E. C. MACDOWELL
( F r o m thr Department of Genetics, Carwcgie Institution of Waslii/igfon,
m a s p t ~ t t g€[arbor, N . Y . )
I n the study of maligmancy, biologists and pathologists are being
clrslmn closer and closer together. 7 t is a commonplace that knowledge
of normal processes is largely iiicrcnsccl by comparison with abnormal
pi-occsses. Whether greater contril~ntionsmill eventually be made hy
paihologists to the problem of living matter, or by biologists t o tho
cancer problem is an interesting question.
In scientific analysis tlie establishment of units is an important step.
Biologists recognize three units: the organism, the cell, and the gene.
The organism is tlie major unit ;, the cell is the architectural unit of the
organism; the gene is a self-perpetuating unit of living matter. It is
possible that the greatest contribution of biologists t o the problem of
malignaiicy will prove t o be the concc?pt of the gene, wliich a t present
is the only objectively defined unit of living matter. By the control of
genes, experimental animals of virtiially nnif orm constitution can be
produced and the baffling varia1)ility of results greatly rednced.
Further, dif'fercnces in genes have mcli significant inflnence in irarious
phases of the stncly of malignant growth that the possildity of their
playing a leading r6le has not 1)ccn eliminated.
This, of coiirse, does not in any way prccludc the existence of other
units, or a re-definition of the gene. I miderstand that a t the moment
students of malignancy ai-e considerably interested in the question of
the nature of the unit of maligiiancy, and it seems not a t all impossible
that they will be able to establish another unit of living matter. Rut
until the r61e of one type of unit is fully understood, so that its activity can be rigidly controlled, the establishment of another unit may
be difficult. However, tlie deter*miniLtionof a unit is only a first step in
analysis. Whether the unit of malignancy is in fact a gene, o r an
organism that grows only in h i ~ h l yspecialized protoplasmic environments, or a molecule tliat multiplies by virtue of integral chemical
union with a moleciile of protoplasm, the basic question remains unanswered: How does this unit produce its results? How does it act
upon normal processes, axid how far do normal cell processes control
the activity of this unit? Geneticists are proceeding towards this
further step of analysis in the study of many traits.
The following discussion is based on a study of mouse lenkemia,
begun in 1928 by a cooperative group in the Department of Genetics
of Carnegie Institution of Washington and in the Department of
1 Read before the America11 Association for Cancer Ilcsearc~li, New York, April 17, 1935.
For discussion scc page 1%.
85
Ptttliology of tlic College of Physicians ancl Surgeons (1). A s the ~ v o r k
is in l)i*ogi*clssthc emphasis r.iglit1.y f;ills iipon problems ratlic'r Ilinii
gcii(wi1 coii('1iisioiis. Tlic i*i,le of geiies, 01' intrinsic factoi's, xvill I)c
coiisitlcrorl, wit ti spccinl w f c ~ i w i ( ~t co their intc.i*rcl;itioris with 11011gcnelic, o r extrillsic+,f;ictors : ( u ) iil inoculated hosts, ( b ) in trnnxmittctl
lcukcmic cclls, and finally ( r ) in ciises of spontaiieons lculwmin. Siicccss in diffcrcntinting two caategorics of variables, such :is intrinsic 0 1 '
c~striiisicfactors, tlepends upon tlie dcgrce of control that ci\ll 1)c cstn1)lislied. Chemists clcmand thc highest purity of their rcngents. Hiit
workcrs with sinimals sometimcs accept hctcrogeneoiis mnteri:il, 11ot
realiziiig t l ~ i ~significaiit
t
clarification of their 1-esnlts might follow the
use of gciietically purc matcrials. I n the midst of the bowilderiiig
NO CELLS
INOCULATED
~oo~ooo~ooor
I\
10,000,000
1,000,000
005E
CELLS
-
v5
AVERAGL
INlERVAL
BEFORE DEATH
LINE I
1105T5
-
STRAIN C5B
I \
-
Ip0,oOo -
difficulties of biological research, onc great aid stands ont : that is, that
genctic constitution can be highly controlled e r c n when tlic constituent
genes are not identified. This is accomplished by intensive i n h e d i n g
of indiviclnally pedigi-ecd animals. According to the calculations of
Wright (a),no mattci. how diverse the genes of two parents, if matings of their descentlarits a r e always made between brother and sister,
I)y tlic time the 11th generation has been reached all the offspring from
one pair will liavc virtnally tlic same set of genes (over 95 per cent
homozygosis), whatever their properties may be. This method of
geiic?ticpiirifictltion parallels thc 1)actcriological method of purification
hy succcssive platings of minute samples. I n the case of inbreeding,
the 1)roccss is slower, since thc selection of the two parents to cnt.1-y
tlic strain to tlic next gcncration is made blindly.
