646
Double Knockouts
Production of Mutant Cell Lines in Cardiovascular Research
Richard M. Mortensen
Double knockouts by homologous recombination is a method Tor producing cell lines with an inactivating
mutation in any desired gene. The biochemical analysis of genetically altered cell lines has been important
in determining the function of specific proteins. Until recently, mutant cell lines have been produced by
random mutagenesis and then selection for a particular phenotypic change. Recent technological
advances in gene targeting by homologous recombination now enable the production of mutants in any
desired gene. Diploid cells contain two copies or alleles of each gene encoded on an autosome (nonsex)
chromosome. In most cases, both alleles must be inactivated to produce a phenotypic change in a mutant
cell line, hence the term "double knockout." We and others have described the production of mutationally
altered cell lines by inactivating both alleles by the production of two targeting vectors, two separate
homologous recombination events, and selection. A simpler procedure, involving considerably less effort
and time, has been used to inactivate several or-subunits of G proteins and other genes. This method
facilitates the inactlvation of more than one gene in a single cell line. (Hypertension. 1993;22:646-651.)
KEY WORDS
• genetics • cell line • recombination, genetic • mutation
T
he analysis of mutant organisms and cell lines has
been important in determining the function of
specific proteins. Until recently, mutants have
been produced by random mutagenesis and then selection for a particular phenotypic change. Recent technological advances in gene targeting by homologous recombination in mammalian systems now enable the
production of mutants in any desired gene. 10 This
technology can be used to produce mutant mouse
strains and mutant cell lines. Diploid cells contain two
copies or alleles of each gene encoded on an autosome
or nonsex chromosome. In most cases, both alleles must
be inactivated to produce a phenotypic change in a
mutant. The conversion to homozygous mutations is
accomplished by breeding in the case of mouse strains
and by direct means for cell lines.
To produce a mutant mouse strain, the desired
mutation is first introduced into cloned DNA sequences. This targeting construct is then transfected
into a cultured embryonic stem (ES) cell line. These cell
lines are derived from the inner cell mass of normal
blastocysts and maintain the ability to differentiate into
every cell type. Homologous recombination will occur
in a small number of these cells, resulting in introduction of the mutation into the genome. Once identified,
Received March 17, 1993; accepted in revised form June 21,
1993.
From Brigham and Women's Hospital, Harvard Medical
School, Boston, Mass.
Presented as a part of the Sunday Afternoon Program "Manipulating Genes to Understand Cardiovascular Diseases: Principles,
Methodologies, and Applications" at the American Heart Association's 65th Scientific Sessions, New Orleans, La, November 15,
1992.
Reprint requests to Richard M. Mortensen, MD, PhD, Endocrine-Hypertension Division, 221 Longwood Ave, Boston, MA
02115.
these mutant clones can be microinjected into a normal
blastocyst to produce a chimera. A chimera can have
tissues, including the germ line, with contribution from
the normal blastocyst and from the injected ES cells.
Breeding these germ-line chimeras yields animals heterozygous for the mutation introduced into the ES cell.
Heterozygotes can be bred to produce homozygotes.
Homologous recombination can also be applied to
produce mutant cultured cell lines. Although previously
inactivation of both copies of a gene required two
rounds of homologous recombination and selection, 46
inactivation of many genes now only requires the production of a single targeting construct.7 These mutants
can be analyzed for phenotypic changes to determine
the function of the gene. This method facilitates the
inactivation of more than one gene in a single cell line.
Gene Targeting of the First Allele
There are two basic configurations of constructs for
homologous recombination, termed insertion and replacement constructs (Fig 1A). In insertion constructs,
the sequences in the targeting vector are introduced
into the homologous site, interrupting the normal structure of the gene. The region of homology to the target
gene is cloned as a single continuous sequence, and the
construct is linearized by cleavage at a unique restriction site in the region of homology. Homologous recombination adds sequences to the target gene. As a result,
the normal gene can be regenerated by an intrachromosomal recombination event.
