Oncogene (1997) 14, 2943 ± 2950
 1997 Stockton Press All rights reserved 0950 ± 9232/97 $12.00
ayk1, a novel mammalian gene related to Drosophila aurora centrosome
separation kinase, is speci®cally expressed during meiosis
Amiel Yanai, Eli Arama, Gillar Kil®n and Benny Motro
Department of Life Sciences, Bar-Ilan University, Ramat-Gan 52900, Israel
A novel murine gene, designated ayk1, which encodes a
putative serine/threonine kinase has been cloned and
characterized. The predicted catalytic domain of the
protein is highly similar to that of Drosophila aurora
(62.9% identity), and to that of Saccharomyces Ipl1
(49.4% identity). All three proteins also have very basic
calculated isoelectric points (higher than 10). aurora has
been recently shown to be crucial for centrosome
separation and chromosome segregation, while Ipl1 is
essential for yeast viability and accurate chromosome
segregation. The results of Northern analysis and in situ
RNA localization support a similar role for ayk1. The
gene is speci®cally expressed in meiotically active cells,
and during spermatogenesis, ayk1 transcripts accumulate
just before the ®rst meiotic division. Much lower levels
are found in mitotically active cells. We propose that
Ayk1, aurora and Ipl1 belong to a distinct new subfamily
of kinases. These results suggest that the pathways
controlling chromosome segregation are evolutionary
conserved, and that similar control mechanisms operate
in mitosis and meiosis.
Keywords: serine/threonine protein kinase; centrosome
separation; meiosis
Introduction
Accurate segregation of chromosomes in mitosis is
critical for cell viability and normal function, and
missegregation can lead to oncogenesis. Impaired
chromosome segregation during meiosis would results
in uneuploidy, the most common class of chromosome
abnormality found in human pregnancies (Hassold and
Jacob, 1984). Genome segregation in higher eukaryotes
is carried out by a bipolar spindle composed of two
centrosomes, a dynamic array of microtubules and
diverse associated proteins. The vertebrate centrosome
consists of a centriole pair and a pericentriolar
microtubule-organizing center that nucleates microtubule assembly, which has recently been shown to be
based on a g-tubulin ring (Zheng et al., 1995; Moritz et
al., 1995). The centrosome undergoes dynamic changes
and movements in a cell-cycle dependent way. In
cycling cells, the centrosome is replicated during the S
phase, while its separation into a bipolar spindle is
delayed until prophase/prometaphase of mitosis. The
centrosomes move further apart during late anaphase
when the spindle elongates.
The mechanism of vertebrate centrosome separation
is not well understood. It has been proposed that both
pushing forces between sliding polar microtubules
emanating from opposite spindle poles, and pulling
forces of each centrosome mediated by astral microtubules interacting with cell cortex, contribute to
centrosome movements in vertebrates (Ault and
Rieder, 1994). Genetic and biochemical studies point
to the critical role of dynein- and kinesin-related
microtubule motor proteins in centrosome separation,
and they are presumably responsible for generating the
separating forces (see for reviews Endow and Titus,
1992; Hoyt, 1994; Moore and Endow, 1996). However,
centrosome movement is normally a cell cycledependent process and it is not clear how it is
controlled. Recently, a mutation in Drosophila,
designated aurora, that a€ects centrosome separation
and behavior has been described (Glover et al., 1995).
Amorphic alleles in aurora result in pupal lethality
which is probably due to a failure in centrosome
separation, and consequently a failure in formation of
a bipolar spindle. The mutant gene was cloned and has
been shown to encode a novel serine-threonine kinase
(Glover et al., 1995), suggesting that reversible protein
phosphorylation plays a central role in controlling
Drosophila centrosome separation.
We report here on the cloning and characterization
of a novel mammalian protein kinase gene, designated
ayk1, which is highly similar to the Drosophila gene
aurora. The predicted amino acid sequence shows
62.9% identity to aurora in the kinase domain, and
lesser similarity exists in the amino and the carboxy
terminals. Ayk1 is also similar to yeast Ipl1 (49.4%
identity in the catalytic domain), which has also been
shown to be involved in chromosome segregation
(Francisco et al., 1994). Expression analysis, by
Northern and in situ RNA localization, revealed that
the predominant sites of ayk1 expression are the
gonads. Interestingly, in the murine testis, highest
levels of ayk1 RNA are con®ned to a speci®c subset
of spermatocytes, and much lower levels are seen in
mitotically dividing spermatogonia. These results
suggest that the pathways controlling centrosome
movements are conserved from yeast to mammals,
and that common control mechanisms operate in
mitosis and meiosis.