1'relirninni.y to the active s t i d y of leukemia, several strains of ill-
87
GENETIC ASPECTS O F MOUSE LEUKEMIA
dividually pcdigrced mice were inbred by brother-and-sister mating
in thc Department of Genetics. Hy the time the cooperative project
was started this process had pilcd up about 1-1 gencrations of common
ancestors, 2 per generation, of all the mice within each strain. This
preliminary work of purification occupied six years. These purified
strains are the biological analogues of purified chemical compounds,
and with thcse the work on leukemia brought highly consistent results.
This analogy with chemicals is grossly misleading so far as it implies
that constaiicy of genetic constitution means uniformity of individuals.
F o r each iiidividual is a chemical cxperimeiit. The reactions of given
chemical substances depend upon the conditions of the experiment-so
the somatic characteristics of tlie individual with a given genetic conm
N
-
wpRox'MAT~
w.w
STANDARD
80,000,000r
CELLS
1 : 4,095
19.530
4,080-
1 . 65,536
1.220-
305
-
1 : 1,048,576
76
-
1:4.194,304
19-
I
-
SURVIVAL
HOST5
STRAIN C 5 8
48
8
4
1
I
*Ox
E'IG. 2.
Vs.
LINE I
-
I : 16,384
1 : 262.144
- DOSE
I
40h
PERCFNT
I
60%
I
80%
1
100%
SU R V I V A L
INVERSE
RELATIONBETWEEN DOSAGEOF LEUKEMIC
CELLS (LOGARITHMIC
PERCKKTAGE
OF SURVIVAL
(ARITIIMETIC SCALE)
Dots are averages, with nurribcr of inice iiidicnteil for each dot.
SCALE)
AND
Straight linc drawn by eye.
stitution dcpcnd upon extrinsic conditions. The problem before us is
the evaluation of the relative cf'fectiveness of intrinsic and extrinsic
factors in bringing about ccrtaiii somatic conditions related to leukemia.
GENETICASPECTSOF INOCULATED
HOSTS
When the first transfer of tissue from a spontaneous case of leukemia was made to 8 young mice of tlie same strain, each one developed
lcukemia. This has been repeated with 20 spontaneous cases, with
success in every case. The series of successive transfers from each
spontancons case is called a Zi~e. Numcrous lincs have been dropped
a t different times, but a group of three has been carried through 260
to 500 transfers during the past six years. Throughout, the rule has
becn 100 per cent success in each transfer, with only r a r e exceptions.
This applies only Lo host mice from the strain in which the line origiiiated arid to standard teclinic of transfer. By varying cither technic
or tlic strain of Iiost, divcrsc Imt r)retlictilWc results are obtaiiicd. With
all other coiiditioiis the same, differences in tlie amount of matcritrl
transplanted will g!ve any dcsired rcsult from complotc succcss to
complete failure (Fi,lqs. 1-2). Again, hy the use of hosts of different
geneticully pnre st raiiis, wit11 all otller coiitlitioiis coilsttint (tlial is,
the same lcwkcmic cc~lls,line I, Fig. 3, illlil the massive standard close),
a similar gradcd series of results was found.
The accompnying charts indicate clearly enough that genetic constitution alonc does not ensure uniformity of results, any inore than
heredity alone, or envii~oiimcntalone, can produce an organism. Thcre
STRAIN
OF
HOST
STRAIN OF HOST vs
StOll --
CELLS * LINES 1 . C AND L
SURVIVAL
DOSE
I
STANDARD
1iiic.s coniicctillg pniirts xervc iiierrly t o iilcutify ~ ~ ~ i iand
i t s cmpliasize differential d i l tions of hnst str:tiiir :tiid tlic diff crciit lilies of 1eukc.mic cells. Host strains arr:rngcd 011 chart
iii order of incrcasirig hurriv:il with liuc I cclls.
