![[Feedback]](/icons/banner/feedbackACT.gif)
![[For Subscribers]](/icons/banner/subscriberACT.gif)
![[Archive]](/icons/banner/archiveACT.gif)
![[Search]](/icons/banner/searchACT.gif)
![[Contents]](/icons/banner/tocACT.gif)
Previous Article | Next Article 
Molecular and Cellular Biology, August 2000, p. 6147-6158, Vol. 20, No. 16
0270-7306/00/$04.00+0
Copyright © 2000, American Society for Microbiology. All rights reserved.
Functional Collaboration between Different
Cyclin-Dependent Kinase Inhibitors Suppresses Tumor Growth with
Distinct Tissue Specificity
David S.
Franklin,1,
Virginia L.
Godfrey,2
Deborah A.
O'Brien,1,3,5
Chuxia
Deng,6 and
Yue
Xiong1,4,5,*
Lineberger Comprehensive Cancer
Center,1 Department of Pathology and
Laboratory Medicine,2 Departments of
Cell Biology and Anatomy and Pediatrics,3
Department of Biochemistry and
Biophysics,4 and Program in
Molecular Biology and Biotechnology,5 University
of North Carolina at Chapel Hill, Chapel Hill, North Carolina
27599-7295, and Laboratory of Biochemistry and Metabolism,6
National Institute of Diabetes, Digestive and Kidney Diseases,
National Institutes of Health, Bethesda, Maryland 20892
Received 7 January 2000/Returned for modification 6 March
2000/Accepted 12 May 2000
 |
ABSTRACT |
The presence of two families of seven distinct mammalian
cyclin-dependent kinase (CDK) inhibitor genes is thought to mediate the
complexity of connecting a variety of cellular processes to the cell
cycle control pathway. The distinct pattern of tissue expression of CDK
inhibitor genes suggests that they may function as tumor suppressors
with different tissue specificities. To test this hypothesis, we have
characterized two strains of double mutant mice lacking either
p18INK4c and p27KIP1 or p18INK4c
and p21CIP1/WAF1. Loss of both p18 and p27 function
resulted in the spontaneous development by 3 months of age of at least
eight different types of hyperplastic tissues and/or tumors in the
pituitary, adrenals, thyroid, parathyroid, testes, pancreas, duodenum,
and stomach. Six of these hyperplastic tissues and tumors were in
endocrine organs, and several types of tumors routinely developed
within the same animal, a phenotype reminiscent of that seen in
combined human multiple endocrine neoplasia syndromes. The p18-p21
double null mice, on the other hand, developed pituitary adenomas,
multifocal gastric neuroendocrine hyperplasia, and lung
bronchioalveolar tumors later in life. G1 CDK2 and CDK4
kinase activities were increased in both normal and neoplastic tissues
derived from mice lacking individual CDK inhibitors and were
synergistically stimulated by the simultaneous loss of two CDK
inhibitors. This indicates that an increase in G1 CDK
kinase activity is a critical step during but is not sufficient for
tumor growth. Our results suggest that functional collaborations
between distinct CDK inhibitor genes are tissue specific and confer yet
another level of regulation in cell growth control and tumor suppression.
| |
INTRODUCTION |
More than a dozen tumor suppressor
genes have been identified thus far by virtue of their genetic
mutations in human cancers. Some appear to function in a specific cell
type, such as BRCA1 and BRCA2 in breast and ovarian cancer, Smad4
(Dpc4), APC, and Smad2 in colon cancer, and Menin in type 1 multiple
endocrine neoplasia (MEN1). Other tumor suppressors, notably p53 and
Rb, are mutated in a wide range of tumor types, indicating a more general function in tumor suppression (21, 32, 36).
Conceptually, genes that negatively regulate the growth-suppressing
activity of either p53 or pRb may be proto-oncogenes, as exemplified by the observation that MDM2 (26) and cyclin D1
(20), negative regulators of p53 and pRb, respectively, are
frequently activated in human cancers and promote tumor growth when
targeted for transgenic expression in mouse mammary tissues (22,
35). Likewise, genes that function to activate or to retain the
growth suppression activity of either p53 or pRb are candidate tumor
suppressors (32, 33). Indeed, the ARF-INK4a locus, the
second most frequently disrupted locus in human cancers next to p53
(17, 25), encodes two separate proteins, ARF and
p16INK4a, that positively regulate p53 and Rb, respectively
(33). The high frequency of genetic alterations and the
often nonoverlapping mutational pattern among the genes within each of
these two pathways have led to the notion that functional inactivation
of both the p53 and pRb pathways may be necessary for the development
of different types of cancer.
Two families of cyclin-dependent kinase (CDK) inhibitors, totaling
seven genes, have been identified in mammalian cells. Their similar
biochemical activity in blocking CDK enzymes and maintaining the
growth-suppressive activity of Rb predict a tumor suppression function
for CDK inhibitor genes, yet only the p16INK4a gene has
been directly linked to tumor growth by genetic alterations found in
human cancers (17, 25) and by the early development of
spontaneous tumors in mice lacking p16 (31). Neither
mutational analysis in human tumors nor phenotypic examination of
genetically targeted mice lacking any of the other individual CDK
inhibitor genes has provided strong evidence for a direct role for any
of the other CDK inhibitors as tumor suppressors. Mice lacking
p21CIP1/WAF1 (6),
p27KIP1 (10, 19, 23),
p57KIP2 (40, 41), or
p18INK4c (11) do not develop
spontaneous tumors at an early age. However, potential tumor
suppression functions were suggested by the observations that cells
lacking p21 are defective in a DNA damage-induced, p53-mediated
G1 checkpoint (3, 6), that mice lacking either p18 or p27 slowly develop intermediate-lobe pituitary tumors later in
life (10, 11, 19, 23), that p27 heterozygous mice have a
higher tumor incidence when challenged with 
-irradiation
(9), and that Rb+/
-p27/ mice
developed more aggressive pituitary adenoma and thyroid C-cell
carcinomas than the Rb+/ mice (27). The lack
of more widely spread tumors in these single-knockout mice may, in
part, be due to a redundant or overlapping function for many of these
CDK inhibitors in specific tissues. To test the possibility that
different CDK inhibitor genes may functionally collaborate to suppress
tumor growth with different tissue specificities, we have characterized
the tumorigenesis of two strains of double mutant mice lacking either
p18 and p27 or p18 and p21.
| |
MATERIALS AND METHODS |
Creation of double null mice.
