The RIF1-Long splice variant promotes G1 phase 53BP1 nuclear bodies to protect against replication stress

Human RIF1 functions in DNA replication and damage repair. Of clinical interest, RIF1-depleted cells are highly sensitive to replication inhibitors such as Aphidicolin, but the reasons for this sensitivity have been enigmatic. Here we show that RIF1 must be present both during replication stress and in the ensuing recovery period to promote cell survival. RIF1 Long and Short isoforms are produced by alternative splicing. We find that RIF1-Long alone can protect cells against replication inhibition, but RIF1-Short is completely incapable of mediating protection. Consistent with its isoform-specific role in enabling survival, RIF1-Long is specifically required to promote the formation of the 53BP1 nuclear bodies that protect unrepaired damage sites in the G1 phase following replication stress. Overall, our observations show that RIF1 is needed at several cell cycle stages to support genome maintenance following replication insult, with the RIF1-Long isoform playing a previously unsuspected but crucial role in damage site protection during the ensuing G1 phase.

(a derivative of GFP 35 ) ( Fig. 2A) 36 . The expressed construct remains under control of the endogenous RIF1 promoter. Western blot analysis indicated that expression levels of mAC-RIF1 in the absence of Auxin were similar to those of endogenous untagged RIF1 (Fig. 2B).
Treatment for 24 hr with DOX and Auxin led to near-complete degradation of RIF1, as visualised by Western blotting (Fig. 2B) and microscopically (Fig. 2C). Using flow cytometry analysis of the RIF1-fused mClover tag, we established minimum concentrations of DOX and Auxin to allow effective depletion (Fig. S1A,B). Degradation was largely complete after 3 hr of Auxin treatment (Fig. S1C), while expression was restored to almost normal levels 5 hr after Auxin removal (Fig. S1D). Together, these results confirm the construction of a cell line allowing rapid depletion and re-expression of RIF1.
We visualised mAC-RIF1 based on Clover fluorescence, to test whether tagged protein retained the behaviour of endogenous RIF1. In live-cell imaging experiments we observed a pattern of numerous mAC-RIF1 foci throughout S phase nuclei (Fig. 2D, top row), often with 3-6 prominent foci superimposed on a pattern of more numerous smaller foci, consistent with previous studies 8, 30,37 . In prometaphase mAC-RIF1 exhibits a localisation pattern similar to that described for kinetochores (Fig. 2D,  RIF1 signal was absent for a short period at metaphase, quickly followed by reappearance of numerous smaller foci on telophase chromosomes coupled with localisation at anaphase bridges.
The mAC-RIF1 fusion therefore retains the localisation and functional characteristics of the endogenous RIF1 protein. Its highly dynamic behaviour indicates that RIF1 functions in several cell cycle phases to maintain chromosome stability.

