Synthesis and utilization of 8-azidoguanosine 3'-phosphate 5'-[5'-32P]phosphate. Photoaffinity studies on cytosolic proteins of Escherichia coli.

A family of guanosine 3',5'-phosphorylated nucleotides have been postulated to have pleiotypic regulatory properties in prokaryotes during the stringent response. To study proteins which may interact with nucleotides of this homologous series, a photoactive analog of guanosine 3',5'-diphosphate has been synthesized. The analog, 8-azidoguanosine 3'-phosphate 5'-[5'-32P]phosphate, proved to be an effective photoaffinity probe for two nucleotide-binding proteins of Escherichia coli sonicates. It predominately photolabels two proteins with approximate molecular weights of 86,000 and 65,000 (p86 and p65, respectively). The Kd for p65 was approximately 10 microM; that for p86 was not determined. The nucleotide-binding sites were characterized by photolabeling in the presence of various nucleotides. The nucleotides guanosine 3',5'-dipyrophosphate, guanosine 3'-monophosphate 5'-diphosphate, and GTP were most effective at decreasing photoincorporation into p86; guanosine 3'-diphosphate 5'-monophosphate was least effective, with guanosine 3',5'-diphosphate and GMP having an intermediate effect. ATP increased photolabeling of p86. However, ATP was one of the best of the nucleotides studied at decreasing photolabeling of p65, although guanosine 3'-monophosphate 5'-diphosphate, guanosine 3',5'-diphosphate, and GMP appeared only slightly less effective. The relative lack of effectiveness of guanosine 3'-diphosphate 5'-monophosphate inhibiting photolabeling of either protein supports observations that this nucleotide does not have a regulatory role in E. coli. The results presented indicate that the 8-azidoguanosine analogs of this homologous series will prove to be effective probes for studying the protein-nucleotide interactions involved in the stringent response.

A family of guanosine 3',5'-phosphorylated nucleotides have been postulated to have pleiotypic regulatory properties in prokaryotes during the stringent response. To study proteins which may interact with nucleotides of this homologous series, a photoactive analog of guanosine 3',5'-diphosphate has been synthesized. The analog, 8-azidoguanosine 3"phosphate 5'-[5'-32P]phosphate, proved to be an effective photoaffinity probe for two nucleotide-binding proteins of Escherichia coli sonicates. It predominately photolabels two proteins with approximate molecular weights of 86,000 and 65,000 (p86 and p65, respectively). The Kd for p65 was approximately 10 pM; that for p86 was not determined. The nucleotide-binding sites were characterized by photolabeling in the presence of various nucleotides. The nucleotides guanosine 3',5'-dipyrophosphate, guanosine 3'-monophosphate 5'-diphosphate, and GTP were most effective at decreasing photoincorporation into p86; guanosine 3"diphosphate 5'-monophosphate was least effective, with guanosine 3',5'-diphosphate and GMP having an intermediate effect. ATP increased photolabeling of p86. However, ATP was one of the best of the nucleotides studied at decreasing photolabeling of p65, although guanosine 3'-monophosphate 5'-diphosphate, guanosine 3',5'-diphosphate, and GMP appeared only slightly less effective. The relative lack of effectiveness of guanosine 3"diphosphate 5'-monophosphate inhibiting photolabeling of either protein supports observations that this nucleotide does not have a regulatory role in E. coli. The results presented indicate that the 8-azidoguanosine analogs of this homologous series will prove to be effective probes for studying the protein-nucleotide interactions involved in the stringent response.
Many microorganisms respond to nutritional deprivation by the synthesis of "magic-spot'' compounds, i.e. pppGpp' and * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ppGpp (1). During this stringent response, the tetraphosphate regulates cellular metabolism such that growth processes are inhibited but adaptive responses are encouraged. For example, stable RNA synthesis is inhibited, but some transcription is enhanced; protease activity is increased to provide amino acids for synthesis of new (adaptive) enzymes; and metabolic activities, in general, are slowed down while this process of adaptation is occurring (2). Since the discovery of these guanosine polyphosphates in 1969 (3), many mechanistic details have been elucidated, but many questions still remain unanswered.
