LINEAR INHERITANCE IN TRANSDUCTIONAL CLONES JOSHUA LEDERBERG Department of Genetics, University of Wisconsin, Madison, Wisconsin Reprinted from Genptics Vol. 41, No. 6, November, 1956 Printed in U.S.A. Reprinted from GENETICS Vol. 41, No. 6, November, 1956 Printed in U.S.A. LINEAR INHERITANCE IN TRANSDUCTIONAL CLONES! JOSHUA LEDERBERG Department of Genetics, University of Wisconsin, Madison, Wisconsin Received July 6, 1956 HE rule of inheritance in cell lineages is the transmission of an undiminished legacy to each of a geometrically increasing family of descendants. Episodes of mutation or segregation may intervene, but further descendants will again follow this rule of clonal heredity, which is the corollary of equal division. But other rules of inheritance are known—for example, the entailment of estates in land and the traditional law of primogeniture in titles of nobility—-whereby a legacy must pass undivided through a single line of descent through the generations. This paper will have to do with biological analogies of linear inheritance which have appeared in experiments on the transduction of motility-genes in Salmonella. Transduction is a mechanism of genetic recombination which is notable for the transfer of hereditary fragments from one cell to another (Symposrum 1955, LEDER- BERG 1956a). In these experiments, a temperate bacteriophage serves as vector for the fragments, which are furnished by the disruption of the chromosomes of a bac- terial host as it supports the growth of the phage. When this crop of phage is applied to cells of a suitably marked recipient strain, some (~ 10-6) of these cells yield a transformed clone which carries a given marker from the donor. In previous studies, the transformed clones have exhibited the same genotypic stability as did the parents. However, the selective methods which were used to isolate the rare recombinants would overlook transductional effects that did not yield substantial clones of the new types. These studies included auxotrophic, fermentative, resistance, serological and motility markers, and each one for which a suitable selective technique was available was subject to transduction in much the same fashion. The following experiments are a follow-up of observations on “motility trails” (see paragraph 1. 1) initiated by Dr. Bruce STOCKER during a research visit to this laboratory (STOCKER, ZINDER and LEDERBERG 1953). After his return to England, Dr. STOCKER began microscopic studies on these trails; the immediate concern here was the problem of segregation and crossing over in transductional clones. However, the two studies proved to be operationally inseparable. Iam indebted to Dr. STOCKER for an unreserved exchange of materials, information and manuscript drafts through- out these studies. In the main, the terminology also follows his suggestions. A con- cordance of my results and interpretations with his (StocKER 1956b) is given at the close of this paper. Glossary and symbols. The central concept of this paper is that of a line (adj. linear or unilinear) which signifies a single, unbranched, finite or infinite chain of descent 1 Paper No. 604 of the Department of Genetics. This work has been supported by research grants from the National Cancer Institute, (C-2157), Public Health Service, from the National Science Foundation, and from the Graduate School of the University of Wisconsin with funds allocated by the Wisconsin Alumni Research Foundation. 846 JOSHUA LEDERBERG \ KRY Fictre 1.—Diagrammatic representations of hereditary transmission within a clone?. A, holo- clone. B, segregating clone. C, line. D, a pattern of delayed development, see 4.5. The large circles may be taken as +, the small as —. * (fig. 1C), with regard to a given quality. A line must correspond to a sequence of unequal fissions, at each stage of which only one product propagates the line. A clone may be multilinear if its descent can be resolved into a number of lines. By extension, a clone or a cell may be loosely referred to as line or linear if it contains or initiates a line. In previous discussions, semiclone and chain have been used as synonyms of dine. As will be seen, trails are the overt manifestations of linear inheri- tance of motility in a clone growing in semisolid agar medium. Clone is taken to mean the progeny from a single cell, and often implies the regular appearance of a trait throughout that progeny. When the meaning is not given by the context, holoclone (see fig. 1A) will be used for the latter sense. Exogenole signifies a chromosome fragment, explicitly the one given over in trans- duction. A — x B, or its equivalent, B x—A, are abbreviations for transduction from genotype A (donor) to B (recipient). These and other terms are discussed in more detail] elsewhere (Morsr, LEDERBERG and LEDERBERG 1956b). >A uniform convention for the numbering of cells in a lineage would be helpful. The scheme shown in A is adapted from Jennines (1908; cf. ZELLE 1951) and has the advantage that the num- ber of digits is the generation number, and that the family relationships are readily visualized. For very large pedigrees a sequence of bixary numbers could be replaced by the corresponding decimal, but the generation number must then be specified. In C, the line shown is 0-1-12-121. In principle, however, any single line could be written 0-1-11-111-. .. as has been adopted in table 2. LINEAR INHERITANCE IN TRANSDUCTIONAL CLONES 847 Following the introduction, this paper comprises the following: Section 1 (["1.1 to 1.3) reviews the experimental procedures. Section 2 gives experimental detail on the microscopic analysis of cell pedigrees, and section 3 on plating experiments. Section 4 brings together ancillary observations. The interpretive analysis of the data and speculations are deferred to section 5. Section 6 compares DR. STOCKER’S data with these. Many readers may wish to proceed to the recapitulations of the experimental sections, placed for convenience at 2.0 and 3.0, and to the discussions of sections 5 and 6, before reviewing the details. 