Fluorescently Labelled ATP Analogues for Direct Monitoring of Ubiquitin Activation

Abstract Simple and robust assays to monitor enzymatic ATP cleavage with high efficiency in real‐time are scarce. To address this shortcoming, we developed fluorescently labelled adenosine tri‐, tetra‐ and pentaphosphate analogues of ATP. The novel ATP analogues bear — in contrast to earlier reports — only a single acridone‐based dye at the terminal phosphate group. The dye's fluorescence is quenched by the adenine component of the ATP analogue and is restored upon cleavage of the phosphate chain and dissociation of the dye from the adenosine moiety. Thereby the activity of ATP‐cleaving enzymes can be followed in real‐time. We demonstrate this proficiency for ubiquitin activation by the ubiquitin‐activating enzymes UBA1 and UBA6 which represents the first step in an enzymatic cascade leading to the covalent attachment of ubiquitin to substrate proteins, a process that is highly conserved from yeast to humans. We found that the efficiency to serve as cofactor for UBA1/UBA6 very much depends on the length of the phosphate chain of the ATP analogue: triphosphates are used poorly while pentaphosphates are most efficiently processed. Notably, the novel pentaphosphate‐harbouring ATP analogue supersedes the efficiency of recently reported dual‐dye labelled analogues and thus, is a promising candidate for broad applications.


Introduction
Ubiquitination -t he covalent modification of proteins by the 76 aminoa cid protein ubiquitin (Ub)-i saubiquitous protein modification with fundamentalr oles in numerous cellular processes including protein degradation,D NA damage repair,c ell cycle regulation and gene expression. [1,2] Malfunction of the ubiquitination system contributes to ab road variety of human diseases like cancer, diabeteso rn eurodegeneratived isorders. [3,4] For the attachment of Ub to substrate proteins,t he consecutive action of at least three classes of enzymes is needed. In the first step, Ub is activatedb yaubiquitin-activating enzyme (E1) at the consumption of ATP. Thereby Ub is ade-nylated and then transferred to the active-site-cysteine of the E1 to form at hioester with the C-terminal glycine carboxylate of Ub [5] (Figure 1A). By transthiolation, Ub is transferred to a ubiquitin-conjugating enzyme (E2). Finally,b yt he aid of ubiquitin-protein ligases( E3), Ub is covalently connected to the targetp rotein by forminga ni sopeptide bond with the eamino group of alysine residue.
In humans, two E1s (UBA1 and UBA6) for Ub are known. [6,7] Quantifying and following E1 activity in real-time is of great importance to study for example, effectors, but meanst od o Figure 1. A) Ubiquitin(Ub) activation by E1 (UBA1) with ATP. Adenylated Ub is loaded on UBA1 by the formationo fat hioester.B)Concept of mono-labelled ATPa nalogues as UBA 1s ensors that are investigated in this study: Wheni nc lose proximity to the nucleobase adenine, the fluorescent dye is quenched as ar esult of photoinduced electrontransfer (PET). After enzymatic release of the phosphatechain, the fluorescence is restored. so are sparse. Available E1 activity assays include SDS-PAGE analysiso fE 1/E2-Ub thioester conjugates by Western Blot, [8] radio-labelling of the involved proteins with 125 [I] [9] or 32 [P], [10] FRET between Ub and E1 [11] or enzyme-coupled spectrophotometric assays for phosphate determination. [12] Drawbackso f these assays are that they are either laborious, do not allow continuousr ead-out or are dependento na dditional enzymes of the downstream cascade.T of ill this gap, we have recently developed at ime-resolved ATPase sensor( TRASE) based on a Fçrster resonance energy transfer (FRET) pair embedded within the ATPs caffold to continuously monitor ATP-dependent enzymes. [13,14] This ATPF RET probe sensoru ses two fluorescent dyes, ad onor and an acceptor dye, that are attached to the terminal phosphate group and to adenine, respectively.B ye mploying the TRASE assay,w ef ound that g-modified triphosphates are poorly accepted by UBA1 whereas d-modified tetraphosphates turned out to be betters ubstrates. [15] In order to improvet he substrate properties of ATP-based E1 sensors, we synthesized and investigated new ATPa nalogues that contain only one dye and differ in the length of the phosphate chain for their propensity to visualize E1 activity.I ndeed, we identified af luorescentd ye with an acridone core structure that is efficiently quenched by the canonical nucleobaseadenine.
While triphosphatesw ere poorly processed, we found that also in this case acceptance is improved by elongation of the phosphate chain, that is, tetraphosphates are better accepted than triphosphates. In fact, elongation of the phosphate chain to pentaphosphate resulted in even bettera nalogues enabling us to follow the activation of Ub by both enzymes, UBA1 and UBA6, in real-time.

