A Novel Family of TRF (DNA Topoisomerase I-Related Function) Genes Required for Proper Nuclear Segregation
Irene B. Castao
Ben U. Sadoff
David J. Fitzhugh
Michael F. Christman
Department of Radiation Oncology, University of California
San Francisco, CA 94143, USA
We recently reported the identification of a gene,TRF4 (for DNA topoisomerase related function), in a screen for mutations that are synthetically lethal with mutations in DNA topoisomerase I (top1). Here we describe the isolation of a second member of theTRF4 gene family, TRF5. Overexpression of TRF5 complements the inviability of top1 trf4 double mutants. The predicted Trf5 protein is 55% identical and 72% similar to Trf4p. As with Trf4p, a region of Trf5p is homologous to the catalytically dispensable N-terminus of Top1p. The TRF4/5 function is essential as trf4 trf5 double mutants are inviable. A trf4 (ts) trf5 double mutant is hypersensitive to the anti-microtubule agent thiabendazole at a semi-permissive temperature, suggesting that TRF4/5 function is required at the time of mitosis. Examination of nuclear morphology in atrf4 (ts) trf5 mutant at a restrictive temperature reveals the presence of many cells undergoing aberrant nuclear division, as well as many anucleate cells, demonstrating that the TRF4/5 function is required for proper mitosis. Database searches reveal the existence of probable Schizosaccharomyces pombe and human homologs of Trf4p, indicating that TRF4 is the canonical member of a gene family that is highly conserved evolutionarily.
Chromosomes are involved in dynamic cellular processes, such as
DNA replication, transcription, chromatin assembly and genetic
recombination, that lead to the formation of local domains of
torsional stress (reviewed in 1). In Saccharomyces cerevisiae DNA
topoisomerases I and II (topo I and II) appear to act together during
DNA replication as a swivel to prevent the formation of positive
supercoils ahead of the DNA replication fork. In S.cerevisiae under
conditions where both topo I and II are inactivated, DNA replication
stops rapidly (2), with elongation of new DNA chains continuing for
only a few thousand nucleotides (3). Similar results have been
obtained for Schizosaccharomyces pombe (4).
* To whom correspondence should be addressed
Transcription can also lead to the formation of locally
supercoiled domains in DNA. In the twin domain model (5),
movement of a transcription complex along the helical backbone
generates positive supercoils ahead of the complex and negative
supercoils behind the complex. It has been suggested that
transcription is a major determinant of supercoiling in vivo (6). In
S.cerevisiae, topo I and II appear to function together as a swivel
for rRNA transcription and, to a lesser extent, for mRNA
transcription (2). Transcription from a strong promoter can lead
to hypernegative supercoiling of plasmids in top1 mutants (6,7),
suggesting that topo I normally removes negative supercoils
formed during transcription. These results and others have led to
the suggestion that another major role for topo I is to relieve
torsional stress generated during transcription.
Despite the considerable evidence for the involvement of topo I
in both DNA replication and transcription, topo I is not essential
in either S.cerevisiae or S.pombe. Null mutations in the gene
encoding the only type I DNA topoisomerase activity detectable
in crude extracts cause only modest growth defects (4,8,9).
Furthermore, the overall rates of both DNA and RNA synthesis
are normal in top1 mutants (2), supporting the suggestion that
topo II activity can substitute for topo I activity in these crucial
processes (2,4,10). In S.pombe, conditional top2 mutants have
been isolated that are inviable at the permissive temperature in
combination with a top1 null mutation (11). In S.cerevisiae, top1
null mutants display a synthetic growth defect in combination
with some top2 alleles (10) and we have identified alleles of
TOP2 in a screen for top1 synthetic lethal mutations (12). These
results demonstrate that some of the functions of type I and type
II topoisomerases overlap. It remains unclear, however, whether
topo I has additional functions distinct from those of topo II. The
biological function of a second type I topoisomerase, topo III,
(13,14), remains unclear. Null mutations in TOP3 result in
hyper-recombination and slow growth (15,16), but top1 top3
double mutants are viable. Association of Top3p with Sgs1p, a
DNA helicase, has been proposed to result in formation of a
eukaryotic reverse gyrase (17).