The second type of construct is more commonly used
and is termed a replacement construct. There are two
regions homologous to the target gene on either side of
a mutation (usually a positive selectable marker, see
below). Homologous recombination proceeds by a double crossover event that results in the replacement of
the endogenous sequences with the construct se-
Mortensen
Homologous Recombination
647
A CONSTRUCT TYPES
REPLACEMENT CONSTRUCT
INSERTION CONSTRUCT
2*
3*
x
4—•-•—»
I
1
1
2 I 3
2 3*
2' 3
B ENRICHMENT BY POSITIVE-NEGATIVE SELECTION
HOMOLOGOUS RECOMBINATION
2'
RANDOM INSERTION
3*
TK
2*
CELL PHENOTYPE:
3*
G418R
GANCS
Homologous sequences
construct and genomtc
Non-homologous sequences
positive selectable marker
genomtc sequences
•
•
•
negative selectable marker E2
vector sequences
—
FIG 1. A: Diagrams show two configurations for homologous recombination. Homologous sequences in the construct are
indicated by an asterisk. Replacement constructs substitute their sequences (exon 2* neo, exon 3*) for the endogenous sequences
(cxons 2 and 3). Insertion constructs add their sequences to the genome, resulting in tandem duplication and disruption of the
normal gene structure. Insertion constructs could also incorporate interruption of an exon by neo. B: Diagram shows enrichment
by positive-negative selection using the herpes simplex virus thymidine kinase gene (HSV- TK). Crossover on either side of the neo
gene results in bss of the TK if homologous recombination occurs. Random insertion tends to preserve the TK. Presence of the TK
gene can be selected against because any cell expressing the gene will be killed by gancyclovir (GANC). The construct is shown
linearized so that the plasmid vector sequences are attached to the TKgene. This configuration helps preserve the integrity of the
TKgene. Superscript R denotes resistance to the antibiotic; S denotes sensitivity.
quences. No duplication of sequences occurs, so the
normal gene cannot be regenerated.
Nearly all constructs for homologous recombination
rely on the positive selection of a drug resistance gene
(eg, neo) that is usually also used to produce the
interruption in the gene. When either an insertion or
replacement construct is linearized, the positive selectable marker is flanked by two regions homologous to the
target gene. These regions of genomic sequence provide
the substrates for homologous recombination. Gener-
ally, the homology regions should be greater than 1 kb
on each side of the neo gene, with a total homology of 6
kb or greater. The degree of homology between the
construct and the target genome can have a dramatic
effect on the rate of homologous recombination in two
ways. First, the DNA used to construct the targeting
vector must be from the same species as the cell in
which the mutation is to be introduced. Homologous
recombination requires stretches of exact homology.
Because different animal strains may differ just as
648
Hypertension
Vol 22, No 4
October 1993
Construct
Homologous
rvcomUnant
TfSBl
±±
Nail
8 1 kb
FIG 2. Diagram shows structure of the targeting construct
used to inactivate the a^ gene. The construct has approximately 7 kb of homology and the neo gene interrupting exon
6. Homologous recombination is detected by Southern analysis by digestion ofgenomic DNA with Nsi / and hybridization
with a probe (BamHI-BamHI) outside the region of homology
with the construct (size of predicted hybridizing fragments
shown as solid bars). Adapted from Mortensen et al.7
individual outbred animals differ, there may be a mismatch on average every 500 bp; therefore, it is best if
the homology segments are derived from DNA that is
isogenic with the target DNA (ie, from the same strain
of animal or made from the cell line directly). The
nonisogenic mismatch is sufficient to dramatically decrease the rate of homologous recombination.89 However, constructs derived from nonisogenic DNA often
work well. Secondly, the longer the homologous regions,
the higher the rates of recombination (within limits).
The exact length at which creating longer constructs will
not increase recombination rates is controversial but
may be up to 15 kb. 810 Furthermore, fidelity of recombination can be lower if the length of homology is
shorter than 1 kb on a side.11
For the a^ gene, the construct was devised as a
replacement vector using positive-negative selection (illustrated in Fig IB and explained below). The neo was
inserted into the BamHl site in exon 6, and the a$ gene
was interrupted at the Neo I site at the initiating ATG
codon in exon 1 (Fig 2).5 The neo coding sequence was
expressed using a promoter active in ES cells. Neither of
these constructs was isogenic with the target DNA,
because they were derived from a BALB/c genomic
library and the ES cells are derived from the 129Sv
murine strain. The construct is introduced into ES cells
by electroporation. Selection of the cells in G418 eliminates the great majority of clones that do not stably
incorporate the construct. However, many of the clones
will incorporate the construct through random integration in the genome.
