Development and Characterization of PAR-Trackers: New Tools for Detecting Poly(ADP-ribose) In Vitro and In Vivo

ADP-ribosylation (ADPRylation) is a reversible post-translation modification resulting in the covalent attachment of ADP-ribose (ADPR) moieties on substrate proteins. Naturally-occurring protein motifs and domains, including WWEs, PBZs, and macrodomains, act as “readers” for protein-linked ADPR. Although recombinant, antibody-like ADPR detection reagents containing these readers have facilitated the detection of ADPR, they are limited in their ability to capture the dynamic nature of ADPRylation. Herein, we describe and characterize a set of poly(ADP-ribose) (PAR) Trackers (PAR-Ts) - optimized dimerization-dependent or split-protein reassembly PAR sensors in which a naturally occurring PAR binding domain, WWE, was fused to both halves of dimerization-dependent GFP (ddGFP) or split Nano Luciferase (NanoLuc), respectively. We demonstrate that these new tools allow the detection and quantification of PAR levels in extracts, living cells, and living tissues with greater sensitivity, as well as temporal and spatial precision. Importantly, these sensors detect changes in cellular ADPR levels in response to physiological cues (e.g., hormone-dependent induction of adipogenesis without DNA damage), as well as xenograft tumor tissues in living mice. Our results indicate that PAR Trackers have broad utility for detecting ADPR in many different experimental and biological systems.


ADP-ribosylation (ADPRylation) is a regulatory post-translational modification (PTM) of
proteins that results in the reversible attachment of ADP-ribose (ADPR) units on substrate proteins Luscher et al., 2018). Members of the PARP family of ADPribosyltransferases (ARTs) (Ame et al., 2004;Vyas et al., 2013) function as "writers" to catalyze the transfer of ADPR moieties from oxidized b-nicotinamide adenine dinucleotide (NAD + ) to a variety of amino acids (Asp, Glu, Ser, Arg, Lys) in substrate proteins Luscher et al., 2018). Mono(ADP-ribosyl) transferases (MARTs) add a single ADPR moiety to their substrates through a process called mono(ADP-ribosyl)ation (MARylation) (Challa et al., 2021), whereas poly(ADP-ribosyl) transferases (PARPs) add branched or linear chains of multiple ADPR moieties through a process called poly(ADP-ribosyl)ation (PARylation) Luscher et al., 2018). Site-specific ADPRylation of substrate proteins by PARP enzymes can have important functional consequences, including alteration of the biochemical or biophysical properties of the substrate protein or creation of new binding sites for ADPR binding domains that drive protein-protein interactions (Gibson and Kraus, 2012;Gupte et al., 2017). As such, ADPRylation can control a wide variety of cellular and biological processes, including DNA repair, DNA replication, gene expression, and RNA biology, as well as inflammatory responses and cell differentiation. (Challa et al., 2021;Gibson and Kraus, 2012;Gupte et al., 2017;Luscher et al., 2018).
Although recent developments in mass spectrometry-based identification of ADPRmodified amino acids have enhanced the study of specific ADP-ribosylation events on target proteins (Daniels et al., 2015), the lack of a complete set of immunological tools that recognize the diverse forms of ADPR has hampered progress in studying ADPRylation. Anti-ADPR polyclonal antibodies have been reported, but the specificity and utility of these antibodies has not been assessed broadly (Bredehorst et al., 1978;Kanai et al., 1974Kanai et al., , 1978Meyer and Hilz, 1986;Sakura et al., 1978). Instead, the PARP field has relied on the anti-PAR monoclonal antibody 10H, which binds to PAR chains longer than ten ADPR units (Kawamitsu et al., 1984). Although useful, this antibody has left the field blind to mono-and oligo(ADP-ribosyl)ation.
The recent development of recombinant site-specific and broad-specificity antibodies to ADPR has been a major advance (Bonfiglio et al., 2020). We recently described the generation and characterization of a set of recombinant antibody-like ADP-ribose binding proteins, in which natural ARBDs have been functionalized with the Fc region of rabbit immunoglobulin . Collectively, the ADPR detection tools described here are useful for cellular and biochemical assays, but they are not useful for exploring the dynamics of ADPRylation in cells and in animals. In fact, ADPRylation is a rapid process that can occur within minutes and can be removed by various 'erasers' including ARH3 and PARG (Barkauskaite et al., 2013;Gupte et al., 2017). Therefore, developing tools to measure ADPR dynamics in cells and in vivo is critical for better understanding the various biological processes mediated by ADPR.
To overcome this limitation, several split-protein reassembly approaches have been applied to PAR detection. With this approach, non-functional fragments of a split-fluorescent protein or luciferase are induced to reassemble through the direct interaction of fused ADPR binding domains (Furman et al., 2011;Krastev et al., 2018;Lee et al., 2021;Serebrovskaya et al., 2020). These include the PBZ modules of aprataxin PNK-like factor (APLF) (Ahel et al., 2008) attached to each half of split firefly luciferase (split-Fluc) (Furman et al., 2011), PBZ modules with split Venus GFP (Krastev et al., 2018), and WWE domains with Turquoise and Venus, allowing for Förster
In the work described herein, we developed a set of PAR Trackers (PAR-Ts) -optimized dimerization-dependent and split-protein reassembly PAR detection tools that have broad utility for both in vitro and in vivo studies. The PAR-Ts contain a WWE fused to both parts of dimerization-dependent GFP (ddGFP) (Alford et al., 2012) or split Nano luciferase (NanoLuc) (Wang et al., 2020) with LSSmOrange. The ddGFP version (PAR-T ddGFP) allows for real time assessment of dynamic PAR production in vitro and in living cells, while the split NanoLuc version (PAR-T NanoLuc) allows detection of PAR production in tissues in living mammals.

RESULTS
The use of functionalized ARBDs to detect ADP-ribosylation has been a useful approach Gibson et al., 2017;Timinszky et al., 2009). In this study, we further developed our previous ADPR detection reagents  to expand their utility to in vivo applications as ADPR sensors. To achieve this, we used a systematic approach including in vitro characterization of the sensors and validation of their utility in vivo.

