SPRTN protease and checkpoint kinase 1 cross-activation loop safeguards DNA replication

The SPRTN metalloprotease is essential for DNA-protein crosslink (DPC) repair and DNA replication in vertebrate cells. Cells deficient in SPRTN protease exhibit DPC-induced replication stress and genome instability, manifesting as premature ageing and liver cancer. Here, we provide a body of evidence suggesting that SPRTN activates the ATR-CHK1 phosphorylation signalling cascade during physiological DNA replication by proteolysis-dependent eviction of CHK1 from replicative chromatin. During this process, SPRTN proteolyses the C-terminal/inhibitory part of CHK1, liberating N-terminal CHK1 kinase active fragments. Simultaneously, CHK1 full length and its N-terminal fragments phosphorylate SPRTN at the C-terminal regulatory domain, which stimulates SPRTN recruitment to chromatin to promote unperturbed DNA replication fork progression and DPC repair. Our data suggest that a SPRTN-CHK1 cross-activation loop plays a part in DNA replication and protection from DNA replication stress. Finally, our results with purified components of this pathway further support the proposed model of a SPRTN-CHK1 cross-activation loop.


Introduction
The timely completion of DNA replication is essential for genome integrity and preventing the onset of cancer, premature ageing and developmental disorders 1,2 . DNA replication is constantly threatened by many factors, including DNA lesions, collisions with the transcription machinery and repetitive DNA sequences. Cells have evolved robust DNA damage tolerance and DNA damage response pathways to cope with the different lesions and obstacles that challenge the progression of DNA replication forks 3-6 . In particular, the ATR (Ataxia telangiectasia Rad3 related)-CHK1 signalling cascade is the major regulator of the response to replication stress 7,8 . This cascade performs multiple functions in response to DNA replication stress, including regulating DNA replication origin firing, stabilising stalled replication forks, and delaying mitotic entry by preventing CDK1/2 hyperactivation 5,9 .
The main stimulus for CHK1 activation is replication protein A (RPA)-coated single-stranded (ss) DNA that typically forms upon DNA replication fork stalling due to uncoupling of the Cdc45-Mcm2-7-GINS (CMG) helicase complex from DNA polymerases 5 . This ssDNA-protein structure recruits the ATR-ATRIP protein kinase complex which activates CHK1 by phosphorylating serines 317 and 345 to expose the catalytic N-terminal domain of CHK1 10 . Upon phosphorylation, CHK1 is released from chromatin by an as yet unknown mechanism and spread throughout the nucleus and cytoplasm to regulate the activity of its substrates and safeguard genome stability 9,11,12 . The CHK1 signalling pathway can also be activated in S-phase by ssDNA generated after 5'-3' end resection of double strand DNA breaks (DSB) 13 . However, given that CHK1 activity is required for physiological DNA replication fork progressionwhen long stretches of ssDNA are scarce -the question of how CHK1 is activated during steady-state DNA synthesis remains unanswered [14][15][16][17][18] . Identifying the mechanisms that regulate CHK1 signalling under physiological conditions is therefore essential to understand how cells survive and preserve genomic stability 19 .
We and others have recently identified the SPRTN metalloprotease as being a constitutive component of the DNA replication machinery and essential for DNA replication [20][21][22][23][24] . The essential role of SPRTN in safeguarding genome stability is demonstrated both in human disease and in animal models. Monogenic, biallelic SPRTN germline mutations cause Ruijs-Aalfs Syndrome (RJALS), a rare disease characterised by genomic instability, premature ageing and hepatocellular carcinoma 20,25,26 . Furthermore, SPRTN haploinsufficient mice develop RJALS-like phenotypes, while complete SPRTN knockout is embryonic lethal 27 . SPRTN has recently been identified as an essential core fitness gene in humans [27][28][29][30] . Finally, downregulation of SPRTN in zebrafish severely impairs normal embryonic development and increases embryonic lethality 25 .
The source of genome instability in both RJALS patient cells and SPRTN-deficient human and mouse cells was recently demonstrated to arise from replication stress caused by the accumulation of replication-blocking DNA-protein crosslinks (DPC) 20,22,23 . DPCs are formed by various aldehydes including the well-known DPC inducing agent formaldehyde (FA), which are by-products of metabolic processes such as lipid peroxidation or histone and DNA demethylation [31][32][33] . As SPRTN protease activity is required to cleave DPCs, defective SPRTN protease activity results in profound replication stress, visualised as increased fork stalling and significantly reduced DNA replication fork velocity. Strikingly, we observed a severe G2/M-checkpoint defect in SPRTN-deficient cells treated with genotoxic agents that interfere with DNA replication 25 . The G2/M checkpoint was, however, completely functional after the induction of non-replication-associated DNA strand breaks using ionising radiation, suggesting that SPRTN-defective cells lack the ability to activate CHK1 in response to replication stress when replication forks are still intact 25 .
Here, we demonstrate that SPRTN stimulates CHK1 function during physiological DNA replication and vice versa. SPRTN proteolytic activity (proteolysis) evicts CHK1 from replicative chromatin/sites of DNA replication what allows physiological CHK1 function during steady-state DNA replication.
We also show that SPRTN proteolysis cleaves the regulatory/inhibitory C-terminal domain of CHK1 in vitro and in vivo, and thus releases active N-terminal CHK1 products. These N-terminal CHK1 products, when ectopically expressed, are sufficient to stabilise DNA replication forks, rescue embryonic development and genome stability that are compromised by depletion of SPRTN to approximately 30% of wildtype cells. This rescue is dependent of CHK1 phosphoryalting the Cterminus of the residual SPRTN, further promoting SPRTN recruitment to chromatin for the removal of DPCs in front of DNA replication forks. In summary, we show that a SPRTN-CHK1 crossactivation loop is essential for steady-state DNA replication, embryonic development and survival and is evolutionarily conserved in vertebrates.

