• Keine Ergebnisse gefunden

1 Theoretical Background

1.3 PAR Synthesising Enzymes and their Function

For a long time, ARTD1 was believed to be the be the only PARylating enzyme. The founding member accounts for 75-90% of cellular PAR formation upon genotoxic stress and has therefore been most extensively studied.[14] It was formerly known as poly(ADP-ribose) polymerase 1 (PARP1) and the term ‘PARP’ for these enzymes is still frequently used in the research literature.[8]

With the help of structural and computational analysis, the other ARTD family members have been identified and can be subdivided into four subfamilies on the basis of their domain structures (Table 1):[24-25] the DNA-dependent ARTDs, the tankyrases, the CCCH ARTDs, the macro-ARTDs, and some members with distinct domain structures, that do not fit into one of these classes. DNA-dependent ARTDs exhibit N-terminal DNA binding domains and their activity is regulated by binding to discontinuous DNA structures. Tankyrases comprises large ankyrin domain repeats mediating protein-protein-interactions. The CCCH ARTDs can bind to RNA with the help of their Cys-Cys-Cys-His containing zinc-fingers and macro-ARTDs contain additional macrodomain folds that help to localise them at sites of PARylation. Not all ARTD family members were found to catalyse PARylation. Indeed, only ARTD1, ARTD2, ARTD5 and ARTD6 were proven to form polymers, whereas the others exhibit MARylation activity or are either catalytically inactive or require further, yet unknown factors to trigger ADP-ribosylation. Derived from structural and computational data, ARTD3 and ARTD4 are predicted to form polymers as well, but this activity has not been confirmed yet.[12, 16]

ARTDs are involved in the regulation of many cellular processes.[6, 16, 31] By binding to DNA strand breaks, ARTDs are involved in DNA repair and recruit different DNA repair factors. As most DNA breaks occur due to genotoxic stress, ARTDs are essential for cell survival and genome maintenance. Through the modification of histones and binding to other nucleosomal proteins as well as by the modulation of RNA, ARTDs regulate chromatin compaction and have severe impact on gene expression.[31] Moreover, ARTD activity has also been found at centrosomes, microtubules and telomeres, which links PARylation with cell division and aging. 

In contrast to these cytoprotective functions, ARTD overactivity can lead to pathophysiological effects, up to the point of cell death by depletion of NAD+ and ATP pools.[32] These effects are linked to a variety of disease state as found in oxidative stress, inflammatory, and metabolic diseases, neurodegenerative disorders, and heart failure.[33]

Moreover, ARTD1 expression is upregulated in various types of human cancers, and this has been linked to resistance of cancer cells to genotoxic treatment.[34-35] Thus, targeting tumour suppressor mechanisms such as DNA repair is an attractive approach in cancer therapy to sensitise tumour cells and to overcome acquired resistance. Some cancers that have mutations in the BRCA1 gene (BReast CAncer 1) rely exclusively on ARTD1-mediated DNA

Theoretical Background

repair and thus inhibiting ARTD1 can kill these cancer cells selectively. In 2014, olaparib was approved as first ARTD1 inhibitor for the monotherapeutic treatment of BRCA-mutated ovarian cancers.[33]

Because PAR synthesising enzymes are in the focus of this project, the DNA-dependent ARTD1 and ARTD2 as well as the tankyrases ARTD5 and ARTD6 are discussed in more detail.

Table 1. Overview of ARTD family members.[12, 16, 31]