Genetic disruptions of the p18
(11), p21 (6), and p27 (19) loci have
been previously described. Mice deficient for both p18 and p21 were
created by mating p18/ and p21/ mice
(6). The resulting F1
p18+/-p21+/ mice were crossed to create the
double null genotype. The creation of the
p18/-p27/ strain as well as the
intermediate "3/4" mutant strains
(p18+/-p27/ and
p18/-p27+/) have been previously
described (11). All of the p18-p21 and p18-p27 genotypes
(wild type, p18/, p21/,
p27/, p18/-p21/, and
p18/-p27/) are on a mixed C57BL/6-129
genetic background. For each genotype, intercrosses have been carried
out to F9 or F10 generations without any
alteration of the observed phenotypes resulting from genetic background
effects. All genotypes were confirmed by Southern blot analysis
(6, 11, 19).
Anatomic and histologic analysis.
Many animals were
sacrificed and subjected to complete necropsy at the first indication
of morbidity (weight loss, dehydration, ataxia, or failure to thrive).
Other animals (primarily wild-type or single null animals) were
sacrificed as age-matched controls. Several animals were sacrificed
between 12 and 17 months of age to analyze age-related tumor
progression. Body and organ weights were measured for every animal
during necropsy. For histological analysis, tissues were fixed and
processed as previously described (11). Testes were fixed in
Bouins' solution.
Antibodies and immunochemistry.
Antisera for p18, p21, p27,
CDK2, CDK4, and CDK6 and procedures for immunoprecipitations,
immunoblotting, and kinase assays have been previously described
(11, 12, 16, 28, 38). Protein lysate concentrations were
determined by the Bradford assay and equalized for each experiment. All
Western analysis and kinase assays were performed at least twice using
independent sets of 3-month-old tissue lysates. Equal loading of
lysates was further verified by tubulin (Neomarkers, Freemont, Calif.)
Western blot analysis. Procedures for calcitonin (Dako Corporation,
Carpinteria, Calif.) immunohistochemistry were carried out according to
the manufacturer's suggested protocol. Leydig cells were selectively immunostained with a monoclonal antibody, LC-6H6 (15), which recognizes an antigen on the surface of Leydig cells, or with an
antibody against estrogen sulfotransferase (34), a Leydig cell cytosolic enzyme. Immunostaining was visualized using the avidin-biotin-immunoperoxidase method with diaminobenzidine as the
substrate (Vectastain ABC; Vector Laboratories, Burlingame, Calif.).
Calcitonin and Leydig-specific immunostaining was performed on tissue
sections from several animals of all genotypes.
Kinase assays.
Kinase assays have been previously described
(11). Briefly, cell lysate was prepared in ice-cold NP-40
lysis buffer from different tissues with different genotypes and
precipitated with a specific antibody for 2 h at 4°C with
rotation. Five micrograms of affinity-purified anti-mouse CDK4 antibody
or 1 µl of crude anti-CDK2 serum was used to immunoprecipitate 2 mg
of cell lysate. Protein A-agarose beads were added and incubated for
1 h with rotation at 4°C to precipitate immunoglobulin. The
beads were washed twice with NP-40 lysis buffer and once in kinase
assay buffer. The washed beads were resuspended in 25 µl of kinase
assay buffer containing 5 µCi of [-32P]ATP and 2 µg of GST-pRbC137 substrate (a fusion protein of
glutathione-S-transferase [GST] and the C-terminal 137 amino acids of pRb) for CDK4 or 4 µg of histone H1 substrate
(Boehringer Mannheim) for CDK2 kinase assays. The reaction mix was
incubated at 30°C for 30 min, and the reaction was terminated by
adding 20 µl of 2× loading dye (100 mM Tris-HCl [pH 6.8], 4%
sodium dodecyl sulfate [SDS], 20% glycerol, 0.2% bromophenol blue,
0.2 M dithiothreitol). Ten microliters of CDK2 or CDK4 reaction samples
was resolved by SDS-15% polyacrylamide gel electrophoresis (PAGE). To
verify equal loading, the gel was stained with Coomassie blue to
visualize immunoglobulin and substrate proteins before it was dried and
exposed to X-ray film. Following exposure to X-ray film, the dried gels
were exposed to phosphorimaging plates. 32P incorporation
was quantitated on a PhosphorImager.
| |
RESULTS |
Generation of p18-p27 and p18-p21 double null mice.
We
previously found that the progression of pituitary tumors was greatly
accelerated by the simultaneous loss of both p18 and p27 genes and that
double null mice invariably died by 3 months of age (11). To
determine whether p18 and p27 may also function in suppressing tumor
growth in other tissues, we carried out a more detailed histological
analysis in the p18-p27 double null mice (n = 33),
especially in tissues where loss of either p18 or p27 had previously
been shown to cause detectable abnormal growth. To determine the
specificity of functional collaboration between different CDK inhibitor
genes, we also generated double mutant mice lacking both p18 and p21
genes and examined both the incidence and spectrum of tumor development
(n = 18). Because the accelerated mortality of p18-p27
double null mice might potentially exclude the detection of
later-developing tumors, we also generated mice retaining one allele of
either p18 or p27 (p18/-p27+/ or
p18+/-p27/; hereafter referred to as 3/4
mutant mice). The 3/4 mutant animals have an extended life span
(average, 9 months) beyond that of double null mice, thereby allowing
examination for later-developing tumor growth. The p18-p21 double null
mice can live beyond 14 months without detectable increase in
mortality, and thus no 3/4 mutant p18-p21 mice were analyzed.
p18-p27 mutant mice were generated by crossing p18/
mice with mice disrupted in the cyclin-CDK inhibition domain of p27
(11, 19). Mice lacking p18 and p21 genes were generated by
crossing p18/ mice with mice deleted for p21 exon 2, which removes 90% of the p21 coding sequence (6). Various
genotypes were verified by Southern blot analysis (Fig.
