Basis for Resistance to 3-Deazaaristeromycin, an Inhibitor of S-Adenosylhomocysteine Hydrolase, in Human B-Lymphoblasts"

Clones resistant to 3-deazaaristeromycin, a potent inhibitor of S-adenosylhomocysteine hydrolase, were selected from a nucleoside kinase-deficient derivative of the WIL-2 human B-lymphoblastoid cell line. The resistant clones took up 3-deazaaristeromycin and showed no alteration in the level of S-adenosylhomo- cysteine hydrolase activity or in the sensitivity of the enzyme to inhibition by 3-deazaaristeromycin. 'How- ever, they displayed markedly elevated S-adenosyl-methionine content during growth in 3-deazaaristero- mycin and, following prolonged selection, enhanced export of S-adenosylhomocysteine. As a result they maintained a high ratio of S-adenosylmethionine to S-adenosylhomocysteine and thus were resistant to the inhibition of S-adenosylmethionine turnover and transmethylation caused by 3-deazaaristeromycin. Ex- panded S-adenosylmethionine pools declined over several weeks of nonselective growth, suggesting a meta- bolic adaptation rather than a mutational mechanism. No alterations in S-adenosylmethionine synthetase ac- tivity were found in the 3-deazaaristeromycin-resist-ant clones. S-Adenosylhomocysteine export appeared to be carrier-mediated and largely unidirectional. The resistant clones showed a 5-fold increased rate of S- adenosylhomocysteine export compared with parental cells, but a similar K,,, for intracellular S-adenosylhom-ocysteine, estimated to be -1 mM. Our results highlight the opposing effects of S-adenosylmethionine and S-adenosylhomocysteine on transmethylation and sug- gest that the ability to elevate S-adenosylmethionine pools and to export S-adenosylhomocysteine may pro- vide for homeostatic control of transmethylation in lymphoid cells when S-adenosylhomocysteine hydro- lase activity is limited.

enosylhomocysteine (AdoHcy) as a product and competitive inhibitor of all transmethylation reactions are potentially regulators of methylation-dependent cellular processes (1,2). In mammalian cells, the only metabolic route of AdoHcy elimination is hydrolysis to Ado and homocysteine, a reversible and thermodynamically unfavorable reaction catalyzed by S-adenosyl-L-homocysteine hydrolase (AdoHcyase; EC 3.3.1.1) (3,4). Nucleoside analog inhibitors of AdoHcyase with antiviral and antineoplastic activity have been identified (5, 6). Naturally occurring nucleosides can also influence AdoHcyase activity with potentially important clinical consequences.
AdoHcyase is inhibited by Ado (7)(8)(9) and inactivated by dAdo (10)(11)(12). Thus, impaired catabolism of AdoHcy may contribute to the failure of lymphoid development caused by inherited deficiency of adenosine deaminase (EC 3.5.4.4). AdoHcyase activity is significantly decreased in erythrocytes and marrow cells of patients with adenosine deaminase deficiency (13-15), and AdoHcy accumulates to levels capable of inhibiting methylation reactions in the lymphoblasts of leukemia patients treated with the adenosine deaminase inhibitor 2'-deoxycoformycin (16,17). Studies of AdoMet utilization in uitro suggest that lymphocytes may have a greater requirement for transmethylation than other cell types (18), so that diminished AdoHcyase activity might be particularly detrimental to immune function.
In order to dissociate their AdoHcyase-dependent and nucleotide-dependent actions, we have previously studied Ado and dAdo in adenosine deaminase-inhibited mutants of the WIL-2 human B-lymphoblastoid cell line that lack Ado kinase and deoxycytidine kinase (9,11,19). To more specifically investigate the consequences of selective AdoHcyase inhibition, we have isolated clones of Ado kinase-deoxycytidine kinase-deficient WIL-2 cells resistant to 3-deazaaristeromycin (C3Ari), a potent inhibitor of AdoHcyase that undergoes only limited deamination and phosphorylation (20,21). C3Ariresistant clones exhibited an unusual, longterm adaptation in AdoMet metabolism, and enhanced cellular export of AdoHcy. These findings highlight the opposing roles of AdoMet and AdoHcy as regulators of transmethylation and suggest a basis for homeostatic control of this essential process in lymphoid cells.
for mycoplasma by determining the ability of culture medium to catalyze the conversion of adenosine to adenine in the presence of an inhibitor of adenosine deaminase (23). Isolation of cells resistant to C3Ari is described under "Results." Effect of Nucleosides on Cell Growth-Cell cultures (initially IO5 cells/ml) were grown for 3-5 days in the absence (control) or the presence of nucleoside analogs, and cell titer was determined daily with a Coulter Model ZBI particle counter (Coulter Electronics, Hialeah, FL). After subtracting the initial cell density from the final cell density (on day 3 or 4), "relative growth" was calculated as the ratio of the final cell density in wells containing nucleosides to this value for the control. Doubling times were estimated from the period of linear growth (semilog plots of cell density versus time) over the course of these experiments.
