1. Cell Biology

SIRT6 is a DNA double-strand break sensor

  1. Lior Onn
  2. Miguel Portillo
  3. Stefan Ilic
  4. Gal Cleitman
  5. Daniel Stein
  6. Shai Kaluski
  7. Ido Shirat
  8. Zeev Slobodnik
  9. Monica Einav
  10. Fabian Erdel
  11. Barak Akabayov
  12. Debra Toiber  Is a corresponding author
  1. Department of Life Sciences, Ben-Gurion University of the Negev, Israel
  2. The Zlotowski Center for Neuroscience, Ben-Gurion University of the Negev, Israel
  3. Department of Chemistry, Ben-Gurion University of the Negev, Israel
  4. Division of Chromatin Networks, German Cancer Research Center (DKFZ), BioQuant, Germany
  5. Centre de Biologie Intégrative, CNRS UPS, France
https://doi.org/10.7554/eLife.51636
Share this article
Cite this article
  1. Lior Onn
  2. Miguel Portillo
  3. Stefan Ilic
  4. Gal Cleitman
  5. Daniel Stein
  6. Shai Kaluski
  7. Ido Shirat
  8. Zeev Slobodnik
  9. Monica Einav
  10. Fabian Erdel
  11. Barak Akabayov
  12. Debra Toiber
(2020)
SIRT6 is a DNA double-strand break sensor
eLife 9:e51636.
https://doi.org/10.7554/eLife.51636
  2. Download BibTeX
  3. Download .RIS

Abstract

DNA double-strand breaks (DSB) are the most deleterious type of DNA damage. In this work, we show that SIRT6 directly recognizes DNA damage through a tunnel-like structure that has high affinity for DSB. SIRT6 relocates to sites of damage independently of signaling and known sensors. It activates downstream signaling for DSB repair by triggering ATM recruitment, H2AX phosphorylation and the recruitment of proteins of the homologous recombination and non-homologous end joining pathways. Our findings indicate that SIRT6 plays a previously uncharacterized role as a DNA damage sensor, a critical factor in initiating the DNA damage response (DDR). Moreover, other Sirtuins share some DSB-binding capacity and DDR activation. SIRT6 activates the DDR before the repair pathway is chosen, and prevents genomic instability. Our findings place SIRT6 as a sensor of DSB, and pave the road to dissecting the contributions of distinct DSB sensors in downstream signaling.

eLife digest

DNA is a double-stranded molecule in which the two strands run in opposite directions, like the lanes on a two-lane road. Also like a road, DNA can be damaged by use and adverse conditions. Double-strand breaks – where both strands of DNA snap at once – are the most dangerous type of DNA damage, so cells have systems in place to rapidly detect and repair this kind of damage.

There are three confirmed sensors for double-strand break in human cells. A fourth protein, known as SIRT6, arrives within five seconds of DNA damage, and was known to make the DNA more accessible so that it can be repaired. However, it was unclear whether SIRT6 could detect the double-strand break itself, or whether it was recruited to the damage by another double-strand break sensor.

To address this issue, Onn et al. blocked the three other sensors in human cells and watched the response to DNA damage. Even when all the other sensors were inactive, SIRT6 still arrived at damaged DNA and activated the DNA damage response. To find out how SIRT6 sensed DNA damage, Onn et al. examined how purified SIRT6 interacts with different kinds of DNA. This revealed that SIRT6 sticks to broken DNA ends, especially if the end of one strand slightly overhangs the other – a common feature of double-strand breaks. A closer look at the structure of the SIRT6 protein revealed that it contains a narrow tube, which fits over the end of one broken DNA strand. When both strands break at once, two SIRT6 molecules cap the broken ends, joining together to form a pair. This pair not only protects the open ends of the DNA from further damage, it also sends signals to initiating repairs. In this way, SIRT6 could be thought of acting like a paramedic who arrives first on the scene of an accident and works to treat the injured while waiting for more specialized help to arrive.

Understanding the SIRT6 sensor could improve knowledge about how cells repair their DNA. SIRT6 arrives before the cell chooses how to fix its broken DNA, so studying it further could reveal how that critical decision happens. This is important for medical research because DNA damage builds up in age-related diseases like cancer and neurodegeneration. In the long term, these findings can help us develop new treatments that target different types of DNA damage sensors.

Introduction

DNA safekeeping is one of the most important functions of the cell, allowing both the transfer of unchanged genetic material to the next generation and proper cellular functioning. Therefore, cells have evolved a sophisticated array of mechanisms to counteract daily endogenous and environmental assaults on the genome. These mechanisms rely on the recognition of the damaged DNA and its subsequent signaling. This signaling cascade triggers responses such as checkpoint activation and energy expenditure, and initiates the DNA repair process (Bartek and Lukas, 2007; Bartek and Lukas, 2003; Ciccia and Elledge, 2010; San Filippo et al., 2008; Hoeijmakers, 2009; Iyama and Wilson, 2013; Jackson and Bartek, 2009; Lieber, 2008; Madabhushi et al., 2014). If DNA damage is not properly recognized, all downstream signaling will be impaired.

Among the various types of DNA damage, the most deleterious are double-strand breaks (DSBs), which can cause translocations and the loss of genomic material. Until now, very few DSB sensors have been identified, among them poly ADP-ribose polymerase-1 (PARP1), the MRN complex (MRE11, RAD50, NBS1) and Ku70/80 complex. All of these sensors initiate downstream signaling cascades which usually lead to the activation of specific repair pathways, such as homologous recombination (HR) or classical non-homologous end joining (C-NHEJ) (Andres et al., 2015; Sung et al., 2014; Woods et al., 2015). How a specific repair pathway is chosen is not fully understood, but it is known that the identity of the DSB sensor influences the outcome. For example, the MRN complex is associated with HR, whereas Ku70/80 is associated with C-NHEJ. Once DNA damage is recognized, transducers from the phosphoinositide 3-kinase family (e.g., ATM, ATR, and DNA-PK) are recruited to the sites of damage. They initiate a broad cascade, recruiting and activating hundreds of proteins which regulate the cellular response, including cell cycle progression, transcription, and metabolism. Ultimately, this response will determine whether the cell will live, senesce, or die. Failure to recognize and repair DSBs may lead to tissue ageing and disease (Ciccia and Elledge, 2010; San Filippo et al., 2008; Gasser et al., 2017; Ribezzo et al., 2016; Shiloh, 2014).

Sirtuin 6 (SIRT6) is a chromatin-bound protein from a family of NAD+-dependent deacylases and ADP-ribosylases. Through these functions, SIRT6 regulates DNA damage repair (DDR), telomere maintenance, and gene expression (Feldman et al., 2013; Jiang et al., 2013; Kugel and Mostoslavsky, 2014). The importance of SIRT6 to DNA maintenance is exemplified in SIRT6-KO mice phenotypes, which include accelerated ageing, cancer and neurodegeneration (Kaluski et al., 2017; Stein and Toiber, 2017; Tasselli et al., 2017; Zorrilla-Zubilete et al., 2018; Zwaans and Lombard, 2014). SIRT6-deficient cells exhibit genomic instability, increased aerobic glycolysis and defects in DNA repair, among other phenotypes (Kugel and Mostoslavsky, 2014; Stein and Toiber, 2017; Tasselli et al., 2017). Moreover, it was recently shown that the capacity of SIRT6 to repair DSB, but not to perform nucleotide excision repair (NER), is directly linked to longevity (Tian et al., 2019).

We have shown previously that SIRT6 is one of the earliest factors recruited to DSBs, arriving at the damage site within 5 seconds and allowing the opening of chromatin at these sites by recruiting the chromatin remodeler SNF2H (Toiber et al., 2013). In addition, the silencing of SIRT6 resulted in impaired downstream signaling, affecting the recruitment of key repair proteins such as Ku80, BRCA1 and 53BP1, among others, which are involved in both NHEJ and HR (Bunting et al., 2010; Chen et al., 2017; Daley and Sung, 2014; Escribano-Díaz et al., 2013; Gupta et al., 2014; McCord et al., 2009; Tang et al., 2013; Toiber et al., 2013). These studies indicate that SIRT6 plays important roles at very early stages of the DDR. The prominent role of SIRT6 in the early steps of DNA damage signaling raises the fascinating possibility that it is also directly involved in DSB sensing. In this work, we show that SIRT6 is indeed a DSB sensor, able to detect broken DNA and to activate the DNA damage signaling, revealing its key role in DNA repair initiation.

Results

SIRT6 arrives at sites of damage independently of other sensors or signaling

First, we set out to investigate the relationship between SIRT6 and the three known DSB sensors, PARP1, MRE11 (of the MRN complex), and Ku80 (of the Ku complex). PARP proteins are among the fastest known enzymes to arrive at DSBs, and their absence is known to impair the recruitment of DSB repair enzymes such as MRE11, NBS1 and Ku80 (Haince et al., 2008; Yang et al., 2018). We inhibited PARP activity by supplementing cells with Olaparib, and tracked SIRT6 recruitment to sites of laser induced damage (LID) by live-cell imaging. Interestingly, SIRT6 recruitment was found to be independent of PARP activity. SIRT6 arrived at the damage sites even when PARP proteins were inhibited, while the recruitment of the macro-H2A macro domain, which was used as a control, depended entirely on PARylation (Figure 1A–C, Figure 1—figure supplement 1A–C).

Subsequently, we silenced MRE11 and observed impaired NBS1 recruitment but no effect on SIRT6 (Figure 1D–F, Figure 1—figure supplement 1D–F). Ku80 silencing resulted in the expected defects in Ku70 recruitment, but did not impair SIRT6 arrival, in fact even larger amounts of SIRT6 were recruited to the site of damage (Figure 1D–F, Figure 1—figure supplement 1 G-I). Moreover, when we tested the effect of SIRT6-KO (Figure 1—figure supplement 1J) on the recruitment of MRE11 and Ku80, we found that while MRE11 recruitment was defective (Figure 1G–I), Ku80 was not affected by the lack of SIRT6 (Figure 1J–L). This suggests that SIRT6 may have a role in MRN recruitment or residency at DSB, but that the Ku complex is independent of it. Next, we silenced ATM and H2AX, which are both involved in DDR signaling (Figure 1—figure supplement 2A). Even though this produced defective signaling, as shown by decreased DDR signaling (Figure 1—figure supplement 2B–D), SIRT6 arrived at the sites of damage independently of these factors (Figure 1—figure supplement 2E–G).

Figure 1 with 2 supplements see all
Download asset Open asset
SIRT6 arrives at sites of damage independently of other repair factors.

(A–C) Imaging and AUC for SIRT6-GFP in cells with or without Olaparib. (A) Live imaging recruitment upon UV laser-induced damage (LID) shown by SIRT6-GFP in U2OS +/– Olaparib. Representative experiment examining SIRT6 recruitment to LID (n[+Ola]=23, n[–Ola]=23). (B) SIRT6 accumulation in same experiment as panel (A). (C) Average area under the curve (AUC) for cells +/– Olaparib in three replicate experiments. Error bars are the standard error of the mean (SEM) (n[+Ola]=38, n[–Ola]=39, p>0.05). (D–F) Imaging and AUC for SIRT6-GFP accumulation in shControl, shKu80 or shMRE11 Hela cells. (D) Average AUC from three experiments. Error bars are the SEM (shControl: n = 50; shKu80: n = 50, p<0.0005; shMRE11: n = 52, p>0.05). Accumulation of SIRT6-GFP (E) and imaging (F) from a representative experiment examining SIRT6 recruitment after LID (n[shControl]=28; n[shKu80]=30, n[shMRE11]=30). (G–I) MRE11-Cherry accumulation in response to LID in SIRT6 WT and KO U2OS cells. MRE11-Cherry imaging (G) and accumulation (H) in a representative experiment (n[WT]=20, n[KO]=16). (I) Mean AUC for three replicate experiments. Error bars are the SEM (n[WT]=36, n[KO]=33, p<0.0005). (J–L) Ku80-GFP accumulation in response to LID in SIRT6 WT and KO U2OS cells. Ku80-GFP imaging (J) and accumulation (K) in a representative experiment (n[WT]=17, n[KO]=17). (L) Mean AUC for three replicate experiments. Error bars are the SEM (n[WT]=33, n[KO]=33, p>0.05).

These results indicate that SIRT6 recruitment is independent of known DSB sensors and is upstream of ATM and H2AX phosphorylation. To understand whether SIRT6 is recruited through by signaling initiated at the sites of damage themselves, we tested whether it can be recruited by the initiation of a DNA damage response in the absence of actual DNA damage (lack of DSBs). To answer this question we took advantage of a tethering assay in which we used U2OS cells containing 256x lactose operator (LacO) repeats in their genome (Shanbhag et al., 2010; Tang et al., 2013). We transfected these cells with chimeric proteins containing lactose repressor (LacR) conjugated to known DDR-initiating repair enzymes (scheme in Figure 2A; Soutoglou and Misteli, 2008). In this system, the mere presence of ATM (ATM-LacR-Cherry) on chromatin initiates the DDR, as shown by H2AX ser-139 phosphorylation (γH2AX) (Figure 2—figure supplement 1A–B; Soutoglou and Misteli, 2008). However, in this system with no actual DNA damage, ATM failed to recruit SIRT6 to the LacO site, even though signaling was taking place and H2AX was phosphorylated (Figure 2B–C). As a control, we showed that known interactors such as SNF2H and Ku80 (McCord et al., 2009; Toiber et al., 2013) did recruit SIRT6 to the tethering sites (Figure 2B–C, Figure 2—figure supplement 1C–D). Moreover, MRE11 and NBS1 also recruited SIRT6 to the LacO site (Figure 2—figure supplement 1C–D), suggesting that there is either direct interaction between these sensors and SIRT6 or that they work together in a DDR complex.

Figure 2 with 1 supplement see all
Download asset Open asset
SIRT6 is not recruited by signaling.

(A) Schematic representation of the ‘Tethering assay’. Recruitment can occur through DDR signaling (ATM-LacR-Cherry) or through direct protein–protein interaction (SNF2H-LacR-GFP). (B, C) Recruitment of SIRT6-GFP/SIRT6-Cherry to LacO sites by ATM-LacR-Cherry (n = 30, p>0.05), SNF2H-LacR-GFP (n = 85, p<0.005) and GFP-LacR (n = 85). The bar chart in panel (B) depicts averages for3–6 experiments. Error bars are SEM.

Taken together, these results indicate that SIRT6 arrives at the sites of damage independently of MRE11, Ku80 and PARP activity, and that signaling itself is not sufficient to bring SIRT6 to the damage sites in the absence of actual DNA damage.

SIRT6 binds DNA DSBs directly

The findings described so far suggest that SIRT6 responds selectively to the actual damage, and that silencing or inhibiting key factors in the DDR do not affect its fast recruitment. Therefore, we tested whether SIRT6 could detect the actual DNA break on its own. We first measured SIRT6 capacity to bind naked DNA by electrophoretic mobility shift assay (EMSA). We found that SIRT6 was able to bind naked DNA without preference for a sequence (we tested different oligos and restricted sites, see Table 1) (Figure 3A–B, Figure 3—figure supplement 1A). We studied the preference of SIRT6 for several DNA damage structures, including dsDNA with blunt or overhanging ends as well as RNA. SIRT6 has the ability to bind them all, but it binds RNA with much lower affinity (Figure 3—figure supplement 1B). SIRT6 exhibits the highest affinity towards ssDNA (Kd = 1.39 μM), showing binding affinity values similar to those for MRE11 (Kd ~1 μM) (Williams et al., 2008) and Ku80 (Kd = 0.4 μM) (Arosio et al., 2002). Interestingly, on the basis of the curve fitting, SIRT6 seems to bind ssDNA at one site as a monomer. By contrast, there seems to be a cooperative effect when testing blunt and sticky-end DNA (Hill Slope greater than 1), suggesting that for open-ended dsDNA, two molecules of SIRT6 participate in binding, each SIRT6 molecule binding one DNA strand (Figure 3A–B, Figure 3—figure supplement 1A, scheme in Figure 3C). As all of the DNAs used in the EMSA were open-ended, we developed an additional DNA-binding assay based on the co-immuno-precipitation of a plasmid (IP-qPCR).