is not a trait in any organism that does riot result from the cooperation
of genetic :ml non-genetic factors. In some cases the influence of nongenetic factors is not apparent, iuld tlie trait is called genetic; in otlicr
cases the i r i f l u c n c c ~ of iioii-~eiieticfactors is so obvious that t2ic contributiun of genetic factors is overlooked and thc trait is called nongenetic. But tlicse are only tlic cxtrcmcs of a continuous series in
which all iiitcrgraclcs of relative potcricy of genetic and lion-gcnctic
factors a r c includctl. Iieredity provides a reaction system, but t Iic
operation of this sy m is subject to the influciice of extrinsic variitbles.
Thus, to ciisiire completc uniformity of results, uniformity of riongenetic variables is also necessary.
In two of tlic strains inocnlatcd with the standard docs of line 1
cells (Fig. 3) some of thc mice were susccptihle and some survived
inoculation, and yct cacll of these strains w a s genetically uniform.
From this i t m;iy he coiiclutlccl that uncontrollcd non-gcnctic variables
were turning the balancc Ixtween susceptibility and resis tancc. This
~ ~ 1 1by blwding survivors and finding that Ilici r
conclusioii has 1 ~ ~ tested
offspring, when inoculated, gax7e the same proportion of susccptihlc
animals as in tlic previous generation. In these two strains the genetic
80
GENETIC ASPECTS O F MOUSE LEIJKEMIA
constitution was sufficientlypotent to produce susceptibility only under
certain of the extrinsic conditions encountered. The relative potency
of genetic and non-genetic factors in these strains is roughly indicated
by tlie respective percentages, 26 per cent and 77 per cent. I n other
words the weaker the genetic potency the stronger becomes the cffectivexless of non-genetic variables.
Tlie chart (Fig. 3 ) illustrates one of the most confusing situations
in the study of heredity; a trait (such as susceptibility to leukemic cells
of liiic I) may be determined by different p n e t i c constitutions, which
give different effectiveness t o the same variations in extrinsic factors.
INHICHITANCE O F SUSCEPTIBILITY T O INOCULATION
w m r LINE I
A s CONSTITUTPD
JAS. 'X-MAR. '31
r""l
25 MATINGS 31-60%(b14
25 MATINGS 0 -5 X Oad
FIG.4. EVIDELCE
TIIAT THE SUSCEPTIRLE
AND NON-STSCFSTIRLE
STRAINS
OF HOSTSDIFFER
BY O N E I'AIR
OF
MENDELI.4N GENES, WITII DONINANCE
0%
SUS('EPTIBIL1TY
Close agrcenieiit with expected 3 : 1 ratio in F2 and 1: 1ratio in backcross (B.C.) to ncgntive parent strain; genetic sigiiific:iiiee of backcross (I: 1 ) ratio verified by inoculating 1325
offspring of 50 of these backcross iiiicc, 2 13.C.
Instances of this kind are becoming more and more numerous. They
usually involve a physiological threshold, as in the case of different
strains of guinea-pigs with extra toes and different strains of blind
mice.
I n two of the strains in Fig. 3 ((2% and StoLi), the genetic constitution was sufficiently potent t o give the same result in all mice within
each strain, following inoculation with the standard does of line I cells ;
in one strilin this result was susceptildily, in the other resistance.
These sharply contrasted strains provided opportunity to test, by
hybridization, the nature of the genetic difference, so f a r as it involved
susceptibility to this particular line of leukemia under the given range
of extrinsic conditions. This analysis indicated that the balancc between susceptibility and resistance was turned primarily by one dominant gene for snsccptibility (Fig. 4 ) .