Name Alternative Name Subclass Size

(aa) Subcellular

Localisation Enzymatic

Activity Structural Motifs

ARTD1 PARP1

DNA-dependent 1014 Nuclear PARylation,

branching

WGR, zinc-fingers,

BRCT

ARTD2 PARP2

DNA-dependent 570 Nuclear >> Cytosolic PARylation WGR

ARTD3 PARP3

DNA-dependent 540 Nuclear > Cytosolic MARylation WGR

ARTD4 PARP4, vPARP 1724 Cytosolic (vault

particles) > Nuclear MARylation BRCT ARTD5 Tankyrase 1,

TNKS1, PARP5a Tankyrase 1327 Cytosolic >> Nuclear PARylation, OARylation

PARP6 Tankyrase 1166 Cytosolic >> Nuclear PARylation, OARylation

macro-ARTD 854 Cytosolic >> Nuclear no activity

reported Macro-domain

ARTD10 PARP10 1025 Cytosolic >> Nuclear MARylation

ARTD11 PARP11 331 Nuclear and

Cytosolic MARylation WWE

ARTD12 PARP12, ZC3HDC1

CCCH-ARTD 701 Cytosolic (stress

granules) >> Nuclear MARylation zinc-fingers, WWE

ARTD15 PARP16 630 Cytosolic MARylation

ARTD16 PARP8 854 Cytosolic MARylation

ARTD17 PARP6 322 Cytosolic MARylation

ARTD18 TPT1 unknown no activity

reported

1.3.1 ARTD1

The founding member is a nuclear, DNA-dependent ARTD and it was shown that ARTD1 activity is increased up to 500-fold[36] upon binding to short single-stranded DNA, thereby PARylating itself as main target. The process, where ARTD1 serves as its own acceptor, is called automodification.[37] Other important target proteins include tumour suppressor p53 and several histone proteins.[38] ARTD1 fulfils manifold functions inside of cells, most prominently in DNA repair.

ARTD1 is a 113-kDa-protein that displays a characteristic three-domain structure (Figure 3).[24-25, 39] The N-terminal DNA-binding domain consists of two homologous zinc-finger domains (Zn1, Zn2) and a third, distinct zinc-zinc-finger (Zn3). Crystal and NMR structures unravelled that Zn1 is important for binding and recognition of single- and double-stranded DNA lesions, while Zn2 is important for binding of single strand breaks. Zn3 plays a role for ARTD1 activation and interdomain communication. In addition, the DNA-binding domain comprises a nuclear localisation signal (NLS) and a caspase three cleavage site (Cas).[25]

The second, so-called automodification domain contains a BRCT fold. The BRCT protein interaction motif (BRCA1 C-terminus), which is found in several proteins that regulate cell-cycle checkpoints and DNA repair, is rich in lysine-residues that can be auto(ADP-ribos)ylated.

Figure 3. Structure of human ARTD1. (A) Schematic representation of human ARTD1 domains.[39] (B) Crystal and/or NMR structures of the ARTD1 domains in the absence of DNA. Shown are the NMR structures of Zn1 and Zn2 domains (PDB code 2dmj and 2cs2), the NMR structure of the Zn3 domain (PDB code 2jvn), the NMR structure of the BRCT fold (PDB code 2cok), the NMR structure of the WGR domain (PDB code 2cr9), and the crystal structure of the catalytic domain (PDB code 1a26).

Reproduced with permission.[39] (C) A model of the approximate positioning of all domains.

Reproduced with permission.[39] Structure depictions can be made using Pymol (www.pymol.org).

NLS… nuclear localization signal, Cas… Caspase cleavage site, Zn… zinc-finger, BRCT… BRCA1 C-terminus interaction motif, WGR… Trp-Gly-Arg domain, HD… helical subdomain, ART… ADP-ribosyl-transferase subdomain. Copyright © 2013 Published by Elsevier Ltd.

Theoretical Background

A WGR domain is adjacent to the catalytic domain and comprises a well-conserved region of the amino acid sequence Trp-Gly-Arg responsible for interdomain contacts.[24]

The C-terminal catalytic domain is composed of two subdomains, the helical subdomain (HD) and the ADP-ribosyl-transferase subdomain (ART). The HD consists of six ɑ-helices with connecting linkers that have regulating functions. The ART domain comprising a β-α-loop-β-α fold is the most conserved region and shares a high level of homology with bacterial toxins such as the diphtheria toxin. The ART domain is required for the binding, positioning and processing of NAD+.[24, 39]

In the absence of DNA damage, ARTD1 is believed to exist in an extended ‘beads-on-a-string’ architecture, where the catalytic domain exhibits a rigid conformation and a basal ARTD activity. Upon binding to DNA, interdomain contacts are formed, which allow the catalytic domain to become more flexible and thus being more efficient in producing PAR chains.[39] The more ADP-ribose units are added onto ARTD1, the higher the negative charge becomes. At one point, the negative charge becomes repulsive and ARTD1 and DNA form heterodimers, share several common nuclear binding partners and display some redundant functions in DNA repair and maintenance of genomic stability.[6] This redundancy is supported by knockout experiments in mice, where single knock-out mice are viable, but very sensitive towards genotoxic stress, while the double gene knockout is lethal in an embryonic state.[42]

Nevertheless, some evidence occurred that ARTD2 may fulfil some more specialised tasks.