Results and Discussion
Recently,w ed eveloped g-modified ATPa naloguest hat are suitable as model compounds for monitoring enzymatic ATP consumption by fluorescence lifetime readout. Compared to other probes, these compounds contain only one instead of two fluorophores, which promises better enzymatic acceptance. Fluorescence lifetime changes between the intact and the cleaved ATPa naloguesa re caused by the quenching of fluorescenceb yt he nucleobase adenine [16] ( Figure 1B). The quenching is caused by photoinduced electron transfer (PET). In PET,t he efficiencyf or electron transfer rates and thus for quenching can be estimated by using the Rehm-Weller equation and/or by comparing the involved highest occupied molecular orbital (HOMO) energyl evels. [17][18][19] The PET process starts with excitation of the acceptor chromophoreb yi rradiation (step 1i nF igure 2). This promotes an electron into the lowest unoccupied molecular orbital (LUMO) of the acceptor. In case the HOMO of an eighbouring donor molecule in the immediate vicinity is higheri ne nergy than the HOMO of the acceptorm olecule, an electron is transferredf rom donor to acceptor (step 2) and fluorescencei sq uenched. The cycle is closed by transfer of an electron from the acceptorL UMO to the donor HOMO.T hisp rocess can only take place for distances between acceptor and donor on an anometre scale, that is, in the intact ATPa nalogue. Enzymatic cleavage of the phosphate chain, however,l eads to an immediate separation of the dye-adenine pair by diffusion such that fluorescenceo f the dye is restored.
PET-based quenchers that have previously been employed in biological assays are guanine (À5.33 eV;c alculated from its redox potential) [20] and tryptophan (À4.90 eV;c alculated from its redox potential), [21] both of which possess high HOMO energies. [22][23][24][25][26] By contrast, duet oi ts lower HOMO energy (À5.78 AE 0.01 eV), [16] adenine has only rarely been used as quencher.T o develop new suitable dye-adenine pairs,w eu sed photoelectron spectroscopy in air (PESA) to determine HOMO energy levels for hydrophilicd yes. With this approach, we recently were able to investigate different fluorophores like rhodamines and BODIPYsf rom dry thin films that suited for fluorescence lifetimer eadout when attached to ATP. [16] While we could demonstrate enzymatic processing andr eal-time detection of these compounds, their main drawback was the relatively short lifetime of these dyes. Thism adet he detection of lifetime changes especially under biologically relevant conditions difficult.
In order to overcome this shortcoming, we now investigated fluorescent dyes with long fluorescent lifetimes. Acridone and quinacridone as well as their derivatives are interesting candidates for our purpose, especially because they are extremely photostable, [27] have al ong fluorescencel ifetimeo f1 4nsa nd 22 ns, respectively,a nd show no spectral pH dependency in the biological relevant range from pH 5-9. [28] However,o nly few examples are reported where in particular acridoned erivatives under the name Puretime1 4( PT14) were used for fluorescence lifetime-based biological assays. [29][30][31][32] For functionalization and attachment of acridoneo rq uinacridone to ATP, we followed ak nown synthesis strategy.I nitially,w ec reated water-soluble compounds from the organic pigments by introducing sulfonica cid residues( Scheme 1). [28] We expected that sulfonation would also lower the energy of the frontier orbitals that in turn would result in efficient quenching of the dye when being in close proximity to adenine.
With the water-soluble compounds 2.1, 2.2 and 4 in hand, we performed PESA to obtain insights into the HOMO energies in air.T ob ee fficientlyq uenched by adenine, the HOMO energies of the fluorescent dyes should be lower than that of ATP (À5.78 AE 0.01 eV) [16] (Figure 3).
As af irst proof of concept, we determined the fluorescence lifetimes of Ap 3 -Dye (5)i nt he intact and cleaved state. As a model enzyme for cleavage, we used the ATPh ydrolysing phosphodiesteraseIfrom Crotalus adamanteus (Snake Venom Phosphodiesterase, SVPD). Before cleavage, we measured a fluorescencel ifetimev alue of 8.71 AE 0.10 ns. By additiono f SVPD,t he triphosphate is cleaved and the quencher adenine is released as AMP and separated from the fluorophore. Here, we measured avalue of 15.64 AE 0.25 ns, corresponding to an absolute lifetime change of 6.93 AE 0.27 ns. With typical experimental errors in fluorescencel ifetimed etermination in the range of 100 ps, this differencei sw ell suited to use lifetimes for quantification of ATPa naloguec leavage. Additionala bsorptiona nd emission spectra recorded before and after cleavage with SVPD did not revealaspectral shift assuming al ow interaction between the sulfonated dyea nd the nucleobase adenine (data are shown in the SupportingI nformation).