Ilaving found that genes a r c working, the question arises : How (lo
they bring about their effects 2' Do they providc some necessary cultural material for the growth of the particular cells, or do they dctermine a rate of response on the part of the host? What is the nrtliurc
of natnral resistance 'Y Although various approaches to tlicsc qncstions a r e being made, only onc will be mentioned. The study of veiy
dilute dosos of lcukcmic cells leads to the conchision tlint the host is
TABLN
I: Resistance in ("6X I l 0 s . h Intluced l
q and against Cplls oj Line I
_ _ _ _ _-
Survive staiid:ud dose
Later doses
Standard (50 montjlis)
Repeated dilute doses increasing to
standard
Preliminary experiments
3 4 dilute doses
5-6 dilute doses
5-S dilute doses, 1st dose subcutnneous
3195
Number
Per cent
0 per cetlt,
0
7
11
0
52
60
45
48.2 per retit
76.8 per cent
93.7 per cent
0
24
100.0 per cent
not passive even in strains in which the deaths, followirig the stnntlarcl
dose, a r e 100 per cent; for the survivors of small doses have acquired ii
wsisf ance tliwt e ~ i be
i built up, by increasingly lieavici. (loses, to tlic
point of rcsistiiig the massive standard close (Table I). With the massive standard dose given first, the evidence of ally resistance is swcq~t
ttsitle by lhe great wave of lcukc!mic cells. Resistaiice appears to dcpent1 upon the specific interrelNtioiisliip between tlic host and the leukemic cells. Hosts of the strain of origin are not sufficiently different
from the IeuBemic cells to develop effectivc resistance to the standard
dosc ; foreign hosts usually a r c sufficiently stimulated to develop resistaiicc in tlie first transfers of the line, but in later transfers, when the
virulence of cells is greater, the resistance of hosts in certain foreign
strains may not be tlevcloped in time to save the life of the hosts. If
thcise is no genetic difference between tlie immunizing cells and the liost,
we fail to obtain resistance. This appears in preliminary experiments
in which resistance has been induced by embryo skin (Table 11). The
difference bctweeri strains of embryo skin is not due to a lack of some
specific resisting substance in C58, since C58 skin is effective in n different host strain. Both the natural power of hosts to resist leukemic
inoculation and the power of embryo skin to induce resistance in susc.cpfihlP hosts clcpciid ul)oii H cliflorentid rcllation bctwccn thv hosts
and the material inoculated.
The clear cut diffcrcncv.? between the effectiveness of embryo skin
from these strains provides opportunity to make the usual genetic tests
91
GENETIC ASPECTS O F MOUSE LEUKEMIA
of the nature of this difference. When these two strains are crossed,
embryo skin of the first generation hybrids induces resistance as successfully as embi-yo skin from pure StoLi. It is of interest to recall
that these same hybrids, used as hosts, were 100 per cent susceptible
to the same line of leukemic cells.
The foregoing discussion has been limited to one kind of leukemic
cell, namely line I. When leukemic cells from different lines were
used (lines E and L, Fig. 3 ) , the results were different, for a genetic
constitution that makes a mouse susceptible to one line of leukemic
cells may make it resistant t o another line.
Strain of
Embryo skin
StoLi
C58
C58
C58 X StoLi
No. of exps.
Survived
5
35
0
36
36
Died
Hosts
C58
C58
89
C58
7
5
4
0
41
0
0
GENETICASPECTSOF LINESOF LEUKEMIC
CELLS
So far, attention has been focussed upon hosts which show constancy of characteristics within a strain and diversity between strains.
I n turning to lines of leukemic cells, the same situation appears : within
a line, constancy over considerable periods ; between lines, diversity.
Genetic control involving genes accounts for the results in the hosts.
What accounts for the parallel results in the leukemic cells?
While there is no doubt of the genetic relation of host mice in pedigreed strains, the relationship of leukemic cells in successive transfers
can be determined only hy special study. Such a study was made by
Potter and Richter ( 3 ) using all significant organs of 68 mice taken at
regular intervals after intraperitoneal inoculation. Each organ was
spread out on microscope slides in serial sections. It was possible t o
trace the leukemic cells step by step from the mass of original inoculum
to the final stage of infiltration. The inoculated cells migrated from
their original position, continually dividing as they spread further and
further through the body, following fascia1 planes or lymph channels
or penetrating directly through membranes into organs and nodes.
I n this primary invasion the blood vessels were not used. The spleen
was infiltrated before any significant appearance of the leukemic cells
in the blood. In this whole process the lymphoid tissues of the host
showed no increased activity; the enlargement of the organs was due
to the invading cells, identified by the characteristics of the line inoculated.
Thus the pedigree of the leukemic cells used for transfer has been
92
E. C. MACDOWELL
traced directly to the cells prcviously inoculated, and the conclusion
may be drawn that this pedigree continues back to thc spontaneous casc
of leukemia that gavc rise to the line. Extended cfYfFor'ts l o scparato till
active agent from the living cells were unsuccessful, although a wide
variety of methods were employed, including anaerobic filtrccit'ion.