A nucleotide photoaffinity probe may be defined as a nucleotide derivative which has affinities for binding sites and biological activity comparable to the unaltered nucleotide. Exposure to certain wavelengths of light converts the analog to a very reactive intermediate, typically a nitrene or carbene, which may result in covalent incorporation into the binding site if it is bound to a protein. There are certain advantages to using photoprobes over conventional chemical probes. One advantage is that K,, Kd, and K, values can be determined in the absence of activating light. Another advantage is that complex systems such as ribosomes, membranes, or even whole-cell sonicates can be studied. In this way, the in vivo situation may be more closely approximated and information is obtained that might be lost in a purified system. Many nucleotide photoaffinity probes have been synthesized and used successfully. The photoprobes [32P]8-N3~AMP and [ y -32P]8-N3ATP have been employed to study the mechanisms of action of CAMP-dependent protein kinase (4-6), and ["PI 8-N3cGMP has been used to study cGMP-dependent protein kinase (7). Photoactive analogs of GTP, i.e. [y-32P]8-N3GTP and [&y-32P]8-N3GTP,were used to study tubulin polymerization (8). Also, these nucleotide photoprobes may be successfully applied to many other biochemical problems. Considering the ubiquity and great importance of nucleotides to cellular metabolism and the general applicability of the photoaffinity approach, this technique will be more frequently utilized in the future.
Of the classical magic-spot compounds, only MSI, ppGpp, is known to be biologically active; MSII, pppGpp, appears only to serve as precursor of ppGpp. Recently, a third magicspot compound, ppGp(MSIII), was discovered in Escherichia coli. This nucleotide has been proposed to regulate transcription (9)(10)(11). Under certain conditions, other members of this guanosine nucleotide family may be found in vivo: pGpp and pGp appeared in E. coli mutants with a defective (p)ppGpp degrading enzyme (12). It is entirely possible that these nucleotides could occur in wild-type bacteria if the hydrolase were inhibited. To properly study magic spot-protein interactions, it will be necessary to synthesize a complete series of 8-azidoguanosine nucleotides, starting with [5'-32P]p8N3Gp, including the two triphosphates, and finally pp8NGpp. The present study involves synthesis and use of the first photoprobe of this homologous series, 8-azidoguanosine 3"phosphate 5'-[5'-32P]phosphate, to examine the potential usefulness of these analogs to detect and study proteins that interact with the magic-spot class of nucleotides.

MATERIALS AND METHODS
Guanosine nucleotides were obtained from P-L Biochemicals. DEAE-cellulose was purchased from Pharmacia. Polyethyleneimine TLC plates (PEI-cellulose F, 0.1-mm precoated plastic sheets) were manufactured by EM. Wild-type E. coli K12 (ATCC 10798) were obtained from the American Type Culture Collection. RNase A (from bovine pancreas, type IA, 71 units/mg) was purchased from Sigma; DNase I (2129 units/mg) from Worthington. Spectrapor dialysis tubing with a cutoff of 3500 daltons was used.
Preparation of E. coli Sonicates-The bacteria were grown at 37 "C to AsW = 0.9 in 1 liter of nutrient medium, pH 7.2, containing 5 g of yeast extract (Difco), 8 g of nutrient broth (Difco), 10 g of NaC1, and 246 mg of MgS0,/7Hz0. The cells were centrifuged at 14,727 X g for 10 min in a Sorvall RC58 centrifuge using the GSA head. The pellets were resuspended in 5 ml of 30 mM Tris/HCl, pH 8, 20% (w/v) sucrose. After addition of lysozyme to 100 pg/ml and EDTA to 10 mM, the cells were incubated 30 min on ice. The cells were sonicated for 2 min on ice at 100 watts (Braunsonic 1510 with microprobe) after addition of MgC12 to 20 mM. Following sonication, DNase I and RNase A were added to 5 pg/ml each. The sonicate was then dialyzed against 1 liter of 10 mM Tris, pH 7.4, 10% sucrose, 5 mM EDTA, 10 mM MgC12, 120 mM KC1,20 mM NaCl at 4 "C for 5.5 h. The resulting dialyzed sonicate was stored as 1-ml aliquots at -20 "C. The protein concentration of this sonicate was determined to be 54 mg/ml by the Bio-Rad procedure.