RIF1 is needed during and after replication stress to promote cell proliferation
Prolonged auxin-induced degradation of RIF1 caused sensitivity to Aphidicolin, as expected ( Fig. 3A). To investigate when RIF1 function is required to protect against the effects of Aphidicolin, we synchronised cells first in G1 phase with Lovastatin 41,42 , and 8 hr after release from Lovastatin added Aphidicolin to induce replication stress (Fig. 3B, upper timeline). The reversible CDK1/cyclin B1 inhibitor RO-3306 43 was simultaneously added, to induce a temporary G2 arrest and prevent cells from proceeding into mitosis. After 28 hr, Aphidicolin and RO-3306 were removed to allow release, and then 4 hr later cells were plated for CFA measurement of cell viability 43 . Synchronisation timings were optimised using flow cytometry analysis of cell cycle progression (Fig. 3C, Fig. S3A).
Within the above synchronisation procedure, we either depleted RIF1 during the S phase Aphidicolin treatment period and re-expressed it for the recovery period (condition I, We performed similar conditional depletion experiments in asynchronous cell populations (as outlined in S3C), and observed comparable results (Fig. S3D, S3E), in that depletion of RIF1 either during or after Aphidicolin treatment led to sensitivity.
To summarise, these results imply that RIF1 must be present during both treatment and recovery to protect cells from the effects of replication stress induced by Aphidicolin.
We found that while the mAC-RIF1-L cell line showed resistance to Aphidicolin very similar to that of the parent mAC-RIF1 cells (Fig. 4C, black and red bars), the mAC-RIF1-S isoform in contrast conferred little protection against drug, producing sensitivity similar to that of cells lacking RIF1 altogether (Fig. 4C, grey and open bars). This result established that only RIF1-L can protect cells from replication stress caused by Aphidicolin, and that RIF1-S is ineffective in this role.
To confirm this finding in a different cell line, we used HEK293-derived stable cell lines with siRIF1-resistant cDNA constructs encoding either RIF1-L or RIF1-S, expressed under DOX control (Fig. S4A,B,C) 3 . We found that also in this cell line RIF1-L was able to protect against Aphidicolin treatment, while RIF1-S could not (Fig. 4D, filled red and grey bars).
Examining the previously described mechanisms through which RIF1 directly controls DNA replication, RIF1-L and RIF1-S appeared equally functional. In particular, both isoforms are equally effective in preventing hyperphosphorylation of the MCM complex ( Fig. 4E) and protecting blocked replication forks (Fig. 4F). RIF1-S can therefore repress origin activation and protect nascent DNA but cannot safeguard cells from Aphidicolin treatment, implying that responding to replication stress demands a further RIF1-mediated mechanism to promote cell survival, probably one that operates after the period of stress (Fig. 3) and that specifically requires RIF1-L (Fig. 4C, D).
In both its origin repression and nascent DNA protection functions, RIF1 acts as a PP1 substrate-targeting subunit 3,18,44 . To further test for separability of the effect of RIF1 in replication stress survival from its known roles in replication control, we investigated whether PP1 interaction is essential for RIF1-L to protect against Aphidicolin treatment. We used the HEK293 cell line expressing a version of RIF1-L mutated at the PP1 interaction motifs to prevent PP1 interaction 3 . This 'RIF1-L-pp1bs' protein was almost as effective as wild-type RIF1-L in conferring resistance to Aphidicolin (Fig. S5), indicating that RIF1-L acts largely independent of PP1 function in protecting cells from the effects of Aphidicolin.
This independence from PP1 reinforces the evidence that the function of RIF1 in protecting from replication stress is distinct from its previously known roles in replication control, which do require PP1.

RIF1-Long promotes 53BP1 nuclear body formation in G1 phase
We therefore considered other routes through which RIF1 might promote survival after Aphidicolin treatment, focusing especially on events occurring after the replication stress period itself. UFBs form after replication stress, but we found no clear difference in localization of RIF1-L and RIF1-S to UFBs in mitotic cells (not shown). A further consequence of replication stress is the formation of large 53BP1 nuclear bodies in the subsequent G1 phase, which protect unreplicated DNA damaged by chromosome breakage at mitosis 27,28 . We examined the formation of 53BP1 nuclear bodies in cells lacking RIF1-L, RIF1-S, or both RIF1 isoforms 12 hr after Aphidicolin treatment, limiting our analysis to G1 phase cells by counting only those that were cyclinA2-negative. The parental (mAC-RIF1) cell line showed an elevated fraction of cells with multiple large 53BP1 nuclear bodies (  S6A bottom left panel). The two RIF1 isoforms therefore differ in their effectiveness in promoting 53BP1 nuclear body formation following replication stress, with RIF1-L but not RIF1-S functional in this role. Since 53BP1 nuclear bodies are known to protect DNA damaged as a consequence of Aphidicolin treatment 27 , the defect in 53BP1 body formation when RIF1-L is not available is likely to be a major factor in the replication stress sensitivity of RIF1-deficient cells, and can explain the isoform specificity of the replication stress protection function of RIF1.