The continuity of the lineage of leukemic cells in transfers gives a
basis for the maintenance of leukemic characteristics and for the general diversity between lines that has appeared. Besides cytological
differences and differences in virulence, in host requirements (illus-
11
30
3.
I
3
1
I
295
300
I
I
40
45
1
I
305 . 310
I
50
I
55
I
60
I
1
I
315
320 325
TRANSFER
I
65
I
70
1
75
I
80
1
I
I
I
330
335
340
345
'
BEFORE DEATHIN SUCCESSIVE
TRANSFERS
FROM FOUR
DIFFERENT
LINESOF
FIQ.5. INTERVALS
INOCULATED
LEUKEMIC
CELLS
Each dot represents interval between iiioculatioii and death of one mouse; animals killed
as donors excluded. Vertical scale (days) indicated by horizontal lines at five, or five and tell
days. Middle two cell lines average about the same, but show clear difference in variability.
Hosts all from strain C58, picked at random, four to eight weeks old.
trated in Fig. 3), and in rates of cell divisiom, dissemination, and metabolism, lines of cells show characteristic n f f i d i e s for different organs.
Spleens may be large or small, abdominal muscles clear or densely infiltrated, mesenteries heavily encrusted or normal; all nodes o r only
one may be enlarged ; in one line liver lesions may be characteristically
large, in another line, the kidney lesions ; most lines develop high blood
counts, but in one the counts reach over 1,000,000 and in another are not
above normal. Usually leukemic cells divide wherever they go, but
the cells of one line, although present in large numbers, do not divide
in the blood. As an illustration of the subtilty of demonstrable differ-
93
GENETIC ASPECTS O F MOUSE LEUKEMIA
ences between lines, the intervals before death in two lines may average
the same, but in one line show slightly greater variability, or again,
one may average four days, another five, and another seven (see Fig.
5, in which each dot represents one mouse). These differences persist
for many transfers, but within a line the interval before death is
subject to persisting change at any time and usually in the direction of
increased virulence.
The number of underlying differences upon which these observational differences depend cannot be stated, but certain of the line
HOST
-
U R G E CCLI.8
C 58
Y
80
-
70
-
60
25-
-
10
20-
-40
PROPORTION O F LARGE CELLS
EACH POINT OASCD ON 800 MLMURLMCNTI
DAYS OETORL
DEATH
15
10
0
8 8
-
0
-
-
INTERVALS BEFORE DEATH
00
EACH DOT
0
0
-
.
.
ONE W l P L
30
m
a
o
5-
1
1
1
1
1
1
1
I1
12
13
(4
15
16
17
1
1
10 19 20 21
TRANSFER
1
1
1
22
1
1
1
1
1
1
23 24 25 26 27 28
FIG.6. ARKUPTCHANGEIN INTERVAL
BEFORE DEATH,
ASSOCIATED
WITR A PRECEDING
GRADUAL
RISE IN THE PROPORTION OF ONE OF FOURSIZE-CLASSES OF 'PHE TRANSPLANTED
LEUKEMIC
CELLS (LINE, M LIV.)
characteristics are unquestionably correlated. This appears clearly
in the relation of virulence and cell size. The more virulent the line,
the larger the characteristic cell. Within a line, changes in virulence
are accompanied by changes in cell size. In the case of an especially
abrupt change in virulence, the change in cell size was found to have
preceded the clinical manifestation and to have developed gradually,
through a series of several transfers, until the concentration of the
larger type cell had reached a threshold of clinical effectiveness (Fig.
6). Thus the dependence of the change in virulence upon the change
in cell size was established and a phenomenon of mutational abruptness
was found to depend upon a gradual approach to an effective threshold.
The presence of specific genes can be determined only by cross-
Generation
*From the pedigree
P
e-
book of C.C.Little.
*
91284
1
f
&280*
92119 x d2362*
2
3
92541 x a540
4
I
I
94582 r
d4580
5
96959 x do956
6
7
-%
A
9ll723 x d11722
8
914601 X dl4597
9
917833 x dl7929
I
39649 x do645
1
I
d21041
10
021044
X
z
11
12
928208 x a8212 x 929457
13
933404 X 653171
I
I
A
14
934445 I $34444
15
938210 x ~38208x 938209
16
17
943836 X d43833 X 946192
I
I
I
B
A
t'IG.