Findings from other groups showed that modified pentaphosphates are even better substrates for nucleic acid polymerases than shorter congeners [35][36][37] which motivated us to additionally synthesize an e-modified alkylated pentaphosphate. Describeds ynthesis routes of alkylated pentaphosphates that follow P V -N activation and subsequents ubstitution often lead to unwanted side-products. [35,[38][39][40] Therefore, we decided to explore ak nown iterative polyphosphorylation approach strategy. [41] The reaction is conducted in several steps without purification in one pot. Only modified monophos-  phates are neededa ss tarting materials, which can be easily obtained in high yields or are even commercially available. Scheme 3s hows an overview of the synthesis of Ap 5 -Dye. Diisopropylamino dichlorophosphine (7)i sr eactedf irst with pyrophosphate to form ac yclic pyrophosphoryl P-amidite (8) which is coupled to 6-azidohexyl phosphate by subsequent oxidation to form 1-(6-azido)hexyl phosphoryl cyclotriphosphate (9). The cyclic trimetaphosphate is then opened by adding adenosine monophosphate as nucleophile and MgCl 2 to yield e-O-6-azidohexyl)-adenosine-O5'-pentaphosphate (10)i naonepot reaction with an overall yield of 11 %. Reduction of the azide followed by coupling of the activated S 2 acridone using its NHS ester yielded the desired compound (Ap 5 -Dye (11), 28 %o ver two steps,S cheme 3). For the Ap 5 -Dye 11,t he fluorescencel ifetime of 9.86 AE 0.19 ns is slightly higher than that observedf or the tri-and tetraphosphates 5 and 6 (8.71 AE 0.10 ns and 8.74 AE 0.03 ns, respectively). The same fluorescence lifetimeo ft he free dye was measured when 10 was treated with SVPD as for (5 + 6)( 15.64 AE 0.25 ns). The reduced quenching for longer distances between dye and quencher pairs is in accordance to the literature where efficient PET-quenching takes place on as ub-nanometre scale. [42] With all three compounds in hand, we tested them towards their performance in an E6AP auto-ubiquitination assay that was previously shown to be well-suited to qualitatively evaluate the acceptance of ATPa nalogues by UBA1 (Figure 4). [15,43] All reaction mixtures were pre-treated with recombinant shrimp alkaline phosphatase( rSAP) which dephosphorylates all terminally bound phosphate groups,f or example, of ATPt oi ts nucleoside, [44] while it leaves terminally modified nucleotides unaffected (like for Ap n -Dye). This ensures that the observed activity is due to the Ap n -Dye analogue and does not originate from potential contaminations of natural ATP. After preincubation, rSAP was inactivated by heating the mixture to 65 8Cf or 5minutes.Inthe first lane in Figure 4A and B, reactions are de-picted in which ATPi sh ydrolysed by rSAP and, thus, autoubiquitination of E6AP is not observed. As positivec ontrol,a n additional amount of ATPw as added after inactivation of rSAP which results in the formation of poly-ubiquitinated forms of E6AP (E6AP-Ub n )a nd ac oncurrent decrease of free Ub and non-modified E6AP,r espectively.T he results obtained clearly indicatet hat Ap 3 -Dye (5)i sp oorly processedb yU BA1, while the extension of the phosphoanhydridec hain increasesa ctivity.I nf act, Ap 5 -Dye (11)w as found to be an excellent substrate for UBA1, superior also to Ap 4 -Dye (6).
Similar to UBA1, UBA6 is knownt os upport E6AP autoubiquitination.  Hence, to determine the ability of UBA6 to employ our ATPa nalogues, we also followed the formation of polyubiquitinated E6AP.A sb efore,r SAPw as used to remove terminally bound phosphates and an additional amount of ATP was added as positive control after heat inactivation of rSAP, which resulted in the formation of polyubiquitinated E6AP ( Figure 4B). Within 90 minutes, formation of E6AP-Ub n was neither observedw ith Ap 3 -Dye (5)n or with Ap 4 -Dye (6)w hereas the pentaphosphate analogue (11)w as accepted as substrate. However,t he efficiency appeared to be lower than for UBA1.
Next, we investigated whether real-time experiments are feasible by using UBA1 as well as UBA6 ( Figure S2.1 and S2.2). As tandard multiwell plate reader was used for this purpose and the increasei ni ntensity at 450 nm of S 2 acridone was followed.T he same picture as already seen in the E6AP autoubiquitination assay described above was observed. With Ap 3dye (5)n oi ncrease in fluorescencei ntensity and thus no activated Ub was detected whereas the tetraphosphate and the pentaphosphate are processed by UBA1 and UBA6. Again, the pentaphosphate 11 shows its superior properties over Ap 4 -Dye (6). With UBA1 it is able to adenylate Ub about 2t imes faster ( Figure S2.1) and with UBA6 even 6times faster ( Figure S2.2).
