A METHOD FOR FATE MAPPING THE FOCI OF LETHAL AND BEHAVIORAL MUTANTS I N DROSOPHILA MELANOGASTERI JAMES R. FLANAGAN Department of Genetics, University of Washington, Seattle, Washington 98195
Manuscript received April 12, 1976 Revised copy received December 10, 1976 ABSTRACT
A new method of mosaic fate mapping, called focusing, is introduced. Its advantages are that it allows a mapping, on the blastoderm surface, of the site of action of functions defined by either pre-adult lethal or behavioral mutations. Moreover, it does not require that the mosaics used be 50% of one genotype and 50% of the other. Methods for quantitative evaluation of the results of focusing, and a comparison of this method with others, are discussed.-Focusing is applied to the analysis of a new mutant, doomed (symbol: d m d ) , a distally located X-linked recessive in D. melanogaster, that causes the deaths of males and females around the time of eclosion. The dmd phenotype among eclosing flies is first seen as the loss of thoracic motor coordination and as ether sensitivity. Fate mapping by the method of focusing places the site of action of the dmd+ function in the same region of the map as that of the thoracic neural ganglia primordia, but not that of muscle, suggesting the possibility that the effect of dmd is on a thoracic neural function rather than a muscular one.
G E N E T I C mosaic fate mapping is a method originally suggested by STURTEVANT (1929) and first put in practice by GARCIA-BELLIDO and MERRIAM (1969). In its original form, the method was chiefly of developmental interest, since it showed the spatial relations among the various primordia of adult structures. Since then, a variant of the idea has been introduced by HOTTA and BENZER (1972), extending the scope of fate mapping to the analysis of behavioral mutants in Drosophila. The use of the method as such is to indicate what primary effect of a mutant, e.g., on brain us. musculature, causes the observed behavioral defect. In this report the scope of fate mapping is enlarged to include the analysis of lethal mutants and a few other situations. A lethal such as d m d (doomed), which expresses its phenotype only when the fly has become an adult, is of some interest if only because of the rarity of this class; d m d is one of but three such reported. The others are the fourth chromosome recessive 2(4)18 (HOCHMAN 1973) and drop-dead ( d r d ) , an X linked recessive (HOTTAand BENZER1972). The mutant drd has its primary effect on the brain, as evidenced by morphological abnormalities in mutant brains, and by placement of its site of action, by fate mapping, in the blastoderResearch sponsored jointly by grants G M 09965 and GM 00182 from the Public Health Service Genetics 85: 587-607 April, 1977
588
J.
R.
FLANAGAN
mal region from which the brain is derived (HOTTA and BENZER1972). 1(4)18 causes loss of locomotor coordination and early death. I n this report I show that dmd is a peri-eclosion lethal that causes a lesion in a thoracic function that is probably neural. It was necessary to develop a new method of fate mapping for the analysis of and BENZERin the case of the dmd. The method of fate mapping used by HOTTA adult lethal (behavioral) mutant, drd. is not applicable to the analysis of dmd or of any pre-eclosion o r peri-eclosion lethal. Focusing was developed to obviate the need for the classes of flies not recovered in such cases and to eliminate the assumption that the mosaics used must be 50% of each genotype (so called “50: 50 mosaics”). For these rea$ons, focusing is more generally applicable than is the method of HOTTA and BENZER.Y. HOTTA’S method of “contour mapping” (unpublished-see below for a description) is applicable to the analysis of lethals, but yields different information than does focusing. The relative merits of these two methods of analyzing lethals will be discussed. RESULTS A N D ANALYSIS
The origin and location of dmd: The mutant dmd, recovered in an EMStreated X chromosome, when in homozygous or hemizygous condition, invariably causes death within a few hours of eclosion. The mutation maps to the tip of the X chromosome less than 0.4 map units from the y locus. (This value is an overestimate, since a number of wild-type flies die soon after eclosion even though they are dmd+, and thus appear to be a recombinant class.) This was indicated in the first instance by the observation that, although y dmd-bearing males, with o r without a normal Y chromosome. die shortly after eclosion, such males carrying a Y chromosome which bears the tip of the X ( y + Y : LINDSI~EY and GRELL1968) are phenotypically normal. Thus, dmd should be almost inseparable from y. This is not inconsistent with the mapping of dmd with respect to the y locus. The mapping shows that the dmd phenotype involves a small genetic region, perhaps a single gene. The time of expression of dmd: Most dmd-bearing individuals eclose and die a short time thereafter (within one day). The distribution of these deaths, given as the percent which remain alive at various times after eclosion, is shown in Figure 1 f o r 72 dmd males. In order to find out when those dmd flies that are not recovered as adults die, several crosses were made to generate either dmd males, homozygous dmd females, or only dmd+ progeny in order to compare the viabilities of embryos, larvae, and pupae and to compare the progeny recovered both as dead late pupae (pharate adults) and as eclosed adults. These crosses, along with the numbers of progeny recovered in the various classes, are presented in Table 1. I n all three crosses, the parental female carries the y dmd chromosome balanced by M-5 (Zn(l)sCS1Lsc8E-k S, scsl sc8 w aB ) . I n the first cross listed in Table 1, all progeny are phenotypically dmd+ (except the nondisjunctional y male class) because the male parent carries a dmd+ allele on both its X and Y chromosomes. The
589
F A T E M A P P I N G IN DROSOPHILA
TIME AFTER E C L O S I O N (HOURS)
FIGURE 1.-The survival distribution of 72 dmd males that eclosed.