Assessment of Nucleoside Analogs as Substrates for AdoHcyase-Nucleosides (200 pM) were incubated with 6 pg of purified human placental AdoHcyase (27) and 2.5 mM homocysteine, in 25 mM Tris-HCl, pH 7.4, 1 mM EDTA, 1 mM dithiothreitol, 1 p~ 2'-deoxycoformycin at 37 "C for 60 min. Reactions were terminated by addition of 2 M HC10, to give 0.25 M, and placed on ice. Extracts were centrifuged and supernatants analyzed by high pressure liquid chromatography (HPLC) as described below. S-Homocysteinyl derivatives were identified on the basis of their having UV absorption characteristics of the parent nucleoside, their position of elution, and the dependence for their formation on both AdoHcyase and homocysteine.
Effect of Adenosine Analogs on Intracellular Concentration of AdoMet, AdoHcy, and Other Metabolites-Cells in log phase were resuspended in fresh medium at 0.3-0.7 X 106/ml and allowed to equilibrate at 37 "C for 4 h before addition of nucleosides (lox to 1 0 0~) or water. At various times aliquots of suspension were transferred to tubes on ice for 5 min, centrifuged, and an aliquot of medium was stored at -70 "C. The remaining medium was aspirated and the cells were resuspended in -1 ml of phosphate-buffered saline/bovine serum albumin, counted, and an aliquot containing 1.5-4 X lo6 cells was pelleted in a microcentrifuge for 1 min at 4 "C; the supernatant was aspirated and the cell pellet was frozen on dry ice and stored at -70 "C. Extracts of medium and cells were prepared and analyzed by HPLC as described below.
HPLC Analysis of Nucleosides, AdoMet, AdoHcy, and Polyamines-Cell pellets were extracted on ice with 0.25 M HC104. Medium samples were extracted on ice by addition of 2 M HCIOl to 0.25 M. Extracts were clarified by centrifugation at 4 "C in a microcentrifuge and analyzed directly by HPLC as described (28) with minor modifications. Briefly, 50-100 p1 of acid extract was fractionated on a CIS Novapak column (Waters Associates) at ambient temperature with a complex gradient elution at a flow rate of 1 ml/min, using the following mobile phases: Mobile phase A = 100 mM NaH2P04, pH 2.55, 8 mM heptanesulfonic acid, 0.05 mM Na,EDTA; 11 ml of acetonitrile was added to 1 liter of this buffer to give the final composition. Mobile phase B = 200 mM NaH2POI, pH 3.10, 11.4 mM heptanesulfonic acid; 700 ml of this buffer was mixed with 300 ml of acetonitrile to give the final composition. From the initial conditions (85% A, 15% B) a linear gradient was run in 15 min to 58% A and 42% B, followed by a second linear gradient to 40% A and 60% B in 3 min, and then a third linear gradient to final conditions of 25% A, 75% B. The final conditions were maintained for 3 min, followed by return to initial conditions over 1 min. The column was then reequilibrated for 9 min before the next injection. For some analyses a CIS pBondapak column (Waters Associates) was substituted and the flow rate was adjusted to 1.5 ml/min. The column effluent was monitored at 260 and 280 nm and adenine-containing compounds were quantified by comparing their peak areas to those of external standards.
