Coupled ribonucleoside diphosphate reduction, channeling, and incorporation into DNA of mammalian cells.

There is rapid and specific channeling of ribonucleoside diphosphates into DNA through reactions beginning with ribonucleotide reductase and terminating with DNA polymerase. Lysolecithin-permeabilized Chinese hamster embryo fibroblasts in culture rapidly reduced ribonucleoside diphosphates by ribonucleotide reductase action when dithiothreitol was provided as a reducing agent and incorporated these deoxynucleotides into DNA. The radioactive label provided in ribo-CDP was not diluted by added deoxyribo-CTP during its incorporation into DNA, showing that the ribo-CDP does not pass through a deoxy-CTP pool. Under the conditions that permitted rapid incorporation of ribonucleoside diphosphates, deoxynucleoside triphosphates were very poorly incorporated. Ribonucleotide reductase with the rate-limiting enzyme for the overall process. The Km values for the reductase reaction and the overall process were similar and low enough for saturation by in vivo pools. Natural feedback inhibitors dATP or dTTP inhibited incorporation of labeled ribo-CDP into deoxyribonucleotides and into DNA to the same extent. Ribonucleotide reductase behaved like other enzymes that are associated in a rapidly sedimenting form. It was concentrated in the nucleus during S phase, and most of the enzyme activity in these nuclear extracts was co-sedimented with DNA polymerase on sucrose density gradients. These data support the hypotheses that a physically associated complex of enzymes (replitase) catalyzes the production of deoxynucleotides and their incorporation into DNA in S phase cells.

There is rapid and specific channeling of ribonucleoside diphosphates into DNA through reactions beginning with ribonucleotide reductase and terminating with DNA polymerase. Lysolecithin-permeabilized Chinese hamster embryo fibroblasts in culture rapidly reduced ribonucleoside diphosphates by ribonucleotide reductase action when dithiothreitol was provided as a reducing agent and incorporated these deoxynucleotides into DNA. The radioactive label provided in ribo-CDP was not diluted by added deoxyribo-CTP during its incorporation into DNA, showing that ribo-CDP does not pass through a deoxy-CTP pool. Under the conditions that permitted rapid incorporation of ribonucleoside diphosphates, deoxynucleoside triphosphates were very poorly incorporated.
Ribonucleotide reductase was the rate-limiting enzyme for the overall process. The K,,, values for the reductase reaction and the overall process were similar and low enough for saturation by in vivo pools. Natural feedback inhibitors dATP or dTTP inhibited incorporation of labeled ribo-CDP into deoxyribonucleotides and into DNA to the same extent. Ribonucleotide reductase behaved like other enzymes that are associated in a rapidly sedimenting form. It was concentrated in the nucleus during S phase, and most of the enzyme activity in these nuclear extracts was co-sedimented with DNA polymerase on sucrose density gradients. These data support the hypotheses that a physically associated complex of enzymes (replitase) catalyzes the production of deoxynucleotides and their incorporation into DNA in S phase cells.
The replicative synthesis of DNA by mammalian cells exhibits numerous features that are not readily understandable if the reactions of precursor synthesis and polymerization are catalyzed by soluble enzymes. DNA synthesis is abruptly initiated in each cell cycle (at the beginning of S phase) after several hours of RNA and protein synthesis (during GI phase), and yet neither DNA polymerase nor its substrates are limiting prior to initiation (1,2 ) . Chain elongation, in vivo, by incorporation of dNTPs' occurs at a rate several times faster than for most macromolecular biosynthetic processes, including the chemically similar process of RNA synthesis (see Refs. 3 and 4). The average concentrations of dNTPs existing in a cell during the S phase are quite low (5) relative to the average K , of DNA polymerase for in situ semiconservative DNA synthesis (6) which is 4-to 5-fold higher. Furthermore, radioactivity from rNDPs was incorporated into DNA of permeabilized cells more effectively than from dNTPs (7), under the conditions suitable for the incorporation of each nucleotide (8). Under conditions which included dithiothreitol as a reducing agent to permit reduction of rNDPs (7), dNTPs were incorporated even less rapidly.
