The checkpoint transcriptional response

Abstract

The replication checkpoint signaling network monitors the presence of replication-induced lesions to DNA and coordinates an elaborate cellular response that includes ample transcriptional reprogramming. Recent work has established two major groups of replication stress-induced genes in Saccharomyces cerevisiae, the DNA damage response (DDR) genes and G₁/S cell cycle (CC) genes. In both cases, transcriptional activation is mediated via checkpoint-dependent inhibition of a transcriptional repressor (Crt1 for DDR and Nrm1 for CC) that participates in negative feedback regulation. This repressor-mediated regulation enables transcription to be rapidly repressed once cells have dealt with the replication stress. The recent finding of a new class of CC genes, named “switch genes,” further uncovers a mode of transcription regulation that prevents overexpression of replication stress induced genes during G₁. Collectively, these findings highlight the need for mechanisms that tightly control replication stress-induced transcription, allowing rapid transcriptional activation during replication stress but also avoiding long-term hyperaccumulation of the induced protein product that may be detrimental to cell proliferation.

Introduction

Stress generated during the process of DNA replication is arguably one of the most important challenges that proliferating cells need to cope with in order to ensure cell survival and the maintenance of genomic integrity.^1-5 Every cell going through S-phase experiences a certain level of replication stress generated either through the collisions of the replication fork with transcriptional machineries localized in highly transcribed regions of the DNA (e.g., rDNA and tRNA genes), through the replication of “hard-to-replicate” repetitive DNA sequences, by limited supplies of dNTPs, or when the replication fork encounters DNA templates damaged by endogenous or exogenous factors (e.g., reactive oxygen species and UV radiation, respectively).^6-12 Transcriptional activation of selected genes is an important part of the cellular response that allows cells to better cope with DNA replication stress. However, this transcriptional activation needs to be tightly regulated, as overexpression of some proteins may adversely affect the ability of cells to progress in the cell cycle once the stress ceases.

Here we discuss our current understanding of the molecular circuitries that control transcription levels of DNA replication stress-induced genes. Our recent work, published in back-to-back papers with the Wittenberg group in The EMBO Journal, establishes two major groups of replication stress-induced genes in Saccharomyces cerevisiae, the DNA damage response (DDR) genes and G₁/S cell cycle (CC) genes. In both cases, transcriptional activation is mediated via checkpoint-dependent inhibition of a transcriptional repressor (Crt1 for DDR and Nrm1 for CC) that participates in negative feedback regulation.^13,14 This repressor-mediated regulation enables transcription to be rapidly repressed once cells have dealt with the replication stress and shares similarities with transcriptional repression mechanisms found in bacteria and mammals.^15,16 Our recent discovery of a select group of CC promoter genes that undergo a transcriptional factor switch during the G₁-to-S transition (which we named the “switch genes”) further illustrates the need for tight control of replication stress-induced genes.

Sensing Replication Stress via DNA Damage Checkpoint Kinases

Replication stress-induced transcriptional responses are mostly regulated by phosphorylation-mediated signals emanated by DNA damage checkpoint kinases. DNA damage checkpoint kinases are evolutionarily conserved proteins that constantly monitor DNA structure during any stage of the cell cycle.^17,18 The involvement of checkpoint kinases in specifically monitoring DNA structures during DNA replication is often referred to as the replication checkpoint. Many reviews specifically focus on the DNA checkpoint kinases and their multiple roles in the DNA damage response, and we refer to them for more detailed information about their role in controlling transcription-independent responses.^19-21 At the heart of the DNA damage checkpoint in human cells lays the phosphatidyl-inositol-3-kinase-like sensor kinases ATR/ATM and the downstream effector kinases CHK2/CHK1. ATR and ATM are at the top of the signaling cascade and act partially redundantly to phosphorylate common substrates at S/T-Q motifs.²² ATR and ATM phosphorylate and control the activity of CHK1 and CHK2.^17,23-26 ATM and CHK2 are mainly activated by DNA double-strand breaks (DSBs), whereas ATR and CHK1 respond to a wide variety of replication-induced lesions that lead to the accumulation of single-stranded DNA (ssDNA). Although there is some degree of crosstalk between the two kinases, ATR has a preponderant role in replication checkpoint signaling. In mammalian cells, ATR and CHK1 are essential kinases that phosphorylate an extremely complex network of substrates; however, how these phosphorylations regulate the replication stress response is not fully understood.²⁷