In order to investigate further the in vivo functions of topo I, we
have perfomed a genetic screen to identify mutations affecting
gene products that perform overlapping or dependent functions
(18) with topo I and, thereby, to further elucidate which processes
in the cell require topo I. We have identified four complementation
groups of mutants with this phenotype called TRF, for DNA
topoisomerase I-related function (12). The predicted Trf4 protein
shares a region of homology with the N-terminus of Top1p.
We report here the isolation of a TRF4 homolog, TRF5. DNA
sequence analysis reveals that the predicted Trf4 and Trf5 proteins
are 57% identical and 72% similar over their entire lengths.
Overexpression of TRF5 complements the inviability of top1 trf4
double mutants. A trf4 (ts) trf5 double mutant is hypersensitive to
the anti-microtubule agent thiabendazole at a semi-permissive
temperature and grossly defective in nuclear division at a restrictive
temperature, demonstrating that TRF4/5 function is required for
proper mitosis. Database searches reveal the existence of probable
S.pombe and human and homologs of TRF4/5, indicating that this
novel gene family is evolutionarily conserved.
MATERIALS AND METHODS
Yeast strains (Table 1) were transformed using the lithium acetate
method (19). The TRF5 gene was isolated from a genomic DNA
library consisting of random S.cerevisiae DNA Sau3A fragments
inserted into the BamHI site of YEp24. Escherichia coli cells
(strain DH5a ) were transformed by electroporation (20). Small
scale plasmid DNA preparations were made by the boiling lysis
method (21) or using Qiagen columns (Qiagen, Chatsworth, CA).
Yeast strain construction and crosses
CY667. Strains CY726 and CY445 were crossed and sporulated.
CY726. Derived from CY429 following EMS mutagenesis.
CY855 and CY857. Plasmid pCB470 was digested with SacI to
release a trf4-101::HIS3 fragment. This fragment was used to
disrupt TRF4 in the TRF4/TRF4 diploid CY858. The resulting
diploid was sporulated and several tetrads were dissected. One of
the His+, cold-sensitive spores was designated CY855; another
cold-sensitive His+ spore was designated CY857. Disruption at the
TRF4 locus in CY855 was confirmed by Southern blot analysis.
CY869. pCB469 was digested with NotI and XhoI to release
trf4-102::TRP1 and the resulting fragment was used to disrupt
TRF4 in CY184 to create CY869.
CY870, CY872 and CY874. Plasmid pSH5 was digested with
BamHI and PstI to release a trf5-3::LEU2 fragment. This
fragment was used to disrupt TRF5 in the TRF5/TRF5 diploid
(CY858). Disruption at the TRF5 locus at one allele was confirmed
by Southern blotting. The resulting diploid (CY870) was
sporulated and several tetrads were dissected. One of the Leu+
spores was designated CY874 (trf5-3::LEU2). A sister spore that
is TRF5+ was designated CY872.
CY908. Plasmid pSH5 was digested with BamHI and PstI to
release a trf5-3::LEU2 fragment. This fragment was used to
disrupt TRF5 in the trf4-101::HIS3/trf4-101::HIS3 homozygous
diploid (CY892) to generate CY908. Disruption of TRF5 was
confirmed by Southern blotting. The resulting diploid was
sporulated and tetrads were dissected to test for synthetic lethality
between the trf4 and trf5 mutations.
MATa ade2-1 ura3-1 his3-11,15 trp1-1 leu2-3,112 can1-100 rDNA::URA3
MATa ade2-1 ura3-1 trp1-1 his3-11,15 leu2-3,112 rDNA::ADE2.
MATa top1-7::LEU2 ade2-1 ura3-1 trp1-1 his3-11,15 leu2-3,112 rDNA::ADE2
MATa top1-7::LEU2 ade2-1 ura3-1 trp1-1 his3-11,15 leu2-3,112 ade3::hisG
MATa top1-7::LEU2 ade2-1 ura3-1 his 3-11,15 trp1-1 leu2-3,112 ade3::hisG/pBS3
CY429 switched to MATa with pCB10.