Table 1
DNA sequences used in the EMSA assay.
DNA sequences used in the EMSA assays
ssDNA5′GGGAAAGTTGACGGGAGGGTATTGGAGGTTAGTGGAGGTGAGTGG3′
ssDNA5′CCACTCACCTCCACTAACCTCCAATACCCTCCCGTCAACTTTCCC3′
dsDNA-Blunt5′CCACTCACCTCCACTAACCTCCAATACCCTCCCGTCAACTTTCCC3′
dsDNA-recessed5′CCACTCACCTCCACTAACCTCCAATACCCTCCCGTCAAC3′
dsDNA-recessed5′ACCTCCACTAACCTCCAATACCCTCCCGTCAACTTTCCC3′
dsDNA-Blunt5′AAGGTCGACACCACCTTTGAGAGCGCGCGGCCCACGCAGACCCACATGGCGCTGGTGCAGCTGGAGCGCGTGGGCCTCCTCCGCTTCCTGGTCAGCCAGAACGTCGACAAA3′
dsDNA-recessed5′TCGACACCACCTTTGAGAGCGCGCGGCCCACGCAGACCCACATGGCGCTGGTGCAGC
TGGAGCGCGTGGGCCTCCTCCGCTTCCTGGTCAGCCAGAACG		3′
RNA5′GCGAAGUCUUCGU3′
Figure 3 with 2 supplements see all
Download asset Open asset
SIRT6 binds DNA with no intermediates.

(A–B) Gel retardation assay of 32 P-5' end-labeled single-strand DNAs and sticky ended dsDNAs as a function of increasing concentrations of SIRT6-His (ssDNA, Kd = 1.48 ± 0.52; sticky dsDNA, Kd = 3.59 ± 0.17). (C) Suggested model of SIRT6 binding to ssDNA as a monomer or open ssDNA ends of dsDNA as a dimer. (D) Ability of SIRT6-Flag to bind to the DNA of circular, blunt-ended and sticky-ended cleaved plasmids. The bar chart depicts averages for three replicate experiments (error bars show SEM), after logarithmic transformation. (E) SIRT6-Flag DNA-binding ability for an open-ended +plasmid +/- NAD. Data are averages from four experiments, with error bars representing SEM (after logarithmic transformation). (F, G) Dimerization of SIRT6 at the LacO site, represented by the recruitment of SIRT6-Cherry by SIRT6-LacR-GFP (n = 181, p<0.005) or GFP-LacR (n = 104). Data are averages from four experiments, with error bars representing SEM.

In brief, flag-tagged repair proteins were purified and incubated with DNA, then immunoprecipitated along with the DNA that they bound. The DNA was later purified and its enrichment was measured by qPCR. Proteins were incubated either with a circular plasmid or with the same plasmid presenting blunt or sticky ends. As expected, NBS1, which does not bind DNA by itself (Myler et al., 2017), did not bind either plasmid (open or closed ends). By contrast, SIRT6 and MRE11 had high affinity to liner DNA, but they showed almost no binding to closed plasmids (Figure 3—figure supplement 1C). Moreover, SIRT6 exhibited a higher affinity for sticky ends, structures that show a high resemblance to DSBs, over blunt ends (Figure 3D). In addition, it did not distinguish between 3' or 5' overhangs (Figure 3—figure supplement 1D). These assays indicate that SIRT6 does not function by binding intact DNA or a particular sequence, but rather by binding to open DNA ends, and particularly to ssDNA. It is important to note that this capacity is independent of the presence of NAD+, the known cofactor of SIRT6 (Figure 3E), and the binding of DNA per se, does not activate SIRT6 catalytic activity (Figure 3—figure supplement 1E). Moreover, SIRT6 was able to protect the open ends of DNA from exonuclease activity (ExoI), preventing exonuclease cleavage just as in the case of MRE11 and implying that SIRT6 specifically binds to DNA ends (Figure 3—figure supplement 1F–H).

SIRT6 binds DNA ends as a dimer

Our EMSA results indicate that SIRT6 binds ssDNA with no cooperativity, suggesting a single binding site. By contrast, when the substrates were dsDNA oligos, we found the Hill coefficient to be greater than 1, indicating cooperativity (Figure 3A–B, Figure 3—figure supplement 1A). These results suggest that a single molecule of SIRT6 binds ssDNA. Even so, given two ssDNAs, such as would be present at an open-ended DSB, one SIRT6 molecule will interact with another, allowing a dimer of SIRT6 to bind a single molecule of dsDNA that has two open ends on a single side, 5′ and 3′ (see schematic Figure 3C). Together, the two SIRT6 molecules show cooperativity.

Interestingly, the known crystal structure of SIRT6 presents a dimer conformation (Jiang et al., 2013; You et al., 2017). To further characterize the structure of SIRT6 in a solution, we used size exclusion chromatography-multi-angle light scattering (SEC-MALS) and small-angle X-ray scattering (SAXS). Importantly, both methods showed that SIRT6 tends to aggregate; however, when using SEC-MALS, we noted that the aggregation was significantly reduced by the presence of DNA oligomers (Figure 3—figure supplement 2A), which suggests that SIRT6 is stabilized by (and favors) DNA interactions. SAXS data provide a low-resolution structure of SIRT6, presumably corresponding to a tetramer (Figure 3—figure supplement 1B–D), supporting the model suggested by the EMSA results (with dimers at the 5′ and 3′, a tetramer). The result obtained by SAXS does not exclude the presence of SIRT6 dimers or trimers in solution (see Table 2). Last, we measured dimerization in vivo by taking advantage of SIRT6-LacR-GFP localization at LacO sites and the recruitment of SIRT6-RFP, observing a significant co-localization of both SIRT6 molecules (Figure 3F–G), indicating that the bound SIRT6-GFP recruits the soluble SIRT6-RFP.

Table 2
Theoretical Rg (Å) derived from the SAXS data.
Theoretical Rg (Å)
SIRT6 dimer27.14
SIRT6 tetramer36
SIRT6 hexamer40

Overall, our predictions suggest that the SIRT6-DNA complex is organized in dimers, probably at each end of the DNA oligomers. Moreover, on the basis of the reconstructed SAXS structure, we show a compaction of SIRT6 in the presence of DNA, suggesting a conformational change (Figure 3—figure supplement 1B–D).

SIRT6 binds ssDNA through its core domain, which forms a ‘tunnel-like’ structure

SIRT6 has not been previously reported in the literature to be a DNA binding protein, so we aimed to identify the domain involved in ssDNA binding. To this end, we first analyzed the SIRT6 structure to find a potential DNA-binding domain. We found a region within the core domain (28 a.a.) that had potential to bind DNA (Figure 4A–C). We purified full-length SIRT6 (SIRT6 FL) and a fragment of the core domain alone (core: from a.a. 34 to 274). Both were able to bind DNA with similar affinities, indicating that the core domain is the main domain responsible for DNA binding (Figure 4D).

To understand which amino acids could be involved in the DSB binding, we mapped them to the known structure of SIRT6 (http://dnabind.szialab.org/). The model points to a subset of amino acids that are more likely to be involved in DNA binding. Surprisingly, these amino acids are concentrated near a physical structure that resembles a tunnel (Figure 4A). This tunnel is narrow and could accommodate ssDNA (Figure 4E), but not larger dsDNA. Without an open end, normal undamaged DNA could not enter this tunnel, but broken DNA ends could. Therefore, we hypothesized that the destruction or disruption of the tunnel would impair SIRT6 DNA-binding capacity. To test this hypothesis, we generated several point mutations of the amino acids in the tunnel-like structure of SIRT6 (Figure 4—figure supplement 1A–B). Purified SIRT6-MBP point-mutants were tested by EMSA to estimate their DNA-binding ability. As predicted, single point mutations in key amino acids at the tunnel (including the catalytic dead mutant H133Y) impaired the DNA- binding capacity (Figure 4F–G). The only mutant that showed no effect on binding was D63Y, in which the mutated amino acid did not impair the charge as strongly as the D63H mutation. Interestingly, mutations in D63 had previously been reported to provoke the loss of SIRT6 function in cancer, and have recently been shown to be lethal in humans (Ferrer et al., 2018; Kugel et al., 2015).

Figure 4 with 1 supplement see all
Download asset Open asset
SIRT6 binds DSB through its core domain.

(A) Predicted DNA-binding site based on the published SIRT6 structure, (http://dnabind.szialab.org/). Highlighted in yellow are the predicted DNA-binding amino acids in the SIRT6 core domain; red highlights show the tunnel-forming amino acids that were mutated. (B) Schematic representation of the SIRT6 core domain. (C) List of amino acids that are predicted to participate in the ‘tunnel-like’ structure. (D) DNA binding of an open-ended plasmid by full-length SIRT6 (p<0.0005) and by the SIRT6-core domain (p<0.005). Data are the log of averages from three experiments (with error bars respresenting SEMs). (E) SIRT6 ssDNA-binding prediction, based on the known SIRT6 structure with bound ssDNA. (F, G) Gel retardation assay of 32 P-5′ end-labeled ssDNAs with SIRT6-MBP mutants. Data are averages from three experiments (with error bars representing SEMs). (H) Ability of Flag-tagged mammalian Sirtuins to bind the DNA of circular and linear plasmids. Data are averages from 4–7 experiments (with error bars representing SEMs) (*, p <0.05; **, p <0.005; ***, p <0.0005).

As our prediction shows that the SIRT6 DNA-binding domain is in close proximity to its catalytic domain, we set out to examine how these mutations would affect SIRT6 catalytic activity. We performed a Fluor-de-lys assay to assess the mutant deacylation activity, using a H3K9-myristolatted peptide. Most mutants showed a decrease in SIRT6 activity compared to SIRT6-WT; however, A13W mutation showed increased SIRT6 activity (Figure 4—figure supplement 1C). This finding indicates that DSB binding and SIRT6 deacylation activity are not completely linked. However, given the close proximity of the two domains, they may share some of their functions because of the similarity of ssDNA and NAD+ molecules (ssDNA is a polymer of nucleotides; NAD+ consists of two nucleotides joined through their phosphate groups).

DNA binding ability is conserved among other Sirtuins

The core domain of SIRT6, where its DNA-binding domain is located, is conserved among all Sirtuins. Therefore, we tested whether other mammalian Sirtuins could bind DSB as well. Our results indicate that all Sirtuins have some capacity to bind broken-ended DNA, but some do it with a significantly lower affinity (Figure 4H, Figure 4—figure supplement 1D). Only SIRT7 showed binding capacity towards circular DNA, as previously described (Gil et al., 2013). It is also important to note that we tested mouse and human SIRT6 (mSIRT-Flag, hSIRT6-His) and found that both bind linear, but not circular DNA (Figure 4H, Figure 4—figure supplement 1D).

SIRT6 can initiate DNA damage response

As shown above, SIRT6 directly recognizes DNA breaks and arrives at the sites of damage independently of DDR signaling. Nonetheless, DNA damage recognition per se cannot activate the DDR. Therefore, we set out to examine whether SIRT6 also has the capacity to initiate the DDR through downstream signaling. To that aim, we took advantage of the previously described tethering assay using SIRT6-LacR-GFP/Cherry chimeras. Remarkably, SIRT6 has the same ability to induce the activation of the DDR as MRE11, measured by its capacity, compared to that of LacR-GFP/Cherry, to activate the phosphorylation of H2AX at the LacO site. Interestingly, the SIRT6 catalytic mutant SIRT6-HY was also able to initiate the DDR, raising the possibility that SIRT6 DDR initiation capacity is independent of its catalytic activity (Figure 5A–B).

Figure 5 with 1 supplement see all
Download asset Open asset
SIRT6 can initiate the DNA damage response.

(A, B) Initiation of the DDR, measured by co-localization of MRE11-LacR-Cherry (n = 136, p<0.005), SIRT6-LacR-GFP (n = 243, p<0.0001) or SIRT6 HY-LacR-GFP (n = 71, p<0.0001), compared to GFP-LacR (n = 310). Data are means for 4–9 experiments (error bars are SEMs). (C, D) Live imaging upon laser-induced damage (LID) of SIRT6-WT-GFP (n = 20) or SIRT6-HY-GFP (n = 20) in SIRT6 KO U2OS cells (n = 20). (D) Accumulation over time in 3 s intervals. (E) Average area under the curve of for three LID experiments (error bars are SEMs) (n[S6-WT]=40, n[S6-HY]=39, p<0.0001). (F) Initiation of DDR by full-length SIRT6-LacR-GFP (n = 127, p<0.005), Core-LacR-GFP (n = 66, p>0.05) or GFP-LacR (n = 64). Data are averages from 3–4 experiments (error bars show SEMs). (G) Initiation of DDR by SIRT6-LacR-GFP (n = 127, p<0.05), SIRT1-LacR-GFP (n = 44, p>0.05), SIRT2-LacR-GFP (n = 67, p<0.005), SIRT7-LacR-GFP (n = 68, p<0.005) or GFP-LacR (n = 64). Data are averages from 3–10 experiments (error bars show SEMs).

Nevertheless, because SIRT6 can generate dimers, endogenous SIRT6 could dimerize in the cells with SIRT6-HY-LacR, allowing the activation of the DDR. To test this possibility, we used nicotinamide (NAM) to inhibit endogenous SIRT6 activity (Figure 5—figure supplement 1A). However, even when the endogenous SIRT6 was inhibited (shown by the increase in H3K56ac), LacR-SIRT6-HY was still able to activate the DDR, supporting the evidence of DDR initiation that is independent of SIRT6 catalytic activity (Figure 5—figure supplement 1B).

It is important to highlight that SIRT6-HY has 50% less DNA-binding capacity to DSB than wildtype SIRT6 (Figure 4F–G); however, in this assay, it is forced to bind to the DNA through the LacR domain. In fact, we predicted that SIRT6-HY would fail to bind DNA if it was not tethered to chromatin through the LacR domain. To prove this hypothesis, we tested SIRT6-HY recruitment to DSBs in vivo using laser-induced damage in SIRT6-KO U2OS cells. Using SIRT6-KO cells rules out any contribution that an interaction with the endogenous SIRT6 might have. As expected, we found that unlike SIRT6-WT, SIRT6-HY does not arrive at sites of damage (Figure 5C–E). This finding strengthens our hypothesis that DNA binding is an important step in the role of SIRT6 in DSB repair, and that the residue that is defective in the SIRT6-HY mutant is critical for SIRT6-DSB binding.

To study whether SIRT6 activity and initiation capacity are separate, we tested Core-LacR-GFP, which has an active catalytic domain but lacks the C and N terminus of SIRT6 (Tennen et al., 2010). We observed that Core-LacR-GFP failed to activate the DDR (Figure 5F, Figure 5—figure supplement 1C), suggesting that other domains play a more prominent role in initiating signaling.

Moreover, we tested the initiation capacity of LacR-SIRT1, SIRT2 and SIRT7 in the tethering assay, because all of these Sirtuins have the ability to localize to the nucleus and have been associated with DNA repair (Jeong et al., 2007; Li et al., 2016; Paredes and Chua, 2016; Rifaï et al., 2018; Vazquez et al., 2017; Zhang et al., 2016). Remarkably, SIRT2 and SIRT7 could initiate the DDR, but SIRT1 could not (see note in 'Materials and methods') (Figure 5G, Figure 5—figure supplement 1D). Although other Sirtuins have some binding activity and some initiation capacity, SIRT6 is unique for having both.

Taken together, these experiments indicate that although SIRT6 binds DNA through its core domain, the activation of downstream signaling does not require the catalytic activity of SIRT6, but its N and C terminus are required for DDR activation.

Last, we tested whether SIRT6 could recruit repair factors of the DDR cascade and whether it shows a preference for a certain repair pathway. Although we observed a more prominent effect of SIRT6 on the recruitment of the HR initiator MRE11 rather than that of the NHEJ initiator Ku80 (Figure 1G–L), it was previously reported that SIRT6 affects both repair pathways (Chen et al., 2017; Mao et al., 2011; McCord et al., 2009; Tian et al., 2019; Toiber et al., 2013). Indeed, we noticed that SIRT6 deficiency results in impaired recruitment of both 53BP1 and BRCA1 to the sites of laser-induced DSBs, suggesting impaired activation of both NHEJ and HR (Figure 6—figure supplement 1A).

