Regulation of topoisomerase II by phosphorylation: a role for casein kinase II

Eukaryotic DNA topoisomerase II is a nuclear enzyme which controls DNA topology by transiently breaking and rejoining DNA, relaxing both positive and negative supercoils and resolving knotted and catenated DNA molecules (reviewed by Gellert, 1981; Wang, 1985). The DNA decatenation activity of topoisomerase II is essential for the condensation of interphase chromatin into metaphase chromosomes (Uemura et al., 1987; Newport and Spann, 1987; Adachi et al., 1991; Wood and Earnshaw, 1990), the separation of sister chromatids in anaphase (DiNardo et al., 1984; Holm et al., 1985; Uemura and Yanagida, 1984, 1986; Shamu and Murray, 1992) and for proper chromosome segregation at meiosis I (Rose et al., 1990). Because of these specific roles in mitosis and meiosis, it is noteworthy that topoisomerase II is a major component of the metaphase chromosomal scaffold (Earnshaw et al., 1985; Gasser et al., 1986) and that it localizes to the synaptonemal complex and adjacent chromatin in pachytene cells from chicken, mouse and yeast (Moens and Earnshaw, 1989; Klein et al., 1992). Topoisomerase II also cofractionates with interphase nuclear scaffolds or matrices from various species (Berrios et al., 1985; Berrios and Fisher, 1988; Cardenas et al., 1990). Although it has been well established that topoisomerase II relaxes supercoils introduced in DNA templates by RNA and DNA polymerases during transcription and replication, particularly in cells defective for DNA topoisomerase I (Brill and Sternglanz, 1988; Kim and Wang, 1989a,b; Brill et al., 1987), any structural role topoisomerase II might have in the organization of the interphase nucleus remains to be clarified.

The essential role topoisomerase II plays at mitosis in the condensation and decatenation of DNA suggests that the enzyme activity may be regulated through the cell cycle. Considerable effort in recent years has been directed at elucidating the mechanisms involved in this regulation. The focus of this Commentary is to review the progress and to discuss models concerning topoisomerase II regulation by phosphorylation.

Three modes of post-translational regulation of topoisomerase II have been described: poly ADP-ribosylation, protein turnover and phosphorylation. While the situation has become somewhat confused in mammalian cells, due to the discovery that there are two DNA topoisomerase II isozymes (IIa and IIP; Drake et al., 1989), it appears that in yeast and Drosophila melanogaster there is but one enzyme, which more closely resembles the type IIa topoisomerase. For the sake of simplicity, we will assume that the studies we cite on topoisomerase II regulation in mammalian cells reflect primarily the more abundant a-form of DNA topoisomerase II, and that they are therefore comparable to studies in lower eukaryotes.

Regulation by poly ADP-ribosylation has been primarily studied in vitro. Topoisomerase II is known to be inactivated in vitro by poly ADP-ribosylation (Darby et al., 1985), a post-translational modification commonly found on nuclear proteins. It is not known, however, if topoisomerase II is poly ADP-ribosylated in vivo. The role of protein turnover in topoisomerase II regulation has been studied with chicken lymphoblastoid and mouse 3T3 cells. This work has shown that the turnover of topoisomerase II protein is closely related to the cell cycle and the proliferation status of these cells (Heck et al., 1988; Saijo et al., 1992). Topoisomerase II synthesis occurs as cells progress from G1 to S phase, resulting in a maximal level of protein at late G2 and mitosis. Following mitosis, about one third of the protein is degraded, at least in chicken cells, with a half-life of 1.8 h (Heck et al., 1988). Similarly, very low levels of topoisomerase II protein are detectable in resting or differentiated cells (as low as 300 copies per nucleus), and this level increases when cells are stimulated to divide (Heck and Earnshaw, 1986; Saijo et al., 1992). Whether this is a general mechanism of regulation is still in question, since in HeLa and yeast cells topoisomerase II levels are relatively stable throughout the cell cycle, fluctuating only 2-to 3-fold between G1 and M phases (Gasser et al., 1986; Kroll and Rowe, 1991; Cardenas et al., 1992). Moreover, in Drosophila embryos and yeast, the level of topoisomerase II does not seem to vary significantly with the proliferation state of cells, in contrast to the situation in vertebrate cells (Whalen et al., 1991; Cardenas et al.,1990).