Using dimerization-dependent GFP-based sensors to detect PAR in vitro
Dimerization-dependent GFP is a genetically encoded sensor that was initially developed to study protein interactions (Alford et al., 2012). In this system, a pair containing a quenched GFP (ddGFPA) and a non-fluorogenic GFP (ddGFPB) form a heterodimer with improved fluorescence (Alford et al., 2012). The reversible complementation of ddGFP pairs, unlike irreversible split fluorophores, is ideal to monitor dynamic signaling events, such as PARylation (Villalobos et al., 2007). Hence, we sought to design ddGFP-based fluorescence sensors for PAR (fluorescent PAR-Trackers or PAR-T ddGFP) (Table 1) to enable us to perform live cell imaging, with a high signal-noise ratio. To achieve this, we fused various ADPR binding domains (ARBDs) H2O2 to induce DNA damage and activate PARP-1, or an inhibitor of the PAR glycohydrolase PARG (i.e., PDD00017273) to increase PAR. We observed an increase in fluorescence when lysates from cells treated with either H2O2 or PARG inhibitor (PDD00017273) were used, as well as a profound increase in fluorescence when lysates from cells treated both H2O2 and PARG inhibitor were used (Figure 1 -figure supplement 1, I and J). The H2O2-and PARG inhibitorstimulated signals were reduced with PARP inhibitor treatment (Figure 1 -figure supplement 1, I and J). Together, these data suggest that the PAR-T ddGFP sensors can specifically recognize PAR, and that they exhibit a good dynamic range in vitro.

Using PAR-T ddGFP sensors to detect PAR in live cells
Having confirmed the specificity of the PAR-T sensors, we next sought to test their utility in live cell imaging. We expressed the PAR-T sensors in HeLa cells using doxycycline (Dox) induction and performed live cell imaging after subjecting the cells to H2O2-mediated PARP-1 activation ( Cancers are heterogenous tissues with spatial variation in nutrient availability and cellextrinsic stressors (Dagogo-Jack and Shaw, 2018). PAR is enhanced by stressors, such as DNA damage or hypoxia, but the spatiotemporal dynamics of PAR production in cells have remained unclear due to a lack of efficient detection methods. Thus, we asked if PAR levels vary spatially in groups of cells. We performed live cell imaging in three-dimensional (3D) cancer spheroids using MCF-7 human breast cancer cells expressing the WWE-based PAR-T-ddGFP sensor. We observed a heterogeneous distribution of PAR throughout the spheroid, which was inhibited by the PARP inhibitor, Niraparib (Figure 3, A and B). We also performed a time course of Niraparib treatment in 3D cancer spheroids to visualize spatio-temporal changes in PAR levels over time.
The results indicate that the PAR levels in cells at the core of the spheroids are relatively resistant to Niraparib treatment, since the PAR levels in these cells decrease at a lower rate compared to the PAR levels in the cells in the outer layer of the spheroid (Figure 3, C and D). These results show that the based PAR-T-ddGFP sensor can be used for live cell imaging to evaluate spatial and temporal changes in PARylation in cancer cells.

Developing a highly sensitive split-luciferase PAR-T sensor
Since we demonstrated that WWE-domain based PAR-T sensors can specifically detect PAR in biochemical assays and living cells, we sought to develop a set of highly sensitive PAR-T sensors that can detect PAR in vivo. Fluorescent sensors are not well suited for detection in vivo due to auto-fluorescence of tissues that can cause high background signals. Instead, luminescencebased approaches are preferred for in vivo applications due to lack of auto-luminescence in tissues (Tung et al., 2016). Hence, we developed a set of luminescent PAR-T sensors to detect PAR levels in vivo. We first generated sensors based on split firefly luciferase (Maita et al., 2014) using various combinations of the ARBDs. As before, the WWE domains consistently performed better in detecting an increase in PAR levels with PARG inhibitor treatment and a decrease in PAR levels with a PARP inhibitor treatment using either cell lysates ( Although split firefly luciferase-based approaches were capable of detecting signals from cells in vitro, the signal intensity from the split firefly luciferase-based PAR-T sensors were ~10,000 fold less than intact firefly luciferase, which makes it difficult to use this sensor in vivo. This limitation could be due to the bulkiness of firefly luciferase that may interfere with the function of the domains fused to them in complementation assays (Wang et al., 2020;Yano et al., 2018). Therefore, we decided to employ a Nano luciferase (NanoLuc)-based split luciferase complementation system (Dixon et al., 2016;Hall et al., 2012), which is a smaller, brighter, and more stable luciferase compared to firefly luciferase ( Figure 4A).
To quantitatively assess the activity of this luminescent PAR-Tracker (PAR-T NanoLuc), we expressed it in a Dox-dependent manner in human breast cancer cells that also stably express firefly luciferase (MDA-MB-231-Luc cells) ( Figure 4A). In this way, we had an internal standard (i.e., the signal from the firefly luciferase), which allowed us to account for changes in cell viability or tumor size in these experiments. We first tested if there was cross reactivity of the two luciferases (NanoLuc and firefly luciferase) to the substrates; we observed specific detection of firefly luciferase with D-Luciferin and NanoLuc with furimazine with no cross-reactivity ( Figure   4, B-D). Moreover, the luminescence of PAR-T NanoLuc is only 30-fold lower than intact firefly luciferase ( Figure 4C). PARP-1 depletion reduced the luminescence from PAR-T NanoLuc with little effect on the luminescence of firefly luciferase (Figure 4, E-G). Interestingly, knockdown of PARP-2 had no effect on luminescence from PAR-T NanoLuc. Nevertheless, the luminescent PAR-T NanoLuc sensor is highly sensitive and can be used to detect PAR in 1,000 cells with a dynamic range of approximately three-fold (minimum to maximum) ( DNA damaging agents, such as UV irradiation and gamma irradiation, activate PARP-1 and promote auto and trans PARylation of PARP-1 and other DNA damage repair proteins, respectively, that are recruited to sites of DNA damage (Ray Chaudhuri and Nussenzweig, 2017).
Since the PAR-T sensor can detect H2O2-induced PARP-1 activation (Figure 2), we assessed whether it can detect radiation-induced PARP-1 activation. We subjected the MDA-MB-231-Luc cells to Dox-induced expression of PAR-T NanoLuc and then exposed the cells to UV radiation.
We observed that UV radiation induced PARP-1 activation as assessed by an accumulation of PAR on Western blots ( Figure 5A). This was further enhanced by inhibition of PARG, whereas inhibition of PARP-1 blocked the UV-induced PARP-1 activation ( Figure 5A). UV radiation of PARG inhibitor-treated cells enhanced PAR-T luminescence, whereas UV radiation of PARP inhibitor-treated cells reduced the PAR-T luminescence ( Figure 5, B and C). None of these treatments affected the luminescence from firefly luciferase ( Figure 5, B and C).
In a similar manner, we also performed a time course of UV-mediated PARP1 activation using live cell luminescence assay with the PAR-T NanoLuc sensor. We subjected the MDA-MB-231 cells to Dox-induced expression of PAR-T NanoLuc or intact Nano luciferase and then exposed the cells to UV radiation. Consistent with the previous experiment, we observed a timedependent increase in PAR-T NanoLuc signal in vehicle-treated cells, but not in Niraparib-treated cells. Interestingly, UV-mediated increases in PARP1 activation were more spontaneous in PARG inhibitor-treated cells ( Figure 5D). The PAR levels under basal (-UV) conditions were low, resulting in only a 50% decrease in PAR-T NanoLuc signal with Niraparib treatment ( Figure 5, A-C). The decrease in PAR-T NanoLuc signal was greater when UV-treated cells were pre-treated with Niraparib, which is consistent with the results from Western blot analysis ( Figure 5A).

