prevalence of nucleoside diphosphate kinase autophosphorylation in human colon carcinoma versus...

MOLECULAR OZRCINOGEIVESIS 2: 168- 178 (1989)

Prevalence of Nucleoside Diphosphate Kinase Autophosphorylation in Human Colon Carcinoma Versus Normal Colon Homogenates Brian Francis, Jean Overmeyer, William John, Ernest Marshall, and Boyd Haley' Lucille t? Markey Cancer Center, University of Kentuckx Lexington, Kentucky

The G-regulatory proteins of adenylate cyclase, tubulin, and the ras oncogene protein product require the production of GTP from ATP in order to exert their effects within the cell. This implies that the activity of nucleoside diphosphate kinase (NDPK) plays a major role in the regulation of cellular events requiring GTP and that the level of activity of this enzyme is critical. This report presents a simple method for trapping a specific isozyme of NDPK in its high-energy phosphorylated form (NDPK-P) using EDTA and demonstratesthatthis NDPK-P is tenfold higher in malignant colon tumor tissue than in normal colon tissue. This autophosphorylation of the 21,000 and 24,000 M, subunits of NDPK occurs rapidly a t O'C, will use either [Y-~~PIATR [y32P]GTR or the corresponding 8-azidopurine photoprobes, is intramolecular, displays saturation effects, and is prevented from forming if GTPyS is added. Dephosphorylation in the homogenate occurs rapidly upon addition of Mg2+ or any nucleoside-5'-diphosphate. The subunits autophosphorylated in the homogenates are mostly in the soluble phase, and they comigrate with the subunits of pure NDPK from human erythrocytes. Cross-addition of normal and malignant homogenates does not decrease the level of autophosphorylation of NDPK, which indicates that the level of NDPK-P may be a quantitative measure of the level of this specific NDPK isozyme form. Assays for NDPK activity show correspondingly elevated levels in the malignant homogenates. Using western blot and photoaffinity labeling techniques, we distinguished the NDPK-P subunits from two closely migrating GTP-binding proteins. These were identified as the ras gene protein product and a 20,000 M, protein, which comigrates identically with ADP-ribosylating factor (ARF). The ARF also comigrates in a tight band that is phosphorylated by [Y~~P]ATP or [Y-~~PIGTP when Mg2+ is present.

Key words: Malignant, EDTA, intramolecular, ras, photoaffinity probe

INTRODUCTION Nucleoside diphosphate kinases (NDPK. E.C. 2.7.4.6.)

have been reported to copurify with proteins whose bio- logical activity is regulated by GTF? These include tubulin [I ,21, elongation factor 1 [3], and a G protein of 20,000 M, which has properties similar to ~ J S p21 [4]. Also, adenylate cyclase (AC) from brain has copurified as a minor contaminant with NDPK [5], prompting speculation that brain NDPK may be associated with an AC-regulating com- plex that contains G proteins [6]. This idea is supported by a report of increased AC activity in W40-transformed normal rat kidney cells, which was associated with an increase in membrane-associated NDPK activity [71. It is reasonable to consider the hypothesis that NDPK activity would be associated with GTP and would require complexes to convert GDP to GTP more rapidly, thereby assisting the regulatory mechanism.

Our preliminary studies on colon tissue demonstrated an elevated phosphorylation of 21,000 and 24,000 M, proteins in malignant versus nonmalignant homogenates in experiments using GTF] ATF] 8N3GTF) or 8N3ATP as the phosphoryl donor [8]. These phosphorylations are similar to the reported phosphorylation of NDPK subunits of 19,000 and 21,000/22,000 M, in HeLa 53 cells [9,101 and Ehrlich ascites tumor cells [ I I] with the exception that excess EDTA

prevented the formation of NDPK in its high-energy phosphorylated form (NDPK-P) in these reports. However, in this report we observed that EDTA enhanced and stabilized this phosphorylation in malignant colon homogenates. In this study the 21,000 and 24,000 M, doublet, which is rapidly phosphorylated by y3'P-labeled GTP, ATP, or their respective 8-azido photoaffinity analogs in colon carcinoma homogenates, represented the subunits of an isozyme of NDPK whose high-energy intermediate(NDPK-P)could be formed by autophosphorylation in the presence of excess EDTA. Also, the level of this autophosphorylation of NDPK averaged tenfold greater in malignant colon tissue than in normal colon tissue; uninvolved colon tissue from patients with colon cancer displayed a level slightly higher than the normal but markedly less than in the excised carcinoma.

GTP phosphorylation of proteins is rare when compared to ATP phosphorylation, so it was important to distinguish any possible GTP-interacting protein from the NDPK subunits.

'Corresponding author: Lucille I? Markey Cancer Center, Univer- sity of Kentucky, 800 Rose Street, Lexington, KY 40536-0093.

Abbreviations: NDPK. nucleoside diphosphate kinase; 8N3GTP, 8-azidoguanosine-5'-triphosphate; 8N3ATP, 8-azidoadenosine-5'-triphosphate; AC, adenylate cyclase; PCA, percholoric acid; PSM, protein solubilizing mixture; SDS, sodium dodecyl sulfate; PAGE, poly- actylamide gel electrophoresis.

0 1989ALAN R. LISS, INC.

NDPK-PIN CARCINOMA AND NORMAL COLON 169

One GTP-interacting protein that could migrate in this region is the rasprotein productof approximately21,OOO M,(p21), which selectively binds GTP 11 21. The cellular proto- oncogenic form of ras p21 is phosphorylated on a serine residue only to a minute extent [13]. Also, certain ADP- ribosylating factors (ARFs) with values in the 19,000-21,000 M, region bind GTP i14.151. Elevated c-ras p21 has been shown in some, but not all, coloredal malignancies [16,17], and little is known about the levels of ARF. Therefore, photolabeling experiments with [32P]8N3cTe comigration of purified p21 and ARF on SDS-PAGE, and western blot studies were done. The studies eliminated the possibility that the 21,000-24,000 M, NDPK subunits comigrated with either of these proteins. The variation in levels of rasexpres- sion in colorectal malignancies contrasts with the consis- tent elevation of NDPK-P observed in these malignant tissues and suggests that certain isozymes of NDPK may play an important role in carcinogenesis.