In order to test SIRT6's ability to recruit these and other DDR factors to the sites of damage, we took advantage of the tethering system once more. Our results show that SIRT6 can recruit proteins that are involved in HR, such as MRE11, NBS1, ATM and BRCA1, as well as proteins that are involved in NHEJ, such as Ku80, Ku70 and 53BP1 (Figure 6A–B). As a control, we tested co-localization with CDT1, a nuclear protein that does not participate in the DDR. As expected, CDT1 was neither recruited by SIRT6 nor by GFP alone.

Figure 6 with 3 supplements see all
Download asset Open asset
SIRT6 can recruit enzymes of both the NHEJ and HR repair pathways.

(A, B) Percentage repair enzymes that are co-localized with SIRT6-LacR-GFP/Cherry at LacO sites. IF with Flag antibody. Data are averages for 3–6 experiments (with error bars representing SEMs) (*, p<0.05; **, p<0.005; ***, p<0.0005; ****, p<0.00005).

As SIRT6 DDR activation is independent of its catalytic activity, we further examined whether it is needed for DDR protein recruitment. Taking advantage of the tethering assay, we observed that both SIRT6-WT and SIRT6-HY recruited 53BP1 and BRCA1, meaning that the recruitment is independent of SIRT6 catalytic activity (Figure 6—figure supplement 2A–B).

53BP1 and BRCA1 can antagonize each other, and a change in their concentration within the cell may influence the recruitment capacity. Therefore, we used IF to test whether overexpression of these proteins results in a different outcome from that produced by the endogenous proteins. However, the results were very similar, suggesting that the recruitment is independent of the amount of protein in the cell, and that an additional layer of regulation would influence the recruitment (Figure 6—figure supplement 2C–D).

The tethering assay can detect both protein–protein interaction or recruitment through signaling. To differentiate these two possibilities, we inhibited DDR signaling by supplementing the media of the cells with Wortmannin, thus inhibiting ATM, ATR and DNA-PKc (scheme in Figure 6—figure supplement 3A). Our results indicate that when these kinases are inhibited (shown by a reduction in γH2AX levels), the recruitment of both 53BP1 and BRCA1 to the LacO site is reduced (Figure 6—figure supplement 3B–D). However, the recruitment of the DDR initiators Ku80 and MRE11 is not affected by Wortmannin, suggesting that their recruitment is based on protein–protein interactions and not on downstream signaling alone (Figure 6—figure supplement 3E). These results indicated that SIRT6 participates in DDR activation through the initiation of signaling and the recruitment of various proteins, which lead to the different DNA-repair pathways.

Discussion

In this work, we discovered a novel function for the chromatin factor SIRT6 as a DSB sensor that is able to bind DSBs and initiate the cellular DDR.

We showed that SIRT6 can bind DNA with high affinity for ssDNA and open-ended dsDNA. We believe that the binding occurs through a tunnel-like structure in the protein core domain, close to its catalytic site. This structure could only fit ssDNA, and whereas other proteins require resection for ssDNA identification, 3–4 bases are enough for SIRT6.

By generating several point mutations in the hypothesized DNA-binding site, we managed to reduce the DNA-binding capacity of SIRT6, also reducing the catalytic activity. However, A13W and D63Y mutations raise the possibility that, despite the proximity of these sites, these abilities are distinct ones. The D63Y mutation had no effect on DNA binding, but it caused a significant reduction in SIRT6 catalytic activity. A13W mutation, on the other hand, resulted in an increase in catalytic activity along with a slight reduction in DNA binding.

In addition, we showed that SIRT6 can arrive at the sites of DSBs independently of the known sensors MRE11 and Ku80 and of PARP activity, and can activate the DDR on its own. We also observed that its catalytic activity is not necessary for DDR initiation when it is already bound to the DNA, as shown by the ability of SIRT6-HY-LacR to initiate the DDR. However, because the binding capacity in the HY mutant is reduced, we believe that SIRT6-HY is not able to bind and remain attached to the DNA (as shown by its impaired recruitment to laser-induced damage sites), and therefore that all DDR initiation would be impaired by this mutant. Interestingly, even though the initiation of the DDR occurs when SIRT6 is catalytically inactive, it cannot be initiated by the active core-domain alone. These results suggest a complex relationship between binding capacity and activation, in which binding per se cannot result in DDR signaling.

Given that the core domain, which contains both the catalytic domain and the DNA-binding domain of SIRT6, is conserved among Sirtuins, we also showed that other Sirtuins share the ssDNA-binding capacity (but with different affinities). This is especially interesting as Sirtuins are present in the cell at different cellular locations (cytoplasm, nucleus and mitochondria) and have different catalytic activities (deacetylases, deacylases, and ADP ribosylases) (Liszt et al., 2005). This suggests that the DSB-binding capacity could be relevant in other cellular compartments, for example, in mitochondrial DNA repair. When nuclear SIRT2 and SIRT7 were forced to localize to the DNA by the LacO-LacR tethering assay, they were also able to initiate the DDR. However, SIRT7 lacks the broken-DNA binding specificity and SIRT2 has a poor binding capacity, which would impair their roles as DNA damage sensors.

These findings open new possibilities for the cellular functions of the Sirtuin family; nevertheless, we believe in the uniqueness of SIRT6 as it possesses all of these abilities at once.

The placing of SIRT6 as a sensor of DSBs might explain why the lack of SIRT6 gives rise to one of the most striking phenotypes in humans, monkeys and mice, including phenotypes that are typically associated with genomic instability such as premature ageing, accelerated neurodegeneration, tissue atrophy and cancer (Kugel and Mostoslavsky, 2014; Tasselli et al., 2017). In particular, SIRT6 is involved in several repair pathways. As a sensor and DDR initiator, its absence would have deleterious effects on the whole downstream DDR signaling. Our results point out that its role begins as a DSB sensor (although it may recognize other DNA lesions), recognizing and initiating the DDR independently of other factors. SIRT6 has multiple functions in the context of chromatin (Kugel and Mostoslavsky, 2014), including transcriptional regulation. Thus, it might seem somewhat paradoxical that it can initiate the DDR response by merely binding to damage sites.

It is not particularly clear how SIRT6 can selectively activate the DDR when bound to DNA damage sites but not when bound to sites of transcription regulation. A possible explanation could rely on the fact that transcription factors are very dynamic, and they usually bind chromatin transiently (Hager et al., 2009). Therefore, we speculate that SIRT6, similarly to MRE11, probes the DNA transiently, and that even though it is constantly present in chromatin, its binding to unbroken DNA is not as tight as when it encounters broken DNA (as seen in the binding assays) (Myler et al., 2017). Tighter binding of SIRT6 might allow stabilization through protein interactions and modifications, analogous to the processes that occur with MRE11, NBS1, ATM and other DDR proteins. It is also possible that SIRT6 undergoes a conformational change when bound to broken DNA. However, our tethering system suggests that its continuous presence in chromatin (in the absence of broken DNA to bind) is sufficient to initiate the DDR cascade.

Interestingly, unlike other factors, SIRT6 recruitment and kinetics are not affected by PARP activity, making it independent of PARylation and giving it an advantage over other factors that require PARylation for fast recruitment (Mao et al., 2011). This feature could be relevant as an adjuvant therapy in cancer treatment (Beck et al., 2014; Haince et al., 2008).

It is also important to note that although SIRT6 can recruit proteins of both HR and NHEJ and its deficiency affects both pathways, SIRT6 KO impaired the recruitment of MRE11, but not Ku80, to sites of laser-induced damage. It is possible that the Ku complex does not require SIRT6 for recognition, yet it may require SIRT6 chromatin remodeling activity in later repair steps as NHEJ repair is affected by the lack of SIRT6. Alternatively, as in the case of PARP1 and the MRN complex, SIRT6 might compete with the Ku complex for DSB binding and DDR initiation (Myler et al., 2017; Yang et al., 2018) .

Our findings place SIRT6 at the beginning of the DDR response as a novel DSB sensor, but how it affects the DSB repair pathway choice still needs to be investigated. Nevertheless, as there is significant cross-talk between the pathways (seen, for example, by the involvement of the HR initiator MRE11 in NHEJ [Xie et al., 2009]), it is possible that it has roles in both.

In conclusion, we have demonstrated that SIRT6 has a role as an independent DNA damage sensor. This is critical for the initiation of the DSB-DNA damage response and hence for the support of genomic stability and health.

Materials and methods

Key resources table
Reagent type
(species) or resource		DesignationSource or
reference		IdentifiersAdditional
information
Recombinant DNA reagentATM-LacR-CherrySoutoglou and Misteli, 2008
Recombinant DNA reagentCDT1-TagRFPThermoFisherP36237
Recombinant DNA reagentCMV-FlagToiber et al., 2013
Recombinant DNA reagentMRE11-FlagWu et al., 2008
Recombinant DNA reagentMRE11-LacR-CherrySoutoglou and Misteli, 2008
Recombinant DNA reagentmRFP-SIRT6Kaidi et al., 2010 - retracted
Recombinant DNA reagentNBS1-FlagWu et al., 2008
Recombinant DNA reagentNBS1-LacR-Cherry
Recombinant DNA reagentpcDNA3.1(+)Flag-His-ATM-WTAddgene31985
Recombinant DNA reagentpcDNA5-FRT/T0-Flag-53BP1Addgene52507
Recombinant DNA reagentpDEST 3x Flag-pcDNA5-FRT/T0-BRCA1Addgene52504
Recombinant DNA reagentpEGFP-C1-FLAG-Ku70Addgene46957
Recombinant DNA reagentpEGFP-C1-FLAG-Ku80Addgene46958
Recombinant DNA reagentpEGFP-C1-FLAG-XRCC4Addgene46959
Recombinant DNA reagentpEGFP- SIRT6Kaidi et al., 2010 - retracted
Recombinant DNA reagentpET28 hSIRT6-HisGertman et al., 2018
Recombinant DNA reagentpHPRT-DRGFPAddgene26476
Recombinant DNA reagentpMal-C2-hSIRT6-WTGertman et al., 2018
Recombinant DNA reagentpMal-C2-hSIRT6-A13WThis paper
Recombinant DNA reagentpMal-C2-hSIRT6-D63HThis paper
Recombinant DNA reagentpMal-C2-hSIRT6-D63YThis paper
Recombinant DNA reagentpMal-C2-hSIRT6-H133YGertman et al., 2018
Recombinant DNA reagentpMal-C2-hSIRT6-D188AThis paper
Recombinant DNA reagentpMal-C2-hSIRT6-D190WThis paper
Recombinant DNA reagentpMal-C2-hSIRT6-I217AThis paper
Recombinant DNA reagentpQCXIP-Cherry-LacRThis paper
Recombinant
DNA reagent		pQCXIP-SIRT6 Core-GFP-LacRThis paper
Recombinant DNA reagentpQCXIP-GFP-LacRAddgene59418
Recombinant DNA reagentpQCXIP-KU80-GFP-LacRThis paper
Recombinant DNA reagentpQCXIP-mSIRT6-Cherry-LacRThis paper
Recombinant DNA reagentpQCXIP-mSIRT6-GFP-LacRThis paper
Recombinant DNA reagentpQCXIP-mSIRT6-H133Y-GFP-LacRThis paper
Recombinant DNA reagentpQCXIP-SIRT1-GFP-LacRThis paper
Recombinant DNA reagentpQCXIP-SIRT2- GFP-LacRThis paper
Recombinant DNA reagentpQCXIP-SIRT7- GFP-LacRThis paper
Recombinant DNA reagentSIRT1-FlagZhong et al., 2010
Recombinant DNA reagentSIRT2-FlagAddgene13813
Recombinant DNA reagentSIRT3-FlagAddgene13814
Recombinant DNA reagentSIRT4-FlagAddgene13815
Recombinant DNA reagentSIRT5-FlagAddgene13816
Recombinant DNA reagentSIRT6 CoreTennen et al., 2010
Recombinant DNA reagentmSIRT6-WT-FlagZhong et al., 2010
Recombinant DNA reagentSIRT7-FlagAddgene13818
Recombinant DNA reagentSNF2H-WT-GFP-LacRKlement et al., 2014
AntibodyAlexa Fluor
488 AffiniPure Donkey Anti-Rabbit IgG (H+L)		Jackson Immunoresearch711-545-152IF (1:200)
AntibodyAlexa Fluor
594 AffiniPure
Donkey Anti-Rabbit IgG (H+L)		Jackson Immunoresearch711-585-152IF (1:200)
AntibodyAlexa Fluor
488 AffiniPure Donkey Anti-Mouse IgG (H+L)		Jackson Immunoresearch715-545-150IF (1:200)
AntibodyAlexa Fluor
594 AffiniPure Goat Anti-Mouse IgG (H+L)		Jackson Immunoresearch115-585-062IF (1:200)
Antibody53
BP1		Snata-Cruz Biotechnologysc-22760IF (1:300)
AntibodyBRCA1Snata-Cruz Biotechnologysc-7298WB (1:1000)
AntibodyFlagSigma-AldrichF1804IF (1:900), WB (1:1000)
AntibodyFlag BeadsSigma-AldrichA2220IP
Antibodygamma H2A.X (phospho s139)abcamab2893IF(1:3000), WB (1:1000)
AntibodyGoat Anti-Rabbit IgG H and L (HRP)abcamab6721WB (1:10000)
AntibodyHistone H3abcamab1791WB (1:5000)
AntibodyHistone H3 (acetyl K56)abcamab76307WB (1:1000)
AntibodyHSC 70Snata-Cruz Biotechnologysc-7298WB (1:1000)
AntibodyKu80Cell Signaling#2180WB (1:1000)
AntibodyMRE11abcamab214WB (1:1000)
Antibodyphospho-ATM (Ser1981)Cell Signaling#5883WB (1:1000)
Antibodyphospho-Histone H2A.X (Ser139)Millipore05–636IF (1:1500)
AntibodyRabbit Anti-Mouse IgG H and L (HRP)abcamab97046WB (1:10000)
AntibodySIRT6abcamab62739WB (1:1000)
AntibodyTubulinMerckMAB1637WB (1:1000)
Peptide, recombinant proteinSIRT1 HumanPROSPECPRO-1909DNA binding assay
Peptide, recombinant proteinSIRT3 HumanPROSPECPRO-462DNA binding assay
Peptide, recombinant proteinSIRT5 HumanPROSPECPRO-1774DNA binding assay
Peptide, recombinant proteinSIRT6 HumanPROSPECPRO-282DNA binding assay

Cell cultures

Request a detailed protocol

All cells were cultured in DMEM and 4.5 g/l glucose, supplemented with 10% fetal bovine serum, 1% penicillin and streptomycin cocktail and 1% L-glutamine. Cells were cultured with 5% CO2 at 37°C.

All lines were confirmed to be Mycoplasma-free using a hylabs Hy-mycoplasma Detection PCR Kit with internal control (Cat No. KI 5034I).

Cells were authenticated by the Biochemical Core Facility of the Genomics Center at the Technion-Israel Institute of Technology.