Photolabeling Experiments-Samples in the wells of an immunospot plate in an ice bath were photolyzed for 5 min a t a distance of 10 cm in a ChromatoVue Model CC20 light box (Ultra-Violet Products, Inc.). The wavelength of the light is 254 nm and the intensity is 160 microwatts/cm. The 5O-pl samples contain 2 p1 of E. coli sonicate (108 pg of protein) and various concentrations of [5'-3zP] p8N3Gp and competing nucleotides (e.g. ATP, GTP, ppGpp, ppGp, pGpp, pGp, and GMP) in 10 mM Tris, pH 7.4,5 mM EDTA, 10 mM MgClP, 120 mM KCI, and 20 mM NaCl. Photolyzed samples were solubilized with 25 pl of 25% sucrose, 2.5% sodium dodecyl sulfate, 2.5 mg/100 ml pyronin Y, 25 mM Tris/HCI, pH 8.0, 2.5 mM EDTA, and 15.4 mg/ml DTT. The entire 7 5 4 sample was loaded onto the gel; in addition, the wells of the spot plate were washed with a 1:3 dilution of the solubilizing solution, and this wash was added to the gel as well. The Laemmli-type polyacrylamide gel consisted of a 4% stacking gel and a 7-14% running gel and was run for about 4 h at a 32-mA constant current, as previously reported (4).
The major radioactive peak following the minor [3zP]Pr peak was pooled, rotary-evaporated, and identified a s [ Y -~~P J A T P .
After several additions of methanol to the residue followed by evaporation (to remove triethylamine), the [y-32P]ATP was redissolved in a small volume of methanol. Specific activity was determined by UV spectrophotometry and liquid scintillation counting. The reaction was monitored, and the radiochemical purity of the ATP was demonstrated with PEI TLC plates developed with 0.38 M KHzPOI, pH 3.4.
TLC-The migration of standards and reaction products were determined using PEI-cellulose F sheets and two solvent systems, as follows.
the procedure published for 8-N3GMP except that the bromination buffer was 1 M phosphate, pH 3.4 (13) (see Fig. 1). The enzymatic conversion of 8N3Gp to [5'-32P]p8N,Gp was based on the reports of Lillehaug and Kleppe (16,17). To a plastic tube was added 1 pmol of 8N3Gp and 0.3 pmol of [y-32P]ATP in methanol. This solution was evaporated and the residue redissolved in 250 p1 of 12 mM MgC12, 125 mM KCI, and 84 mM Tris, pH 8.9. Ten units of polynucleotide kinase ( 2 pl) were added, and the reaction mixture was incubated for 2 h at 37 "C. T o remove unreacted [32P]ATP, glucose was added to 0.1 M concentration to 0.5 mg of hexokinase (Sigma type 111, from yeast; 15 units/mg), and the solution was incubated 15 min a t 25 "C. The solution was then loaded onto a column (2 X 37 cm) of DEAE-cellulose (previously equilibrated with triethylammonium bicarbonate), and the analog eluted with 400 ml of 0-0.5 M triethylammonium bicarbonate. The (radioactive) coiumn elution profile is shown in Fig. 2 and consists of six peaks. (The elution can be followed with a UV monitor; however, UV light destroys the azide and even small losses of nanomolar quantities of analog are significant).
Identification of the components in each peak was done by comparing: (a) UV spectra, before and after photolysis; ( b ) UV spectra before and after treatment with DTT; (c) reactiv- ity of the products with 3"nucleotidase; and ( d ) TLC, with standards, of peak components and products produced from them by the above treatments. Representative reactions of the pGp derivative may be seen in Fig. 3; the results of the peak analyses are shown in Table I. First, peak 1 (12% of 32P) was shown to consist of [32P]P04 and gluco~e-6-[~~P]P, while peak 5 was determined to be unreacted [y3'P]ATP. Peak 2 (5% of 32P) was shown to be [32P]8-N3GMP, which is probably generated by contaminating 3'-nucleotidases or phosphatases in the polynucleotide kinase. The peak 2 material was not modified by incubation with 3"nucleotidase; it was photoactive and sensitive to DTT; and the R F values and UV spectra of it and its reduced product correspond to 8-N3GMP and 8-NH2GMP, respectively. Peak 3 material (3% of 32P) was not identified. Its R F value proves that it is not 8-N3GDP. Peak 4 (13% of 32P) was not photoactive or sensitive to DTT. Compounds with similar Rp values and UV spectra could be readily generated from peak 6 compound by either photolysis or exposure to temperatures of 37 "C or higher. It appears to be the thermal or photo breakdown product(s) of [5'-32P]p8N3Gp (see below). Peak 6 (63% of "P) contained one component of over 95% radiochemical purity. It was identified as the desired product, [5'-32P]p8N3Gp, by its UV spectrum and photosensitivity and by the fact that 3"nucleotidase converted it to 8-N3GMP. This identification is also