1. MATERIALS, METHODS AND PRELIMINARY OBSERVATIONS 1.1 General procedures and background observations have been given at length (ZINDER and LEDERBERG 1952; SrockER ef al. 1953; LEDERBERG and EDWARDS 1953; LEDERBERG and Irno 1956) and will be recapitulated only briefly. In 1897, Hiss had discovered that a very soft agar permitted motile bacteria to spread throughout the medium, while nonmotile varieties stayed where put. The technique was redis- covered several times, and has since been widely adopted in enteric bacteriology. The spreading cloudy growth of a motile clone is called a swarm (figs. 7, 8) and is the most characteristic result of transduction of motility to a nonmotile strain (STockeEr ef al. 1953). In the same experiments, trails are seen: these are groups of small colonies strung out through the soft agar for some millimeters from the point where the treated bacteria of the nonmotile strain had been planted (fig. +). The trails were thought to mark the path of a motile cell wherever it divided and left behind nonmotile progeny. The first observations suggested that the trails, and therefore the corresponding lines of inheritance of motility, were always unbranched, but later evidence has weakened this conclusion. At the time, however, the trails were explained by the linear transmission through the recipient clone of a damaged exogenote which could no longer replicate, but could still confer motility on the cell which carried it. Since transduction here fell short of a transformed clone, it was described as abortive. To test the hypothesis of abortive transduction and to supple- ment plating experiments, cell lineages from transformed motile cells were studied more directly in pedigrees controlled by micromanipulation. 1.2 The manipulative procedures, especially the use of the oil chamber, follow bE Fonsrune (1949), The principal media were Difco penassay broth and NGA, “nutrient gelatin agar”, G.8 per cent gelatin, 0.4 per cent agar with a broth base. Microclones were routinely held at room temperature (22 + 2°C) and examined at 150 magnifications darkfield. This was conveniently obtained with a 15 X ocular and a 10 X objective (not necessarily phase contrast) in combination with a Bausch and Lomb LWD phase condenser carrying a 43 X annulus. For closer study at 645 X, a matched 43 & dark phase contrast objective was swung into place. 1.3 Most of the experiments were of the form SW-623 —x SW-666. The cultures are described more fully elsewhere (LEDERBERG and Epwarps 1953) but both are derived from a monophasic S. paratyphi B. SW-666 is Fla; Hy (fiagellaless, hence nonmotile; phase-1 flagellar determinant b); SW-623 is Fla,t Hy‘. The transducing phage was PLT22 adapted by serial passage on SW-623. SW-666 was originally chosen for these experiments because of the previously demonstrated linkage of the 848 JOSHUA LEDERBERG A, to the Fla, factor, a consideration which is immaterial except where it is empha- sized below. EXPERIMENTAL RESULTS AND CONCLUSIONS 2. Experiments with microclones 2.0 Recapitulation. Partial pedigree analyses were made of the progeny from iso- lated motile initials. A single initial might generate many motile offspring during the first ten or fifteen generations. When these motile offspring were isolated and propagated, they usually constituted strict hereditary lines (for motility) for as many as thirty fissions longer. In further pedigree analyses, the number of lines which issued from two sister cells was unequally partitioned. That is, if “2” repre- sents the potentiality to produce many motile offspring, E was unequally, perhaps linearly, transmitted at cell division. Formazan granules were also transmitted linearly, but the possible inherent polarity of the bacteria, as marked by formazan, could not be correlated with the linear transmission of E. About four per cent of motile initials generated stable transductional clones. These often showed an early segregation of motile and nonmotile subclones. The concurrence of motile and multi- linear subclones was also noted. 2.1 Isolation of motile initials. The first step in each experiment was the isolation of infrequent motile znitial cells from SW-623 —x SW-666. Equal volumes of an overnight broth culture of SW-666 and stock lysate of SW-623 were mixed and incu- bated for about 90 minutes to give an input of about 10° bacteria, 10! phage per ml. Between 90 and 150 minutes, 10‘ to 105 cells became motile. These initials were readily isolated with the help of a trapping droplet. Small drops of the treated cul- tures were deposited on a cover glass under oil, adjacent to drops of clear broth. The droplets were then fused in pairs, permitting the motile initials to swim into the traps where they could be clearly seen and individually caught. The interval between mixture and isolation was two to three hours, which might allow as many as three fissions depending on a variable initial lag. Subsequent pedigrees are therefore likely to be truncated at the origin. After most of the experiments were completed, it was found that isolation was simplified by spinning down the transductional mixture after about 60 minutes and trapping from droplets of concentrated sediments. Hundreds of initials could be trapped in a short time by this procedure, which was used in experiments below 2.16, 2.18, 3.1, 3.7, 4.1. 2.2 Undivided clones. The simplest experiment was to plant the motile initials in individual droplets and examine the microclones the next day, when they usually contained about 10* cells each. Table 1 shows the results from 384 viable isolations, including some pedigrees which have been summed to give the total yields. In addi- tion, about ten per cent of the cells isolated were inviable. In these cases, a long fila~ ment (“snake”) often persisted which might remain motile for hours or days but never divide. Other clones contained one or several ghosts which may have originated by phage lysis. 2.3 Only fifteen (four per cent) of the clones contained a preponderance of motile cells which would relate them to swarms, i.e., holoclonal inheritance of motility These will be taken up later 2.22. LINEAR INHERITANCE IN TRANSDUCTIONAL CLONES 849 TABLE 1 Distribution of yields of motile cells in 384 single microclones 7 | Yield:....... 0 1 2 4/5 6 | 6 rs 9} 10 11-15:16-20121-30 >30 Swarm Number of | | clones:....| 100 | 59 | 39} 28] 17 | 7 | 9/6 | 6 4 6 25 | 23 19 | 21 15 2.4 The remaining clones all had a limited number of motile individuals, ranging with decreasing frequency from 0 to about 100. The distribution of these numbers was highly skewed, as can be seen from table 1. For numbers in the range 0 to 9, the distribution is approximately exponential, each class being proportional to 0.8*; while the larger numbers follow a nearly flat distribution. Except as an indication of nonhomogeneity, the quantitative significance of this distribution is not apparent, and we shall be concerned only with its qualitative features. 