For determination of the specific radioactivity, as well as the concentration, of AdoMet, AdoHcy, polyamines, and methionine in cells incubated with [14C]methionine (see below) a portion of the HPLC effluent was mixed continuously with scintillation fluid ( with acid and the specific radioactivity of the AdoMet produced was determined by radiochemical HPLC analysis as described. Turnover and Utilization of Cellular AdoMet-Cells were resuspended at 0.8-3 X 106/ml in RPMI 1640 medium containing 10% horse serum, 20 mM HEPES, pH 7.4, and adjusted to 10 p~ methionine. After equilibration at 37 'C for 4 h, ~-[3,4-'~C]methionine (57 mCi/mmol; Research Products International) was added to give 10 pCi/ml (final methionine concentration was 20-30 p~ including contribution from serum). Incubation was continued at 37 "C. Aliquots were removed immediately (t = 0) and at various times (see Fig. 7, "Results"), chilled in an isopropanol-ice slurry (-15 "C) for 20 s and centrifuged in a microcentrifuge for 45 s. Medium was removed for analysis, and the cells were washed with ice-cold phosphatebuffered saline/bovine serum albumin, centrifuged again, and the pellet was frozen on dry ice and stored at -70 "C. Alternatively, aliquots of cell suspensions were placed on top of a 0.6-ml oil cushion (90% silicone oil, 10% light paraffin oil) in a 1.5-ml microcentrifuge tube, and samples were centrifuged for -45 s at 14,000 rpm. The medium (above the oil) was collected and stored at -70 "C. The oil was aspirated and the cell pellet was frozen and stored as above. The two methods of harvesting gave equivalent results. At each time point methionine incorporation into cellular protein was measured; an aliquot of cell suspension made 10% in trichloroacetic acid was filtered onto GFC filters (Whatman). After three washes with 10% trichloroacetic acid followed by 95% ethanol, filters were dried and counted in a toluene-based scintillation fluid. Aliquots of medium were also counted directly to follow methionine uptake by cells.
Previous studies of AdoMet turnover and consumption in WIL-2 lymphoblasts demonstrated that 80% equilibration between extracellular and intracellular methionine occurs in -1 min, and the rate of methionine turnover is >lo-fold the rate of AdoMet turnover (18). In estimating AdoMet turnover we have made the simplifying assumption that this equilibrium is established at the time of addition of radiolabeled methionine. The present studies used ["Clmethionine and a final concentration of methionine in medium of 20-30 pM, while German et al. (18) added tracer amounts of [35S]methionine to medium containing 100 p~ methionine. The conditions used in the present studies permit comparable estimation of AdoMet turnover since: 1) WIL-2 maintains a normal doubling time for 3 days in as little as 10 p~ methionine; 2) >75% of medium methionine remained at the end of labeling; 3) AdoMet concentration in cells did not change during labeling; and 4) the specific radioactivity of AdoMet approached a constant value, i.e. equilibrium was established between the AdoMet pool and extracellular methionine.
Based on the above considerations (for a more detailed discussion, see Ref. 18). the relationship between radiolabel in AdoMet and extracellular methionine is given by where SA AdoMet and SA methionine are, respectively, the specific radioactivities of AdoMet and extracellular methionine, t is time, and k is the fractional turnover rate of the AdoMet pool. As noted previously (18), the specific radioactivity of AdoMet in lymphoblasts sometimes plateaus at 70-85% of the specific radioactivity of extracellular methionine. The value approached by the AdoMet pool is then substituted for the experimentally measured specific radioactivity of extracellular methionine, yielding where SA AdoMet, and SA AdoMet,.. are, respectively, the specific radioactivities of the AdoMet pool at time t and at a time when the specific radioactivity of AdoMet has reached a constant value. AdoMet consumption was estimated by multiplying the turnover rate ( k ) and the Adomet pool size (the mean of five to seven measurements of AdoMet content made at the beginning, during, and end of each experiment).

RESULTS
Characteristics of Adenosine Analogs and 3-Deazaaristeromycin-resistant (C3Arir) Clones-C3Ari and the related compounds f aristeromycin (Ari) and 3-deazaadenosine (@Ado) all inhibit AdoHcyase, but differ in their ability to undergo deamination, phosphorylation, and AdoHcyase-catalyzed condensation with L-homocysteine. We examined these aspects of metabolism in relation to their growth inhibitory effects on WIL-2 lymphoblasts. C3Ari and C3Ado were not deaminated by adenosine deaminase, while Ari was deaminated slowly (data not shown), as reported previously (30). C3Ari was converted to an S-homocysteinyl derivative by human placental AdoHcyase at a rate of 0.037 pmol/min/mg, about twice the rate with Ari and -1% the rates with Ado and C3Ado (nucleosides were 200 p~ and homocysteine of 2.5 mM). S-Nucleosidylhomocysteine derivatives were detectable in HPLC analyses of extracts of lymphoblasts incubated with Ari and C3Ari, but never exceeded 10% of AdoHcy in these cells. Metabolism by adenosine deaminase and AdoHcyase does not appreciably influence the cytoxicity of Ari and C3Ari. Thus, their toxicity was not enhanced by either an adenosine deaminase inhibitor or homocysteine thiolactone, both of which potentiate Ado toxicity to WIL-2 (9); the toxicity of C3Ado, an active substrate for AdoHcyase but not adenosine deaminase, was potentiated by homocysteine thiolactone but not 2'-deoxycoformycin (Table I).