SPRTN deficiency leads to aberrant CHK1 activity
Analysis of RJALS patient and SPRTN-depleted cells revealed that SPRTN protease activity is essential for DNA replication fork progression, cell cycle progression, and G2/M checkpoint activation after DNA replication stress but not ionising radiation 20,23,25 . These results suggest that SPRTN bridges DNA replication and G2/M-checkpoint regulation. To re-evaluate these findings, we performed analysis of DNA replication using the DNA fiber assay (Fig. 1a). Depletion of SPRTN by three independent siRNA sequences caused severe DNA replication stress in human embryonic kidney 293 (HEK293) cells, visualised as a reduction in DNA replication fork velocity and an increased frequency of fork stalling ( Fig. 1b-d). Short treatment of control cells with a low dose of hydroxyurea (HU), a drug that limits the cellular dNTP pool, was used as a positive control of DNA replication stress phenotypes. In general, as a response to DNA replication stress, cells suppress dormant origin firing, as was visible after HU treatment (Fig. 1e). Interestingly, dormant origin firing was more than 3-fold higher in SPRTN-depleted cells when compared to HU-treated cells. As firing of dormant origins is tightly regulated by the CHK1 kinase 34,35 which also controls the G2/Mcheckpoint 36,37 , we asked whether CHK1 signalling was defective in SPRTN-inactivated cells. ATR-CHK1 signalling, visualised by CHK1 S345 phosphorylation, was not activated in SPRTN-depleted cells (Fig. 1f,g) despite the severe DNA replication stress phenotypes . Accordingly, CHK1 kinase activity was not activated as demonstrated by the lack of CHK1 serine 296 (S296) phosphorylation, the residue that CHK1 auto-phosphorylates once its kinase activity has been stimulated by ATR 38,39 . Furthermore, despite being in similar cell cycle stages , SPRTN-depleted cells exhibited less CHK1-S296 and -S345 phosphorylation than even unchallenged control cells and ~9-10-fold less compared to HU-treated control cells, which showed similarly reduced replication fork velocity and elevated levels of fork stalling as SPRTN-depleted cells (Fig. 1f, g). Similar findings were observed in two other human cells lines: HeLa (Fig. 3b, c) and U2OS (data not shown). These results show that SPRTN-inactivated human cells fail to activate a robust CHK1 response despite exhibiting severe replication stress phenotypes that would ordinarily be expected to elicit such a response, namely CHK1 chromatin eviction and consequent activation of CHK1 signalling 11 . Indeed, SPRTN-inactivated cells accumulate over two-fold more CHK1 on chromatin, similar to cells treated with the DNA-protein crosslinking agent formaldehyde (FA), but opposite to cells treated with HU, where CHK1 is released from chromatin ( Supplementary Fig. 1d, e).
To further validate this observation, we treated HEK293 cells with UCN-01, a well-characterised CHK1 inhibitor 15,40,41 . UCN-01 treatment in control cells caused a severe reduction in replication fork velocity and increased the frequency of new (dormant) origin firing and fork stalling ( Fig. 1h-j).
UCN-01, however, had no additive effect on replication fork velocity, new origin firing, or fork stalling in SPRTN-depleted cells. In addition, in comparison to wt cells, SPRTN-haploinsufficient HeLa (DSPRTN) cells were relatively but significantly less sensitive to UCN-01 treatment (Fig. 1k).
This further suggests that CHK1 is not fully activated in SPRTN-defective human cells. Altogether, these results highlight the importance of CHK1 activity during steady-state DNA replication 14,15,18 , demonstrate an epistatic relationship between SPRTN and CHK1, and reveal a severe defect in CHK1 kinase activation in SPRTN-defective cells despite them suffering from severe DNA replication stress. We concluded that SPRTN-defective cells lack the optimal (physiological) CHK1 activity required for DNA replication and genome stability, most probably due to their inefficiency in evicting CHK1 from chromatin, and thus activating a steady-state CHK1 signalling cascade.