MATERIALS AND METHODS Tissue Homogenization

Fresh human colon carcinoma surgical specimens were frozen in phosphate-buffered saline at - 70°C. Wet tissue (1 00 mg) was washed with buffer (1 0 mM Tris, 5 mM KCI, pH 7.1) and homogenized at 0°C using a glass tissue grinder containing 2 mL of the same Tris buffer supplemented with pepstatin A (0.5 pg/mL), leupeptin (0.5 pg/mL), and aprotonin (0.5 pg/mL). The resulting homogenates were spun for 1 min in an Eppendorf centrifuge to remove par- ticulate debris. Aliquots of the resulting supernatant, usually 50 pL(80-120 pg), were used in experiments. Protein concentrations varied from 1.6 to 2.4 mg/mLas measured using the Bio-Rad protein assay (Bio-Rad Laboratories, Rich- mond, CA).

Purified Protein and Nucleotides

Nucleoside diphosphate kinase purified from human erythrocytes, pepstatin A, leupeptin, aprotonin, and nucleoside di- and triphosphates were obtained from Sigma Chemical Co., (St Louis, MO). Purified recombinant c-ras p2 1 and monoclonal anti-ras p2 1 antibody Y 13-259 were provided by Dr. Y. 5. Cho-Chung (National Institutes of Health). ARF was provided by Dr. J. Moss (National Insti- tutes of Health).

Radiolabeled Nucleotides

8N3GTP was synthesized by a modification of the procedures of Czarnecki et al. [181, as described by Potter and Haley [19]. [y-32P18N3GTe [ T - ~ ~ P I ~ N ~ A T ~ [ Y - ~ ~ P I G T ~ and [Y-~~PJATP were prepared according to Khatoon et al. [20]. [cY-~~PI~N~GTP was synthesized by procedures described by Potter and Haley [19]. Specific activities of the nucleoside triphosphates ranged between 5 and 18 mCi/pmol.

Phosphorylation and Photolabeling Reactions

Phosphorylation and photolabeling reactions were performed on ice using Eppendorf tubes. Nucleotides and buffer, together with Mg2+, Ca2+, or EDTA as indicated, were mixed and tissue homogenate was added to initiate

the phosphorylation reactions. Reactants were mixed rapidly using either an airstream or the micropipette that had contained the homogenate. After different periods of incubation with or without ultraviolet (UV) irradiation, reactions were terminated by addition of 7% perchloric acid (PCA) or a protein-solubilizing mixture (PSM) containing Tris (20 mM), sodium dodecyl sulfate (SDS; lo%), urea (2 M), and DTT (2.5%) at pH 8.0 [19]. PCA-terminated mixtures were left for 10 min at 0°C prior to centrifugation and addition of PSM. All phosphorylation reactions were carried out at pH 7.1, and the pH was measured using a microelectrode. Samples were irradiated with a hand-held UV lamp (7500 pW/cm2).

Denaturing Gel Electrophoresis

Homogenate proteins in PSM were separated on one- dimensional SDS gels using 10% or 12% polyacrylamide. After the gels were dried, autoradiograms were prepared on Dupont Cronex film with DuPont Cronex Quanta 111 inten- sifying screens (Dupont Instruments, Wilmington, DE). Films wereexposedforvarying lengthsoftimeat - 70"C.Appar- ent molecular weights were estimated using protein markers from Bio-Rad. Quantification of the amounts of radioactivity in labeled proteins of the homogenates was carried out by densitometry using an LKB 2202 UltroScan Laser Densitometer (LKB Instruments, Inc., Rockville, MD) or by liquid scintillation counting of the excised gel portions.

Measurement of Radioactive Nucleotide Levels by TLC

Aliquots of reaction mixtures containing 32P-labeled nucleotides before and after addition of tissue homogenate were spotted on PEl-Cellulose F plastic TLC sheets (EM Reagents). The TLC sheets were eluted with 400 mM ammo- nium bicarbonate, and the migration of radioactivity shown by autoradiography was compared with the Rf values of ATP (0.38). GTP (0.181, ADP (0.551, AMP (0.681, and Pi (0.85) standards. Radioactivity was determined by liquid scintillation counting of sections from TLC sheets.

Western Blot

Colon carcinoma proteins separated by SDS-polyacrylamide gel electrophoresis (PAGE) were transferred to nitro- cellulose and challenged first with monoclonal antibody Y13-259 against ras p21 and then with ['251]-protein A before x-ray film exposure.

Phosphoamino Acid Analysis

Proteins were extracted from SDS gel slices, hydrolyzed, and analyzed by one-dimensional TLC electrophoresis, according to the procedure of Hunter and Sefton [22]. Hydroxylamine treatment was done by the procedure of Lipmann (251.

RESULTS

Early experiments showed a phosphorylation in CO-

Ion carcinoma homogenates that was stabilized in the presence of EDTA concentrations, which abolished the phosphorylation of the great majority of other proteins.

I 70 FRANCIS ETAL.

I 2 3 4 5 6 7 8 9 1 0 1 1 1 2

21 - 14-

INCUBATION TIME (sacs)

Figure 1. Autoradiogram of an SDS gel containing proteins from a malignant colon tissue homogenate separated by PAGE after incubation with 5 pM [Y-~'P]ATP for 10,30, or 60 s in the presence of 0.5 mM Mg'+, 0.5 mM Ca2+ or 5 mM EDTA or without added divalent cations or chelating agents, as described in