7.
C
LEUKEMIA
WAR
FORMALLY
STARTED
Letters A, B and C rcfor t o pcdigreos coiitinued in figures 8 a, b, and o.
PEDIGREE OF STRAIN
C58,
SHOWING INBREEDING BEFORE STUDY OF
94
GENETIC ASPECTS O F MOUSE LEUKEMIA
95
breeding. Since lines of leukemic cells cannot be cross-bred, the direct
determination of the r61e of genes in leukemic cells is impossible. Since
other contestants f o r the important honor are unknown, one might say
that genes win by default. But such country-club rules are dangerous
in the laboratory.
Whethcr all characteristics of lines of leukemic cells are transmitted
from cell generation to cell generation by the same o r different types of
genetic units, whether these units are genes, or semi-living substances,
o r molecules of living protoplasm, or in viruses o r mutagens, and
whether these units are related t o the primary organization of the cell
intimately or superficially, the properties of leukemic lymphocytes are
genetic-under the same conditions, persistent differences from normal
and between lines. And with this, the genetic problem of leukemic cells
becomes parallel with numerous other genetic problems in asking: Are
genes the only units for the transmission of inherited traits!
GENET1 C
ASPECTSOF SPONTANE~US
TIEUKEMIA
The final and major genetic problem conceriis the spontaneous origin
of leukemia (4). This presents the greatest difficulties of all; in it,
many of the difficulties of the preceding studies are combined. If more
questions have been raised than answered by the analysis of various
processes experimentally separated, how much more confusing the
attempt t o analyze their simultaneous activity in spontaneous cases !
For the genetic problem, the problem of origins, is not answered by
the determination of a symbolic formula of units. It also asks how
these units bring about their results.
It is highly regrettable that, outside the immediate circle of geneticists, there seems to be an impression that the gene is self-sufficient
and is either dominant o r recessive. Especially as applied to neoplasia, this misunderstanding has led t o erroneous conclusions both on
the part of hostile critics and ardent believers. Dominance is only a
special case at the end of a continuous series of interrelations between
pairs of genes. No gene can produce its effect without cooperation of
many other genes ; its apparent primary effect, as well as its so-called
dominance, may be altered by changing the associated genes. And, to
repeat, genes and extrinsic conditions cooperate in all cases ; when
genes are relatively powerful they are largely independent of variations in the conditions; when they are relatively weak, variations in
extrinsic conditions become effective. I n the incidence of spontaneous
leukemia the same variable relationship between the potency of genetic
and non-genetic influences is found as in the case of the susceptibility
of different strains of host mice t o inoculation with the same line of
leukemic cells, and, as in that case, the possibility of correct interpretation depends upon genetic purification of the strain before the special
study was started. Slye’s (5) pedigree of spontaneous mouse leukemia
represents the process of genetic purification, but it stops at the point
at which a critical study should begin. As her data stand, there is no
1461601
$’I(?.
8 ~ .PEDIGREE OF A BANDOM SAMYLE OF MICE OF STRAIN C58 LIVINGMORET H A N 81s
LABORATORY
CONDITIONS
MONTHSAND HELD UNTIL DEATHUNDER NORMAL
Long vertical brackets connect offspring from oiie father meted t o his sisters; mothers’
numbers iii brackets to left of her offspring; successive offspriiig pcdigreo iiumbers repeat
oiily last o m (or t w o ) digits. Heavy bars, microscopic diagnosis, by Dr. M. N. Richter, of
some leukemic disease ; circles, iiegstivo for loukemia; short dash, diagnosis uuccrtaiii, mostly
duo to post-mortem changes.
96
GENETIC ASPECTS O F MOUSE LEUKEMIA
97
possibility of defining the relative potency of genetic and non-genetic
variables, so that any interpretation is hypothetical. If, as in the
present study, non-genetic variables were influential, the seeming agreement of certain of her results with mendelian ratios would deny the
hypothesis which she claims they verify.