SPRTN regulates CHK1 signalling pathway under physiological conditions
To test this conclusion, we investigated the signalling pathway downstream of CHK1 by monitoring the total levels of the CHK1 target, protein phosphatase Cdc25A 3 . Upon phosphorylation by CHK1, Cdc25A is degraded by the proteasome, as observed in cells treated with HU ( Fig. 1f and Supplementary Fig. 1f). Consequently, CDK1/2 are hyper-phosphorylated and become inactive, which leads to intra S-phase and G2/M-checkpoint activation and cell cycle arrest 42,43 . CDK1 and 2 drive S-phase progression and the G2/M cell cycle transition, but hyper-activation of CDK1/2 negatively influences DNA replication fork stability and causes premature mitotic entry 44,45 . Hence, the ATR-CHK1-Cdc25-CDK1/2 pathway is necessary to regulate cell cycle progression during DNA synthesis. Due to faulty CHK1 activation, SPRTN-depleted cells hyper-accumulated Cdc25A ( Fig.1f and Supplementary Fig. 1f), which in turn dephosphorylates and hyper-activates CDK1/2, as was visible by the increased phosphorylation of total CDK1/2 substrates in HEK293 cell extracts ( Supplementary Fig. 1g, h).

CHK1 overexpression corrects DNA replication stress and genome instability in SPRTNdefective cell
To assess whether the failure to activate CHK1 signalling could explain the DNA replication phenotypes and G2/M defects observed in SPRTN-depleted cells ( Fig. 1) 25 , we ectopically expressed CHK1-wild type (wt) or its phosphorylation (phospho)-defective variants (CHK1-S317A or CHK1-S345A) (Fig. 2a, b and Supplementary Fig. 2a). CHK1-wt, but not CHK1-S317A or CHK1-S345A, restored DNA replication fork velocity, suppressed new origin firing and rescued replication fork stalling in SPRTN-depleted cells. Moreover, ectopic expression of CHK1-wt in SPRTN-deficient cells also corrected chromosomal instability, measured by the number of chromosomal aberrations on mitotic chromosomes ( Fig. 2c and Supplementary Fig. 2b). These results further support our initial observation that SPRTN-inactivation leads to an impaired CHK1 signalling cascade, resulting in severe DNA replication stress, a defective G2/M checkpoint and the accumulation of chromosomal aberrations in SPRTN-deficient cells 25,27 .

The SPRTN-CHK1 axis is essential for development and genome stability in zebrafish embryos
To investigate and validate our observations so far on an organismal level, we took advantage of the zebrafish model system. We have previously shown that morpholino (MO)-mediated depletion of SPRTN in zebrafish embryos causes severe development defects and accumulation of DNA damage 25 . When fertilized eggs with SPRTN MO were co-injected with capped RNAs encoding for GFP-CHK1-wt, both developmental retardation and DNA damage (the latter analysed by gH2AX accumulation) were rescued   Supplementary Fig. 2c). Conversely, reconstitution with RNAs encoding phospho-defective variants of GFP-CHK1, S317A or S345A, failed to rescue the phenotypes of SPRTN depletion in early zebrafish embryos. Altogether, these data suggest that the restoration of CHK1 activity is able to compensate for SPRTN-deficiency in human cells and zebrafish embryos, and that the SPRTN-CHK1 axis is conserved in vertebrates.

SPRTN does not contribute to CHK1 activation after DNA double strand break formation
Interestingly, when SPRTN-depleted cells were challenged with HU, CHK1 was activated to the same extent as in HU-treated control cells (Fig. 3a, lanes 2, 4, 6 and 8). This CHK1 activation was concomitant with phosphorylation of ATM, CHK2 and RPA, indicating DSB formation 36,46 .
Moreover, the total level of RPA remained unchanged in these cells and phosphorylation of RPA ruled out the possibility that severe replication stress in SPRTN-inactivated cells results in RPA exhaustion 47 . Similar findings were observed in haploinsufficient HeLa SPRTN cells (D-SPRTN) (Fig. 3b,c). Taken together, our data suggest that SPRTN-deficient human cells have a fully functional ATR-CHK1 signalling pathway in the context of DSB repair, but that SPRTN is needed to activate this pathway during physiological/steady-state DNA replication.
In addition, immunofluorescence-coupled microscopy analysis of ssDNA formation in S-phase cells (CldU positive) by BrdU staining under native conditions further suggested that SPRTN-defective cells do not form extensive ssDNA formation ( Fig. 3d-f), a platform for canonical and robust CHK1 activation, despite enduring severe DNA replication stress. Similar results were obtained with FAtreatment. HU treatment was used as a positive control for DNA replication stress-induced ssDNA formation. However, when SPRTN-depleted or FA-treated cells were exposed to a high dose of Camptothecin (CPT) or HU, a striking increase in ssDNA (visualised as BrdU foci), generated by 5'-3' end resection of DSBs ( Supplementary Fig. 3a, b), was observed. This effect ruled out the possibility that this BrdU/ssDNA assay was compromised in SPRTN-deficient cells owing to reduced BrdU incorporation due to defective DNA replication ( Supplementary Fig 1a-c). Altogether, these results suggest that SPRTN is essential for activation of physiological CHK1 response during steadystate DNA synthesis, most probably due to removal of bulky DPCs in front of DNA replication forks that prevent CMG helicase progression and consequently ssDNA formation, but dispensable for CHK1 activation after DSB formation, when a huge amount of ssDNA is formed by 5'-3' end resection.