These observations can be accounted for on the basis of a multienzyme complex which functions to channel rNDPs through a pathway starting with ribonucleotide reductase (2'deoxyribonucleoside diph0sphate:oxidized thioredoxin 2'-oxidoreductase, EC 1.17.4.1) and terminating with DNA polymerase. Evidence for such complexes has accumulated over 10 years, in prokaryotic and eukaryotic cells, both virus infected and uninfected (see Ref. 9; also Refs. [10][11][12][13][14]. We have demonstrated that in mammalian cells more than a half-dozen enzymes of the DNA synthetic pathway can be co-sedimented on sucrose gradients. We proposed a multienzyme complex that we have named replitase ( 7 ) . It was found in mammalian cells that were synthesizing DNA semiconservatively and located in the nuclear fraction. Cells that had not initiated DNA synthesis did not possess replitase, and the corresponding enzymes were found unassociated and in the cytoplasm. Such a complex might function to rapidly and specifically incorporate rNDPs into DNA (following their reduction) and exclude dNTPs.
We now report that permeabilized cells rapidly incorporate label from rNDPs into DNA. The rate is limited by the initial step, catalyzed by ribonucleotide reductase. The reduced nucleotide (rCDP) does not become mixed with an added pool of dCTP. We find that ribonucleotide reductase exists in the nucleus during S phase and that it can be sedimented with the other enzymes of replitase. These results indicate precursors are channeled through the replitase complex into DNA. Restrictive (isoleucine-deficient) conditions were used to synchronize the growth of CHEF/18 cells in culture as described (7). Cell synchrony was determined by two different criteria (7): 1) flow microfluorometry and 2) autoradiography. More than 90% of Chinese hamster embryo fibroblast cells maintained in isoleucine-deficient medium for 40 h were in a quiescent state with a GI/Go DNA content, without revealing any morphological signs of deterioration. Upon replacement of isoleucine-deficient medium with complete medium these cells progressed synchronously into S phase, all being in S phase in about 12 h. In all the experiments S phase cells were used unless otherwise noted.

Materials
Permeabilization of Cells by L-Lysophosphatidylcholine-Synchronized cells in S phase were trypsinized and collected by centrifugation. These cells were then washed and permeabilized as described by Miller et al. (17). The final concentration of L-lysophosphatidylcholine used to permeabilize the cells was 250 p g / d .
DNA Replication in Permeabilized Cells-When DNA synthesis from rNDPs was measured the incubation mixture in 0.3 ml contained 50 mM Hepes (pH 7.4). 10 (18) and DNA synthesis (8). Incubations were carried out for 10 min at 37 "C, and acidprecipitable alkali-resistant material was prepared by a method modified from Fridlender et al. (19). 75-pl aliquots of incubation mixture were added to 15 pl of 60% perchloric acid and 0.1% sodium pyrophosphate and kept on ice for 15 min. This mixture was diluted by adding 1 ml of distilled H20 and centrifuged to pellet acid-precipitable material. This pellet was extracted with 0.1 ml of 0.2 M NaOH and incubated at 37 "C for 30 min. This whole process of perchloric acid precipitation and NaOH extraction was repeated again. Then 75-pl aliquots of the NaOH suspension in 5 ml of Biofluor (New England Nuclear) were counted for radioactivity with a Beckman scintillation counter LS9000. One unit of specific activity is 1 pmol/min of incorporation.
When DNA synthesis from dNTPs was measured, the incubation mixture was essentially as described by Castellot  and dGTP, and 1 X lo7 permeabilized cells. Further steps were the same as above.
dTTP and rCDP Turnover during DNA Replication in Permeabilized CHEFI18 Cells-75-pl aliquots of incubation mixture for DNA replication, containing either ['HIdTTP when dNTPs were used as precursors or ['HH]rCDP when rNDPs were used as precursors, were added to 150 pl of ice-cold methanol. After sitting on ice for 2 h, each mixture was centrifuged and 10 pl of supernatant along with appropriate markers were spotted on polyethyleneimine cellulose chromatographic thin layers. Thymidine and cytosine nucleotides were separated and quantitated as described by Mathews (20).