supported by the selectivity of the polynucleotide kinase reaction, i.e. this enzyme is only known to add one phosphate. The late elution from the DEAE column also indicates it is the most negatively charged (most highly phosphorylated) species present. Peak 6 was pooled and evaporated to dryness at 20 "C, dissolved in anhydrous methanol, evaporated to dryness three times (to remove excess triethylammonium bicarbonate), and extracted and stored at -20 "C in anhydrous methanol. [5'-32P]p8N3Gp appears to be much more sensitive to reduction and/or thermal breakdown than the corresponding 8-N3GMP analog. After the above isolation of this analog, with less than 5% detectable radiolabeled impurities in the elution peak, there appeared two new contaminating substances of variable quantity with R F values (solvent B) of 0.31 and 0.24. These compounds could be produced from [5'-32P]p8N3Gp by reduction with DTT ( R F = 0.31) or photolysis ( R F = 0.24) as shown in Table I. Analog isolated by this procedure is over 80% radiochemically pure; the major contaminants are produced during the evaporation of the combined fractions in triethylammonium bicarbonate buffer. The analog could be further purified to over 95% radiochemical purity using a Waters high-pressure liquid chromatography system consisting of two Model 6000A pumps and a model 660 solvent programmer. The column, packed with polyethyleneimine (2.5 X 20 mm, J. T. Baker Chemical Co.), was eluted with a linear gradient of 25-250 mM sodium phosphate, pH 7.5, at a flow rate of 0.5 ml/min. The reduced analog, p8NH2Gp, elutes first a t 44 min, followed by p8N3Gp elution at 46 min. If present, p8BrGp elutes a t 55 min. Purity of the pNsGp is supported by the expected Amax at 278 nm and an isobestic point at 243 nm (Fig. 4). Both the X , , , and the isobestic point would be affected if significant contamination by either of the corresponding 8-bromo or 8-NH2 analogs was present. Both of these analogs have X, , , values in this UV region ( Fig. 1 and Table I) and would be the most likely compounds to cochromatograph with p8N3Gp.
Digestion of [5'-32P]p8N3Gp with 3"nucleotidase confirmed the assigned structure of the analog since this specific enzyme produced [32P]8-N3GMP (Table I). However, after prolonged digestion, varying small amounts of some refractory compounds remained with R F values (TLC system B, Table  I) of 0.33 and 0.25. DTT converted the faster-migrating compound ( R F = 0.33) to the slower-moving compound ( R F = 0.25). Using techniques previously described and a two-dimensional TLC system utilizing cellulose plates (developed in the first dimension with isobutyric acid/concentrated NH40H/Hz0, 66:1.7:33, and in the second dimension with saturated (NH4),S04/8.2% sodium acetate/isopropyl alcohol, 8018:2) reported elsewhere (9), the refractory compounds were identified as [5'-32P]5'-p8N3G2'p ( R F = 0.33) and [5'-32P]5'-p8N2HG2'p ( R F = 0.25). Since polynucleotide kinase does not accept nucleotide 2'-phosphates as substrates (16), the 2' isomer must have been produced from the [5'-''P] p8N3Gp by incubation in the alkaline reaction mixture. The presence of this analog did not appear to affect the photolabeling experiments. This is probably due to their low concentration, and the observation that the 2' isomer of ppGpp is reportedly not synthesized in uiuo (18) and is not known to be biologically active.
Photolabeling with [5'-32P]p8N3Gp- Fig. 5 shows an autoradiograph of a gel containing [5'-32P]p8N3Gp-photolabeled protein from E. coli sonicates. Only two major bands with apparent molecular weights of 86,000 and 65,000 (referred to as p86 and p65, respectively) are visible after photolysis at 5 KM analog concentration (sbt 2). Even at 30 pM [5'-32P] p8N3Gp, these are still the major bands; however, several minor bands appear at this higher concentration (slot 7). Slot 1 contains the Coomassie Blue-stained protein profile from which the autoradiograph was made. The Coomassie Blue bands which coincide with the bands on the autoradiograph are marked p86 appears to be one of the visible Coomassie  Fig. 4; DTT, reduction with DTT, pH 7.5, as shown in Fig. 3 Ultraviolet spectrum was determined on a Beckman Model 25 spectrophotometer. Ten pl of a stock analog solution (in methanol) were diluted with 1 ml of distilled water, and the spectrum was determined before and after (30, 60, 120, and 300 s) photolysis with a hand-held Mineralight UV lamp (254 nm, 190 microwatts/cm; Ultra-Violet Products, Inc.) Extinction coefficient of p8NGp is 1.2 X lo4 + 5% at Amax of 278 nm.