For instance, it was shown that ARTD2 is selectively activated by 5’-phosphorylated DNA strand breaks and it is suggested that its activity is stimulated in response to more specific DNA repair intermediates or at particular repair stages.[43] From a structural point of view, the domain architecture outside the catalytic domain and the WGR domain differs substantially from ARTD1 (Figure 4). Instead of a complex DNA binding domain, ARTD2 contains only a short N-terminal region (NTR). The NTR comprises a nucleolar localisation sequence (NoLS) and a putative nuclear localisation signal (NLS) and was found to be involved in protein-protein interactions.[43] Previous reports[40] identified the NTR as an DNA-binding domain, but newer reports suggest that the NTR has rather contributing functions.[43] The NTR was found to be unstructured and particularly important for the binding of and activation on single stranded DNA strand breaks, whereas the WGR and catalytic domains concurrently mediate the binding to DNA damage sites.[43]

1.3.3 Tankyrases ARTD5 and ARTD6

ARTD5[44] and ARTD6[45-46], which are functionally and structurally similar, are also termed tankyrase 1 and 2. The name is derived from TRF1-interacting ankyrin-related ADP-ribose polymerase, because ARTD5 was first discovered as a regulator of telomeric DNA through its interaction with and modification of telomere repeat-binding factor 1 (TRF1).[6, 24] In contrast to ARTD1 and ARTD2, tankyrases were found to be mainly expressed in the

cytosol. Most protein interaction partners are shared by both tankyrases, which suggest that they have similar functions. Using siRNA knockdowns, it was shown that ARTD5 has essential regulatory function in mitotic segregation, whereas ARTD6 knockout mice exhibited normal telomeres, but a perturbed metabolism due to the smaller size of mice.[47] Some reports associated tankyrase-mediated PARylation activity with ubiquitin ligase activity. This suggests a link towards regulation of protein turnover and signalling pathways. A prominent target is Axin1, which is involved in Wnt signalling and thus being of high interest for cancer research.[24]

The catalytic domain of the tankyrases is located at the C-terminus. In contrast to ARTD1 and ARTD2, the tankyrases do not exhibit an ɑ-helical regulatory subdomain. Typical for the tankyrases are a series of structural modules, the ankyrin repeats (ANK), through which protein interaction is mediated. ARTD5 contains 24 of these repeats, whereas ARTD6 consists of 16 modules. Each ankyrin repeat comprises of a loop-helix-loop-helix-loop structure that is similar to the human protein ankyrin. A sterile-alpha motif (SAM) domain separates the N-terminal ankyrin repeat region and the C-terminal catalytic region of the tankyrases. Its function has not yet been determined, but it is suggested to serve as multimerisation domain. ARTD5 has an additional N-terminal region, which is missing in ARTD6. This region with presumably regulatory function is termed HPS domain, due to its low sequence complexity consisting of repeats of His, Pro and Ser.[6, 24]

Figure 4. Schematic representations of PAR synthesising ARTDs.[6, 39] NLS… nuclear localization signal, Cas… Caspase cleavage site, Zn… zinc-finger, BRCT… BRCA1 C-terminus interaction motif, WGR… Trp-Gly-Arg domain, HD… helical subdomain, ART… ADP-ribosyl-transferase subdomain, NoLS… nucleolar localisation sequence, NTR… N-terminal region, HPS… His, Pro and Ser rich domain, ANK… ankyrin, SAM… sterile-alpha motif.

Theoretical Background

1.3.4 Detection of PARylation

To understand the complex and functional processes of PARylation, it is crucial for scientists to develop tools that can detect and quantify this posttranslational modification.

However, this task remains challenging due to the following reasons.

First, PAR was found to be attached on a wide range of protein sites, such as basic and acidic protein side chains. This variability creates different chemical and enzymatic reactivity and stability of the covalent bonds formed. Furthermore, the polymer itself is rather labile due to the diphosphate backbone, which consists of highly energetic anhydride bonds and is prone to nucleophilic attacks. Moreover, the high negative charge makes the handling and purification of PAR and PARylated proteins challenging. Additionally, the polymer is very heterogeneous and consists of differently sized, linear or branched chains. Last but not least, PARylation underlies a dynamic equilibrium, where chains are attached, elongated or removed with fast kinetics and making it hard to ‘catch’ the polymer.[27]

Remarkable efforts have been undertaken to develop tools and assays for studying determine both ARTD and PARG[48] activity and allows for the in vitro screening of enzyme inhibitors. Although successfully established, special training in dealing with radioactivity is needed and cellular applications are hampered by the lack of a spatially resolved read-out.

Another breakthrough was the development of PAR-specific antibodies by Kawamitsu et al.[49] in 1984 and the establishment of immunocytochemical protocols.[50] Detecting PAR with the help of the 10H-anti-PAR antibody and coupling to other read-outs such as chemical luminescence or fluorescence has evolved as the second standard method in PAR research.