Plasmids and transfections

Request a detailed protocol

To prepare pQCXIP-msirt6-GFP-LacR, mouse sirt6 without a stop codon was amplified by PCR and introduced in frame with GFP-LacR into the AgeI site of plasmid pQCXIP-GFP-LacR (Addgene, 59418).

pQCXIP-mSIRT6-H133Y-GFP-LacR was prepared by Quick Change Site-directed mutagenesis of mSIRT6 flanked by AgeI sites in pGEM, and after sequencing, introduced to the AgeI site in frame with the fused GFP-LacR of pQCXIP-GFP-LacR (Addgene, 59418).

pQCXIP-Cherry-LacR was prepared by excision of the AgeI/XhoI GFP fragment of pQCXIP-GFP-LacR and exchanged with AgeI/XhoI mCherry amplified from pDEST-mCherry-LacR-BRCA1 (Addgene, 71115).

pQCXIP-mSIRT6-Cherry-LacR was prepared by introducing the AgeI mSIRT6 from pQCXIP-KU80-GFP-LacR and by introducing KU80, amplified from pEGFP-C1-FLAG-Ku80 (Addgene, 46958), into the AgeI site of pQCXIP-GFP-LacR in frame with GFP.

pQCXIP-hSIRT1-GFP-LacR was prepared by inserting the amplified SIRT1 from SIRT1-Flag (Mostoslavsky Lab) with the AgeI site in frame with the GFP-LacR of plasmid pQCXIP-GFP-LacR (Addgene, 59418).

pQCXIP-hSIRT2-GFP-LacR was prepared by inserting the amplified SIRT2 from SIRT2-Flag (Addgen #13813) with the AgeI site in frame into the GFP-LacR of plasmid pQCXIP-GFP-LacR (Addgene, 59418).

pQCXIP-hSIRT7-GFP-LacR was prepared by inserting the amplified SIRT7 from SIRT7-Flag (Addgen #13818) with the AgeI site in frame into the GFP-LacR of plasmid pQCXIP-GFP-LacR (Addgene, 59418).

pQCXIP-Core hSIRT6-GFP-LacR was prepared by inserting the amplified 233 amino acid (aa) core region from aa 43 to aa 276 of human SIRT6 and introducing it into the AgeI site of pQCXIP-GFP-LacR (Addgene, 59418) with an additional methionine before aa 43 and in frame with the GFP-LacR of the plasmid.

pMal-C2-hSIRT6 A13W, D63H, D63Y, W188A, D190Wand I217A were prepared by Quick Change Site-directed Mutagenesis on pMal-C2-hSIRT6. The mutation was affirmed by sequencing.

All PCRs were performed with Hot start, KAPA HiFi #KM 2605 or abm Kodaq #G497-Dye proofreading polymerases. All clones were sequenced for validation, and expression of the fluorescent fusion proteins were checked by transfection into cells. All transfections were performed using PolyJet In Vitro Transfection (SignaGen, SL100688), according to the manufacturer's instructions.

Immunofluorescence

Request a detailed protocol

U2OS cells were washed with PBS and fixed with 2% paraformaldehyde for 15 min at room temperature, followed by an additional wash. Quenching was then performed with 100 mM glycine for 5 min at room temperature (RT). Cells were permeabilized (0.1% sodium citrate, 0.1% Trition X-100 [pH 6], in deionized distilled water [DDW]) for 5 min and washed again. After 1 hr blocking (0.5% BSA, 0.1% Tween-20 in PBS), cells were incubated with primary antibody diluted in blocking buffer over night at 4°C. The next day, cells were washed three times with wash buffer (0.25% BSA, 0.1% Tween-20 in PBS), incubated for 1 hr with secondary antibody (diluted in blocking buffer 1:200) at RT and washed three more times. Cells were then DAPI stained for three minutes at RT and washed with PBS twice before imaging.

Tethering assay

Request a detailed protocol

U2OS cells containing 256X LacO sequence repeats in their genome were transfected with plasmids of chimeric LacR-DDR enzyme-GFP/Cherry proteins. Cells were either co-transfected with a second plasmid of a fluorescent/Flag-tagged protein or immuno-stained (see 'Immunofluorescence') for an endogenic protein.

Cells expressing both proteins of interest and exhibiting visible foci of LacR-DDR-GFP/Cherry at LacO sites were located using an Olympus IX73 fluorescent microscope, whereas co-localization between both proteins was assessed visually using Olympus CellSens Software. Co-localization is defined as the common localization of large foci of the two proteins of interest at the LacO site. Co-localization was assessed as either positive (1) or negative (0). From this analysis, the percentage of cells that exhibit co-localization (positive cells) was calculated, and defined as ‘percentage of co-localization between two proteins’. The co-localization percentage for each protein of interest was compared to the co-localization percentage with LacR-GFP/Cherry as a control.

Notes: the pQCXIP-Ku80-GFP-LacR plasmid used in this assay contains Ku80 that was acquired from Addgene (cat. #46958) and contains the D158G mutation.

The pQCXIP-SIRT1-GFP-LacR plasmid used in this assay contains SIRT1 that was obtained from the Mostoslavsky lab (Zhong et al., 2010). This protein variant is lacking 79 amino acids in the N-terminus.

Immunoprecipitation (IP)

Request a detailed protocol

Flag-tagged proteins were purified from transfected HEK293T cells. Cells were collected and washed with PBS. Cell disruption was performed in lysis buffer (0.5M KCl, 50 mM Tris-HCl [pH 7.5], 1% NP40, 0.5M DTT, 200 mM TSA and protease and phosphatase inhibitors in DDW) by 10 min rotation at 4°C. Cell debris were sedimented by 15 min centrifugation at 21,000 g. Lysate was collected and added to ANTI-FLAG M2 Affinity Gel (Sigma-Aldrich, A2220) beads for 2 hr rotation at 4°C. Beads were then washed three times with lysis buffer and once with SDAC buffer (50 mM Tris-HCl [pH 9], 4 mM MgCl, 50 mM NaCl, 0.5 mM DTT, 200 mM TSA and protease and phosphatase inhibitors in DDW). Proteins were released by flag-peptide.

Expression and purification of recombinant SIRT6 in Escherichia coli

Request a detailed protocol

Expression and purification of His-tagged and MBP-tagged proteins in E. coli were performed as previously described by Gertman et al. (2018).

Fluorescence recovery after photobleaching (FRAP)

Request a detailed protocol

FRAP experiments (laser-induced damage) were performed as previously described by Toiber et al. (2013). In brief, cells were plated in Ibidi µ-Slide eight-well glass bottom plates (Cat. No.: 80827) and transfected with the desired fluorescent plasmid. Pre-sensitization with Hoechst (1 mM) was done for 10 min before the experiment. FRAP experiments were carried out using a Leica SP5 microscope (German Cancer Research Center (DKFZ) and BioQuant, Heidelberg, Germany) or using a LSM880 microscope (Ben Gurion University, Be’er Sheva, Israel) with a 63X oil immersion objective. Images were acquired in a 512 × 512 format with a scan speed of 1,400 Hz. Circular bleach spots of 2 µm diameter were positioned either at a damage site or at a distant reference site. Spots were bleached with an argon laser of 488 nm with a power of 1 mW in the back aperture of the objective. Images were taken at 3 s intervals, with three baseline images taken before bleaching. Acquisition before bleaching was used for normalization of each cell intensity (average of the baseline intensity of the whole cell nucleus prior to DNA damage). Images analysis and fluorescence assessment were performed using ImageJ 1.52i software.

To assess protein amounts and to compare between the different conditions, area under the curve was calculated using a MATLAB pipeline.

DNA-binding assay

Request a detailed protocol

Open-ended plasmids were prepared in advance by incubating DR-GFP plasmids with EcoRV for blunt ends, KpnI for 3' over hang or SalI for 5' over hang according to manufacture instructions. Circular plasmids were subjected to the same conditions with no restriction enzyme.

To achieve protein–DNA binding, flag-tagged proteins that were previously immunoprecipitated were incubated at 37°C for 1 hr with same amount of circular or open-ended DNA, 1:5 of 5X deacetylation buffer (50 mM Tris HCl [pH 8], 50 mM NaCl, 4 mM MgCl2 and 0.5 mM DTT in DDW), and 1:50 50X protease inhibitors in DDW.

ANTI-FLAG M2 Affinity Gel (Sigma-Aldrich, A2220) beads were blocked with 5% BSA supplemented with 1X deacetylation buffer (with 1% phosphate inhibitors) by rotation for 1 hr in 4°C. Beads were then centrifuged (1000 g, 3 min, 4°C) and buffer was changed to clean deacetylation buffer 1X. Beads were then distributed equally between all samples.

To achieve beads–protein binding, samples were rotated for 2 hr in 4°C. After rotation, samples were centrifuged (1000 g, 3 min, 4°C) and washed 3 times with 1 ml of wash buffer (0.1% SDS, 0.5% Triton x-100, 2 mM EDTA, 20 mM Tris-HCl [pH8] and 150 mM NaCl in DDW).

Protein–DNA complexes were then released by two rounds of 20 min vortexing at room temperature with 100 μl elution buffer (0.1M NaHCO3 and 1% SDS in DDW).

For His-tagged proteins (acquired from PROSPEC), the assay was performed using HisPur Ni-NTA Resin (ThermoFisher, 88221) under the same conditions with the appropriate buffers (binding buffer — 20 mM Tris HCl [pH 8], 150 mM NaCl, 10% PMSF, 1% phosphatase inhibitors; wash buffer — 20 mM Tris HCl [pH 8], 150 mM NaCl, 20 mM imidazole; elution buffer — 20 mM Tris HCl [pH 8], 150 mM NaCl, 500 mM imidazole).

Notes: the SIRT1-Flag used in this assay was obtained from the Mostoslavsky lab (PDMI: 20141841). This protein variant is lacking 79 amino acids in the N-terminus.

The SIRT1-His used in this assay was acquired from PROSPEC (https://www.prospecbio.com/sirt1_human). This SIRT1 is a 280 aa poly-peptide (aa 254–495).

DNA isolation

Request a detailed protocol

1:1 vol of phenol:chloroform:isoamyl alcohol (25:24:1) was added to the eluted DNA from the DNA binding assay, vortexed and centrifuged at RT for 5 min at 17,000 g. The top aqueous layer was then isolated and washed with 1 vol of chloroform: isoamyl alcohol (24:1). Samples were then centrifuged under the same conditions, and the top aqueous layer was isolated. 1/10 vol 3M NaOAc, 30 μg glycogen and 2.5 volumes ice cold 100% EtOH were added to each sample, followed by incubation for at least 30 min at −80°C. After incubation, DNA was precipitated by centrifugation at max. speed for 30 min at 4°C, supernatant was discarded and the pellet was washed with 500 µl 70% ice-cold EtOH. Samples were then centrifuged at max. speed for 30 min at 4°C, before the supernatant was discarded and the DNA pellet was air dried before re-suspension with ultra-pure water.

Quantitative PCR

Request a detailed protocol

For relative quantification of the DNA isolated from all of the DNA-binding assays performed, qPCR was performed using SsoAdvanced Universal SYBER Green Supermix (BIO-RAD, 172–5274) according to the manufacturer's instructions.

Primers used for DR-GFP plasmid amplification:

  • Forward: 5'-TCTTCTTCAAGGACGACGACGGCAACT-3'

  • Reverse: 5'-TTGTAGTTGTACTCCAGCTTGTGC-3'

Exonuclease assay

Request a detailed protocol

DR-GFP plasmid was cut with restriction enzymes generating linear DNA with blunt (EcoRV) or overhanging ssDNA (SalI or KpnI). DNA cleavage was confirmed by agarose gel electrophoresis. 10 µg of the restricted DNA was incubated with BSA, NBS1, MRE11 or SIRT6 purified proteins in NEB exonuclease buffer for 0 to 20 min. ExoI was then added to the samples. Samples of each reaction were taken at 0, 10 and 20 min. DNA was purified using a Qiagen PCR purification kit. The purified DNA was run on 0.8% agarose gel, and the amount of DNA was assessed by image analysis using ImageJ 1.52i software and normalized to the amount of the DNA at the 0′ time point.

Fluor de lys (FDL) activity assay

Request a detailed protocol

Fluor de lys assay with SIRT6-point mutant-MBP proteins was performed as previously described by Gertman et al. (2018).

NAD+ consumption assay

Request a detailed protocol

Purified SIRT6-Flag was incubated at 37 °C for 3 hr with either PstI digested pDR-GFP (DSB), ssDNA or a H3K56 acetylated peptide with 2.5 mM NAD+ and HEPES buffer (50 mM HEPES [pH 7.5], 100 mM KCl, 20 mM MgCl2, and 10% glycerol). After incubation, samples were supplemented with 1 μM 1,3-propanediol dehydrogenase (1,3-PD) and 170 mM 1,3-propanediol for an additional 3 hr incubation. NAD+ consumption by SIRT6 was assessed by NADH levels produced by 1,3-PDase activity, by measuring its absorption at 340 nm. To monitor spontaneous NAD+ consumption in the presence of PstI digested pDR-GFP, ssDNA or H3K56 acetylated peptide, the assay was conducted without SIRT6, and each treatment was normalized to its control.

EMSA

Request a detailed protocol

SIRT6 in storage buffer (20 mM Tris-HCl [pH 7.4], 150 mM NaCl and 50% glycerol) was equilibrated with DNA (or RNA) for 20 min on ice. The buffer composition of EMSA was optimized to obtain the maximum resolution for resolving DNA/RNA. Reactions (final volume 10 μL) were resolved by electrophoresis at 4°C through native gel containing 5% (for blunt-end and sticky-end DNA), 8% (for ssDNA) and 10% (for RNA) polyacrylamide (29:1 acrylamide: bisacrylamide) in 1X TBE buffer. Autoradiographs of the dried gels were analyzed by densitometry using Fujifilm PhosphorImager. The signal was quantified by ImageQuant TL.

GraphPad Prism 7 was used to estimate apparent Kd value for ssDNA (one site, specific binding fit, y = Bmax[SIRT6]/(Kd + [SIRT6]) and for blunt-end and sticky-end DNA (specific binding with Hill slope, y = Bmax[SIRT6]h/(Kdh + [SIRT6]h).

SAXS

Request a detailed protocol

SAXS data were collected at BioSAXS beamline BM29 (ESRF, Grenoble, France), possessing a Pilatus 1M detector. The scattering intensity was recorded in the interval 0.0035 < q < 0.49 Å−1. The measurements were performed at 20°C. SIRT6 (alone or in the presence of dsDNA) was measured at a concentration of 0.5 mg/ml, as it tends to aggregate at higher concentrations. The scattering of the buffer was also measured and subtracted from the scattering of the samples by using Primus (Konarev and Svergun, 2018).

Konarev and Svergun (2018) PyMOL (https://pymol.org/) was used to extract the structures of the SIRT6 dimer and tetramer from the available crystal structure (PDB ID code: 3pki). CRYSOL (Svergun et al., 1995) was then used to compute the artificial SAXS spectra of each protein species. These spectra served as a reference for the reconstitution of experimental SAXS data.

Values for the radius of gyration (Rg) and the maximum particle dimension (Dmax) were derived from distance distribution function P(r), using in-house script (Akabayov et al., 2010). This script was designed to perform an automatic search for the best fitting parameters in GNOM (Svergun, 1992). In the end, DAMMIN (Svergun, 1999) was used to reconstruct the molecular envelope on the basis of the best GNOM fit (obtained from the script analysis and refined manually). E models were calculated and averaged using DAMMAVER (Volkov and Svergun, 2003).

SEC-MALS

Request a detailed protocol

A miniDAWN TREOS multi-angle light scattering detector, with three detector angles (43.6°, 90° and 136.4°) and a 658.9 nm laser beam (Wyatt Technology, Santa Barbara, CA), with a Wyatt QELS dynamic light scattering module for determination of hydrodynamic radius and an Optilab T-rEX refractometer (Wyatt Technology), were used in-line with a size exclusion chromatography analytical column, Superdex 200 Increase 10/300 GL (GE, Life Science, Marlborough, MA) equilibrated in buffer (50 mM tris, 150 mM NaCl and 4 mM MgCl2 [pH 8.0]).

Experiments were performed using an AKTA explorer system with a UV-900 detector (GE), at 0.8 ml/min. All experiments were performed at RT (25°C).

Data collection and mass calculation by SEC-MALS analysis were performed with ASTRA 6.1 software (Wyatt Technology). The refractive index of the solvent was defined as 1.331 and the viscosity was defined as 0.8945 cP (common parameters for PBS buffer at 658.9 nm). dn/dc (refractive index increment) value for all samples was defined as 0.185 mL/g (a standard value for proteins). For the SIRT6 experiment, 150 ul 4.5 mg/ml human-SIRT6-His was injected. For SIRT6+DNA, 200 μl human-SIRT6-His + 50 ul DNA was injected after 1 hr incubation at 37°C.

Statistical analysis

Request a detailed protocol

Statistical analysis was done using GraphPad Prism 7. Analysis included either one-way or two-way ANOVA followed by a post-hoc Dunnet test or a Tukey test, respectively. Significance was set at p<0.05.