Budding yeast is a very useful system to study the replication checkpoint due to a high degree of conservation of the main checkpoint kinases. In addition to its powerful genetics, the use of budding yeast allows the deletion of the functional counterparts of human ATR and CHK1 (Mec1 and Rad53, respectively) in the presence of a rescuing deletion of the ribonucleotide reductase inhibitor Sml1.²⁸ In contrast, deletion of these essential kinases leads to lethality in most model organisms including human tissue culture cells. The architecture of the pathway in budding yeast and humans is strikingly similar. In the same way to ATR and CHK1, Mec1 and Rad53 are central to the replication stress response, and dysfunctions in these kinases cause hypersensitivity to replication stress and induce genomic instability.^{17,18,24,29-31} Mec1 and Rad53-dependent signaling is crucial to mediate stabilization and repair of replication forks, inhibition of late origin firing, increased production of deoxyribonucleotides (dNTPs), among other responses.^{1,9,10,14,28,31-38} While the dramatic loss of viability observed in cells lacking Mec1 and/or Rad53 exposed to replication stress is related to translation-independent functions, the identification of multiple functional connections between Mec1/Rad53 signaling and transcription highlight the central importance of regulating transcription during replication stress.^30,31,39,40

Checkpoint-Dependent Transcriptional Regulation in Yeast

Since the very early work in the 80s monitoring gene expression in response to DNA damage, it was known that genotoxicity leads to repression and activation of different subsets of genes.^41-45 This became clear in the late 90s with the use of high-throughput microarray technologies.⁴⁶ Genome-wide expression analysis in yeast has revealed that in response to replication stress induced by methyl methane sulphonate (MMS), ionizing radiation (IR), hydroxyurea (HU) and UV radiation, hundreds of genes are either induced or repressed.^46-48 Consistent with the notion that the DNA damage checkpoint is a central trigger for most of the transcriptional responses to replication stress, the vast majority of the observed replication-specific changes in expression were indeed found to be dependent on the Mec1/Rad53 kinase-signaling pathway.^47,49 This pathway can be divided into two major branches, one mediated by the most downstream checkpoint protein kinase Dun1, a kinase that is directly phosphorylated and activated by Rad53, while the other branch is Dun1-independent (Fig. 1A).^13,14,50,51

Figure 1. (A) Checkpoint-dependent transcriptional pathways in budding yeast. Upon genotoxic stress, Rad53 de-represses transcription of DNA damage response (DDR) and G₁/S cell cycle (CC) genes via Dun1/Crt1 and Nrm1/MBF pathways, respectively. Crt1 and Nrm1 auto-regulation is indicated. (B) Crt1 and Nrm1 activity is regulated by a negative feedback loop. Upon DNA replication stress, Crt1 and Nrm1 repressors (indicated by the letter “R”) are phosphorylated in a checkpoint dependent manner. Phosphorylation releases Crt1 and Nrm1 from target promoters including their own promoters, leading to transcription activation. Increased transcription results in a continues production of unphosphorylated Crt1 and Nrm1 during the stress response, which allow for rapid downregulation of transcription after genotoxic stress is resolved.

Transcriptional Regulation of DNA Damage Response (DDR) Genes

The first set of co-regulated genes induced as part of the DNA replication and damage checkpoint response was identified in S. cerevisiae and named the DNA damage response (DDR) genes. This small group of genes includes genes that encode for the subunits of ribonucleotide reductase (RNR2, RNR3 and RNR4) and additional genes whose functions are not well understood, such as HUG1, FSH3, YLR345W and NTH2.^14,52 The initial study, performed in the Elledge lab, showed that the mechanism of DDR transcriptional regulation involves Dun1-dependent phosphorylation and inactivation of the transcriptional repressor Crt1. Crt1 is bound to the promoters of DDR genes, as well as to its own promoter, and is released upon phosphorylation. Therefore, Dun1-dependent phosphorylation of Crt1 results in de-repression and consequent activation of DDR genes.¹⁴ The Dun1-Crt1 regulatory circuit is not conserved in higher eukaryotes, and in budding yeast it represents only a small part of the checkpoint-dependent transcriptional response.