Figure 1 shows a representative autoradiogram of the time course of phosphorylation by [Y-~~PIATP of a colon carcinoma homogenate at o"C, pH 7.1, with addition of varying amounts of divalent metal ions and EDTA. The addition of 0.5 mM Ca2+ or Mg2+ generally seemed to enhance the level of phosphorylation of all proteins, including a 20,000 M, species. However, the exception was a doublet of 21,000 and 24,000 M, proteins whose phosphorylation decreased with time. These same two proteins increased in level of phosphorylation in the presence of 5 mM EDTA. The same effect of EDTA on this doublet was also observed if [y-32P]-labeled GTP. 8N3ATP or 8N3GTP was used. The level of phosphorylation of the 21,000 and 24,000 M, proteins is graphically displayed in Figure 2. It shows that EDTA seemed to prevent the dephosphorylation of these proteins, whereas the presence of divalent metal ions, especially Mg2+, seemed to enhance this pro- cess. To confirm that Mg2+ and Ca2+ promoted dephosphorylation, maximum phosphorylation was obtained in the presence of 5 mM EDTA and [Y-~~PIATP. Mg2+, or Ca2+ was then added in excess of the EDTA, and the rate of dephosphorylation was followed. The results showed a 75-80% decrease within 30 s (data not shown) compared with the rate when no Mg2+ or Ca2+ was added. Experi- ments to determine the kinetic parameters of the phosphorylation of the 21,000 and 24,000 M, proteins showed almost identical properties for both proteins. Saturation of 32P incorporation occurred at 8 to 12 pM [Y -~~PIAT~

4 2 4 4 2 I

10 30 60 10 30 60 10 30 60 10 30 60 - - - - - - 0.5 05 0.5 - - -

- - - 0.5 05 05 - - - - - 5 5 5

Materials and Methods. At the left of lane 1 are shown the positions of proteins used as molecular weight markers. On the right of lane 12 the positions of the 21,000- and 24,000-M, labeled proteins are shown.

GTP, or the corresponding 8N3 purine analogs. Apparent K, values for all of the nucleoside triphosphates ranged from 5 to 7 pM in the homogenate.

Phosphorylation time courses were carried out using malignant and uninvolved colon homogenates to compare the rates and final levels of phosphorylation of the 21,000 and 24,000 M, proteins. The autoradiogram shown in Fig- ure 3 displays the higher level of phosphorylation observed in a carcinoma compared with an uninvolved homogenate. Phosphorylation was essentially complete within 10-20 s and remained level for over 2 min with AT? GTP. or their 8N3 purine analogs as substrates. Experiments using malignant and uninvolved tissues from the same patients showed the same discrepancy as seen in Figure 3, eliminating sam- ple source as the cause of this elevated phosphorylation. Also seen in the autoradiogram of Figure 3 are phosphorylations of 66,000 and 44,000 M, proteins, which were hardly detectable in the uninvolved tissue.

When homogenate from similar experiments was frac- tionated after phosphorylation by centrifugation at 100,000 g for 30 min, more than 95% of the 32P-labeled protein was recovered in the soluble fraction, showing that this was probably a cytosolic or a loosely associated membrane protein.

These results suggest that in some cases, colon carcinoma tissue 21,000 and 24,000 M, proteins could be more extensively phosphorylated than the corresponding uninvolved colon proteins. To determine whether this was a


I I I 1 15 30 45 60

B

- I

I I I 1 15 30 45 60

TIME (secs)

Figure 2. Time coursesof phosphorylation of 21,000 M, protein (A) and 24,000 M, protein (B) without added divalent cations or EDTA (no addition) and with added 5 mM EDTA, 0.5 mM

Ca2+, or 0.5 mM Mg2+. Relative 32P incorporation was measured by densitometry from the autoradiogram of Figure 1.

I 2 3 4 5 6 7 8 9 10 I1 12

21-

INCUBATION TIME (secs)

4 24 4 2 I

-(I-1

10 20 30 45 60 120 10 20 30 45 60 120

TISSUE TYPE TUMOR UNINVOLVED

Figure 3. Autoradiogram of an SDS gel containing homogenate proteins from colon carcinoma and uninvolved tissues separated by PAGE after phosphorylation using 5 FM [y-32P]8N3GTP in the presence of 5 mM EDTA for different times at 0.C.

widely occurring phenomenon, we surveyed in a blind study 17 colon carcinoma, 9 uninvolved colon, and 1 1 normal colon tissuesfor 21,000 and 24,000 M, protein phosphorylation using [y-32P]8N3GTP for 30 s at 0°C in the presence of 5 pM EDTA. The tissue with the highest level of phosphorylation was assigned loo%, and the rest of the samples were normalized from this value. Statistical results

from this experiment are shown in Table 1. They demonstrate that the level of phosphorylation was greater in malignant colon tissue than in uninvolved tissue and that this difference was even more marked when the comparison is made with normal colon tissue. On the average the colon carcinoma tissues had a phosphorylation level six- to sev- enfold higher than uninvolved tissue and nine- to tenfold

172 FRANCIS ETAL.

Table 1. Phosphorylation of 21,000 and 24,000 M, Proteins

Tissue Protein Carcinoma Uninvolved Normal 2 1,000

Range 12-100 2-18 2-6 Mean 39 6 4 SD 25 4 2

24,000 Range 9- 100 2-1 2 2-1 2 Mean 38 6 4 SD 25 4 2

The relative levels of 32P incorporation were converted to a scale in which the highest level observed was taken as 100. The range, mean and standard deviations (SD) shown in the table were derived from thevalues obtained.

higher than normal tissue. The 66,000 M, protein phosphorylation was also highly elevated in colon carcinoma tissues but was not usually as extensive as the 21,000 and 24,000 M, phosphorylations in the presence of EDTA.

After establishing the validity of the elevated phosphorylation observation, experiments were designed to char- acterize this activity and to determine its mechanism and catalytic source. First, an experiment involving cross-mixing of malignant and uninvolved homogenates was devised to determine if either tissue contained a component that activated or inhibited the observed phosphorylations. The resulting homogenate mixtures were incubated with [y-32P18N3GfP for 60 s at 0°C in the presence of EDTA or MgZf . The results from the experiment with EDTA are shown in Table 2. Qualitatively similar results were obtained in the presence of Mg2+. The effect of cross-mixing was additive, indicating that neither tissue contained a component that decreased or increased the formation of the phosphorylated form of these proteins, whether or not Mg2+ was present. This contrasted with the 45% decrease in 66,000 M, when equal amounts of uninvolved and malignant homogenate were mixed (data not shown).