DISCUSSION
We sought in this study to understand mechanisms through which RIF1 protects against interruption to replication. In testing the function of the RIF1 isoforms, we found that RIF1-L is able to protect against replication stress while RIF1-S cannot. This deficiency of RIF1-S function was initially surprising, since RIF1-S appears competent to fulfil the known functions of RIF1 in DNA replication management-in particular RIF1-S is able to support replication licensing and control MCM phosphorylation (Fig. 4E) 3 , and to protect against nascent DNA degradation (Fig. 4F). Consistently however, all of these known functions of RIF1 in replication control (promotion of replication licensing, control of MCM phosphorylation, and nascent DNA protection) depend on PP1 recruitment by RIF1 3,18,21 ; while we find that protection against replication stress does not require PP1 interaction (Fig.   S5), again suggesting that the role of RIF1 in protecting from replication stress might involve previously undescribed mechanisms.
Testing the effects of conditional depletion in synchronised cultures revealed that RIF1 is still needed after a period of replication stress to guard against toxicity, implying that protection from Aphidicolin involves a further function of RIF1. We therefore investigated whether RIF1 operates in post-S phase replication stress response pathways, a line of enquiry that revealed a new function for RIF1 in promoting the assembly of 53BP1 nuclear bodies ( Fig. 5). This requirement for RIF1 for 53BP1 nuclear body assembly represents a surprising role reversal from the order of protein assembly in double-strand break repair, where RIF1 recruitment depends on 53BP1 1,2,4 . Remarkably, we found that RIF1-L but not RIF1-S can function in promoting 53BP1 body formation, potentially explaining the specific requirement for RIF1-L in protecting against replication stress, since 53BP1 body assembly represents an important step in correct handling of stress-associated damage to enable ongoing proliferation. RIF1-L may directly promote 53BP1 body assembly, or possibly assist with the transit of damaged sites through to G1 phase to allow such protective bodies to form.
Presently we do not understand the molecular mechanistic differences between RIF1-L and RIF1-S, or how the apparently small difference between the isoforms (the inclusion or exclusion of just 26 amino acids) either permits or prevents 53BP1 nuclear body formation.
One intriguing possibility is that RIF1-L is involved in the phase separation of 53BP1 45 recently described as important for assembling G1 phase nuclear bodies. The findings described here explain however that RIF1 contributes to recognising under-replicated and unrepaired sites for special protection and handling later in the chromosome cycle-in particular for delayed replication that guards against unscheduled recombinational repair to prevent the formation of pathological intermediates, as recently described 46 . Overall, this study highlights the multifunctional role of RIF1 in ensuring chromosome maintenance to promote the survival and proliferation of cells after replication stress, emphasising the importance of RIF1 for determining response to replication-inhibiting chemotherapeutic drugs.
The HCT116 mAC-RIF1 cell line was constructed as described 33,36 . HCT116 mAC-RIF1-L, HCT116 mAC-RIF1-S and HCT116 RIF1 KO cells were constructed as described below. HCT116 mAC-RIF1 mCherry-PCNA was constructed by introducing the mCherry-PCNA construct under control of the EF1 alpha promoter using the piggyBac system 47 .

HEK293 cell lines and culture conditions
HEK293-derived cell lines were cultivated in Dulbecco's Modified Eagle's Minimal medium supplemented with 10% foetal bovine serum (tetracycline-free), 100 U/ml penicillin, and 100 µg/ml streptomycin at 5% CO2 and ambient O2 at 37°C. Appropriate antibiotics were added for selection of integrated constructs.
To construct cell lines, pOG44 48 and pcDNA5/FRT/TO-based plasmids carrying the RIF1-L or RIF1-L-pp1bs gene were mixed in 9:1 molar ratio and used for transfection of Flp- In T-Rex 293 cells (Invitrogen) with Lipofectamine 3000 (Invitrogen). Transfections and hygromycin B selection of stably transfected cells were performed as described by the manufacturer. Clones were tested for doxycycline-dependent induction of GFP fusion proteins by western blot and microscopy.
To assess the effect of ectopically expressing RIF1, cells were transfected with either control siRNA or siRNA against human RIF1. 2 days later, cells were split with addition of 1 µg/ml DOX then incubated for 24 hr to induce expression of GFP-RIF1 variant proteins.
siRNA transfection was carried out using Lipofectamine RNAiMAX (Invitrogen) as described by the manufacturer. siRNA used were Human RIF1 siRNA (Dharmacon, D-027983-02) and Control siRNA against Luciferase (Dharmacon, D-001100-01). Synonymous base mutations in the ectopically expressed GFP-RIF1 constructs make them resistant to siRNA targeted against endogenous RIF1 1 . RIF1 expression was assessed by western blot using RIF1 and GFP antibodies.