Both C3Ari and Ari inhibited the growth of WIL-2 by 50% at concentrations between 1 and 10 p~; the ECso of C3Ado was 30-45 p~ (Fig. 1, Table I). C3Ari and C3Ado were about equally toxic to WIL-2 and K2B, its Ado kinase-and deoxycytidine kinase-deficient derivative, while Ari was more toxic to WIL-2 than K2B (Fig. 1). These results suggest a phosphorylation-dependent component of Ari toxicity in the wild type cells, while growth inhibition by C3Ari and C3Ado depends on neither Ado kinase nor deoxycytidine kinase. We decided to use K2B for isolating C3Ari-resistant cells in order to minimize the possibility of nucleotide-dependent selection pressure and to permit the eventual use of resistant cells for studies of nucleotide-independent effects of analogs that might undergo adenosine kinaseor deoxycytidine kinasedependent phosphorylation in WIL-2.
Selection for C3Ari resistance was carried out by exposing K2B over several months to increasing concentrations of C3Ari in mass culture, followed by cloning in 75 p~ C3Ari. Several resistant clones able to grow in the presence of 350 pM C3Ari were then maintained in medium containing 100 p~ C3Ari. A clone designated C3Ari"l (clone 1) was selected  Cells (-1 X lo5 cells/ml) were incubated in medium containing the indicated concentration of nucleoside. After 3-4 days, cell number was determined with a Coulter particle counter on aliquots of each culture. Relative growth is expressed as described under "Experimental Procedures." Data presented are from a single representative determination for each cell line and nucleoside. Experiments were repeated at least three times. A, C3Ari; B, Ari; C, C3Ado. for study; data are also presented for C3Arir-9 (clone 9) with similar properties. Both C3Arir clones were -25-fold less sensitive to C3Ari than parental cells (Fig. 1, Table 11); the doubling time of the resistant clones was prolonged by <2fold in the presence of -500 PM C3Ari (Fig. 2). Compared with K2B, the C3Ari' clones were also -25-fold less sensitive to Ari, and -3-fold less sensitive to Ado and to adenine arabinoside, an inactivator of AdoHcyase (10); neither clone was appreciably resistant to C3Ado (Fig. 1, Table 11). These growth characteristics have been stable over a 3-year period of observation and have been documented in cells maintained under nonselective conditions for from 10 to 50 doublings.  Fig. 3). The ratio of intracellular AdoMet to AdoHcy for K2B was 12.5 f 4.8. A fall in the AdoMekAdoHcy ratio, or methylation index (MI), correlates with the degree of inhibition of methylation reactions, and cell growth, in lymphoblasts exposed to various conditions that elevate AdoHcy (7,9,31). Our initial studies in the period after cloning suggested that the C3Ari' phenotype was due to an ability to maintain very high levels of AdoMet, and hence a near normal MI, rather than to a diminished inhibitory effect of C3Ari' on AdoHcy hydrolysis.
In five experiments carried out after removal to drug-free medium for periods from several days to several months, exposure to C3Ari (5-125 PM) caused a similar dose-and timedependent accumulation of AdoHcy in K2B and the C3Ari' clones. The AdoMet content and MI of the clones remained 2-4.5-fold higher than the parent during reexposure to C3Ari. A representative experiment involving a 4-h incubation with 5-80 p~ C3Ari is shown in Fig. 4. At the highest drug concentration, AdoHcy in K2B and clone 1 increased from basal levels of 9-12 to 230 and 210 pmol/106 cells, respectively; AdoMet in K2B increased from 200 to 315 pmol/106 cells, compared with an increase from 515 to 745 pmol/106 cells in clone 1. In this experiment, the MI of K2B fell from -16 in the absence, to 4.5 in the presence of 5 ~L M C3Ari (roughly the ECso), and to 1.4 with 80 pM C3Ari; in clone 1 the MI fell from a basal value of >20 to 3.5 at 80 pM C3Ari.