The rapid rate of phosphorylation of these proteins at 0°C suggests that it may be an intramolecular event. A dilution experiment, similar to that used by Rosen et a1.1231

Table 2. Protein Phosphorylation in Mixtures of Malignant and Uninvolved Colon Hornogenates

Volume of homogenate (pL)

Malignant Uninvolved tissue tissue 2 1,000 M, 24,000 M, 0 30 15 20

30 0 100 100 30 5 100 95 30 10 100 100 30 20 110 110 30 30 115 120 30 40 125 120

32P incorporated into the 21,000 and 24,000 M, proteins was measured as described in Materials and Methods. The values given represent the relative level of 32P incorporated taking that for malignant tissue by itself as 100.

to demonstrate autophosphorylation of type I I CAMP- dependent kinase, was performed using constant concentrations of EDTA and [y-32P]8N3GTP while varying the homogenate concentration by increasing the total volume of the phosphorylating mix. After 10 s the protein was precipitated by 3% PCA at O'C, the mixture was left for 10 min and centrifuged, and the entire pellet was dissolved in PSM and subjected to SDS-PAGE. No decrease in 32P incorporation was observed between samples left for 5 to 20 min in the presence of PCA at 0°C (data not shown). Protein levels recovered were monitored by densitometry and varied less than 10%. 32P incorporation was determined as before, and the results are displayed in Figure 4, which shows that dilution had very little effect. If a bimo- lecular mechanism had been involved, decreases would have been expected to follow the small dashed lines. A unimolecular reaction would have followed the large dashed lines with no decrease. These data suggest that the phosphorylation occurs by an intramolecular mechanism or a closely associated kinase.

Nucleotides are rapidly hydrolyzed in homogenates and their effects may be lost in the preparation of these samples. Experimenb were designed to test the effects of added nucleotides within the 10-30 s with 0°C incubations that were routine in our studies. The rate of hydrolysis of 5 pM [Y-~~PIGTP or ATP in the presence of 5 mM EDTA was determined to be less than 5% after 5 min. Under these conditions the formation of phosphorylated protein could be inhibited by over 95% by adding 50 p,M GTP. ATP. GDP, ADe or the corresponding 8N3 purine analogs. Fifty micro- molar GTPyS prevented phosphorylation by 85%, whereas a like addition of GMP or cGMP had no significant effect. Inhibition of 32P labeling of the proteins in a concentration- dependent manner by ATP and GTP is shown in Figure 5. Both nucleotides caused a 50% decrease in the 1-5 p M

I = I 0

a N m I 1 W

F U J W Lz

24K "

REACTION VOLUME ( ) I )

Figure 4. Dependence of phosphorylation of 24,000 M, protein (closed circles) and 21,000 M, protein (closed boxes) on protein concentration. The same amount of malignant homogenate protein in different total reaction volumes was incubated for 10 s with 5 pM [y3*P]8N3GTP in the presence of 5 mM EDTA.

NDPK-PIN CARCINOMA AND NORMAL COLON

6'

4

2 2 0 I-

g 0 a a: 0 0 '

12 f a N 7)

w 10 > I-

w

- a - 1 8 a

6

4

2

0

A . 2 4 K

B. 21K

I L I

5 10 15 20 CONCENTRATION fpM)

Figure 5. Phosphorylation of 24,000 M, (A) and 21,000 M, (6) proteins of colon carcinoma homogenate using 5 pM [y3 P8N3C;TP and 5 mM EDTA for 30 s at 0°C in the presence of increasing concentrations of GTP (open boxes) or ATP (open circles).

range with GTP, which appear to be slightly more effective with both proteins.

The ability of GTP and GDP, but not GMP, to prevent the observed autophosphorylation and the similarities of the 21,000 and 24,000 M, proteins to reports on NDPK in other systems suggested that this phosphorylation may be NDPK-F? However, reports from other laboratories stated that the NDPK-P formation was prevented by the presence of excess EDTA [ l l ] . The phosphorylation and molecular weight values of purified NDPK from human erythrocytes were compared with those of the 21,000 and 24,000 M, colon carcinoma homogenate proteins. As shown in Figure 6, purified NDPK forms a NDPK-P intermediate in the presence of 5 mM Mg2+ or 5 mM EDTA and this phosphorylated intermediate has two subunits that comigrate with the 21,000 and 24,000 M, phosphorylated homogenate proteins. Addition of tenfold excess ADP. ATP, GDP, and GTP to the purified NDPK prevented NDPK-Pfor- mation in the presence of Mg2+ or EDTA (data not shown). The dependence of purified NDPK autophosphorylation on the concentration of [y-32P]8N3GTP was studied in the presence of 5 mM EDTA using a 10-s incubation period at 0°C. Saturation was observed at 8 p M and a K, of 3 p M was calculated (data not shown).

66 +

45 +

31 -+

21 +

I 2 3

I 73

COLON NUCLEOS IDE CARCINOMA DI PHOSPHATE

HOMOGENATE K I NASE

Mg2+ Mg2+ E DTA

Figure 6. Phosphorylation of colon carcinoma hornogenate proteins (lane 1) in the presence of 5 mM Mg2+ and purified nucleoside di hos hate kinase (lanes 2 and 3) in the Presence of 5 rnM MgP+ o r 5 mM EDTA using 5 pM [y-32P]8NsGTP for 60 s at 0°C.