MATa top1-7::LEU2 trf4-1 his3-11,15 trp1-1 ade2-1 ade3/pBS3
MATa top1-7::LEU2 trf4-1 his3-11,15 trp1-1 ura3-1 ade2-1ade3::hisG
MATa trf4-101::HIS3 ura3-1 ade2-1 his3-11,15 trp1-1 leu2-3,112 rDNA::ADE2
MATa trf4-101::HIS3 ura3-1 ade2-1 his3-11,15 trp1-1 leu2-3,112 rDNA::URA3
Diploid from CY184 X CY143
MATa trf4-102::TRP1 ura3-1 ade2-1 his3-11,15 trp1-1 leu2-3,112 rDNA::ADE2
Diploid CY858 disrupted with trf5-3::LEU2 at one allele
MATa ura3-1 ade2-1 his 3-11,15 trp1-1 leu2-3,112 rDNA::ADE2
MATa trf5-3::LEU2 ura3-1 ade2-1 his3-11,15 trp1-1 leu2-112,3 rDNA::ADE2
Diploid from CY855 CY857
MATa trf4-101::HIS3 trf5-3::LEU2 ura3-1 ade2-1 his3-11,15 trp1-1 leu2-3,112 rDNA::ADE2/pCB432
MATa trf4-101::HIS3 trf5-3::LEU2 ura3-1 ade2-1 his3-11,15 trp1-1 leu2-3,112 rDNA::URA3/pCB639 (trf4-ts2)
Media and growth conditions
Yeast strains were routinely grown in YEP with 2% glucose
(YPD). Thiabendazole (60 m g/ml) was added to YPD agar just prior
to pouring the plates. Unless otherwise stated, cells were grown at
DNA sequencing and mapping
A 3.0 kb fragment of genomic DNA containing TRF5 was cloned
into YEplac112 and sequenced by the chain termination method
(22) using a Sequenase kit (US Biochemicals, Cleveland, OH) and
[35S]dATP. The 3.0 kb XhoIXhoI insert from pSH4 was
sequenced on both strands using synthetic primers. To map TRF5,
this fragment was radiolabeled and hybridized to filters containing
l clones spanning the yeast genome (23).
Plasmid constructions (Table 2)
pSH2. pCB426 was digested with BamHI and the 10.6 kb fragment
carrying TRF5 was ligated into the BamHI site of YEplac112.
pSH3. pSH2 was digested with EcoRI and the 4.2 kb fragment
was ligated into the EcoRI site of YEplac112.
pSH4. pSH2 was digested with XhoI and the 3.0 kb fragment was
ligated into the SalI site of YEplac112.
pSH5. The LEU2 gene was excised from YEp13 by cutting with
XhoI and SalI and ligated into the SalI site of pSH4.
pSH6. pSH4 was digested with SalI and the ends were filled in
with Klenow enzyme then religated to create a +4 frameshift (f.s.)
mutation in TRF5.
pCB432. The plasmid containing the library clone of TRF4
(pCB427) was digested with SnaBI and partially with HindIII to
release a 3.7 kb fragment that was subsequently inserted into the
HindIII (left) and SmaI (right) sites of pRS316. The resulting
construct was designated pCB432.
pCB469. pCB432 was digested with BamHI, the first 288 amino
acids of Trf4p were removed and the TRP1 fragment was excised
from pCB53 with a BamHI/BglII digest and ligated to the deleted
TRF4 gene in pCB432, creating pCB469.
pCB635. A PCR product containing TRF4 and 396 nt upstream
of the ATG was cloned into the BamHI site of pRS314.
pCB693. Identical to pCB635 (above) except that it carries the
G317D mutation (trf4-ts2) in TRF4.
Precise mapping of the TRF5 locus
CY726 (top1-7::LEU2 trf4-1) carrying pBS3 (TOP1 URA3
ADE3 2u) was transformed with one of several subclones (pSH2,
pSH3, pSH4, pSH5 or pSH6), grown non-selectively for the
TOP1-containing plasmid (pBS3) and plated on 5-fluorotic acid
(5FOA) at 30 C. 5-FOA medium provides a URA3+
counterselection, so that cells that are unable to lose pBS3 will not grow
on this medium. The 10.6 kb fragment containing TRF5 (pSH2)
permitted growth of Ura segregants, indicating that TRF5
complements the synthetic lethality of top1 trf4 double mutants
in addition to the trf4-1 cs phenotype. The subclone containing a
4.2 kb EcoRI fragment (pSH3) did not complement. However, a
plasmid containing the 3.0 kb XhoIXhoI fragment (pSH4) did,
indicating that this fragment contained the TRF5 gene.
TRF4 URA3 CEN (library)
trf4101::HIS3 URA3 CEN
trf4-ts2 (G317D) TRP1 CEN
TOP1 ADE3 URA3 2u
Isolation of a temperature-sensitive trf4 allele
Plasmid pCB635 was mutagenized in vitro with hydroxylamine
(24) and transformed into yeast strain CY924 (trf5
trf4/pTRF4.URA3) by selection for Trp+. Transformants were
grown non-selectively for the pTRF4.URA3 plasmid and replica
printed to 5-FOA plates at 24 and 37 C. Colonies able to give rise
to 5-FOA-resistant segregants at 24 C but not at 37 C were
identified. Plasmid pCB693 was recovered from 5-FOA-resistant
segregants of trf4 ts candidates grown at 24 C and used to
retransform CY924 to confirm that trf4 ts activity was plasmid
associated. For one such isolate, the entire TRF4 open reading
frame was sequenced. The only mutation found was a GA
transition, resulting in a missense mutation of Gly317 to Asp
(G317D or ts2).
TRF5 overexpression complements a mutation in trf4
A 2m yeast genomic library was screened for plasmids that
suppressed the cs phenotype of a trf4-1 mutant strain
(CY667/pBS3) in order to clone the TRF4 gene. In addition to
TRF4 (12), a clone containing a different 10.6 kb insert based on
restriction mapping (pCB426) was identified. This gene was
designated TRF5. Subsequently, TRF5 was found to complement
the synthetic lethality between trf4 and top1 based on its ability
to allow a top1 trf4 double mutant to survive following loss of a
TOP1.URA3 plasmid and, thereby, to give rise to 5-FOA-resistant
segregants (Fig. 1A). Complementation of top1 trf4 synthetic
lethality was used to localize TRF5 to a 3.0 kb region (Fig. 1B).
A SalI site within this 3.0 kb region was used to generate a LEU2
Figure 3. Homology between the Trf5p C-terminus and the Top1p N-terminus. The alignment was generated using the GENWORKS sequence analysis software from
Intelligenetics. Boxed residues are those with a cost of 3.
insertion mutation (pSH5) and a frameshift mutation allele
(pSH6) in TRF5. Neither of these trf5 alleles complemented the
inviability of a top1 trf4 double mutant (Fig. 1A). These results
indicate that TRF5 is contained within the 3.0 kb XhoIXhoI
fragment on pSH4. Hybridization of this region to filters
containing DNA from contigs covering 90% of the yeast genome
(23) revealed that TRF5 mapped to the left arm of chromosome
XIV ~ 40 kb centromere distal to MET2.
TRF4 and TRF5 define a novel gene family
The 3.0 kb XhoIXhoI fragment containing TRF5 was sequenced
on both strands (GenBank accession no. U47282). The predicted
TRF5 open reading frame encodes a 625 amino acid protein (Fig. 2)
whose sequence is 57% identical and 72% similar to Trf4p. A
search of the GenBank database reveals the existence of a clear
S.pombe homolog of TRF4/5 identified in the S.pombe genome
project (Fig. 2). The predicted S.pombe homolog, called
SPAC12G12.13c, is 33% identical and 56% similar to Trf4p over
its C-terminal 683 amino acids. However, it also has a 570 amino
acid N-terminal extension relative to Trf4p that is highly
homologous to the G-b protein family and completely unrelated to Trf4p
or Trf5p. The function of this protein in S.pombe is not known.
Short regions of possible homology to Trf4p were also found
in two human expressed sequence tags (ESTs; Fig. 2). The first
human EST (accession no. H85548) encodes a predicted peptide
of 131 amino acids (H85548.pep in Fig. 2) that is 41% identical
and 60% similar to a region of Trf4p. The second EST (accession
no. H90950) encodes a predicted 130 amino acid peptide
(H90950.pep in Fig. 2) that is 44% identical and 65% similar to
a different region of Trf4p. The two EST peptides overlap but are
not identical at the DNA sequence level, indicating that they are
probably derived from different genes. In addition, two short
predicted peptides (59 and 24 amino acids) derived from a
Drosophila STS (accession no. G01479) are 42% identical and
61% similar to Trf4p (not shown). This sequence may traverse an
intron/exon boundary, since it not derived from cDNA. The
function of these putative TRF4 homologs is not known.
Examination of the Trf4p-related sequences with gcg programs
such as MOTIFS and with BLOCKS motif databases also failed
to identify meaningful signature regions.
In addition, limited sequence homology was also found between
a 91 amino acid N-terminal region of S.cerevisiae Top1p and a 92
amino acid C-terminal region of Trf5p (Fig. 3). These regions are
33% identical and 58% similar. The alignment, while limited, is
likely to be significant as it is 6.06 SD above the mean significance
of 20 randomized alignments of the same two sequences using the
gcg sequence analysis program BESTFIT. We previously reported
homology between the N-terminus of Top1p and the N-terminus
of Trf4p (12). Thus, a Top1p-related region exists at the N-terminus
of Trf4p, but at the C-terminus of Trf5p. The N-terminus of Top1p
is not required for catalytic activity (25) and may be involved in
some other aspect of topo I function.
TRF4 and TRF5 define an essential function that is
required at the time of mitosis
A null allele of TRF5 was constructed by inserting LEU2 into
TRF5 to make trf5-3::LEU2 (Fig. 1B). This construct was used to
disrupt one allele of TRF5 in a trf4/trf4 diploid (CY892).
Following sporulation, 40 tetrads were dissected. All tetrads
segregated two viable and two inviable spores, and all viable
spores were Leu, indicating that the trf4 trf5 double mutant is
inviable (Fig. 4). Because trf4 shows synthetic lethality with top1
(12), we tested for a genetic interaction between trf5 and top1.
However, in contrast to top1 trf4 double mutants, the top1 trf5
double mutant was viable (data not shown).
To examine the basis of the defect in cells deficient in TRF4/5
function, we constructed a conditional allele of trf4 by in vitro
mutagenesis with hydroxylamine (see Materials and Methods).
DNA sequence analysis showed the trf4 ts allele to be a change
of Gly317 to Asp. The mutation lies in a region conserved among
TRF4, TRF5 and SPAC12G12.13c. The trf4 (ts) trf5 double
mutant grows somewhat more slowly than the wild-type at
29.5 C, but arrests growth at 37 C (Fig. 5). At a semi-permissive
temperature of 32 C, the trf4 (ts) trf5 double mutant displays an
increased sensitivity to the anti-microtubule drug thiabendazole
(Fig. 5), suggesting that the common TRF4/5 function may occur
at the time of mitosis.
ase I) gene products serve overlapping or dependent functions in
S.cerevisiae, as the double mutant, but neither single mutant, is
inviable (12). TRF5 was isolated based on its ability to
complement the cold sensitivity of a trf4-1 point mutation and
encodes a second member of the TRF4 family. TRF5
overexpression also complements the inviability of a top1 trf4 double mutant
and the predicted Trf5 protein has a region that is homologous to
the N-terminus of Top1p. Putative TRF4 homologs from
S.pombe, Drosophila and human were identified in database
searches. trf4 trf5 double mutants are inviable, demonstrating that
TRF4/5 function is essential. A conditional trf4 (ts) trf5 mutant
was constructed and shown to be hypersensitive to the microtubule
poison thiabendazole when grown at a semi-permissive
temperature, as well as grossly defective in nuclear division at a
restricitve temperature, indicating that TRF4/5 function is
required at the time of mitosis.
Percentage of cells with each
To determine whether TRF4/5 function is required at mitosis,
a direct examination of nuclear morphology was performed on
the trf4 (ts) trf5 double mutant following a shift to a restrictive
temperature (37 C) for 3 h (Fig. 6). Three hours after a shift to
a restrictive temperature (37 C), 50% of the trf4 (ts) trf5 cells that
are undergoing nuclear division display an aberrant nuclear
morphology in which the dividing nucleus is primarily within the
mother cell (arrows in Fig. 6). This is similar to the so called cut
phenotype observed in certain S.pombe mutants that are defective
in nuclear division, in which nuclear division is stalled and the
nucleus is unequally distributed between mother and daughter
cells (4). In addition, anucleate and binucleate cells resulting from
a defective nuclear division are observed at greater frequency in
trf4 (ts) trf5 mutants than in the wild-type (Fig. 6, 37 C panel).
Quantitation of the defects in nuclear divison showed that in trf4
(ts) trf5 mutants shifted to 37 C for 3 h, 15 times more anucleate
cells and 18 times more cut-like cells are observed than in an
otherwise isogenic wild-type strain (Table3). Some of the trf4 (ts)
trf5 defects are apparent even at the permissive temperature. For
example, cells of the trf4 (ts) trf5 mutant are considerably larger
than TRF+ cells even at 24 C. In addition, the trf4 (ts) trf5 mutant
shows six times more anucleate cells and six times more cut-like
nuclei than the wild-type at 24 C. These results indicate that
TRF4/5 function is required for proper nuclear division.
We report the identification of a novel gene, TRF5, that is related
to the S.cerevisiae TRF4 (DNA topoisomerase I-related function)
gene product. The TRF4 and TOP1 (encoding DNA
Both Trf4p and Trf5p show a region of limited but significant
homology to the catalytically dispensable N-terminus of Top1p.
This region is not conserved among type I DNA topoisomerases
a50% of the cells that are in nuclear divison.
and its role in Top1p function is not known. The high percentage
of both positively and negatively charged residues in these regions
(average 35%) may indicate that their function is to mediate
interaction with chromatin, as similar regions are commonly found
in HMG proteins (26).
The Trf4 family of proteins does not show significant overall
homology with other proteins in the GenBank database (release 91)
that would suggest a biochemical activity, with the exception of
the S.pombe homolog, which shows a significant homology to
G-b proteins via its 570 amino acid N-terminal extension. G-b
family members encode proteins involved in many processes
(27), including initiation of DNA replication (CDC4),
chromosome separation (CDC20) and transcriptional repression (TUP1).
A 40 amino acid repeated sequence containing consecutive Trp
and Asp residues, termed the WD40 repeat, is found between five
and eight times in most family members and appears to mediate
proteinprotein interaction (28).
TRF4/5 function is required at the time of mitosis for proper
nuclear division. A critical and poorly understood event in the
chromosome cycle that occurs at mitosis is chromosome
condensation. Thus, a plausible hypothesis is that TOP1, TRF4 and
TRF5 are involved in mediating mitotic chromosome condensation.
This model is consistent with the proposed mitotic function for
TRF4/5. In addition, null mutations in trf4 are inviable in
combination with a mutation in SMC1 (29), the yeast homolog of
XCAP-C (30), a Xenopus protein that is required for mitotic
chromosome condensation in vitro (I. B. Castao and M. F.
Christman, unpublished observations). Taken together, these
results suggest that the TRF4 gene family may encode proteins
that mediate mitotic chromosome structure in some manner.
We thank Pius Brzoska and Nikki Levin for critically reading the
manuscript and Pragati Bakshi for help in preparing the
manuscript. I.B.C. was supported in part by a fellowship from the
Organization of American States and S.H.P. was supported by an
NIH training grant (CA09215) post-doctoral fellowship. This
work was supported by National Institutes of Health grant
R01GM46877 to M.F.C.