Transcriptional Regulation of G₁/S Cell Cycle (CC) Genes during Replication Stress

Recent work revealed that the transcription of a large group of cell cycle-dependent genes known as G₁/S genes is also induced as part of the DNA replication stress transcriptional response in S. cerevisiae.^13,51 Under unperturbed conditions, the transcription of G₁/S cell cycle genes peaks during the transition from G₁ to S-phase, and as cells progress into S-phase the transcription of G₁/S cell cycle genes is inactivated (for review on G₁/S genes, see ref. 53). Under replication stress, a number of G₁/S genes exhibited sustained transcription in S phase. This group of replication stress-induced G₁/S genes includes close to 100 genes and consequently represents one of the largest single groups of co-regulated targets among DNA-damage induced genes.^13,51 It is well established that G₁/S transcription depends on the SBF and MBF transcription factor complexes that regulate over 200 cell cycle genes.^53,54 SBF and MBF are comprised of the common regulatory component Swi6 and a specific DNA binding protein, Swi4 or Mbp1, respectively.^53,55 Swi4 and Mbp1 recognize distinct DNA sequence elements named SCB and MCB, respectively, which define SBF and MBF target promoters. Whereas the majority of G₁/S targets are regulated by either SBF or MBF, a small subset is thought to be regulated by both.^56-58 SBF has been shown to be required for activating transcription during G₁, while MBF is involved in repressing G₁/S transcription outside of G₁, thereby confining expression to the G₁ phase of the cell cycle.^56,59,60

Our recent work, published in The EMBO Journal, revealed that CC genes that depend on MBF for inactivation once cells progress into S phase are transcriptionally induced upon replication stress. These two studies show that MBF targets, but not SBF targets, are upregulated via checkpoint-dependent inactivation of the Nrm1 co-repressor. The regulation of Nrm1 was found to be Rad53-dependent but Dun1-independent, and is clear that it represents a pathway that is parallel to the Rad53-Dun1-Crt1-DDR pathway downstream of Rad53. Interestingly, similarly to Crt1, Nrm1 also targets its own promoter and represses its expression.

Checkpoint Interference of Negative Feedback Loops

As described above, during DNA replication stress, part of the G₁/S cell cycle transcription program is maintained in an active state by interference of a negative feedback-loop involving the transcriptional repressor Nrm1.^61,62 Interestingly, the repressor Crt1, involved in regulation of the DNA damage-inducible genes also participates in a negative feedback response. The capacity of both Nrm1 and Crt1 to repress their own expression allows rapid downregulation of the transcription of their target after cells recovery from a DNA replication arrest (Fig. 1B). Overall, interference of transcriptional repressors that participate in a negative feedback, as part of the checkpoint transcriptional response, appears to be conserved throughout evolution. In E. coli, LexA mediates feedback regulation during the SOS response and mammalian Mdm2, a negative regulator of p53, mediates feedback regulation of transcription during recovery from the DNA damage checkpoint response.^15,16 The conservation of this particular network wiring shows that the transcriptional response initiated by DNA damage and replication stress needs to be rapidly repressed once those problems have been rectified.