second cross generates d m d males, the y male class, and the third homozygous d m d females, the y female class. From these data it can be seen that the y d m d males are recovered 88% as frequently as are wild-type [2(0.9)100/(1.09 -I-0.95)], and homozygous d m d females are recovered 81% as frequently as are wild-type females [2(0.85) 100/ (1.O f 1.1) 1. These are the recoveries of both pharate and eclosed adults; eclosed adults represent about two-thirds of the total in each case. There were no significant differences among these crosses in their egg hatch o r larval pupation frequencies, and therefore the remaining d m d individuals must have died during pupation. This establishes that all d m d eggs hatch and larvae pupate; about 15% of the;e die before the late pupa stage, 25% die as late pupae, and the remaining 60% die within 24 hours of eclosion. Other properties of the dmd phenotype: Visible effects of d m d on enclosed adults were examined just prior to death to see if this would suggest the nature of the function altered in d m d flies. The first defect seen is loss of coordination of leg movements; however, flies remain alive long after the legs cease movement. I n bilaterally mosaic flies (for construction, see next section) that are heteroTABLE 1 Crosses to determine the relaiiue recovery of dmd flies as adults
+
135 (0.88)
81 (1.0)
82 (1.01)
68 (0.87)
y dmd ___
86
74
61
y+Y
(1.0’)
(0.86)
(0.71)
Y x -__
M-5,B ydmd
Y x __-
M-5,B
x
__-
___
ydmd
154 (1.0)
(1.11)
M-5,B -ydmd
y+ Y
+
Y
171
2 (0.01)
73 (0.90) 0 (0.0)
168 (1.09) 0 (0.0)
82
(0.95)
The upper number is the total number of progeny recovered both as late pupae (pharate adults) and as adults. The lower number, in parentheses, is the relative rate of recovery of each class compared to that of the common class-the B 0 0 .
590
J . R. F L A N A G A N
zygous d m d + on one side and hemizygous dmd on the other, the initial loss of motor functions is seen only on the side that is hemizygous dmd. Thus d m d is autonomous with respect to its initial effects; but the dysfunction owing to d m d on one side of the mosaic is followed by premature death. Finally, dmd flies are highly ether sensitive. When etherized even slightly, no d m d flies recover; they do, however, recover from lengthy CO, anesthetization. A cursory histological examination of eclosed d m d males was done. Newly dead adults were imbedded in plastic, serially sectioned ( 4 p sections), the sections stained with Richardson's stain and examined at low power with phase contrast optics. No obvious defects were found in the integrity of the flight musculature, various parts of the brain, nor of the thoracic nervous system. Identifying the tissue in w h i c h the dmd+ function acts: The tissue in which a function defined by a mutation acts is known as the focus of that mutation. The focus of d m d can be identified by making individuals mosaic for the allele of d m d they carry and then determining which tissues must contain the d m d + allele to insure that the mosaic will survive more than one day of adult life. The mosaics used here were generated by mating y d m d / y + Y males with females carrying the unstable ring-X chromosome, I n ( l ) w v c (HINTON 1955). which is y + and d m d + , and then collecting the mosaic progeny which bear yellow as well as brown cuticular structures. The property of the ring-X chromosome of interest here is that it is frequently lost from one of the daughter cells at the first cleavage division in the embryo, so that about 50% of the cells of the resulting individuals have no y + d m d + X chromosome. The results of this cross, and a control cross in which there is no d m d allele, are given in Table 2. Also given is the number of mosaics that survived more than one day after eclosion. If the d m d + allele is required in only one type of tissue, then it can be expected to be used in two such sites per fly, because the organism is bilaterally symmetrical. As a first indication that it may be required in only two such sites, and to exemplify the calculation of the fraction of dmd mosaics conceiued which survive TABLE 2 The frequency of suruival of flies mosaic for dmd
??
Cross
In(l),wvc ___ Y
W
Progeny
dd
y X-
Y
In(l),wvc y dmd ~X -__ Y W Y+ y
Y??
754 (1.0) 1795 (1.0)
Y'?
?
ydd
Gynandromorphs Recovered total Survivors'
475 (0.63)
109 (0.15)
181 181 (0.24) (0.24)
883
205 (0.11)
269 1013 (0.15) (0.057)
(0.49)
The unstable ring-X (In(l),wvc) of HINTON(1955) was used to produce gynandromorphs composed of dmd+/dmd+ and dmd+/O tissue in the case of the control (first cross), or of dmd/dmd+ and dmd/O tissue in the case of the experimental (second) cross. The upper number under each class is the total number of newly eclosed adults recovered; the lower number, in parentheses, represents this as a fraction of the y female class. Flies living more than one day as adults are considered to be effectively dmd+, having survived the period during which dmd causes death. +
F A T E M A P P I N G I N DROSOPHILA
591
more than one day after eclosion, consider the following. It is inferred from the average frequency that a bristle on a control mosaic is brown (therefore y+ dmd+-bearing) that the frequency that any cell at the blastoderm stage contained the dmd+ allele is 0.49. Assuming, for the moment, that the dmd+ allele is required in both of two foci per individual to effect survival and that the genotypes of these two sites in the blastoderm are independent, then, since the probability that either site carries the dmd+ allele is 0.49, one expects that 24% of all mosaics produced in the experimental cross will survive more than one day of adulthood. In order that the ratio between the total mosaics and the y females (a common class between crosses) may be the same in the experimental as in the control cross, there must have been 431 mosaics in the dmd-cross before the occurrence of any deaths caused by dmd. Of these, 103 mosaics, or 24%,survived more than one day after eclosion. This result suggests, though it does not prove, that the dmd+ function is required in only two sites. or foci, per blastoderm. Note that one is concerned, from the point of view of mosaic analysis, only with the number of foci as they exist at the blastoderm stage, rather than with the and number of sites in the adult that are descended from those. As do HOTTA BENZER,I will assume a two-focus model in order to ask the more detailed question of where in the blastoderm these foci are, with the hope that this knowledge will suggest the type of tissue in which the dmd+ function is required. Also, as in the case of the HOTTA and BENZERanalysis, the only support for the two-focus assumption will be in how well the method works; that is, how strongly the data can be shown to map to a single site per side in the blastoderm. Fate mapping: To identify the foci of dmd, one seeks their positions on the blastoderm relative to those of the blastodermal origins of visible adult structures (bristles and other cuticular parts). This will be done using a procedure that is ultimately based upon the mosaic fate mapping method developed by GARCIABELLIDOand MERRIAM (1 969). With their method, one makes a two-dimensional representation of one side (right-left) of the blastoderm. On this map is sought the location of one of dmd's foci. Mapping the adult structures involves first scoring their colors ( y versus y+) in XX-XO mosaics (diplo/haplo-X mosaics or gynandromorphs) and then using the frequency with which members of each pair of structures differ in hue on a side of a mosaic as a measure of the distance between the structures' blastodermal origins. The frequency, expressed as a percentage, is known as the number of Sturts between the blastodermal sites. These distances are used to triangulate a planar map of the blastoderm (see GARCIA-BELLIDO and MERRIAM 1969, for a defense of this procedure). This triangulation is performed automatically by a computerized least-squares procedure (FLANAGAN 1976) which also results in a quantitative measure of how well each mapped site is placed. (The programs which perform the least-squares procedures and their explanations are available from the author on request.) In this procedure, a set of rectangular coordinate locations (in Sturt units) is assigned to each structure. From the coordinates of two sites, one can predict the distance between them that should be observed were the sites in fact at those locations. This distance is compared to the observed
