57.9 kb remains because the appropriate end fragments and partials could not be resolved. In addition, two regions of the HindIII map (38.8–45.65 and 91.9–99.5 kb) are uncertain, because confirming data was not available, and errors could have been made in interpreting the complex patterns.
The PstI map was derived from a similar analysis using partial digests of end-labeled BglII and SalI fragments and from two-dimensional analyses where XhoI(32P) or KpnI(32P)-BglII fragments were separated in the first dimensions, cut with PstI and analyzed in the second dimension. In addition, a two-dimensional analysis of end-labeled SalI fragments partially digested with EcoRI allowed us to complete the EcoRI map.
Initial alignment with the genetic map relied on six different clones of T4 DNA, obtained from Ted Young and Wai Mun Huang. The portion of the genetic map carried on each clone had been determined by marker-rescue experiments (cf. Mattson et al. 1977) and is indicated on the resultant map (Table 4). Plasmid DNA from each clone was labeled as described in Materials and Methods. Each of the labeled DNAs was then used as a hybridization probe to analyze one-dimensional separations of assorted restriction digests of T4 DNA which had been transferred to diazophenylthioether paper (“blots”) as described in Materials and Methods. The patterns of hybridization to fragments obtained by digestion with XhoI, BglII, EcoRI, HindII or PstI (or various combinations of these enzymes) allowed unambiguous assignment of each cloned fragment to a position on the restriction map we had derived. The major lack of precision here results from ambiguity as to the exact positions of genes within the cloned fragments.
Several methods were used to confirm and refine the alignment at various points arount the map:
As discussed by Wood and Revel (1976), recombinational hot spots prevent a perfect correlation between the physical and genetic maps of T4. The map of Edgar and Wood (1966) is based on corrected recombination frequencies. Wood and Revel (1976) refined the relation between physical and genetic distances by adjusting the map distances based on physical measurements of the sizes and locations of various mapped deletions and the molecular weights of protein products. A map derived by Mosig (1976) gives physical distances based on marker-rescue using incomplete genomes, a technique indicating the approximate midpoints of genes. These previous attempts to correlate the physical and genetic maps allow a fairly accurate alignment of the genetic map with the restriction map once a few genetic markers have been localized on the physical map. Since the restriction map provides the true physical map, exact localization of additional genetic markers on this map should eventually allow a precise comparison of physical and genetic distances.
The alignment of our restriction map with the genetic map (see above) allowed us to check our data by comparing it to data obtained from cloned fragments of T4. Cloned fragments whose restriction map was known and whose genetic composition was known from marker-rescue experiments could be matched to both the restriction map and the genetic map. Table 5 summarizes our comparison between the positions of cloned fragments on the restriction map and their positions on the genetic map of Wood and Revel (1976). The correlation is very good, almost never differing by more than 2 kb. Only in three short stretches is the physical distance substantially less than the genetic distance: between genes 43 and 45 (1.5vs 4.5kb), between genes 45 and 47 (2.0 vs 3.9 kb) and between genes td and 32 (1.7 vs 3.2 kb). Our correction largely eliminates the previous gaps unassigned to genes in those three regions of the genome. (For the region around gene 32, Hänngi and Zachau (1980) and Mosig (personal communication) also have evidence for such an anomaly in the genetic map). It is possible that these three sites represent recombinational hot spots.
We would like to thank Bruce Alberts in whose laboratory this work was initiated and gratefully acknowledge his advice and help with the manuscript. We also thank Ursula Hibner for her advice and generous donation of T4 DNA polymerase, Rae Lynn Burke for help with the manuscript, Ted Young and Wai Mun Huang for cloned T4 DNA fragments, G. Wilson for his kind donation of the strains, Burt Guttman for his help with the construction and illustration of the map, and the many individual who shared information with us prior to publication (L. Albright, K. Carlson, U. Hänggi, K. Jakobs, J. Karam, R. Marsh, N. Murray, D. Pribnow, H. Revel, W. Rüger, H. Takahashi, J. Velten, and G.S. Wilson).
The work was supported by NSF grants PCM 7825677 to P.H.O. and PCM 7905626 and an NIH grant GM 24020 to Bruce Alberts.