2.5 The microclones therefore give a partial corroboration of the hypothesis of abortive transduction of motility, in so far as transformed motile cells are isolated which do not transmit the trait regularly to their descendants. The frequent class of clones, containing one motile individual is most readily understood: each com- prises a line still in being after 13 fissions (2% ~ 104), while the most frequent, zero class consists of lines that had terminated some time between the first and the 13th fission. Most of the experiments are directed at an understanding of the remaining two thirds of clones which have more than one motile individual: whether these are multilinear, and if so the patterns of distribution or generation of the lines. 2.6 Progeny of intermediate isolates. These experiments were done to confirm the linear inheritance of motility for a number of generations, and to ascertain whether the clones containing many motile cells could be resolved into lines, or whether they would show a continuing pattern of irregular transmission which might be neither linear nor clonal. The expression “‘isolation at 7,’ means that a motile individual was reisolated from a clone whose total population indicated a history of & fissions. As indicated in the previous paragraph, many isolates were made at about 9 to #3; the history of these pedigrees is summarized in figure 2. 2.7 As shown in figure 2, linear inheritance was followed strictly in every reisolate but one after #, that is, no branching (production of two motile progeny from one cell) need have occurred after my or could have occurred after m5. This range, mn to m5, follows because the clones could not be examined at each fission. Many of the clones were already resolved into one or more lines between the first and tenth fis- sions. The exceptional clone (4 in figure 2) must have had a branch not earlier than Nyy and possibly as late as #7; the figure represents a progeny from a cell isolated at mis which gave a subclone containing 18 motile among 5,000 nonmotile cells. As 18 ~ 24, and 5000 ~ 2”, this subclone must have branched not earlier than z = 15 + = 19, and not later than # = 15 + 12 = 27. Ten of the motile cells at #2; were isolated; one formed a line for at least 13 additional generations (749); the others were nonmotile or inviable at the next examination at 39 to #41. 2.8 The outstanding examples of continued linear inheritance were a pair of motile cells isolated from the same clone at #7; which gave regular asymmetric division for 850 JOSHUA LEDERBERG aneee SSS 0 fo 20 50 40 500 10 20 30 40 50 Ficure 2.—Duration of motile lines generated by isolated initials. The scale is the number of generations elapsed from the time of isolation. The heavy horizontal bars show the minimum dura- tion of a single line, from the last point at which any branching could have occurred to the earliest point at which the line could have terminated. The light bars and extensions show (to the left) the earliest point at which branching could have ceased, and (to the right) the longest that each line might have lasted. The overlap represents intervals during which individual clones were not examined. A vertical tick on a light bar is equivalent te a short heavy bar, that is, a single motile cell was observed and isolated at this point, but had no motile progeny when its subclone was re- examined. Many lines were terminated before motility had ceased; when examinations were con- tinued after the cessation of motility in a line, the end of the study is indicated by ~. An arrow indicates a swarm. Vertical bars connect lines that had been isolated from the same clone; they do not represent the derivation of one line from another. an additional 45 and 52 fissions, respectively (fig. 2B). These lines were observed and reisolated at intervals of a few fissions until they finally terminated, in one case by cessation of growth, in the other by the gradual loss of motility. (This experiment lasted seven days (March 17-23, 1954), the clones being held at 10°C to slow their growth at night.) 2.9 These pedigrees, from intermediate isolates, suggest that clones with many motile cells are indeed multilinear, i.e., can be resolved into simple lines; although branching may persist as late as the third decade of fissions, it is usually not observed after the first. Once established, the lines were propagated for a variable interval (from one to 42 fissions) being terminated either by the death or immotility of the line cell, 2.10 On a number of occasions, a motile line was watched throughout one or more fission intervals. The dividing cell remained active until the moment of separation, at which point one daughter continued to move, while the other remained stationary. The motile daughter was the cell which continued the line on further division. LINEAR INHERITANCE IN TRANSDUCTIONAL CLONES 851 2.11 Early lineages from motile initials and partition of line numbers at fission. It is manifestly impossible to make a complete pedigree analysis of a multilinear clone for more than a few generations, if for no other reason than that one cover glass will not hold more than a few dozen isolates, and that at least a minute is needed, under the best circumstances, to separate two daughters into separate drops. As the clones grew, it became increasingly difficult to complete the handling of a generation before it divided again, and to maintain a coherent record of its disposition. On the other hand, concentration of effort on a single clone was unrewarding because the majority of initials generate none, one or very few lines. No procedure was found (cf. STOCKER, and 4.9) by which the minority of cells that would generate multilinear or swarm- clones could be detected in advance of growing out their progeny. A number of clones were, however, followed for a few fissions to answer the specific question whether the partition of lines (i.e., the total number of motile cells which ultimately appear in sib subclones) was random at the division of a multilinear cell. Figure 3 represents these partitions most of which come from single observations at an early fission. The coordinates of each point give the larger yield as abscissa, the smaller as ordinate. Thus the point (40, 2) refers to a fission at which one daughter gave a clone (usually read at a size of 10% to 10*) containing 40 motile individuals, the other daughter clone giving 2. In a few cases, the partition could be followed for two or three fissions. Only a few of the partitions represent divisions later than mo for the reasons given in 2.7. 