Preparation of Lysates-Gentle preparation of nuclear and cytoplasmic lysates is essentially as described elsewhere (7). Any modifications made in this method are described in corresponding figure legends.
Enzyme Assays-Ribonucleotide reductase activity in either nuclear or cytoplasmic lysates was measured in a reaction mixture of 50 m M Hepes (pH 7.21, 8 mM dithiothreitol, 10 m MgCI2, 6 mM rATP, 60 p M FeCb, 8 nm NaF, 100 p~ [I4C]rCDP (8000 cpm/nmol), and 60 p1 of lysate fraction (made from 1 X IO8 nuclear or cytoplasmic particles/ml) in a final volume of 150 p1. After incubating at 37 "C for 30-60 min, the reaction was terminated by heating for 3 min in boiling H20. The deoxycytidine formed was measured using Dowex 1 borate columns as described by Steeper and Steuart (21). When this enzyme was measured in permeabilized cells either the same reaction mixture or a modified DNA replication reaction mixture (described above) was used, as described in the appropriate figure legends.
DNA polymerase activity was measured as described elsewhere (7). Protein concentration was estimated by the method of Lowry et al. (22).

RESULTS
Incorporation of rNDPs into D N A of Permeabilized Cells-Incorporation of [3H]rCDP into DNA was rapid ( Fig.  1) as previously reported (7). The V,,,,, was 2.5 units, calculated from a double reciprocal plot that gave a straight line (Fig. 2 ) . This value is comparable to the in vivo rate of about 5 units (17). The K, for this process was 100 p~, determined in terms of added rCDP. However, rCDP was rapidly converted to mono-and triphosphates in these reaction mixtures (Fig. 3). Thus, the average rCDP concentration was lower during this 10-min assay, and correspondingly the K , was nearer to 50 ,EM.
The incorporation of rNDPs into DNA could be limited by their rates of reduction. We, therefore, measured kinetics of ribonucleotide reductase in the same preparation. Using rCDP as a substrate V,,,,, was 3.0 units and K , was 100 ~L M (Fig. 2 ) (uncorrected). This value is similar to the K , of 94 ,UM reported for crude sonicates of DON hamster fibroblast cells (23) and 130 p~ for permeabilized CHO cells (24) ribonucleotide reduc-  When the rate of rCDP reduction was measured the incubation mixture used was the same as described for DNA synthesis in the presence of rNDPs. Therefore, the rate expressed is the sum of dC that aDDeared in the nucleotides and DNA. tase. Lower values, ranging from 7 pM for rCDP to 80 pM for rADP have been reported for highly purified ribonucleotide reductase obtained from Molt-4F cultured human cells (25).
Comparison of these K , values with reported intracellular ribonucleotide pool concentrations in the range of 0.2 to 1 mM (1,26) leads to the conclusion that pools in cells could give a maximal rate for such an incorpation of rNDPs into DNA.
If ribonucleotide reductase is rate limiting for incorporation of rNDPs into DNA, then any inhibition of this activity should decrease incorporation into DNA, and the two activities should be inhibited in proportion. We, therefore, tested inhibitions with dATP and dTTP, compounds that participate in the elegant feedback control of ribonucleotide reductase (27). With permeabilized CHEF/18 cells 200 p~ dATP inhibited [14C]rCDP incorporation into DNA by about 90% (Table I); dTTP similarly inhibited [I4C]rCDP incorporation over a concentration range (Fig. 4), reaching 50% at 150 p~ d T T P (Fig. 4). In good agreement with these data, assays showed quantitatively similar inhibitions of rCDP reduction in the presence of dATP or dTTP (Table I).