Comparison of assay performance using the PAR-T sensor and conventional PAR detection reagents
Next, we performed a set of assays to compare the performance of the PAR-T sensors to conventional PAR detection reagents (WWE-Fc and PAR antibody) in a variety of assays. In sum, we compared: (1) Western blotting with WWE-Fc versus fluorescence assay with PAR-T ddGFP, which were performed in conjunction with ARH3-mediated degradation of PAR in vitro ( Figure   6A); (2) Western blotting with WWE-Fc versus live cell luciferase assay using PAR-T NanoLuc, which were performed in conjunction with UV-induced DNA damage in MDA-MB-231 cells ( Figure 6B); (3) ELISA with PAR antibody versus fluorescence assay with PAR-T ddGFP, which were performed using immobilized PAR ( Figure 6C); and (4) Immunofluorescence with WWE-Fc versus live cell imaging using PAR-T ddGFP, which were performed using H2O2-mediated PARP1 activation in 293T cells ( Figure 6, D and E).
We compared the dynamic ranges of the PAR-T sensors with the other reagents (WWE-Fc and PAR antibody) in various PAR detection assays, such as Western blotting and ELISA ( Figure   6F). We found that Western blotting with WWE-Fc had the highest dynamic range for detection of PAR (8-fold), but the dynamic range of live cell luciferase assay with PAR-T NanoLuc was comparable (6-fold) ( Figure 6F). While PAR-T ddGFP in a modified fluorescence assay had a larger dynamic range than PAR antibody in an ELISA (6-fold vs 3.5-fold) when the assays were performed using immobilized PAR, it had a lower dynamic range when used for live cell imaging (4.4-fold). This can be explained, in part, by the higher autofluorescence of cells, which can diminish the dynamic range of the PAR-T ddGFP sensors. Thus, this sensor requires further optimization to increase the signal-noise ratio. Nevertheless, the performance of PAR-T ddGFP in live cells is comparable to that of an immunofluorescence assay with WWE-Fc (4.4-fold vs 5fold).

Detection of PAR production from PARP-1 activation under physiological conditions
We previously showed that PARP-1 catalytic activity decreases during the initial differentiation of preadipocytes (Huang et al., 2020;Luo et al., 2017;Ryu et al., 2018). Thus, adipogenesis is a unique biological process to study the dynamics of PAR accumulation from changes in PARP-1 activity under physiological conditions. We used the PAR-T NanoLuc sensor to investigate changes in PARP-1 activity during early adipogenesis of murine preadipocytes (i.e., 3T3-L1 cells). We observed a decrease in the signal from PAR-T NanoLuc by 12 hours of differentiation and a greater reduction in PAR-T NanoLuc signal by 24 hours of differentiation ( Figure 7), consistent with our previous observation that PARP-1 activation decreases precipitously during adipogenesis (Huang et al., 2020;Luo et al., 2017;Ryu et al., 2018). These results further highlight the high sensitivity of PAR-T NanoLuc sensor, which can be used to study physiological changes in PAR levels during biological processes, such as adipogenesis.

Detection of PAR production from PARP-1 activation in vivo
To assess the in vivo utility of PAR-T NanoLuc, we established xenograft tumors in NOD/SCID/gamma (NSG) mice using MDA-MB-231-Luc cells and induced the expression of PAR-T NanoLuc with Dox ( Figure 8A). Similar to the in vitro experiments, we detected an increase in PAR levels with the PAR-T NanoLuc sensor when the gamma-irradiated mice were treated with PARG inhibitor, but the signal was decreased when the gamma-irradiated mice were treated with PARP inhibitor ( Figure 8B and Figure 8 -figure supplement 1A). We normalized the signal from PAR-T to luminescence from firefly luciferase to confidently measure the differences in PAR levels, while accounting for the variability in tumor sizes ( Figure 8C).
We also measured PAR accumulation in breast cancer cells injected into C57/BL6 mice without establishing xenograft tumors over a 24 hour time course post injection, with or without PARG inhibitor treatment (Figure 8 -figure supplement 1, B-D). PAR accumulation was readily detected in the breast cancer cells injected into the mice in the absence of treatment. Upon treatment with PARG inhibitor, the luminescence from PAR-T NanoLuc increased significantly by 6 hours and then diminished by 24 hours. These results demonstrate that the PAR-T NanoLuc sensor has sufficient sensitivity to detect dynamic changes in PAR production in tissues of living animals in vivo.