Results
Isolation and characterization of ayk1 cDNA
Correspondence: B Motro
Received 30 August 1996; revised 11 March 1997; accepted 11 March
1997
In an e€ort to identify novel murine protein kinases of
the NIMA-related family (Nek kinases) (Osmani et al.,
1988; Letwin et al., 1992; Schultz and Nigg, 1993), we
Ayk1, a novel aurora-related mammalian kinase
A Yanai et al
2944
designed degenerate primers tailored to characteristic
amino acids of this group. The primers were used in a
RT ± PCR cloning procedure with di€erent sources of
polyA(+) RNAs (see Materials and methods). While
succeeding in cloning novel murine members of this
family (submitted for publication), a 335 bp fragment
of a cDNA encoding a putative unrelated protein
kinase was cloned from adult testis mRNA. This
fragment was used as a probe on testis, brain and
embryonic murine cDNA libraries. Several testis and
embryonic clones and one brain clone were isolated
and were analysed further. Restriction fragment
analysis and partial nucleotide sequence analysis
revealed that all clones probably encode overlapping
a
b
Figure 1 Structure of ayk1 clones and the sequence of ayk1
cDNA. (a) Schematic representation of the structure of the
predicted Ayk1 kinase is shown above the individual clones
isolated from testis (tay2 and tay1), brain (bay3) and fetal (fay8)
mouse cDNA libraries. In the diagram thin lines represent the
predicted 5' and 3' untranslated regions, open rectangles the
coding region, and the hatched box the catalytic kinase domain.
The product of the initial PCR reaction, and the regions used as
probes in Northern and in situ analyses, and for cloning of
additional 5' clones are also shown. (b) Nucleotide and deduced
amino acid sequence of mouse ayk1. Nucleotides are numbered
on the left and amino acids on the right. Highly conserved amino
acids in the kinase catalytic domains are underlined. Also
underlined are the ®rst in-frame methionine codon and the
polyadenylation signal (aataaa). Amino acids with a continuous
underline are possible targets for phosphorylation by PKA, and
amino acids in bold represent the putative PEST domain
fragments of the same cDNA, designated ayk1 (for
aurora-family kinase 1). Therefore, the two longest
testis clones, one brain and one embryonic clone (tay1
and tay2 ; bay3 and fay8, respectively) (Figure 1a) were
sequenced. The composite nucleotide and inferred
amino acid sequences are shown in Figure 1b. The
sequence spans 1656 nt, ends with a poly(A) tail, and
contains the consensus polyadenylation signal AATAAA 18 nucleotides upstream of the polyA sequence
(underlined in Figure 1b). The combined sequence
begins with a long open reading frame, and the ®rst inframe ATG codon (beginning at nucleotide 49)
conforms to Kozak's consensus (Kozak, 1984),
suggesting that it serves as an initiator methionine.
To con®rm that our composite cDNA represents a fulllength cDNA, we screened a primary embryonic cDNA
library with ayk1 5' probe (5' probe in Figure 1a), and
isolated 24 independent clones. PCR and sequencing
analyses revealed that none of the clones have longer 5'
sequences, and two of the clones ®nished at the exact
same 5' nucleotide found in the testicular clone tay1.
These results suggest that either there is a strong stop
for reverse transcriptase at this site, or more likely, that
this is an authentic 5' end of the RNA. Making this
assumption, ayk1 encodes a predicted polypeptide of
395 amino acids, with a calculated molecular mass of
44.8 kDa, and calculated isoelectric point of 10.2.
Residues 124 to 374 of the predicted open reading
frame contain all the highly conserved motifs characteristic of the 12 subdomains of protein kinases
(underlined in Figure 1b) (Hanks and Hunter, 1995).