SPRTN proteolysis releases CHK1 from replicating chromatin
Given that SPRTN enables DNA replication by proteolysis of its chromatin-bound substrates 20,22 and that, in SPRTN-defective cells, CHK1 is not activated during physiological conditions ( One, by using cell fractionation we demonstrated that a small amount of CHK1 is constantly bound to chromatin ( Fig. 4a, b). Additionally, CHK1 significantly hyper-accumulated on chromatin isolated under stringent and denaturing conditions in SPRTN-inactivated cells during S-phase progression (Fig. 4c,d) suggesting that a small protion of CHK1 is also covalently attached to chromatin. Two, in pull-down experiments using purified SPRTN and CHK1 proteins we demonstrated that these two proteins interact directly in vitro (Fig. 4e). Three, by co-immunoprecipitation experiments from HEK293 cells expressing SPRTN-wt or a RJALS patient protease defective variant (SPRTN Y117C), we showed that both SPRTN wt and Y117C form a physical complex with CHK1 and other components of DNA replication machinery such as PCNA and MCM3 48 , further suggesting that SPRTN and CHK1 are components of the DNA replication machinery in vivo (Fig. 4f). Four, we took advantage of iPOND to demonstrate SPRTN-dependent CHK1 dynamics at replicative chromatin (Fig. 4g,h). iPOND directly demonstrated that both CHK1 and SPRTN are present at/around sites of DNA replication forks (nascent DNA) and travel with the fork as confirmed by their absence from mature chromatin (after the thymidine chase) (Fig. 4h). Importantly, inactivation of SPRTN led to a strong accumulation of CHK1 on both nascent and mature chromatin (Fig. 4i, compare lanes 2 and 3 with 4 and 5).
In the light of the above observations we used iPOND to directly assess if SPRTN proteolysis plays an essential role in CHK1 release/eviction from sites of (around DNA replication) replicative chromatin. We analysed the abundance of CHK1 by iPOND in SPRTN-depleted cells expressing an empty vector (EV), SPRTN-wt or the SPRTN protease-deficient variant E112A (Fig. 4j, k). Although both SPRTN variants were equally expressed, SPRTN-wt strongly evicted CHK1 from nascent (DNA replication sites) and mature chromatin but SPRTN-E112A did not. A similar effect was observed on total chromatin isolated by biochemical fractionation (Supplementary Figure 4a, b). Altogether, these results directly demonstrate that the SPRTN-CHK1 complex is an integral part of the DNA replication machinery and that SPRTN proteolysis evicts CHK1 from replicating chromatin during DNA synthesis.

SPRTN protease cleaves CHK1
As SPRTN directly interacts with (Fig. 4e) and evicts CHK1 from sites of DNA replication in a protease-dependent manner ( Fig. 4g-k), we investigated whether SPRTN proteolytically cleaves CHK1. Purified Flag-CHK1 was incubated with SPRTN-wt or a protease-deficient variant (E112A) in vitro in the presence of dsDNA, an activator of SPRTN metalloprotease activity (Fig. 5a). SPRTN-wt but not SPRTN-E112A cleaved CHK1 in at least three sites, generating three cleaved N-terminal fragments [Cleaved Products (CPs) 1, 2 and 3], as detected by Flag antibody (Fig. 5a). Similar Nterminal CHK1 fragments were obtained on endogenous or over-expressed CHK1 protein in vivo (Fig. 5b,c). Strikingly, these cleaved CHK1 products were completely absent in cells treated with three independent siRNA sequences targeting SPRTN (Fig. 5c). Altogether, these results further support the conclusion that CHK1 is a substrate of SPRTN in vitro and in vivo. As FA induces DPC formation and consequently increases SPRTN recruitment to chromatin (see also Fig. 6a-f) to proteolytically cleave chromatin-bound substrates, we wondered whether cleaved fragments of CHK1 are also enhanced in FA-treated cells ( Fig. 5d and Supplementary Fig. 5 a). Indeed, increasing concentrations of FA stimulated the formation of cleaved N-terminal fragments of endogenous CHK1 ( Fig. 5d) or over-expressed CHK1 ( Supplementary Fig. 5 b), which mimic the SPRTN-dependent products of CHK1 cleavage in vitro and in vivo.
We estimated the size of these fragments by their molecular weight and designed constructs mimicking these CPs for their expression in vertebrates (Fig. 5e). SPRTN-cleaved CHK1 fragments are kinase active as they were able to phosphorylate SPRTN or the well-characterised CHK1 substrate Cdc25A in vitro ( Fig. 5f, g) 49 .
Most notably, the N-terminal CHK1 products were fully functional in vivo, as their ectopic expression was able to restore DNA replication fork velocity and suppress new origin firing and fork stalling in SPRTN-depleted cells ( Fig. 5h and Supplementary Fig. 5b). Importantly, the recovery of DNA replication velocity with the N-terminal CHK1 product CP3 was sensitive to the CHK1 kinase inhibitor UCN-01 but resistant to the ATR kinase inhibitor VE-821 ( Supplementary Fig. 5c, d), and this rescue was completely dependent on the kinase active centre, located at aspartic acid 130 (D130) ( Supplementary Fig. 5e). These results further suggest that the N-terminal CHK1 fragments, which mimic the cleaved CHK1 fragments generated by SPRTN proteolysis, are kinase active. Similarly, expression of the N-terminal CHK1 product CP3 corrected developmental defects and suppressed DNA damage in SPRTN-depleted zebrafish embryos to the same extent as full-length, wild-type CHK1 (Fig. 5i, compare with Fig. 2f, g).