7 ? 7 7 e . ° ° 3 e Se e e See e 7 - “ff e 3 3 oo 3 e e e @ ese e — 7 $ e Zo 08 of or” Teer pecergprryey v ] J | ¥ 0 lo x 20 30 4 Fictre 3.—Partition of potential lines at cell division. Each point represents a fission after which the two daughters were allowed to form clones, and the yield of motile lines in each was enu- merated. The smaller yield is always given as ordinate. For details of the method of plotting, see 2.12. 852 JOSHUA LEDERBERG 2.12 The figure presents only those partitions where a total of eight or more lines was at stake. The data are plotted on a modification of MosreLLer and TuKEY’s (1949) co-ordinate paper. The ordinate scale is shifted to ~/ y + 1 rather than Vy so that the ‘‘shortest distances” can be read directly from the plotted points. The “2a” band which lies parallel to the expected line, « = y, thus should exclude only five percent of the observations. The observed partitions are clearly unequal, one daughter tending to inherit most of the potential lines. The most nearly equal splits of a large stake were 19,11 and 30,9; grossly discrepant splits such as 36,0 or 40,1 were more common. 2.13 These inequalities are consistent with the hypothesis (SrocKEeR 1956b) that the multilinear clones are resolvable into two orders of linear inheritance, the simple motility lines already described, and in some clones, a line defined by the ability to generate many of these simple lines. Following STockER, descents with many motile cells will be designated as the ‘‘E line” (E for exceptional), leaving open the question whether the descendancy is strictly linear (see 5.3). 2.14 Linear inheritance of formazan residue. Linear inheritance is the expected con- sequence of the passive handing down of a particle at cell division. An almost trivial instance of a line stems from observations on the bacterial reduction of tripheny]- tetrazolium chloride. The reduction product is triphenylformazan, a fat-soluble, water-insoluble red pigment which is usually deposited as a single conspicuous granule near one pole of the bacterium (LEDERBERG 1948; WEIBULL 1953). The transmission of this granule at cell division has been seen to be linear in direct pedi- gree isolations, and by other procedures 4.7, 2.15 Correlation of motility with fermazan-lines. The mere fact of unipolar deposi- tion of formazan speaks for a polarity of cellular organization in Salmonella (and other enteric bacteria) which belies the superficial antero-posterior symmetry of the rod. An effort was made to correlate the unequal division of a cell into formazan- carrying Fz* and not-carrying Fz~ daughters, with the unequal division of motility and of E lines. However, it proved to be difficult to induce the deposition of formazan in cells of microclones: in general, it is taken up only by cells in the stationary phase, and it was therefore impractical to study the correlation of formazan with motility in intermediate isolates. Nevertheless, cells already Fz+ were amenable to trans- duction, so that pre-stained initials could be isolated. It was however noted that recipient populations, in which half the cells were Fz+, gave motile initials of which only one to two percent were Fz+. A similar negative correlation was found between motility and Fz* in experiments with motile clones of Salmonella and of Escherichia co. Such clones invariably contain a proportion of temporarily nonmotile indi- viduals which is increased after formazan-staining. However, those motile cells which do carry a unipolar granule appear to be as vigorous and viable as the controls. Pos- sibly the formazan is toxic when it is deposited above a threshold level. No preferred orientation (formazan anterior or posterior) was noted; an individual motile cell might reverse its orientation at intervals of a few seconds. 2.16 Motile initials were collected from transductions to prestained recipients and followed for three or four fissions along the formazan line. That is, at successive fissions, the Fz+ cell was separated for further observation, and the Fz~ sib set aside LINEAR INHERITANCE IN TRANSDUCTIONAL CLONES 853 TABLE 2 Yield of motile progeny in pedigree lines of formazan-stained initials Yield in cell no. Initial no. 2 12 112 Fz = itt 1. 4 25 0 inviable 2. 15 1 2 0 3. 23 0 0 4. 18 0 2 2 S. 13 1 1 2 6. 2 3 44 1 7. 0 22 0 1 8. swarm — 0 22 9. 2 18 6 10. 1 2 42 (2) 11. 3 1 12 12. 1 1 0 0 13. 3 0 0 0 14. 1 0 1 0 15. i 0 1 0 16. 3 0 0 0 17. 3 0 2 1 18. 0 1 1 inviable 19. 2 0 2 0 20. 2 0 0 0 From pre-stained initials, the line of cells (1-11-111) carrying a formazan granule was followed for three fissions, the successive sibs (2-12-112) being transferred to individual droplets. Lines No. 3, 9 and 11 grew more slowly or rapidly than the others, and were therefore not separated at a third fission. Lines 1 to 9 are summarized in 2.16. Line 11 is excluded because of the uncertainty whether 111 or 112 would have been multilinear. In line 10, the granule was no longer discerned after the first fission, and the “formazan-line” was continued only arbitrarily. 12. to 20. are some of the pedigrees with few motile progeny. In line 8, the initial cell had already divided twice when the isolate was reexamined, and cells 2 (= 21 + 22) and 12 were pooled. Unfortunately, the swarm that resulted was lost before it could be analysed for homogeneity. The numbering of cells follows figure 1A. and allowed to form a microclone. Finally, both sibs were set aside, and all the clones were scored for number of motiles the following morning. The sequences of table 2 may be thought of as progressive halvings of the initial cell, from the center to the polar granule. With three fissions, a random disposition would lead to the occurrence of E in successive segments (i.e., 2, 12, 112, 111, see fig. 1A) in the proportions 4:2:1:1. The eight £-clones of this experiment were found in the proportions 4:3:1:0, which is concordant with the expectation, i.e., the £ quality is not specifically asso- ciated with a part of the initial cell that is marked by Fz polarity. The possibility of mutual exclusion of £ and the Fz-marked segment was considered, but another clone (which also contained a swarm 2.22) included a multilinear progeny from the Fz- terminal segment. 