DISCUSSION
Naturally occurring ADPR binding domains (ARBDs) have been invaluable tools for developing novel ADPR detection reagents and sensors Gibson et al., 2017;Timinszky et al., 2009). In this study, we developed a set of PAR sensors that are useful tools for in vitro assays, live cells, and tissues in living animals (Table 1). To this end, we constructed a ddGFP-based fluorescent PAR-Tracker (PAR-T) that can be used for in vitro assays and live cell imaging, to track dynamic changes in PAR levels in a single cell ( Figure 2) and spatially resolve PAR accumulation in tissue-like spheroids ( Figure 3). In addition, we made a split NanoLucbased luminescent PAR-T containing LSSmOrange that can be used to track physiological changes in PAR levels during differentiation (Figure 7). It is extremely sensitive and can be used to detect PAR in xenograft tumors in living mice ( Figure 8).
However, the aforementioned approaches for PAR sensors have the following limitations.
(1) they are irreversible, thus limiting their use for measuring dynamic changes in PAR levels, (2) the requirement for measuring filtered light emissions in FRET-based sensors results in low signal intensities and narrow dynamic ranges that limit their utility in a variety of applications, (3) PBZ domains detect branched PAR chains synthesized predominantly by PARP-2 (Chen et al., 2018), and (4) they are unable to measure PAR levels in vivo. To overcome these limitations, we developed a set of PAR sensors in which ARBDs are fused to ddGFP or split luciferase. Assembly of dimerization-dependent fluorophores and split luciferase proteins are reversible and have a higher signal intensity (Alford et al., 2012;Luker et al., 2004). In our screen for the best ARBDs for PAR detection, the WWE domain from RNF146 and the macrodomain from AF1521 exhibited superior detection of PAR production by activated PARP-1, compared to the PBZ domains from APLF or the macro H2A.1 macrodomain (Figure 1 -figure supplement 1E and Table 1). Thus, although useful, the currently available PAR sensors have multiple opportunities for improvement.

Detection of PAR in tissues in living animals
A major goal of this work was to generate a sensor with sufficient stability and sensitivity to allow detection of PAR in tissues in living animals. We faced several challenges in doing so.
For example, fluorescent sensors are not optimal for use in vivo due to high auto-fluorescence of tissues, thus we had to use a luminescent sensor to increase the sensitivity of PAR detection in vivo. Although intact luciferase can be used in cells as reporter, achieving usable signals from split luciferase is technically challenging because (1) it is difficult to express, (2) the luminescence of split luciferase is typically 100-1000 fold less than intact luciferase, and (3) the wavelength emitted by luciferase exhibits poor penetration in tissues. A broadly useful PAR detection tool would need to overcome these limitations and allow real-time dynamic observations in cells and detection in tissues.
To this end, we optimized several aspects of the sensor to achieve the highest sensitivity: (1) we used NanoLuc, the smallest and brightest luciferase available (Wang et al., 2020), (2) we added LSSmOrange to stabilize the C-terminal fragment of NanoLuc (Schaub et al., 2015), (3) we used the Nano-Glo live cell substrate to be able to perform these assays in live cells, (4) we used Dox-inducible constructs to avoid any effects of expression of these constructs on cell viability, and (5) we developed a dual luciferase assay to quantify PAR levels more accurately. The blueshifted emission of NanoLuc (at 460 nm) diffuses rapidly in tissues and hence it is not optimal for imaging deep tissues (Schaub et al., 2015). Therefore, we fused the C-terminal domain of NanoLuc with LSSmOrange fluorescent protein (red-shifted GFP variant), which has a higher fluorescence quantum yield that is compatible for excitation by the light emitted by NanoLuc.
Upon addition of the luciferase substrate, the bioluminescence energy from NanoLuc excites the LSSmOrange fluorophore, shifting the emission energy to 570 nm via BRET reaction (Schaub et al., 2015).
In summary, we generated a set of PAR-Trackers with significant improvements in functionality over our previous detection reagents. The PAR-T NanoLuc sensor can detect PAR levels in as few as 1000 cells with a good dynamic range of detection (Figure 4 -figure supplement 2) and it can also detect PAR in tissues in living animals (Figure 7).

Perspective and future advances for PAR sensors
Current studies on PARylation are limited to in vitro biochemical assays and end-point cellular assays. Moreover, the techniques routinely used to measure PAR levels require laborious and time consuming assays, such as Western blotting, ELISA, immunofluorescence, or immunohistochemistry. The high sensitivity and low signal to noise ratios of the PAR-Trackers described here enable spatial and temporal monitoring of PAR levels in cells and in animals.
Moreover, these techniques do not require exogenous ligands and involve limited manipulation to cells that can limit artifacts caused by sample handling, such as lysing or fixing of cells.
Generating animal models with tissue-specific expression of the PAR-T NanoLuc sensor will enable monitoring PAR levels in specific cell types in vivo.
Although the dynamic range of the PAR-T sensors are comparable to other available PAR detection tools, the sensors can be improved further by optimizing the assay conditions and the design of the PAR-T sensors. For example, our data suggests that in plate-based fluorescence assays, immobilizing PAR on the well increases the sensitivity of detection by PAR-T ddGFP.
The reduced dynamic range of PAR-T ddGFP in live cell imaging assays can be attributed to high background autofluorescence from the cells and culture medium. To minimize this background signal, future PAR-T fluorescent sensors should be developed using fluorophores that have low background signal, such as those that emit fluorescence at a far-red wavelength. In addition, the ddGFP-A fragment has low fluorescence intensity, possibly contributing to reduced dynamic range. To avoid this, an optimized split-fluorescence sensor should be used. Similarly, the background signals contributing to reduced dynamic range in the assays using the PAR-T NanoLuc sensor can be reduced by optimizing the luciferase-fluorophore reporter pair used in the sensor.    Inc. He is also coholder of U.S. Patent 9,599,606 covering the ADP-ribose detection reagent used herein, which has been licensed to and is sold by EMD Millipore.
PATENT K.W.R., S.C., and W.L.K. have a patent pending for the PAR-T sensors described herein.

DATA AVAILABILITY
All unprocessed Western blot image data and data from individual replicates from the fluorescent and luminescent assays can be found here: DOI: 10.17632/x9j73tdb5r. The work does not contain any high complexity, high content "omics" data.