5 92
J. R. FLANAGAN
distance. The squared difference between these distances, divided by the observed distance, and summed over all such comparisoins between a particular site and all others, is denfied as the error-of-placement of that site. The values of such errars, one for each site mapped, are minimized by the least-squares procedure. A map using this procedure with data from the control mosaics of Table 2 is shown in Figure 2. The distribution of errors-of-placement for the sites on this p a p is shown in Figure 3. This distribution is important as a reference point representing the results of a relatively simple and direct method of ascertaining distances between blastodermal sites to which one can compare the results of a more complicated procedure such as focusing. The sites shown in Figure 2 will be used as landmarks in locating the focus of dmd, once the distances between these sites and the focus are known. Since the dmd phenotype of each mosaic is assumed to depend on the genotypes of both the right and left foci, the phenotypes of these two foci are not independently scorable. Following the convention of HOTTA and BENZER,if both foci must carry the dmd+allele to insure survival, then the mutant (hemizygous dmd) focus is called domineering. If, on the other hand, both foci must lack the dmd+ allele to insure that death ensues, then the mutant focus is called submissive. It is assumed that a mutant focus is either strictly domineering or strictly submissive. In either case, the analysis is complicated by the fact that the frequency that any landmark (A in Figure 4) is y among mosaics with a dmd+
50-
2
.
”
TER
5 6
T
r 30-
L2.
.L3 L?.
.Ph
*SS
.AR
STE
.
3
2
PY.
GEN.
. 4 5
.ORP
.
6
I 0
50
30
70
-
STURTS
FIGURE 2.-A least-squares fate map of data from unstable-ring-X loss mosaics. Abbreviations used are: ADC-anterior dorsocentral, ANP-anterior notopleural, AR-arista, ASC-anterior scutellar, CO-costa, GEN-genetalia, HU-humeral, IV-inner vertical, L (1-3) -legs, OC-ocellar, ORAanterior orbital, ORP-posterior orbital, OV-outer vertical, PA-palpus, PDC-posterior dorsocentral, PNP-posterior no topleural, PRO proboscus, PS-presutrals, PSA-posterior supra-alar, PCS-posterior scutellar, PV-post-vertical, SP-sternopleural, SS-sternal seta, STE (2-6) -sternites, TER (2-7) tergites, VB-vibrissae, W-wing.
593
F A T E M A P P I N G I N DROSOPHILA
ERROR
FIGURE 3.-The distribution of the landmarks with respect tc the error on their map positions. Pairwise comparison of map positions generates predicted distannces which differ from the observed distances, so that the error on a position is defined as: E = Z (predicted - observed) */ observed, summed over n = 59 comparisons, defines the error of placement for each of the 60 landmarks.
phenotype depends both on the distance, S, between A and focus on the same side of the blastoderm and on the distance, L, between the foci. The mosaic analysis of lethals and that of behavioral mutants share this comand BENZERsolves this problem for the case of plexity. The method of HOTTA behavioral mutants when the mosaics generated are half male (haplo-X) and half female (diplo-X) (or, more generally, half one genotype and half the other, the so-called 50: 50 mosaic). When mosaics are other than 50: 50 there are, in their analysis, more unknowns than can be solved. Similarly, in the case of a pre- or peri-eclosion lethal, certain of their classes of mosaics are not recovered (i.e., those which have the mutant behavior) ; this results in too few knowns to solve
0
\
0
8
8
0
\
\
8
\
8
I
8
f
8
I
8
I
\ I
I
I I I
I I
L
I I
8
$
8
\
8 8
8
0
8
8
8
0
LANDMARK \
0 0
FIGURE 4.-A dorsal view of the blastoderm as envisioned in the theory of fate-mapping the focus of a behavioral or lethal mutation.
594
J. R. FLANAGAN
the unknowns. The method of focusing is applicable to mutants with both preadult and adult phenotypes, whether or not 50: 50 mosaics are used. The method of focusing: The theoretical aspects of the method are dealt with first; after these are explained, sample calculations will be shown using d m d as an example. The first step of focusing is to calculate. from the directly observable quantities, two parameters ( q and 4’) for each landmark; the manner in which these are calculated depends only on whether one is dealing with an adult or a pre-adult phenotype. The second step is to calculate the distance between a landmark and the focus of the mutant from the values of q and q’ for that landmark. The calculations depend on whether the focus is domineering or submissive, a matter that will be decided upon inspection of the relative values of q and q’. Also derived is a calculation of the distance between right and left foci applicable to all mosaic analysis situations. This distance varies from region to region of the map, presumably because the blastoderm is not box-shaped (as the planar-mapping assumption implies) but is rounded like a cigar. Therefore, sites on the blastoderm far from the dorsal-ventral midline (separating right from left halves) will have larger interlandmark o r interfocus distances than will sites closer to this midline. (Note that because of a convenience in calculation, when distances are referred to as S or L, as in the equations to follow, the distance scale is from 0.0 through 1.0, rather than the 0.0 through 100 scale of Sturts which I use for the map scales because of convention.) Calculating q and q’: All mosaics can be classified as indicated in Figure 5 into one of four categories (W, X, Y, and Z). The value of q is defined as the ratio of the size of class Z to the sum of those of classes Z and Y. The value of q’ is defined as the ratio of the size of class W to the sum of those of classes W a n d X. When all mosaics are recoverable, as in the case of an adult behavioral mutant, the sizes of all these classes are known, and the parameters q and q’ can be calculated by the equations in Figure 5A. Notice that the penetrance of the mutant in the homozygous or hemizygous state may be less than 1.0; that is, some flies which (genetically) should show mutant behavior do not. Making the assumption that its penetrance in such individuals applies also to the mosaic situation (actually this is the meaning of having assumed that a mutant focus is either strictly domineering or strictly submissive), then the values q and are calculated by the equations of Figure SA, which allow for a penetrance less than 1.O. When dealing with a lethal analysis, however, classes W and Y are not recovered; only a fraction, T, of the mosaics are recovered. In order to deal with this situation, the value of T must be found by comparing the results of a cross in which the flies are mosaic only f o r some indicator phenotypes (such as those of y and yf) as a control, with those of a cross in which they are also mosaic for the lethal (experimental). As was shown earlier for the data in Table 2, this comparison reveals the total number of mosaics produced out of which the surviving mosaics are recovered. The ratio of the surviving number to the total number is equal to T ; this is also equal to the sum of classes X and Z (whose sizes are expressed as fractions of the total). (The quantity F in Figure 5 is defined as the sum of classes Y and Z; its value for a particular landmark is y+
595
F A T E M A P P I N G IN DROSOPHILA
I
behavior:
MOSAIC CLASSES W+X+Y+Z
= 1.0
landmork
F=Y+Z P = PENETRANCE
I\
BEHAVIORAL MUTANTS Pz1.0:
q
=+ -w - 1-F
B
IN GENERAL :
q'= 7%
LETHAL MUTANTS f = Z/(X+Z) P=1.0 : T=X+Z
FIGURE 5.-Calculating
IN GENERAL :
-
T = 1 P + P(X+Z)
the parameters necessary for focusing.