Within the catalytic domain, two sequences that
distinguish serine/threonine (S/T) kinases from tyrosine
kinases are present. In subdomain VIB tyrosine kinases
usually have AARN while S/T kinases have KPEN,
with the lysine being highly conserved. Ayk1 conforms
fully to the S/T consensus (a.a. 249 ± 252). In
subdomain VIII tyrosine kinases are characterized by
PXXWXAPE while in most S/T kinases, and in Ayk1,
only the PE exists. The highly conserved motifs suggest
that Ayk1 encodes a functional S/T kinase. Comparison
of the predicted amino acids of the putative catalytic
domain with protein kinases in the Genbank/EMBL
databases revealed closest homology to an unpublished
Xenopus laevis sequence designated Eg2 (XLP46APK,
accession number Z17206) (83.3% identity; 91.6%
similarity), and to Drosophila melanogaster aurora
(Glover et al., 1995) (62.9% identity; 79.3% similarity) and Saccharomyces cerevisiae Ipl1 (Francisco et al.,
1994) (49.4% identity; 70.1% similarity). Much lesser
homology (35 ± 37% identity) exists with several c-AMP
dependent protein kinases (PKAs). The high levels of
homology between Ipl1, aurora and Ayk1 (Figure 2)
suggests that the three kinases constitute a distinct
subfamily of serine/threonine kinases. However, the
sequence similarity outside the kinase domain is much
lower (in the N-terminal domain Ayk1 share 26.3%
identity with aurora and 22.4% identity with Ipl1). In
addition, Ipl1 kinase has an about 100 a.a. N-terminal
domain which carries possible phosphorylation sites for
both PKA and p34cdc2 (Francisco et al., 1994), that do
not exist in the two other kinases. However, in the
catalytic domain itself, Ayk1 contains two sequences
which match the consensus for sites phosphorylated by
PKA (R/K-R/K-X-S/T) (Pearson and Kemp, 1991)
(Figure 1b).
Ayk1, a novel aurora-related mammalian kinase
A Yanai et al
Intestine
Spleen
Brain
Testis
Ovary
Placenta
17.5 dpc
16.5 dpc
8.5 dpc
2945
28S —
Figure 2 Alignment of the protein kinase catalytic domain of
Ayk1 with those of Drosophila aurora and Saccharomyces Ipl1.
The kinase subdomains are indicated on the top of the sequences,
according to Hanks and Quinn (1991). Amino acids identical in
all three sequences are indicated in the consensus line
Interestingly, mutation of either ipl1 or aurora
results in a failure to accurately segregate chromosomes. Temperature-sensitive mutants for ipl1 missegregate chromosomes at the restrictive temperature and
the kinase is essential for completion of mitosis and
yeast viability (Francisco et al., 1994). The aurora gene
was cloned by chromosomal walking combined with
complementation assays in a ¯y mutant for centrosome
separation and chromosome segregation (Glover et al.,
1995). The role of these genes in the cell cycle suggests
a similar function for Ayk1. Therefore, we determined
whether ayk1 patterns of expression are consistent with
such a role.
18S —
— ayk1
— α-tubulin
Figure 3 Northern blot analysis of ayk1 expression in embryonic
and adult tissues. 12 mg of total RNA extracted from the
indicated murine tissues were blotted on a Magna nylon
membrane (MSI), and hybridized to the ayk1 probe indicated in
Figure 1a (see Materials and methods). The placental RNA was
extracted from 12.5 dpc embryos. The migration of the 28S and
18S ribosomal RNAs is indicated at the left. The same blot was
rehybrydized to a-tubulin probe for comparison of RNA amounts
used
Expression of ayk1 in embryonic and adult mouse tissues
ayk1 expression during gametogenesis
To determine the size and distribution of ayk1
transcripts, Northern blot analysis was performed on
total RNA from di€erent embryonic stages and adult
tissues. The probe spans 601 3' nucleotides, consisting
almost entirely of 3' noncoding region. In all sources of
RNA examined, ayk1 marked only one sized transcript
of about 1.8 kb (Figure 3). Throughout embryonic
development high levels were detected, with slightly
higher levels during earlier stages (Figure 3, and not
shown). In adult tissues robust expression was found in
testis, and much lower levels were detected in ovary,
intestine and spleen. The lowest levels were found in
12.5 dpc (days post coitum) placenta and brain.
To identify the embryonic cells expressing ayk1, insitu RNA hybridization was performed on cryosections
of embryos at successive developmental stages. Early in
development, at 7.5 ± 9.5 dpc, ayk1 was found to be
expressed evenly all over the embryo. At later stages,
ayk1 transcripts were mainly restricted to proliferating
zones. For example, in the central nervous system, the
expression was con®ned to the ventricular zone (Figure
4A,B), and we were unable to detect any signal in adult
brain slides (not shown). The highest levels at mid
embryonic development (13.5 dpc) were observed in
the liver, lung, kidney and back (trapezius) muscle, all
regions in active proliferation (Figure 4C,D and not
shown). At later stages (15.5 ± 17.5 dpc), lower levels
were detected in these tissues (not shown).
