Fluorescence Lifetime Imaging of Protein-Specific PARylation in Cells

As most of the functions of PARylation are mediated through covalent and non-covalent binding of proteins, it would be as well of great benefit to study these interactions on a protein-specific level. For this purpose, proximity-based FRET approaches have been very useful. One of the most robust methods to determine FRET is the measurement of fluorescence lifetimes.^[100] Within her PhD thesis, Annette Buntz constructed a so-called fluorescence lifetime imaging microscope (FLIM) being able to resolve spatially fluorescence lifetime changes in the dimension of living cells.^[166] Thus, we opted to implement the new NAD⁺ analogue 11 in combination with eGFP-tagged proteins into this FLIM-FRET technique and to detect ADP-ribosylation in a protein-specific manner as it has already been useful for other posttranslational modifications.^[167-168] All in vitro experiments were performed and optimised by Annette Buntz and are described in detail in her PhD Thesis.^[166] For the final cell-based experiments described in the following, Eva Gwosch and I assisted in sample preparation and execution. The recombinant purified eGFP-tagged ARTD1 was provided by Sascha Beneke, University of Konstanz. The plasmid coding for eGFP-tagged ARTD1 and macroH2A was already available in the group of Prof. Dr. Ferrando-May.

First and for proof-of principle, the automodification of ARTD1 was chosen as an example for the detection of a covalent PAR interaction. For this purpose, ARTD1 was fused to the FRET donor eGFP, while the TMR label present on the NAD⁺ analogue functions as FRET acceptor. This FRET pair has a calculated Förster Radius of 5.8 nm and can effectively be used to detect donor-acceptor distances of 2 to 10 nm. To confirm that the eGFP-tag does not impair ARTD1 activity, wild type ARTD1 and eGFP-ARTD1 were compared in the usual ADP-ribosylation assays (Figure 28) indicating no different in vitro behaviour.

Dye Modified NAD⁺ Analogues

Figure 28. SDS PAGE analysis comparing wild type ARTD1 (wt) with eGFP variant in ADP-ribosylation of histone H1.2 (A) and automodification of ARTD1 (B). Left panel shows Coomassie Blue staining; right panel shows TMR-fluorescence signals. The total concentration of NADs was 1 mM. (A) Reactions were performed with natural NAD⁺ (lane 1 and 4), with a 9:1 mixture of natural NAD⁺ and NAD⁺ analogue 11 (lane 2 and 5), without enzyme (lane 3) or with ARTD inhibitor olaparib (I, lane 6). (B) As a control, the same amount of ARTD1 was loaded (lane 1 and 5). Reactions were performed with natural NAD⁺ (lane 2 and 6), with a 9:1 mixture of natural NAD⁺ and NAD⁺ analogue 11 (lane 3 and 7), without enzyme (lane 4) or with ARTD inhibitor olaparib (I, lane 8).

When eGFP-ARTD1 is modified with TMR-containing ADP-ribose, FRET can be detected by a reduction in the fluorescence lifetime of the eGFP-label, whereas modification with natural NAD⁺ has no effect on the lifetime (Figure 29A). Annette performed a set of in vitro experiments, where consecutive modification of eGFP-ARTD1 with multiple TMR-labelled ADP-ribose units led to a prominent and steady decrease of the eGFP fluorescence lifetime, indicative of increasing FRET. She used either the purified protein or lysates of HEK293T cells expressing eGFP-ARTD1 and monitored PARylation of ARTD1 in real-time. The effect was not observed, when the reaction was performed in the absence of dsDNA or upon addition of ARTD-inhibitor olaparib^[149].

After successful in vitro measurements, cellular imaging was advanced. For this purpose, HeLa cells were transfected with the eGFP-ARTD1 plasmid overnight and transiently expressed the protein the next day. Then, these cells were treated with the delivery mix containing 11 and Pep-1 as described earlier (Chapter 5.4). Unfortunately, this double treatment stressed the HeLa cells and it was difficult to find doubly labelled, viable cells.