In a wide range of organisms including Drosophila, sponges, chicken, mouse, human and yeast (Ackerman et al., 1988; Rottman et al., 1987; Heck et al., 1989; Saijo et al., 1990; Kroll and Rowe, 1991; Cardenas et al., 1992), topoisomerase II has been shown to be phosphorylated in vivo. A number of protein kinases, including casein kinase II (CKII; Ackerman et al., 1985), protein kinase C (PKC; Sahyoun et al., 1986; Rottman et al., 1987), Ca²⁺/calmodulin-dependent protein kinase (Sahyoun et al., 1986; Cardenas et al., 1992) and p34^cdc2 kinase (Cardenas et al., 1992) can modify topoisomerase II in vitro, and in all cases phosphorylation stimulates the ATP-dependent decatenation activity. While topoisomerase II is a suitable substrate for many kinases in vitro, it is essential to ask which of these kinases actually has access to topoisomerase II in the cell nucleus.

The phosphorylation of topoisomerase II by PKC in vitro prompted an examination of whether this occurs under conditions where PKC activity is stimulated in intact cells. Because inhibition of topoisomerase II activity by novobiocin and m-AMSA blocks the phorbol ester-induced differentiation of HL-60 cells (Sahyoun et al., 1986), a process in which PKC activation by diacylglycerol is thought to play a central role, it has been speculated that topoisomerase II activation by PKC mediates some of the nuclear effects of this kinase (Sahyoun et al., 1986). Phosphorylation of topoisomerase II by PKC was also proposed as a mediator of the mitogenic stimulation triggered upon exposure of dissociated sponge Geodia cydonium cells to aggregation factor (Rottman et al., 1987). However, in both of these studies no direct evidence is given to show that topoisomerase II is phosphorylated by PKC in vivo, and indeed, there is no evidence to date that PKC is localized to the nucleus. In the case of HL60 cell differentiation, the block imposed by anti-topoisomerase II drugs could be attributed to the cell cycle arrest effected by these drugs (reviewed by Liu, 1989). In conclusion, a role for PKC in the regulation of topoisomerase II in vivo remains to be established.

Cell cycle phosphorylation studies in chicken lymphoblastoid and yeast cells revealed that phosphorylation of topoisomerase II increases 6-to 10-fold as cells progress through G2 and enter mitosis (Heck et al., 1989; Cardenas et al., 1992). In mouse 3T3 cells phosphorylation was also observed to peak in late G2, and to remain elevated in mitosis, mimicking protein levels yet occurring independently of protein synthesis (Saijo et al., 1992). These results, and the finding that soluble mitotic extracts from HeLa cells contain more topoisomerase II activity than S-phase extracts (Estey et al., 1987; see Tricoli et al., 1985, for an alternative observation), support models in which phosphorylation activates topoisomerase II just prior to and during mitosis.

An extensive study with metabolically ³²PO4-labelled yeast cells arrested at different points of the cell cycle has identified casein kinase II as the major kinase responsible for phosphorylation in vivo (Cardenas et al., 1992). In addition, the phosphoacceptor sites were mapped within the yeast topoisomerase II protein. Two-dimensional phosphopeptide maps revealed that the same topoisomerase II tryptic peptides are phosphorylated in G1 and in mitosis, but that the levels of modification vary significantly. Some acceptor sites are more highly phosphorylated at mitosis than in G1, while the converse is true for some of the major G1 phosphoacceptors sites (Cardenas et al., 1992). Most importantly, all but two of the major in vivo phosphoacceptor peptides comigrate with peptides phosphorylated by CKII in vitro.

The two phosphorylated peptides that do not comigrate with the CKII-modified peptides could reflect differences in the in vitro versus in vivo specificity of casein kinase II, or may result from phosphorylation by another kinase(s). The latter possibility seems to be in contradiction to the observation that no phosphate is incorporated into topoisomerase II in cells lacking CKII activity (Cardenas et al., 1992). However, in several instances, CKII phosphorylation was shown to be required for phosphorylation by a second kinase (see Roach, 1991, for a review). An intriguing possibility is that CKII phosphorylation of topoisomerase II is a requirement for phosphorylation by a mitotic-specific kinase, such as p34^cdc2 kinase, for which yeast topoisomerase II is also a good substrate in vitro (Cardenas et al., 1992). Further peptide mapping studies will be required to identify which kinases other than CKII are involved in topoisomerase II modification in intact cells.