Cell culture and treatments
HeLa, 293T, 3T3-L1, and MCF-7 cells were obtained from the American Type Cell Culture, and MDA-MB-231-Luc cells were obtained from Dr. Srinivas Malladi, UT Southwestern Medical Center. Fresh cell stocks were regularly replenished from the original stocks, verified for cell type identity using the GenePrint 24 system (Promega, B1870), and confirmed as mycoplasma-free using a commercial testing kit every three months.
HeLa, 293T, and MCF-7 cells were cultured in DMEM (Sigma-Aldrich, D5796) supplemented with 10% fetal bovine serum (Sigma, F8067) and 1% penicillin/streptomycin. 3T3-L1 cells were cultured in DMEM (Cellgro, 10-017-CM) supplemented with 10% fetal bovine serum (Atlanta Biologicals, S11550) and 1% penicillin/streptomycin. For the luciferase assays, 5,000 3T3-L1 cells were plated in each well of a 96-well format plate. For the induction of adipogenesis, the 3T3-L1 cells were grown to confluence and then cultured for two more days until contact inhibited. The cells were then treated for two days with MDI adipogenic cocktail containing 0.25 mM IBMX, 1 µM dexamethasone, and 10 µg/mL insulin for 12 hours or 24 hours as indicated. Expression of PAR-T NanoLuc was induced by treating the 3T3-L1 cells with doxycycline (Dox) for 24 hours before the luciferase assay was performed.

Vectors for ectopic expression and knockdown
The vectors described below were generated using the oligonucleotide primers described in the next section. All constructs were verified by sequencing.
Mammalian expression vectors. The plasmids for Dox-inducible expression of the ddGFP PAR-T constructs were generated using a cDNA for ddGFP-A (Addgene, 40286) or ddGFP-B (Addgene, 40287). cDNAs for the PAR binding domains were amplified from previously published pET19b constructs . The cDNAs were assembled and cloned first into pCDNA3 and then into pInducer20 or pET19b using Gibson assembly (NEB, E2621). The split luciferase constructs were synthesized as gene blocks (Integrated DNA Technologies), and then cloned into the pInducer20 vectors using Gibson assembly.

Western blotting
Protein concentrations of the cell lysates were determined using a Bio-Rad Protein Assay Dye Reagent (Bio-Rad, 5000006). Volumes of lysates containing equal amounts of total protein were boiled at 100°C for 5 minutes after addition of 1/4 volume of 4x SDS-PAGE Loading Solution (250 mM Tris, pH 6.8, 40% glycerol, 0.04% Bromophenol Blue, 4% SDS), run on 6% polyacrylamide-SDS gels, and transferred to nitrocellulose membranes. After blocking with 5% nonfat milk in TBST, the membranes were incubated with the primary antibodies described above in 1% non-fat milk in TBST with 0.02% sodium azide, followed by anti-rabbit HRP-conjugated IgG (1:5000) or anti-mouse HRP-conjugated IgG (1:5000). Immunoblot signals were detected using an ECL detection reagent (Thermo Fisher Scientific, 34577, 34095).

Purification of PAR-T sensor proteins expressed in bacteria
His-tagged PAR detection reagents in the pET19b-based bacterial expression vector were expressed in E. coli strain BL21(DE3-pLysis). The transformed bacteria were grown in LB containing ampicillin at 37 °C until the OD595 reached 0.4−0.6. Recombinant protein expression was induced by the addition of 1 mM IPTG for 3 hours at 37 °C. The cells were collected by centrifugation, and the cell pellets were flash-frozen in liquid nitrogen and stored at -80 °C until further use. The frozen cell pellets were thawed on ice and lysed by sonication in Ni-NTA Lysis Buffer (10 mM Tris-HCl pH 7.5, 0.15 M NaCl, 0.5 mM EDTA, 0.1% NP-40, 10% glycerol, 1 mM PMSF, and 1 mM β-mercaptoethanol). The lysates were clarified by centrifugation at 15,000 rpm using an SS34 rotor (Sorvall) at 4°C for 30 minutes. The supernatant was incubated with 1 mL of Ni-NTA resin equilibrated in Ni-NTA Lysis Buffer at 4 °C for 2 hours with gentle mixing. The resin was collected by centrifugation at 4 °C for 10 minutes at 1,000 x g, and the supernatant was removed. The resin was washed five times with Ni-NTA Wash Buffer (10 mM Tris-HCl pH 7.5, 0.3 M NaCl, 0.2% NP-40, 10% glycerol, 15 mM imidazole, 1 mM PMSF and 1 mM βmercaptoethanol). The recombinant proteins were then eluted using Ni-NTA Elution Buffer (10 mM Tris-HCl pH 7.5, 0.2 M NaCl, 0.1% NP-40, 10% glycerol, 500 mM imidazole, 1 mM PMSF, and 1 mM β-mercaptoethanol). The eluates were collected by centrifugation at 4°C for 10 minutes at 1,000 x g, and dialyzed in Ni-NTA Dialysis Buffer (10 mM Tris-HCl pH 7.5, 0.15 M NaCl, 10% glycerol, 1 mM PMSF, and 1 mM β-mercaptoethanol). The dialyzed proteins were quantified using a Bradford protein assay (Bio-Rad), aliquoted, flash-frozen in liquid N2, and stored at −80°C.

Detection of PAR by PAR-T sensors in vitro
The following methods were used to measure PAR levels using the recombinant PAR-T constructs in vitro.