If the 21,000 and 24,000 M, proteins that displayed elevated phosphorylation in colon carcinoma homogenates represented an elevated level of NDPK-P formation, then these homogenates should display elevated NDPK activity. This activity can be measured using TLC to follow the transfer of 32P from [ Y - ~ ~ P I ~ N ~ G T P to ADP through the NDPK-P intermediate to form [Y-~~PIATF? Several such experiments were performed, and an example is shown in Figure 7. [ Y - ~ ~ P I ~ N ~ G T P was used as the starting nucleotide to decrease the effects of endogeneous adenylate kinase, for which it is not a substrate (data not shown). In the presence of EDTA, [Y -~~PI~N~GTP is hydrolyzed quite slow- ly, but upon addition of ADP and MgZf the hydrolysis is very rapid and there is immediate conversion of ADP to [y-32P]ATP. which reaches a maximum within 30 s and is rapidly used itself to produce [P-32PlADP through the action of adenylate kinase. At 20 min most of the GTP and ATP were hydrolyzed to leave primarily [32PlPi and [P-32P]ADP This result shows that NDPK activity can be measured using this technique of monitoring [Y-~~PIATP production within 30 s but that the values will represent less NDPK activity

174 FRANCIS ETA L.

COLON CARCINOMA

TIME (mins)

Figure 7. Time course of synthesis of [Y-~~PIATR [p-32PlADP and [3ZP]orthophosphate from [y-3zP]8N3GTP (5 pM) and ADP (5 FM) at 0°C in the presence of 5 mM Mg2+ (open symbols) or

Table 3. Comparison of 21,000 and 24,000 M, Protein Phosphorylation and NDPK Activity of Malignant and

Uninvolved Colon Homogenates

Phosphorylation NDPK Activity Tissue A* B* C+

homogenate 21,000 M, 24,000 M, [Y-~~PIATP Formation

Malignant 1 45 35 80 2 100 100 100 3 35 45 100 4 15 20 65 5 25 35 75 6 25 30 75

7 10 10 60 8 10 10 50 9 5 5 45

10 <5 <5 40 11 15 15 55 12 10 15 55 13 5 5 45 14 5 5 35

Uninvolved

5 mM EDTA (closed symbols) and colon carcinoma tissue homogenate.

Colon carcinoma tissues with elevated phosphorylation of the 21,000 and 24,000 M, proteins displayed about 60% higher total NDPK activity than uninvolved colon tissues.

A study of the chemical nature of the NDPK-P phosphate bond was done to determine the amino acid residue involved. First, the 21,000 and 24,000 M, [32P]NDPK-P proteins from a carcinoma homogenate were eluted from excised portions of SDS gels, hydrolyzed, and tested for the presence of phosphorylated serine, threonine, or tyrosine residues [22]. Less than 1 YO of the radioactivity was found in phosphoserine and the remaining radioactivity in orthophosphate, which showed that these amino acids were probably not involved. The phosphoproteins also tested negatively for the presence of an acyl phosphate by reaction with hydroxylamine [25].

For experimental results presented in Table 4, colon carcinoma hornogenate proteins were incubated at 0°C for 60 s with 5 pM [y-32P18N3GTP and 5 mM EDTA prior to SDS-PAGE. Without fixing or staining, the gel was exposed to x-ray film to locate the phosphorylated 21,000 and 24,000 of proteins. Slices containing pairs of 2 1,000 and 24,000

*Percentage of maximum level of 32P incorporation observed. tNormalized percent converSion of [y-32P]8N3GTP into [Y-~~PIATP (for total NDPK activity the same tissue homogenates were mixed with 5 mM Mg2+. 5 pM [V -~~PI~N~GTP and 5 pM ADP for 30 sat 0°C).

than actually exists because of the effects of other ATPases and adenylate kinase. With this caveat, the level of NDPK-P formed was compared with the level of NDPK activity using six malignant and eight uninvolved colon homogenates from different patients. The results are shown in Table 3.

Table 4. pH Stability of 21,000 and 24,000 M, Colon Carcinoma Phosphoproteins

32P incorporated cprn x pH Oh 0.25h 1 h 3 h 12h 18h

1 2.0 1.1 0.5 0.3 0.1 0.1 3 2.1 1.9 1.3 0.8 0.4 0.2 7 2.1 2.0 1.4 0.9 0.5 0.3

11 2.0 2.0 1.4 1.1 0.6 0.4 13 2.2 2.2 1.9 1.6 1.1 0.7

NDPK-P IN CARCINOMA AND NORMAL COLON 175

M, 32P-labeled phosphoproteins were excised from the gel, washed with 3 x 20 mL 10 mM Tris buffer pH 7.1 (1 0 min each) and the 32P in the slices measured by Ceren- kov counting to give the zero time point levels. The gel slices were then immersed at 23°C in 20 mL of HCI (pH 11, 50 mM sodium formate (pH 3), 10 mM Tris (pH 7). Ten millimolar 3-(cyclohexylamino)- 1 -propanesulfonicacid (pH 11) or NaOH (pH 13) and a t various times they were removed and counted for 32!?

At 23°C it took I S min for 50% of the 32P bound to the proteins in an SDS-gel slice to be hydrolyzed at pH 1, whereas at pH 13 it took 12 h. These results are consis- tent with a high-energy histidine phosphate bond that is preferentially hydrolyzed under acidic conditions [241.

Since the phosphorylation of the 21,000 and 24,000 M, proteins appears to involve an intramolecular mechanism and does so with quite high affinity, we attempted to photolabel these binding sites with [d2P1 and [ Y - ~ ~ P I - 8N3GTP and the corresponding 8N3ATP probes. Under no conditions could we demonstrate photoinsertion into these proteins. This was probably because of the rapid autophosphorylation that displaced the nucleoside triphosphate from the active site and the fact that the base was not

specifically involved in the binding site and hence not close enough to active site amino acid residues to photoinsert. The latter is supported by the ability of all y-32P-nucleoside triphosphates tested to form the phosphorylated proteins.

Because of the reported relationship between the ras- oncogene protein product (p21) and NDPK [4], experiments were designed to resolve p21 from the 21,000 M, subunit of NDPK. On our gel system both the molecular weight of ras monoclonal interaction to endogenous p21 and the molecular weight of purified p21 wasabout 25,000, slightly higher than the NDPK 24,000 M, subunit.

