GUNTHER SCHLAGER
     Department of Systematics and Ecology, University of Kansas, Lawrence, Kansas 66045
                                Manuscript received September 21, 1973


              Response to two-way selection for systolic blood pressure was immediate
          and continuous for about eight generations. In the twelfth generation, the
          High males differed from the Low males by 38 mmHG; the females differed
          by 39 mmHg. There was little overlap between the two lines and they were
          statistically significant from each other and from the Random control line.
          There appeared to be no more additive genetic variance in the eleventh and
          twelfth generahons. Causes for the cessation of response are explored. This is
          probably due to a combination of natural selection acting to reduce litter sizes
          in the Low line, a higher incidence of sudden deaths in the High line, and loss
          of favorable alleles as both selection lines went through a population bottleneck
          in the ninth generation.-In the eleventh generation, the selected lines were
          used to produce F,, F,, and backcross generations. A genetic analysis yielded
          significant additive and dominance components in the inheritance of systolic
          blood pressure.

          need for animal models of human diseases has prompted a good deal of
TFe:earch      designed to uncover these models. For defects caused by single
genes this has entailed searching large animal colonies for specifically detectable
anomolies and then, by the appropriate genetic crosses, to produce a pure-bred
colony of afflicted animals. For diseases dependent on the expression of many
genes, selective breeding of extreme segments of the animal population should,
in a number of generations, lead to a colony of animals each with a higher (or
lower) value for the trait than the majority of individuals of the original base
population. The end result of such a breeding program would depend largely on
the number of loci at which alleles are segregating and at which the alleles con-
tributing to the desired direction of the trait are available for fixation. The more
“plus alleles” available at these loci in the base generation, the further the selec-
tion response should go. The methods of selection will be highly dependent on the
objectives of the program. If the animal model is the sole objective, intense selec-
tion pressure with moderate inbreeding would be efficacious. It would be ques-
tionable whether or not to carry a simultaneous control line as more space and
effort could be devoted to maintaining a larger population to intensify the selec-
tion pressure. If genetic information (heritability, number of loci, mode of in-
heritance) is the objective, two-way selection, a concurrent random-bred control,
avoidance of inbreeding and patience may be the answer.
   Previously reported selection programs in rodents were concerned with the
production of an animal model. I n the selection program to be described, artificial
Genetlcs 76: 537-549 March, 1974.
538                                      G.   SCHLAGER

selection for high and low systolic blood pressure levels was carried out in a syn-
thesized mouse population with the primary objective of determining the genetic
characteristics of the trait. Response to selection pressure for this trait has varied
in a number of different organisms but has been successful in elevating blood
pressure levels in chickens, turkeys, rabbits, and rats. There is one reported case
where selection for elevated pressure was not successful ( STURTEVANT           1953).
Previous attempts to select for low systolic blood pressure are limited to chickens
and turkeys (see review by SCHLAGER        1972). Studies of the nature of the in-
heritance of systolic blood pressure in experimental animals involve the rat
 (PHELAN   1968; KNUDSEN aZ. 1970; TANASEal. 1970; LOUIS al. 1969;
                              et                      et                  et
OKAMOTO al. 1966) and mouse (SCHLAGER
           et                                       1968, 1971; SCHLAGER WEI-and
                       and              1968). It is generally agreed that more than
a single pair of alleles is involved in determining blood pressure levels although
as few as two were found to be sufficient to explain the susceptibility to the hyper-
tensive effects of chronic salt ingestion (KNUDSEN al. 1970). I n genetic crosses
between strains or selected lines, or between selected lines and controls, the Fl
blood pressures tend to be intermediate between the parental strains but some
degree of dominance has clso been demonstrated.