1) and PCR (data not shown). One of the
characteristic phenotypes of both p18- and p27-deficient mice is the
development of gigantism and widespread organomegaly (10, 11, 19,
23). Many of these organs (e.g., adrenal, thyroid, and testis)
from p18-p27 double mutant mice displayed an even greater
disproportionate organomegaly (data not shown). Like the mating between
p18+/ and p27+/ mice (11),
mating between p18+/ and p21+/ mice
produced all genotypes at the anticipated Mendelian ratios, indicating
that embryos with partial or complete loss of p18 and p21 functions are
viable (data not shown). Thirty
p18/-p21/ mice monitored between 5 and
14 months of age appeared developmentally normal and did not exhibit
any apparent enhancement of the p18/ gigantism and
organomegaly phenotypes (data not shown). In distinct contrast to
p18/-p27/ mice, which invariably died
at between 3 and 4 months of age, p18/-p21/ mice demonstrated no
decreased survival. These results suggest that p18 functionally
collaborates with p27 but not with p21 in controlling organ size and
body weight.
View larger version (25K):
[in this window]
[in a new window]
|
FIG. 1.
Genotypic analysis of CDK inhibitor genes. Southern blot
analysis of p18, p21, and p27 loci. Genomic DNA from the indicated
genotypes was digested with EcoRV (p18), BglII
(p21), or EcoRI (p27) and hybridized with the indicated DNA
probes. The wild-type (WT) and targeted null (Mut.) alleles are
indicated.
|
|
Different spectra of tumor development in p18-p27 and p18-p21
double null mice.
In addition to the previously described
pituitary adenoma (11), the p18-p27 mutant mice, including
both double null and 3/4 mutant animals, frequently developed
hyperplasia, adenomas, and/or carcinomas in seven other tissues: the
adrenal, thyroid, and parathyroid glands, testis, stomach, duodenum,
and pancreas (Table 1). With the
exception of pituitary adenomas, these tumors were rarely seen in
either single null strain. Invariably,
p18/-p27/ animals died around the age
of 3 months and the 3/4 mutant animals died at an average age of about
9 months from massive pituitary neoplasms. These results further
confirm the functional redundancy between p18 and p27 genes and
revealed a much broader function of these two CDK inhibitor genes in
tumor suppression than previously recognized.
Examination of p18-p21 double null mice (n = 18)
revealed a different and more confined spectrum of tumor growth (Table
2). Except for one animal, all of the
p18-p21 double null mice developed pituitary pathology (hyperplasia or
adenomas) by the age of 12 to 13 months (Table 2). At this age, about
half of the p18/ mice showed pituitary hyperplasia and
the remainder had pituitary adenomas. Although only a few p21 null mice
developed slight pituitary hyperplasia, loss of p21 function
accelerated the incidence and progression of pituitary pathology
initiated by the loss of p18. It should be noted that the loss of p18
and p21, like the loss of p18 and p27, enhanced the pituitary tumor
phenotype in this tissue. Therefore, in this case, the accelerated
pituitary tumor growth represents a functional collaboration between
p18 and p21 rather than distinct tissue specificity from that of p18
and p27. Tissue specificity of tumor suppression is more apparent when comparing the p18-p27 double mutant mice with p18-p21 double mutant mice in other tissues, where clear tissue differences are observed (e.g., thyroid or duodenum in p18-p27 mice and lung in p18-p21 mice).
Nearly all p18-p21 double null mice developed multifocal gastric
neuroendocrine cell hyperplasia. Gastric neuroendocrine cell hyperplasia was detected in a few p21/ mice and was not
found in any p18/ mice. At a lower incidence, p18-p21
double null mice developed lung bronchioalveolar adenoma or carcinoma.
Two p18-p21 double null mice also exhibited hepatic nodular
hyperplasia. These results provide the first genetic evidence for a
tumor suppression function of p21. In six tissues where p18-p27 double
mutant mice developed frequent hyperplasia or tumors (adrenal, thyroid,
testis, parathyroid, pancreas, and duodenum), there is no evidence that
p18-p21 losses of function collaborate in tumor development (Table 2).
Conversely, p18 and p27 do not seem to collaborate significantly in the
formation of gastric neuroendocrine cell hyperplasia, lung tumors, or
liver hyperplasia. The distinct tumor spectra of p18-p27 and p18-p21 double null mice indicate that functional collaboration between different CDK inhibitors suppresses tumor growth with different tissue specificities.
p18-p27 double mutant mice develop multiple types of endocrine
tumors.
Notably, five of the seven tumors that occur in p18-p27
double mutant mice (pituitary adenoma, adrenal pheochromocytoma,
thyroid C-cell adenoma, parathyroid adenoma, and testis Leydig cell
adenoma), as well as the pancreatic islet cell hyperplasia, all involve endocrine tissues (Table 1 and Fig.
2). This affected
tissue spectrum overlaps that seen in patients with type I (MEN1,
pituitary, pancreatic, and parathyroid tumors) or type II (MEN2,
thyroid C-cell carcinomas, pheochromocytomas, and parathyroid tumors) MEN syndromes. One characteristic feature of MEN syndromes is the
concurrent development of multiple endocrine tumors in the same
patient. All of the p18/-p27/ and most
of the 3/4 mutant animals developed multiple endocrine tumors. Most
double null animals possessed three or four tumors and had additional
hyperplastic tissues. In one p18/-p27/
animal, five simultaneous tumors were detected. The most frequently occurring hyperplastic tissues and/or tumors in
p18/-p27/ mice were the pituitary
adenoma and carcinoma (31 of 33, 94%), adrenal medullary hyperplasia
and pheochromocytoma (23 of 23, 100%), thyroid C-cell hyperplasia and
adenoma (14 of 16, 88%), and testis interstitial cell hyperplasia (10 of 10 males, 100%), all four developing with near or complete
penetrance (Table 1). Endocrine lesions found at a lower incidence were
parathyroid hyperplasia and adenomas (3 of 8, 38%,
p18/-p27+/ only) and pancreatic islet
cell hyperplasia (3 of 9 [33%] in p18/-p27+/ and 1 of 5 [20%] in double
null mice). While the p18-p27 mouse tumors may not exactly mimic those
seen in MEN patients (e.g., adenomas in mice instead of carcinomas in
humans), the affected cell types (e.g., thyroid C cells, adrenal
medullary cells, or pancreatic islet cells) and the spectrum of
simultaneous multiple endocrine tissues are strikingly similar. It
remains to be determined whether the development of mouse endocrine
tumors and human MEN syndrome have a similar molecular mechanism.