The autoradiogram in Figure 8 shows the results of photolabeling and phosphorylation experiments using a colon carcinoma homogenate. Proteins of 21,000 and 24,000 M, autophosphorylated in the presence of EDTA (lanes 3,4, and 5) are compared with those photolabeled with [y-32P]8N3GTP (lane 2) or [R-~~PI~N~GTP (lanes 6-1 0). Addition of excess nonradioactive ATP to reactions containing [32P]8N3GTP prevented labeling by phosphorylation or photoinsertion into proteins that contained ATP binding sites. However, photolabeling involving sites specific for GTP was retained and [32P18N3GTP site occupancy enhanced because of reduced 8N3GTP hydrolysis. For exam-

I 2 3 4 5 6 7 8 9 10

45-

31-

21 - 1 4 4

h J

TEMPERATURE ("C) TIME (min)

[32P] 8N3GTP

EDTA Mg2+

ATP ( ImM)

+ + + + + - + - - - 37 37 37 24 0 37 37 37 37 37 I I I I I 5 2 I I I I I i x i s ~ s c o C . c U

- + - - t + + - - - + + - - - + t + + - + + - - - + + - + +

Figure 8. Autoradiogram of an 50Sgel containing colon carcinoma proteins separated by PAGE after treatment with 5 JLM [Y -~~P] (lanes 1-5) or [ u - ~ ~ P I ~ N ~ G I T (lanes 6-10). Preincubations were f o r 1,2, or 5 min, as indicated, followed by photolysis for

1 min (lanes 2 and 6-10), procedures done at the temperature indicated. ATP (1 mM) was added before initiation of preincu- bation, which was started by the addition of [32P]8N3GTP to the completed reaction mix.

176 FRANCIS ETAL.

ple, in Figure 8, lane 8 lacks added 1 mM ATP, and the decrease in photoincorporation into specific GTP binding proteins is evident (notably at 20,000 and 31,000 M, values). Also evident is the increased photoinsertion into proteins at 38,000 and 44,000 M, values, which photolabeled with both 8N3GTP and 8N3ATP (data not shown) and represent proteins not absolutely specific for GTP Both [y-32P]8N3GTP and [ c K - ~ ~ P I ~ N ~ G T P photolabet the same proteins. The protein near 53,000 M, value migrates identically to the p-subunit of tubulin, a known GTP binding protein that has a fast exchange rate when not tied up in microtubules. The 40,000-44,000 M, proteins may be the G-regulatory proteins of AC. The 30,000-31,000 M, photolabeled protein remains unidentified. The protein at 19,000-20,000 M, value photolabeled with kinetics similar to purified ADP-ribosylation factor (ARF) and comigrated with this factor on SDS-PAGE (data not shown). This 19,000-20,000 M, 8N3GTP-photolabeled protein did not comigrate with the 21,000 M, subunit of NDPK (compare lanes 2 and 3) nor did it react to antibody against ras p2 1. The30,OOO-31,000 and 19,000-20,000 M,proteinswere not consistently photolabeled in all colon malignancies to the extent shown in Figure 8.

DISCUSSION

The autophosphorylation of two proteins of 21,000 and 24,000 M, by ATP and GTP was consistently elevated when we compared malignant and normal tissues from colon and used excess EDTA to trap the phosphorylated inter- mediates. These phosphoproteins represented the subunits of NDPK trapped in their high-energy phosphoryl form (NDPK-P), which is used to convert nucleoside diphosphates to triphosphates. Evidence for this includes: (a) the ability to use [cK-~~P]ATP or [Y-~~PIGTP to form the [32PlNDPK-P and transfer this [32P]phosphate to a nucleoside diphosphate, (b) the comigration of [32PlNDPK-P subunits from purified NDPK with the [32Pl-phosphorylated proteins observed in the homogenates, (c) the observations that pure NDPK-P subunitsand homogenate phosphoproteins behave in the same way with regard to prevention of NDPK-P formation by appropriate nucleoside di- and triphosphates, (d) the fact that NDPK-P phosphoproteins are formed by a protein dilution independent mechanism indicating an intramolecular autophosphorylation, (e) the observation that malignant homogenates with elevated phosphorylation of the 21,000 and 24,000 M, proteins also showed elevated NDPK activity, (f) both purified NDPK and the homogenate proteins use the 8N3purine deriva- tives of ATP and GTP as effectively as the native nucleotide but neither one is photolabeled, and (9) the phosphate that is covalently bound to these proteins is not bound to serine, threonine, tyrosine, or an acyl phosphate. It is rapidly hydrolyzed in acidic solution as has been found for purified NDPKfrom a variety of sources in which the phosphate is linked to histidine [26-301.

While these data support the conclusion that the phosphorylated proteins we observed trapped by the presence of excess EDTA were NDPK-P, it should be noted that oth- ers have reported that NDPK from Ehrlich ascites tumor

cells requires Mg2+ to form NDPK-P and that the presence of EDTA prevents by over 99% the formation of this species [ l 11. However, purified NDPK from human erythrocytes was able to form the NDPK-P form in the presence of excess EDTA (Figure 6). A possible explanation for this discrepancy could be that two or more different isozyme forms of NDPK that behave differently in this respect are involved. Our observation that while the formation of [32P]NDPK-P was elevated about sixfold in malignant versus uninvolved colon, the total NDPK activity was only -60% higher supports this concept and indicates that the [32P]NDPK-P we saw formed in the presence of EDTA was only a fraction of the total NDPK present. Also, we observed that the formation of [32PlNDPK-P reached 50% saturation at approximately 5 p M using [Y-~*P]ATP or the corresponding radiolabeled GTP, 8N3ATP or 8N3GTP This is a very low value considering that the reported K, values of NDPKs from several sources varied between 50 and 200 p M for GTP and 80 to 3000 p M for ATP 1311. Additionally, a 50% decrease in [32PlNDPK-P produced by [-y-32P]- 8N3GTP was obtained by the presence of less than 5 pM GTP or ATP (Figure 51, again indicating a very tight binding affinity for this NDPK. Since human erythrocytes contain at least six different isozymic forms of NDPK, each exhibiting differentaffinitiesfor GTPand ATP [311, it is possible that the [32PlNDPK-P we observed represents only one specific species of several present.