To enrich for homologous recombinant colonies, the
herpes simplex virus thymidine kinase gene (HSV-TK)
is included outside the regions of homology. Cells
lacking the TK gene can be selected by treatment of a
cell culture with gancyclovir.12 This method for further
enrichment of homologous recombinants is termed positive-negative selection (Fig IB). It is applicable to
replacement constructs only. The presence of this thymidine kinase gene makes cells sensitive to acyclovir
and its analogues such as gancyclovir. HSV-TK enzyme
can activate these drugs, which are then incorporated
into growing DNA during synthesis, causing chain termination and death of the cell. Sequences outside the
regions of homology will be lost during homologous
recombination because of crossover. However, all sequences in the construct will tend to be retained if
random integration occurs, because recombination usually occurs at the ends of the construct. The presence of
the HSV-TK gene can be selected against by gancyclovir. Therefore, homologous recombinants will be neo
resistant and gancyclovir resistant, whereas random
integration of the construct will give neo resistance but
also gancyclovir sensitivity. In some cases, the TK will
be inactivated without homologous recombination so
that surviving clones must be screened for the true
homologous recombinants. Other markers that are lethal to cells containing them have also been used, eg,
diphtheria toxin.13 The degree of enrichment varies but
is generally fivefold to 10-fold (0,2, threefold to fivefold;
a-,, 10-fold).
The CCQ targeting vector was transfected into the cells,
selected in G418 and gancyclovir, and the surviving
colonies screened for homologous recombination by
Southern analysis. Correct homologous recombination
results in a change in the size of the hybridizing band
using a probe homologous to a region not contained in
the construct (Fig 2). The probe, for detection of
homologous recombination by Southern analysis, can
usually be obtained from sequences in the original
genomic clone that were not used for the construct.
With the use of a probe outside the sequences in the
construct, only the endogenous gene and homologous
recombinants will be detected, not the construct randomly integrated into the genome. The probe should
detect no more than a few bands on a Southern blot of
genomic DNA.
Only clones that have undergone homologous recombination are analyzed further. Usually, no other recombination events have occurred in these clones. However,
it is prudent to test for a second copy of neo to ensure
that a random integration did not occur in the same
clone, because this could cause insertional mutagenesis
or the failure of selection of double knockouts using the
single construct method (below).
Double Construct Method
The second allele can be inactivated by a second
round of gene targeting (Fig 3, Double Construct Method). Because the cells already express resistance to
G418, a second selectable marker must be used to
interrupt the exon. The hygromycin B resistance gene
(hyg) has been used either as a fusion protein* or driven
by its own promoter.5-6 The second round is identical to
the first. The second targeting construct can replace
either the normal allele or the allele that has already
been inactivated by neo. Homozygous mutants are determined by Southern analysis and the presence of both
the hyg- and neo-altered alleles.
This method is time consuming in that it requires the
creation of a second targeting construct, and the rate of
homologous recombination with the second construct will
be similar to the first. For genes that undergo homologous
recombination at low rates, screening a second time would
be labor intensive. Most importantly, there are a very
Mortensen
FIRST TARGETING
649
DOUBLE CONSTRUCT
METHOD
SINGLE CONSTRUCT
METHOD
•
1
CREATE CONSTRUCT
n»o
Homologous Recombination
CREATE CONSTRUCT
TK
hyg
TRANSFECT
TK
TRANSFECT
SELECT
HGHO418
\
Homologoua MquanoM
Non-homotogou» taquano—
pootthro NwtiibtB motor
I K f l l • lltli HKl ! ! • * •
vtdof Mquanc««
promotor saquano*
SELECT
LOW 0418
QANC
SELECT LOW
HYQROUYCIN B
QANC
SCREEN
SCREEN
D
n
SCREEN
•
0418
n
HXMOlygote
HYGROMYCJN B R
HonoiygoM
FIG 3. Diagram shows scheme for production of homozygous mutants using two separate targeting constructs or a single
construct. Resulting cell lines from the two construct method are resistant to both G418 and hygromycin B, whereas using a single
construct the homozygous cell is resistant only to G418. TK, herpes simplex virus thymidine kinase gene (HSV-TK); GANC,
gancyclovir; superscript R, resistance to the antibiotic.
limited number of usable selectable markers so that the
inactivation of both alleles of two or more genes will be
difficult or impossible. This is particularly important when
investigating the function of a family of genes that may
have overlapping phenotypes, because more than one
member of a family would need to be inactivated to
observe the desired phenotype.