                                 MATERIALS A N D METHODS

    The base population for the BPI series was derived from an eight-way cross of the inbred
lines (LP/J, SJL/J, BALB/cJ, C57BL/6J, 129/J, CBA/J, RF/J, and BDP/J) followed by three
generations of random mating. A total of 165 mice were measured in generation zero from which
the three lines, “High,” “Low,” and “Random”, were begun. The lines were closed after the
first generation. In the earlier generations, 20 matings were established for each line, avoiding
double first-cousin and sib-matings. One of the 20 pairs of mice was mated in each of the three
lines on each of 20 working days to ensure that the offspring of the three lines would be exposed
to comparable environmental differences during post-partum holding and measurement periods.
The mice were weaned i t holding cages, sexes separated, with no more than three mice t o a
standard 11” x 5“ x 6” stainless steel cage.
    A combination of family- and individual-merit selection was practiced. The blood pressures
were ranked from lowest to highest in the two selection lines and within each sex the mice with
the highest blood pressure were selected to propagate the high line and those with the lowest, the
low line. However, no more than four individuals from the same family were chosen. Mice
for the continuation of the random line were selected without any knowledge of their blood
pressures but not with any strict randomization procedure.
    After the sixth generation, a number of ccmpmmises were introduced into the design due to
budgetary considerations. The number of matings were reduced and the restriction on mating
sibs and double first-cousins was not strictly enforced. Between the sixth and seventh generations,
the mice were moved from Bar Harbor, Maine to Lawrence, Kansas, where they were then
reared in plastic disposable cages (10.5” x 8” x 4.5”). A change in food from Old Guilford to
Purina Laboratory Chow was also made at this time. Better control of temperature was intro-
duced during generation 9, where the fluctuations were limited to about a IO” range between 70”
F and 80” F most of the year. A constant light-cycle of 12 hours light and 12 hours dark was
maintained throughout the entire experiment.
    Systolic blood pressures were measured with a Physiograph IV (Narco-Biosystems, Inc.) by
occluding the flow of blood in the tail and detecting the return of pulse distal to the cuff upon
deflation. The mice were unanesthetized but restrained i n a small holder mounted on a thermo-
statically controlled warming plate. The temperature of the plate was maintained at 37.5” f 1.OD.
The mouse was introduced into the holder, allowed to become accustomed to the restraint for a
few minutes and then blood pressure determinations were made at half-minute intervals. The
                             SELECTION FOR BLOOD PRESSURE                                   539
validity of the method was determined by simultaneoas direct (carotid artery) and indirect
measurements under anesthesia ( SCHLAGER WEIBUST
                                             and           1967). Five measurements were taken
on each of three days. During the earlier generations, these measurements were taken a t 100,
125, and 150 days, plus or minus five days. After generation 6, the three days generally fell
between 100 and 150 days of age, although a few mice were measured as early as 85 days. A
mean blood pressure was calculated for each day and the three means were then averaged to give
an overall measure of systolic blood pressure for a mouse. This overall measure was used in all
calculations and as the basis for selection.
    F , and Backcrosses: Four pairs of matings were established from the mice used to propagate
selection generation eleven. Two were High females mated to Low males and the other two were
the reciprocal crosses. Forty-four F, progeny were produced and these in turn were used to
propagate seven F, families and four families in each of the two backcrosses. In all of the back-
crosses, the females were F,'s since these would be younger, more vigorous females than those
of the reciprocal lines, which were well over 250 days old by this time. Offspring from these
crosses were all weaned and measured as described above f w the selection lines.