Taken together, these recent results clearly establish CKII as the major kinase phosphorylating topoisomerase II in vivo. In all species examined, CKII is a tetramer formed by two a, or an a and an a’, catalytic subunits and two 0 regulatory subunits (aa’02 or a.202) (for recent review see Pinna, 1990). In yeast the two a subunits are encoded by two different genes, CKA1 and CKA2 (Chen-Wu et al., 1988; Padmanabha et al., 1990). Double mutants in which both CKA1 and CKA2 have been disrupted are inviable, and cka1-cka2-8 temperature-sensitive mutants shifted to the non-permissive temperature arrest growth as half unbudded and half large-budded cells, depending on where the cells were in the cell cycle at the time of the temperature shift (D. Hanna and C. V. C. Glover, personal communication). The large-budded cell arrest phenotype is characteristic of G2-M arrest and could arise from failure of CKII to activate topoisomerase II and other enzymes required to complete DNA replication. In the cka1-cka2-8 ts mutant the phosphorylation of topoisomerase II is greatly reduced at the non-permissive temperature, confirming that CKII plays a major role in this modification (Cardenas et al., 1992). This role for casein kinase II in the activation of topoisomerase II was previously proposed from one-dimensional phosphopeptide maps of the Drosophila enzyme (Ackerman et al., 1988), and is expected to hold true for many species, given that both topoisomerase II and CKII have been highly conserved through evolution. Consensus sites for CKII phosphorylation are present in the C termini of human (a form), Schizosaccharomyces pombe and Drosophila type II topoisomerases (Cardenas et al., 1992).

Two mechanisms might account for the increased phosphorylation of topoisomerase II at mitosis: either CKII activity is stimulated or an interphase-specific phosphatase is inactivated. The first model is supported by the observations that the CKII a and 0 subunits are phosphorylated in intact chicken and human cells at sites phosphorylated in vitro by p34^cdc2 kinase (Litchfield et al., 1991, 1992). Interestingly, these phosphoacceptor sites are hyperphosphorylated in cells arrested at mitosis (Litchfield et al., 1992). Although in these studies CKII activity derived from cells arrested in interphase or mitosis was similar, in a previous report phosphorylation in vitro of Xenopus laevis CKII by p34^cdc2 kinase resulted in 1.5-to 5-fold stimulation in CKII activity (Mulner-Lorillon et al., 1990). Alternatively, rather than simply stimulating activity, one could imagine that the mitotic-specific phosphorylation by p34^cdc2 modulates the affinity of CKII for some substrates or enhances the nuclear localization of a particular isozyme (eg. Yu et al., 1991), thus favoring substrates hyperphosphorylated at G2-M.

The second possibility is that the phosphates on CKII-target sites are constantly and rapidly turning over. Thus a balance between CKII and a nuclear phosphatase would determine the net level of topoisomerase II phosphorylation. One would then propose that a phosphatase must be inactivated as cells enter mitosis, and as a result the net modification of a substrate like topoisomerase II would be elevated. One such candidate for this function is protein phosphatase 1 (PP1). A recent study of protein phosphatase 1 in Xenopus embryos shows that the activity peaks in interphase and decreases shortly before the onset of mitosis (Walker et al., 1992). This study stresses the role of the phosphatase in determining the timing of exit from S phase and entry into M phase. This is supported by the fact that overexpression of PP1 in a S. pombe weel^-/cdc25ts double mutant at the restrictive temperature causes a G2 arrest (Booher and Beach, 1989). Several other studies have documented a role for PP1 in the exit from mitosis, as judged by the arrest in mid-mitosis of ppl- mutants (Doonan and Morris, 1989; Axton et al., 1990; Kinoshita et al., 1990) and the mid-mitotic block induced by a mutation in a positive regulator of PP1 (Ohkura and Yanagida, 1991). The regulation of the phosphatase may itself be brought about through the interplay of positive regulators and an inhibitor (Inh-2), whose levels were shown to oscillate through the cell cycle (Brautigan et al., 1990).

A computer search for the CKII consensus substrate motif, Ser/Thr-X-X-Glu/Asp or generally an acidic amino acid (see Pinna, 1990 for review), identified 23 potential phosphorylation sites in topoisomerase II. 18 of these 23 sites lie within the last 350 C-terminal amino acids of topoisomerase II and most are clustered such that their modification would result in multiply phosphorylated tryptic peptides (Fig. 1). Immunoprecipitation of CKII-labelled and hydroxylamine-cleaved topoisomerase II with domainspecific antibodies demonstrated that all of the CKII phosphorylated sites are indeed located within the C-terminal 350 amino acids (Cardenas et al., 1992). Based on twodimensional phosphoamino acid analysis and two dimensional phosphopeptide patterns, the CKII target sites in topoisomerase II have been tentatively assigned to four multiply phosphorylated tryptic peptides at amino acid positions 1085 to 1121, 1359 to 1289, 1350 to 1382, and 1398 to 1429 (peptides 6, 8, 9, and 10, shown in Fig. 1). The other six potential phosphate-accepting peptides did not comigrate with the peptides phosphorylated by CKII in vivo or in vitro. These C-terminal acceptor sites are arranged in a hierarchal phosphorylation cascade, such that kinase action at a C-terminal site generates an adjacent, upstream site for CKII modification, which in turn generates yet another consensus site (Pinna, 1990; Roach, 1991, for reviews).

Eukaryotic type II topoisomerases contain roughly 1400 amino acids, of which the first 900 N-terminal amino acids show a high degree of sequence homology with bacterial gyrase (Fig. 1). Regions of significant homology can also be aligned up to amino acid 1170 in the Saccharomyces cerevisiae gene, but thereafter homology with the bacterial gyrase disappears, and even among eukaryotic species the region from amino acid 1170 to the end of the protein is highly divergent (Caron and Wang, 1992). Thus, a priori, this C-terminal, phosphate-accepting domain could represent a part of the protein involved in regulating the aspects of topoisomerase II function which are unique to eukaryotic cells and which may require species-specific partners.

Analyses of truncated forms of yeast topoisomerase II has provided intriguing insights about the importance of the C-terminal domain and its phosphorylation. Drastic truncations up to amino acid 979 of the S. cerevisiae TOP2 gene, or up to amino acid 922 in the S. pombe Top2 gene, do not allow rescue of top2 ts or deletion strains (Shiozaki and Yanagida, 1991; M. E. C. and S. M. G., unpublished). In contrast, a C-terminal truncation of the S. pombe topoisomerase II to amino acid 1198, which removes most of the potential CKII phosphoaceptor sites, partially complements a S. pombe top2ts mutation and a null mutant, allowing conditional growth, yet fails to complement another top2cs mutation (Shiozaki and Yanagida, 1991). Surprisingly, protease digestion studies show that a proteolytic core of the enzyme that extends only to amino acid 1220 retains the full decatenation activity of topoisomerase II (Shiozaki and Yanagida, 1991). Similarly, a truncation of the S. cerevisiae enzyme to amino acid 1167, produces a topoisomerase II dimer that is fully active in vitro, but which complements a top2 disruption poorly in vivo (Caron, P. and Wang, J., personal communication). Paradoxically, when purified, these C-terminally truncated enzymes are highly active in vitro (Shiozaki and Yanagida, 1991; Caron, P. and Wang, J., personal communication), while the full-length, dephosphorylated topoisomerase II is nearly inactive (Cardenas et al., 1993; Saijo et al., 1990). Since phosphorylation by CKII suffices to reactivate the full-length enzyme, we propose that the C-terminal 259 amino acids serve as a negative regulatory domain that can be neutralized by phosphorylation. Without this inhibitory domain (eg. among the prokaryotic or truncated eukaryotic enzymes) topoisomerase II should not require phosphorylation to be active, and should be insensitive to the stimulatory effects of the kinases. This hypothesis remains to be tested.

A precedent for this type of regulation has recently been described for human DNA ligase I (Prigent et al., 1992). Expression of the full-length human DNA ligase I gene in Escherichia coli cells is unable to rescue conditional lethal ligase mutants. However, expression of an N-terminally truncated DNA ligase I allows the mutant E. coli cells to grow at the restrictive temperature (Kodama et al., 1991). When both forms of ligase l were purified from E. coli, only the truncated form was active in vitro (Kodama et al., 1991) , yet phosphorylation of the full-length E. coli-made enzyme on CKII acceptor sites in the N terminus rendered the enzyme active (Prigent et al., 1992). As for topoisomerase II, it was shown that mammalian DNA ligase l is a phosphoprotein in intact cells. A comparison of the model presented for DNA ligase I, and the model we propose for yeast topoisomerase II is presented in Fig. 2.