“Switch Genes:” A New Class of G₁/S Cell Cycle Genes

Our work established that only G₁/S genes that depend on MBF for repression during S phase are induced by replication stress. Interestingly, we found that this set of genes included not only G₁/S genes solely regulated by MBF throughout the cell cycle, but also a new subset of G₁/S genes regulated by SBF during G₁ and by MBF during S-phase.¹³ This subset of genes was named “switch genes” based on the SBF-to-MBF switch at their promoters during the G₁-to-S transition. The SBF-to-MBF switch was first established for the G₁/S gene TOS4. Since Tos4 protein accumulates in response to replication stress, like MBF-dependent genes, but was previously identified as a target of SBF, detailed analysis was performed to establish its transcriptional mode of regulation. This revealed that peak expression of TOS4 during G₁ depends on SBF, whereas repression outside of G₁ requires MBF. Chromatin immunoprecipitation (ChIP) experiments showed that SBF binds the TOS4 promoter during G₁ but is subsequently replaced by MBF once cells progress into S phase. Analysis of the TOS4 promoter revealed that it contains an overlapping MCB/SCB motif that maintains both consensus-binding sequences (Fig. 2A). Further analysis established that in the absence of Swi4, Mbp1 is able to further bind the TOS4 promoter during G₁, indicating that Swi4 competes for binding at the overlapping site during G₁. Since Swi4 leaves G₁/S target promoters once cells progress into S phase, in a Clb/CDK dependent manner, Mbp1 is able to bind the TOS4 promoter at the overlapping MCB/SCB site. Analysis of SBF and MBF target promoters identified in three genome-wide binding studies, resulted in a list of 44 genes containing an overlapping MCB/SCB site, which were annotated as putative switch genes (Table 1).^54,63,64 Additional analysis of these genes confirmed that the majority are cell cycle-regulated, induced in response to HU treatment and display an SBF-to-MBF switch at their promoters. But what is the biological relevance of this SBF-to-MBF switch mechanism? We propose that G₁/S genes that need to be induced in response to replication stress, but whose constitutive expression is detrimental to cell cycle progression, would benefit from an SBF-to-MBF switch regulation. In circumstances where MBF function is compromised, this switch regulation would work as a “fail-safe” mechanism, allowing the switch-genes to behave as a replication stress-inducible gene during S phase but, differently from a MBF-only regulated gene, will avoid constitutive high expression and hyper-accumulation (Fig. 2B). Consistent with this notion, switch genes are significantly enriched for dosage-sensitive genes reported to result in cell cycle delay and/or growth defects when overexpressed.^65,66 Moreover, analysis of transcription induction upon replication stress showed that this group of genes is enriched for genes that are upregulated during replication stress conditions.^13,51

Figure 2. (A) The prototypical SBF (SCB) and MBF (MCB) overlapping motif identified in the TOS4 promoter (-232 to -214 from the ATG start codon). Grey shade represents SCB/MCB overlapping region. (B) SBF-to-MBF switch mechanism avoids protein accumulation outside of S-phase. During replication stress, G₁/S genes regulated by MBF or SBF-to-MBF switch are repressed during G₁ and being kept active during S-phase. However, in case of MBF malfunction the SBF-to-MBF switch on G₁/S promoters provides a fail-safe mechanism that allows gene expression during S-phase but prevents loss of periodicity and hyper-accumulation of dosage sensitive genes outside of S-phase. Star represents MBF malfunction.

Table 1. List of putative switch genes.