For all DNA binding assay results, statistical analysis was preceded by logarithmic transformation to overcome large variance between the different experiments. Statistical analysis was performed on the transformed data as described.

Data availability

All the data generated or analyzed during this study are included in the manuscript and supporting files.

References

Decision letter

  1. Katrin Chua
    Reviewing Editor; Stanford University, United States
  2. Jessica K Tyler
    Senior Editor; Weill Cornell Medicine, United States

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This study by Onn and colleagues uncovers a new role of the sirtuin enzyme SIRT6 as a DNA damage sensor, which can directly bind to DNA double-strand breaks (DSBs) and is critical for initiating cellular DNA damage responses. The findings support the model that SIRT6 functions very early after DNA damage and precedes commitment to particular DNA repair pathways. The study also identifies a tunnel-like structure within the SIRT6 protein, which may be responsible for recognizing broken DNA ends. SIRT6 is of great interest in aging biology; overexpression of SIRT6 in mice can extend lifespan, and SIRT6-mediated modifications of histones and other proteins regulate key signaling pathways relevant for cellular and organismic homeostasis. SIRT6 has previously been implicated in promoting DNA repair via multiple mechanisms, including recruiting DNA repair factors to sites of DNA damage and interacting with specific chromatin factors. The new conclusions in this study add to this area and improve the granularity of how we understand SIRT6 function in DNA repair processes. The work provides important novel contributions to the molecular understanding of SIRT6 function and the study of DNA damage repair.

Decision letter after peer review:

[Editors’ note: the authors submitted for reconsideration following the decision after peer review. What follows is the decision letter after the first round of review.]

Thank you for submitting your work entitled "SIRT6 is a DNA Double-Strand Break Sensor" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by a Senior Editor. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife at this time. However, if you are able to address in full the reviewers' concerns in a substantially revised manuscript, we do encourage you to consider resubmitting to eLife.

The consensus of the reviewers is that the study proposes interesting points that would in principle be of sufficient novelty to merit publication in eLife, including the very early role of SIRT6 in the DDR and the direct binding of SIRT6 to broken DNA. However, there were major technical concerns, detailed in the accompanying reviews, which were assessed as not possible to address within the 2 month revision window of eLife. These include, but are not limited to: (1) the lack of essential controls demonstrating full inactivation of ATM or RNF8; (2) apparently conflicting data and insufficient methodological detail in SIRT6 localization analyses; (3) unconvincing data failing to fully support conclusions from tunnel mutations studies and inhibition of SIRT6 catalytic activity. In addition, the manuscript would benefit from considerable editing, as there are numerous areas with grammatical or other linguistic errors. Thus, the reviewers' concerns are substantial. However, if the authors can fully address all of these concerns with new experimental work, we would be open to consider the work in a new submission.

Reviewer #1:

This manuscript by Onn and colleagues presents evidence that the sirtuin SIRT6 is able to directly bind to DNA to initiate double-strand break (DSB) responses. SIRT6 is of great interest in aging biology; systemic overexpression of SIRT6 extends mouse lifespan, and it has been reported to regulate several key signaling pathways relevant for cellular and organismic homeostasis. SIRT6 has been implicated in promoting DNA repair via multiple mechanisms: e.g. activating PARP, promoting DNA-PK localization to sites of DNA damage, interaction with specific chromatin factors, etc. The conclusions of the authors, if fully substantiated, would add significantly to this area, and represent important novel contributions to sirtuin biology and the study of DNA damage repair. However, there are important areas where additional experimentation is needed to fully substantiate the conclusions. Specifically:

1) For Olaparib studies, the authors need to show that PAR levels are diminished under their treatment conditions. They also need to document better the reduction in ATM levels in response to ATM siRNA (Figure 1—figure supplement 1B is not convincing in this regard).

2) Some of the SIRT6 localization experiments (Figure 2C, top middle panel) are not entirely convincing. In general, these experiments are quite confusingly presented and hard to follow, as the authors keep switching colors (SIRT6 is sometimes green, and sometimes red), and not every panel is labeled.

3) In Figure 5E, why does LacRSIRT6 sometimes localize to a single point in the cell (top left), and sometimes show mostly diffuse localization (top right).

4) If the authors want to substantiate the interaction of SIRT6 with Mre11 and NBS (Figure 5E), they need to assess interaction of the endogenous proteins, not just overexpressed tagged transgenes.

5) It would be helpful for the authors to show in detail how they believe DNA interacts with the tunnel on SIRT6 in their model. In particular, two of the apparently key SIRT6 residues are aspartates, which are negatively charged and would repel the negatively charged backbone on DNA.

6) I don't think the NAM experiment (S5A-B) is sufficient to rule out catalytic SIRT6 roles, particularly since the authors don't actually directly test SIRT6 activity in response to NAM treatment. A much cleaner system would be to reconstitute SIRT6 KO cells with WT or appropriate site-directed mutants of SIRT6.

7) Likewise, the manuscript would be much stronger if the authors showed that some of their DNA-binding mutants maintained a "canonical" function of SIRT6 in vivo (e.g. acH3K9 or acH3K56 acetylation), but failed to rescue DNA damage resistance of SIRT6 KO cells in vivo.

Reviewer #2:

In their manuscript entitled "SIRT6 is a DNA Double-Strand Break Sensor," Onn and colleagues propose that SIRT6 plays a previously uncharacterized role as a DNA damage sensor, which is critical for initiating the DNA damage response (DDR). They further report that other sirtuins share DSB binding capacity and DDR activation. Overall, the authors suggest that SIRT6 functions very proximal in the DDR, prior to repair pathway selection. Their findings are potentially relevant and support previous reports on the involvement of SIRT6 in the DDR. However, we felt that some of their findings, as outlined below, were not sufficiently supported by their data or that more information needs to be included with their current report before it can be recommended for publication.

Substantive concerns

In Figure 1, formation of large SIRT6-GFP foci in response to laser-mediated UV irradiation is shown. No information on timing is given. Moreover, it is not directly shown that these Sirt6 foci correspond to DSB foci (e.g., via gH2AX or 53BP1 co-staining). It is not discussed what fraction of cells actually show accumulation of SIRT6 in foci post UV irradiation. Under these laser irradiation conditions, how many DSB foci per cell are generated?

A major conclusion of this manuscript is that SIRT6 DSB recruitment is independent of DDR factors. These conclusions are based on knockdown or chemical inhibition of factors. The ATM depletion data shown in Figure 1—figure supplement 1B is not compelling. Additional experiments showing inhibition of ATM function are necessary to confirm DDR-independent recruitment of SIRT6 to DSBs. Similarly, it should be confirmed that RNF8-dependent processes are indeed inactivated under the experimental conditions of RNF8 knockdown. It should further be stated how many cells were analyzed per experiment shown in Figure 1A and B.

Reviewer #3:

This study proposes an intriguing model that SIRT6 directly binds DNA breaks and acts at the very top of the DDR cascade. In principle, aspects of this model would improve the granularity of how we understand SIRT6 function in DNA repair processes. However, previous studies, including from the corresponding author, already proposed a very early function of SIRT6 in DDR, and the present manuscript falls short of providing more than an incremental advance. There are also considerable concerns that some important controls are lacking, and some data do not strongly support the conclusions and/or are overinterpreted. Thus, overall, the work is not appropriate for eLife.

1) A key conclusion of the study is that SIRT6 is recruited to DNA damage sites independent of known DDR factors ATM, H2AX, RNF8, and Parp. In the case of Parp, the data are clear; and MacromKate2 is an appropriate positive control for Parp inhibition. By contrast, in the cases of ATM, H2AX, and RNF8, it is not clear that these factors are functionally inhibited. The immunoblots (Figure 1—figure supplement 1) are not convincing that ATM or RNF8 are adequately depleted, and some functional demonstration that signaling by these factors is abrogated is necessary to conclude that SIRT6 recruitment to DSBs is indeed independent of the factors. Regarding Parp, previous work has shown that SIRT6 is upstream of Parp (Mao et al., 2011), so the finding is not surprising.

2) In Figure 2 and Figure 2—figure supplement 1, the authors claim that in tethering assays, SIRT6 colocalizes with multiple DDR factors (SNF2H, MRE11, Ku80, NBS1) but not ATM. It is very difficult to conclude this from the data shown. It is not at all clear how the co-localization was scored, since the SIRT6 signal is largely diffuse, and there certainly does appear to be yellow (signal overlap) in the ATM-SIRT6 image. More importantly, the tethering assay may be overinterpreted if used to make conclusions about whether one factor is upstream of another; instead, it may simply be a readout for protein interactions. Indeed, in Figure 5, reciprocal assays tethering SIRT6 are used to claim that SIRT6 initiates signaling of downstream factors. By the same logic, then the data in Figure 2 would indicate that SIRT6 is recruited by MRE11/NBS1/Ku80 etc. Such interpretations are problematic.

3) The notion that ssDNA binds SIRT6 in a tunnel-like structure is intriguing. However, it is surprising that mutation of key residues have relatively minor effects on DNA binding in vitro (Figure 4E). Moreover, mutation in the N-terminus, which in Figure 4D is shown not to be necessary for binding, has just as much effect as the putative tunnel forming residues.

4) In Figure 5 and Figure 5—figure supplement 1, it is troubling that the diffuse signal of LacR-SIRT6 varies greatly in different panels. In some the entire nucleus has strong signal, whereas in others, the nucleus has virtually no signal except for the lacR-recruited dot. What is the basis for this, and it would seem to make the data problematic.

[Editors’ note: further revisions were suggested after the authors resubmitted, as described below.]

Thank you for submitting your article "SIRT6 is a DNA double-strand break sensor" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Jessica Tyler as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

This manuscript by Onn and colleagues proposes that the mammalian sirtuin protein SIRT6 directly binds to damaged DNA to initiate DNA damage responses. These findings, if fully substantiated, would add to the molecular understanding of SIRT6 function, and make important contributions to sirtuin biology and the study of DNA damage repair.

At present, however, the manuscript falls short of providing a sufficiently clear picture of the mechanism(s) and physiological relevance of SIRT6 as a DSB sensor. It is viewed that several major issues must be addressed in order for the conclusions to be rigorously convincing. There is concern over the heavy reliance on the LacO-based "artificial" system to follow the role of SIRT6 as a DNA break sensor and the rigor of the experimental interpretations generally. There is also a tendency in the manuscript to over interpret/ overstate conclusions, and a careful editing will be required to avoid this. We have the following requested revisions.

Essential revisions:

1) The authors use the PARP inhibitor Olaparib to show that SIRT6 binds DSBs independent of PARP1. However, these experiments do not address if PARP protein (independent of activity) might recruit SIRT6; to address this, the authors need to deplete PARP1 by RNAi or CRISPR knockout, and test if this affects SIRT6. Further, the authors state that SIRT6 binding to DSBs is independent of other proteins. Yet, at least in vivo this could be mediated by PARP1 or other DNA binding proteins that were not examined (potentially WRN).

2) The authors base their model heavily on the LacO tethering system, but there remains major concern that results using this system are insufficient to justify their conclusions regarding DNA damage signaling and protein interactions of SIRT6. For example, to make any statements about SIRT6 interacting with other DDR factors, the LacO system is insufficient, as acknowledged by the authors. The authors failed to perform co-IP experiments, which is a concern. Did they try to IP from both directions? Also after damage? If not, care must be taken not to overstate their conclusions from the tethering assay.

3) The authors have not adequately addressed the prior concerns over the scoring of "co-localization" in the tethering assays. As pointed out before, in Figure 2B, for example, the graph shows ~55% "recruitment" of SIRT6 by SNF2H-LacR but only background levels by ATM-LacR. However, the primary data in 2C don't reflect this at all. Indeed, there seems to be clear overlap (yellow dot) in the ATM samples, at least as clear if not clearer than for the SNF2H tethering. It's not at all obvious how such data could lead to the graph, and suggests the scoring is highly subjective. The authors' response merely describes the scoring process, but does not deal with this discrepancy.

4) There remains concern over the basis on which the authors draw conclusions about what factor recruits what, and this affects the main conclusion that SIRT6 is initiating at the top of the DSB response. In the response, they say their data "suggest that SIRT6 arrives independently and can recruit these factors, or it can be recruited by them when they initiate the signal." This suggests, as was raised in previous review, that the tethering assays that are employed are insufficient to draw conclusions about which factor is upstream of another. More important, the point by the authors undercuts the strength of their model that SIRT6 is at the very top.

5) Figure 3A-C. the authors use EMSA to calculate the Kd value and cooperativity for SIRT6-DNA binding. This method is semi quantitative and therefore additional quantitative methods should be used such as Surface Plasmon Resonance (SPR) and Biolayer interferometry (BLI). Moreover, given the fast interaction of SIRT6 with the damaged site (5 seconds), a more reliable Kd value based on the methods listed above can provide additional information for binding kinetics (kon and Koff) which is needed in order to justify their claims.

6) Several SIRT6 ChIP seq experiments were published, which involves sonicating DNA and the creation of DNA breaks. If SIRT6 binds to ssDNA as the authors propose, it is surprising that ChIP-seq studies were able to identify clear peaks in vivo. This suggests that the ssDNA binding may not occur physiologically in vivo.

7) For reasons outlined in the rebuttal letter, prior data in Figure 1 is no longer included. The new data are less compelling. Overall effects are very moderate in size. Also, in Figure 1B, it is not clear what n = 10 refers to. Is it cells or perhaps independent experiments? Error bars are not included, and p-values are not shown for Figure 1 data. This is of concern and needs to be addressed. We noted that p-values are stated for other data (but not Figure 1 data) in the Transparent Reporting Form.

8) The authors state in their rebuttal that the U2OS cells used were "very unstable" and that "too many gH2AX foci were present making the experiments impossible to quantify or rely on. When cells were too old or damaged, we thawed new ones from old passages and began the whole experiment again". This seems quite arbitrary and made us concerned about the overall validity and rigor of the studies. At a minimum, more specific criteria for what constitutes cells being "too old or damaged" would need to be clearly described in the manuscript.

9) The paper needs much better editing. The authors indicate that the revised manuscript was sent for "editorial revision" but similar issues as pointed out in the prior reviews remain.

Aside from typos, there are numerous mistakes in the reference list. For example, Tang is mentioned 3 times in a row, Kaidi et al. was retracted, etc.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "SIRT6 is a DNA double-strand break sensor" for further consideration by eLife. Your revised article has been evaluated by Jessica Tyler as the Senior Editor, and a Reviewing Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:

1) The revised manuscript now describes more accurately that Parp activity (versus Parp1 protein) is dispensable for SIRT6 binding to DSBs; however, it is problematic that the new submission still fails to show that Parp1 protein does not recruit SIRT6 to breaks. Although knocking-down or CRISPR KO of Parp1 may require trouble shooting for technical reasons, it remains concerning that this essential experiment has not been done, because this limits the conclusions that can then be drawn. The authors' state that Mao et al. previously showed that Parp1 KO MEFs have no defects in SIRT6 recruitment to DSBs; however, this appears incorrect. In fact, Mao et al. demonstrated that SIRT6 and Parp1 interact physically with each other. This makes it a distinct possibility that the SIRT6-Parp1 interaction could in fact mediate recruitment of SIRT6 to DSBs, as previously raised by the reviews (though this is not the focus of the Mao paper). Because this possibility has not been ruled out, the authors cannot rigorously draw the conclusion that SIRT6 binds DSBs "independent of known DSB sensors". For publication in eLife, the authors need to be very careful to limit their conclusions to what their data actually show. For example, they need to rephrase their conclusions (throughout the text), to "SIRT6 recruitment is independent of the MRE11 and Ku DSB sensors and of Parp1 enzymatic activity," or similar statement. If they wish to propose the model that SIRT6 is independent of all known sensors, they will need to state explicitly that this is a speculation, and that their data cannot rule out that recruitment of SIRT6 to DSBs might occur at least in part via its previously described interaction with Parp1.

2) There is also remaining concern that conclusions continue to be overstated in places. In the last sentence of the Discussion ("In conclusion, we have demonstrated that Sirtuins – and mainly SIRT6 – have a role as independent DNA damage sensors."), the conclusion regarding all Sirtuins is not supported by the data. The authors should limit their conclusion to SIRT6. There is not sufficient evidence from their experiments that the other Sirtuins have roles as sensors.

https://doi.org/10.7554/eLife.51636.sa1

Author response

[Editors’ note: the authors resubmitted a revised version of the paper for consideration. What follows is the authors’ response to the first round of review.]