The level of AdoMet in lymphoblasts appears to increase in response to increases in the level of AdoHcy, but the relationship involves a degree of chronic adaptation. Upon short-term exposure to C3Ari, AdoMet pools in WIL-2 and K2B expanded by 1.2-2-fold, rarely approaching 3-fold, over 1-3 h of exposure to C3Ari, after which AdoMet remained stable or fell slightly over 24-48 h (Fig. 4, Table 111, and data  not shown). During chronic growth in 100 NM C3Ari AdoMet in the resistant clones rose to much higher levels (Fig. 3A,  arrow). For example, on one occasion AdoHcy levels in clones 1 and 9, respectively, were 131 and 197 pmol/106 cells; their AdoMet levels at the time were 1400 and 1440 pmol/106 cells (about one-half the level of ATP), yielding MI values of 10.7 and 7.3. After transfer to drug-free medium, AdoHcy in the C3Ari' clones fell to normal within a few hours, but AdoMet and the MI remained 2-4-fold higher than in K2B for several weeks, after which these values were variably elevated and often in the parental range (Fig. 3).
In preliminary experiments we found <%fold difference in AdoMet synthetase activity, measured at saturating or subsaturating substrate (methionine, ATP) concentrations, in extracts of K2B, clone 1 and clone 9. There was no difference in sensitivity of AdoMet synthetase from K2B and clone 1 to product inhibition by AdoMet; neither enzyme was stimulated or inhibited by C3Ari or AdoHcy at concentrations up to 0.5 mM, in reactions containing a range of AdoMet concentrations. Similar results were obtained with extracts from cells maintained under nonselective conditions and in 100 PM C3Ari. AdoHcy Efflux-After 6-7 months of continuous growth of clone 1 in 100 ~L M C3Ari, a second basis for resistance to C3Ari became apparent. During chronic growth in drug, clone 1 continued to maintain very high levels of AdoMet. However, upon short-term (up to 48 h) incubation with C3Ari (after growth from several days to several months in drug-free medium) the resistant cells now consistently accumulated less AdoHcy than parental cells. For example, in the experiment presented in Table 111, AdoHcy levels in K2B exposed to 5-100 ~L M C3Ari rose 4-15-fold, while in clone 1 AdoHcy rose to only about one-third of these levels. AdoMet pools in K2B and clone 1 differed by <25% throughout this experiment. We addressed some possible mechanisms for the diminished ability of C3Ari to elevate intracellular AdoHcy in chronically selected clone 1 cells. Uptake of C3Ari was not grossly altered (Table IV). The level of AdoHcyase activity, and the sensitivity of AdoHcyase to inhibition by C3Ari, were similar in extracts of parent and resistant cells (Table V), and they contained similar amounts of AdoHcyase protein detectable by immunoblot analysis with monoclonal antibodies to placental AdoHcyase (data not shown).
Because these results suggested that AdoHcy hydrolysis should be inhibited to the same degree in the resistant and parental cells, we measured the level of AdoHcy in culture medium of cells exposed to C3Ari. For the experiment described in Table 111, the total AdoHcy production (AdoHcy in cells plus AdoHcy in culture medium) was indeed greater for clone 1 than for K2B, but the resistant clone excreted a greater fraction of the AdoHcy at all concentrations of C3Ari (Fig. 5A). Fig. 5 3 shows that, for this experiment and for another, similar experiment, clone 1 released AdoHcy to the medium at lower intracellular AdoHcy levels than K2B. In experiments not presented we found similar AdoHcy excretion by K2B and WIL-2 (detectable only in the presence of an inhibitor of AdoHcyase). Since AdoHcy efflux was not   examined in experiments conducted immediately after cloning, it is unclear when increased AdoHcy export began to occur, or whether the rate of export increased gradually or as a single-step event in the course of selection. We explored the relationship between the intracellular concentration of AdoHcy and the rate of AdoHcy export (Fig. 6). A range of intracellular AdoHcy concentrations was first established by incubating K2B and clone 1 with an adenosine deaminase inhibitor, various concentrations of Ado, and 200 p M homocysteine thiolactone. Then the cells were transferred to medium lacking these compounds and containing 200 p M C3Ari (to limit AdoHcy hydrolysis) and the intracellular and extracellular concentrations of AdoHcy were followed over 30 min (AdoHcy efflux had been shown to be linear for 60 min in preliminary experiments not shown). In both cell lines intracellular concentrations of AdoHcy ranged from near 0.1 to near 10 nmol/106 cells after the initial incubation. As intracellular AdoHcy increased, the rate of AdoHcy efflux increased and then saturated in both cell types, indicating a carrier-mediated process. Half-maximal rates of efflux occurred at intracellular AdoHcy concentrations of approximately 2 nmol/1O6 cells for clone 1  An indication of the efficiency of AdoHcy export by clone 