Encouraged by these positive results, we determined the velocity of the UBA1 reaction depending on the Ub concentration. Additionally,w ec ompared the herein reported analogues with the best earlierr eported doubly labelledA TP analogue (Cy5-Ap 4 -Cy3) (for structure see Supporting Information). Interestingly, at high Ub concentration we observed for all three compounds ar eaction inhibition (see Supporting Information Figure S2.3-S2.5), while at low substrate concentration,M ichaelis-Menten kinetics were observed. Ap ossible explanation for this observation is that the UBA1-SH bindings ite is non-covalently occupied by as econd Ub upon increasingU bc oncentration, thus, inhibiting the transfer of adenylated Ub to the active-site cysteine of UBA1 by forming at hioester with the Cterminal glycinec arboxylate of Ub. Therefore, we fitted our data ( Figure 5) to ak inetic model describings uch am ode of inhibition [46] [Eq. 1]: The high affinity binding site is described by K M ,w hereas the inhibitory site, which is in general markedly lower in affinity,isdescribed by K i .Wefound for all analoguesaK i of approximately 50 mm and a K M of (4.1 AE 0.5) mm forA p 5 -Dye (11) which is more than 7t imes lower than the K M for Ap 4 -Dye (6) (29.1 AE 3.0) mm and almost 20 times lower than the doubly labelled Cy5-Ap 4 -Cy3 (76.1 AE 6.7) mm.H owever,t he K M for Ub with natural ATPi seven lower (0.2 mm)s uggesting that our analogues may somewhat interfere with Ub binding. [47] Finally,w em easuredt he activation of Ub by UBA1 and UBA6 under conditions, where both thioesters (i.e. UBA1~Ub and UBA6~Ub) are unloaded. To do so, we added UbcH5b and E6AP in the same concentration as for the SDS-PAGE experiment ( Figure 4). As shown in Figure 6, again neither UBA1 nor UBA6 are able to activate Ub with Ap 3 -dye (5)a sc ofactor, while elongation of the phosphate chain rescues activity as already seen in the SDS-PAGE analysis. Ap 4 -Dye (6)a nd Ap 5 -Dye (11)a re both linearly processed which shows once more the suitability of our setup. Moreover,t he extension of the tetraphosphate chain to Ap 5 -Dye (11)i ncreases acceptance of both UBA1 and UBA6 significantly.W hen all cognate enzymes are present,U BA1 activates Ub approximately 2.9 times faster with Ap 5 -Dye (11)a sA TP source compared to Ap 4 -Dye (6). This tendency is also seen with UBA6 where the activation with Ap 5 -Dye (11)i se ven 4.9 times faster than with Ap 4 -Dye (6).

Conclusions
In conclusion, we developed and exploredn ovel fluorescently labelleda denosine tri-, tetra-, and pentaphosphates. The ATP analogues bear as ingle acridone-based dye at the terminus of the phosphate chain. Most importantly,t he dye's fluorescence is quenched by the adenine residue of the ATPa nalogue.F luorescence is restored upon cleavage of the phosphate chain and dissociation of the dye from the adenosine moiety.T hereby the activity of ATPc leaving enzymes can be followed.
In comparison to our earliera pproaches, in this approach only one dye modification is appendedt ot he ATPa nalogue. This has several advantages.O bviously,t he synthesis towards the probes is simplified making these sensors more readily available. Another advantage is the absence of any modification at the nucleobase therebyr endering the analogues to be superiorlyp rocessed by the enzymes investigated here. This might be due to the fact that upon usage by the adenylateforminge nzymes investigated here, the formed reactive adenylated ubiquitin species is identical to that with natural ATP.
By elongation of the phosphate chain, we were able to increase acceptance and reactionv elocity significantly and demonstrated this for UBA1 and UBA6 that accept Ap 5 -Dye (11) Figure 5. Steady-state kineticsofUBA1 activatedUbasafunction of Ub concentration. The fittingc urves represent the nonlinearleast squares best fit to the described equation. K i = 50 mm. best. We could also show that the herein presented analogues can be readily used in real-time assays to follow Ub activation by UBA1 and UBA6. Notably,u sing the developed disulfonated acridone, the read-out can be both fluorescence intensity and lifetime. These characteristics make the herein developed ATP analogues versatilely applicable for future uses, for example, in the high-throughput screening for effectors of E1 enzymes.