View larger version (64K):
[in this window]
[in a new window]
|
FIG. 2.
p18-p27 double mutant mice developed multiple
endocrine tumors. (A) Malignant pheochromocytoma in
p18/-p27/ adrenal glands. Adrenal
cross sections of 3-month-old mice were stained with hematoxylin and
eosin. (B) Thyroid C-cell phenotype. The thyroid glands were
immunostained for calcitonin. C-cell hyperplasia (CH) in
p18/-p27/ and C-cell adenomas (A)
in p18/-p27+/ mice are shown. (C)
Parathyroid phenotypes. Hematoxylin and eosin staining of parathyroid
hyperplasia (PH) and adenoma (PA) in a 7-month-old
p18/-p27+/ mouse. (D) Pancreas
phenotype. Hematoxylin and eosin staining of pancreatic islet cell
hyperplasia in a 9-month-old
p18/-p27+/ mouse. Several normal (N)
and three hyperplastic (ICH) islets are indicated. (E) Testis
interstitial cell phenotype. (Left) Interstitial cell adenoma in the
testis of a 9-month-old p18/-p27+/
mouse stained with hematoxylin and eosin. (Right) Interstitial cell
hyperplasia in the testis of a 9-month-old
p18/-p27+/ mouse immunostained with
monoclonal antibody LC-6H6, which recognizes a Leydig cell
constituent.
|
|
By gross examination, adrenal glands of
p18/-p27/ animals were enlarged
compared with age-matched single null or wild-type animals (data not
shown). All p18/-p27/ mice developed
adrenal pathology (n = 23, Table 1). Two 3-month-old p18/-p27/ mice exhibited severe
medullary hyperplasia, while adrenals from the remaining 23 p18/-p27/ mice contained bilateral
pheochromocytomas (91%) (Fig. 2A). These tumors were detected as early
as 1 month and invaded the adrenal cortex in some animals and, in one
instance, metastasized to a distal pelvic nerve (data not shown). No
adrenal pathology was detected in age-matched 3-month-old wild-type
(n = 7), p18/ (n = 8),
or p27/ (n = 9) animals (data not
shown). Although adrenal pathology was detected in some
p18/ or p27/ animals, the age of
detection was never less than 6 months for p27/ mice or
12 months for p18/ mice. The earlier age of onset and
greater incidence of adrenal phenotypes in p18-p27 double mutants
clearly demonstrate that p18 and p27 functionally collaborate in
adrenal tumorigenesis.
The thyroid glands of p18/-p27/ animals
appeared enlarged compared with age-matched single null or wild-type
animals (data not shown). Thyroids from most of the 3-month-old
p18/-p27/ animals revealed C-cell
hyperplasia (13 of 16, 81%) (Fig. 2B, right). In addition, three 3/4
mutant mice (3 of 28, 11%) and one
p18/-p27/ mouse (1 of 16, 6%)
developed thyroid C-cell adenomas (Fig. 2B, left). The C-cell origin of
these lesions was verified by immunohistochemistry using an antibody to
calcitonin, a hormone produced in thyroid C cells (Fig. 2B). Thyroid
follicular hyperplasia was occasionally seen in
p18/-p27+/ mice (data not shown). Thyroid
glands of seven age-matched 3-month-old wild-type, eight
p18/, and nine p27/ animals were
histologically normal (data not shown). The p18/ and
p27/ mice that developed thyroid pathology were all
over 12 months (p18/) or 6 months
(p27/) of age.
Parathyroid hyperplasia and/or adenomas were detected in 38% (3 of 8)
of the p18/-p27+/ mice, a lower incidence
than seen in the other endocrine tissues (Table 1, Fig. 2C). The
average age at detection was 6.8 months. The absence of this phenotype
in p18/-p27/ mice (0 of 6) may reflect
the premature mortality (3 months) of this genotype, while the
longer-living p18/-p27+/ mice
(approximately 9 months) survive long enough to develop this phenotype.
In addition, 33% of the p18/-p27+/ (3 of
9) and 20% of the p18/-p27/ (1 of 5)
mice displayed pancreatic islet cell hyperplasia (Fig. 2D). The age at
onset of islet cell hyperplasia was 8.9 months for
p18/-p27+/ mice and 2.9 months for
p18/-p27/ mice. The lower incidence of
islet cell hyperplasia in p18/-p27/
mice again may reflect premature mortality and the time required to
develop pancreatic islet cell hyperplasia. With the exception of one
22-month-old wild-type mouse, the pancreatic phenotype was not detected
in any wild-type (16 of 17), p18/ (0 of 15),
p27/ (0 of 14), or
p18+/-p27/ (0 of 3) mice. While one
wild-type and one p18/-p27/ each
developed pancreatic islet cell hyperplasia, the increased rate of
incidence (6% versus 20%, respectively) and the early age at onset
(22 months versus 3 months, respectively) in the double mutant genotype
clearly validate the double null result.
In addition to the endocrine tumors commonly seen in MEN patients, the
p18-p27 double mutant mice also developed hyperplasia and tumor
phenotypes in another endocrine organ, the testis. Gross examination of
p18-p27 double null males revealed a marked increase in testicular size
by 3 months of age. While p18/ and p27/
males also have enlarged testes (10, 11, 23), this phenotype was more pronounced in p18/-p27/ males.