The Northern analysis pointed to the gonads as the
major site of ayk1 expression. The identity of the
expressing cells was assessed by in situ analysis. In
female mice proliferation of oogonia and entrance into
meiosis occur during late embryonic development. At
birth, the oocytes are arrested at prophase of the ®rst
meiotic division, and are designated resting oocytes.
Following gonadotropic stimulation, the oocyte's size
increases and the surrounding granulosa cells proliferate. The mature oocyte is stimulated by luteinizing
hormone to resume the ®rst meiotic division and to
ovulate, while the second division is dependent on
fertilization. In situ hybridization to adult ovaries
revealed that there was no expression of ayk1 in
primary resting oocytes. Expression in the oocyte
started at around the transition to large growing
follicle, as staged by the appearance of more than
two layers of granulosa cells, and high levels persisted
until ovulation (staging according to Peters, 1969). In
addition, ayk1 RNA was quite abundant in the
proliferating granulosa and thecal cells of the growing
follicle, and in the young corpus luteum. No signi®cant
signal was found over interstitial cells or granulosa
cells of resting follicles (Figure 4E,F).
Identi®cation of the testicular cells expressing ayk1
took advantage of the well characterized development
of the prepuberal testis (Bellve et al., 1977), as outlined
below. At birth, the testis contains nonproliferating
Ayk1, a novel aurora-related mammalian kinase
A Yanai et al
2946
Figure 4 ayk1 expression during embryogenesis and in the ovary. (A) Bright- and (B) dark-®eld view of a cross section of the brain
of 14.5 dpc embryo. The expression is con®ned to the ventricular (proliferating) zone. c.p., choroid plexus; IV, fourth ventricle; st,
striatum; th, thalamus. (C) Bright- and (D) dark-®eld view of a cross section of 14.5 dpc embryo. The liver (li) and lung (lu) display
signi®cant ayk1 expression. di, diaphragma; he, heart. (E) Bright- and (F) dark-®eld view of an adult ovary. Silver grains are
prominent over large oocytes (marked by arrows), and lower levels are seen over young corpus luteum (c.p.)
gonocytes. Shortly afterwards, spermatogenesis is
initiated, at day 6 the ®rst primitive type A
spermatogonia appear, and by day 8 type B
spermatogonia are found. The ®rst meiotic pachytene
spermatocytes emerge at day 14, and at around days
18 ± 20 secondary spermatocytes, and soon after
haploid spermatids, appear (Bellve et al., 1977). The
haploid round spermatids are transformed into
elongated spermatids and spermatozoa in the spermiogenesis process that produces the ®rst mature
spermatozoa at around day 30. Using this information, we followed by Northern and in situ hybridization
the expression of the gene in testes of di€erent
developmental ages in order to de®ne the stage at
which the germ cells start to express ayk1.
The Northern analysis revealed very low levels of
expression in 8 to 17 days old postnatal testis, dramatic
8
10
12
14
17
21
30
AYK1
GAPDH
Figure 5 Northern blot analysis of ayk1 expression in the
developing testis. 10 mg of total RNA were extracted from testes
at the indicated days after birth, blotted on a Magna nylon
membrane (MSI) and hybridized to the ayk1 probe. Exposure time
shown was 24 h and longer exposure revealed bands in all lanes.
As a control, the same blot was rehybrydized to GAPDH probe
Ayk1, a novel aurora-related mammalian kinase
A Yanai et al
2947
Figure 6 Expression of ayk1 RNA in the developing postnatal testis. (A) Bright- and (B) dark-®eld view of a section of 17 days
postnatal (dpn) testis hybridized to ayk1 antisense RNA probe. The low levels of signal are not con®ned to the seminiferous tubules.
(C) Bright- and (D) dark-®eld view of a section of 21 dpn testis. ayk1 hybridization signal is con®ned to only some of the tubules.