The difference between the levels of [32PlNDPK-P formed in colon carcinoma homogenates and in uninvolved or normal colon tissue could have been caused by factors that activate the formation of NDPK-P in malignant homogenates or prevent its formation in normal homogenates, even if both tissues contained the same absolute level of this specific isozymic NDPK. Also, the amount of NDPK in the NDPK-Pform might have differed between malignant and normal tissues before [y3*P1GTP or ATP was added, resulting in less [32P]NDPK-P being observed even though the total NDPK-P levels were the same or higher. However, our results suggest that the difference was due to an elevated level of this species of NDPK since (a) the NDPK-P was very unstable in homogenates without EDTA and would not likely exist except for a very short time after the addition of a nucleoside triphosphate (Figures 1 and 2), (b) the cross-mixing experiments of malignant and uninvolved homogenates displayed additive effects on NDPK-P formation, which indicated that no factors (e.g., ATP or ADP) existed in the uninvolved tissue a t levels that could prevent formation of [32PlNDPK-P, (c) likewise, nothing existed in the malignant homogenate that activated NDPK-P formation in uninvolved homogenate (Table 2). (d) the time course of the NDPK activity showed that added nucleotide triphosphate was rapidly converted to Pi and nucleoside monophosphate (Figure 71, which does not affect NDPK-P formation; it is therefore unlikely that endogenous nucleotides that could affect this phosphorylation existed in these homogenates at any significant level, (e) the tissues that showed elevated [32PlNDPK-P formation also showed higher NDPK activity (Table 3). Thus, while this evidence is not conclusive it does strongly suggest that


the level of [32P]NDPK-P formation may be a measure of the level of the NDPK isozyme that forms [32PlNDPK-P in the presence of excess EDTA. This is also supported by the relatively large spread of values of NDPK-P levels observed in the malignant homogenates (Table I), which probably reflects the varied percentage of malignant versus normal cells in the surgically removed section.

It is possible that this particular NDPK isozyme may be associated with different complexes in malignant and normal cells in a manner that either augments or prevents its autophosphorylation, even though the levels are the same. However, such complexation could also affect NDPK activity. This may make the determination of [32P]NDPK-P levels a more reliable method for deter- mining actual levels of this isozymic form of NDPK since all species could contribute to NDPK activity but only this specific isozyme forms a NDPK-P in the presence of EDTA.

A 20,000 M, GTP binding protein that behaves like ras p21 has been reported to copurify with NDPK of Ehrlich ascites cells [4]. There are several reports that present data supporting the elevated expression of ras p21 in the majority of colorectal cancers [ 161 and a high level of ras gene mutations in the same tissues 132,331. However, other work- ers have demonstrated that the level of ras p21 expression is not correlated with proliferation and malignancy [ I 71. We found that under the conditions used for the phosphorylation experiments, ra5 p21 is not photolabeled by 8N3GTP in the colon carcinoma homogenates, even though it was present on western blot. Considering the low copy number and the slow guanosine nucleotide exchange rate of ras p21, it is likely that the photoprobe was not significantly bound to enough p21 molecules to allow us to observe any photoinsertion. We observed elevated NDPK-P in the presence of EDTA in 17 of 17 carcinomas. In comparison, the level of ra5 p21 was reported to be elevated in 9 of 17 primary colon carcinomas [I 61. We have also observed increased [32P]NDPK-P in malignant tissues from human breast, skin, kidney, and ovarian cancers compared with the corresponding normal tissues (data not shown). In addition, it has been shown that NDPK activity increases in mouse natural killer cells when cellular proliferation is induced by interleukin 2 [341. This indicates that specific isozymes of NDPK may play an important role in all rapidly dividing cells. Under- standing the association of NDPK with specific GTP binding proteins may lead to a better understanding of the mechanism action.

Any speculation on the involvement of NDPK-P in regulation of cellular events should be done with caution since NDPK exhibits marked heterogeneity on isoelectric focus- ing with each of these species capable of having greatly different K, values for ATP and GTP [31]. Further work to understand the involvement of NDPK-P in neoplasia requires that its specific isozyme characteristics be determined and its possible association with specific protein complexes be identified, especially those that may contain G-regulatory proteins. Experiments on colon and breast tumors are cur- rently under way.

1.

2.

3.

4.

5.

6.

7.

8.

9

10

11

12

13

14

15

16

17

18.

19.

20

REFERENCES

ACKNOWLEDGMENTS

This work was supported by grant GM 35766 from the National Institutes of Health to Boyd E. Haley.

Received December 8. 1988; revised March 30, 1989; accepted April 24, 1989.