Single Construct Method
A simpler, more efficient method can be used to
obtain homozygous knockout cells that are resistant to
only a single antibiotic (Fig 3, Single Construct Method). This method has been applied using a neomycin
resistance gene that contains a point mutation resulting
in lower phosphotransferase activity.14 If the wild-type
gene is used, it may result in cells that are resistant to
antibiotics even at very high levels. Once a heterozygous
targeted clone has been obtained, a homozygous cell
line can be isolated from these heterozygous cells by
selecting cells that are resistant to higher concentrations
of G418 than the homozygous cells. Cells that have lost
heterozygosity contain two interrupted alleles and
therefore contain two neo genes, which make the cells
more resistant to G418. The heterozygous cells are
selected at higher concentrations of G418 (>1.0 mg/
mL). The cells are cultured for 7 to 10 days until single
surviving colonies are detected. For some genes, higher
G418 concentrations may be required, presumably depending on the relative expression of the neomycin
resistance gene at different genomic sites. The relative
resistance of the heterozygous cells to G418 will also
depend on the promoter used to drive the expression of
neo or on the integration site so that other clones may
require more or less G418. If cells overgrow plates and
no single colonies are obtained, the cells should be
replated using higher G418 levels (some clones have
required greater than 10 mg/mL). The surviving clones
are screened by Southern blot analysis exactly as performed for the isolation of the heterozygous cell line
except that the Southern blot of the DNA derived from
a double knockout (homozygous) clone will now completely lack the band corresponding to the gene segment
found in the normal parent cell. Examples of Southern
analysis of two clones homozygous for a^ gene inactivation are shown in Fig 4 (lanes 3 and 4).
The frequency of homozygous clones differs between
heterozygous clones because loss of heterozygosity ap-
Nsl1
1 2
8.1 —
3
4
*~
6.5— - •
FIG 4. Southern analysis shows DNA isolated from embryonic stem cells: wild type (+/+), heterozygous for gene
disruption with neo (+/-), and homozygous for the inactivated allele (—/—). Similar to previously published data.7
650
Hypertension
Vol 22, No 4 October 1993
Loss of Heterozygoslty In Embryonic Stem Cell Lines
[G418], mg/mL
2.0
1.0-1.5
Gene
(Parent Une)
aa (CCE)
a0 (CC1.2)
TCR-a (CCE)
Myhc-b (CC1.2)
Cell Une
No. Cells
Plated*
No. of
Colonies
% Homozygote
(No. Analyzed)
No. of
Colonies
% Homozygote
(No. Analyzed)
17E10
2X10 4
23
100% (21)
2
0%(2)
18D3
2X10
4
175
12% (26)
40
ND
32-32
1x10=
100
53% (17)
17
88% (17)
32-28
1X105
27
4% (28)
0
1A4
5x10"
37
43% (23)
9
76
7% (76)
5
6-6
5X10
6-22
5X10 5
44% (9)
54
24% (54)
2
50% (2)
ND indicates not determined. Delta from Mortensen et al. 7
•Efficiency of plating 40% to 80%.
pears to be a random event. Several separately derived
heterozygous clones should be expanded in culture and
subjected to selection with higher levels of G418, because
there appears to be some clone-to-clone variation in
resistance to this drug. The frequency of homozygous
clones may also be influenced by the time in culture before
selection because the homozygous clones are continuously
produced and tend to accumulate with time. Although the
frequency of homozygous clones varies from 4% to 100%,
it is often around 50% (Table). These cells retain their
normal karyotype, and so this method may be useful for
studying genes in development.
The mechanism by which these homozygous cells are
produced is unclear. Possibilities include chromosomal
loss and duplication, nondisjunction, or even gene conversion. Previous descriptions of spontaneous production of homozygous clones have shown that the mechanism may vary with cell type. 1517 Further definition of
the mechanism in ES cells will require cells that are
heterozygous at multiple loci.