                                 RESULTS AND DISCUSSION

   Response to selection: The average systolic blood pressure for the High, Low,
and Random lines for 12 generations of selection are shown in Figures 1 and 2.
The separation between the selected lines occurred immediately and the differ-
ence increased with succeeding generations until the 8th to 12th generation.
Fluctuations in response are evident in both sexes and these are for the most
part parallel in the three lines. These are undcubtedly related to environmental
factors, although attempts to quantify these have been unsuccessful. The fluctua-
tions are more marked in the later generations corresponding to the less stringent
environmental and procedural controls imposed in the animal facilities at the
University of Kansas. The fluctuations both in the earlier and later generations
are more pronounced among the females than in the males.
   The response in terms of divergence between the selected lines was regressed
on cumulated selection differential as shown in Figure 3. Realized heritability
was higher in the males (14.1 %) but not statistically different from that of fe-
males (11.LE%). It is evident from the plot that selection pressure was continously
exerted on these lines and it was rather constant in its magnitude throughout the
   Selection limits: Response to selection may cease for a variety of reasons. The
most common reasons are those associated with the absence of additional additive
genetic variation and those associated with physiological limits to a trait with the
consequence that natural selection counteracts any artificial selection pressure.
Physiological limits per se have probably not been reached in this selection pro-
gram since much higher blood pressures are tolerated in other mammals. How-
ever, limits may be imposed by the method of measurement which may act in the
same manner as physiological limits. An evaluation of deaths in the 12th genera-
tion revealed a significantly greater mortality among High males than among
either the Lows or Randoms. Among High males almost 30% die in the restrain-
ing chamber before or during the blood pressure determination. This compares to
10% in Randoms and 14% in Lows. This difference is significant at the P < 0.05
level ( x z = 6.61 at 2 degrees of freedom). Among females the percentages are
9% Highs, 0% Randoms, and 11% Lows, which are not statistically different
544                                               G . SCHLAGER


              120-                                                                    HIGH
             ! 110-

             P                                                                        RANDOM
             0                                                                        LOW
             F     80-

                    00  r     I   2   3   4   5     6
                                                        7   8       9       1011 12

       1.-Response to selection for systolic blood pressure in male mice.






                                  .   I   *   .     . .
                                  2 3 4 5 6 7 8 9 1 0 1 1 1 2
                                                                *       I    .   .    .
       2.-Response to selection loor systolic blood pressure in female mice.
                             SELECTION FOR BLOOD PRESSURE                                   541

          0                   IO0           200             300
                             CUMULATED SELECTION DIFFERENTIAL
   FIGURE    3.-Response to selection regressed on cumulated selection differential. Estimates of
heritability were 1 . % in males and 11.4% in females.

from one another. If the males with extreme blood pressure levels are more likely
to die while being measured, this would effectively eliminate a sizeable propor-
tion of the upper end of the distribution from selection.
   Theoretically, selection limits can be predicted by the response in early gen-
erations. DEMPSTER955) has shown that the total advance made in a selection
program should equal 2N times the gain in the first generation, where N is the
effective population size. This prediction assumes that the rate of fixation is low
and that the genes act additively. If dominance is present, the calculation will
result in an underestimation. Previous work has shown that the genes affecting
blood pressure in mice act additively in crosses between inbred strains with rela-
tively high and low blood pressures (SCHLAGER       1968; SCHLAGER WEIBUST
1967) but crosses between these selected lines showed some degree of dominance
 (see below). The estimate of total selection response based on this calculation is
probably an underestimation. The average divergence between the two selected
lines was 7 mmHg in the first generation. The harmonic mean of the number
of parents during the entire study was 17.6, giving an estimate of total divergence
oi 123 mmHg. The actual response was about one-third this prediction.
   ROBERTSON   (1960) has shown that half of the total response should be achieved
in no more than 1.4N generations, and that if the half-life is short of this value,
then the majority of alleles favorable to the direction of the selection will have
been fixed in the population. The response to blood pressure falls far short of the
predicted 25 generations using 1.4N.
   The number of parents used to propagate the next generation varied consider-
ably throughout this study. During the first six generations, 18 to 38 parents were
used as compared to 4 to 18 in the remainder of the study. The High line went
542                               G. SCHLAGER

