Fluorescence of the tri-cyclic adenine and isoguanine derivatives and their ribosides: possible analytical applications*

Fluorescent tri-cyclic purine analogs, derivatives of isoguanine and adenine, were examined as potential sub- strates of purine-nucleoside phosphorylase. It was found previously that etheno- derivatives of both compounds are ribosylated in phosphate-free media, but ribosyla- tion places in some instances differ from purine N9. New ribosides are examined as potential substrates of human blood PNP and indicators of this enzyme. Of these, N 6 -riboside of 1,N 6 -etheno-adenine was found the most promising.


INTRODUCTION
Tri-cyclic analogs of the canonical purines are frequently used as fluorescent probes in enzymological research. The best known example is 1,N 6 -etheno-adenosine (εAdo) and analogs (Leonard, 1984;Leonard, 1985), but other fluorescent derivatives are also known (Virta et al., 2004;Wang et al., 2017). Tri-cyclic analogs and their ribosides are characterized by moderate biological activity, but some of them reveal promising anti-viral properties (Janz-Wechmann et al., 2015). They are known to react with many enzymes of purine metabolism (Leonard, 1984), and are important intermediates in the process of chemical mutagenesis induced by vinyl chloride (Singer & Kuśmierek, 1982;Chatterjee & Walker, 2017).
Scheme 2. Structure of artificial PNP substrates examined in this work (R, ribose).
(1) N 6 -β-D-ribosyl-1,N 6 -ethenoadenine, (2) N 6 -β-D-ribosyl-1,N 6ethenoisoguanine, and products of their phosphorolysis: (3) 1,N 6ethenoadenine and (4) 1,N 6 -ethenoisoguanine. Note that the purine numbering is applied. not necessarily identical with those produced chemically, with ribosylation sites varying for different PNP forms. The purpose of the present paper is examination of substrate properties of new fluorescent ribosides toward the human blood enzyme, and their analytical potential in clinical investigations. We have previously identified two such substrates, isomeric ribosides of 2,6-diamino-8-azapurine (Wierzchowski et al., 2014), and now extend our work to include tri-cyclic ribosides with different spectral characteristics.
Human PNP, as well as other mammalian forms of this enzyme, belongs to the second (trimeric) class within the broad family of PNP (Bzowska et al., 2000), and its substrate specificity is different than that of the hexameric (bacterial) forms. In particular, the trimeric forms of PNP are inactive towards adenosine and some derivatives (Bzowska et al., 2000;Yehia et al., 2017). We therefore considered only those ribosides, which were previously shown to be substrates for the calf enzyme (Stachelska-Wierzchowska et al., 2018, 2019. We have chosen some non-canonical ribosides of etheno-adenine and etheno-isoguanosine (Scheme 2).
Blood samples were obtained as leftovers from glucometric measurements of one of the authors (J.W.). 10 µL blood samples were lysed in 0.5 mL of 2 mM phosphate buffer, pH 7, containing 0.5 mM EDTA and an aliquot (~0.2 mM) of dithiothreitol. The hemolysates were kept at 5°C.