Reviewer #1:

[…] 1) For Olaparib studies, the authors need to show that PAR levels are diminished under their treatment conditions. They also need to document better the reduction in ATM levels in response to ATM siRNA (Figure 1—figure supplement 1B is not convincing in this regard).

MacroH2A is commonly used as a control for Olaparib inhibition (Sherry et al., 2017), as reviewer #3 noted. Given that we cannot perform immunofluorescence at such early time points (these experiments are in live imaging), we show lack of Parylation in live imaging through the lack of MacroH2A recruitment.

Regarding the shATM blot, we replaced the previous blot with a new one with better visible difference in ATM levels. It was achieved by using a different and better antibody (Figure 1—figure supplement 1H).

Author response image 1
Download asset Open asset
Validation on defective repair cell-lines.

(A) Foci quantification of ɤH2AX (in shATM and shH2AX) or 53BP1 (in shRNF8) for DDR impairment assessment in shHela cells from Mostoslavsky’s lab. Not included in the manuscript. (B) Blot of nuclear protein extraction of Hela cells knocked-down for ATM. (C) ɤH2AX foci quantification in Control (n[-IR]=283, n[+IR]=301), shH2AX (n[-IR]=284, n[+IR]=322) and shATM (n[-IR]=299, n[+IR]=385) Hela cells. The graph presents the percentage of cells that present either 0-3, 4-10 or above 10 foci per nucleus, with or without exposure to irradiation. The graph depicts an average of 6 experiments. (D) Samples of the IF of ɤH2AX in sh- Hela cells upon irradiation that were used in the quantification.

To further prove the independence of SIRT6 on other DDR factors, we silenced MRE11 and Ku80 which are the two main DSB sensors (in addition to PARP proteins). As expected, we observed defective recruitment of the main proteins in their complexes- NBS1 (for MRE11) and Ku70 (for Ku80). However, the silencing of neither MRE11 nor Ku80 affected SIRT6 recruitment. (Figure 1C-D, Figure 1—figure supplement 1C-F).

Interestingly, when we tested the effect of SIRT6-KO on the recruitment of MRE11 and Ku80, we found that while the recruitment of MRE11 was defective, Ku80 was not affected by the lack of SIRT6, suggesting a more prominent role of SIRT6 in HR. (Figure 1—figure supplement 1G, 1E-H).

2) Some of the SIRT6 localization experiments (Figure 2C, top middle panel) are not entirely convincing. In general, these experiments are quite confusingly presented and hard to follow, as the authors keep switching colors (SIRT6 is sometimes green, and sometimes red), and not every panel is labeled.

We have improved this issue by adding a label to each image so as to avoid confusions. Since in this work we used a large verity of plasmids and proteins, we were not always able to obtain every plasmid in every color. To overcome this issue, we obtained both SIRT6-GFP and SIRT6-RFP and chose which one to use according to the color of the other protein in each experiment. For the same reason, we also generated both SIRT6-LacR-GFP and SIRT6-LacR-RFP, as well as GFP-LacR and Cherry-LacR.

It is important to note that these experiments were manually quantified by two different students (several of the experiments were quantified by a second student in a blind manner to assure a proper quantification process).

Since the amount of noise in the assay could be quite large, we set a number of rules on what to quantify and how to do it:

Quantified cells were chosen to obtain the following characteristics:

1) Cells must be co-transfected (meaning, they express both red and green colors to avoid lack of co-localization due to the lack of one of the proteins from the cells)

2) Strong signaling and large Lac-Protein dot (that was clearly not background, or the protein was not localized at the LacO repeats).

3) No cross channel fluorescence (we observed minimal to zero crossover and set a threshold to ignore this low false signal)

4) U20S were very unstable and several times too many ϫH2AX foci were present making the experiments impossible to quantify or rely on. When cells were too old or damaged, we thawed new ones from old passages and began the whole experiment again.

3) In Figure 5E, why does LacRSIRT6 sometimes localize to a single point in the cell (top left), and sometimes show mostly diffuse localization (top right).

SIRT6 is a nuclear/chromatin bound protein spread in the nucleus. We usually see the entire nucleus in green as in the sample of the diffused localization. In some pictures, the LacR dot is very bright and clear, and little background is observed. In those cases, the pictures were taken with lower exposure times, and there appears to be no SIRT6 in the nucleus, though in the background there was. In some pictures with lower expression or higher background, we used a longer time of exposure. The causes of these differences depend on: the levels of expression, the time after transfection and the position in the well (well borders tend to have higher background intensity). This does not represent a difference in SIRT6 behavior (as some SIRT6-GFP can be seen in the second top left corner (new Figure 6B).

Importantly, all of the pictures were taken before saturation and avoiding channels cross-talk since, as explained before, these experiments were quantified in a qualitative manner and the fluorescence intensity was not relevant for the co-localization. We allowed ourselves to take different exposure times following the rules detailed before.

It is also important to note that SIRT6 does not form foci and even after irradiation SIRT6 still appeared diffused, covering the entire nucleus. To prove it localizes to the sites of damage, we previously used other methods such as laser-induced damage in live imaging, as shown in this manuscript. But we also showed it by laser induced damage and IF, and ChIP on sites cleaved by ISceI (tested before in (Toiber et al., 2013)). Thus, an apparently large SIRT6 foci in this system is due to the accumulation of a large number of LacR-SIRT6 proteins in the LacO site 256 repeats.

4) If the authors want to substantiate the interaction of SIRT6 with Mre11 and NBS (Figure 5E), they need to assess interaction of the endogenous proteins, not just overexpressed tagged transgenes.

After several attempts, we failed to Co-IP endogenous SIRT6 with endogenous MRE11 and NBS1 due to excessive background with various antibodies, therefore, we removed the Flag-IP. However, we also took a different approach to distinguish between interaction, and recruitment dependent on signaling. We performed the tethering assay while inhibiting DDR signaling using wortmannin- an ATM, ATR and DNA-PKc inhibitor. This treatment led to a reduction in H2AX phosphorylation and in co-localization of SIRT6-LacR with both 53BP1 and BRCA1. In this experiment, we did not observe the same reduction in co-localization with MRE11 and Ku80. The most reasonable explanation for these results is that the recruitment of these factors is based on direct protein-protein interactions or complex formation, and not on DDR signaling. However, we cannot confirm this is a protein-protein interaction, therefore, we removed the IP and the paragraph from this work, and added only the wortmannin experiment. (Figure 6—figure supplement 3A-E).

5) It would be helpful for the authors to show in detail how they believe DNA interacts with the tunnel on SIRT6 in their model. In particular, two of the apparently key SIRT6 residues are aspartates, which are negatively charged and would repel the negatively charged backbone on DNA.

We propose the following spatial model of SIRT6. Amino acids predicted to bind to DNA are colored in red. The tunnel is located at the bottom part, with a single DNA strand drawn inside (Figure 4E).

As for the negatively charged amino acids, aspartate could be involved in binding the bases and backbone (binding of Asp to Cys). Usually, Asp is part of a sequence that binds DNA through the divalent metal ion (Mg (II)), and mediates the binding to the phosphate backbone of nucleic acids (Akabayov et al., 2011). At this point, and without a crystal structure, it is impossible to fully determine how the binding is performed. We do believe, however, that the binding is influenced by hydrophobic interactions. We would like to point out that our structural model should be regarded as a hypothetical model, which we have tried to make clear in the text.

Author response image 2
Download asset Open asset
Aspartate basepairing with DNA bases and backbone.

6) I don't think the NAM experiment (S5A-B) is sufficient to rule out catalytic SIRT6 roles, particularly since the authors don't actually directly test SIRT6 activity in response to NAM treatment. A much cleaner system would be to reconstitute SIRT6 KO cells with WT or appropriate site-directed mutants of SIRT6.

The U20S cellular system developed by Roger Greenberg’s lab is resistant to various antibiotics, thus the knockout of SIRT6 in these cells failed. However, taking into account our findings, we believe that this action is not necessary. To prove that SIRT6 catalytic activity is not important for DDR initiation, we used the catalytic mutant SIRT6 H133Y, which is known to be catalytically dead. This was confirmed by a fluor de lys assay performed in this work as well (Figure 4—figure supplement 1C).

To overrule any catalytic activity performed by possible dimerization of SIRT6- HY-LacR with the endogenous SIRT6 present in the cells, we inhibited its activity using Nicotinamide (NAM).

We have now added the control showing an increase in the acetylation of the SIRT6 target H3K56ac, indicating that NAM treatment for 12 and 24 hours inactivates Sirtuins activity. Again, no change was found in SIRT6-HY-LacR ability to initiate the DDR, suggesting that this ability is not dependent on SIRT6 catalytic activity (Figure 5—figure supplement 1A-B). The previous experiment was replaced by this new figure.

7) Likewise, the manuscript would be much stronger if the authors showed that some of their DNA-binding mutants maintained a "canonical" function of SIRT6 in vivo (e.g. acH3K9 or acH3K56 acetylation), but failed to rescue DNA damage resistance of SIRT6 KO cells in vivo.

It would be most interesting to find separate functions, however, most of the mutants affect the catalytic activity of SIRT6 (Figure 4—figure supplement 1C) as well as its binding activity. We think that SIRT6 binding domain to DNA is very close to the binding of NAD, since both NAD and ssDNA have a very similar chemical composition. We hypothesize that this is the reason why most of the mutations in the tunnels also affect the catalytic activity of SIRT6. However, we do observe two exceptions: A13W and D63Y. On the one hand, D63Y mutation had no effect on DNA binding and it causes a significant reduction in SIRT6 catalytic activity. A13W mutation, on the other hand, resulted in an increase in catalytic activity along with a reduction in DNA binding. We raise this point in the discussion of the paper as well. However, with the lack of a definitive proof (such as crystal structure) we believe that making any unequivocal conclusions on whether these two abilities are related to one another or not is out of the scope of this paper.

Reviewer #2:

Substantive concerns

In Figure 1, formation of large SIRT6-GFP foci in response to laser-mediated UV irradiation is shown. No information on timing is given. Moreover, it is not directly shown that these Sirt6 foci correspond to DSB foci (e.g., via gH2AX or 53BP1 co-staining). It is not discussed what fraction of cells actually show accumulation of SIRT6 in foci post UV irradiation. Under these laser irradiation conditions, how many DSB foci per cell are generated?

In these experiments, laser induced damage is used in a live imaging technique which generates DNA damage at a single (see scheme) relatively large dot on cells pre-sensitized with BrdU or Hoechst, thus generating double strand breaks. Cells are transfected with a florescent protein plasmid, and the recruitment of this protein to sites of damage is measured in vivo. Images were taken at 3-5 second intervals for 60 seconds after photobleacing. Before bleaching, 3 images were taken for each cell to measure the florescence baseline, see Author response image 3.

Author response image 3
Download asset Open asset
Scheme of Laser Induced damage (LID) experiment.

In this method we do not generate DNA damage foci. Therefore, there are no quantifications of the percentage of cells that show foci, but we show an analysis on a single cell that was measured over time. The experiment was repeated in several cells (note the N = in each experiment) and the results are presented as the average of the measurements.

The method was explained in (Toiber et al., 2013) and in here for clarification. It was adapted to microscope use for the following experiments:

“Fluorescence Recovery after Photobleaching (FRAP)

FRAP experiments (Laser induced damage) were performed as previously described at Toiber et al., 2013. […] Images analysis was performed using ImageJ 1.52i software.”

In Figure 1A-B and Figure 1—figure supplement 1A-B, U2OS Cells were incubated with 10µM BrdU or 10µM BrdU + 10µM Olaparib over-night, before going through the same procedure.

A major conclusion of this manuscript is that SIRT6 DSB recruitment is independent of DDR factors. These conclusions are based on knockdown or chemical inhibition of factors. The ATM depletion data shown in Figure 1—figure supplement 1B is not compelling. Additional experiments showing inhibition of ATM function are necessary to confirm DDR-independent recruitment of SIRT6 to DSBs. Similarly, it should be confirmed that RNF8-dependent processes are indeed inactivated under the experimental conditions of RNF8 knockdown.

See response to reviewer #1.

It should further be stated how many cells were analyzed per experiment shown in Figure 1A and B.

This information was added to the manuscriptoiber51636Authorresponseimage2.epsoiber51636Authorresponseimage2.eps

Reviewer #3:

1) A key conclusion of the study is that SIRT6 is recruited to DNA damage sites independent of known DDR factors ATM, H2AX, RNF8, and Parp. In the case of Parp, the data are clear; and MacromKate2 is an appropriate positive control for Parp inhibition. By contrast, in the cases of ATM, H2AX, and RNF8, it is not clear that these factors are functionally inhibited. The immunoblots (Figure 1—figure supplement 1) are not convincing that ATM or RNF8 are adequately depleted, and some functional demonstration that signaling by these factors is abrogated is necessary to conclude that SIRT6 recruitment to DSBs is indeed independent of the factors. Regarding Parp, previous work has shown that SIRT6 is upstream of Parp (Mao et al., 2011), so the finding is not surprising.

Regarding the first concern, please see response to reviewer #1.

Regarding the relationship between SIRT6 and PARP, Mao and associates (Mao et al., 2011) indeed deal with this question in their work; however, they mainly focus on the relationship under oxidative stress induced by H2O2. Moreover, they do not address the question of which protein arrives to sites of damage first. PARP1 is the fastest known enzyme to be recruited to sites of DSBs (Yang et al., 2018), and is responsible, at least in part, for the recruitment of the MRN complex (Haince et al., 2008). Taking that into consideration, along with the fact that SIRT6 was not reported to be a DSB binding protein before, its arrival to sites of damage independently of PARP1, even if not surprising, is a significant finding.

2) In Figure 2 and Figure 2—figure supplement 1, the authors claim that in tethering assays, SIRT6 colocalizes with multiple DDR factors (SNF2H, MRE11, Ku80, NBS1) but not ATM. It is very difficult to conclude this from the data shown. It is not at all clear how the co-localization was scored, since the SIRT6 signal is largely diffuse, and there certainly does appear to be yellow (signal overlap) in the ATM-SIRT6 image. More importantly, the tethering assay may be overinterpreted if used to make conclusions about whether one factor is upstream of another; instead, it may simply be a readout for protein interactions. Indeed, in Figure 5, reciprocal assays tethering SIRT6 are used to claim that SIRT6 initiates signaling of downstream factors. By the same logic, then the data in Figure 2 would indicate that SIRT6 is recruited by MRE11/NBS1/Ku80 etc. Such interpretations are problematic.

Reviewer #3 is correct to point out that in the tethering assay the data can be interpreted as protein-protein interactions (direct interaction: protein-protein; or recruitment: signaling induced co-localization, but not through direct interaction). However, the early arrival of SIRT6, together with the new results of MRE11 and Ku80 silencing, allow us to suggest that SIRT6 arrives independently and can recruit these factors, or it can be recruited by them when they initiate the signal. This would allow SIRT6 to be a sensor, but also to form protein-complexes with other sensors, revealing the importance for both HR and NHEJ.

A better explanation on how co-localization was scored can be found in the response to reviewer #1. However, it is important to note that SIRT6 tends to be evenly dispersed in the nucleus and does not form foci, thus, a visible SIRT6 foci in the tethering assay can only be explained by the accumulation of a large number of SIRT6 molecules at the LacO site. Since there is no actual DNA damage in this site (as shown by GFP-LacR control), we can assume that the accumulation of DDR proteins is initiated by the system.

To better understand which co-localizations occur in response to DDR signaling and which occur due to protein interactions, we performed the tethering assay inhibiting DDR signaling with wortmannin- an ATM, ATR and DNA-PKc inhibitor. While signaling was inhibited (seen by a reduction in co-localization with ϫH2AX), as well as the recruitment of both 53BP1 and BRCA1, the recruitment of MRE11 and Ku80 was unaffected (Figure 6—figure supplement 3A-E). These findings suggest that while there is a direct interaction between SIRT6 and MRE11 or Ku80, the most downstream factors arrive upon signaling.