Why does topoisomerase II have a mechanism for the negative regulation of its activity? Clearly the relaxation activity of DNA topoisomerases (type I or II) is required throughout the cell cycle, as long as there is transcription or replication occurring, and it is not clear that an excess of topoisomerase activity would be detrimental to the cell. Nonetheless, the observation that C-terminal truncations of yeast topoisomerase II do not fully complement null mutants (see above) suggests that either the down-regulation itself is important, or that the C terminus has a secondary role. Perhaps when the C terminus of topoisomerase II is phosphorylated at metaphase, it becomes extended or its conformation changes (see Fig. 2), such that it can interact with other proteins or other molecules of topoisomerase II. A similar event could also occur for DNA ligase 1; i.e. phosphorylation of its N terminus might promote interaction with other proteins of the replication complex. The dephosphorylation of topoisomerase II, which appears to occur in late mitosis, may both down-regulate its activity and counteract this protein-protein interaction. Such contacts could allow topoisomerase II to aid in chromosome condensation, which occurs precisely when the enzyme is maximally phosphorylated. While speculative, this hypothesis is now readily testable.

CKII has been proposed to be an effector of signals to the nucleus that modulate cell proliferation. Although alternative results have been presented (see Litchfield et al., 1992), it has been reported that the stimulation of intact cells with mitogenic agents, including epidermal growth factor, insulin and insulin-like growth factor, increases CKII activity (Sommercorn et al., 1987; Klarlund and Czech, 1988; Ackerman et al., 1990). While various interpretations might be given for these results, it is nonetheless convincingly shown that CKII modifies a large number of nuclear proteins known to play key roles in cell growth and proliferation control. These substrates include the proto-oncogene proteins c-Myb, c-Fos, c-Jun and c-ErbA, the oncogene product E1a, the tumor suppressor p53, SV40 large T antigen, the cAMP response element binding factor, serum response factor, Max, DNA ligase I, DNA topoisomerase II, and the nucleolar substrates B23, nucleolin, mUBF and potentially Nopp140 (reviewed by Meisner and Czech, 1991; Hunter and Karin, 1992; see also Berberich and Cole, 1992; Voit et al., 1992; Prigent et al., 1992; Meier and Blobel, 1992; Ackerman et al., 1988; Cardenas et al., 1992). For many of these substrates, phosphorylation by CKII is known to influence their biological activities, yet the results of these modifications are not uniform, even within the same functional class of substrates.

For example, elimination of the CKII phosphorylation sites in the proto-oncogene proteins c-Myb, c-Fos and c-ErbA is associated with proto-oncogene to oncogene conversion. In contrast, in v-ErbA the same modification markedly reduces transformation efficiency (see Meisner and Czech, 1991, for review). For SV40, large T antigen mutagenesis of the acceptor serine had little effect on SV40 transforming potential (Barbosa et al., 1990). In the case of serum response factor, phosphorylation by CKII increased its DNA dissociation constant from the serum response binding element, which is located upstream of genes like c-fos, whose transcription is activated upon stimulation of cells with growth factor (Manak and Prywes, 1991; Marais et al., 1992). Like c-Myb (Lüscher et al., 1990), the binding of c-Jun to DNA is negatively regulated by CKII (Lin et al., 1992), while DNA topoisomerase II and DNA ligase 1 are positively regulated by similar modifications, and the acidic C-terminal tail of mUBF is an essential positive regulatory domain activated by CKII phosphorylation (Voit et al., 1992). Thus, despite advances implicating CKII in either the control or response to cell growth, the question remains: is casein kinase II an “intelligent” kinase, or are the critical cell cycle and proliferation signals controlled through specific phosphatases and modulators that do the fine-tuning on a ubiquitous, broad-spectrum and more-or-less unregulated kinase? At least as far as DNA replication goes, we can expect that the study of well characterized, essential nuclear enzymes, like DNA topoisomerase II and DNA ligase 1, will shed some light on the function of casein kinase II in the successful completion of replication prior to mitotic division.

We would like to thank Drs Paul Caron and James Wang for communicating and allowing us to include unpublished results, and Drs Tomas Lindahl, James Wang and Joseph Heitman for comments on the manuscript. M. E. C. thanks the Roche Research Foundation for its support. Research in the laboratory of S. M. G. was supported by a Human Frontiers grant, the Swiss National Science Foundation and the Swiss Cancer League.