Ribonucleotide reductase might reduce rNDPs and then release the dNDP products into a general metabolic pool upon which DNA polymerase draws. We would then expect decreased incorporation of [I4C]rCDP into DNA of permeabilized cells if nonradioactive dCTP is added because of dilution of the labeled dCTP intermediate. But if rCDP is incorporated into DNA through a channeling process that excludes dNTPs there should be major effects due to dilution of label upon including dNTPs in the reaction mixture. This experiment must be done with rCDP and dCTP, since only dCTP does not strongly inhibit ribonucleotide reductase (Table I), in contrast to the other dNTPs (27). Addition of up to 200 ~L M dCTP to a reaction containing 200 p~ [ 14C]rCDP only slightly (8%) decreased incorporation of label into DNA (Fig. 4). Thus, rCDP must be converted to dCDP and incorporated into DNA without mixing with the general dCTP pool.
It is possible that dCDPs formed due to a "free" ribonucleotide reductase activity are released into a general metabolic pool in the incubation mixture, and these noncompartmentalized dNDPs are then channeled into DNA due to a tight coupling between deoxynucleoside diphosphate kinase (NDP kinase) and DNA polymerase. In order t o clarify this and to understand if rNDP channeling into DNA was in fact due to the direct interaction between ribonucleotide reductase and the rest of the replication complex (containing DNA polymerase and NDP kinase) we have tested whether dNDPs added exogenously can equilibrate with radioactive dNDPs produced by ribonucleotide reductase from the corresponding ribo-NDPs. As shown in Fig. 4, exogenously added dCDP, similarly to exogenously added dCTP, did not equilibrate with endogenously generated radioactive dCDP and did not decrease the incorporation of [ 14C]rCDP into DNA during an initial 10-min incubation. These observations reveal that during active DNA replication in lysolecithin-permeabilized cells at least three enzymes, ribonucleotide reductase, NDP kinase, and DNA polymerase, remain tightly coupled to channel rNDPs into newly replicating DNA, preventing free diffusion of endogenously generated dNDP and dNTP intermediates.
Intracellular Localization of Ribonucleotide Reductase-The simplest explanation for the above channeling results is that ribonucleotide reductase is physically closely associated with later enzymes in the pathway leading to DNA synthesis. We have reported earlier that during the S phase a half-dozen of these enzymes are found in the nucleus and are in a rapidly sedimenting fraction which we named replitase (7). In contrast, in cells not synthesizing DNA these enzymes are in the cytoplasm and are not rapidly sedimented.
For rNDPs to be channeled into DNA through a sequence initiated by reduction, ribonucleotide reductase must be a component of replitase. Ribonucleotide reductase should then appear in the nucleus around the beginning of S phase, as do other replitase enzymes (7). Fig. 5 shows this is the case. In a synchronized population, S phase commenced at 6 h. Up to this time, ribonucleotide reductase activity was distributed between cytoplasm and nucleus. But after 6 h all extra enzyme, which was rapidly formed, appeared in the nucleus. Ribonucleotide reductase must be synthesized on ribosomes in the cytoplasm, yet it rapidly migrated into the nucleus a t the beginning of S phase, as did other replitase enzymes (7). ]rCDP incorporation into DNA in permeabilized CHEF/18 cells. Synchronized CHEF/18 cells in S phase were permeabilized and incubated for DNA synthesis in the presence of rNDPs and appropriate effectors (dTTP or dCDP or dCTP). Incubations were carried out for 10 min at 37 "C and acid-precipitable and alkali-resistant material was prepared. 0, represent the repeated experiment for the effect of dCTP. Kinetics of Incorporation of dNTPs into DNA-Permeabilized cells rapidly incorporate dNTPs into DNA, as previously reported (6,17). From Fig. 1, V,,, was about 1.2 units, and K,,, was 50 ~L M in agreement with earlier reports (6). Stability of dTTP was relatively high compared to rCDP (Fig.  3). Deoxyribopyrimidine triphosphatase activity was reported to be absent from uninfected baby hamster kidney cells (28). The substrates added thus remained at relatively constant concentrations and provided a suitable basis for calculation of K,,,. The pools of dNTPs reported for Chinese hamster ovary cells in S phase (5 to 20 p~) ( 5 ) are well below this K,,, value and are inadequate to give a rate of incorporation comparable to the rate of DNA synthesis ( 5 units) maintained by intact cells.