#	ORF (SGD)	Switch Gene	Function	Up in HU?		Overexpression (Phenotype)	Cell Cycle Gene?
									de Oliveira 2012	Travesa 2012		Travesa 2012	Spellman 1998	Orlando 2008
1	YBR070C	ALG14	Glycosylation	√	√	not observed	√	√	√
2	YBR071W	YBR071W	Unknown	-	√	vegetative growth: decreased rate	√	√	√
3	YBR162C	TOS1	Unknown	X	-	not observed	√	X	√
4	YCR065W	HCM1	Transcription factor	√	√	not observed	√	√	√
5	YDL003W	MCD1	Cohesin subunit	√	√	vegetative growth: decreased rate	√	√	√
6	YDL055C	PSA1	Cell wall biosynthesis	X	√	not observed	√	√	X
7	YDL127W	PCL2	Cyclin	X	-	not observed	√	√	√
8	YDR222W	YDR222W	Unknown	√	X	vegetative growth: decreased rate	√	X	√
9	YDR224C	HTB1	Histone H2B	-	X	vegetative growth: decreased rate	√	√	√
10	YDR225W	HTA1	Histone H2A	-	X	vegetative growth: decreased rate	√	√	√
11	YDR501W	PLM2	Homolog of TOS4	-	-	vegetative growth: decreased rate	√	√	√
12	YER001W	MNN1	Glycosylation	-	√	not observed	√	√	√
13	YER111C	SWI4	Subunit of the SBF complex	√	√	abnormal cell cycle progression in G2	√	X	√
14	YER112W	LSM4	RNA processing	-	X	not observed	√	X	X
15	YGL096W	TOS8	Putative transcription factor	X	-	vegetative growth: decreased rate	√	X	X
16	YGL179C	TOS3	Kinase: activates Snf1p	√	√	not observed	√	X	√
17	YGR140W	CBF2	Kinetochore protein	√	X	vegetative growth: decreased rate	√	√	√
18	YGR151C	YGR151C	Unknown	-	√	not observed	√	√	X
19	YGR153W	YGR153W	Unknown	-	-	not observed	X	X	X
20	YIL123W	SIM1	Unknown	√	-	not observed	√	X	√
21	YIL141W	YIL141W	Unknown	√	-	not observed	√	√	X
22	YJL187C	SWE1	Kinase: inhibits Cdc28	√	√	vegetative growth: decreased rate	√	√	√
23	YJR054W	ERM6	Vacuolar protein	X	-	vegetative growth: decreased rate	√	√	√
24	YKL008C	LAC1	Ceramide synthesis	√	X	not observed	√	√	√
25	YKL102C	YKL102C	Unknown	X	-	not observed	√	X	X
26	YLR183C	TOS4	Checkpoint effector, HDAC regulator	√	√	abnormal cell cycle progression in G1/S	√	√	√
27	YLR332W	MID2	Cell wall integrity	-	-	vegetative growth: decreased rate	√	X	X
28	YMR144W	YMR144W	Unknown	-	√	not observed	√	√	√
29	YNL231C	PDR16	Lipid homeostasis	√	√	vegetative growth: decreased rate	√	√	√
30	YNL278W	CAF120	Transcription factor	√	X	vegetative growth: decreased rate	√	X	√
31	YNL283C	WSC2	Cell wall integrity	√	-	vegetative growth: decreased rate	√	√	√
32	YNL301C	RPL18B	Ribosomal protein	X	-	not observed	√	X	X
33	YOL007C	CSI2	Unknown	X	√	vegetative growth: decreased rate	√	√	√
34	YOR247W	SRL1	Cell wall integrity	√	X	vegetative growth: decreased rate	√	√	√
35	YOR342C	YOR342C	Unknown	X	-	not observed	√	X	√
36	YPL126W	NAN1	Ribosomal protein	-	-	not observed	√	X	X
37	YPL127C	HHO1	Histone H1	√	√	abnormal cell cycle progression in G1	√	√	√
38	YPR204W	YPR204W	DNA helicase	√	-	not observed	√	X	X
39	YKL007W	CAP1	Actin polymerisation inhibitor	-	-	not observed	√	X	√
40	YPL267W	ACM1	APC inhibitor	-	√	vegetative growth: decreased rate	√	√	√
41	YBR161W	CSH1	Glycosylation	-	-	vegetative growth: decreased rate	√	√	√
42	YGL027C	GLS1	Glycosylation	-	-	vegetative growth: decreased rate	√	√	X
43	YGR221C	TOS2	Budding	-	√	not observed	√	√	√
44	YNL300W	TOS6	GPI-dependent cell wall protein	-	√	not observed	√	√	√

Switch Genes as Dosage-Sensitive Checkpoint Effectors

TOS genes.