2.17 Clones with few lines from the same experiment are also tabulated but show no striking features. 854 JOSHUA LEDERBERG 2.18 Manifestation of linked transduction; tests for reciprocal crossing over. As al- ready mentioned the Fla~ locus is linked to another marker, 4, the parental cou- plings in these experiments usually being FlatH,'—x Fla-H’. To this point we have considered only the motility phenotype (Flat or Fla~). The serotype (/7;' or H,°) of motile clones is readily detected with absorbed antiserums by agglutination tests, or by the inhibition of motility in NGA or under the microscope. The Fla—H, linkage is exhibited by a proportion of clonal swarms which were FlatII 1’ as well as others which were Fla+H,>. The serotype of non-clonal initials and lines is now in question. The first trials showed a complete inhibition of motile initials equally by 6 and i serum, and in NGA as well as in microclones. However, control experiments Pla*Hy’—x Fla~H,” also showed inhibition by i serum, which must therefore be at- tributed to a delicate cross-reaction between 6 and i (not observed in agglutination tests or in inhibition of swarms, and not necessarily a flagellar reaction, cf. LEDER- BERG and Tino 1956, 93.14) rather than the necessary presence of the 7 antigen on the motile initials. As a comparable non- Fla~ stock is not available the specificity of inhibition by serum has not been verified. 2.19 Other serums (a, c, d, k, r and 1.2) were then tested on the b—x } controls, and all were found non-inhibitory, in contrast to i and }. Further experiments were therefore conducted with the transductions FlatiZ,* (SW-940)—x Fla~H\. With this system, all initials were completely inhibited by 1:1000 8 antiserum (figure 11) and were all also partially or completely inhibited by anti-a. In microclones all cells (except a few which generated 8 swarms) were slowly but completely immobilized by a serum. In dilute NGA + a serum (see 3.3) stationary colonies and a few short trails were seen (fig. 12). This result is evidence, not otherwise available, of the homogeneity of the exogenotes, i.e., that each one that carries the Flat marker also carries the coupled Hy. The occurrence of recombinant clones FlatH 1” may therefore be attributed to crossing over ina transient, initial heterogenote Fla~Hyb/ex Fla+H,4 (cf. Morse ef al. 1956b; DeMEREC and DEMEREC 1956). 2.20 Intermediate motile isolates from multilinear clones have also been tested separately with b or a serum. Most were inhibited but in one test, two of four cells isolated were unaffected by @ serum, and may have had a pure & serotype. These terminal non-a lines may account for the residual trails seen in NGA platings with this antiserum. 2.21 A search was made for reciprocal crossovers, ie., Fla~H\* among nonmotile progeny in multilinear clones and co-segregants (2.22) in clones containing motile transformations (both Hy? and H,2). These are detected as isolates capable of yielding motile, non-6 recombinants when tested x—FlatH? in NGA plus 6 antiserum. None were found in a total of about 100 tests. 2.22 Clones wiih swarms: segregation. As stated, 2.3 and table 2, about four percent of motile initials gave clones which contain ten percent or more motile individuals. These have also been replated on NGA and verified to initiate swarms. Several dozen individuals from these progenies have also been allowed to form clones and proved to be regularly motile, without continued segregation. They therefore correspond to stable transductions of motility. However, the initial clones are often mixtures of LINEAR INHERITANCE IN TRANSDUCTIONAL CLONES 855 stably motile and stably nonmotile cells, shown in terms of proportions of motile to total cells: Pure + ~w+ ' ~let ~Met swarms support a cross-over model of incorporation, similar to that which is more directly inferred for E. coli. 5.13 Asymmetry in cell division: bacteria, Enteric bacteria appear to be simple rods, lacking head-tail asymmetry. Other bacterial species are more clearly differentiated, spectacularly in Caulobacter which has a long stalk, terminated by flagella at one end (HouwInk 1955). At cell division, the body of the cell is split, concurrently with the development of a flagellum and ultimately a stalk at the distal (free) pole. These observations also put in question the possibility of genetic continuity of the flagellar ‘ Anlage, (Bisset 1956) unlike the division processes described in many protozoa. Bac- teria with unipolar flagellation, like those studied by Lerrson (1951) pose the same problem of the morphogenetic relationship of the old and new flagella. At any rate, each fission of such a cell may be considered to engender one new cell (in respect to the flagellum) and leave one old one. Whether the sequence of oldest cells in a series of fissions will constitute a detectable line can be thought of as depending on how completely a mature flagellar apparatus is resynthesized by the young cell in an inter- 866 JOSHUA LEDERBERG fission interval. The motility lines would represent the extreme case where no such synthesis is possible owing to genetic defect; intermediate situations are represented by spontaneous trails 4.1, and by the observations on Pseudomonas aeruginosa 4.5 and E. coli 4.4, 5.14 The overt symmetry of Salmonella and other bacillary species may then be only superficial; suggestions of head-tail or corresponding mother-daughter differ- ences come from the unipolar deposition of formazan 2.14 and the basophilia of one or both poles, interpreted by Bisset (1956) as indicative of a growing point. Differ- ences in the reaction of sib daughter cells to staining procedures (PENNINGTON 1950) and to antibiotics (Linz 1954; Smires, WELcH and ELrorp 1948) are also indicative of unequal division. However, these scattered observations have not yet been corre- lated to the extent that this conception of the organization of the bacterial cell may be regarded as more than a speculative working hypothesis. 