CHK1 stimulates recruitment of SPRTN to chromatin
Our results so far suggest that SPRTN is needed for proper activation of CHK1 during physiological DNA replication (Fig. 1). This CHK1 activation occurs by both (i) SPRTN proteolysis-dependent eviction of CHK1 from replicative chromatin  and (ii) by simultaneous cleaving off the inhibitory/regulatory C-terminal parts of CHK1 ( Fig. 5a-e). Paradoxically, ectopic expression of CHK1 restored various tested phenotypes of SPRTN-deficiency both in human cells and zebrafish embryos (Figs. 2, 5h, i and Supplementary Fig. 5b, d). As siRNA and MO-mediated SPRTN depletion is never 100% efficient, we speculated that ectopic CHK1 expression stimulates, by phosphorylation ( Fig. 5f), the remaining SPRTN to process DPCs during DNA synthesis. Indeed, biochemical analysis demonstrated that SPRTN-inactivated cells still retain approximately 20-30% of SPRTN protein ( Fig.   6a, b). To investigate this possible cross-talk between SPRTN and CHK1, we monitored SPRTN recruitment to chromatin after ectopic CHK1 expression (Fig. 6c, d). The amount of SPRTN on chromatin was substantially increased after CHK1 overexpression, suggesting that CHK1 induces SPRTN recruitment to chromatin. Importantly, the residual endogenous SPRTN in D-SPRTN cells (≈30% of control cells; Fig. 6a, b) hyper-accumulated on chromatin after CHK1 overexpression, almost to the same level as in untreated wild-type cells (Fig. 6c, d). To further support these results, we induced SPRTN-SSH expression in doxycycline-inducible HEK293 cells and ectopically coexpressed CHK1-wt or the N-terminal CHK1 cleaved fragments (CP1, 2 and 3). Ectopic expression of CHK1-wt stimulated SPRTN-SSH accumulation on chromatin. Strikingly, SPRTN-SSH recruitment to chromatin was even further enhanced after ectopic expression of the N-terminal CHK1 fragments ( Fig. 6e, f). These results explain why the ectopic expression of CHK1-wt or its N-terminal fragments restores the phenotypes arising from SPRTN-deficiency in human cell lines and zebrafish embryos.

CHK1 phosphorylates the C-terminal regulatory part of SPRTN
To identify CHK1 phosphorylation sites ( Fig. 5f) on SPRTN in vivo, we performed mass spectrometry. SPRTN-E112A-SSH was ectopically expressed in HEK293 cells co-transfected with either CHK1-wt or the kinase-defective CHK1-S345A, and anti-SSH precipitates were isolated under denaturing conditions (Fig. 6g, Supplementary Fig. 6a). The SPRTN protease dead mutant (E112A) was used to avoid SPRTN auto-cleavage activity 20 (Fig 5a, lower panel). Mass spectrometry analysis identified three main CHK1 phosphorylation sites on SPRTN -serines 373, 374 and 383 -( Fig. 6h and Supplementary Fig. 6b, c) located in the C terminal part of SPRTN. These CHK1 phosphorylation sites on SPRTN correspond to a consensus target motif of CHK1 (R/K-x-x-p(S/T) 50 . To validate our results, we isolated SPRTN-wt-SSH or -S373A, -S374A and -S383A variants from HEK293 cells treated with DMSO or the CHK1 inhibitor UCN-01. SPRTN isolates were analysed by Western blot using an antibody that recognises CHK1 consensus sites (Fig. 6i, j). SPRTN-wt was phosphorylated in control cells but this phosphorylation signal was strongly decreased when cells were treated with UCN-01. Importantly, SPRTN phosphorylation was strongly reduced (S373A or S383A) or almost completely abolished (S374A) even when only one of the three identified serines (S373, S374 and S383) was mutated. This suggests that the C-terminal, regulatory part of SPRTN is phosphorylated by CHK1 at these specific serines under steady state conditions.