3) The notion that ssDNA binds SIRT6 in a tunnel-like structure is intriguing. However, it is surprising that mutation of key residues have relatively minor effects on DNA binding in vitro (Figure 4E).

These experiments were carried out with a single point mutant to avoid instability or the complete misfolding of the protein. Since these are just single point mutations, we do not expect that they would completely abolish SIRT6 DNA binding ability, but rather reduce it with an almost 50% reduction for SIRT6-HY mutant.

Moreover, mutation in the N-terminus, which in Figure 4D is shown not to be necessary for binding, has just as much effect as the putative tunnel forming residues.

Figure 4D shows the core domain (not the N terminus), where most of the residues predicted to interact with DNA are present, as well as the ones we mutated (Figure 4A-C, S4A).

4) In Figure 5 and Figure 5—figure supplement 1, it is troubling that the diffuse signal of LacR-SIRT6 varies greatly in different panels. In some the entire nucleus has strong signal, whereas in others, the nucleus has virtually no signal except for the lacR-recruited dot. What is the basis for this, and it would seem to make the data problematic.

See response to reviewer #1.

[Editors’ note: what follows is the authors’ response to the second round of review.]

Essential revisions:

1) The authors use the PARP inhibitor Olaparib to show that SIRT6 binds DSBs independent of PARP1. However, these experiments do not address if PARP protein (independent of activity) might recruit SIRT6; to address this, the authors need to deplete PARP1 by RNAi or CRISPR knockout, and test if this affects SIRT6. Further, the authors state that SIRT6 binding to DSBs is independent of other proteins. Yet, at least in vivo this could be mediated by PARP1 or other DNA binding proteins that were not examined (potentially WRN).

1a) We agree with the reviewer that we cannot rule out direct interaction of PARP1 with SIRT6. To clarify this point, we rephrase the interpretation of our results (PARP activity).

We tried to generate PAPR1-KO cells, but were unsuccessful (see Author response image 4). However, Mao et al1 show no effect in SIRT6 recruitment in MEFs KO for PARP1.

We would like to emphasize, however, that this lack of SIRT6 dependence on PARP1 activity is contrasted by the fact that MRE11 and NBS1 recruitment to DSBs does require PARP1 activity2. On a technical note, Olaparib inhibits all PARPs (PARP1, PARP2 and PARP3), which could not be accomplished by KO of PARP1 alone.

Author response image 4
Download asset Open asset
Western blot of Hela cells after attempt to create PARP1 KO using a CRISPR-Cas9 system.

1b) When addressing SIRT6 recruitment to sites of DSBs and its DNA binding ability, it is important to take into consideration the timing of protein recruitment to sites of damage.

The most relevant proteins to test are the first proteins to arrive to DSB sites. These are the other DSB sensors, which also initiate the DNA damage response. If SIRT6 arrival is independent of these proteins (as indeed verified as shown in figure 1 and Figure 1—figure supplement 1), then it is not likely that SIRT6 would be dependent on proteins that arrive at the site of damage at later time points. It is an interesting question whether WRN and other proteins influence SIRT6 repair efficiency, kinetics and other parameters of repair; however, these are beyond the scope of this paper.

Therefore, we suggest that with the elimination of the known sensors (PAPR inhibition, MRE11 and KU80), SIRT6 arrival should be independent of the activity of the hundreds of other proteins that participate downstream in the DNA damage response.

2) The authors base their model heavily on the LacO tethering system, but there remains major concern that results using this system are insufficient to justify their conclusions regarding DNA damage signaling and protein interactions of SIRT6. For example, to make any statements about SIRT6 interacting with other DDR factors, the LacO system is insufficient, as acknowledged by the authors. The authors failed to perform co-IP experiments, which is a concern. Did they try to IP from both directions? Also after damage? If not, care must be taken not to overstate their conclusions from the tethering assay.

We agree with the reviewer that our data do not resolve the possible contribution of direct and indirect protein interactions to SIRT6 DNA damage activity. We have indeed attempted several versions of the endogenous Co-IP from both directions as well as post irradiation, but these attempts were unsuccessful because of the poor quality of the antibodies.

However, we used a different approach to differentiate between recruitment triggered by protein-protein interactions and signaling-based recruitment. We performed the LacO tethering assays using the ATM/ATR/DNA-PK inhibitor Wortmannin. We show that addition of Wortmannin inhibited DDR signaling (manifested in reduction of ɤH2AX levels) and the recruitment of downstream factors, such as 53BP1 and BRCA1. However, the inhibition by Wortmannin had no effect on the recruitment of MRE11 and Ku80, suggesting that their recruitment occurs independently of the signaling propagated by ATM/ATR/DNA-PK kinases. As far as we know, any protein recruitment that occurs prior to this stage is based on direct interaction or complex formation, supported also by the reciprocal recruitment of SIRT6 and these factors.

We therefore suggest in the manuscript that protein interactions could take part in the SIRT6 signaling/complex formation (Figure 6—figure supplement 3), without definitively claiming that such interactions exist. In any case, we do not think that this diminishes the importance of our findings and conclusions, since the capacity of SIRT6 to initiate recruitment remains relevant regardless of the contribution of protein interactions.

We would like to highlight that one of the most important finding of our work is not whether the interaction is direct or indirect, but rather that the recruitment of SIRT6 is not triggered by the DDR signaling (e.g., via ATM). This, in turn, suggests that SIRT6 does not function downstream to ATM. Signaling itself is not sufficient to recruit SIRT6 in the absence of open-ended DNA. This, together with our biochemical assays, indicates that SIRT6 directly binds open-ended DNA, suggesting that the mechanism is the direct binding of DSBs by SIRT6.

We disagree that we base our model heavily on the LacO system, since many of our findings are based on other several assays (laser induced damage, EMSA, DNA binding assay, SAXS, end protection assay, etc.). Moreover, taking into account previous publications indicating that silencing of SIRT6 completely impaired the recruitment of BRCA1, RPA, 53BP1, SNF2H, H2AUb, and the ability to repair DNA damage itself through both HR and NHEJ, strengthens our point of SIRT6 as a DSB sensor.

3) The authors have not adequately addressed the prior concerns over the scoring of "co-localization" in the tethering assays. As pointed out before, in Figure 2B, for example, the graph shows ~55% "recruitment" of SIRT6 by SNF2H-LacR but only background levels by ATM-LacR. However, the primary data in 2C don't reflect this at all. Indeed, there seems to be clear overlap (yellow dot) in the ATM samples, at least as clear if not clearer than for the SNF2H tethering. It's not at all obvious how such data could lead to the graph, and suggests the scoring is highly subjective. The authors' response merely describes the scoring process, but does not deal with this discrepancy.

The tethering assay was developed in the lab of Tom Misteli and published in Science in 20083. Ever since, it has been widely used by his lab and others to show causality in DNA damage response initiation and protein recruitment4–8.

In order to alleviate the doubts of the reviewers, it is important to note that if two fluorescent proteins (GFP and RFP) co-exist in the nucleus in similar intensities, the cell will most likely present a “yellow dot” every time a merged image is presented. However, this is not considered as a co-localization. In our scoring process that will be exemplified below, co-localization is only positive if in the exact position of the LacR protein there is a clear “dot” in the other channel.

To better explain the method, we have added here a more detailed explanation of the experimental procedure, data collection and the analysis process that is now part of the Materials and methods.

“Tethering assay

U2OS cells containing 256X LacO sequence repeats in their genome were transfected with plasmids of chimeric LacR-DDR enzyme-GFP/Cherry proteins. Cells were either co-transfected with a second plasmid of a fluorescent/ Flag tagged protein or Immuno-stained (See Immunofluorescence) for an endogenic protein.”

Out of the transfected cells, those chosen for the experiment had to adhere to the following conditions:

1) Cells exhibiting a large visible focus of LacR-DDR-GFP/Cherry at LacO sites (located using fluorescent microscopy).

2) Cells must express both proteins of interest, meaning they should exhibit both red and green colors.

3) No cross channel fluorescence (we observed minimal to zero crossover and set a threshold to ignore this low false signal).

After cell selection, co-localization percentage was assessed. Co-localization is defined as the common localization of large foci of the two proteins of interest at the LacO site. Nuclear proteins often seem to be evenly diffused throughout the entire nucleus; thus, unless 2 defined foci have been located, this is not defined as co-localization.

However, to better explain the scoring process and the different variations that could be seen in these experiments, we performed a computational analysis for numerous cells, using ImageJ 1.52p software (see examples in Author response image 5). In the analysis, we:

1) Draw a line crossing the nucleus at the focus formed by the LacR- chimeric protein.

2) Measure the fluorescence intensity at every pixel along the line for both filters.

Author response image 5
Download asset Open asset
Co-localization analysis.

(A) LacR-ATM with SIRT6-GFP: Both are nuclear proteins that are usually found evenly spread across the entire nucleus. While we can see a sharp peak where the ATM-LacR focus is located, we do not see a peak of SIRT6-GFP at the same cellular location, which means there is no co-localization of the two proteins in this image. (B) LacR-SNF2H with SIRT6-RFP: A clear peak in the graph of the SNF2H-LacR is accompanied by a clear peak of the SIRT6-RFP, meaning positive co-localization. Since the red signal is much weaker than the green, the SIRT6 peak is much smaller than the SNF2H one. (C) LacR-Ku80 with SIRT6-RFP: In this case the SIRT6-RFP signal is much stronger that the Ku80-LacR GFP signal; however, we can still observe clear peaks at the same location. (D) LacR-GFP with SIRT6-RFP: In this negative control condition, there are no parallel peaks in the graph. However, when looking at the merged image we see that both cells are”yellow”. This just means that both signals are equally as strong (as can be seen in the graph as well) and that both proteins are spread throughout the entire nucleus; however, this is not a co-localization. (E-F) LacR-SIRT6 with 53BP1: When comparing endogenous 53BP1 (E) to the exogenous (F), we can see that using an endogenous antibody usually gives a “cleaner” higher peak with less background; however, in both cases the co-localization is clear.

Colocalization quantification:

Co-localization was assessed as either positive (1) or negative (0). From this analysis, the percentage of cells that exhibit co-localization (positive cells) was calculated, and defined as “percentage of co-localization between two proteins”. The graphs that appear in the paper present the average of several individual experiments with SEM, while “n” is the number of cells used.

We are aware of a concern recently brought up concerning this method 9, in which it was claimed that the LacO array is susceptible to DNA damage. However, we address this issue by including a negative control of GFP/mCherry-LacR protein in all of our experiments.

Lastly, to avoid any further confusion that might arise from the images selected for Figure 2 of the paper, we replaced the ATM-LacR image with a different one, hopefully making our claim clearer.

After performing hundreds of these experiments over many years we are highly trained at assessing co-localization; thus we feel comfortable about quantifying the experiments manually. Moreover, many of the experiments were quantified blindly by more than one student in order to reduce subjectivity and assure a proper quantification process.

4) There remains concern over the basis on which the authors draw conclusions about what factor recruits what, and this affects the main conclusion that SIRT6 is initiating at the top of the DSB response. In the response, they say their data "suggest that SIRT6 arrives independently and can recruit these factors, or it can be recruited by them when they initiate the signal." This suggests, as was raised in previous review, that the tethering assays that are employed are insufficient to draw conclusions about which factor is upstream of another. More important, the point by the authors undercuts the strength of their model that SIRT6 is at the very top.

There seem to be some misunderstanding about the suggested model. Just like MRE11 and Ku80 arrive to sites of damage independently of each other11,12, we suggest that SIRT6 can arrive independently of them as well. There is no factor at the top of all DSB repair pathways, and we do not state otherwise. We have edited the text accordingly to prevent misunderstandings.

Nonetheless, we believe that we have proven that SIRT6 is a sensor, just as MRE11 and KU80 are. We used several lines of experiments (Laser induced damage, biochemical assays and tethering) in addition to published data that allow us to support our final conclusion.

1) SIRT6 arrives to sites of DSBs in less than 5 seconds10.

2) SIRT6 arrives to sites of DSBs independently of the known sensors- MRE11 and Ku80.

3) Unlike other upstream factors, SIRT6 arrives to sites of DSBs independently of PARP activity.

4) SIRT6 deficiency results in defective recruitment of a DDR initiator- MRE11.

5) SIRT6 is not recruited by DDR signaling alone- unlike most downstream factors, ATM-LacR signaling does not recruit it.

6) SIRT6 binds ssDNA and sticky-ended DNA with no intermediates in-vitro.

7) Like other DDR initiators, SIRT6 can initiate the DNA damage response even in the absence of actual damage (shown by using SIRT6-LacR at the teething assay).

8) SIRT6 can recruit DDR proteins to the LacO array.

9) SIRT6 deficiency results in defective DDR protein recruitment, demonstrated by the defective recruitment of both 53BP1 and BRCA1.

10) SIRT6 deficiency results in defective repair by both HR and NHEJ10.

11) SIRT6 silencing prevents recruitment of BRCA1, 53BP1, RPA, H2Bub12010.

When what we expect of a sensor is to: first, recognize and bind DSBs by itself in a fast manner; second, activate DDR protein signaling and recruitment; and last, affect the actual DNA repair, we believe that – when taking all of these findings into consideration – SIRT6 is a DSB sensor.

5) Figure 3A-C. the authors use EMSA to calculate the Kd value and cooperativity for SIRT6-DNA binding. This method is semi quantitative and therefore additional quantitative methods should be used such as Surface Plasmon Resonance (SPR) and Biolayer interferometry (BLI). Moreover, given the fast interaction of SIRT6 with the damaged site (5 seconds), a more reliable Kd value based on the methods listed above can provide additional information for binding kinetics (kon and Koff) which is needed in order to justify their claims.

Although in theory SPR can provide the kinetic parameters of binding (Kon, Koff), it is not a good method for measuring DNA-protein binding due to its generally high binding affinity and strong electrostatic nature. These features give rise to several practical complications:

1) Mass transfer limits on kinetics, where the rates of transfer of components from the injected solution to the immobilized component are slower than the association reaction.

2) Very slow dissociation rates with potential rebinding during the dissociation phase.

3) Limited time for the association reaction due to volume limitations in the injection syringe.

Biolayer interferometry is a relatively new technique and is not widely used. Therefore, we do not know how suitable it is for DNA-binding interactions.

Gel shift assay is a commonly used and highly acceptable method for the evaluation of nucleic acid-protein interactions. One great advantage of this method is that it is not sensitive for binding, meaning that only tightly bound DNA-protein complexes are observable. This is a robust method for Kd that is around a μM range. In our case, it is quite reliable (more reliable than Plasmon resonance.) We believe that more accurate data does not provide additional insight to our findings; these methods are not simple and are beyond the scope of this paper.

6) Several SIRT6 ChIP seq experiments were published, which involves sonicating DNA and the creation of DNA breaks. If SIRT6 binds to ssDNA as the authors propose, it is surprising that ChIP-seq studies were able to identify clear peaks in vivo. This suggests that the ssDNA binding may not occur physiologically in vivo.

When performing ChIP-seq, cells are fixed with formaldehyde prior to their sonication. Thus, DSB sensors as well as other proteins are fixed to their cellular location and are not recruited to the breaks generated through the sonication process. However, if some of them would still bind, DNA damage would occur in the cell randomly (as would occur by sonication as well), thus we do not expect to see an enriched peak, but possibly more “random noise”.

7) For reasons outlined in the rebuttal letter, prior data in Figure 1 is no longer included. The new data are less compelling. Overall effects are very moderate in size. Also, in Figure 1B, it is not clear what n = 10 refers to. Is it cells or perhaps independent experiments? Error bars are not included, and p-values are not shown for Figure 1 data. This is of concern and needs to be addressed. We noted that p-values are stated for other data (but not Figure 1 data) in the Transparent Reporting Form.

The only experiment that was removed from Figure 1 was the downstream factor RNF8. Moreover, MRE11 and KU80 silencing were added, both more relevant since they are the actual known DSB sensors. We fail to understand how that missing panel (RNF8) makes the data less compelling.