Among the genes found to be regulated by a SBF-to-MBF switch, seven were previously identified in the Iyer ChIP-chip study as targets of SBF (TOS1, 2, 3, 4, 6 and 8).⁵⁴ Why these genes containing MCB/SCB overlapping motifs were previously identified as SBF targets is still unclear. However, it is well established that the enrichment for promoter DNA in ChIP analysis of MBF targets when pulling down Mbp1 is much lower than that of SCB target promoters when pulling down Swi4. This is assumed to be a reflection of the cross-linking methodology, rather than a reflection of the actual binding. Since TOS1-4, 6 and 8 have both SCBs and MCBs, they were only annotated as SBF genes after the genome-wide ChIP analysis, probably because signals for SBF were much higher than for MBF pulldowns. Of the Tos proteins, TOS4 was found to encode for a Forkhead-associated (FHA) domain-containing protein. Interestingly, upon replication stress, Tos4 accumulates in a checkpoint-dependent manner and interacts with the Hos2 and Rpd3 histone deacetylases (HDAC) subunits, which are part of two major HDAC complexes.¹³ Interaction between Tos4 and the HDACs is mediated by its FHA-domain, and this interaction is critical for Tos4 function. Genetic evidence suggests that Tos4 is important for cell viability during replication stress and functions in a parallel pathway to that of the checkpoint protein kinase Dun1.¹³ Although the role of Tos4 during replication stress response is still unclear, we speculate that in response to genotoxic stress, Tos4 might be involved in either promoting or inhibiting recruitment of HDACs to certain gene promoters or modulating the activity of certain HDACs. The function of the TOS genes remains largely unknown. However, whereas the TOS genes share no sequence homology, they are part of a very select group of genes linked by their expression profile and the mechanism of transcriptional control both during the cell cycle and in response to replication stress. Based on this, it is anticipated that the proteins encoded by these genes have a specific role outside of the G₁ phase of the cell cycle and in response to checkpoint activation.

MCD1, SWE1 and ACM1.

Among the genes controlled by the SBF-to-MBF switch, we also identified MCD1, a gene that encodes a subunit of the yeast cohesin complex, which is required for cohesion, DSB repair and recombination between sister chromatids during mitosis. Previous work has shown that Mcd1 is a checkpoint target, and the checkpoint-dependent phosphorylation of Mcd1 was shown to be important for cohesion during G₂/M.⁶⁷ Like MCD1, SWE1 is another switch gene that also plays an important role during G₂/M transition. SWE1 encodes a protein kinase involved in the inhibition of the cyclin-dependent kinase (CDK) Cdc28. Swe1-dependent inhibition of CDK is a central mechanism to regulate cell cycle transition during G₂/M.⁶⁸ The ACM1 switch gene plays a role in a later cell cycle transition, namely the exit of mitosis, by inhibiting the anaphase-promoting complex (APC) via competitive inhibition. The APC coactivator Cdh1, which is involved in the recruitment of target substrates to the APC, is inhibited from binding its targets by Cdc28-dependent stabilization of Acm1.⁶⁹

TOS4, MCD1, SWE1 and ACM1 are cell cycle-regulated, and, corroborating our hypothesis that switch genes expression needs to be tightly regulated, overexpression of these genes was linked to growth defect and/or cell cycle arrest.^65,66 Further analysis of the other switch genes, such as the TOS genes and eight putative ORFs, will show their potential role as checkpoint effectors.

Conclusion

With the discovery of the SBF-to-MBF switch in budding yeast we have identified a group of genes involved in the checkpoint response that require timely inactivation of transcription. The switch mechanism not only allows the upregulation of genes that are important for the checkpoint response during S-phase, but it also prevents the overexpression of these dosage-sensitive genes outside of G₁ in case of MBF malfunction. Furthermore, preliminary work from the de Bruin lab in fission yeast indicates that, whereas G₁/S transcription levels are maintained at a high level in response to genotoxic stress, repression outside of G₁ is required to maintain genome stability (C. Caetano and RdB). Together with the Nrm1 negative feedback mechanism of transcriptional repression, which rapidly shuts off transcription of MBF targets, the SBF-to-MBF switch provides an important regulatory mechanism to tightly control protein accumulation not only during S-phase, but also throughout the whole cell cycle. Consequently, switch genes are enriched for cell cycle-regulated, dosage-dependent and replication stress-induced genes. These features could direct the future characterization of this select group of co-regulated genes, many of which are of unknown function. In addition, future research will shed light on how widely used this mode of regulation is and the importance of this particular network wiring for cell survival in response to stress and during evolution.