In the present study the origin of the strain of mice which proved
to yield spontaneous leukemia is shown by the pedigree of brother-bysister matings (Fig. 7). All the mice in this strain that have been
studied are descended from one pair in the 14th successive inbred generation. Beginning in 1928, a random sample of over 600 mice was
held until natural death. Before this, the incidence of leukemia in this
strain was unknown. I n the pedigree of these animals (Fig. 8 a, b, c)
the diagnoses are grouped in three main classes : a heavy Oar indicates
some form of a considerable range of leukemic diseases, usually involving lymphoid cells ; zero is negative for leukemia ; a dash indicates
uncertain diagnosis, due in most cases to post-mortem changes. Of
the definite diagnoses, 90 per cent were of “leukemia.”
As previously indicated, the long inbreeding is evidence of genetic
uniformity. This is confirmed by the facts, first, that the negative
cases are distributed at random among different branches of the pedigree, and second, that negative and doubtful animals as parents gave
the same proportion of leukemic offspring as leukemic parents. A
random sample of 70 mice from later generations of this strain gave
the aame proportion of spontaneous cases of leukemia as the first random sample. Doubtful cases were virtually eliminated by killing when
the mice were sick instead of waiting until actual death. I n contrast
to this, an equally inbred strain was found in which leukemia occurred
in less than 2 per cent of a sample of some 300 mice. This strain is
spoken of as negative. The above observations indicate that the genetic
constitution in one strain leads to leukemia under the great majority
of conditions encountered, while the genetic constitution of the other
strain required extremely rare conditions for leukemia t o develop.
These strains were cross-bred. The first generation hybrids, with
fathers from the leukemic strain, indicate that the genetic difference is
sufficient to reduce the incidence of leukemia by about one half of that
in the leukemic strain. I n these hybrids again the difference between
the leukemic and non-leukemic animals was not genetic. I n the first
generation of a cross between any two pure strains, all the individuals
are hybrid, but equally hybrid; they all have the same genetic constitution. I n this case the reduction of the total heredity from the leukemic
strain by one half has reduced its relative potency and correspondingly
increased the potency of non-genetic variables in deciding the issue for
each individual. The conditions encountered by slightly more than
half of the mice do not permit the development of leukemia. The
genetic uniformity in this first hybrid generation was checked by breeding both positives and negatives with the pure negative strain, and
finding that whatever the somatic condition of the hybrid parent, the
proportion of leukemic offspring remained the same.
46611
ao
41118
49717
PO
,
1
FIG.8b
See Fig. 8a for explanation.
98
GENETIC ASPECTS O F MOUSE LEUKEMIA
99
These offspring, constituting the first back-cross, are genetically
diverse, for this generation is the result of segregation and recombination of all the genes by which the pure strains differ. If non-genetic
factors are relatively insignificant, the somatic conditions exhibited
by animals in a back-cross usually iiidicate the genetic ratio and thus
tthe number of gene differences involved. In the present case the r61e
of non-genetic factors is significant and becomes more so as the leukemic heredity is divided up. With the hybrids as fathers, the incidence in this first back-cross is about half that in the first generation of
hybrids ;, another reduction by one half of the total heredity from the
FIG.8c
Sce Fig. 8a for explanation.
leukemic strain is accompanied by a corresponding reduction in the
incidence of leukemia. Thus with transmission through males, the
incidence of leukemia in the first hybrid generation and the first backcross is closely correlated with the proportion of total heredity from
the leukemic strain. This result could be interpreted by numerous
hypotheses, involving aay n u m b e r of gerzes desired, but no conclusion
can be drawn. The somatic conditions of the animals in this generation give no proof of the underlying genetic ratio. To obtain this, it
will be necessary to classify the animals in this first back-cross according to the varying proportions of leukemic offspring that each one may
produce in the second back-cross to the pure negative strain, as was
done in a previous instance.
There remains one genetic aspect of spontaneous leukemia whose
general significance may eventually supersede all other phases of this
study. This is a difference in results in hybrid generations according
as the leukemic heredity is transmitted by mothers or fathers (6). The
incidence of spontaneous leukemia in the first hybrid generation was
19 per cent higher when the mother came from the leukemic strain; and
100
E. C. MACDOWELL
in the back-cross, 36 per cent higher when the mother was the hybrid
parent. These diff'orences are statistically significant, being 4.5 and 7.0
times their respective probable errors (Table 111). This repeats for
leukemia a result reported by Little for breast tumors. Daughters
from reciprocal matings show the same difference in incidence of lenkemia as the sons. This eliminates sex linkage, which gives different
results in sons and daughters. Nor is this an accidental result due
t o making the cross with a single mating that might include an aberrmt
LONGEVITY AND LEUKEMIA
IN RECIPROCAL CROSSES
80
-
70
-
5 60-
z
L!