HCT116 mAC-RIF1
To construct miniAID-mClover-fused RIF1 stable cell lines, HCT116 cells expressing the auxin-responsive F-box protein Oryza sativa TIR1 (OsTIR1) under the control of a Tet promoter were transfected using FuGENE HD (Promega) with a CRISPR/Cas9 plasmid targeting nearby the 1st ATG codon of the RIF1 gene (5'-TCTCCAACAGCGGCGCGAGGggg-3') together with a donor plasmid based on pMK345 36 , that contains a cassette (hygromycin resistance marker, self-cleaving peptide P2A, and mAID-mClover 36 ) flanked by 500bp homology arms. Two days after transfection cells were diluted in 10 cm dishes, to which 100 µg/mL of Hygromycin B Gold (Invivogen) was added for selection. After 10-12 days, colonies were picked for further selection in a 96well plate. Bi-allelic insertion of the donor sequence was checked by genomic PCR. Clones were tested for RIF1 expression and AID-mediated degradation of RIF1 by western blot, flow cytometry and microscopy.
To induce degradation of miniAID-mClover-fused RIF1, OsTIR1 expression was first induced by 0.2 µg/ml DOX added to the culture medium, to produce a functional SCF (Skp1-Cullin-F-box) ubiquitin ligase that directs degradation of an AID-tagged protein 33,34 .
After 24 hr, 10 µM Auxin (indole-3-acetic acid; IAA) was added to the culture medium to promote the interaction of mAC-RIF1 with SCF-OsTIR1, driving ubiquitination and mAC-RIF1 degradation. To suppress premature degradation of RIF1 in the presence of DOX, 100 µM of the TIR1 inhibitor Auxinole was added 36,49 . In subsequent depletion experiments we used a regime in which DOX was first added in the presence of Auxinole, and then Auxinole was replaced with Auxin. Optimisation of DOX and Auxin concentrations is shown in Figure   S1A and B. Unless otherwise stated, the above-mentioned drug concentrations were used throughout. with exon 31 (mAC-RIF1-L) or without (mAC-RIF1-S), followed by a Neomycin resistance marker ( Figure 4A). Transfection and clonal selection were carried out as described above.
Clones were tested for RIF1 expression by western blot.

HCT116 RIF1 KO
HCT116 cells were transfected with the same CRISPR/Cas9 plasmid that was used to construct HCT116 mAC-RIF1 cells, targeting near the 1 st ATG codon of the RIF1 gene (sequence as above). Simultaneously transfected was a donor plasmid containing a hygromycin resistance marker flanked by 500bp homology arms. Transfection and clonal selection was carried out as described above. Clones were tested for loss of RIF1 expression by western blot.