amcng the control mosaics. The quantity f is defined as the frequency that this landmark is y+ among surviving mosaics of the experimental cross. The calculation of q and # from these quantities is shown in Figure 5B. As before, equations are also given f o r the case of a penetrance less than 1.0. I n that case, which is however, the fraction of mosaics which survive is merely Tobserved, greater than the fraction which, genetically, should survive. It must be corrected as follows to give T for calculations: T = 1.0 4-P (Tobserved - 1.0). Calculating distances: The distances of interest are that between the foci, L, and that between a landmark and a focus, S. The former is the easier to calculate. To do so requires knowing the average frequency that a landmark is yf among control mosaics; this quantity is referred to as Favgin calculating L in Figure 6. The mosaic can be divided (in theory but not directly by observation) into the three classes of Figure 6 according to the genotypes of the foci per blastoderm (the symbol 8 designates the hemizygous dmd state and the symbol 0 designates the heterozygous d m d + state). Under the domineering model (Figure 6A) the value of T , as defined, is also equal to the probability that both foci are female (diplo-X), o r to class NI in Figure 6. The size of NI is related to Favgand L as in Figure 6A. These relations reduce to equation 1 of Figure 6A giving L as a function of T and Fa,,&.Similarly, for the submissive model, the value of T is also equal to the probability that it is not the case that both foci are male, or
596
J. R . F L A N A G A N
G e n o t y p e s of Foci
:
Fraction
A
of.
1 and 2
Mosaics
s
d
6
0
D o m i n e e r i n g Model = F a v g (1 - 1l2L;V LM )
J L
9
Fa,,
0
"
= M t &L
t = d i s t a n c e between foci
FIGURE 6.-Calculating carried the
+ allele and
'I
the distance ( L ) between foci. The genotype designator '' $! " means
8" means the opposite.
FIGURE 7.-The types of mosaics represented by parameter q (panel A) and q' (panel B); the ovals represent dorsal views of blastoderms, the open circles a landmark, and the closed circles the foci of the mutant being fate mapped. The shaded areas contain the wild-type alleles of y and of the behavioral mutant.
F A T E M A P P I N G IN DROSOPHILA
597
to 1.0 - 1.Equation 2 of Figure 6B relates L to T and Favgf o r the submissive model. The values of q and q’ f o r a landmark are related t o the distance between that landmark and the focus on the same side of the blastoderm (S in Figure 4).I n order to see this relationship, it is convenient to interpret q and q’ as follows. The value of q can be interpreted as the probability that a mosaic will survive, given that landmark A (any particular landmark) is female ( y + ) . The events this might represent in terms of the position of the border between male and female nuclei in a mosaic relative to the positions of the landmark and the foci depends on whether one is dealing with a domineering or a submissive mutant focus. Figure 7A deals with the definition of q. If landmark A (open circle) is female (shaded in Figure 7 ) and the focus (closed circles) is domineering, then the border must not pass among three sites, i.e., the landmark and the two foci, if the mosaic is to survive. If the focus is submissive, then any of the events listed under the heading submissive in Figure 7A will allow the mosaic to survive. T h e value of q’, on the other hand, can be interpreted as the probability that a mosaic will not survive given that landmark A is male. The events this might represent are dealt with in Figure 7B. The probabilities of any of the events in Figure 7 can be written as proportional to the product of either S or 1 -S with L or 1-L. For instance, the probability of the event in Figure 7A under the heading domineering is proportional to the probability that the mosaic border does not pass between the landmark and the nearer focus, 1-S, multiplied by the probability that this border does not pass between the foci, I-L. The reasons that it is not equal to this product are two. The first is that the probabilities of these events are not the same when landmark A is male as when female, unless one uses 50:50 mosaics. The second reason is that the product of the probabilities of the two requirements, l-S and l-& is exactly correct only if these two requirements are independent (they may be, but need not always be so). The relationships among q, q’, S and L are shown for the domineering model in Figure SA and for the submissive model in Figure 8B. These relations become equations upon introducing proportionality constants. Moreover, the equations for q and q‘ are reduced to a single equation relating S to q and q’ and an unknown quantity: K (equations 1 and 2 for the domineering and submissive models. respectively). Determining which model is correct is straightforward in that, if a mutant focus is domineering, then q‘ must always be greater than q; if submissive, then the opposite is true. O n the value of K: First note that it is assumed that one value of K will suffice for the calculation of S for all landmarks. It will be shown that this assumption is in fact nearly true, in that a single value for K gives very good results in control-simulations and in the analysis of dmd. The proper value of K to be used can be arrived at in one of two ways, the first being the more intuitively straightforward, but the second being the easier and more accurate method. The first way in which the value of K can be arrived at is through control simulations of lethal foci. From a set of control mosaics, one selects those mosaics
598
J. R. FLANAGAN
A
B
Dominee ing M o d e l
Manuscript received April 12, 1976 Revised copy received December 10, 1976 ABSTRACT
A new method of mosaic fate mapping, called focusing, is introduced. Its advantages are that it allows a mapping, on the blastoderm surface, of the site of action of functions defined by either pre-adult lethal or behavioral mutations. Moreover, it does not require that the mosaics used be 50% of one genotype and 50% of the other. Methods for quantitative evaluation of the results of focusing, and a comparison of this method with others, are discussed.-Focusing is applied to the analysis of a new mutant, doomed (symbol: d m d ) , a distally located X-linked recessive in D. melanogaster, that causes the deaths of males and females around the time of eclosion. The dmd phenotype among eclosing flies is first seen as the loss of thoracic motor coordination and as ether sensitivity. Fate mapping by the method of focusing places the site of action of the dmd+ function in the same region of the map as that of the thoracic neural ganglia primordia, but not that of muscle, suggesting the possibility that the effect of dmd is on a thoracic neural function rather than a muscular one.
G E N E T I C mosaic fate mapping is a method originally suggested by STURTEVANT (1929) and first put in practice by GARCIA-BELLIDO and MERRIAM (1969). In its original form, the method was chiefly of developmental interest, since it showed the spatial relations among the various primordia of adult structures. Since then, a variant of the idea has been introduced by HOTTA and BENZER (1972), extending the scope of fate mapping to the analysis of behavioral mutants in Drosophila. The use of the method as such is to indicate what primary effect of a mutant, e.g., on brain us. musculature, causes the observed behavioral defect. In this report the scope of fate mapping is enlarged to include the analysis of lethal mutants and a few other situations. A lethal such as d m d (doomed), which expresses its phenotype only when the fly has become an adult, is of some interest if only because of the rarity of this class; d m d is one of but three such reported. The others are the fourth chromosome recessive 2(4)18 (HOCHMAN 1973) and drop-dead ( d r d ) , an X linked recessive (HOTTAand BENZER1972). The mutant drd has its primary effect on the brain, as evidenced by morphological abnormalities in mutant brains, and by placement of its site of action, by fate mapping, in the blastoderResearch sponsored jointly by grants G M 09965 and GM 00182 from the Public Health Service Genetics 85: 587-607 April, 1977
588
J.
R.
FLANAGAN
mal region from which the brain is derived (HOTTA and BENZER1972). 1(4)18 causes loss of locomotor coordination and early death. I n this report I show that dmd is a peri-eclosion lethal that causes a lesion in a thoracic function that is probably neural. It was necessary to develop a new method of fate mapping for the analysis of and BENZERin the case of the dmd. The method of fate mapping used by HOTTA adult lethal (behavioral) mutant, drd. is not applicable to the analysis of dmd or of any pre-eclosion o r peri-eclosion lethal. Focusing was developed to obviate the need for the classes of flies not recovered in such cases and to eliminate the assumption that the mosaics used must be 50% of each genotype (so called “50: 50 mosaics”). For these rea$ons, focusing is more generally applicable than is the method of HOTTA and BENZER.Y. HOTTA’S method of “contour mapping” (unpublished-see below for a description) is applicable to the analysis of lethals, but yields different information than does focusing. The relative merits of these two methods of analyzing lethals will be discussed. RESULTS A N D ANALYSIS
The origin and location of dmd: The mutant dmd, recovered in an EMStreated X chromosome, when in homozygous or hemizygous condition, invariably causes death within a few hours of eclosion. The mutation maps to the tip of the X chromosome less than 0.4 map units from the y locus. (This value is an overestimate, since a number of wild-type flies die soon after eclosion even though they are dmd+, and thus appear to be a recombinant class.) This was indicated in the first instance by the observation that, although y dmd-bearing males, with o r without a normal Y chromosome. die shortly after eclosion, such males carrying a Y chromosome which bears the tip of the X ( y + Y : LINDSI~EY and GRELL1968) are phenotypically normal. Thus, dmd should be almost inseparable from y. This is not inconsistent with the mapping of dmd with respect to the y locus. The mapping shows that the dmd phenotype involves a small genetic region, perhaps a single gene. The time of expression of dmd: Most dmd-bearing individuals eclose and die a short time thereafter (within one day). The distribution of these deaths, given as the percent which remain alive at various times after eclosion, is shown in Figure 1 f o r 72 dmd males. In order to find out when those dmd flies that are not recovered as adults die, several crosses were made to generate either dmd males, homozygous dmd females, or only dmd+ progeny in order to compare the viabilities of embryos, larvae, and pupae and to compare the progeny recovered both as dead late pupae (pharate adults) and as eclosed adults. These crosses, along with the numbers of progeny recovered in the various classes, are presented in Table 1. I n all three crosses, the parental female carries the y dmd chromosome balanced by M-5 (Zn(l)sCS1Lsc8E-k S, scsl sc8 w aB ) . I n the first cross listed in Table 1, all progeny are phenotypically dmd+ (except the nondisjunctional y male class) because the male parent carries a dmd+ allele on both its X and Y chromosomes. The
589
F A T E M A P P I N G IN DROSOPHILA
TIME AFTER E C L O S I O N (HOURS)
FIGURE 1.-The survival distribution of 72 dmd males that eclosed.
second cross generates d m d males, the y male class, and the third homozygous d m d females, the y female class. From these data it can be seen that the y d m d males are recovered 88% as frequently as are wild-type [2(0.9)100/(1.09 -I-0.95)], and homozygous d m d females are recovered 81% as frequently as are wild-type females [2(0.85) 100/ (1.O f 1.1) 1. These are the recoveries of both pharate and eclosed adults; eclosed adults represent about two-thirds of the total in each case. There were no significant differences among these crosses in their egg hatch o r larval pupation frequencies, and therefore the remaining d m d individuals must have died during pupation. This establishes that all d m d eggs hatch and larvae pupate; about 15% of the;e die before the late pupa stage, 25% die as late pupae, and the remaining 60% die within 24 hours of eclosion. Other properties of the dmd phenotype: Visible effects of d m d on enclosed adults were examined just prior to death to see if this would suggest the nature of the function altered in d m d flies. The first defect seen is loss of coordination of leg movements; however, flies remain alive long after the legs cease movement. I n bilaterally mosaic flies (for construction, see next section) that are heteroTABLE 1 Crosses to determine the relaiiue recovery of dmd flies as adults
+
135 (0.88)
81 (1.0)
82 (1.01)
68 (0.87)
y dmd ___
86
74
61
y+Y
(1.0’)
(0.86)
(0.71)
Y x -__
M-5,B ydmd
Y x __-
M-5,B
x
__-
___
ydmd
154 (1.0)
(1.11)
M-5,B -ydmd
y+ Y
+
Y
171
2 (0.01)
73 (0.90) 0 (0.0)
168 (1.09) 0 (0.0)
82
(0.95)
The upper number is the total number of progeny recovered both as late pupae (pharate adults) and as adults. The lower number, in parentheses, is the relative rate of recovery of each class compared to that of the common class-the B 0 0 .
590
J . R. F L A N A G A N
zygous d m d + on one side and hemizygous dmd on the other, the initial loss of motor functions is seen only on the side that is hemizygous dmd. Thus d m d is autonomous with respect to its initial effects; but the dysfunction owing to d m d on one side of the mosaic is followed by premature death. Finally, dmd flies are highly ether sensitive. When etherized even slightly, no d m d flies recover; they do, however, recover from lengthy CO, anesthetization. A cursory histological examination of eclosed d m d males was done. Newly dead adults were imbedded in plastic, serially sectioned ( 4 p sections), the sections stained with Richardson's stain and examined at low power with phase contrast optics. No obvious defects were found in the integrity of the flight musculature, various parts of the brain, nor of the thoracic nervous system. Identifying the tissue in w h i c h the dmd+ function acts: The tissue in which a function defined by a mutation acts is known as the focus of that mutation. The focus of d m d can be identified by making individuals mosaic for the allele of d m d they carry and then determining which tissues must contain the d m d + allele to insure that the mosaic will survive more than one day of adult life. The mosaics used here were generated by mating y d m d / y + Y males with females carrying the unstable ring-X chromosome, I n ( l ) w v c (HINTON 1955). which is y + and d m d + , and then collecting the mosaic progeny which bear yellow as well as brown cuticular structures. The property of the ring-X chromosome of interest here is that it is frequently lost from one of the daughter cells at the first cleavage division in the embryo, so that about 50% of the cells of the resulting individuals have no y + d m d + X chromosome. The results of this cross, and a control cross in which there is no d m d allele, are given in Table 2. Also given is the number of mosaics that survived more than one day after eclosion. If the d m d + allele is required in only one type of tissue, then it can be expected to be used in two such sites per fly, because the organism is bilaterally symmetrical. As a first indication that it may be required in only two such sites, and to exemplify the calculation of the fraction of dmd mosaics conceiued which survive TABLE 2 The frequency of suruival of flies mosaic for dmd
??
Cross
In(l),wvc ___ Y
W
Progeny
dd
y X-
Y
In(l),wvc y dmd ~X -__ Y W Y+ y
Y??
754 (1.0) 1795 (1.0)
Y'?
?
ydd
Gynandromorphs Recovered total Survivors'
475 (0.63)
109 (0.15)
181 181 (0.24) (0.24)
883
205 (0.11)
269 1013 (0.15) (0.057)
(0.49)
The unstable ring-X (In(l),wvc) of HINTON(1955) was used to produce gynandromorphs composed of dmd+/dmd+ and dmd+/O tissue in the case of the control (first cross), or of dmd/dmd+ and dmd/O tissue in the case of the experimental (second) cross. The upper number under each class is the total number of newly eclosed adults recovered; the lower number, in parentheses, represents this as a fraction of the y female class. Flies living more than one day as adults are considered to be effectively dmd+, having survived the period during which dmd causes death. +
F A T E M A P P I N G I N DROSOPHILA
591
more than one day after eclosion, consider the following. It is inferred from the average frequency that a bristle on a control mosaic is brown (therefore y+ dmd+-bearing) that the frequency that any cell at the blastoderm stage contained the dmd+ allele is 0.49. Assuming, for the moment, that the dmd+ allele is required in both of two foci per individual to effect survival and that the genotypes of these two sites in the blastoderm are independent, then, since the probability that either site carries the dmd+ allele is 0.49, one expects that 24% of all mosaics produced in the experimental cross will survive more than one day of adulthood. In order that the ratio between the total mosaics and the y females (a common class between crosses) may be the same in the experimental as in the control cross, there must have been 431 mosaics in the dmd-cross before the occurrence of any deaths caused by dmd. Of these, 103 mosaics, or 24%,survived more than one day after eclosion. This result suggests, though it does not prove, that the dmd+ function is required in only two sites. or foci, per blastoderm. Note that one is concerned, from the point of view of mosaic analysis, only with the number of foci as they exist at the blastoderm stage, rather than with the and number of sites in the adult that are descended from those. As do HOTTA BENZER,I will assume a two-focus model in order to ask the more detailed question of where in the blastoderm these foci are, with the hope that this knowledge will suggest the type of tissue in which the dmd+ function is required. Also, as in the case of the HOTTA and BENZERanalysis, the only support for the two-focus assumption will be in how well the method works; that is, how strongly the data can be shown to map to a single site per side in the blastoderm. Fate mapping: To identify the foci of dmd, one seeks their positions on the blastoderm relative to those of the blastodermal origins of visible adult structures (bristles and other cuticular parts). This will be done using a procedure that is ultimately based upon the mosaic fate mapping method developed by GARCIABELLIDOand MERRIAM (1 969). With their method, one makes a two-dimensional representation of one side (right-left) of the blastoderm. On this map is sought the location of one of dmd's foci. Mapping the adult structures involves first scoring their colors ( y versus y+) in XX-XO mosaics (diplo/haplo-X mosaics or gynandromorphs) and then using the frequency with which members of each pair of structures differ in hue on a side of a mosaic as a measure of the distance between the structures' blastodermal origins. The frequency, expressed as a percentage, is known as the number of Sturts between the blastodermal sites. These distances are used to triangulate a planar map of the blastoderm (see GARCIA-BELLIDO and MERRIAM 1969, for a defense of this procedure). This triangulation is performed automatically by a computerized least-squares procedure (FLANAGAN 1976) which also results in a quantitative measure of how well each mapped site is placed. (The programs which perform the least-squares procedures and their explanations are available from the author on request.) In this procedure, a set of rectangular coordinate locations (in Sturt units) is assigned to each structure. From the coordinates of two sites, one can predict the distance between them that should be observed were the sites in fact at those locations. This distance is compared to the observed
5 92
J. R. FLANAGAN
distance. The squared difference between these distances, divided by the observed distance, and summed over all such comparisoins between a particular site and all others, is denfied as the error-of-placement of that site. The values of such errars, one for each site mapped, are minimized by the least-squares procedure. A map using this procedure with data from the control mosaics of Table 2 is shown in Figure 2. The distribution of errors-of-placement for the sites on this p a p is shown in Figure 3. This distribution is important as a reference point representing the results of a relatively simple and direct method of ascertaining distances between blastodermal sites to which one can compare the results of a more complicated procedure such as focusing. The sites shown in Figure 2 will be used as landmarks in locating the focus of dmd, once the distances between these sites and the focus are known. Since the dmd phenotype of each mosaic is assumed to depend on the genotypes of both the right and left foci, the phenotypes of these two foci are not independently scorable. Following the convention of HOTTA and BENZER,if both foci must carry the dmd+allele to insure survival, then the mutant (hemizygous dmd) focus is called domineering. If, on the other hand, both foci must lack the dmd+ allele to insure that death ensues, then the mutant focus is called submissive. It is assumed that a mutant focus is either strictly domineering or strictly submissive. In either case, the analysis is complicated by the fact that the frequency that any landmark (A in Figure 4) is y among mosaics with a dmd+
50-
2
.
”
TER
5 6
T
r 30-
L2.
.L3 L?.
.Ph
*SS
.AR
STE
.
3
2
PY.
GEN.
. 4 5
.ORP
.
6
I 0
50
30
70
-
STURTS
FIGURE 2.-A least-squares fate map of data from unstable-ring-X loss mosaics. Abbreviations used are: ADC-anterior dorsocentral, ANP-anterior notopleural, AR-arista, ASC-anterior scutellar, CO-costa, GEN-genetalia, HU-humeral, IV-inner vertical, L (1-3) -legs, OC-ocellar, ORAanterior orbital, ORP-posterior orbital, OV-outer vertical, PA-palpus, PDC-posterior dorsocentral, PNP-posterior no topleural, PRO proboscus, PS-presutrals, PSA-posterior supra-alar, PCS-posterior scutellar, PV-post-vertical, SP-sternopleural, SS-sternal seta, STE (2-6) -sternites, TER (2-7) tergites, VB-vibrissae, W-wing.
593
F A T E M A P P I N G I N DROSOPHILA
ERROR
FIGURE 3.-The distribution of the landmarks with respect tc the error on their map positions. Pairwise comparison of map positions generates predicted distannces which differ from the observed distances, so that the error on a position is defined as: E = Z (predicted - observed) */ observed, summed over n = 59 comparisons, defines the error of placement for each of the 60 landmarks.
phenotype depends both on the distance, S, between A and focus on the same side of the blastoderm and on the distance, L, between the foci. The mosaic analysis of lethals and that of behavioral mutants share this comand BENZERsolves this problem for the case of plexity. The method of HOTTA behavioral mutants when the mosaics generated are half male (haplo-X) and half female (diplo-X) (or, more generally, half one genotype and half the other, the so-called 50: 50 mosaic). When mosaics are other than 50: 50 there are, in their analysis, more unknowns than can be solved. Similarly, in the case of a pre- or peri-eclosion lethal, certain of their classes of mosaics are not recovered (i.e., those which have the mutant behavior) ; this results in too few knowns to solve
0
\
0
8
8
0
\
\
8
\
8
I
8
f
8
I
8
I
\ I
I
I I I
I I
L
I I
8
$
8
\
8 8
8
0
8
8
8
0
LANDMARK \
0 0
FIGURE 4.-A dorsal view of the blastoderm as envisioned in the theory of fate-mapping the focus of a behavioral or lethal mutation.
594
J. R. FLANAGAN
the unknowns. The method of focusing is applicable to mutants with both preadult and adult phenotypes, whether or not 50: 50 mosaics are used. The method of focusing: The theoretical aspects of the method are dealt with first; after these are explained, sample calculations will be shown using d m d as an example. The first step of focusing is to calculate. from the directly observable quantities, two parameters ( q and 4’) for each landmark; the manner in which these are calculated depends only on whether one is dealing with an adult or a pre-adult phenotype. The second step is to calculate the distance between a landmark and the focus of the mutant from the values of q and q’ for that landmark. The calculations depend on whether the focus is domineering or submissive, a matter that will be decided upon inspection of the relative values of q and q’. Also derived is a calculation of the distance between right and left foci applicable to all mosaic analysis situations. This distance varies from region to region of the map, presumably because the blastoderm is not box-shaped (as the planar-mapping assumption implies) but is rounded like a cigar. Therefore, sites on the blastoderm far from the dorsal-ventral midline (separating right from left halves) will have larger interlandmark o r interfocus distances than will sites closer to this midline. (Note that because of a convenience in calculation, when distances are referred to as S or L, as in the equations to follow, the distance scale is from 0.0 through 1.0, rather than the 0.0 through 100 scale of Sturts which I use for the map scales because of convention.) Calculating q and q’: All mosaics can be classified as indicated in Figure 5 into one of four categories (W, X, Y, and Z). The value of q is defined as the ratio of the size of class Z to the sum of those of classes Z and Y. The value of q’ is defined as the ratio of the size of class W to the sum of those of classes W a n d X. When all mosaics are recoverable, as in the case of an adult behavioral mutant, the sizes of all these classes are known, and the parameters q and q’ can be calculated by the equations in Figure 5A. Notice that the penetrance of the mutant in the homozygous or hemizygous state may be less than 1.0; that is, some flies which (genetically) should show mutant behavior do not. Making the assumption that its penetrance in such individuals applies also to the mosaic situation (actually this is the meaning of having assumed that a mutant focus is either strictly domineering or strictly submissive), then the values q and are calculated by the equations of Figure SA, which allow for a penetrance less than 1.O. When dealing with a lethal analysis, however, classes W and Y are not recovered; only a fraction, T, of the mosaics are recovered. In order to deal with this situation, the value of T must be found by comparing the results of a cross in which the flies are mosaic only f o r some indicator phenotypes (such as those of y and yf) as a control, with those of a cross in which they are also mosaic for the lethal (experimental). As was shown earlier for the data in Table 2, this comparison reveals the total number of mosaics produced out of which the surviving mosaics are recovered. The ratio of the surviving number to the total number is equal to T ; this is also equal to the sum of classes X and Z (whose sizes are expressed as fractions of the total). (The quantity F in Figure 5 is defined as the sum of classes Y and Z; its value for a particular landmark is y+
595
F A T E M A P P I N G IN DROSOPHILA
I
behavior:
MOSAIC CLASSES W+X+Y+Z
= 1.0
landmork
F=Y+Z P = PENETRANCE
I\
BEHAVIORAL MUTANTS Pz1.0:
q
=+ -w - 1-F
B
IN GENERAL :
q'= 7%
LETHAL MUTANTS f = Z/(X+Z) P=1.0 : T=X+Z
FIGURE 5.-Calculating
IN GENERAL :
-
T = 1 P + P(X+Z)
the parameters necessary for focusing.
amcng the control mosaics. The quantity f is defined as the frequency that this landmark is y+ among surviving mosaics of the experimental cross. The calculation of q and # from these quantities is shown in Figure 5B. As before, equations are also given f o r the case of a penetrance less than 1.0. I n that case, which is however, the fraction of mosaics which survive is merely Tobserved, greater than the fraction which, genetically, should survive. It must be corrected as follows to give T for calculations: T = 1.0 4-P (Tobserved - 1.0). Calculating distances: The distances of interest are that between the foci, L, and that between a landmark and a focus, S. The former is the easier to calculate. To do so requires knowing the average frequency that a landmark is yf among control mosaics; this quantity is referred to as Favgin calculating L in Figure 6. The mosaic can be divided (in theory but not directly by observation) into the three classes of Figure 6 according to the genotypes of the foci per blastoderm (the symbol 8 designates the hemizygous dmd state and the symbol 0 designates the heterozygous d m d + state). Under the domineering model (Figure 6A) the value of T , as defined, is also equal to the probability that both foci are female (diplo-X), o r to class NI in Figure 6. The size of NI is related to Favgand L as in Figure 6A. These relations reduce to equation 1 of Figure 6A giving L as a function of T and Fa,,&.Similarly, for the submissive model, the value of T is also equal to the probability that it is not the case that both foci are male, or
596
J. R . F L A N A G A N
G e n o t y p e s of Foci
:
Fraction
A
of.
1 and 2
Mosaics
s
d
6
0
D o m i n e e r i n g Model = F a v g (1 - 1l2L;V LM )
J L
9
Fa,,
0
"
= M t &L
t = d i s t a n c e between foci
FIGURE 6.-Calculating carried the
+ allele and
'I
the distance ( L ) between foci. The genotype designator '' $! " means
8" means the opposite.
FIGURE 7.-The types of mosaics represented by parameter q (panel A) and q' (panel B); the ovals represent dorsal views of blastoderms, the open circles a landmark, and the closed circles the foci of the mutant being fate mapped. The shaded areas contain the wild-type alleles of y and of the behavioral mutant.
F A T E M A P P I N G IN DROSOPHILA
597
to 1.0 - 1.Equation 2 of Figure 6B relates L to T and Favgf o r the submissive model. The values of q and q’ f o r a landmark are related t o the distance between that landmark and the focus on the same side of the blastoderm (S in Figure 4).I n order to see this relationship, it is convenient to interpret q and q’ as follows. The value of q can be interpreted as the probability that a mosaic will survive, given that landmark A (any particular landmark) is female ( y + ) . The events this might represent in terms of the position of the border between male and female nuclei in a mosaic relative to the positions of the landmark and the foci depends on whether one is dealing with a domineering or a submissive mutant focus. Figure 7A deals with the definition of q. If landmark A (open circle) is female (shaded in Figure 7 ) and the focus (closed circles) is domineering, then the border must not pass among three sites, i.e., the landmark and the two foci, if the mosaic is to survive. If the focus is submissive, then any of the events listed under the heading submissive in Figure 7A will allow the mosaic to survive. T h e value of q’, on the other hand, can be interpreted as the probability that a mosaic will not survive given that landmark A is male. The events this might represent are dealt with in Figure 7B. The probabilities of any of the events in Figure 7 can be written as proportional to the product of either S or 1 -S with L or 1-L. For instance, the probability of the event in Figure 7A under the heading domineering is proportional to the probability that the mosaic border does not pass between the landmark and the nearer focus, 1-S, multiplied by the probability that this border does not pass between the foci, I-L. The reasons that it is not equal to this product are two. The first is that the probabilities of these events are not the same when landmark A is male as when female, unless one uses 50:50 mosaics. The second reason is that the product of the probabilities of the two requirements, l-S and l-& is exactly correct only if these two requirements are independent (they may be, but need not always be so). The relationships among q, q’, S and L are shown for the domineering model in Figure SA and for the submissive model in Figure 8B. These relations become equations upon introducing proportionality constants. Moreover, the equations for q and q‘ are reduced to a single equation relating S to q and q’ and an unknown quantity: K (equations 1 and 2 for the domineering and submissive models. respectively). Determining which model is correct is straightforward in that, if a mutant focus is domineering, then q‘ must always be greater than q; if submissive, then the opposite is true. O n the value of K: First note that it is assumed that one value of K will suffice for the calculation of S for all landmarks. It will be shown that this assumption is in fact nearly true, in that a single value for K gives very good results in control-simulations and in the analysis of dmd. The proper value of K to be used can be arrived at in one of two ways, the first being the more intuitively straightforward, but the second being the easier and more accurate method. The first way in which the value of K can be arrived at is through control simulations of lethal foci. From a set of control mosaics, one selects those mosaics
598
J. R. FLANAGAN
A
B
Dominee ing M o d e l