1 was the finding that in the presence of 5 p M 2'-deoxycoformycin, but in the absence of an AdoHcyase inhibitor, clone 1 excreted >1 nmol of AdoHcy/106 cells in 24 h, while AdoHcy excretion by K2B was not detectable (cO.09 nmol of AdoHcy/ lo6 cells) ( Table VI). Under these conditions in adenosine deaminase-inhibited, adenosine kinase-deficient cells, a low level of AdoHcy accumulation results from the effect of endogenously produced Ado on the equilibrium of the reversible AdoHcyase reaction (11). Enhanced AdoHcy efflux also enabled clone 1 to maintain lower levels of AdoHcy than K2B when the cells were incubated with Ari and C3Ado. Clone 1 accumulated less C3AdoHcy from C3Ado and excreted more  6. Export of AdoHcy as a function of intracellular AdoHcy. Intracellular AdoHcy was increased to varying levels (9) by incubating suspensions of K2B (squares) and C3Arir clone 1 (circles) (4-5 X lo6 cells/ml) for 90 min at 37 "C in medium containing 5 p~ erythro-9-(2-hydroxy-3-nonyl)adenine (a reversible inhibitor of adenosine deaminase), 200 pM homocysteine thiolactone, and 0-200 PM Ado. Aliquots (9 ml) of these suspensions were then centrifuged and the medium aspirated completely; the cells were then resuspended in 2 ml of medium lacking Ado, homocysteine thiolactone, and erythro-9-(2-hydroxy-3-nonyl)adenine, but containing 200 p M C3Ari to inhibit hydrolysis of AdoHcy. Immediately following resuspension (to) and after 10 and 30 min of further incubation at 37 "C, 0.55-ml samples were centrifuged through an oil cushion and extracts of cells and medium were prepared and analyzed for AdoHcy ("Experimental Procedures"). The rate of AdoHcy excretion was determined from the net increase in medium AdoHcy during incubation of the resuspended cells (amount at 10 or 30 min minus the amount present at to). The average of the AdoHcy export rate measurements for the two time points (picomoles/106 cells/min) is plotted against the intracellular concentration of AdoHcy at to. The possibility that cell lysis contributed significantly to AdoHcy in medium was excluded since <2% of cellular purine nucleoside phosphorylase activity was released to the medium during the 30-min incubations for K2B and clone 1 cells that had been exposed to 0 or 100 p~ Ado.

TABLE VI
Efflux of AdoHcy from K2B and clone 1 3 X lo6 cells were incubated for 24 h in medium containing the indicated compounds. Cells and medium were extracted and analyzed for AdoHcy and 3-deazaadenosylhomocysteine (C3AdoHcy) as described under "Experimental Procedures." The clone l cells used in this experiment had been maintained in 100 p~ C3Ari for >6 months but had been cultured in C3Ari-free medium for 31 days before the experiment.
AdoHcy (C'AdoHcy) (nmol/106 cells) C3AdoHcy than K2B ( Table VI). Effect of PAri on AdoMet Turnover and Utilization-The ability to maintain higher MI despite inhibition of AdoHcy hydrolysis should permit C3Ari' cells to carry out AdoMetmediated transmethylations at concentrations of C3Ari that would inhibit these reactions in K2B. To test this hypothesis, we compared the effects of C3Ari on AdoMet turnover and utilization in cells labeled with the AdoMet precursor ~-[3,4-"Clmethionine.
In preliminary studies with WIL-2, experimental conditions were established such that total uptake of methionine was linear for 2 h (-1 nmol/106 cells/h), and <20% of medium methionine was used by 3 h; levels of AdoMet and AdoHcy were not altered, and incorporation of methionine into protein was linear throughout the labeling period. We measured the approach to constant specific activity of the AdoMet pool and from these data estimated the fractional AdoMet turnover rate and rate of total AdoMet utilization to be 0.045-0.081 min" and 7.7-9.0 pmol/106 cells/min, respectively, in reasonable agreement with values of 0.076 min" and 7.4 pmol/106 cells/min reported by German et al. ( not affect the rates of protein or polyamine synthesis by more than 8%. Therefore, it can be concluded that inhibition of AdoMet utilization in cells exposed to C3Ari is due entirely to inhibition of transmethylation, with negligible effect on polyamine synthesis.* A representative study of AdoMet turnover in K2B and clone 1 is shown in Fig. 7. In the untreated cells, AdoMet was 1.8-fold higher in clone 1 than K2B, 295 versus 160 pmol/1o6 cells. Fractional turnover rates of AdoMet pools were 0.049 min" and 0.027 min" for K2B and clone 1, so that basal rates of AdoMet consumption for K2B and clone 1 were nearly identical, 7.4 and 7.3 pmol/min/106 cells, respectively. A 4-h incubation with 100 p~ C3Ari increased intracellular AdoHcy from basal levels of 35 pmol/106 cells to 297 (K2B) and 82 (clone 1) pmol/106 cells; AdoMet increased by 2-fold to 326 pmol/106 cells in K2B, and by 37% to 404 pmol/106 cells in clone 1. AdoMet and AdoHcy levels were constant during the subsequent turnover studies. C3Ari did not affect the plateau value for specific activity of AdoMet,., (Fig. 7, A and B ) .