through a narrow population bottleneck in the 9th generation when only two
pairs of mice were fertile. In the Low line the smallest number was four pairs in
the same generation. The premature cessation of predicted response may be due
to the loss of many favorable alleles as the two lines went through these bottle-
   Comparison of selection results: There are two comparable selection programs
in rodents which used systolic blood pressure levels as the basis for selection.
SMIRK   and HALL    (1958) and PHELAN      (1968) reported on the development of a
New Zealand (NZ) strain of rats with genetic hypertension that was derived
from a single pair of rats of the Wistar strain. Six sublines propagated by sib-
matings were established by more than 20 generations of selection. During this
period the average change in systolic blood pressure in the hypertensive strain
was an increase of about 2 mmHg per generation, resulting in a difference of
50-60 mmHg above the control in the best subline. Heritability was estimated
 at about 15% for males and 10% for females. This may be a conservative esti-
mate of heritability, as there was an indication that there was a tendency for the
blood pressure levels to plateau after about 15 generations in some of the sublines.
    Hypertension was also common in the Japanese spontaneous hypertension rat
 (SHR) resulting from a selection program by OKAMOTO AOKI (1963). Again
the strain was propagated by sib-mating from a single pair of Wistar-derived rats
with elevated systolic blood pressure. Rapid response to selection occurred during
the first four generations and continued selection pressure yielded little further
response. Systolic blood pressures exceeding 200 mmHg are common in the SHR
and the strain shows almost 100% incidence of hypertension. The difference be-
 tween the selected line and the controls is about 70 mmHg in males and 60 mmHg
in females (OKAMOTOal. 1966). Realized heritability was about 20% in males
 and 28% in females.
    The selection programs and progress in these two rat experiments differed in
 a number of ways. The rats of the NZ lines were anesthetized during measure-
 ments while those of the SHR were not. B U ~ A G ,   MCCUBBIN      and PAGE    (1971)
 demonstrated that systolic pressure in the unanesthetized rat’s tail is always
 lower by about 30 mmHg than that in the aorta. I n anesthetized rats, reports
 have shown consistently lower pressures (SHULER,       KUPPERMAN HAMILTON
 1944; FREGLY    1963) or identical pressures (FRANGIPANE       and APORTI    1969) to
 those in the carotid artery. The difference in measuring technique of anesthetiz-
 ing or not would accentuate the diff ereiices found in mean systolic blood pressure.
    Both the NZ and SHR lines were begun from a single-pair samples from
 Wistar and Wistar-derived substrains and in both experiments sib-mating was
 practiced almost exclusively. One would assume that there was only a small
 sample of alleles available for selection in either o i these two lines, yet response
 was rapid in the SHR and very gradual in the NZ.Estimates of realized herit-
 ability in the NZ lines were about half that of the SHR line. These rats are also
 very different physiologically ( (PHELAN    1968).
    In our selection program, the mice were unanesthetized as in the SHR pro-
  gram, but the response was gradual as in the NZ program. The selection lines
                           SELECTION FOR BLOOD PRESSURE                          543
were derived from a more heterogeneous base population (8-way cross) and
double-cousin and sib-matings were avoided wherever possible. Theoretically,
more alleles should be available for selection in our mouse experiments and their
fixation should be more gradual, yet the response was not as great as in either of
the rat experiments. In retrospect, the eight strains used to produce the base pop-
ulation may not have been the ideal genetic pool for selection. These strains were
originally chosen because they were different in origin and maximum genetic
heterogeneity was the goal in the 8-way cross. Strains with more extreme systolic
blood pressures than these eight were later found in a survey of 20 strains
 ( SCHLAGER,   unpublished) and a more genetically heterogeneous base population
could have been produced with different samples of strains.
    Systolic blood pressures of inbred strains at comparable ages are not available
but the strains used in the 8-way cross were measured at 8-10 months of age,
as were the other strains in the survey. The difference between the highest
 (BALB/cJ) and lowest (BDP) of the eight strains was 25 mm Hg and the range
of all strains measured was 38 mm Hg. However, the selected high line in this
experiment was more than 10 mm Hg higher than any of the inbred strains.
    Litter size: The High line had larger litters than the Low line in all but one
generation (Figure 4).This difference between mean number born varied in
magnitude but was greater than one mouse in three generations out of the first
six. This suggests that there may have been some effect of natural selection
against extremely low blood pressures in the Low line acting as a maternal in-
fluence on the trait or the smaller litter size may reflect the loss of those embryos
with extreme blood pressure. The Random line had intermediate litter sizes dur-
ing the first six generations after which it tended to have larger litters than either
of the selected lines.
    Body weight: Body weights were taken each time the blood pressure was de-
termined on a mouse. There was a general trend toward decreasing body weights