SPRTN-CHK1 cross-activation loop
To biologically validate the biochemical evidence that CHK1 phosphorylates SPRTN and thus regulates SPRTN's function, we depleted SPRTN in HEK293 cells and tested the effects of ectopically-expressing variants of SPRTN that either could or could not be phosphorylated by CHK1.
To demonstrate our model in which a SPRTN protease-CHK1 kinase cross-activation loop processes DPCs in front of the DNA replication fork, we measured the progression of DNA replication fork over FA-induced DPCs by DNA fiber assay (Fig. 8a). Transient and low dose FA treatment strongly reduced DNA replication fork velocity, but this could be significantly reversed by expression of either CHK1-wt or SPRTN-wt. This demonstrates that both SPRTN and CHK1 prevent DPC-induced DNA replication fork slowing and stalling.
As final confirmation that CHK1-overexpression indeed promotes the removal of DPCs in SPRTNdefective cells, we isolated total DPCs from wt and D-SPRTN Hela cells. D-SPRTN cells contain about 2-to 3-fold more DPCs than wt cells (Fig. 8b, c). However, ectopic expression of CHK1-wt, but not of kinase deficient variant CHK1-S345A, reduced the levels of DPCs in D-SPRTN cells to roughly the same level as in wt cells. Altogether, these results support our model ( Fig. 8d) proposing that a SPRTN-CHK1 cross-activation loop works in DPC removal during physiological DNA replication to prevent DNA replication stress and preserve genomic stability.

Discussion
Our results demonstrate the existence of an evolutionarily-conserved protease-kinase axis during physiological DNA replication that is essential for DPC proteolysis and prevention of DNA replication stress. We propose a model (Fig. 8d) wherein SPRTN protease and CHK1 kinase work in a cross-activation loop that operates during steady-state DNA synthesis. SPRTN protease activity releases CHK1 from replicating chromatin and this regulates DNA replication fork progression, suppression of dormant origin firing and prevents replication fork stalling, as well as the fine-tuning of CDK1/2 activity. These processes are essential for genome stability and embryonic development under physiological conditions as demonstrated in human cell lines and zebrafish embryos, respectively. In turn, CHK1 phosphorylates SPRTN at C-terminally located serines. SPRTN further accumulates on chromatin to proteolyse DPCs and thus allow unperturbed DNA synthesis. Our findings demonstrate that: (i) the SPRTN protease is tightly regulated and recruited to chromatin during DNA synthesis by CHK1 and at the same time (ii) SPRTN proteolysis evicts, cleaves and thus consequently activates CHK1 during steady-state DNA replication to fine-tune DNA replication and thus prevent DNA replication stress, one of the main causes of genome instability 2,51 .

Regulation of the SPRTN protease by CHK1 phosphorylation
The pleiotropic nature of SPRTN protease activity and its tight association with the replisome suggest that SPRTN must be tightly regulated. So far, the regulation of SPRTN has only been investigated upon exposure to genotoxic conditions such as UV light and FA. SPRTN accumulation at UVinduced DNA lesions strongly depends on its PCNA-binding PIP box and ubiquitin-binding UBZ domain 52-54 . On the other hand, SPRTN activation in response to DPCs induced by 1 mM FA involves SPRTN deubiquitination, but does not depend on the PIP box and UBZ domains 21 .
However, how SPRTN is regulated under physiological conditions is unclear. This is a critical question as SPRTN is an essential mammalian gene from early embryogenesis and is physically present at DNA replication forks and possesses pleotropic protease activity. It has been proposed that ssDNA promotes SPRTN-dependent substrate cleavage while dsDNA induces SPRTN auto-cleavage to negatively regulate SPRTN activity 21 . There are two arguments against this model: (i) SPRTN is an essential enzyme for DPC repair during DNA replication, yet DPCs prevent the helicasepolymerase uncoupling and ssDNA formation that would be required to activate SPRTN-dependent substrate cleavage in this model, (ii) SPRTN protease activity towards DNA-binding substrates is also stimulated by dsDNA 20,22 , Our results suggest a model in which SPRTN protease activity is regulated by CHK1 kinase activity during physiological DNA replication. CHK1 constitutively phosphorylates SPRTN at the Cterminally located serines S373, S374 and S383. We show that CHK1-dependent SPRTN phosphorylation is essential for DNA replication and embryonic development under physiological conditions. We demonstrated that CHK1-dependent phosphorylation of SPRTN stimulates SPRTN recruitment to chromatin, and that these phosphorylation events are essential to regulate DNA replication, embryonic development and the release of CHK1 from chromatin. Further investigation should elucidate the phosphorylation dependant regulation of SPRTN.