Under these conditions consumption of AdoMet by K2B fell to 4.0 pmol/min/106 cells (Fig. 7A), and in clone 1 to 5.7 pmol/rnin/lO' cells ( Fig. 7B). Based on the considerations outlined in the previous paragraph, we estimate that transmethylation was inhibited in K2B by 54%, and by 28% in clone 1.
The increased ability of clone 1 to export AdoHcy raised the possibility that these cells might also have an enhanced ability to take up AdoHcy. However, significant uptake of AdoHcy by either cell line seems unlikely since addition of 240 p~ AdoHcy to incubation medium had no effect on AdoMet turnover (Fig. 7). For comparison, the intracellular concentrations of AdoHcy in C3Ari-treated K2B and clone 1 were approximately 180 and 50 p~. The fact that exogenous AdoHcy did not potentiate the inhibitory effect of C3Ari on AdoMet utilization also implies that efflux of AdoHcy is unaffected by extracellular AdoHcy, i.e. that AdoHcy export can probably occur against a concentration gradient.

DISCUSSION
Resistance to C3Ari could have arisen from defective uptake of C3Ari, increased expression of AdoHcyase or a decrease in its sensitivity to inhibition by C3Ari, or from some other mechanism that alleviates toxic consequences of AdoHcyase inhibition. Considerable evidence suggests that inhibition of methyltransferases, some of which have Ki values for AdoHcy in the submicromolar range, is a major cause of AdoHcy toxicity (2,5,7,9,11,33). However, 100 p M C3Ari apparently inhibited the growth of a mouse macrophage cell line by preventing the homocysteine-dependent regeneration of tetrahydrofolate from 5-methyltetrahydrofolate, leading to a block in de nouo purine nucleotide and thymidylate synthesis (21). AdoHcy at high concentration may affect CAMP (34) and phosphoinositide metabolism (35), though the relationship of these effects to AdoHcy toxicity is unclear. We found neither a defect in C3Ari uptake nor any change in the expression or properties of AdoHcyase in C3Ari' cells. WIL-2 lymphoblasts do not remethylate homocysteine (18), and in studies not presented neither homocysteine nor hypoxanthine and thymidine protected WIL-2 or K2B from C3Ari toxicity.
Gordon et al. (32) suggested that C3Ari could inhibit polyamine synthesis by directly inhibiting AdoMet decarboxylase. They reported that 4 PM C3Ari inhibited decarboxylation of AdoMet in HeLa cells. However, we observed no effect on spermidine or spermine synthesis in WIL-2 incubated with 50 p~ C3Ari, a concentration 10 times the ECw.
We have not explored the effects of C3Ari on CAMP or phosphoinositide metabolism.
C3Ari-resistant clones displayed two properties, expanded AdoMet pools and an enhanced ability to excrete AdoHcy, that implicate inhibition of methylation as the basis for C3Ari (and AdoHcy) toxicity. Each of these metabolic alterations had the effect of maintaining a high ratio of AdoMet to AdoHcy despite a block in AdoHcy hydrolysis, diminishing the inhibition of overall cellular transmethylation by C'Ari. These results are consistent with the previous finding of an inverse correlation between the AdoMet:AdoHcy ratio (MI) and degree of inhibition of specific DNA and RNA methylation reactions in human lymphoid cells (9,16).