(E) Bright- and (F) dark-®eld view of a section of adult testis. High levels of expression are seen in only some of the tubules, and
higher magni®cation of this section (G) emphasizes that the signal is localized to pachytene spermatocytes, and no grains are seen
over the interstitial Leydig cells, peripheral spermatogonia or the inner elongated spermatids. (H) Adult testis hybridized to ayk1
sense RNA (control) probe. The calibration bar in A indicates 100 mm, and A ± F and H are of the same magni®cation. In G the bar
indicates 25 mm
Ayk1, a novel aurora-related mammalian kinase
A Yanai et al
2948
appearance at 21 days, and similar levels in adult testis
(Figure 5). To localize the testicular cells expressing
ayk1, in situ hybridization was performed on sections
of di€erent developmental ages. To allow direct
comparison of ayk1 levels, all the testes from the
di€erent ages were embedded in a single cryomold and
processed on a single slide. In agreement with the
results from the Northern analysis, low levels of
expression were found in the mitotic period at day 6
testis, and they were not con®ned speci®cally to the
spermatogonia compartment. Only low levels were
detected in leptotene/zygotene and pachytene spermatocytes, at day 12 and 17, respectively (Figure 6A,B
and not shown). However, at day 17 some rare tubules
showed much higher levels (not shown). In contrast, by
day 21 about a third of the seminiferous tubules
revealed very high levels of ayk1 transcripts, while the
rest retained low levels of expression (Figure 6C,D).
Similar patterns were found in adult testis, where the
strong signal for ayk1 was restricted to about third of
the tubules (Figure 6E,F). The timing of ayk1
appearance suggests that the gene is turned on to
very high levels toward the end of the pachytene stage,
just prior to the ®rst meiotic division. The similar
percentage of tubules expression ayk1 at 21 days and
adults indicates that ayk1 is expressed until the round
spermatid stage. This interpretation is supported by
analysis of the cells expressing ayk1 within the adult
testis. The hybridization signal was present over some
of the tubules containing pachytene spermatocytes
(apparently late pachytene) and over round spermatids, but no signal above background was detected over
elongated spermatids (Figure 6G). The expression was
almost exclusively con®ned to germ cells, and interstitial and Sertoli cells did not exhibit signi®cant signals.
Discussion
In this study we identi®ed a novel murine gene that is
highly related to Saccharomyces cerevisiae Ipl1
(Francisco et al., 1994) and Drosophila melanogaster
aurora (Glover et al., 1995) kinases. Even though
kinase activity has not yet been demonstrated for any
of these three proteins, the genes probably encode
active serine/threonine kinases, since the predicted
amino acid sequences contain all the conserved
hallmarks of protein kinases, and motifs characterizing the serine/threonine group. The high overall
similarity between the kinase domains, and the
conservation of certain motifs, suggest that these
kinases belong to a novel subfamily, distantly related
evolutionarily to the PKA family (Figure 7). Diagnostic stretches of amino acids include the motifs
RRE(I/V)EIQ (domain III), IHRDIKPEN (domain
VIB), and CGT(I/V)DYL(S/P)PEM (domain VIII). All
three proteins are very basic, having calculated
isoelectric points higher than 10. Although only low
homology exists in both termini, it is remarkable that
the three kinases share very short carboxy termini (5,
12 and 28 amino acids in aurora, Ipl1 and Ayk1,
respectively). Interestingly, Ayk1 (but not Ipl1 or
aurora) has a PEST-rich domain at the C-terminus
(11 out of 17 residues) bordered by basic residues,
resembling motifs found in proteins destined for rapid
controlled degradation (Rogers et al., 1986).
Figure 7 Phylogenetic tree of Ayk1, aurora and Ipl1 with
representatives of the major serine/threonine families. The tree
was created using the GCG programs of the Wisconsin Sequence
Analysis Package, and is based on comparison of the kinase
catalytic domains. The distances between the domains were
calculated using the Distances program according to Kimura
(1983). The tree was drawn using the Growtree program
according to the Neighbor-Joining method (Saitou and Nei,
1987). Branch lengths indicate estimated number of amino acid
substitutions, per 100 amino acids, required to reach hypothetical
common ancestor
The phenotypic consequence of mutations in the Ipl1
and aurora loci is a failure to accurately segregate
chromosomes. However, the precise mechanism responsible for this defect is probably di€erent for the
two gene products. In budding yeast, for mutants of
the Ipl1 gene, it has been shown that the kinase is
absolutely required during mitosis, but not for earlier
stages of the cell cycle. The reason for the missegregation of the chromosomes is not clear, as the mutant
cells exhibit structurally normal mitotic spindles and
microtubule structures (Francisco et al., 1994). In
contrast, in aurora mutants, a clear defect in spindle
formation has been reported (Glover et al., 1995). In
the mitotic cells of the mutant's brains, the centrosomes do not separate, resulting in the formation of a
monopolar spindle around which the chromosomes are
arranged circularly. Polyploid nuclei are common in
aurora mutants, while gain or loss of few chromosomes
is the rule in Ipl1 mutant yeast. Nevertheless, the
overall structural similarity between aurora and Ipl1
(and Ayk1) kinases suggests a common role, and
similar substrates. It is thus tempting to speculate that
these kinases are implicated in phosphorylation of
related spindle-associated proteins important for the
interaction of the spindle with the kinetochore/
Ayk1, a novel aurora-related mammalian kinase
A Yanai et al
chromosome (Ipl1), or the centrosome (aurora). It is
important to note in this regard that it has been shown
in various systems that kinesin and dynein-related
microtubule motor proteins, are crucial for chromosome segregation. Their absence leads in some cases to
disruption of spindle assembly and to disruption of the
spindles association with the chromosome (Zhang et
al., 1990); Theurkauf and Hawley, 1992; Vernos et al.,
1995). In other cases, their absence a€ects more
speci®cally centrosome/pole separation, leading to the
formation of monopolar spindles (Hoyt et al., 1992;
Heck et al., 1993; Vaisberg et al., 1993; Blangy et al.,
1995; Boleti et al., 1996), similar to those obtained in
aurora mutants. It also has been shown that
phosphorylation is a major regulatory mechanism in
controlling motor protein localization and activity
(Liao et al., 1994; Sawin and Mitchison, 1995). It is
thus possible that the di€erent aurora-related kinases
are involved in controlling the activity of various
microtubule motor proteins. However, cellular localization data and biochemical analyses will be needed in
order to evaluate this possibility.
We show in this work that ayk1 is expressed in
embryonic mitotically dividing cells, and in germ cells.
Interestingly, the levels in meiotic germ cells were much
higher than those observed in somatic dividing cells,
and only low (but signi®cant) signal was observed over
dividing spermatogonia at postnatal days 8 and 10.
Similarly, only low levels were seen in NIH3T3
®broblasts (data not shown), and the role, if any, of
Ayk1 in the mitotic cell cycle is not clear. The
expression in spermatocytes increases sharply toward
the completion of prophase, and decays around the
transition of the spermatids from round to elongated.
This expression pattern implies that Ayk1 kinase is not
essential in the preleptotene to pachytene stages, but
probably functions in the meiotic divisions. Analogously, the levels of ayk1 in oocytes rise in secondary
oocytes which are preparing to undergo the ®rst
meiotic division. Taken together, the similarity of
Ayk1 to kinases involved in chromosome segregation,
and its distinct temporal expression, suggest that it
serves a crucial role in mammalian meiotic divisions,
probably in controlling some aspects of chromosome
segregation. Using a dominant-negative approach, we
are now examining whether disruption of Ayk1 activity
leads to chromosomal missegregation and uneuploidy,
and whether mutations in ayk1 gene are present in
various cancers.
Materials and methods
Cloning the ayk1 cDNA
The initial ayk1 fregment was cloned while we were
attempting to clone NIMA-related murine kinases by
RT ± PCR. Reverse transcription of poly(A)+ mRNA from
adult testis was done using random hexamers as primers
(Boehringer Mannheim). For the PCR reaction, degenerate
oligonucleotide primers, designed according to conserved
amino acids in NIMA-related kinases, were synthesized.
The forward primer represented the motif (L/M)KHPNIV
[TG AA(A/G) CA(T/C) CCN AA(A/G) AT(T/C) GT], and
the backward primer the motif GTPYY(L/V/M) [TAN (A/
G)TA (A/G)TANGG NGT NCC]. The PCR was carried
out for 36 cycles using Taq DNA polymerase (Boehringer
Mannheim) and an automated thermal cycler (MJ
Research). Each cycle consisted of 1 min at 948C, 1 min
at 408C and 1 min at 638C. The products were subjected to
electrophoresis through a 1.5% agarose gel, extracted using
QIAquick Gel Extraction kits (QIAGEN), and ligated into
the T-cloning Kit based on the pUC57 vector (MBI
fermentase). The clones were sequenced using the Dye
Terminator Cycle Sequencing Kit (Perkin Elmer) and an
Applied Biosystems Model 373 automated DNA sequencer
at the DNA Sequencing Unit, Biological Services,
Weizmann Institute. One of the clones (Figure 1a)
encoded a putative kinase unrelated to the NIMA family.
It was used as a probe to isolate cDNA clones from murine
cDNA libraries. Several clones were isolated from an adult
testis cDNA library, and from a 10.5 dpc embryonic
library (kindly provided by Uri Nir, Bar Ilan University,
and Maria Rozakis-Adcock, Samuel Lunenfeld Research
Institute, respectively). One clone was isolated from an
adult brain cDNA library. Sequences obtained from the
ends of the clones and restriction analyses suggested that
all clones encode for the same cDNA, and therefore only
two testis (tay1 and tay2), one brain (bay3), and one
embryonic clone (fay8) were subcloned and sequenced
(Figure 1a). In addition, 24 independent cDNA clones were
isolated from the 10.5 dpc embryonic library using a 5'
probe (Figure 1a). The 5' extents of these clones were
determined by PCR, with antisense primer located at
nucleotides 83 ± 102 of ayk1 (GGGAACAGTGGTCTTAACAG), and a primer from the lEXlox vector
(Novagen) (GAGGTTGTAGAAGTTC CGCA). The 5'
side of the ®ve longest cDNA clones were sequenced. The
reported sequence was con®rmed on both strands.
Sequence comparisons
A search for sequences similar to ayk1 was conducted
against the GenBank/EMBL library database using the
FASTA program of the University of Wisconsin Genetic
Computer Group (GCG) Sequence Analysis Software
Package. The percentage of similarity between sequences
was determined using the Gap program. Comparison of the
sequence of Ayk1 kinase to those of aurora and Ipl1 was
done using the Pileup program of the GCG package.
Northern analysis
Total RNA was extracted using the LiCI/urea precipitation
method, as described in Au€ray and Rougeon (1980,
Procedure C). The RNA was subjected to electrophoresis
on a 1.2% agarose gel and transferred to a Magna nylon
membrane (MSI). The ®lter was hybridized with a
randomly primed ayk1 cDNA probe spanning nucleotides
1056-clone according to standard procedures (Sambrook et
al., 1988), and washed in 0.16SSC and 0.1% SDS at 608C.
The ®lter was exposed to Kodak X-OMAT ®lm with
intensifying screens. The size of ayk1 transcript was
determined in comparison to RNA Molecular Weight
Markers (New England Biolabs).
In situ RNA analysis
In situ RNA hybridization was carried out on embryonic
and adult 10 mm cryosections as previously described
(Motro et al., 1991). To allow comparison of the levels
of ayk1 expression in the di€erent stages of the prepuberal
and adult testis, all of the testes from the di€erent time
points were embedded in a single cryomold, cosectioned,
and processed on a single slide. As an antisense probe, an
EcoRI subclone of TAY2 in KS+ (Stratagene) spanning
nucleotides 1056-clone, was linearized at the HindIII site
(at the polylinker), and 36S-labeled RNA was transcribed
by T7 RNA polymerase. As a sense control probe, the
2949
Ayk1, a novel aurora-related mammalian kinase
A Yanai et al
2950
same fragment was labeled after linearization with Xbal,
followed by use of T3 RNA polymerase. Post-hybridization
washings included treatment with 50 mg/ml RNase A at
378C for 30 min, and two stringent washes of 20 min each
at 608C in 0.16SSC. The slides were dipped in Kodak
NTB-2 emultion, exposed for 5 ± 6 days, developed, and
stained with toluidine blue.
LineUp editing program. The pairwise evolutionary
distances between the aligned sequences were calculated
using the Distances program according to Kimura formula
(Kimura, 1983). The Neighbor-Joining algorithm (based on
Saitou and Nei, 1987) in the GrowTree program was used
to reconstruct the phylogenetic tree.
Phylogenetic tree creation
Acknowledgements
The alignment of the catalytic domains of the listed kinases
and the construction of the phylogenetic tree were done
using several GCG programs. The sequences were aligned
using the Pileup program and edited by hand, according to
the kinase subdomains (Hanks and Hunter, 1995) using the
We thank Ron Goldstein, Ron Wides and Jeremy Don for
critical reading of the manuscript, and Uri Nir and Maria
Rozakis-Adcock for providing testis and fetal cDNA
libraries, respectively. This work was supported by the
Israel Cancer Research Fund.
References
Au€ray C and Rougeon F. (1980). Eur. J. Biochem., 107,
303 ± 314.
Ault JG and Rieder CL. (1994). Curr. Opin. Cell Biol., 6, 41 ±
49.
Bellve AR, Cavicchia JC, Millette CF, O'Brien DA,
Bhatnagar YM and Dym M. (1977). J. Cell Biol., 74,
68 ± 85.
Blangy A, Lane HA, d'Herin P, Harper M, Kress M and
Nigg EA. (1995). Cell, 83, 1159 ± 1169.
Boleti H, Karsenti E and Vernos I. (1996). Cell, 84, 49 ± 59.
Endow SA and Titus MA. (1992). Annu. Rev. Cell Biol., 8,
29 ± 66.
Francisco I, Wang W and Chan CSM. (1994). Mol. Cell.
Biol. ,14, 4731 ± 4740.
Glover DM, Leibowitz MA, McLean DA and Pary H.
(1995). Cell, 81, 95 ± 105.
Hanks SK and Quinn AM. (1991). Methods Enzymol, 200,
38 ± 62.
Hanks SK and Hunter T. (1995). FASEB J., 9, 576 ± 596.
Hassold T and Jacobs PA. (1984). Annu. Rev. Genet., 6, 555 ±
565.
Heck MMS, Pereira A, Pesavento P, Yannoii Y, Sparding
AC and Goldstein LSB. (1993). J. Cell Biol., 123, 665 ±
679.
Hoyt MA, He L, Loo KK and Saunders WS. (1992). J. Cell
Biol., 118, 109 ± 120.
Hoyt MA. (1994). Curr. Opin. Cell Biol., 6, 63 ± 68.
Liao H, Li G and Yen TJ. (1994). Science, 265, 394 ± 398.
Kimura M. (1983). The Neutral Theory of Molecular
Evolution. Cambridge University Press, Cambridge.
Kozak M. (1984). Nucl. Acid Res., 12, 857 ± 870.
Letwin K, Mizzen L, Motro B, Ben-David Y, Bernstein A
and Pawson T. (1992). EMBO J., 10, 3521 ± 3531.
Moore JD and Endow SA. (1996). BioEssays, 18, 207 ± 219.
Motro B, van der Kooy D, Rossant J, Reith A and Bernstein
A. (1991). Development, 113, 1207 ± 1221.
Moritz M, Braunfeld MB, Sedat JW, Alberts B and Agard
DA. (1995). Nature, 378, 638 ± 640.
Osmani SA, Pu RT and Morris NR. (1988). Cell, 53, 237 ±
244.
Pearson RB and Kemp BE. (1991). Methods Enzimol., 200,
62 ± 81.
Peters H. (1969). Acta Endocrin., 62, 98 ± 116.
Rogers S, Wells R and Rechsteiner M. (1986). Science, 234,
364 ± 368.
Sambrook J, Fritsch EF and Maniatis T. (1988). Molecular
Cloning: A Laboratory Manual, 2nd Edition, Cold Spring
Harbor: NY, Cold Spring Laboratory Press.
Sawin KE and Mitchison TJ. (1995). Proc. Natl. Acad. Sci.
USA, 92, 4289 ± 4293.
Saitou N and Nei M. (1987). Mol. Cell. Evol., 4, 406 ± 425.
Schultz SJ and Nigg EA. (1993). Cell Growth Di€er., 4, 821 ±
830.
Theurkauf WE and Hawley RS. (1992). J. Cell Biol., 116,
1167 ± 1180.
Vaisberg EA, Koonce MP and McIntosh JR. (1993). J. Cell.
Biol., 123, 849 ± 858.
Vernos I, Raats J, Hirano T, Heasman J, Karsenti E and
Wylie C. (1995). Cell, 81, 117 ± 127.
Zhang P, Knowles BA, Goldstein LSB and Hawley RS.
(1990). Cell, 62, 1053 ± 1062.
Zheng Y, Wong ML, Alberts B and Mitchison T. (1995).
Nature, 378, 578 ± 583.