In vitro auto ADP-ribosylation assays.
In vitro auto ADP-ribosylation assays were performed essentially as described previously . The ADP-ribosylation reactions contained 0.2 µM purified PARP-1 or PARP-3 and 100 ng/µL of sheared salmon sperm DNA (Invitrogen, AM9680). Purified proteins were mixed with the indicated amounts of NAD + in ADP-ribosylation Buffer (50 mM Tris-HCl pH 7.5, 0.125 M NaCl, 12.5 mM MgCl2). The reactions were incubated at room temperature for 15 minutes and terminated by addition of 200 nM PJ34.
For time course experiments, in vitro ADP-Ribosylation assays were performed as described above in the presence of 200 nM of recombinant PAR-T sensor proteins. Two hundred and fifty µM NAD + was added and the reaction was allowed to proceed for 16 minutes. The fluorescence intensities were measured every 30 seconds using a plate reader (CLARIOstar BMG Labtech). A similar in vitro ADP-ribosylation reaction was performed and processed for western blotting as described above.
Tracking PAR degradation in vitro. For measuring the level of PAR degradation in vitro, 5 nM of PARP-1 and 200 nM recombinant PAR-T sensor proteins were incubated in ADPribosylation Buffer along with the indicated amounts of ARH3. The reaction mixture was incubated at 37°C for the indicated amount of time and fluorescence intensities were measured.
Tracking PAR formation in cell extracts. For measuring the levels of PAR in mammalian cell lysates, HeLa cells subjected to Dox-induced expression of the PAR-T sensor proteins were treated with 20 µM PJ34 or 20 µM PDD 00017273 for 2 hours prior to induction of DNA damage by treatment with H2O2 (1 mM; Sigma, 216763) for 10 minutes. The cells were then lysed in Lysis Buffer (20 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% NP-40, 1% sodium deoxycholate, 0.1% SDS) containing 1 mM DTT, 250 nM ADP-HPD (Sigma, A0627), 10 μM PJ34 (Enzo, ALX-270), and 1x complete protease inhibitor cocktail (Roche, 11697498001). Equal volumes of the lysate and ADP-ribosylation Buffer were incubated with 200 nM of recombinant PAR sensor proteins. The reaction mixture was incubated at room temperature for 15 minutes and the fluorescence intensity was measured.

Measuring the levels of immobilized PAR in a plate-based assay.
Plate-based fluorescent assays using PAR-T ddGFP were performed in a manner similar to the ELISA assays according to the manufacturer's protocol (Cell Biolabs, XDN-5114). Briefly, the protein binding strip-wells were coated with the anti-PAR Coating Antibody (1:500 in PBS) overnight. The antibody was removed and the wells were blocked with the Assay Diluent for 2 hours at room temperature. The solution was removed and the wells were washed three times with PBS. The wells were then incubated with the indicated amounts of PAR in Assay Diluent for 1 hour at room temperature with gentle shaking. The wells were then washed three times with 1x wash buffer and incubated with 200 nM sensor protein in ADP-ribosylation Buffer (50 mM Tris-HCl pH 7.5, 0.125 M NaCl, 12.5 mM MgCl2) for 30 minutes at room temperature in the dark and the fluorescence intensity was measured. The intensity from a blank well was used to determine background signal, which was subtracted from the fluorescence signal.

Measurement of in vitro fluorescence changes using spectroscopy.
Purified sensor and control proteins (200 nM) were incubated with the purified proteins or cell extracts in a volume of 100 μL. The samples were incubated for 15 minutes and the fluorescence intensity was measured using a plate reader (CLARIOstar BMG Labtech). Excitation and emission spectra were 488 nm and 530 nm, respectively.

Immunofluorescence and live cell imaging assays
The following methods were used for measuring the PAR levels in live cells using confocal microscopy.
Tracking PAR in live cells. For live cell imaging assays to measure the level of PAR, HeLa cells expressing the fluorescence PAR-T sensors were plated on poly-D-Lysine coated live cell imaging chamber slides (Thermo Fisher, 15411) and cultured in FluoroBrite medium (Thermo Fisher, A1896701) supplemented with 10% FBS (TET tested; Atlanta Biologicals, S103050) and 1% penicillin/streptomycin and 1 mg/mL of Dox. Twenty-four hours later, the cells were pretreated with 20 μM PJ34 for 2 hours followed by treatment with 1 mM H2O2 for the indicated times. Images were acquired using an inverted Zeiss LSM 780 confocal microscope affixed with a 37°C, 5% CO2 incubator.
Image analysis. We used Image J software to subtract background, set thresholds, select the regions of interest (ROIs), and quantify fluorescence intensity. The nuclei were selected using mCherry signal and corresponding intensity of PAR (GFP signal) was quantified. Time series v.3.0 plugin was used to get an average of the ROI intensities of mCherry and GFP.
Generation of 3D cancer spheroids. For generation of cancer spheroids, MCF-7 cells expressing the fluorescence PAR-T sensors were sorted by fluorescence-activated cell sorting (FACS) to obtain a population of cells expressing high levels of mCherry (a marker for nuclei). The 3D cancer spheroids were then generated in Matrigel as described previously (Debnath et al., 2003) with some variations. The bottom of an 8-well chamber slide (Thermo Fisher, 154534) was coated with 90 μL of 100% growth factor reduced Matrigel (Fisher, CB 40230) and incubated at 37°C for 30 minutes to allow the Matrigel to solidify. Then 5,000 cells in 400 μL of complete Fluorobrite medium containing 2% Matrigel was added on top of the Matrigel. The medium was changed every 3 days and the cells were allowed to grow for 2 weeks. Once the cells formed 3D clusters, expression of the PAR-T sensor was induced by treating the cells with 1 mg/mL Dox for 24 hours. For inhibition of PARylation, the cells were treated with 20 μM Olaparib for 24 hours. For the time course experiments, the cancer cell spheroids were generated as described above in a 96-well plate format (Cellvis, P96-1-N). The spheroids were then treated with 20 μM Niraparib and imaged at the indicated times. Images of the 3D spheroids were acquired using an LSM 880 confocal microscope.
Image analysis. Multiple z-stacks were acquired and the Z-stack projections were obtained using Fiji Image J software. The spheroids were selected using the mCherry signal and the intensity of the corresponding of PAR (GFP signal) was quantified. To obtain the intensities of the signals in the outer layer and core of the spheroids, the intensities of whole spheroid and the cores were measured and the following formula was used to calculate the PAR signal from cells within the outer layer: Immunofluorescent staining of cultured cells. 293T cells were seeded into 8-well chambered slides (Thermo Fisher, 154534). Twenty-four hours later, the cells were treated with 20 μM PJ34 for 2 hours followed by H2O2 treatment for 15 minutes. The cells were then washed once with PBS, fixed with 4% paraformaldehyde for 15 minutes at room temperature, and washed three times with PBS. The cells were permeabilized for 5 minutes using Permeabilization Buffer (PBS containing 0.1% Triton X-100), washed three times with PBS, and incubated for 1 hour at room temperature in Blocking Solution (PBS containing 1% BSA, 10% FBS, 0.3 M glycine and 0.1% Tween-20). The fixed cells were incubated with PAR antibody (1:500 Millipore, MABE1031) in PBS overnight at 4ºC. After washing three times with PBS, the cells were incubated with Alexa Fluor 594 goat anti-rabbit IgG (1:500, ThermoFisher, A-21207) in PBS for 1 hour at room temperature. After washing three times with PBS, coverslips were mounted with the VectaShield Antifade Mounting Medium with DAPI (Vector Laboratories, H-1200). All images were acquired using an inverted Zeiss LSM 880 confocal microscope. Fiji Image J software was used to perform the quantification of fluorescence intensities. The ROUT method in GraphPad PRISM was used to identify and remove the outliers.