                0 1
                            I    I   ,

                       2 3 4 5 6 7 8 9 1 0 1 1 1 2
                                          ,    ,   ,     ,   ,   ,   ,

        4.-Mean litter size far the High and LGW
   FIGURE                                       lines.
544                                           G. SCHLAGER

                                               TABLE 1
                   O t h r trm'ts measured during selection for systolic blood pressure

      Trait               Sex          High                     Random                    O

Body weight ( 9 )          M      27.3 f 0.5 (49)*         29.6   +    0.5 (28)     27.7 f 0.4 (50)
                           F      22.0 + 0.4 (44)          24.1   +    0.4 (23)     22.3 f 0.4 (43)
Pulse rate (bpm)           M      605 + 10 (42)            592    f    12 (24)      580 f 1 1 (38)
                           F      617 + 9 (42)             631    +    14 (22)      596 c 9 (35)
Hematocrit ( X )           M      57.1 + 0.5 (20)          53.1   +    0.5 (15)     51.1 f 0.4 (47)
                           F      55.3 + 0.4 (28)          51.9   ?r   1.0 (6)      50.2 f 0.3 (56)
JGI (unweighted)           M      0.32 ?: 0.03 (15)        0.38   ?:   0.04 (15)    0.35 f 0.02 (15)
                           F      0.39 F 0.02 (15)         0.35   +    0.02 (15)    0.32 2 0.03 (15)
JGI (weighted)             M      0.52 + 0.05 (15)         0.55   +    0.05 (15)    0.67 2 0.05 (15)
                           F      0.58 + 0.04 (15)         0.67   f    0.05 (15)    0.49 0.04 (15)

  * Mean 1 standard error (sample size).

in the three lines during the 12 generations, probably due to the inbreeding with
passing generations. The Low line was consistently slightly heavier than the
High line and both selected lines were lighter than the Random line by genera-
tion 12 (see Table l ) . The difference between selected lines was not significant
but both were significantly lighter than the Random (P < 0.01) .
   PHELAN    (1968) reported that the NZ spontaneous hypertension rat was lighter
than the control line starting at four weeks of age in both sexes. This relationship
was not as consistent in the SHR males, and SHR females tended to be heavier
than controls beyond 15 weeks of age (OKAMOTO al. 1966). Clearly in the
mouse, differences in blood pressure between lines are not associated with body
weight at the ages these are being measured.
   Pulse rate: In the twelfth generation samples of pulse rate were taken from
the physiograph tracings. These values are given in Table 1. The pulse rate was
higher in the males of the High line than in the Lows, and the Random line was
intermediate. An analysis of variance showed that these differences were not sig-
nificant (F = 2.84, P > 0.05). In females, the Random line had the highest pulse
rate; again these differences were not significant (F = 1.53,P > 0.05). One can-
not attribute the blood pressure differences to excitability, which certainly would
be reflected in higher pulse rates.
   Hematocrits: An association between hypertension and hematocrit level has
been reported in the medical literature. During the early generations of selec-
tion, there was no evidence of a striking difference between the selected lines. In
the tenth generation, however, significant differences were found among the
lines, with the Random line intermediate to the higher levels in the High line and
lower levels in the Low lines (Table 1) .
   A similar association was previously reported in the A/J and SWR/J inbred
strains of mice were elevated hematocrit was found in the SWR/J strain which
                          SELECTION FOR BLOOD PRESSURE                           545
had the higher blood pressure (SCHLAGER      1968). I n that study, it was concluded
that the association was fortuitous in the sense that the SWR/J strain carried
genes for both the elevated blood pressure and hematocrit level. That conclusion
was based on the absence of a correlation between these two variables in the F,
generation of the crosses between the strains. An F, generation from crosses be-
tween the High and Low selected lines also lacked a significant correlation coeffi-
cient ( r = 0.16, P > 0.05). Hematocrit level appears to be a trait influenced by
few genes which respond rapidly to selection ( SCHLAGER,         unpublished). The
higher hematocrits in the High line may have resulted from the chance fixation
of alleles for elevated levels when the line went through its population bottleneck
in generation 9. However, one could argue that chance fixation of high alleles in
the High line simultaneously with low alleles in the Low line is unlikely.
    Renin granularity: Right kidneys of a sample of 15 mice of each sex and line
of the fifth generation were prepared for histological examination by the methods
described in an earlier paper (SCHLAGER     1968). Both a weighted and unweighted
juxtaglomerular cell index (JGI) were calculated and these results are shown in
Table 1. A two-way analysis of variance for sex and lines main effects was per-
formed for each of the two scores. NQsignificant difference could be found be-
tween sexes or lines in either analysis.
    JGI was previously found to be independent of blood pressure level in lines se-
lected for JGI (RAPP 1965) and in A/J and SWR/J strains of mice which ex-
hibited large differences in blood pressure (SCHLAGER    1968).
    F, and backcrosses: A comparison was made of the selection lines of generation
12 and the F, and segregating generations produced by parental selection lines of
generation 11. These were measured during the same time period, thus avoiding
any effects of the fluctuations in systolic blood pressures seen throughout the
selection program. The analysis of the first degree statistics is summarized in
Table 2. The procedures followed in this analysis were presented in detail by
MATHER JINKSand         (1971). The data were first analyzed in the original scale of
measurement. There were no significant deviations from zero in any of the scal-
ing tests, suggesting that an additive-dominance model was adequate. The genetic
parameters were then estimated by the least squares technique in which the six
equations based on the contributions of each type family to m, the midpoint be-
tween the two selected lines, [ d ] ,the sum of the average effect of the alleles, and
 [ h ], the deviation of the F, from the midparental value as a measure of overall
dominance. The six equations were weighted by the reciprocal of the variance of
the means and multiplied through by the value of the coefficient of the genetic
parameter to yield three simultaneous equations for the three unknowns. These
were then solved by matrix inversion to yield the estimates given in Table 2.
The adequacy of the additive-dominance model was then further tested by com-
puting the expected values of the six generation means and comparing these to
the observed data. A goodness-of-fit test demonstrated an adequate agreement.
To further demonstrate that non-allelic or epistatic interactions was absent. an
attempt was made to fit a model including the genetic components [i], additive X
additive interactions, [ i ] , additive X dominance interactions, and [Z] , domi-
546                                            G . SCHLAGER

                                                 TABLE 2

                     Genetic parameters for the selected lines, P I , F,, and backcrosses

                                                       Original scale               Logarithmic scale
                                                    mean + standard ermr          mean f standard error

         Generation (sample s i z e )
           High         (49)                            124    f    2.7            2.0899 f      0.0096
                        (14)                            128    f    6.9            2.1014 f      0.0236
                        (21 1                           114    f    2.3            2.0536 f      0.0092
                        (401                            116    f    3.6            2.0575 f      0.0138
                        (23)                            106    f    4.8            2.0138 f      0.0194
                               (50)                      89    f    2.0            1.9441 f      0.0095
         Scaling tests
            A                                             9    f    10.1           0.0594 f 0.0490
                B                                        18    f    14.3           0.0299 f 0.0410
                C                                       -23    f    15.5           0.0890 f 0.0599
         Genetic parameters
                                                      107.7 f 1.62                 2.0206 2 0.0065
            [                                          18.1 f 1.64                 0.0733 i 0.0066
            Chl                                         7.2 f 2.82                 0.0397 f 0.0113
      Generation (sample s i z e )
         High                                           117 f 2.2                  2.0631    + 0.0083
                B,                                      118 f 4.8                  2.0659    f 0.0172
                F!                                      114 e 5.0                  2.0524    f 0.0188
                                                        108 f 2.5                  2.0288    f 0.0108
                                                         96 f 3.7                  1.9752    f 0.0172
                                                         85 f 2.1                  1.9264    f 0.0106

         Scal i ng t e s t s
            A                                              7   f     9.2           .0.0283 f 0.0406
                B                                         -5   f    11.2            0.0163 f 0.0401
                C                                         -2   _+   14.5            0.0210 f 0.0588
         Genetic parameters
                m                                     101.5    f    1.46            1.9946   2   0.0065
            [d]                                        15.6    2    1.47            0.0699   f   0.0065
            Chl                                        11.3    f    3.74            0.0597       0.0160

nance x dominance interactions. These were shown to be not significantly dif-
ferent from zero.
   The additive-dominance model also assumes that there is no genotype x en-
vironment interactions. A crude test for the presence of this interaction is the
comparison of the variances of the two parental lines and the F, generation. The
homogeneity of these variances was tested by an F,,, test and showed homogene-
ity of the variances in the females but some heterogeneity in the males (Fmax   =
3.21, P < 0.05). Consequently, the analyses and estimations were repeated using
a logarithmic scale where both the absence of non-allelic interaction and the ab-
sence of genotype x environment interaction were demonstrated. The data, scal-
ing tests, and estimations of the genetic parameters are also shown in Table 2.
                            SELECTION FOR BLOOD PRESSURE                                  547


        L     BCL      Fl       BCH      H               L     BCL      Fl     BCH        H
                       F2                                               F2
        5.--Segregating     and non-segregating generation means in a genetic triangle.

   Regardless of the scale, the results clearly show that both the [ d ] and [h] com-
ponents are significantly different from zero. the [h] component is positive, so
we can conclude that dominance is present with the alleles for elevated blood
pressure being on the average dominant to those for lowering blood pressure. The
ratio of [h] to [ d ] is a measure of the average degree of dominance; this was
higher in females (0.85) than in males (0.54). The relationships among these
six generations can be seen in Figure 5 where the data are plotted in the loga-
rithmic scale.
   TANASE (1970) made genetic crosses between the SHR and each of three
           et aE.
inbred strains of rats. The resulting genetic triangles were much flatter than
those of our data with the F, systolic blood pressure very near the mid-parental
value. The F, blood pressures were consistently lower than the F, in all three
crosses, while the backcrosses to the high SHR strain tended to be nearer the F,
value than to the SHR. The genetic parameters calculated by TANASE                using
MATHER’S    (1949) variances method did show a sizable dominance component,
although in the absence of standard errors, it is difficult to assess its significance.
   The genetic parameters in the mouse cross correspond clearly to what can be
readily seen in the graphic representation of the genetic triangle (Figure 5).
There is a large additive component and a smaller dominance component, both
of which are significant in the male and female data. This differs from the genetic
analysis of blood pressure in crosses between inbred strains (SCHLAGER              and
WEIBUST    1967; SCHLAGER      1968). The F, males in crosses between A/J and
BALB/cJ were intermediate to the parental strains and subsequent backcrosses
gave intermediate means ,between the F, and the parental strains. The female
data did show an F, systolic blood pressure near the high parental strain value.
I n another cross, A/J X SWR/J, the results were similar, with no dominance in
548                                     G. SCHLAGER

the males but some evidence for dominance in the females. Genetic parameters
were estimated in the A/J x SWR/J cross and the dominance component was
not significant. Of these three strains (A/J, BALB/cJ, and SWR/J), only the
BALB/cJ was used in the eight-way cross to produce the base generation for se-
lection. There were evidently alleles contributed by some of the other seven
strains which show dominance for elevated systolic blood pressures.
    I am grateful to DR. THOMAS RODERICK samples of mice from his eight-way cross
stocks. This research was supported in part by grant HE-09331 from the National Heart Institute,
a grant from the Kansas Heart Association, an allocation from the Biomedical Sciences Support
Grant RR-07037, and the General Research Fund of the University of Kansas.