RESULTS AND DISCUSSION
It has been shown previously, that enzymatic ribosylation of some nucleobase analogs with PNP as a biocatalyst leads to non-typical ribosides, with ribose moiety attached not necessarily to purine N9, but also to other nitrogen atoms (Stachelska-Wierzchowska et al., 2013;Stachelska-Wierzchowska et al., 2016;Stachelska-Wierzchowska et al., 2018;Stachelska-Wierzchowska et al., 2019). In particular, while the E. coli PNP directs the ribosyl group predominantly to the N9 of ethenoadenine, the calf enzyme leads to almost exclusively N 6riboside (Stachelska-Wierzchowska et al., 2018). In 1,N 6ethenoisoguanine, the situation is even more complex, since typically mixtures of various ribosides are produced The most interesting, from analytical point of view, were those ribosides which were generated by calf spleen PNP, that is, N 6 -β-d-ribosides. We have shown that these ribosides (see Scheme II, above) were also excellent substrates for the E. coli PNP, and phosphorolytic    N 6 -β-d-ribosyl-1,N 6 -etheno-adenine (1) with blood lysate as a catalyst, in the phosphate buffer, pH 7, at 25°C. Experimental conditions were exactly as in Fig. 2, above. Final curves (points) were obtained by spiking blood sample with the purified E. coli PNP. Black curves refer to purified substrate (1) spectra. Time intervals were 5 minutes for UV absorption, 10 minutes for fluorescence. Fluorescence minimum at 410 nm is due to light re-absorption by hemoglobin (Slater band). Note that the isosbestic and isoemissive points are strictly maintained. reactions were easily followed by UV absorption of fluorescence spectroscopy (Figs. 2 and 3). The new ribosides are fairly stable, and could be stored for months in stock solutions at -5°C (not shown).
Human blood is particularly rich in PNP activity, which is located mainly in erythrocytes (Bzowska et al., 2000). We have shown previously that some phosphorolytic reactions are easily observed spectrally or fluorimetrically using 1000-fold diluted whole blood lysates as a catalyst in ca. 50 mM phosphate buffer (Wierzchowski et al., 2002;Wierzchowski et al., 2014). We have therefore examined the new ribosides as potential substrates for the human PNP, using the same methodology (except optical pathlength of the cuvettes, reduced now to 4 mm).
Experiments with 1000-fold diluted blood lysates has shown that only one of the examined substrates, of N 6 -β-d-ribosyl-1,N 6 -etheno-adenine (1), was readily phosphorolysed (Fig. 4) with calculated reaction rate of ~0.45 µM/min at substrate concentration 66 µM. Phosphorolysis was easily observed spectrophotometrically as well as fluorimetrically. Hemolysate optical background does not interfere with the measurements, except visible re-absorption near the Slater band of hemoglobin at 410 nm (Fig. 4). Blood proteins contribute somewhat to the overall fluorescence with excitation at 275 nm, but with 1000-fold sample dilution this fluorescence, visible as short-wavelength inflection on Fig. 4, right panel, is low and negligible at λ ~ 430 nm, where the measurements are the most accurate.
The second substrate, the highly fluorescent N 6 -β-dribosyl-1,N 6 -etheno-isoguanine (2), was apparently inactive (rate <0.01 µM/min) with the same blood sample, although it was rapidly phosphorolysed by the purified E. coli PNP (Fig. 3) as well as by the calf enzyme. This result was somewhat surprising, since the human enzyme belongs to the same class of trimeric PNP, and shows homology of >70% with calf PNP (Bzowska et al., 2000). At present, we are unable to explain this phenomenon.
The experimental conditions of the presented reactions need to be optimized, for applications to clinical analyses. This refers particularly to buffer pH, substrate concentration (the apparent K m for the human enzyme) and excitation wavelength. With conditions fulfilled, the proposed assay will be probably much more sensitive than those described previously.

CONCLUDING REMARKS
We have described two novel, both fluorescent and fluorogenic, substrates for PNP. One of these, N 6 -β-dribosyl-1,N 6 -etheno-adenine (1), can be used to quantitate PNP activity in human blood. Possible applications include early detection of immunological deficiencies (Grunebaum et al., 2013). The second substrate, N 6 -β-dribosyl-1,N 6 -etheno-isoguanine (2), can be used to selectively detect bacterial PNP activity in biological samples, with possible use in the investigations of the suicidal gene therapy of cancer, utilizing bacterial PNP to generate in situ cytotoxic nucleobase analogs (Karjoo et al., 2017).
Various assays, including fluorimetric, were previously proposed for this enzyme (Bzowska et al., 2000;Wierzchowski et al., 2002;Wierzchowski et al., 2014), but their sensitivity was not always satisfactory, mostly because of slow reaction rates of the artificial substrates. Therefore search for new, more sensitive substrates is continued.