Luciferase assays
Luciferase assays in live cells were performed to measure the levels of PAR in cells. Cells subjected to dox-inducible PAR-T expression were plated into black bottomed 96-well plates in the presence of 1 mg/mL Dox. Twenty-four hours later, the cells were treated as indicated and luciferase assays were performed using the Nano-Glo Live Cell Assay System (Promega, N2011) for Nanoluciferase measurements or by using the Luciferase assay system (Promega, E1500) for measuring firefly luciferase measurements according to the manufacturer's instructions. For dualluciferase assays, images of PAR-T NanoLuc were first obtained using the Nano-Glo Live Cell Assay System. The medium was then changed to PBS containing 100 μg/mL of D-luciferin (Gold Biotechnology, LUCNA-500) to obtain luminescence images of firefly luciferase. The luminescence images were obtained using IVIS Spectrum with an open filter. The luminescence intensities were quantified either by measuring the ROI intensities using IVIS Spectrum or by spectroscopy using a plate reader (CLARIOstar BMG Labtech).
Time course experiments. To detect the time-dependent changes in PARP1 activation upon UV treatment, MDA-MB-231 cells were plated into black bottomed 96-well plates in the presence of 1 mg/mL Dox. Twenty-four hours later, the cells were treated with 20 μM Niraparib or 20 μM PARG inhibitor in FluoroBrite medium (ThermoFisher) supplemented with 10% FBS for 2 hours. The cells were then treated with UV radiation and then luciferase assays were performed using the Nano-Glo Live Cell Assay System and the luminescence intensities were measured in an interval of 1 minute using the CLARIOstar plate reader. The data was shown as a mean of the ratio of luminescence from PAR-T NanoLuc to that of intact nano luciferase.

Enzyme-linked immunosorbent assays (ELISAs)
ELISAs were performed according to the manufacturer's protocol (Cell Biolabs, XDN-5114). Briefly, the protein binding strip-wells were coated with the anti-PAR Coating Antibody (1:500 in PBS) overnight. The antibody was removed and the wells were blocked with the Assay Diluent for 2 hours at room temperature. The solution was removed and the wells were washed 3 times with PBS. The wells were then incubated with the indicated amounts of PAR in the Assay Diluent for 1 hour at room temperature with gentle shaking. The wells were then washed three times with 1x wash buffer and incubated with anti-PAR Detection Antibody (1:1000 in Assay Diluent) for 1 hour at room temperature with gentle shaking. The wells were washed 3 times with 1x wash buffer and incubated with the Secondary Antibody-HRP conjugate (1:1000 in Assay Diluent) for 1 hour at room temperature with gentle shaking. The wells were washed three times with 1x wash buffer, incubated with the substrate solution. The reaction was stopped with Stop Solution when the color change was observed and the absorbance at 450 nm was measured.

Xenograft experiments
All mouse xenograft experiments were performed in compliance with the Institutional Animal Care and Use Committee (IACUC) at the UT Southwestern Medical Center. Female NOD/SCID/gamma (NSG) mice at 6-8 weeks of age were used. To establish breast cancer xenografts, 10 x 10 6 MDA-MB-231-Luc cells with Dox-inducible expression of PAR-T NanoLuc were injected subcutaneously in into the flank of the mice in 100 μL of 1:1 ratio of PBS and Matrigel (Fisher,CB 40230). The weight of the mice and tumor growth was monitored once per week. Once the tumors were formed (~3 weeks post-tumor cell injection), the mice were randomized to receive 25 mg/kg of Niraparib or PARG inhibitor or vehicle in 4% DMSO, 5% PEG 300, 5% Tween80 in PBS intraperitoneally for 5 consecutive days a week (2 days off) for 3 weeks. The mice were placed on a Dox containing diet (625 mg/kg; Envigo) and injected with 100 μg/g of Dox for 5 days a week, for 3 weeks. Bioluminescence imaging of the xenografts was performed using IVIS Lumina. The mice were anesthetized, treated locally with 5 Gy of radiation, followed by injection in PBS of a combination of 40x dilutions of Nano-Glo live cell imaging (Promega, N2011) and Nano-Glo (Promega, N1110) substrates for acquiring Nanoluciferase luminescence. Luminescence from firefly luciferase was then measured by treating the mice with 150 mg/kg D-Luciferin.
Time course assay. For the time course experiment ± PARG inhibitor, 10 x 10 6 MDA-MB-231-Luc cells with Dox-inducible expression of PAR-T NanoLuc were injected subcutaneously into the flank of the C57/BL6 mice in 100 μL of 1:1 ratio of PBS and Matrigel. The mice were immediately treated with 25 mg/kg of PARG inhibitor or vehicle in 4% DMSO, 5% PEG 300, 5% Tween80 in PBS. Bioluminescence imaging of the xenografts was performed using IVIS Spectrum. The mice were anesthetized, followed by injection in PBS of a combination of 40x dilutions of Nano-Glo live cell imaging substrate and Nano-Glo substrate for acquiring Nano luciferase luminescence 6 hours or 24 hours after the injection of the cells.