However, sufficient numbers of cells could be subjected to laser microirradiation as described earlier (Chapter 5.5) by being fast and preparing multiple cell samples. Laser-induced DNA damage led to the enrichment of both eGFP and TMR fluorescence signals at the irradiated sites, indicating that ARTD1 is recruited and fluorescent PAR is formed (Figure 30A). After two to three minutes after laser damage, the cells were fixed and subsequently analysed on the FLIM microscope (Figure 29B). The fluorescence lifetimes of eGFP-ARTD1 were measured for every image pixel and is displayed via intensity-weighted

pseudo-colours (e.g. pixels with higher intensity appear brighter). At DNA damage sites, the eGFP fluorescence signals are very bright due to ARTD1 recruitment. When no fluorescently labelled NAD⁺ was applied, the eGFP fluorescence lifetime was found to be invariant throughout the nucleus. With 11 being present, regions of high eGFP intensity show a significantly reduced fluorescence lifetime of the donor caused by FRET between eGFP-ARTD1 and the TMR-labelled PAR. This observation is confirmed by a quantitative analysis of fluorescence lifetimes measured in 15 cells (Figure 29C).

Figure 29. Fluorescence lifetime-based read-out of protein-specific PARylation. Left: Detection of covalent ARTD1 automodification. (A) Activation of ARTD1 leads to FRET between eGFP and the TMR-label on PAR, measured by a decrease of the eGFP fluorescence lifetime. (B) HeLa cells expressing eGFP-ARTD1 were treated with 11 or left untreated, microirradiated for DNA damage induction, and fixed. Fluorescence lifetime images of eGFP-ARTD1 are intensity-weighed and presented in pseudo-colours. Scale bars: 7 µm. (C) Quantitative analysis of fluorescence lifetimes from 15 cells transfected with eGFP-ARTD1. Right: Detection of the non-covalent interaction of macroH2A-eGFP with PAR. (D) Activation of ARTD1 leads to PARylation and recruitment of macroH2A. Non-covalent interaction results in FRET between eGFP and the TMR-label on PAR, measured by a decrease of the eGFP fluorescence lifetime. (E) Fluorescence lifetime images of macroH2A-eGFP upon DNA photodamage. (E) Quantitative analysis of fluorescence lifetimes from 15 cells transfected with macroH2A-eGFP. (C), (E) Mean values ± SEM are depicted. Statistical significance was assessed with a Two-Way-ANOVA and Bonferroni posttest. The level of significance is given with n.s. … not significant, *** p < 0.001.

Dye Modified NAD⁺ Analogues

Figure 30. Enrichment of eGFP-fusion proteins and TMR fluorescence signals at sites of microirradiation. HeLa cells expressing eGFP-ARTD1 were loaded with 11 and microirradiated as described. Cells were fixed 2-3 min after irradiation with 4% PFA and imaged with a confocal fluorescence microscope. (A) eGFP-ARTD1 is localised in the nucleus. Strong enhancement of both eGFP and TMR signals at laser-irradiated tracks are observed. (B) MacroH2A-eGFP is equally distributed in the cell. Strong enhancement at DNA damage sites can be observed in both fluorescence channels. Scale bars:

7 µm.

Encouraged by these results, it was investigated next, if this approach also enables the detection of non-covalent interactions of PAR with proteins. Methods which have been employed to identify and characterise specific non-covalent PAR-binding protein domains^[169]

cannot be applied in the context of intact cells. To test the hypothesis, the macroH2A-eGFP fusion protein was selected as FRET donor.^[30] MacroH2A is a highly conserved protein domain known to bind ADP-ribose in a non-covalent manner,^[29-30] upon which FRET can be detected (Figure 29D). Thus, HeLa cells expressing macroH2A-eGFP were treated with 11, irradiated, and subjected to FLIM-FRET analysis as described above (Figure 29E). Also here, the macroH2A-eGFP fluorescence lifetimes did not differ between treated and untreated cells in the area surrounding the irradiated track. Additionally, the eGFP fluorescence lifetime at the DNA damage site was only significantly reduced in cells containing 11 (Figure 29F). As a conclusion, also non-covalent interaction of macroH2A-eGFP with fluorescently labelled PAR can readily be visualised by FLIM-FRET microscopy.