Nickerson JA, Wells WW. Association of nucleoside diphosphate kinase with microtubules. Biochem Biophys Res Commun 85: 820-826,1978. Fenningroth SW, Kirschner MW. Nucleotide binding and Phos- phorylation in microtubule assembly in vitro. J Mol Biol 1 15:643- 673,1977. Wrtheimer AM, Kaulenas MS. GDP kinase activity asxxiated with salt-washed ribosomes. Biochem Biophys Res Commun 78:565- 571 1977 -. . , . - . . . Uesaka H, Yokoyama M, Ohtsuki K. Physiological correlation between nucleoside-di phosphate kinaseand the enzyme-associated guanine nucleotide binding proteins. Biochern Biophys Res Commun 143:552-559.1987. Stellwagen E, Baker BB. Proposed structure for brain adenylate cyclase purified using blue dextran-sepharose chromatography. Nature261 :719-720. 1976. Robinson JB Jr. Brems DN. Stellwaaen E. A monoisozvme nucleoside diphosphate kinase capable 6f complete phosphorylation. J BiolChem 256 10769-10773,1981. Kimura N, Johnson GS. Increased membrane-associated nucleoside diphosphate kinase activity as a possible basis for enhanced guanine nucleotide-dependent adenylate cyclase activity induced by picolinic acid treatment of Simian virus 40-transformed nor- malratkidneycells. JBiol Chem258.12609-12617, 1983. Francis BR, Overmeyer JH, Marshall ME, Haley BE. Phosphoryla- tion and photolabeling studies of proteins of 21 kDa and 24 kDa in human tumor and non-tumor homogenates (abstract). Thirty- Ninth Symposium of Fundamental Cancer Research. University of Texas. M.D. Anderson Hospital and Tumor Institute, Program and Abstracts. Houston, TX, 1986, p 33-34. Ohtsuki K, Yokoyama M, lkeuchi T, lshida N, Sugita K, Satoh K. Rapid induction of nucleoside-diphosphate kinase in HeLa 53 cells by human-type interferons. FEBS Lett 196: 145-150, 1986. Ohtruki K. lkeuchi T, Yokoyama M. Characterization of nucleosidediphosphate kinase-associated guanine nucleotide-binding proteins from HeLa 53 cells. Biochem Biophys Acta 882:322-330, 1986. Koyama K, Yokoyama M, Koike T, Ohtsuki K, lshida N. Nucleoside diphosphate kinase from Ehrlich ascites tumor cells. 1 Biochem Tokyo 95:925-935, 1984. Shih TY, Papageorge AG, Stokes PE, Weeks MO, Scolnick EM. Gua- nine nucleotide-binding and autophosphorylating activities associated with the ~21'" protein of Harvey murine sarcoma virus. Nature 287:686-691, 1980. Papageorge A, Lowy D, Scolnick EM. Comparative biochemical properties of p21 ras molecules coded for by viral and cellular ra5 genes. 1 Virol44:509-519,1982. Kahn RA, Goddard C, Newkirk M. Chemical and immunological characterization of the 2 1 -kDa ADP-ribosylation factor of ade- nvlate cvclase. J Biol Chem 263:8282-8287. 1988. Tiai SC,'Noda M, Adamik R, etall Stimulation of choleragen enzy- matic activities by GTP and two soluble proteins purified from bovine brain. J Biol Chem 263:1768-1772. 1988. Gallick GE. Kurzrock R, Kloetzer WS, Arlinghaus RE, Gutterman JU. Expression of p21 ra5 in fresh primary and metastatic human colorectal tumors. Proc Natl Acad Sci USA 82: 1795-2 799,1985. Chesa PG, Rettig WJ, Melamed MR, Old U, Niman HL. Expres- sion of p21 ras in normal and malignant human tissues: Lack of association with proliferation and malignancy. Proc Natl Acad Sci USA 84:3234-3238,1987. Czarnecki J, Geahlen R, Haley B. Synthesis and use of azido photoaffinity analogs of adenine and guanine nucleotides. Meth- ods Enzymol56:642-653, 1979. Potter RL, Haley BE. Photoaffinity labeling of nucleotide binding sites with 8-azidopurine analogs: Techniques and applications. Methods Enzymol91:613-633,1982. Khatoon 5, Atherton R, Al-Jumaily W, Haley BE. Use of nucleotide photoaffinity probes to study hormone action. Biol Reprod 28:61-73,1983.

178 FRANCIS ETA L.

21. Skare K, Black JL, Pancoe WL. Haley B. Determination of the cellular location of cyclic nucleotide binding sites using 8-azido- adnnosine-3’,5’-monophosphate, a photoaffinity probe. Arch Biochem Biophys 180:409-415 1977.

22. HunterT, Sefton BM. Transforming gene product of Rous sarcoma virus phosphorylates tyrosine. Proc Natl Acad Sci USA 77: 131 1- 1315,1980.

23. Rangel-Aldao R, Rosen OM. Mechanism of self-phosphorylation of adenosine 3’:5‘-monophosphate-dependent protein kinase from bovinecardiacmuscle. J Biol Chem 251 :7526-7529,1976.

24. Kreil G, Boyer PD. Detection of bound phosphohistidine in E. coli succinatethiokinase. Biochem Biophp ResCommun 16:551-555. 1964.

25. Lipmann F, Tuttle LC. A specific micromethod for the determination of acyl phosphates. J Biol Chem 159:21-28. 1945.

26. Norman AW, Wedding m, Black MK. Detection of phosphohistidine in nucleoside diphosphate kinase isolated from Jerusalem art- choke mitochondria. Biochem Biophys Res Commun 20:703-709. 1965.

27. Walinder 0, Zetterqvkt 0, Engstrom L. Purification of a bovine liver protein rapidly phosphorylated by adenosine triphosphate. J Biol Chem 243:2793-2798,1968.

28. Walinder 0. Identification of a phosphate incorporating protein from bovine liver as nucleotide diphosphate kinase and isolation

of l-32P-phosphohistidine, 3-32P-phosphohistidine, and N-e3‘P- phospholysine from erythrocytic nucleoside diphosphate kinase, incubated with adenosine triph~sphate-~~P. J Biol Chem 243: 3947-3952,1968.

29. Edlun B, Rask L, Olsson P, Walinder 0, Zetterqvist 0, Engstrom L. Preparation of crystalline nucleoside diphosphate kinasefrom baker‘s yeast and identification of 1 -[32P] phosphohistidine as the main DhosDhodated oroduct of an alkaline hvdrolvsate of enzvme incubated wiih 451-455 19159

adenosine 132Pl triphosphate.’ J Biochem 9: .__’

30. Edlun B. Purification of a nucleosidediphosphate kinase from ea seed and phosphorylation of the enzyme with adenosine lY2P] triphosphate. Acta Chern Scand 25: 1370-1376, 1971.

31. Parks RE Jr, Agarawal RP. Nucleoside diphosphokinases. In: Boyer PD (ed), The Enzymes, voI8. Academic Press, New York. 1973, p 307-33.

32. Bos JL, Fearon ER, Hamilton SR, et al. Prevalence of rasgene mutations in human colorectal cancers. Nature 327:293-297, 1987.

33. Forrester K, Alrnoguera C, Han K, Grizzle W, Pericho M. Detec- tion of high incidence of K-ras oncogenes during human colon tumorigenesis. Nature 327:298-303. 1987.

34. Ohtsuki K, Yokoyama M, lshii F, lshida N. Nucleosidediphosphate- kinase induction by human recombinant IL-2 in mouse NK cells. Biochemlnt 10:13-21, 1985.

prevalence of nucleoside diphosphate kinase autophosphorylation in human colon carcinoma versus...

Documents