In single cell microscopy experiments, since every cell is an individual experiment, n=total number of cells. To our knowledge, stating p-values in this type of experiment is not customary; however, since it was pointed out as a concern, we increased the number of cells in various experiments, adding error bars to all the figures. To analyze the differences between the different graphs we measured the area under the curve. Our results show that while most measurements are characterized by a single time scale, they might differ in their saturation values. Since this makes it impossible to average over the entire data collection, we used an alternative method of analysis based on calculating the area under the graph representing the recruitment vs. time. Physically, this area represents the integrated quantity of recruited protein. These areas, approximated by summing the column of signals, were then analyzed using Prism to extract their p-values. Please see the new figures with SEM and the area under the curve here (Figure 1 and Figure 1—figure supplement 1A). The relevant information was added to the figure legends and the “Transparent reporting form” as well.

8) The authors state in their rebuttal that the U2OS cells used were "very unstable" and that "too many gH2AX foci were present making the experiments impossible to quantify or rely on. When cells were too old or damaged, we thawed new ones from old passages and began the whole experiment again". This seems quite arbitrary and made us concerned about the overall validity and rigor of the studies. At a minimum, more specific criteria for what constitutes cells being "too old or damaged" would need to be clearly described in the manuscript.

As with many cell lines, several passages cause basal levels of DNA damage, increasing the background foci of ɤH2AX. When using the U2OS cells of the tethering assay, if the majority of cells were too “Old” (a term we used before), meaning the cell presented DNA damage without any treatment (exhibited more than 15 ɤH2AX foci per nucleus), cells were discarded and a new vial from an earlier passage was thawed. See example in Author response image 6:

Author response image 6
Download asset Open asset
IF for ɤH2AX in in SIRT6-LacR-GFP expressing U2OS cells.

“New” vs. “Old” cells.

Additional notes:

To better understand whether Lack of proper binding of the HY mutant would affect its recruitment to DNA, we added the lack of recruitment of SIRT6-HY into the laser induced damage. We used SIRT6-KO cells and measured SIRT6-WT or SIRT6-HY to the laser induced damage by live imaging. This experiment proved that SIRT6-HY which is catalytically inactive and has lost its binding capacity by 50% is unable to arrive to the sites of damage (Figure 5C).

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:

1) The revised manuscript now describes more accurately that Parp activity (versus Parp1 protein) is dispensable for SIRT6 binding to DSBs; however, it is problematic that the new submission still fails to show that Parp1 protein does not recruit SIRT6 to breaks. […] If they wish to propose the model that SIRT6 is independent of all known sensors, they will need to state explicitly that this is a speculation, and that their data cannot rule out that recruitment of SIRT6 to DSBs might occur at least in part via its previously described interaction with Parp1.

1) Although we tried, the CRISPER-KO did not work, and it could take time and troubleshooting, therefore we referred to Mao et al. In the paper (in which this recruitment is not the main focus) they present data which shows that PAPR1-KO in MEFs does not affect SIRT6-GFP recruitment to sites of double strand breaks.

However, as we did not do the experiment in our cell lines, and PARP2 and PARP3 could still have influence we agree that the statement should only refer to PARP activity. We revised the test accordingly:

Previous submission: Results:

"Taken together, these results indicate that SIRT6 arrives to the sites of damage independently of other known DSB sensors, and that in the absence of actual DNA damage, signaling itself is not sufficient to bring SIRT6 to the damage sites."

Revised submission:

"… SIRT6 arrives to the sites of damage independently of MRE11, Ku80 and PARP activity, and that in the absence…"

Previous submission: Discussion:

"In addition, we showed that SIRT6 can arrive at the sites of DSBs independently of the known sensors, and activate the DDR on its own."

Revised submission:

"In addition, we showed that SIRT6 can arrive at the sites of DSBs independently of the known sensors MRE11 and Ku80 as well as PARP activity, and activate the DDR on its own."

2) There is also remaining concern that conclusions continue to be overstated in places. In the last sentence of the Discussion ("In conclusion, we have demonstrated that Sirtuins – and mainly SIRT6 – have a role as independent DNA damage sensors."), the conclusion regarding all Sirtuins is not supported by the data. The authors should limit their conclusion to SIRT6. There is not sufficient evidence from their experiments that the other Sirtuins have roles as sensors.

We agree with this comment and the new phrase in the Discussion states:

"In conclusion, we have demonstrated that SIRT6 has a role as an independent DNA damage sensor"

References

1. Mao, Z. et al. SIRT6 promotes DNA repair under stress by activating PARP1. Science332, 1443–6 (2011).

2. Haince, J.-F. F. et al. PARP1-dependent kinetics of recruitment of MRE11 and NBS1 proteins to multiple DNA damage sites. J. Biol. Chem.283, 1197–208 (2008).

3. Soutoglou, E. and Misteli, T. Activation of the cellular DNA damage response in the absence of DNA lesions. Science320, 1507–10 (2008).

4. Batenburg, N. L. et al. CSB interacts with BRCA1 in late S/G2 to promote MRN- and CtIP-mediated DNA end resection. Nucleic Acids Res.47, 10678–10692 (2019).

5. Roukos, V., Burgess, R. C. and Misteli, T. Generation of cell-based systems to visualize chromosome damage and translocations in living cells. Nat Protoc9, 2476–92 (2014).

6. Helfricht, A. et al. Remodeling and spacing factor 1 (RSF1) deposits centromere proteins at DNA double-strand breaks to promote non-homologous end-joining. Cell Cycle12, 3070–82 (2013).

7. Luijsterburg, M. S. et al. A new non-catalytic role for ubiquitin ligase RNF8 in unfolding higher-order chromatin structure. EMBO J.31, 2511–27 (2012).

8. Zolghadr, K. et al. A fluorescent two-hybrid assay for direct visualization of protein interactions in living cells. Mol. Cell Proteomics7, 2279–87 (2008).

9. Jacome, A. and Fernandez-Capetillo, O. Lac operator repeats generate a traceable fragile site in mammalian cells. EMBO Rep.12, 1032–8 (2011).

10. Toiber, D. et al. SIRT6 recruits SNF2H to DNA break sites, preventing genomic instability through chromatin remodeling. Mol. Cell51, 454–68 (2013).

https://doi.org/10.7554/eLife.51636.sa2

Article and author information

Author details

  1. Lior Onn

    1. Department of Life Sciences, Ben-Gurion University of the Negev, Beer Sheva, Israel
    2. The Zlotowski Center for Neuroscience, Ben-Gurion University of the Negev, Beer-Sheva, Israel
    Contribution
    Conceptualization, Formal analysis, Investigation
    Competing interests
    No competing interests declared

  2. Miguel Portillo

    1. Department of Life Sciences, Ben-Gurion University of the Negev, Beer Sheva, Israel
    2. The Zlotowski Center for Neuroscience, Ben-Gurion University of the Negev, Beer-Sheva, Israel
    Contribution
    Formal analysis, Investigation
    Competing interests
    No competing interests declared

  3. Stefan Ilic

    Department of Chemistry, Ben-Gurion University of the Negev, Beer-Sheva, Israel
    Contribution
    Investigation, Methodology
    Competing interests
    No competing interests declared

  4. Gal Cleitman

    1. Department of Life Sciences, Ben-Gurion University of the Negev, Beer Sheva, Israel
    2. The Zlotowski Center for Neuroscience, Ben-Gurion University of the Negev, Beer-Sheva, Israel
    Contribution
    Investigation
    Competing interests
    No competing interests declared

  5. Daniel Stein

    1. Department of Life Sciences, Ben-Gurion University of the Negev, Beer Sheva, Israel
    2. The Zlotowski Center for Neuroscience, Ben-Gurion University of the Negev, Beer-Sheva, Israel
    Contribution
    Investigation
    Competing interests
    No competing interests declared

  6. Shai Kaluski

    1. Department of Life Sciences, Ben-Gurion University of the Negev, Beer Sheva, Israel
    2. The Zlotowski Center for Neuroscience, Ben-Gurion University of the Negev, Beer-Sheva, Israel
    Contribution
    Investigation
    Competing interests
    No competing interests declared

  7. Ido Shirat

    1. Department of Life Sciences, Ben-Gurion University of the Negev, Beer Sheva, Israel
    2. The Zlotowski Center for Neuroscience, Ben-Gurion University of the Negev, Beer-Sheva, Israel
    Contribution
    Investigation
    Competing interests
    No competing interests declared

  8. Zeev Slobodnik

    1. Department of Life Sciences, Ben-Gurion University of the Negev, Beer Sheva, Israel
    2. The Zlotowski Center for Neuroscience, Ben-Gurion University of the Negev, Beer-Sheva, Israel
    Contribution
    Investigation
    Competing interests
    No competing interests declared

  9. Monica Einav

    1. Department of Life Sciences, Ben-Gurion University of the Negev, Beer Sheva, Israel
    2. The Zlotowski Center for Neuroscience, Ben-Gurion University of the Negev, Beer-Sheva, Israel
    Contribution
    Methodology, Project administration
    Competing interests
    No competing interests declared

  10. Fabian Erdel

    1. Division of Chromatin Networks, German Cancer Research Center (DKFZ), BioQuant, Heidelberg, Germany
    2. Centre de Biologie Intégrative, CNRS UPS, Toulouse, France
    Contribution
    Formal analysis, Investigation
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2888-7777

  11. Barak Akabayov

    Department of Chemistry, Ben-Gurion University of the Negev, Beer-Sheva, Israel
    Contribution
    Formal analysis, Investigation, Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-3882-2742

  12. Debra Toiber

    1. Department of Life Sciences, Ben-Gurion University of the Negev, Beer Sheva, Israel
    2. The Zlotowski Center for Neuroscience, Ben-Gurion University of the Negev, Beer-Sheva, Israel
    Contribution
    Conceptualization, Formal analysis, Funding acquisition, Investigation
    For correspondence
    toiber@bgu.ac.il
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1465-0130

Funding

Israel Science Foundation (188/17)

  • Debra Toiber

Ben Gurion University (High-tech, Bio-tech and Chemo-tech scholarship)

  • Debra Toiber

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This work was supported by ISF 188/17 and by the High-tech, Bio-tech and Chemo-tech scholarship of Kreitman School of Advanced Research of Ben Gurion University. We appreciate the plasmids kindly donated by Prof. Misteli and the U20S-LacO cells from Prof. Greenberg. We thank the staff scientist of beamline BM29 of ESRF (Grenoble, France) for providing support and the Israeli Block Allocation Group (BAG) for providing access. We thank Dr Mario Lebendiker and Dr Hadar Amartely from the Protein Purification Facility Wolfson Centre for Applied Structural Biology - The Hebrew University of Jerusalem, for their help with the SEC MALS experiments. We thank Prof. Eyal Gur and Dr Maayan Korman from Ben-Gurion University for their contribution to the development of the NAD+ consumption assay. We thank Prof. Amir Aharoni and Dr Adi Hendler from Ben-Gurion University for their help and advice.

Senior Editor

  1. Jessica K Tyler, Weill Cornell Medicine, United States

Reviewing Editor

  1. Katrin Chua, Stanford University, United States

Publication history

  1. Received: September 4, 2019
  2. Accepted: January 23, 2020
  3. Accepted Manuscript published: January 29, 2020 (version 1)
  4. Version of Record published: March 2, 2020 (version 2)
  5. Version of Record updated: March 6, 2020 (version 3)

Copyright

© 2020, Onn et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 7,882
    Page views
  • 941
    Downloads
  • 62
    Citations

Article citation count generated by polling the highest count across the following sources: Scopus, Crossref, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Lior Onn
  2. Miguel Portillo
  3. Stefan Ilic
  4. Gal Cleitman
  5. Daniel Stein
  6. Shai Kaluski
  7. Ido Shirat
  8. Zeev Slobodnik
  9. Monica Einav
  10. Fabian Erdel
  11. Barak Akabayov
  12. Debra Toiber
(2020)
SIRT6 is a DNA double-strand break sensor
eLife 9:e51636.
https://doi.org/10.7554/eLife.51636

Categories and tags

Research organism

Further reading

    1. Structural Biology and Molecular Biophysics
    2. Cell Biology

    Mechanosensitive pore opening of a prokaryotic voltage-gated sodium channel

    Peter R Strege, Luke M Cowan ... Arthur Beyder
    Research Article Updated

    Voltage-gated ion channels (VGICs) orchestrate electrical activities that drive mechanical functions in contractile tissues such as the heart and gut. In turn, contractions change membrane tension and impact ion channels. VGICs are mechanosensitive, but the mechanisms of mechanosensitivity remain poorly understood. Here, we leverage the relative simplicity of NaChBac, a prokaryotic voltage-gated sodium channel from Bacillus halodurans, to investigate mechanosensitivity. In whole-cell experiments on heterologously transfected HEK293 cells, shear stress reversibly altered the kinetic properties of NaChBac and increased its maximum current, comparably to the mechanosensitive eukaryotic sodium channel NaV1.5. In single-channel experiments, patch suction reversibly increased the open probability of a NaChBac mutant with inactivation removed. A simple kinetic mechanism featuring a mechanosensitive pore opening transition explained the overall response to force, whereas an alternative model with mechanosensitive voltage sensor activation diverged from the data. Structural analysis of NaChBac identified a large displacement of the hinged intracellular gate, and mutagenesis near the hinge diminished NaChBac mechanosensitivity, further supporting the proposed mechanism. Our results suggest that NaChBac is overall mechanosensitive due to the mechanosensitivity of a voltage-insensitive gating step associated with the pore opening. This mechanism may apply to eukaryotic VGICs, including NaV1.5.

    1. Cell Biology
    2. Immunology and Inflammation

    HUWE1 controls tristetraprolin proteasomal degradation by regulating its phosphorylation

    Sara Scinicariello, Adrian Soderholm ... Gijs A Versteeg
    Research Article

    Tristetraprolin (TTP) is a critical negative immune regulator. It binds AU-rich elements in the untranslated-regions of many mRNAs encoding pro-inflammatory mediators, thereby accelerating their decay. A key but poorly understood mechanism of TTP regulation is its timely proteolytic removal: TTP is degraded by the proteasome through yet unidentified phosphorylation-controlled drivers. In this study, we set out to identify factors controlling TTP stability. Cellular assays showed that TTP is strongly lysine-ubiquitinated, which is required for its turnover. A genetic screen identified the ubiquitin E3 ligase HUWE1 as a strong regulator of TTP proteasomal degradation, which we found to control TTP stability indirectly by regulating its phosphorylation. Pharmacological assessment of multiple kinases revealed that HUWE1-regulated TTP phosphorylation and stability was independent of the previously characterized effects of MAPK-mediated S52/S178 phosphorylation. HUWE1 function was dependent on phosphatase and E3 ligase binding sites identified in the TTP C-terminus. Our findings indicate that while phosphorylation of S52/S178 is critical for TTP stabilization at earlier times after pro-inflammatory stimulation, phosphorylation of the TTP C-terminus controls its stability at later stages.

    1. Cell Biology

    Defects in lipid homeostasis reflect the function of TANGO2 in phospholipid and neutral lipid metabolism

    Agustin Leonardo Lujan, Ombretta Foresti ... Vivek Malhotra
    Research Article

    We show that TANGO2 in mammalian cells localizes predominantly to mitochondria and partially at mitochondria sites juxtaposed to lipid droplets (LDs) and the endoplasmic reticulum. HepG2 cells and fibroblasts of patients lacking TANGO2 exhibit enlarged LDs. Quantitative lipidomics revealed a marked increase in lysophosphatidic acid (LPA) and a concomitant decrease in its biosynthetic precursor phosphatidic acid (PA). These changes were exacerbated in nutrient-starved cells. Based on our data, we suggest that TANGO2 function is linked to acyl-CoA metabolism, which is necessary for the acylation of LPA to generate PA. The defect in acyl-CoA availability impacts the metabolism of many other fatty acids, generates high levels of reactive oxygen (ROS), and promotes lipid peroxidation. We suggest that the increased size of LDs is a combination of enrichment in peroxidized lipids and a defect in their catabolism. Our findings help explain the physiological consequence of mutations in TANGO2 that induce acute metabolic crises, including rhabdomyolysis, cardiomyopathy, and cardiac arrhythmias, often leading to fatality upon starvation and stress.