Inactivation of More Than One Gene
in a Single Cell Line
Because the homozygous cells are only resistant to
G418, other selectable markers such as the hygromycin
B resistance gene could be used in further homologous
recombination targeting constructs. Other selectable
markers might also be used to inactivate three or more
genes as long as the presence of two copies can be
selected over the presence of one copy. In this way, the
functions of genes with overlapping phenotypes might
be investigated.
Uses of Mutant Cell Lines
Once mutant cell lines have been produced, the cells
can be analyzed to confirm inactivation of the gene and
for phenotypic differences. Expression can be evaluated
by Northern blot analyses of total RNA or Western blot
analyses of proteins isolated from the cell line. The
interruption of exon 6 in the eta gene leads to complete
absence of detectable eta mRNA by Northern analysis.7
For these ES cells lacking functional eta or a^ genes,
coupling of various receptors to different effectors can
be analyzed provided that the responses are present in
the wild-type cells. So far, no receptors known to couple
through Oj have been identified in ES cells,5 so two
approaches can be taken.
First, receptors can be heterologously expressed in
the undifferentiated ES cells. Cell lines expressing the
receptor would be developed in the wild-type and the
null mutant cell background and tested for coupling to
adenylyl cyclase, phospholipase C, phospholipase A2,
mitogenesis, or other G protein-mediated signals. This
method requires the identification of the desired response in the wild-type cell line.
Alternatively, ES cells offer a unique opportunity
because they are capable of differentiating into any cell
type. ES cells have been shown to differentiate into a
number of cell types that are of interest to investigators
in cardiovascular research. Cardiac cells express cardiac-specific a and /3 cardiac myosin.18'19 They beat and
possess the same K+ channels identified in normal
heart. Vascular channels have also been identified in
differentiating embryoid bodies (which are masses of
cells derived from ES cells that have undergone differentiation into an embryo-like structure) and can even
contain blood elements.20 Provided that the phenotype
can be analyzed with a single or few cells, these differentiated cells can then be used to analyze the phenotype
of the disrupted gene.
ES cells can also differentiate within the entire organism by injection into normal blastocysts (homozygous
mutant cells offer no technical advantage over heterozygous cells if the goal is to obtain a mutant mouse line
through germ-line transmission). For the analysis of
homozygous knockout cells in a developing chimeric
embryo, the cells can be tagged by introducing a gene
that has a histochemically detectable product (such as
0-galactosidase). The contribution of these tagged cells
to tissues could then be determined by analyzing for the
histochemical tag. Identified differentiated cells derived
from the homozygous knockout ES cells could also be
analyzed for altered functions at the single cell level.
Once a particular phenotype has been shown to be
dependent on the presence of a gene product, the
system can be easily used to study the structure-function
relations. For ctj genes, mutants can be reintroduced
into the null cells, and there, function can be studied.
This is particularly useful for studying the G proteineffector interactions.
Mortensen Homologous Recombination
Application of the Single Construct Method to
Other Cell Lines
The suitability of this technique for the production of
types of cells other than ES cells that lack a particular
gene is less certain. Because homologous recombination
and spontaneous loss of heterozygosity are known to
occur in other cell lines, these same methods may be
used to produce cell lines lacking other functional
genes. However, other types of immortalized cells (eg,
lymphocytes or fibroblasts) may not allow the targeting
vector to undergo homologous recombination at the
same frequency as found in ES cells. Still, homologous
recombination at usable rates has been reported in
immortalized cell lines.21 Furthermore, many other
types of immortalized cells are polyploid or aneuploid,
so that particular problems may arise for any one cell
type.
The method described here is a general one that can
be used to produce a large variety of cell lines, each one
of which will lack a particular gene. This approach
cannot be used to produce double knockout cell lines
for genes that are required for cell viability. This
method has been successfully used to produce cells that
lack a a . a o , o-cardiac myosin heavy chain, ^-cardiac
myosin heavy chain, and T-cell o-receptor genes. We
expect that future studies will lead to the production of
many other such cell lines and that the characterization
of these cell lines will lead to the definition of the role
of different gene products in cell structure and function.
Acknowledgment
R.M.M. was supported by a Clinician Scientist Award from
the American Heart Association.
References
1. Mansour SL. Gene targeting in murine embryonic stem cells:
introduction of specific alterations into the mammalian genome.
Genet Anal Tech AppL 1990,7:219-227.
2. Robertson EJ. Using embryonic stem cells to introduce mutations
into the mouse germ line. Biol Reprod, 1991;44:238-245.
3. Zimmer A. Manipulating the genome by homologous recombination in embryonic stem cells. Annu Rev NeuroscL 1992; 15:
115-137.
4. teRiele H, Maandag ER, Clarke A, Hooper M, Berns A. Consecutive inactivation of both alleles of the pun-1 proto-oncogene by
homologous recombination in embryonic stem cells. Nature. 1990;
348:649-651.
651
5. Mortensen RM, Zubiaur M, Neer EJ, Seidman JG. Embryonic
stem cells lacking a functional inhibitory G-protein subunit (alpha
i2) produced by gene targeting of both alleles. Proc Natl Acad Sci
USA. 1991;88:7036-7040.
6. Cruz A, Coburn CM, Beverley SM. Double targeted gene
replacement for creating null mutants. Proc Natl Acad Sci USA.
1991;88:7170-7174.
7. Mortensen RM, Conner DA, Chao S, Geisterfer LA, Seidman JG.
Production of homozygous mutant ES cells with a single targeting
construct. Mol Cell BioL 1992;12:2391-2395.
8. Deng C, Capecchi MR. Reexamination of gene targeting frequency
as a function of the extent of homology between the targeting
vector and the target locus. Mol Cell BioL 1992;12:3365-3371.
9. teRiele H, Maandag ER, Berns A. Highly efficient gene targeting
in embryonic stem cells through homologous recombination with
isogenic DNA constructs. Proc Natl Acad Sci USA. 1992;89:
5128-5132.
10. Hasty P, Rivera PJ, Bradley A. The length of homology required
for gene targeting in embryonic stem cells. Mol Cell BioL 1991;11:
5586-5591.
11. Thomas KR, Deng C, Capecchi MR. High-fidelity gene targeting
in embryonic stem cells by using sequence replacement vectors.
Mol Cell BioL 1992;12:2919-2923.
12. Mansour SL, Thomas KR, Capecchi MR. Disruption of the protooncogene int-2 in mouse embryo-derived stem cells: a general
strategy for targeting mutations to non-selectable genes. Nature.
1988;336:348-352.
13. Yagi T, Ikawa Y, Yoshida K, Shigetani Y, Takeda N, Mabuchi I,
Yamamoto T, Aizawa S. Homologous recombination at c-fyn locus
of mouse embryonic stem cells with use of diphtheria toxin
A-fragment gene in negative selection. Proc Natl Acad Sci USA.
1990;87:9918-9922.
14. Yenofsky RL, Fine M, Pellow JW. A mutant neomycin phosphotransferase II gene reduces the resistance of transformants to
antibiotic selection pressure. Proc Natl Acad Sci USA. 1990;87:
3435-3439.
15. Rajan TV, Moffat LF, Frankel WN. Rate and mechanism of generation of beta 2-microglobulin mutants from a heterozygous
murine cell line. J ImmunoL 1990;145:1598-1602.
16. Potter TA, Zeff RA, Frankel W, Rajan TV. Mitotic recombination
between homologous chromosomes generates H-2 somatic cell
variants in vitro. Proc Natl Acad Sci USA. 1987;84:1634-1637.
17. Wasmuth JJ, Vock HL. Genetic demonstration of mitotic recombination in cultured Chinese hamster cell hybrids. Cell 1984;36:
697-707.
18. Sanchez A, Jones WK, Gulick J, Doetschman T, Robbins J.
Myosin heavy chain gene expression in mouse embryoid bodies: an
in vitro developmental study. J Biol Chem. 1991;266:22419-22426.
19. Robbins J, Gulick J, Sanchez A, Howies P, Doetschman T. Mouse
embryonic stem cells express the cardiac myosin heavy chain genes
during development in vitro. J Biol Chem. 1990;265:l 1905-11909.
20. Wang R, Clark R, Bautch VL. Embryonic stem cell-derived
embryoid bodies form vascular channels: an in vitro model of blood
vessel development. Development. 1992;114:303-316.
21. Accili D, Taylor SI. Targeted inactivation of the insulin receptor
gene in mouse 3T3-L1 fibroblasts via homologous recombination.
Proc Natl Acad Sci USA. 1991;88:4708-4712.