Molecular tools unveil distinct waves of ADP-ribosylation during DNA repair

Abstract

ADP-ribosylation is a complex post-translation modification involved in DNA repair. In a recent Molecular Cell publication, Longarini and colleagues measured ADP-ribosylation dynamics with unprecedented specificity, revealing how the monomeric and polymeric forms of ADP-ribosylation regulate the timing of DNA repair events following strand breaks.

ADP-ribosylation is a complex post-translation modification involved in DNA repair and can occur as monomeric or polymeric forms. In a recent Molecular Cell publication, Longarini and colleagues measured ADP-ribosylation dynamics with unprecedented specificity, revealing how different forms of ADP-ribosylation regulate the timing of DNA repair events following strand breaks.

Main text

DNA strand breaks, estimated to occur 55,000 times a day per cell, are a central challenge our cells must face.¹ Loss-of-function mutations in DNA repair pathways are linked to cancer, neurodegeneration, and accelerated aging. Characterizing the molecular events that occur during DNA repair is crucial to understanding these diseases and could identify therapeutic targets.

PARP1 is an abundant nuclear protein that acts as a first responder to single- and double-stranded DNA breaks. PARP1 directly binds strand breaks and then uses NAD⁺ to transfer ADP-ribose onto itself and nearby proteins, a process called ADP-ribosylation (ADPr). PARP1-mediated ADPr facilitates repair within seconds of a DNA strand break by (1) decondensing chromatin, increasing accessibility to damaged DNA, and (2) recruiting proteins that recognize damaged chromatin via direct binding to the ADP-ribose modifications. The causal relationship between PARP1, ADPr, and DNA repair is well established, leading to the approval of four PARP1 inhibitors for treating breast, ovarian, and prostate cancers.²

After decades of research and clinical investment, we know surprisingly little about how ADP-ribose modifications affect protein function, mainly due to technical challenges stemming from the complex nature of ADPr. ADPr can exist in monomeric or polymeric forms, abbreviated as mono-ADP-ribose (MAR) or poly(ADP-ribose) (PAR), with PAR varying in length from 2 to over 100 ADP-ribose units. Therefore, ADPr shares attributes with other polymeric protein modifications, such as glycosylation and ubiquitination. What sets ADPr apart is its diverse conjugation sites, with the modification reported on nine amino acids (CDEHKRSTY) across thousands of human proteins.³

Despite these technical challenges, recent advancements have provided a clearer view of PARP1 signaling during DNA repair. A key moment was the discovery of histone PARylation factor 1 (HPF1)—a DNA repair factor that forms a dimer with PARP1 at DNA breaks.⁴ HPF1 is not merely a PARP1-interacting protein; it lends catalytic residues to the PARP1 active site, essentially acting as one half of a dimeric PARP1/HPF1 enzyme. As a result, the activity and specificity of PARP1/HPF1 is markedly different from PARP1 alone. While PARP1 primarily modifies itself with PAR in vitro, PARP1/HPF1 can deposit MAR on histones in vitro and in cells. Interestingly, the recruitments of PARP1 and HPF1 to DNA breaks are unique, with PARP1 arriving first and HPF1 arriving later.⁵ These data suggest temporally distinct MAR- and PAR-dependent pathways may exist during strand break repair, involving PARP1/HPF1 and PARP1 alone, respectively. However, this hypothesis was previously untestable due to the lack of reagents differentiating between MAR and PAR.

An important technical advance in monitoring the cellular dynamics of ADPr was the discovery of proteins that specifically bind MAR or PAR. These MAR- and PAR-binding domains were then converted to detection reagents through fusion to fluorescent proteins or antibody Fc regions.⁶ The PAR-binding WWE domain of RNF146 has been particularly successful, as a GFP-WWE fusion protein can track PARylation dynamics after DNA damage with live-cell imaging.⁷ The cellular dynamics of MARylation, however, have not been measured, possibly due to naturally occurring MAR-binding domains also binding to other adenosine-containing metabolites (e.g., ATP). Since the abundance of protein MARylation is relatively low, cross-reactivity with abundant metabolites could result in a prohibitively low signal-to-noise ratio in cellular environments.

To circumvent issues related to naturally occurring ADPr-binding domains, the Matic laboratory used phage display against synthetic MARylated peptides to produce a suite of recombinant antibodies.⁸ Through iterative rounds of selection, they generated antibodies that specifically bound to MAR, or both MAR and PAR, or recognized specific MARylation sites on PARP1 and histones. Crucially, one MAR-specific antibody, AbD33205, only bound to MARylated peptides, not free ADP-ribose, suggesting this reagent could be used as a cellular probe for protein MARylation because it will likely not cross-react with adenosine-containing metabolites. While these studies were a major step forward, they primarily focused on proof-of-concept experiments to demonstrate reagent specificity. The biological roles of protein MARylation during DNA damage, and how they might differ from PARylation, remained unexplored.

In a recent issue of Molecular Cell, Longarini and colleagues⁹ applied these next-generation reagents to study DNA repair in living cells. They began by improving a previously reported MAR-specific antibody (AbD33204⁸) through affinity maturation, which decreased the K_D for a MARylated peptide to ∼284 nM. Additionally, they incorporated SpyTag technology, which allows for conjugation of recombinant antibodies to useful tags (e.g., fluorophores, horseradish peroxidase, Fc regions) with a simple, 1-h protocol.¹⁰ These improvements enabled the authors to detect histone MARylation with western blotting, which was previously only possible when ADPr was induced with DNA-damaging agents.