However, this method provides only indirect measures of PARylation and might also give false-positive signals due to unspecific binding.

Moreover, several other methods have been developed to measure PARylation. Besides radioactive labels, the use of ε-NAD+ as a fluorescent analogue has also been investigated.[51] Although it was metabolised by ARTDs, fluorescence quenching was observed after incorporation into the polymer.[52] As a result, formed polymers have to be either digested in order to free the fluorescence signal or additionally immuno-stained with a specific antibody.[53] As this procedure does not offer any advantages over the standard methods, it is not applied.

In contrast to ε-NAD+, 6-biotin-NAD+ was shown to be sufficiently used as ARTD substrate and is now available in a mixture with natural NAD+ in many commercial ADP-ribosylation kits.[54-55] In addition, it was shown to be incorporated into PAR within oxidatively stressed cells.[56] Despite this breakthrough, it is not used by the PAR research community for intracellular PARylation studies. This might be due to the fact, that it bears a long linker and biotin is a rather bulky modification. Therefore, it might not be a frequently used ARTD substrate and decrease ARTD activity, when too high concentrations are applied.

Recently, a new generation of NAD+ analogues have been developed that are equipped with small, terminal alkyne-reporter groups.[57-59] These derivatives are well accepted substrates of ARTD1 and can be further modified with fluorophores and isolation tags via

copper(I)-catalysed click-reactions after incorporation. To the best of the author’s knowledge, cellular applications have not been reported yet.

Next to the development of NAD+ analogues, other approaches for measuring PARylation have been investigated. For instance, the consumption of NAD+ can be determined by either converting NAD+ chemically into a fluorescent product,[60] by applying NAD+ sensors[61] that fluoresce after binding, or by using an colorimetric substrate[62] that releases nitrophenol after incorporation.

Apart from the NAD+, the PAR polymer itself or digested ADP-riboses units can also be chemically modified. By taking advantage of the intrinsic aldehyde function of the ADP-ribose, a fluorescent analogue[63] can be formed or the aldehyde can react with hydrazine[64]- or aminooxy[11]-containing tags such as biotin or fluorophores. However, this reaction is not suitable to detect PAR attached to lysine residues and it will also label all other aldehyde functionalities being present within a biologic context such as abasic sites in DNA.

Apart from antibodies, other PAR and ADP-ribose sensors have been developed. For instance, smaller ADP-ribose binding macrodomains have been identified, which can also be applied for the detection of MARylation and free ADP-riboses.[29] Furthermore, a turn-on split-luciferase sensor was developed that assembles a functional split-luciferase enzyme upon binding to PAR.[65] Another approach uses a supercharged, green fluorescent protein as sensor for PARylation.[66] Here, the fluorescence of the protein is quenched by Förster resonance energy transfer (FRET), while being electrostatically bound to the PAR polymer.

Another powerful method to detect PARylation is mass spectrometry. Due to the fact, that samples get destroyed during measurement, mass spectrometric investigations rather seek to detect protein sites of PARylation, PARylation targets and interaction partners as well as changes in PARylation levels.[27] Here, scientists benefited from the application of NAD+ analogues in order to isolate PARylated proteins and to distinguish signal from background.

Recent progress has been achieved with the help of a chemical genetics approach, where orthogonal ARTD/NAD+ pairs are generated using a ‘bump-hole’-strategy.[67-70] Thus, only the mutated ARTD can use the applied NAD+ analogue as substrate and allows to read out the PARylation target proteins in an enzyme-specific manner. However, the chemical genetics approach is currently limited to in vitro applications and further analysis has to proof, if these approaches faithfully predict in vivo targets.[71]

Most of the presented detection methods can only be applied for the in vitro study of ADP-ribosylation, because they are either too unselective for biological samples or require non-physiological conditions. To date, only few methods exist for the intracellular, dynamic detection of PARylation. As already mentioned, indirect, steady-state cellular imaging can be achieved by immunostaining with PAR antibodies.[50] Moreover, dynamic studies of PARylation processes was achieved by tagging PAR synthesizing or PAR binding enzymes with fluorescent proteins, such as eGFP-ARDT1[72] or eGFP-ARTD2[43] and eGFP-macroH2A,[30] which thus give an indirect measure of PARylation.

Still, no method offers the possibility to follow PARylation directly at the level of the polymer in real-time and dynamically upon extrinsic stimuli like DNA damage induction, but would be of high interest to understand PARylation on a cellular level.[73]

Theoretical Background