Mean testis weights of mice (± standard deviation) between 1.5 and 3 months of age varied with genotype, from 0.106 ± 0.004 mg in
wild-type to 0.134 ± 0.009 mg in p18/, 0.139 ± 0.023 mg in p27/, and 0.206 ± 0.020 mg in
p18/-p27/ mice. Testes from
p18/ males (40 of 41, 98%) displayed interstitial cell
hyperplasia and/or adenoma not seen in wild-type (n = 27) or p27/ (n = 12) males (Table
1). All p18-p27 double null mice (10 of 10) and all but two of the
p18/-p27+/ males (15 of 17, 88%) also developed interstitial cell hyperplasia. The severity of
interstitial cell hyperplasia was always greater in
p18/-p27/ and
p18/-p27+/ mice than in age-matched
p18/ mice (data not shown). The hyperplastic
interstitial cells immunostained with antibodies that recognize Leydig
cell constituents (Fig. 2E, right), including monoclonal antibody
LC-6H6 (15) and an antibody to estrogen sulfotransferase
(34). Compared with p18/ males, the
immunostaining intensity in
p18/-p27/ and
p18/-p27+/ mice was reduced in
regions where interstitial cells were densely packed with little
cytoplasm, morphological changes consistent with impaired pituitary
function (2). Adenomas were detected in the interstitial
cell compartment in six of the
p18/-p27+/ males between 6 and 9 months of age (6 of 17, 35%). This tumor was detected in only one
p18/ male that was 16 months old (1 of 41, 2%). One
p18/-p27+/ male had an adenoma that
filled ~80% of the volume of one testis (Fig. 2E, left), with a
smaller adenoma present in the other testis. Again, the absence of
interstitial cell adenomas in p18-p27 double null mice may result
from their early mortality. Clearly, even the absence of one p27 allele
is sufficient to accelerate the Leydig cell pathology that results from
the loss of p18 function.
Gastrointestinal tumors and hyperplasia in p18-p27 mice.
p18/-p27/ mutant mice developed two
additional nonendocrine lesions: squamous cell hyperplasia and
papillomas in the forestomach, and duodenal hyperplasia and villous
adenomas (Table 1 and Fig. 3). The
squamous cell lesions were focal and always located near the esophageal
inlet. The mouse forestomach is normally lined with squamous mucosa,
and the hyperplasia observed in
p18/-p27/ and 3/4 null animals is
not a metaplastic occurrence. The duodenal lesions only appeared just
distal to the pylorus in conjunction with proliferation of the
Brunner's glands. Incidences of the gastrointestinal lesions in
p18/-p27/ animals were 43% (6 of
14) for squamous cell hyperplasia, 21% (3 of 14) for forestomach
papillomas, 29% (5 of 17) for duodenal hyperplasia, and 59% (10 of
17) for duodenal villous adenomas. Neither tumors nor hyperplasia was
detected in these tissues in any wild-type (n = 59) or
p18/ (n = 39) animals. The incidence of
these phenotypes in p27/ (n = 24)
animals was significantly less than in the
p18/-p27/ mice.
View larger version (53K):
[in this window]
[in a new window]
|
FIG. 3.
Nonendocrine tumors in p18-p27 double mutant mice.
(A) Gross appearance of stomach squamous cell papillomas (SP) and
duodenal villous adenomas (VA) in 3-month-old
p18/-p27/ mouse. Bar, 5 mm. (B)
Hematoxylin and eosin staining of stomach squamous cell papillomas (SP)
and duodenal villous adenomas (VA) in 3-month-old
p18/-p27/ mouse. (C) Stomach
squamous cell papilloma from panel B. (D) Duodenal villous adenoma from
panel B.
|
|
p18-p21 double null mice develop pathology distinct from
p18-p27 mice.
The p18-p21 double null mice developed different
spectra of tumors and hyperplastic tissues than p18-p27-deficient mice.
The incidence and progression of pituitary adenoma were both greatly increased in p18-p21 double null mice compared with
p18/ mice. By 1 year of age, 44% of
p18/ mice developed pituitary hyperplasia and 44%
exhibited adenomas or carcinomas (Table 2). However, by this same age,
all but one of the p18-p21 double null mice developed either pituitary
hyperplasia (1 of 17, 6%) or adenomas (15 of 17, 88%), with almost
all animals now developing pituitary adenomas. Therefore, in a
p18/ background, loss of p21 function greatly enhanced
the incidence of pituitary pathology as well as the progression from
hyperplasia to adenoma. As in p18- or p27-deficient mice, the pituitary
adenomas that developed in the p18-p21 mice originated from the
intermediate lobe (as determined by histological examination; data not
shown). It should be noted that pituitary tumors from 1-year-old
p18/ or p18/-p21/ mice
were histologically indistinguishable and these mice did not seem to
possess different mortality rates (data not shown). This is in contrast
to the short 3-month life span of p18-p27 double null mice.
Of 18 p18-p21 double null mice, 89% developed multifocal gastric
neuroendocrine cell hyperplasia (Fig.
4A). In one instance, nodules of
neuroendocrine cells extended superficially into the subjacent muscular
layers of the stomach (Fig. 4B). This could be evidence of local
invasion or merely hyperplasia of neuroendocrine cells in a naturally
occurring mucosal diverticulum. While local invasion is one of the few
histologic indications of malignancy for neuroendocrine cell tumors
(carcinoids), the small size and multifocal nature of the lesion in
this mouse are more consistent with hyperplasia than carcinoid tumor.
Two of 17 p18-p21 mice had bronchioalveolar adenoma (data not shown),
and 1 of 17 had a bronchioalveolar carcinoma (Fig. 4C). The presence of
lung tumors in 3 of 17 p18-p21 double null mice (18%) combined with
the lack of detectable tumors in the lungs of any wild-type or single
null animals suggests that these tumors result from loss of p18-p21 functions rather than from normal aging processes. Two p18-p21 double
null mice developed hepatic nodular hyperplasia (2 of 16; data not
shown), a finding that is not uncommon in older mice of this genetic
background (C57BL/6). We therefore cannot dismiss the possibility that
this may be an incidental aging lesion and not entirely due to the
p18-p21 mutations. Except for pituitary adenomas in
p18/ mice, the tumor phenotypes seen in p18-p21 double
null mice were rarely detected in wild-type or either single p18- or
p21-null mice. In the six tissues where p18-p27 double mutant mice
developed hyperplasia and/or tumors (adrenal, thyroid, testis,
parathyroid, pancreas, and duodenum), there is no evidence that p18-p21
loss of function collaborates in tumor development (Table 2). Also, there is no indication that p18 and p27 significantly collaborate in
the formation of gastric neuroendocrine cell hyperplasia, lung tumors,
or liver pathology. The distinct spectra of tumors that form in
p18-p27 or p18-p21 double mice indicate that functional collaboration between different CDK inhibitors suppresses tumor growth
with different tissue specificities.
View larger version (67K):
[in this window]
[in a new window]
|
FIG. 4.
Tumor phenotypes in p18-p21 double null mice. (A)
Hematoxylin and eosin staining of multifocal gastric neuroendocrine
cell hyperplasia in 12-month-old
p18/-p21/ mouse. (B) Hematoxylin and
eosin staining of gastric neuroendocrine cell hyperplasia in a
12-month-old p18/-p21/ mouse showing
extension of neuroendocrine cells into the muscular layer of the
stomach. (C) Hematoxylin and eosin staining of lung bronchioalveolar
carcinoma in 12-month-old p18/-p21/
mouse.
|
|
Gene dosage-dependent tumor suppression in p18-p27 mice.
Comparison of mice with five different genotypes
the
p18/ and p27/ single nulls, the two 3/4
mutants, and the p18/-p27/ double
nullrevealed evidence for gene dosage-dependent tumor suppression by
p18 and p27. In many tissues, both 3/4 mutant genotypes exhibited an
intermediate phenotype of tumor growth, more aggressive than in either
single null animal and in most cases less severe than in the double
null mice. This was manifested in both the incidence of tumor formation
and the rate of tumor growth. First, the incidence of tumor formation
in many tissues of 3/4 null animals was decreased compared to double
null mice but was higher compared to either single null mouse strain
(Table 1). For example, pheochromocytomas occurred in 95% of
p18/-p27/ mice but only in 16% of
the p18/-p27+/ mice and 60% of the
p18+/-p27/ mice despite the longer
life span of 3/4 mutant animals. Forestomach hyperplasia and adenomas
were detected in 62% of the double null mice, with hyperplasia
detected in only 5% of the
p18/-p27+/ mice and 40% of the
p18+/-p27/ mice. Duodenal hyperplasia
and adenomas were detected in 88% of the double null mice, with
hyperplasia only detected in 13% of the
p18/-p27+/ mice. Gastrointestinal
tumors were not seen in 3/4 mutant animals (Table 1). Other than
pituitary adenomas and with the previously noted infrequent exceptions,
tumors either did not develop or were rarely seen in wild-type or
either single null genotypes (Table 1). In some tissues, the 3/4 null
mice developed tumors that were not detected in the double null mice.
As previously mentioned, we interpret this as a result of the early
mortality of the double null animals. Because the 3/4 null mice live
longer (9 months) than the double null mice (3 months), perhaps they have sufficient time for further initiation and/or progression of tumor
formation. Second, the rate of tumor growth in some tissues was slower
in 3/4 mutant animals than in p18/-p27/
mice, but was accelerated compared with either single null mouse strain. The average age at pituitary tumor formation was 2.7 months in
p18/-p27/ mice and 7.4 months in
3/4 mutant mice. Accordingly, the average life span increased from 3 months in the double null mice to an average of 9 months in both types
of 3/4 mice. In comparison, p18/ mice (9 of 31)
developed pituitary tumors at an average age of 11 months, while
p27/ mice (4 of 16) developed this tumor at an average
age of 7.6 months, and both single null genotypes routinely live well
beyond 1 year. Similarly, the pheochromocytoma developed by 2.6 months in p18/-p27/ mice and 7.3 months in
3/4 mutant mice.
Conceivably, enhancement of neoplastic transformation in 3/4 animals
compared with single null mice could be caused by an increased
incidence of loss of heterozygosity that removes the remaining
wild-type allele. Three lines of evidence argue against this
possibility. First, mice heterozygous for p27 display intermediate phenotypes in both body size and organ weights without the loss of
heterozygosity (10, 19, 23). Second, both the p18 and p27
genes appear to be quite stable in the genome and are rarely mutated or
deleted. Third, intermediate phenotypes were seen in both types of 3/4
genotypes and in many different organs (Table 1). Consistent with the
notion that functional loss of the remaining wild-type allele is not
the primary cause for the augmented phenotypes seen in the 3/4 mice, we
have found no evidence for the loss of the remaining p27 allele in
multiple samples from three different types of tumors (pituitary,
adrenal, and thyroid) derived from p18/-p27+/ mice by Southern blot
analysis (data not shown).
Tissue expression of CDK inhibitors.
To confirm that p18, p21,
and p27 are normally expressed in the organs where pathology developed
in the mutant mice, we examined various wild-type tissues by Western
blot analysis for expression of p18, p21, and p27 proteins as well as
their two common targets, CDK4 and CDK6 (Fig.
5). CDK4 and CDK6 were expressed in all
of the tissues examined. Expression of p18 was highest in the
intestines, kidney, liver, pituitary, skeletal muscle, stomach, testes,
and thyroid, although expression was detectable in all of the tissues examined. p21 was expressed in the adrenal, intestines, lung, pituitary, spleen, stomach, testes, thymus, and thyroid, with low
expression detected in the heart and liver. p27 was expressed in all of
the tissues examined, with highest expression in the adrenal gland,
heart, pituitary, and skeletal muscle. Importantly, expression of the
CDK inhibitors was consistent with the tissues where p18, p21, or p27
loss of function caused the development of hyperplastic or tumorigenic
phenotypes. For example, p18-p21 double null mice develop pituitary,
gastric neuroendocrine cell, and lung tumors. Both proteins are
expressed in the pituitary, stomach, and lung. Mice lacking both p18
and p27 exhibit tumor growth in the adrenal, intestines, pituitary,
stomach, testes, and thyroid (including parathyroid), all tissues where
p18 and p27 are readily detectable. It should be pointed out that in
some tissues, the expression of these genes is clearly detected (e.g., p21 in adrenal and p27 in lung), yet loss of CDK inhibitor function does not result in obvious tumor phenotypes. This suggests that p18,
p21, or p27 is not a rate-limiting factor for tumor formation in any
tissue where it is expressed.
View larger version (56K):
[in this window]
[in a new window]
|
FIG. 5.
Expression of CDK and CDK inhibitors in mouse tissues.
Total cell lysates were prepared from the indicated wild-type tissues.
Expression patterns of CDK4, CDK6, p27, p21, p18, and tubulin were
determined by Western blot analysis. Tubulin expression was used to
demonstrate equal loading of protein lysates.
|
|
Simultaneous loss of p18 and p27 increases G1 CDK
kinase activity.
To provide a plausible molecular mechanism for
the development of the tumorigenic phenotype in these CDK
inhibitor-deficient mice, we compared the kinase activity of two
G1 CDKs, CDK2 and CDK4, in mice of various genotypes using
either histone H1 (for CDK2) or GST-pRb (for CDK4) as substrates. We
chose four tissues, adrenal gland, testis, lung, and skeletal muscle,
from 3-month-old animals for this analysis. At this age, hyperplastic
or tumor phenotypes were consistently detected in
p18/-p27/ adrenal gland and testis,
but not in lung and skeletal muscle tissue or in mice of several other
genotypes (Tables 1 and 2). Such analysis would potentially allow us to
determine whether any change in kinase activity precedes or follows
cell transformation.
Compared with wild-type adrenal tissue, H1 kinase activity of CDK2
remained essentially unchanged in p21/ and
p18/ lysate, increased about 2.4-fold in
p27/ lysate (varying between 1.5- and 3-fold in
different experiments), 1.7-fold in
p18/-p21/ lysate, and 6-fold in
p18/-p27/ lysate (Fig.
6A). Rb kinase activity of CDK4 increased
in all five mutant adrenal tissues: 2-fold in
p18/ lysate, 4-fold in p21/
lysate, 6-fold in p18/-p21/ lysate,
6-fold in p27/ lysate, and 12-fold in
p18/-p27/ lysate. The testis CDK2
kinase activity increased two- to fourfold in p18/
lysate, four- to sixfold in p27/ lysate, and
synergistically increased about eightfold in
p18/-p27/ testis lysate (Fig. 6B).
Loss of p21 had little effect on testis CDK2 kinase activity as seen in
either p21/ or
p21/-p18/ lysate. The testis Rb kinase
activity of CDK4 increased two- to threefold in p18/
lysates, twofold in p21/ lysates, four- to sixfold in
p27/ lysates, fourfold in
p18/-p21/ lysates, and six- to
sevenfold in p18/-p27/ lysates.
These results suggest that associated with the development of
hyperplastic growth in both the adrenal gland and testis, the activity
of both CDK2 and CDK4 kinase was increased by the loss of p18 or p27,
and the simultaneous loss of both genes synergistically stimulated the
activity of both CDKs.
View larger version (47K):
[in this window]
[in a new window]
|
FIG. 6.
Loss of CDK inhibitors increases CDK kinase activity.
Cell lysates were prepared from adrenal (A), testis (B), lung (C), and
skeletal (sk.) muscle (D) obtained from 3-month-old mice of the
indicated genotypes. Equal amounts of lysate were immunoprecipitated
(IP) with either anti-CDK2 or anti-CDK4 antiserum as indicated.
Competing peptides were added to the immunoprecipitations to confirm
antibody specificity. Immunocomplexes were assayed for kinase activity
using histone H1 (for CDK2) or GST-pRbC137 (for CDK4) as
substrates. Autoradiographs of the kinase assays are shown on the left.
Incorporation of 32P was quantitated, and relative fold
change in kinase activity is graphed on the right.
|
|
To determine whether the increase in CDK activity is consequent to the
development of tumor pathology, we determined both CDK2 and CDK4 kinase
activity in two additional tissues, the lung and skeletal muscle, that
possessed no observable tumor phenotypes at 3 months of age. The lung
CDK2 activity was increased approximately twofold in
p21/, p27/,
p18/-p21/, and
p18/-p27/ genotypes with no
apparent synergistic effects. Lung CDK4 activity was unchanged in
p18/ or p21/ lysates and was increased
approximately threefold in p27/,
p18/-p21/ and
p18/-p27/ genotypes. In the
skeletal muscle, CDK2 activity was increased approximately twofold in
p18/ and p21/ lysates and approximately
four to sixfold in p27/,
p18/-p21/, and
p18/-p27/ lysates. The Rb kinase
activity of CDK4 was increased approximately 2-fold in
p18/ or p21/ lysates, 7- to 8-fold in
p27/ or p18/-p21/
lysates, and 12-fold in p18/-p27/ lysate.
These results provide a plausible biochemical basisincrease in
G1 CDK2 and CDK4 kinase activityfor the hyperplastic and tumor growth observed in the absence of p18, p21, and p27. Increase in
CDK2 and CDK4 kinase activity is not dependent on and most likely
precedes the development of tumor phenotypes. While an altered
proportion of cell types resulting from hyperplastic growth and/or
tumorigenesis could potentially contribute to the change in kinase
activity in double null tissues, this would not explain the increase in
kinase activity in single null tissues where the hyperplastic
phenotypes were not as pronounced or were absent (e.g., skeletal muscle
and lung). We therefore interpret the increased kinase activity in the
single and double null tissues to be a direct effect of the disruption
of p18, p21, and/or p27. In many cases, simultaneous loss of two CDK
inhibitor genes synergistically increased CDK activity. This is
consistent with accelerated tumorigenesis in double mutant mice and
suggests that the level of CDK kinase activity is a rate-limiting
factor for the tumor growth. These results also indicate that an
increase in G1 CDK activity alone is not sufficient to
initiate tumor growth in some tissues (e.g., lung and skeletal muscle)
and additional genetic changes are required. Finally, an increase in
CDK4 kinase activity in all four p21- and p27-deficient tissues,
adrenal, skeletal, lung, and testis, was somewhat unexpected, given the
recent report that loss of either p21 or p27 impaired the assembly of
cyclin D-CDK4 complex and decreased Rb kinase activity of CDK4 in in
vitro-cultured mouse embryo fibroblasts (5). This finding
raises the possibility that assertion of the role of p21 and p27 as
positive regulators of CDK4 may be dependent on cell types or in vivo
and in vitro kinase assay conditions.
| |
DISCUSSION |
In this report, we provided the first evidence that functional
collaborations between different CDK inhibitor genes suppress tumor
growth with distinct tissue specificities. Previously, collaboration between different CDK inhibitors has been thought to contribute to
differentiation of eye lens fiber cells (42) and muscle
(43), but not in general tissue tumor suppression. Our
results indicate that p18, p21, and p27 can indeed function together as
tumor suppressor genes. We have found that, except for the pituitary,
there is no overlap in the tumor spectrum between
p18/-p27/ and
p18/-p21/ mice. While the loss of both
p18 and p27 resulted in spontaneous development of multiple tumors,
predominantly in endocrine organs, p18-p21 double null mice developed
pituitary adenomas, gastric neuroendocrine cell hyperplasia, and lung
bronchioalveolar tumors. Therefore, with the exception of pituitary
tumors, the cooperative tumor suppressor activities of p18-p21 and
p18-p27 function with tissue specificity. p21 functions as the major
downstream target of p53 to cause G1 cell cycle arrest in
response to DNA damage (3, 6, 8), predicting a tumor
suppression function in vivo, yet mice lacking p21 are free of tumor
growth (6). Our results provide the first genetic evidence
supporting a role of p21 in tumor suppression and an
explanationfunctional redundancy with another CDK inhibitor genefor
the lack of tumor growth in p21 single mutant mice. It will be
interesting to determine the tumor incidence and spectra of mice
lacking p21 and other CDK inhibitor genes, especially the members of
the INK4 family.
Notably, five of the seven tumors in p18-p27 double mutant mice
(pituitary adenoma, adrenal pheochromocytoma, thyroid C-cell adenoma,
parathyroid adenoma, and testis interstitial cell adenoma), as well as
the pancreatic islet cell hyperplasia, develop in endocrine tissues.
This tumor spectrum overlaps in the same cell types as seen in human
patients with MEN syndromes (the combined MEN1 and MEN2 tumor spectra).
MEN syndromes refer to a group of diseases characterized by the
concurrent development of multiple endocrine tumors and are clinically
divided into two major subtypes, MEN1 (pituitary, parathyroid,
and pancreatic tumors) and MEN2 (medullary thyroid C-cell carcinomas,
pheochromocytomas, and parathyroid tumors), based on their unique
combinations of affected endocrine organs (4, 29). All five
endocrine organs that are affected in MEN patients exhibit neoplastic
or hyperplastic phenotypes in the p18-p27 mutant mice. In addition,
like MEN patients, all of the p18-p27 double null animals possessed
multiple endocrine tumors (Table 1). It should be pointed out, however,
that there are distinct differences between the p18-p27 mutant mice
and MEN patients, particularly with nonendocrine pathology. Additional MEN1 pathology includes foregut carcinoids, facial angiofibromas, and
multiple lipomatous tumors. MEN2B patients present additional pathology, including hyperplasia of the intrinsic autonomic ganglia in
the wall of the intestine, disorganized growth of peripheral nerve
axons in the lips, oral mucosa, and conjunctiva, and developmental musculoskeletal abnormalities (e.g., pes cavus, slipped femoral epiphyses, pectus excavatum, and bifid ribs). These abnormalities were
not detected in the p18-p27 mutant mice. Likewise, the hyperplasia and tumors of testis Leydig cells and gastrointestinal tumors detected
in the p18-p27 mice are not found in MEN patients. The molecular
mechanism underlying potential pathological similarities between
p18-p27 mutant mice and MEN patients is not clear. It is tempting
to speculate that p18 and p27 may interact functionally with the MEN1
tumor suppressor gene product, menin, and/or the c-RET
proto-oncogene, two genes consistently mutated in MEN1 and MEN2
patients, respectively. While human MEN syndromes can be clinically
separated into two major subtypes based on the affected endocrine
organs, a division of phenotypes cannot be correlated with different
p18-p27 genotypes. Further genetic and biochemical analysis is
necessary to establish a regulatory role, if any exists, for either
menin or c-Ret in controlling p18 and/or p27 function.
Like p18-p27 double mutant mice, Rb heterozygous mice developed
neoplastic phenotypes in multiple endocrine tissues, including the
pituitary, thyroid, parathyroid, and adrenal glands (24), providing genetic evidence that p18 and p27 genes function to suppress
tumor growth by regulating Rb's tumor suppression function. The
p18-p27 and Rb mutant mice exhibit a tumor spectrum completely different from that seen in mice lacking either p53 (7, 13, 30) or its upstream regulators ATM (1, 39) and ARF
(18), which develop predominantly lymphomas and soft tissue
sarcomas. The molecular basis for such dramatic differences in tumor
spectrum between mice lacking genes in the Rb and p53 pathways is not
yet understood but may be related to different roles of the Rb and p53
pathways in tumor suppression. p53 functions primarily as a checkpoint
gene to monitor genomic integrity, while the function of Rb is to
integrate mitogenic signals to determine whether a cell is to enter S
phase for another round of division or to arrest in G1. By
allowing accumulation of mutations, loss of p53 function would
significantly increase the chance of a cell to become tumorigenic, but
only if the cell is in the active division cycle. A stable cell cycle
arrest would effectively prevent tumor development even if the genomic
integrity checkpoint pathway were impaired. Many endocrine tissues are
slow growing or contain very few dividing cells and therefore may
become more susceptible to tumorigenesis by the inactivation of the Rb
pathway resulting from the loss of both p18 and p27. Mice heterozygous
for both p53 and Rb developed multiple endocrine tumors with a spectrum
similar to those observed in the p18-p27 mice, but more aggressive
in growth (14, 37). Together, these data support a greater
role for the Rb pathway (including p18, p27, and Rb) in suppressing
tumor initiation and for p53 in suppressing tumor progression.
| |
ACKNOWLEDGMENTS |
We thank Andrew Koff for providing p27 mutant mice, Masahiko
Negishi for providing antibody specific for estrogen sulfotransferase, and E. M. Eddy for the LC-6H6 monoclonal antibody. We are
particularly grateful to Tomo Ohta and Dell Yarbrough for helpful
discussions throughout the course of this work. We thank the Animal
Histopathology Core Lab for production of the histologic sections.
D.S.F. is a recipient of National Research Service Awards from the
NIH/NIAMS. Y.X. is a Pew Scholar in Biomedical Science and
recipient of an American Cancer Society Junior Faculty award. This
study was supported by National Institutes of Health grants HD26485 to
D.A.O., CA16086 to the UNC Lineberger Comprehensive Cancer Center and
V.L.G., and CA68377 to Y.X.
| |
FOOTNOTES |
*
Corresponding author. Mailing address: Lineberger
Comprehensive Cancer Center, University of North Carolina, Chapel Hill, NC 27599-7295. Phone: (919) 962-2142. Fax: (919) 966-8799. E-mail: yxiong{at}emailunc.edu
Top
Abstract
Introduction