Synthesk and Utilization of [5 '-32PJp8N3Gp
Blue-stained bands, while p65 is a relatively minor component.
In an attempt to obtain an apparent Kd for [5'-32P]p8N3Gp binding, levels of photoincorporation at the various concentrations were determined by cutting out the band from the gel and measuring the amount of radioactivity by scintillation counting. As can be seen from Although p86 has a lower affinity for the analog than does p65, the specificity of the interaction is indicated by two observations. First, p86 is photolabeled a t very low concentrations of analog (5 p~) ; and, second, p86 is one of only two major bands even a t 30 PM [5'-32P]p8N3Gp. There was no detectable radioactivity incorporated by a sample incubated with the [5'-32P]p8N3Gp but not photolyzed or incubated with prephotolyzed analog. This indicates that the radioactive labeling is photodependent.
The autoradiograph of Fig. 7 shows the results of initial competition experiments. In this type of experiment, various nucleotides in addition to the photoactive nucleotide are added to the sample. If the nucleotide binds to the same site as the analog, the photoincorporation resulting from photolysis will be decreased. The greatest decrease in radioactivity will occur when the tightest-binding ligand is present. Thus, if the [5'-32P]p8N3Gp binds to a G T P site, one would expect  6. Saturation curves from Fig. 3. The amount of radioactivity in both bands in Fig. 3 was determined by liquid scintillation counting of relevant sections of the gel. 0, radioactivity in p86; 0, in p65.
GTP, rather than ATP, to cause maximal decrease in photoincorporation. Of the nucleotides tested, ppGpp (slots 6 and 7), ppGp (slots 8 and 9), and then GTP (slots 2 and 3) are most effective at decreasing [5'-''P]p8N3Gp photoincorporation into p86; pGpp (slots 10 and 11 ) and GMP (slot 12) have little or no effect. ATP consistently increases photoincorporation into p86 (and generally into a band of 33,000 daltons) (slot 5). Photolabeling of p65 is maximally decreased by ATP (slots 4 and 5), ppGp (slot 9), and GMP (slot 12). Both G T P (slots 2 and 3) and ppGpp (slots 7 and 8 ) were much less effective. It should be noted that pGp greatly decreases pho- tolabeling of these two proteins, as expected, and that the adenosine homolog of ppGpp, ppApp, is not as effective as ppGpp (data not shown).

DISCUSSION
The use of 8-azidopurine photoprobes of various 5"phosphorylated adenosine-and guanosine-containing nucleotides has proven to be quite successful. These probes have been used to identify specific nucleotide-binding proteins of CAMP, cGMP, ATP, GTP, and S-adenosylmethionine in biological systems of varying complexity. They have also been used to resolve certain mechanistic aspects of nucleotide-regulated phenomena (see Refs. 15, 19, and 20 and references therein). One of the main advantages of this type of photoprobe is that they may be used to detect and study specific nucleotideprotein interactions in complex systems. They may also be used as molecular markers to aid in the purification as well as identification of the biochemical properties of specific proteins. In the same way, photoprobes of the (p)ppGpp homologous series may be used to localize specific 5'p,Gpn binding proteins in subcellular fractions (membrane, cytosol, or DNA-binding proteins); to examine the modulation of enzyme activity by alterations in 5'p,Gpn-protein interactions caused by other nucleotides or environmental factors such as pH, ionic strength, and ionic composition; and to monitor purification of these proteins.
The synthesis of [5'-"P]p8NaGp presented two serious problems, one involving the behavior of the azido group, and the other the chemistry of the 3"phosphate bond (see Fig. 3). It is difficult to prevent destruction of the azido moiety of the [5'-32P]p8N3Gp. This difficulty stems from many sources. 1) The azido group is thermally labile-prolonged exposure to elevated temperatures converts [5'-''P]p8N3Gp to nonphotoactive compounds (one of which is p8NH2Gp). Therefore,

Synthesis and Utilization of [5 '-32P]p8N3Gp
concentration of azido nucleotide solutions, by rotary evaporation for example, must be done at low temperatures (18 "C or lower).