AdoMet pool expansion during chronic growth in C3Ari can be partly attributed to decreased AdoMet consumption caused by AdoHcy-mediated inhibition of transmethylation (Fig. 7). Despite the rapid return of AdoHcy to normal, AdoMet levels remained elevated for several weeks after removal to C3Arifree medium. In the absence of ongoing inhibition of AdoMet consumption this suggests that increased AdoMet synthesis may also have contributed to pool expansion. We found no increase in AdoMet synthetase activity in extracts of chronically selected cells, and neither C'Ari nor AdoHcy stimulated AdoMet synthetase or diminished its sensitivity to product inhibition by AdoMet. Nevertheless, the enzyme from human lymphoid cells displays very complex kinetics, with some 60 terms in the reaction equation (24). The activity of the enzyme in intact cells may not be accurately reflected under conditions of in vitro assay. Stable AdoMet elevation has previously been observed in Ado-resistant mouse lymphoma cells, which were cross-resistant to C3Ari (33). Minor differences in the level and properties of AdoMet synthetase in the variant cells were reported. AdoMet pool expansion and elevated AdoMet synthetase activity were found in hamster cell lines selected for resistance to cycloleucine, a methionine analog that is both an inducer and an inhibitor of AdoMet synthetase in those cells (36,37).
The attenuation of AdoMet pool expansion during nonselective growth of C3Ari' clones in the present study argues against a simple mutational mechanism. It may be worth considering how prolonged accumulation of AdoHcy during C3Ari selection might cause an adaptation affecting AdoMet synthetase activity in intact cells, which could persist for a time after resumption of normal culture conditions. Several isozymes of AdoMet synthetase, with differing kinetic and physical properties, have been characterized (24, [38][39][40][41][42]. The purified enzyme from human lymphocytic leukemia cells is tetrameric with an azB2, or aa'Bz structure (24). The presence of distinct subunits and electrophoretic heterogeneity of one subunit type raised the possibility of posttranslational modification related to a regulatory function. Conceivably, prolonged elevation of AdoHcy could lead to hypomethylation, and hence activation, of a gene coding for a new AdoMet synthetase isozyme with a higher basal rate of activity. Alternatively, submaximal activity of AdoMet synthetase might normally be maintained by AdoMet-mediated methylation of amino acid residues of a regulatory subunit. Chronic AdoHcy accumulation could lead to subunit undermethylation, thus increasing basal AdoMet synthetase activity. Elimination of C3Ari would lead to a gradual remethylation of the isozyme gene or AdoMet synthetase, eventually extinguishing the elevation in AdoMet.
AdoHcy is normally hydrolyzed efficiently, and some cells depend upon the homocysteine derived from the AdoHcyase reaction for the maintenance of normal levels of methionine and folate. Thus, the need for a system to export intact

Resistance to 3-Deazmristeromycin in B-Lymphoblasts 803
AdoHcy from the cell is not obvious. However, previous reports have demonstrated that AdoHcy was released from hepatocytes and some other cells when marked intracellular AdoHcy accumulation was induced (43,44). We observed that plasma AdoHcy rose from undetectable to 1-2 PM in a pateint who was treated with the combination of 2'-deoxycoformycin and adenine arabinoside (16). This was associated with a 12fold increase in lymphoblast AdoHcy concentration. It has been suggested that AdoHcy export may provide a mechanism for escaping the toxic effects of AdoHcy (45). The present studies directly support this possibility. AdoHcy was released from K2B and WIL-2 when intracellular AdoHcy rose to very high levels, but efflux was insufficient to protect these cells from AdoHcy toxicity. The C3Ari' clones, on the other hand, excreted AdoHcy at much lower cellular concentrations, effectively enough to limit AdoHcy accumulation and toxicity when AdoHcyase was severely inhibited. The nature of the AdoHcy transport mechanism is poorly understood at present. AdoHcy does not easily enter cells (46,47). Our studies indicate this as well: high concentrations of extracellular AdoHcy had no effect on intracellular AdoMet utilization, while lower levels of intracellular AdoHcy induced by inhibition of AdoHcyase clearly inhibited AdoMet turnover and methylation. Our studies suggest that AdoHcy export can occur against a concentration gradient and may therefore be an active, and possibly unidirectional, process. These characteristics resemble the export of CAMP described in several systems (48) but contrast with the nonconcentrative, bidirectional transport of nucleosides, dediated by facilitated diffusion (49-51). In preliminary studies, AdoHcy export was insensitive to the potent nucleoside transport inhibitor nitrobenzylthioinosine.
AdoHcy export may be mediated by a transport system for CAMP, or for methionine, homocysteine, or some other amino acid, or it may represent a previously undefined transporter. C3Ari' cells were isolated by prolonged, stepwise selection, originally intended as a means of isolating overproducers of AdoHcy hydrolase. It is possible that the &fold increase in maximal rate of AdoHcy export is due to amplification in expression of a protein directly involved in AdoHcy export. C3Ari' clones should provide a useful system for further biochemical and genetic analysis of the AdoHcy export system, as well as for study of the regulation of AdoMet pools and the homeostatic control of transmethylation in lymphoid cells.