Longarini and colleagues⁹ then turned their attention toward applying these reagents to live-cell imaging. An important aspect of the recombinant antibodies is their availability as monovalent fragments (Fabs), which are smaller than conventional antibodies (∼50 kDa for a Fab versus ∼150 kDa for an antibody). The authors also leveraged a technique called “bead-loading” to gently permeabilize the plasma membrane, allowing Fabs to enter the cell. Once inside the cytoplasm, the Fabs are small enough to passively enter the nucleus, where they can detect ADPr on the chromatin in real time (Figure 1).

Figure 1.

Technical advancements lead to biological insight

(Top) Schematic of the procedure used to track MARylation in living cells following DNA damage. (Bottom) The two-wave model of ADPr signaling following DNA strand break.

The authors observed the rapid recruitment of their protein MARylation detection reagent to laser-induced DNA strand breaks, confirming that protein MARylation is induced by DNA damage. The signal from protein MARylation peaks around 2 min after the break and then gradually decreases over approximately 10 min. This timeline is strikingly different from the PAR-specific GFP-WWE reagent, which peaks at DNA breaks much sooner (within 10 s) and then decreases at least 2-fold within the first minute.

The biphasic nature of their PAR and MAR detection reagents led the authors to investigate whether protein MARylation serves as a newly created mark for DNA damage or whether it is simply a remnant of PARG degradation of PARylation to MARylation. Experiments with PARG inhibitors revealed that some MARylation on PARP1 and histones depends on PARG activity, while MARylation on other proteins is independent of PARG. Therefore, this second wave of MAR can originate from two sources: deposition of MAR by PARP1/HPF1 or degradation of PAR to MAR by PARG.

To further explore the possible functions of a MAR-dependent second wave, Longarini and colleagues aimed to identify any "MAR readers" that could be recruited to DNA breaks via direct interactions at this later stage of repair. They conducted proteomics experiments using recombinant peptides and nucleosomes containing mono-ADPr at specific serine residues known to be ADP-ribosylated after DNA damage. By capturing proteins in cell lysates with their recombinant mono-ADPr probes and analyzing endogenous chromatin exposed to DNA damage, they identified several dozen potential MAR readers.

Among the hits they identified, RNF114, an E3 ligase implicated in cell cycle control and apoptosis, was particularly intriguing. RNF114 bound to a MARylated peptide in vitro and only bound to PAR weakly, demonstrating MAR specificity. In cells, GFP-tagged RNF114 is recruited to DNA breaks gradually, matching the temporal profile of the MAR detection reagent. The zinc fingers within RNF114 are MAR-interacting domains, and a single cysteine-to-alanine mutation (C176A) within a zinc finger abolishes the RNF114-MAR interaction in vitro and the recruitment of RNF114 to DNA breaks in cells.

To assess the functional impact of MAR-dependent recruitment of RNF114 to DNA breaks, Longarini and colleagues generated a series of knockout and knockdown cell lines and then analyzed their DNA damage responses. RNF114 knockout cells were more sensitive to DNA damage and less capable of repairing telomeric DNA strand breaks induced by FokI nuclease. Mechanistically, RNF114 recruits another E3 ligase, RNF168, that deposits ubiquitin on histone H2A. H2A ubiquitin is then read by 53BP1, a central component of double-strand break repair. Adding back RNF114 WT, but not RNF114-C176A, partially or completely rescued these phenotypes, indicating the RNF114-MAR interaction is an important step in DNA repair signaling.

Collectively, Longarini and colleagues⁹ defined how PARP1, HPF1, MAR, and PAR coordinate DNA repair in living cells. By persisting longer than PARylation, protein MARylation can exert effects on later stages of the repair process by recruiting unique proteins, like RNF114 (Figure 1).

Several unanswered questions remain, including the potential involvement of other MAR readers in DNA repair. The DTX3L/PARP9 heterodimer emerged as a strong candidate, especially given the recent discovery that it ubiquitinates MARylated proteins. This process involves the direct modification of MAR through a non-canonical O-linkage between the C terminus of ubiquitin and the 3′-hydroxyl of the ADPr.¹¹ It is possible that histone MARylation serves as a “primer” for DTX3L/PARP9 ubiquitination, generating histone MAR-ubiquitin conjugates that add another layer of complexity to histone modifications following DNA damage.

Additionally, the authors used techniques that introduce a heterogeneous mixture of single- and double-stranded breaks, with the exception of FokI, which only produces double-strand breaks. It will be interesting to directly compare the kinetics of MAR and PAR with techniques that exclusively produce single- or double-stranded breaks. Considering that MAR and PAR recruit different proteins, the balance between these two forms of ADPr might regulate pathway choice.

The development of recombinant antibodies by the Matic laboratory represents an exciting technological frontier for the field. Development of reagents specific to the recently discovered RNA-ADPr, DNA-ADPr, and ubiquitin-ADPr will uncover additional hidden aspects of ADPr biology.