5.15 Protozoa, Outside the bacteria and simpler algae, very few unicellular or- ganisms divide by simple fission, so that there are many opportunities for asymmetric division and linear transmission. For example, JeENNincs (1908) and McCLexpon (1909) have described the linear inheritance of accidental pellicular defects for as many as twenty-two fissions. This pattern is a simple consequence of the conserva- tion of the surface structure of the Paramecium at cell division. In a related experi- ment, JENNINGS (1937) injured the teeth of Difflugia by surgery and found that both daughters of an injured individual were often defective. The deviation from linear transmission in this case is apparently due to the role of the teeth of the mother cell as models for the deposition of the daughter’s. In an appropriate context, then, silica can simulate a gene. However, successive offspring from the line of cells carrying the original defect were progressively more normal. The experimenta! defect therefore did not generate a defective clone, but one with a variety of defective lines. In all of JENNINGS’ experiments, the morphological description of the propagation of varia- tions averts the postulation of discrete particles. 5.16 Algae. The algae furnish several examples of lines which depend on peculiari- ties of cell division. A desmid, for example consists of two modified hemispheres, “semi-cells” joined by a narrow isthmus in which the single diploid nucleus is lodged. At cell division, the nucleus undergoes mitosis and the semi-cells separate, each one later budding a new semicell to form a new cell. Warts (1950) and Kattio (1951) have described mutations in Micrasterias which seemingly have a cytoplasmic basis, since the progeny of a defective semi-cell remain defective, and of the normal, nor- mal, despite the common nuclear origin. In one case, a reversion was described which resulted in a cell with one normal, one mutant semi-cell. This ‘“dichotypic’’ initial generated a clone in which all of the new semi-cells thenceforth were normal. The clone was thus unilinear for dichotypy: the equivalent ‘particle’ in this instance | may be the whole cytoplasmic architecture of the differentiated, mutant semi-cell which persists after the genotype is altered by mutation. 5.17 The growth of diatoms also generates lines of a sort. The cell is bounded by two half-shells which fit into one another like the lid and base of a Petri dish. In many species, the walls are so rigid that they limit the growth of the daughter cells after division. Therefore, the daughter that receives the inner half-shell of the parent. LINEAR INHERITANCE IN TRANSDUCTIONAL CLONES 867 and uses this for its outer half-shell, is regularly smaller than the other daughter, which maintains the same size as the parent. Each clone therefore contains only one line of cells as large as the parent, and a series of lines of progressively smaller size. The mean cell size of a clone diminishes progressively with time, but can be restored by the formation of (often sexual) auxospores in which the rigid walls are discarded. As the cells become smaller they lose the capacity for auxospore formation and ulti- mately for vegetative division too, and the clones are therefore doomed (WIEDLING 1948; cf. Rizer 1953). A formal scheme of progressively attenuated particles might have been constructed for this finite longevity, but again is less meaningful than the inference from direct observation of the division mechanism. 5.18 Yeast. Two types of lineation are possibly related to the proliferation of yeast by budding. The disproportionate partition of total cytoplasm may account for the lines of normal mother cells, with small-colony-variant buds, described by Erurussi and HoTrincuer (1951) in experiments with acriflavine. Bautz and MARQUARDT (1954) have indicated that anaerobic, Nadi-negative mother cells placed in air remain negative but form positive-buds. On the other hand, SprecELMAN (1951) concluded that the cytoplasm was equally divided between mother and bud from an analysis of another cytoplasmic character. If both observations are correct, it may be neces- sary to assume that acriflavine actively influences the segregation of the postulated plasmids, perhaps by aggregating them. 5.19 A second effect of budding is to leave a “birth scar” on the bud, and add a bud scar to the wall of the mother. Each clone should therefore continue a line of the oldest cell, carrying the most scars. Since a new bud was never observed to form at the site of an old scar, the oldest line, indeed any line, presumably has a finite life- time, (BARTON 1950, BARTHOLOMEW and MrrtTwer 1953), though this might be up to one hundred buddings; twenty-three were observed by Barton. Remarkably, BARTHOLOMEW and MITTWER reported having found one cell with twenty scars: its incidence in the entire clone would be calculated at 2-2° or one in four million. 5.20 Morphogenesis. Morphogenetic differentiation must always be related to un- equal division in the long run. When divergent clones are produced, the process re- sembles segregation; however, stem cells of various kinds (e.g. apical cells in shoot meristems; teloblasts in annelid embryos) form approximations to lines, although the detailed cell linages have not often been worked out. In the renewal of sperma- togonia in the rat, however, CLERMONT and LEBLOND (1953) have secured evidence that the ancestral spermatogonia undergo cycles of fission whereby one daughter (or son?) proliferates and differentiates into eight or sixteen spermatocytes, and the other remains an undifferentiated stem cell. The material foundation of the line of stem cells is unknown. 5.21 Orthoclones and senescence. The line of individuals of a given age in a series of progenies has been designated an “orthoclone” by Lansinc (1948) who has also described progressive changes in old orthoclones of rotifers. 5.22 The relationship of linear inheritance to aging mechanisms may be conceived in terms of individual cells of the line (or orthoclone) or to the average properties of the clone. Thus, SonNEBoRN (1930) demonstrated that the line of head-bearing individuals on repeated fission of the worm Stenostomum has a finite life time, while 868 JOSHUA LEDERBERG the line of tail segments can be propagated indefinitely. The aging of a head may be related to the concentration of differentiated structures at that end of the worm; tail segments regenerate new (i.e. “young’’) heads at fission. Attempts to demonstrate a similar aging process in bacterial cells 4.8 and in yeast 5.19 have been unsuccessful, but have been carried for only a Jimited number of generations. The maternal ortho- clone in yeast should ultimately show the effect of progressive scarring. 5.23 On the other hand, if we define aging as the progressive loss of a function in a clone, we should consider the progressive dilution of the parental character as any uni- or multi-linear clone increases. Thus an initially motile individual of Salmonella generates a clone in which the average motility tends toward zero, and reaches it with the occasional termination of lines. The aging of a diatom clone is a parallel but more obvious example. In these examples, however, the parental line is itself the most juvenile element of the population, in so far as it most closely resembles the initial state. 5,24 Chromosomes. Recent studies on the material (in contrast to the informational) content of chromosomes and of the deoxyribonucleic acid of phage have raised the question whether mitosis is ever equal. The Watson-Crick model of DNA structure (Crick 1954) postulates two complementary polynucleotide helices, which might separate at mitosis and generate the alternative complements. Recent experiments by LevInTHAL (1956) are consistent with this semi-conservative model of replication of phage DNA: after the initial separation, each helix is conserved as such and the material content of the original DNA will be found as two lines in a holoclone carry- ing its genetic information. Analogous studies by Mazta and PLautT (1955) on Crepis chromosomes indicate a single line, as does the suggestion of tinctorial differences between ‘mother and daughter chromosomes” (PROKOFIEVA-BELGOVSKAIA 1946; cf. MULLER’s criticism appended thereto). It is therefore plausible to look for linearly transmitted modifications that have a chromosomal basis. 5.25 Particles. The axiomatic foundation of genetic analysis is the particle. How- ever, the term should not be taken too literally, especially when applied to a mathe- matical rather than a material entity. Particulate forms of inheritance apply to such amorphous items as acreage, titles and protoplasmic lumps, from which we can infer that a particle is essentially a rule of division, which can be stated in physical, legal, or biological terms, and whose domain of validity must not be ignored (cf, LEDER- BERG and LEDERBERG 1956). 6. CONCORDANCE WITH STOCKER (1956b) 6.1 No irreconcilable differences obtain between the two studies, which differ pri- marily in the strains used, and in the emphasis given to the interpretation of the linearity of E. 6.2 In his material, StocKER found a greater discontinuity in the distribution of lines per initial than I show in table 1, and his is flatter for the lower yields. We are in full agreement on the disproportionate partitions of figure 3. He has found six pedigrees (for my one, 2.7) in which an £ cell was recovered late in the clone. His pedigree analysis of the linearity of # is more detailed, but it is impossible for it to LINEAR INHERITANCE IN TRANSDUCTIONAL CLONES 869 be complete 2.11 and the main evidence for this comes from the partition data. I have already commented on possible rare exceptions to E-linearity 5.3. These might be misclassifications due to the overlap of # and non-£ distributions, or experimental errors, rather than valid exceptions. On the other hand they are sufficiently rare that STockErR’s data and mine cannot be shown to be statistically heterogeneous. Finally, we have used slightly different materials and, perhaps, techniques. 6.3 It has not been possible to make a direct test of the assertion that only an £-cell will form a trail in NGA. We agree that the incidence of trail-forming and of multilinear initials is about the same. The progressive increase in yield of trails as the NGA is diluted 3.3 suggests that this agreement in ‘‘standard NGA” is fortuitous, and the stochastic factors (such as entrapment by agar fibrils) are decisive in trail formation. The occurrence of three or four (out of 186) clones with more than one trail, 3.8, argues against either the linearity of EZ, or the one to one correspondence of trails with £ cells. 6.4 I have tried to show 5.5 that the exact linearity of £ is not crucial for its mate- rial interpretations. StOCKER’s preference for 5.6 Bi, the residual exogenote, is en- tirely plausible as a working hypothesis, but the alternative versions cannot be dis- counted at present. SUMMARY 1. A hereditary /ine is an unbranched chain of descent, based on recurrent unequal division. Two aspects of linear inheritance were observed in studies of transduction in Salmonella: the persistent linear transmission of motility itself, and the (approxi- mately) linear transmission of the trait ‘“E”, viz. the generation of many motile lines. 2. Initial motile cells were isolated from transductions of motility to a nonmotile mutant, and the progeny from these initials studied microscopically and in soft agar plates. Four percent of the initial cells gave stable motile recombinants, analogous to the transformations of other traits previously studied. The remaining cells gave descents which contained one or more motile lines. Many of these lines were studied to verify that a motile cell would regularly divide to give one motile, one non-motile offspring. 3. The motile lines are considered to arise from a non-dividing phenotypic residue, perhaps a flagellum, from a previous “abortive transduction” of a motility gene. 4. Some of the clones from motile initials had many (ten to forty) motile lines. The partition of potential lines at cell division was regularly unequal. Thus it was concluded that £ is also inherited in linear or nearly linear fashion. 5. A number of hypotheses for this behavior were considered, notably that either a sterile (non-replicating) chromosome fragment or an intermediate product (pheno- typic residue) of its action is responsible. Methods of distinguishing these hypotheses are discussed, but none are presently available for this material. 6. Various aspects of transduction mechanics (including early segregation in trans- formed clones, range of fragment size, position effect) are reviewed. Conceptual parallels for linear inheritance in genetics, development and ageing, and the semantics of “particle” are speculatively discussed. 