2 50I.
40w
0
2- 3020 -
'
' 0O L
1
400
I
1
100
600
700
LENGTH OF LlFL IN DAVE
OF LEUKEMIA
AND LONGEVITY
IN Two PARENT
STRAINS,
P, ( C = '258,
FIG. 9. INCIDWCE
8 = StoLi.), IN RECIPROCAL
CROSSES OF FIRSTHYBRIDGENERATION, F,, AND
IN BACKCROSS
TO STRAIN
S, B.C.
Lines connecting points serve oiily to identify points in each generation. Leukemia and
shorter life enter the cross from the same strain, but there is not a causal connection, since
three points a t very nearly the same age show great differences in incidence of leukemia, and
two points of nearly the same incidence of leukemia show wide differences in length of lifu
The differences in length of life are shown by the leukemic mice alone, as well a8 by the noiv
leukemic mice.
mouse, f o r 11 males and 17 females from one strain were crossed with
6 males and 21 females from the other. Separating the results by
fathers gives small numbers and overlapping variability, but the difference in distribution of the averages in reciprocal matings remains
evident.
Finally there is a similar significant difference between reciprocaI
matings in length of life (7). ThiR also is not due to sex linkage. Within
a generation, longer life is associated witth less leukemia, but this association is not causal (Fig. 9) because the age difference in reciprocal
matings is shown as clearly by leukemic mice as by negative and doubt-
la1
GENETIC ASPECTS O F MOUSE LEUKEMIA
ful ones, and because the hybrids with 62 per cent leukemics lived as
long as the pure negative strain, while the average length of life of
the hybrids from the reciprocal mating with 43 per cent leukemics was
140 days longer than in the negative strain. Furthermore, in backcross and first generation hybrids with leukemic heredity transmitted
through the same sex the length of life was approximately the same,
whereas the incidence of leukemia was markedly different.
TABLE111: Spontaneous Leukemia in Two Hybrid Generations
(In brackets, number of positive plus negative cases; doubtful cases omitted)
C58
1
C58 X StoLi
I
(C58 X StoLi) X-StoLi
89.6% (606)
Through father
Through mother
42.5% (106)
61.9% (139)
Difference
Diff./P.E.
19.4 f4.3
4.5
c*
19.8% (96)
46.5% (159)
26.7 f3.8
7.0
Thus observations on two traits point to a mechanism of transmission through the mother not shared by the father. Since the chromosomal mechanism is shared by fathers and mothers, the implication is
clear that besides the familiar chromosome transmission, some ponc@omosomal mechanism also is involved and that this is probably
cytoplasmic. Again the question arises, are genes the only genetic
units sb
I have traced some of the analytical problems offered to a geneticist by mouse leukemia (8). To a physician these may seem academic.
The geneticist controls intrinsic factors, but the physician is limited
to extrinsic factors. My impression is that, for both geneticist and
physician, an understanding of the interrelations between extrinsic and
intrinsic factors is of critical importance. However complicated this
interrelation appears to be in leukemia, there can certainly be no doubt
but that extrinsic factors are operating at various points. With their
activity delimited by controlled intrinsic factors, the attempt to identify
these extrinsic factors offers a direct approach to their practical
control.
REFBIRIQNCES
1. RICHTER,
M. N., AND MACDOWDLL,
E. C . : Physiol. Rev. 15: 509, 1935.
WRIGHT,S. : Genetics 6 : 111, 1921.
POTTER,
J. S., AND RICHTER,
M. N. : Arch. Path. 15 : 198, 1933.
MACDOWELL,
E. C,, AND RICHTER,
M. N. : Arch. Path. 20 : 709, 1935.
SLYE,MAUD: Am. J. Cancer 15 : 1361, 1931.
6. MACDOWELL,
E. C.: Am. Naturalist 69: 68, 1935.
7. MACDOWELL,
E. C. : Science 81 : 416, 1935.
E. C. : Scientific Monthly, February 1936.
8. MACDOWELL,
2.
3.
4.
5.