Regulation of CHK1 kinase during physiological DNA synthesis
It is well characterised how replication fork stalling and collapse leads to the formation of RPAcoated ssDNA and subsequent activation of CHK1 5 . However, a large body of evidence suggests that CHK1 is essential during unperturbed DNA synthesis when the ssDNA required for activating this canonical ssDNA-ATR-CHK1 signalling cascade is limited 14,15,18,19 . Our data supports a novel mechanism for CHK1 chromatin release and activation during physiological DNA synthesis. We show that CHK1 associates with chromatin and components of DNA replication machinery and that SPRTN protease activity is required to release CHK1 from replicative chromatin during steady-state DNA synthesis.
We also demonstrated that SPRTN directly binds and cleaves the C-terminal part of CHK1 thus releasing at least three well detected the N-terminal CHK1 fragments (CP1, CP2 and CP3) in vitro.
This effect is also observed with endogenous and ectopically expressed CHK1 in vivo. SPRTN protease cleavage of CHK1 generates active N-terminal CHK1 fragments that are able to phosphorylate SPRTN and restore DNA replication defects and developmental retardation in SPRTNdeficient human cells and zebrafish embryos, respectively.
The fact that CHK1 is a signalling molecule that activates downstream effectors but is mostly in an inactive state leads us to speculate that even a small amount of the CHK1 N-terminal fragments is sufficient to regulate physiological CHK1-signalling pathway activation during steady state DNA replication. In this way cells prevent the robust CHK1 activation usually observed after severe DNA damage that leads to apoptosis. Thus, we propose that SPRTN-dependent release and cleavage of even a small amount of CHK1 at/around DNA replication fork is sufficient to control DNA replication fork progression and ensure genome stability Interestingly, the activation of CHK1 by cleavage of its C-terminal part was also observed in chicken DT40 and various human cells treated with Cisplatin and UV light as well as the well-known DPCinducing agents, Campothecin and Etoposide 55,56 . In some cases, this cleavage was caspasedependent while in others it was not, pointing to the existence of a protease that cleaves and activates CHK1 55,56 . These reports and the herein described results that the N-terminal CHK1 kinase active fragments restore DNA replication and developmental defects in SPRTN-depleted cells and zebrafish embryos, respectevely (Supplementary Fig. 5 b, c) further support our model (Fig. 8d)

Competing interest
The authors declare no competing interests.

Methods
Key resources used in this study are listed and described in Supplementary Table 2. Oligonucleotide sequences are listed in Supplementary Table 3.

Experimental models and cell culture
U2OS, HeLa and HEK293 human cells were maintained in DMEM (Sigma-Aldrich) supplemented with 10% FBS (Gibco) and 100 I.U./mL penicillin / 0.1 mg/mL streptomycin at 37°C in a humidified incubator with 5% CO 2 , and tested for mycoplasma contamination. CRISPR partial knockout Δ-SPRTN HeLa cells 20 were maintained as above. Media for doxycycline-inducible stable HEK293 Flp-In TRex SPRTN-wt or SPRTN-Y117A cell lines 20 was additionally supplemented with 15 µg/ml blasticidin / 100 µg/ml hygromycin. Wild-type zebrafish (Danio rerio EK and AB strains) were kept under standardized conditions in a circulating water system on a 14 hours light and 10 hours dark cycle. All procedures involving zebrafish embryos adhered to current European regulations for the use of animals and were approved by the local authorities of Ulm University.

Bacterial strains
Competent E. coli DH5alpha (Invitrogen) and E. coli Rosetta2 (Merck; Novagen), used for cloning/ plasmid amplification or for protein expression, respectively, were transformed with plasmid and grown in LB medium at 37 ºC.

Subcloning and site directed mutagenesis
The pCINeo/CMV-Flag-CHK1-wt plasmid 11 was used for expression in mammalian cells and to generate all CHK1 variants by site-directed mutagenesis. For the expression of CHK1 in Zebrafish, CHK1 variants were subcloned into pCS2-GFP vector by PCR from previous plasmid and EcoRI/XbaI insertion. Site-directed mutagenesis was performed by plasmid PCR using specific mutagenic primers (Supplementary Table 3

DNA fiber assay
The DNA fiber assay was performed as described previously 20,25 .

Metaphase spreads
Metaphase spreads were performed as described previously 25 . HEK293 were treated with 100 ng/mL colcemid for 2 h, trypsinized, collected, and incubated with 0.4% KCl at 37 °C for 15 min. Following detection of replication foci after denaturation, sample was incubated with anti-CldU antibody (clone BU1, 1:100 dilution) for 1 h, washed twice with PBS, then with high salt buffer (PBS with 200 mM NaCl, 0.2% Tween-20 and 0.2% NP-40 ) for 15 min and then with PBS again, then incubated with anti-rat IgG Cy3 antibody (Jackson Immuno Research, 1:2,000 dilution), and washed again. Finally, coverslips were mounted in mounting medium supplemented with 2x DAPI. Images were taken with a Leica DMRB microscope with a DFC360FX camera.

DNA Protein Crosslink isolation and detection
DNA Protein Crosslinks (DPCs) were isolated using a modified Rapid Approach to DNA Adduct Recovery (RADAR) assay 58 and as described previously 20 . Briefly, 1-2x10 6 cells were lysed in 1 mL of buffer containing 6 M guanidine thiocyanate (GTC); 10 mM Tris-HCl, pH 6.8; 20 mM EDTA; 4% Triton-X100; 1% Sarkosyl and 1% DTT. DNA was precipitated by adding 100% ethanol, washed three times in wash buffer (20 mM Tris-HCl, pH 6.8; 150 mM NaCl and 50% ethanol). DNA was then solubilized in 1 mL of 8 mM NaOH. The DNA concentration was quantified by treating a small aliquot of DNA with proteinase K (Invitrogen) for 3 h at 50°C, followed by detection with PicoGreen dye (Invitrogen) according to manufacturer's instructions. Equal dsDNA loading was confirmed by slot-blot immunodetection with an anti-dsDNA antibody (Abcam). Total DPCs after electrophoretic separation on polyacrylamide gels were visualized by silver staining using the ProteoSilver Plus Silver Stain Kit (Sigma-Aldrich) and specific proteins were detected by Western-blot.

SPRTN purification
SPRTN protein purification was performed as described previously 20 . E. coli was resuspended in lysis buffer (100 mM HEPES, pH 7.5; 500 mM NaCl; 10% glycerol; 10 mM imidazole; 1 mM Tris(2carboxyethyl)phosphine (TCEP); 0.1% dodecylmaltoside (DDM); 1 mM MgCl 2 and protease inhibitor cocktail (Merck)) and sonicated. 1 U Benzonase nuclease was added to lysates before cell debris was pelleted by centrifugation. Lysates were applied to a Ni-sepharose IMAC gravity flow column, washed with two column volumes of wash buffer (50 mM HEPES, pH 7.5; 500 mM NaCl; 10% glycerol; 45 mM imidazole; 1 mM TCEP), and eluted in elution buffer (wash buffer but with 300 mM imidazole). Elution fractions were applied directly to a 5mL Hitrap SP HP column (GE healthcare), washed with wash buffer 2 (50 mM HEPES, pH 7.5; 500 mM NaCl; 1 mM TCEP) and eluted with elution buffer (wash buffer but with 1 M NaCl). The purification tag was cleaved by the addition of 1:20 mass ratio of His-tagged TEV protease during overnight dialysis into buffer A (20 mM HEPES, pH 7.5, 500 mM NaCl, 0.5 mM TCEP). Samples were concentrated by ultrafiltration using a 30 kDa molecular weight cut-off centrifugal concentrator and loaded onto size exclusion chromatography using a HiLoad 16/60 Superdex 200 column (GE Healthcare) at 1 mL/min in buffer A. Protein identities were verified by LC/ESI-TOF Mass spectrometry 20 and protein concentrations were determined by absorbance at 280 nm (Nanodrop) using the calculated molecular mass and extinction coefficients.

In vitro cleavage of CHK1
Recombinant SPRTN was purified from E. coli. Flag-CHK1 was purified from ectopically expressing HEK cells in non-denaturing conditions by immobilization in Flag-M2 affinity agarose beads and several washes containing either 0.5% Triton X-100 or 1 M NaCl, and then eluted with a Flag peptide (Sigma-Aldrich). Cleavage reaction was performed typically in 15 µL volume containing Flag-CHK1, 2 ug SPRTN, and a 100 bp dsDNA oligonucleotide probe (Supplementary Table 3

In vitro CHK1 kinase assay
Flag-CHK1 species (full length or cleavage products) were obtained from 10 8 HEK cells previously transfected with expressing vectors and purified as above. For phosphorylation activity, purified Flag-CHK1 was incubated (37 °C, 2 h) with a substrate (SPRTN: 2 µg; cdc25A: 500 ng) in a kinase reaction buffer (10 mM Hepes, pH 7.4; 0.5 mM ATP; 10 mM MgCl 2 ; 5 mM KCl; 0.1 mM DTT; 20 mM b-glycerophosphate and phosphatase inhibitors) in a total volume of 25 µL. The reaction was stopped by the addition of 5x Laemmli buffer and boiling.

Manipulation and analysis of Zebrafish embryos
Wild-type zebrafish (EK and AB strains) were kept under standardised conditions in a circulating water system (Tecniplast, Germany). Eggs were collected from natural matings. At the one to two cell stage, fertilized eggs were microinjected with a previously described SPRTN MO 25 and with in vitro synthesized, capped RNAs encoding different CHK1 variants fused to GFP, which were prepared from linearized plasmids using the mMessage mMachine SP6 Kit (Ambion). Alternatively, capped RNA encoding SPRTN variants were co-injected. Growth retardation was assessed at 9 hours post fertilization (hpf; 90% epiboly stage under control conditions) using an upright brightfield M125 microscope equipped with a IC80 HD camera (both Leica). Immunofluorescence stainings were essentially prepared as previously described 59 using a rabbit-anti-gH2Ax antibody (Genetex, 1:200 dilution) and imaged at a M205FA microscope equipped with a DFC365FX camera (both Leica).
SPRTN isolates were reduced with 5 mM DTT (Sigma-Aldrich; room temperature, 1 h), alkylated with 20 mM iodoacetamide (Sigma-Aldrich; room temperature, 1 h), and extracted using a double methanol/chloroform precipitation. Protein precipitates were resuspended in 6 M urea (with sonication), and diluted in a buffer containing 25 mM Tris-HCl (pH 8.5) and 1 mM EDTA until the urea concentration was < 1 M. Protein was subsequently digested with rLysC (Promega) for overnight at 37˚C. Peptides were then desalted using a SOLA HRP SPE cartridge (Thermo Scientific Cat# 60109-001) and dried.