870 JOSHUA LEDERBERG LITERATURE CITED BarTHOLoMEW, J. W., and T. Mrrrwer, 1953 Demonstration of yeast bud scars with the elec- tron microscope. J. Bacteriol. 65: 272-280. Barton, A. A., 1950 Some aspects of cell division in Saccharomyces cerevisiae. J. Gen. Microbiol. 4: 84-86. Bautz, E., and H. Marquarnt, 1954 Sprunghafte Anderungen des Verhaltens der Mitochondrien von Hefezellen gegeniiber dem Nadi Reagens. Naturwiss. 41: 121-122. Bisset, K. A., 1956 Cellular organization in bacteria. Symposia Soc. Gen. Microbiol. 6: 1-18. CAVALLI-Srorz, L. L., and J. LEpERBERG, 1956 Isolation of pre-adaptive mutants in bacteria by sib selection. Genetics 41: 367-381. Ciermont, Y., and C. P. Lestonp, 1953 Renewal of spermatogonia in the rat. Am. J. Anat. 93: 475-501. Crick, F. H. C., 1954 The structure of the hereditary material, Sci. American 151: 54-61. Dr Fonsrune, P., 1949 La Technique de Micromanipulation. Paris: Masson Cie. DEMEREC, M., and Z. E. DeMEREC, 1956 Analysis of linkage relationships in Salmonella by trans- duction techniques, Brookhaven Symposia in Biology 8: 75-84. Epurvssi, B., and H. Horrincuer, 1951 On an unstable cell state in yeast. Cold Spring Harbor Symposia Quant. Biol. 16: 75-85. Hiss, P. H., 1897 On a method of isolating and identifying Bacillus typhosus. J. Exptl. Med, 2: 677-700. Horcuxiss, R. D., 1956 The genetic organization of the deoxyribonucleate units functioning in bacterial transformations. In Enzymes, Units of Biological Structure and Function. Edited by O. H. GaeBter. New York: Academic Press. Houwink, A. L., 1955 Caulobacter, its morphogenesis, taxonomy and parasitism, Antonie van Leeuwenhoek 21: 49-64. JENNINGS, H. S., 1906 Behavior of the Lower Organisms. New York: Columbia Univ. Press. 1908 Heredity, variation and evolution in protozoa. J. Exptl. Zool. 5: 577-632. 1937 Formation, inheritance and variation of the teeth in Difflugia corona. J. Exptl. Zool. 77: 287-336. JENNINGS, H. S., and J. H. Crossy, 1901 The mechanism by which bacteria react to stimuli, especially to chemical stimuli. Am. J. Physiol. 6: 31-37. Katiio, P., 1951 The significance of nuclear quantity in the genus Micrasterias. Ann. Bot., Soc. Zool. Bot. Fenn. 24: 1-122. Lansinc, A. I., 1948 Evidence for aging as a consequence of growth cessation. Proc. Natl. Acad. Sci. U. S. 34: 304-310. LEDERBERG, E. M., and J. LepErBerc, 1953 Genetic studies of lysogenicity in Eschertchta colt. Genetics 38: 51-64. LEDERBERG, J., 1948 Detection of fermentative variants with tetrazolium. J. Bacteriol. 56: 695. 1949 Bacterial variation. Ann. Rev. Microbiol. 3: 1-22. 1954 A simple method for isolating single microbes. J. Bacteriol. 68: 258-259. 1956a Genetic transduction. Am. Scientist 44: 264-280. 1956b Conjugal pairing in Escherichia coli. J. Bacteriol. 71: 497-498. LepERBERG, J., and P. R. Epwarps, 1953 Serotypic recombination in Salmonella. J. Immunol. 71: 232-240. LEDERBERG, J., and T. Inno, 1956 Phase variation in Salmonella. Genetics 41: 743-757. LEDERBERG, J., and E. M. LeperBerc, 1956 Infection and heredity. Symposia Soc. Developm. Growth 14: 101-124. Lerrson, E., 1951 Staining, shape, and arrangement of bacterial flagella. J. Bacteriol. 62: 377- 389. Levintuar, C., 1955 The mechanism of DNA replication and genetic recombination in phage. Rend. Ist. Lombardo Sci. 89: 192-199. Lieg, M., 1953. The establishment of lysogenicity in Escherichia coli. J. Bacteriol. 65: 642-651. LINEAR INHERITANCE NI TRANSDUCTIONAL CLONES 871 Linz, R., 1954 La multiplication de Afycobacterium tuberculosis en présence de streptomycine. Ann. Inst. Pasteur 86: 334-346. Mazra, D., and W. PLaut, 1955 The distribution of parental material in chromosome reproduc- tion. Biol. Bull. 109: 335-336. McCienpon, J. F., 1909 Protozoan studies. J. Exptl. Zool. 6: 265-283. Morse, M. L., E. M. Leperserc, and J. Leperserc, 1956a Transduction in Escherichia coli K-12. Genetics 41: 142-156. 1956b Transductional heterogenotes in Escherichia coli. Genetics 41: 758-779. MOsTELLER, F., and J. W. Tukey, 1949 The uses and usefulness of binomial probability paper, J. Am. Stat. Assn. 44: 174-212. NEwcoMBE, H. B., 1953 The delayed appearance of radiation-induced genetic change in bac- teria. Genetics 38: 134-151, PennincTon, D., 1950 A reaction of bacterial cells with formaldehyde. J. Bacteriol. 59: 617-623. PROKOFIEVA-BELGOvSKAIA, A. A., 1946 “Mother” and ‘daughter? chromosomes. J. Heredity 37; 239-246. Rizet, G., 1953 Sur Pimpossibilité @obtenir la multiplication végétative ininterromque et illi- mitée de |’Ascomycéte Podospora anserina. Compt. Rend, 287: 838-840. ROTHSCHILD, 1956 Fertilisation. London: Methuen. Ryan, F. J., 1955 Spontaneous mutation in non-dividing bacteria. Genetics 40: 726-738. Ryan, F. J., and L. K. WAINWRIGHT, 1955 Nuclear segregation and the growth of clones of spon- taneous mutants of bacteria. J. Gen. Microbiol. 11: 364-379. Sites, J., F. V. Weccn, and W. J. Etrorp, 1948 The influence of antibacterial substances on the interaction of bacteria and bacteriophage. II. J. Gen. Microbiol. 2: 220-227, SONNEBORN, T. M., 1930 Genetic studies on Stenostomum incaudatum. J. Exptl. Zool. 57: 57-108. SPIEGELMAN, S., 1951 The particulate transmission of enzyme-forming capacity in yeast. Cold Spring Harbor Symposia Quant. Biol. 16: 87-98. Stocker, B. A. D., 1956a Bacterial flagella. Symposia Soc. Gen. Microbiol. 6: 19-40. 1956b Abortive transduction of motility in Salmonella; a non-replicated “gene”? transmitted through many generations to a single descendant. J. Gen. Microbiol. 15: 575-598. STOCKER, B. A. D., N. D. Zinper, and J. LEDERBERG, 1953 Transduction of flagellar characters in Salmonella. J. Gen. Microbiol. 9: 410-433. Symposium on Genetic Recombination, 1955 J. Cell. Comp. Physiol., 45: Suppl. 2. Waris, H., 1950 Cytophysiological studies on Micrasterias. II. The cytoplasmic framework and its mutation. Physiol. Plantarum 3: 236-246. WerBuLL, C., 1953 Observations on the staining of Bacillus megaterinm with triphenyltetra- zolium. J. Bacteriol. 66: 137-139. WIEDLING, S., 1948 Beitrage zur Kenntnis der vegetative Vermehrung bei Diatomeen. Botan. Notiser: 322-354. Witkin, E. M., 1951 Nuclear segregation and the delayed appearance of induced mutants in Escherichia coli. Cold Spring Harbor Symposia Quant. Biol. 16: 357-372. ZELLE, M. R., 1951 A simple single-cell method for genetic studies of bacteria. J. Bacteriol. 61: 345-349. Zinver, N. D., and J. LeperBerc, 1952 Genetic exchange in Salmonella. J. Bacteriol. 64: 679- 699.