                                     LITERATURE CITED

       R.                      and I. H. PAGE,  1971 Lack of correlation between direct and
   indirect measurements of arterial pressure in unanaesthetized rats. Cardiovascular Res. 5:
   24-3 1.
DEMPSTER, R., 1955 Genetic models in relation to animal breeding problems. Biometrics 11:
            G.                   1969 Improved indirect method for the measurement of systolic
   blood pressure in the rat. J. Lab. Clinic. Med. 73: 872-876.
FREGLY, J., 1963 Factors affecting indirect determination of systolic bbod pressure of rats.
   J. Lab. Clinic. Med. 62: 223-230.
         K.                                   J.      M.          and
   chronic salt ingestion; inheritance of hypertension in the rat. J. Exptl. Med. 132: 976-1000.
       W.              A.              and            1969 Inheritance of high blood-pressure
   i n spontaneously hypertensive rat. Lancet: 1035-1036.
MATHER, 1949 Biometrical Genetics. Methuen, London
       K.              1971 Biometrical Genetics. Cornel1 University Press, Ithaca, New
OKAMOTO, and K. AOKI, 1963 Development of a strain of spontaneously hypertensive rats.
   Japan. Circul. J. 27: 282-293.
                       M.                          Y.            K.
                                        S. NOSAKA, YAMORI, ICHIJIMA, HAEBARA,  H.            M.
               T.               Y.
   MATSUMOTO, MARUYAMA,SUZUKI M. T.~MEGAI, and               1966 Further observations of the
   development of a strain of spontaneously hypertensive rats. Japan. Circul. J. 30: 703-716.
PHELAN, L., 1968 The New Zealand strain of rats with genetic hypertension. N. Z. Med. J.
   67 : 334-344.
RAPP,J. P., 1965 Alteration of juxtaglomerular index by selective inbreeding. Endocrinology
   76 : 486-490.
ROBERTSON. 1960 A theory of limits in artificial selection. Proc. Roy. Soc. B 153: 234-249.
SCHLAGER, 1968 Genetic and physiological studies of blood pressure in mice. I. Crosses be-
   tween A/J, SWR/J, and their hybrids. Can. J. Genet. Cytol. 10: 853-864.          ~, 1971
   Genetic control of blood pressure by more than one pair of alleles. Proc. Soc. Exptl. Biol.
   Med. 136: 863-866. -    ,       1972 Spontaneous hypertension in laboratory animals: A
   review of the genetic implications. J. Hered. 63: 35-38.
                           1967 Genetic control of blood pressure in mice. Genetics 5 5 :
                            SELECTION FOR BLOOD PRESSURE                                 549
SHULER, H., H. S. KUPPERMAN W. F. HAMILTON, Comparison of direct and indirect
       R.                       and                     1944
   blood pressure measurements in rats. Am. J. Physiol. 141 : 625-629.
SMIRK,F. H. and W. H. HALL,
                          1958 Inherited hypertension in rats. Nature 182: 727-728.
STURTEVANT,M., 1953 Artificial selection program for degenerative diseases in rats. Genetics
   38: 696.
                          A.             Y.                            1970 Genetic analysis of
   blood pressure in spontaneously hypertensive rats. Japan. Circul. J. 34: 1197-1212.
         R.                        1968 A genetic study of blood pressure, hematocrit and plasma
   cholesterol i n aged mice. Life Sci. 7:1111-1 119.
                                                               Corresponding editor: E. RUSSELL

To top