PCR primers
The following PCR primers were used to construct pcDNA5/FRT/TO-GFP-RIF1-L: The following PCR primers were used to amplify genomic DNA for the homology arms for the RIF1 KO donor plasmid pMK194: HA1 For: (see above) HA3 Rev: 5' -tcgctgcagcccgggggatcGGGGGCTCTGACCCCTGGCCGTCATGTCGG -3' HA4 For: 5' -aagcttatcgataccgtcgaCTTTGGAAGACCCTTCTGCCTCCCATGGAG -3' HA2 Rev: (see above) The following primers were used to amplify the C-terminal portion of either pcDNA5/FRT/TO-GFP-RIF1-L or pcDNA5/FRT/TO-GFP-RIF1 for the mAC-RIF1-L and mAC-RIF1-S donor plasmids: The following primers were used to amplify the homology arms for the mAC-RIF1-L and mAC-RIF1-S donor plasmids:
The GFP-RIF1 constructs used in this study are based on pcDNA5/FRT/TO-GFP-RIF1 1 , which carries human RIF1-S cDNA with GFP fused at its N-terminus. To construct pcDNA5/FRT/TO-GFP-RIF1-L, a PCR fragment containing RIF1-S cDNA was amplified from pcDNA5/FRT/TO-GFP-RIF1 using primers SH593 and SH594, and cloned into pIRESpuro3 vector (linearised by EcoRV and EcoRI) using In-Fusion HD cloning system, to create plasmid pSH1009. The NheI-NotI fragment of the plasmid pSH1009 was replaced by two PCR fragments amplified by SH572 & SH595 and SH596 & SH597 respectively using In-Fusion HD system, to construct pSH1011 which has RIF1-L cDNA. The NheI-PspOMI fragment of pcDNA5/FRT/TO-GFP-RIF1 was replaced by NheI-NotI fragment of the pSH1011 plasmid to construct pcDNA5/FRT/TO-GFP-RIF1-L.
Construction of a GFP-RIF1-S-pp1bs plasmid was previously described 3 . The GFP-RIF1-L-pp1bs construct was made following a similar strategy.
The plasmid pX330-U6-chimeric_BB-CBh-hSPCas9 from Feng Zhang (Addgene, 42230) 50 was used to construct the CRISPR/Cas vector for the guide RNA (sequence as above) according to the protocol of Ran et al 51 . Donor plasmids were based on pBluescript and constructed as described 33,36 . Primers for amplification of the homology arms and cDNA from RIF1-L and RIF1-S are listed under PCR primers.

Protein extraction and western blotting
To prepare whole cell protein extracts, cells were trypsinised and washed with Dulbecco's µM Auxin was added to the culture medium to degrade RIF1. Cells were incubated for 24 hr after which Aphidicolin was added and cells incubated for a further 24 hr, before washing twice with PBS and replacement with the appropriate Aphidicolin-free medium. For HEK293-derived cell lines, 1 µg/ml supplementary DOX was re-added 72 hr after Aphidicolin removal and cells were then incubated for a further 4 days. In the case of HCT116-derived cell lines, after Aphidicolin removal, cells were incubated for 7 days. At the end of the incubation period, colonies consisting of more than 20 cells were counted using a Nikon Eclipse TS100 microscope.

Flow cytometry
To assess DNA content, cells were recovered by trypsinisation, then fixed with 70% ethanol. Slides were air-dried and mounted with Prolong (Invitrogen). Samples were imaged under a Zeiss Axio Imager and analysed using ImageJ. CldU and IdU tract lengths were measured in double-labelled forks and the IdU/CldU ratio was used as an indicator of nascent DNA degradation.

Live cell imaging
Live cell imaging was performed using a DeltaVision microscope equipped with an incubation chamber and a CO2 supply (GE healthcare Life Sciences). HCT116 cells were cultured in a glass-bottomed dish (MatTek) containing the medium without phenol red at 37 ºC with 5% CO2. To visualize nuclei in live cells, 0.5 µM SiR-DNA (Spirochrome) was added to the medium before observation. Image analysis and quantification were performed using the Volocity software (PerkinElmer). The half-life of mClover signal was calculated using the Prism software (GraphPad).
Cells were treated with 1 µM Aphidicolin for 24 hr after which Aphidicolin was removed and cells were incubated for a further 12 hr. Cells were fixed with either a 10% neutral buffered formalin solution (Sigma, HT-5012) for 10 minutes at room temperature or with 100% methanol for 15 minutes at -20 ºC. Blocking was for 30 minutes with PBST/1%BSA at 4 ºC after which antibody staining was performed. Confocal microscopy was performed using an LSM880 + Airyscan (Zeiss). x63 magnification was used and Z-stacks were imaged (40 slices). Images were processed first to an airyscan image and then to a maximum intensity projection (MIP) using ZEN Black (Zeiss). Image analysis and quantification was performed using CellProfiler (Broad Institute) and statistics were calculated using Prism (Graphpad).

List of antibodies used in this study
The following antibodies were used for western blotting: