Extracellular NAD+ enhances PARP-dependent DNA repair capacity independently of CD73 activity

Changes in nicotinamide adenine dinucleotide (NAD+) levels that compromise mitochondrial function trigger release of DNA damaging reactive oxygen species. NAD+ levels also affect DNA repair capacity as NAD+ is a substrate for PARP-enzymes (mono/poly-ADP-ribosylation) and sirtuins (deacetylation). The ecto-5′-nucleotidase CD73, an ectoenzyme highly expressed in cancer, is suggested to regulate intracellular NAD+ levels by processing NAD+ and its bio-precursor, nicotinamide mononucleotide (NMN), from tumor microenvironments, thereby enhancing tumor DNA repair capacity and chemotherapy resistance. We therefore investigated whether expression of CD73 impacts intracellular NAD+ content and NAD+-dependent DNA repair capacity. Reduced intracellular NAD+ levels suppressed recruitment of the DNA repair protein XRCC1 to sites of genomic DNA damage and impacted the amount of accumulated DNA damage. Further, decreased NAD+ reduced the capacity to repair DNA damage induced by DNA alkylating agents. Overall, reversal of these outcomes through NAD+ or NMN supplementation was independent of CD73. In opposition to its proposed role in extracellular NAD+ bioprocessing, we found that recombinant human CD73 only poorly processes NMN but not NAD+. A positive correlation between CD73 expression and intracellular NAD+ content could not be made as CD73 knockout human cells were efficient in generating intracellular NAD+ when supplemented with NAD+ or NMN.


Supplementary Methods
Cell Line Authentication. LN429, T98G, MDA-MB-231 and MCF-7 cell lines were prepared and sent for authentication to Genetica DNA Laboratories -a LabCorp brand.
Briefly, 75% confluent dishes of each of the tested cells were washed with PBS and trypsynized in order to prepare single cell suspensions. Next, cells were spun down and subsequently the pellets were washed twice in PBS to completely removed traces of trypsin. Cell pellets in 100 µl PBS were snap-frozen and shipped to Genetica on dry ice.
A Short Tandem Repeat (STR, PowerPlex16HS -HUMAN specific; includes a mouse marker for detection of mouse DNA) comparative analysis was performed and a Database Profile together with peak data were obtained (see Supplementary Table S2).

Expression and purification of HiNadN. Escherichia coli BL21(DE3) cells (Stratagene)
were transformed with the expression vector for HiNadN (pET25b-HiNadN), the bacteria were inoculated in 1 liter of 2XTY medium supplemented with 50 μg/ml ampicillin and incubated for 4 h at 30°C under vigorous shaking. The absorbance was constantly monitored until reaching a value of 0.6, the growing cells were then supplemented with 0.3mM IPTG (isopropyl β-D-thiogalactopyranoside) to induce gene expression, and the bacterial culture was further incubated for 3 h at 30°C. The bacterial cells were then collected by centrifugation, washed 1X with PBS and frozen at -20°C. To purify the enzyme, the bacterial pellet from 1 liter of HiNadN expressing culture was resuspended in 50 ml of lysis buffer [10mM KH2PO4 pH6.8, 1000 units of Benzonase ® nuclease and an EDTA-free protease inhibitor cocktail], subjected to two cycles at 1.6 kBar in a refrigerated mechanical disruption system (Basic Z apparatus, Constant Systems). Next, the pellet and supernatant were separated by centrifugation. (NH4)2SO4 was added to the clarified lysate to 30% saturation and the resulting precipitate was removed by centrifugation at 10,000xg for 30 min at 4°C. The soluble fraction was dialyzed overnight at 4°C against a solution containing 10 mM KH2PO4, pH 6.8 and loaded onto a ceramic hydroxyapatite column (CHP) equilibrated with the same solution; elution of the proteins was performed by applying a linear gradient from 10 to 500mM of a KH2PO4 buffer pH6.8. Pure fractions, as determined by standard SDS/PAGE analysis, were pooled, dialyzed against a solution containing 5mM KH2PO4 pH 6.8, 100mM Tris/HCl pH8.5 and 10mM MgCl2, and concentrated by ultrafiltration using a 30000 MWCO (molecular weight cut-off) disposable Vivaspin device (Vivascience) to a final protein concentration of 10 mg/ml, as determined by a Bradford assay (Sigma). Aliquots of highly homogeneous HiNadN were used for enzymatic investigations and stored at -20°C. 31 P HiNaDN Assay. Purified, recombinant HiNadN (10.5 µg) was added to a freshly prepared 5mM solution of NAD + in buffer (100mM Tris buffer/HCl pH8.5; 10 mmol MgCl2; 400µl), diluted to a final volume of 500µl with additional buffer. The sample spectra were performed using decoupled 31 P NMR on a Bruker 400 MHz spectrometer with 126 scans at ambient room temperature. Data are presented as a bargraph plot. Micro-irradiation and fluorescence imaging were performed with a Nikon A1rsi laser scanning confocal microscope (Nikon Instruments), and analysis was performed with FIJI.

Analysis of NAD + catabolic activity of commercial preparations of CD73 and of
HiNadN. We have evaluated commercially available mammalian CD73 protein preparations using phosphorus ( 31 P) NMR in order to test their activity towards NAD + , NMN and AMP. Unfortunately, each enzyme preparation was devoid of any enzymatic activity for all substrates, as demonstrated in Supplementary Fig. S3A-D. In addition, we evaluated the enzymatic activity of HiNadN, the Haemophilus influenza orthologue of the human CD73 1 , by phosphorus NMR ( 31 P-NMR). It is important to emphasize that 31 P-NMR analysis provides accurate quantification of pyrophosphatase and C-5 nucleotidase enzymatic activities of HiNadN on NAD + , to its hydrolytic products: adenosine monophosphate (AMP), β-nictotinamide mononucleotide (NMN) and inorganic phosphate without additional sample manipulation. Based on the study by Garavaglia et al. 1 , it was anticipated that NAD + would be cleaved to AMP and NMN through the pyrophosphatase activity of HiNadN, then further to nicotinamide riboside (NR), adenosine and inorganic phosphate via C-5 nucleotidase activity as outlined in Fig. 2A. However, such progression of events was not observed. At 240 min after the start of the reaction, NAD + was completely consumed but the final reaction products included both NMN (3.4 ppm) and phosphate (2.4 ppm), at matching intensity of integrals ( Supplementary Fig. S4A &   S4B). As such, it could be deduced that AMP had been completely consumed to form adenosine and inorganic phosphate, whereas, NMN had been left untouched by the C-5 nucleotidase activity of the HiNadN enzyme ( Supplementary Fig. S4B). Additional spiking of the 31 P-NMR sample with an excess of pure NMN further confirmed the identity of the peak to be NMN, at 3.4 ppm (Supplementary Fig. S4C). The same 31 P-NMR experiment was repeated using HiNadN in the presence of either NMN or AMP. After 12 h (or even longer) incubation at 37˚C, no hydrolysis (C-5 nucleotidase activity) could be observed for NMN, whereas AMP was rapidly consumed to form inorganic phosphate and adenosine ( Supplementary Fig. S4D). These results suggest that NMN is not a substrate for the enzyme HiNadN and given that it is an orthologue of the human CD73, one can predict that NMN is not likely a substrate for the human enzyme.

Laser micro-irradiation: recruitment of XRCC1 and modulation by PARP1
inhibition. MCF-7/CD73-KO and the control MCF-7/Cas9 expressing cells were cultured in different media conditions (normal FBS, heat-inactivated FBS, and serum-free) and treated with the PARP1 inhibitor ABT888. Cells were irradiated with a 355nm laser and recruitment of XRCC1-mCherry was analyzed. Relative recruitment intensities for each damaged cell were calculated (max fluorescence intensity/pre-damage fluorescence intensity), and compared for each cell line by Two-Way ANOVA (inhibitor treatment was the row factor, and media condition was the column factor). Here we show that media condition does not modify the effect of PARP1 inhibition on recruitment and that PARP1 inhibition completely blocks XRCC1-mCherry recruitment to the site of laser-induced DNA damage ( Supplementary Fig. S7B,C).

Supplementary Data -Full Blot Images
As per journal policy, all gels/blots as shown are from the same gel. If cropped from different parts of the same gel, or from different gels, fields, or exposures then that is indicated. In the figures, cropped gels/blots are displayed. Full-length gels and blots are shown here for the indicated figures. Supplementary Table S1. Cell Lines. Human cell lines developed and used in this study.