Table 1. Nomenclature, composition, and activity of the various PAR-T sensors used in this study.
A summary of the PAR-Tracker sensors generated in this study. The activity of these sensors in in vitro and in cells, and the corresponding figures in which the activities are described, are indicated. The activities are described as low (+), medium (++), and high (+++, ++++). N.D. not determined.

Figure 2. Tracking PAR accumulation in live cells using PAR-T ddGFP.
(A and B) Live cell imaging assay to track PAR formation in response to H2O2 in 293T cells subjected to Dox-induced PAR-T ddGFP expression. The cells were treated with 20 µM PJ34 for 2 hours prior to treatment with 1 mM H2O2 for 15 minutes. The scale bar is 10 µm. Each bar in the graph in (B) represents the mean ± SEM of the relative levels of the fluorescence intensity (n = 3, One-way ANOVA, * p < 0. 01; ns = not significant). (C and D) Live cell tracking of PAR formation in response to H2O2. HeLa cells subjected to PAR-T ddGFP expression were treated with 20 µM PJ34 for 2 hours prior to treatment with 1 mM H2O2 and live cell imaging. The scale bar is 10 µm. Each bar in the graph in (D) represents the mean ± SD of the relative levels of the fluorescence intensity (n = 20 for control and n = 21 for PJ34).

PAR-T ddGFP (WWE-A/WWE-B)
Relative GFP/mCherry  The spheroids were divided into "outer" and "core" sections for quantification as indicated by the white circles.

Relative GFP/mCherry
(Left) Enlargement of the indicated areas from the left panels (yellow, core; pink, outer) as indicated. Each point in the graph in (D) represents the mean ± SEM of the relative levels of PAR-T ddGFP fluorescence intensity normalized to mCherry (n = 5, One-way ANOVA, * p < 0.05 and ** p < 0.01).   Figure 1G) of the time course of in vitro PAR degradation using recombinant ARH3. Each line plot in the graph represents mean ± SEM of relative intensities (n = 3).

PAR-T NanoLuc/Full NanoLuc
(B) Quantitative analysis of Western blot analysis and bioluminescence imaging (shown in Figure  4B) of 231-PAR-T NanoLuc cells treated with 20 µM Niraparib or 20 µM PARG inhibitor for 2 hours prior to UV radiation. Each bar in the graph in represents the mean ± SEM of the relative intensities (n = 3, one-way ANOVA, * p < 0.05, ** p < 0.001 and *** p< 0.0001; ns = not significant).
(C) Measurements of ELISA and fluorescence intensities using 0, 0.625, 1.25 and 2.5 nM concentrations of purified PAR. Each bar in the graph in represents the mean ± SEM of the relative intensities (n = 3, Paired t-test, * p < 0.05, ** p < 0.01 and *** p < 0.001; ns = not significant). (D) Immunofluorescence assay using WWE-Fc to measure PAR formation in response to H2O2 using HeLa cells. The cells were treated with 20 µM PJ34 (versus untreated control, "Un") for 2 hours prior to 15 minutes of treatments with 1 mM H2O2. The images were collected using a confocal microscope.
(E) Quantification of the results in (D). Each bar in the graph represents the mean ± SEM of the relative levels of the fluorescence intensity of PAR normalized to DAPI (n =3 biological replicates with at least 150 cells in total, One-way ANOVA, *** p < 0.0001).
(F) Representation of the dynamic ranges of PAR-T sensors in comparison to other available PAR detection tools as indicated: (a) Western blotting with WWE-Fc versus live cell luciferase assay using PAR-T NanoLuc was performed using UV-induced DNA damage in MDA-MB-231 Luc cells (from panel B); (b) Immunofluorescence with WWE-Fc versus live cell imaging using PAR-T ddGFP was performed using H2O2-mediated PARP1 activation in 293T cells (from panel D); (c) Western blotting with WWE-Fc versus fluorescence assay with PAR-T ddGFP was performed using ARH3 mediated degradation of PAR in vitro (from panel A); (d) ELISA versus fluorescence assay with PAR-T ddGFP was performed using immobilized PAR (from panel C).   Figure 7. Tracking the levels of PAR during adipogenesis. (A and B) Bioluminescence imaging of PAR-T NanoLuc (A) and unsplit NanoLuc (B) in 3T3-L1 cells subjected to adipogenic differentiation for 12 hours or 24 hours. (C and D) Quantification of signals from PAR-T NanoLuc (C) and unsplit NanoLuc (D) during adipogenesis. Each bar in the graph represents the mean ± SEM of the relative levels of the luminescence of NanoLuc (n = 4; t-test, ** p < 0.01 and *** p < 0.001. Comparisons between experimental conditions with the intact NanoLuc are not statistically significant (ns). (B and C) Bioluminescence imaging (B) of xenograft tumors formed using 231-PAR-T NanoLuc cells. The tumors were subjected to 5 Gy IR radiation and treatment with PARP inhibitor (Niraparib) or PARG inhibitor as indicated prior to BLI imaging. Each bar in the graph in (C) represents the mean ± SEM of the relative levels of the ratio of luminescence of NanoLuc to firefly luciferase (n = 6 for the vehicle and PARG inhibitor treatment cohorts and n = 7 for the Niraparib treatment cohort; t-test,* p < 0.05 and ** p < 0.01).  See the separate PowerPoint file containing the videos for these images. The videos will automatically commence and loop when the .pptx file is opened in "presentation" mode.

Figure 2 -figure supplement 2 (video). Tracking the levels of PAR in cells.
Live cell tracking of PAR formation in response to H2O2. Time course images of the live cell assay shown in Figure 2C. HeLa cells subjected to PAR-T ddGFP expression were treated with 20 µM PJ34 for 2 hours prior to treatment with 1 mM H2O2 and live cell imaging. The 15 minute time course of H2O2 treatment was condensed to the time frame show in the video (1 minute sampling interval, 15 frames total). The scale bar is 5 µm.

PAR-T ddGFP mCherry
Click on the frame to access the video control panel. Each bar in the graph in (B) represents the mean ± SEM of the relative levels of the ratio of luminescence of Nano luciferase to firefly luciferase (n = 3, t-test * p < 0.05 and ** p < 0.001).