2) The analog is readily reduced to p8NH2Gp by the presence of even very low concentrations of certain reducing reagents, especially DTT. The 8-azidoguanosine analogs also appear to be reduced slowly by cysteine and mercaptoethanol under the conditions used for the polynucleotide kinase reaction. Unfortunately, this enzyme requires a reducing agent and comes in a buffered solution containing mercaptoethanol (cautionary note; some companies use DTT instead). 3) Extremes of pH are deleterious to the stability of the azide and exacerbate the problems posed by temperature and reducing agents. The enzymatic synthesis described in this report involves exposure to certain levels of these undesirable condkions: the reaction temperature is 37 "C, mercaptoethanol is present, and the pH for optimum polynucleotide kinase activity is 8.9. The temperature and/or the pH may be decreased, but the more amenable conditions decrease enzyme activity; therefore, increased reaction time is required with consequent increased exposure to the reducing agent. The problem with the 3'-phosphate arises because extremes of pH catalyze the exchange of the phosphate between the 2' and 3'-hydroxyls. Gp and BrGp are exposed to pH 3.4 during the bromination procedure and some BrG2'p may be formed, but this presents no problem since the final product, 8N3G2'p, is not a substrate for the polynucleotide kinase. The pH 8.9 environment during the enzymatic synthesis of [5'-32P] p8N3Gp probably causes the observed transesterification. As mentioned above, the pH can be decreased, but then the increased reaction time may result in increased reduction of the azido group. Despite these problems, the [5'-32P]p8N3Gp may be obtained in reasonably pure form and seems to be an effective photoaffinity analog of pGp. As such, it interacts with proteins that also bind ppGpp and ppGp, which are proposed to be regulatory nucleotides.
During normal growth, E. coli maintains a basal level of ppGpp in the low micromolar range; during nutritional stress, the stringent response occurs and ppGpp concentrations may rise to millimolar levels (21). It has also been suggested that basal concentrations of this analog may have a regulatory role (23). Another magic-spot compound, ppGp, is produced under certain circumstances and it, too, may have a role in regulation of biochemical processes such as RNA transcription (9)(10)(11). The ppGpp that is synthesized generally seems to exert its regulatory functions by competing with GTP for its binding sites, although some ppGpp-specific sites are known (22). Based on these data, the strongest prediction one could make about a putative ppGpp photoaffinity analog is that its photoincorporation should be strongly inhibited by GTP and/or ppGpp, and possibly by ppGp. The analog used in this research, [5'-32P]p8N3Gp, fulfills this criterion with respect to p86, as shown in Fig. 7.
The photolabeling of p86 with [5'-3ZP]p8NsGp is about equally inhibited by GTP, ppGpp, and ppGp, but pGpp has no apparent effect. GTP and magic-spot nucleotides apparently compete for the same site, but the nucleotide pGpp, which is not believed to be a biological effector, has no effect on photoincorporation of the analog. Although no K d could be determined from the photolabeling data (since concentrations of analog higher than 30 p~ would lead to too much nonspecific labeling), the fact that this protein was labeled at such low analog concentrations indicates that the K d would probably be under 50 p~.
Photoincorporation into p65 is best reduced by ATP and GMP; GTP, ppGpp, and ppGp are less effective. Since ATP is the most effective competitor, it is possible that the analog is labeling an ATP site. However, one might expect the high affinity binding of the analog to correlate with a corresponding specificity. It is probable that ATP allosterically modulates [5'-32P]p8N3Gp binding or that ATP phosphorylation causes the decreased photolabeling of p65 and the increased labeling of p86. Our recent data' indicate that [y3*P]ATP (and GTP) phosphorylates both of these proteins and that ATP phosphorylation may be the cause of the variations in [5'-32P]p8N3Gp photoincorporation. Why GMP should decrease photoincorporation into p65 is at present unclear. However, the data are not inconsistent with both GMP and pGp binding to this protein in vivo, in which case, the observed GMP competition a t 500 PM against 5 p~ analog would not be surprising.
In summary, under the conditions employed in this study, the analog [5'-32P]p8N3Gp photolabels primarily two proteins in E. coli sonicates. Differential protection against photoincorporation was observed with various nucleotides that one would expect to interact with p,Gp,-binding sites. Although the identity of neither of these bands is known, the use of the photoaffinity probe [5'-32P]p8N3Gp (and also [~-~'P18-N,ATP/GTP) can clearly be of use in further characterization of these proteins. The analog can also be used to study purified proteins, e.g. enzymes known to interact with ppGpp. One such enzyme, RNA polymerase, is inhibited both reversibly and irreversibly (on photolysis) by micromolar levels of p8N3Gp? In addition, [ E I ' -~~P ]~~N~G~ (or the "cold" nucleotide) may also be used to prepare [32P]pp8N3Gpp and ["PI pp8N3Gp by the chemical addition of phosphate (e.g. see Ref. 24). These probes would provide other valuable information concerning the regulatory effects of these nucleotides.