The maximal rate of incorporation of rNDPs into DNA was more rapid than the rate of dNTP incorporation (Fig. I), as previously noted (7). This result is hard to explain if a free pool of dNTPs must be created from rNDPs. When dNTPs were incorporated using the conditions devised for the rNDP assay the rate was decreased by about 2-fold (as previously observed) (7). Systematic changes from rNDP to dNTP assay conditions which differ in concentrations of six components, in duplicate experiments, revealed that the most important factor for inhibition of dNTP incorporation was dithiothreitol, with smaller effects from KC1 or MgC12 (Table 11). Dithiothreitol was shown not to inhibit dTTP incorporation into DNA by a soluble enzyme preparation from these cells (7). We propose that an allosteric interaction exists between ribonucleotide reductase and DNA polymerase in the replitase complex, such that activation of the former enzyme with dithiothreitol (which acts as hydrogen donor) prevents dNTPs from reaching the polymerase active site. Consistent with this hypothesis is our previous finding that HU, a ribonucleotide reductase inhibitor, partly reverses the inhibition of dTTP incorporation created by dithiothreitol (7).
An alternative explanation for these effects is that a competition exists between the supplied dNTPs and those formed from added rNTPs when reductase is active and HU absent, such as to dilute the pool of added dNTPs. To investigate this possibility, the other results shown in Table I11 were obtained. The effects of different conditions on incorporation of dNTPs, using ['HITTP as a label, are shown in the fiist section of Table 111. Incorporation in the absence of dithiothreitol was about half as rapid as for rNDP incorporation (plus dithiothreitol). In the presence of dithiothreitol, ['HI TTP incorporation was only about 40 to 50% of that observed in the absence of dithiothreitol. This inhibition of dithiothreitol cannot be because of competitive production of dNTPs from pyrimidine ribonucleoside triphosphates, since their presence or absence had practically no effect.
In the second section of Table I11 one sees that the rNTPs, using ['HIrATP as a label, are very slowly incorporated into DNA, again reflecting slow conversions to the diphosphates. Ribotriphosphates, therefore, could not provide a precursor pool competitive with added dNTPs, even when ribonucleotide reductase was active. These overall rates are so small that conclusions from their variations are not possible.
The last column showed that [I4C]rCDP incorporation was rapid in the presence of dithiothreitol and much slower when dithiothreitol was omitted or HU was present or when both changes were made. These results support the rate-limiting role of ribonucleotide reductase in the process. Pyrimidine nucleoside triphosphates did not compete with the diphosphates, indicating that they are not readily converted to rNDPs.
Finally, HU dramatically increased the rate of dTTP incorporation in the presence of dithiothreitol, even when pyrimidine ribonucleotides were absent. This HU effect thus could not be to block de nouo production of dNTP.
All of these results are consistent with the hypothesis that accessibility of DNA polymerase to dNTP substrates into these permeabilized cells is allosterically determined by the state of ribonucleotide reductase. They are not consistent with interaction of the two enzymes through modifications of free pools of dNTPs.

DISCUSSION
Our main conclusion from this work is that Chinese hamster cells synthesize DNA by a sequestered pathway that starts with rNDPs (see Fig. 6 for a schematic representation). The rate-limiting step is catalyzed by ribonucleoside diphosphate reductase. The metabolic intermediates of this pathway do not mix with free dNTPs, but rather are "channeled" directly into DNA. Free dNTPs, present in S phase cells at low concentrations, are not effectively incorporated. They may function as allosteric regulators of ribonucleotide reductase or as substrates for repair of DNA.
Kinetic data obtained with permeabilized cells provide evidence for these conclusions. These cells incorporated rNDPs into DNA approximately as rapidly as intact cells incorporated added nucleosides (see also Ref. 17). The K , for the overall process is low enough that the pools of rNDPs found in S phase cells saturate the process to give maximal activity.
This K, was the same as the K,,, for ribonucleotide reductase.
The overall reaction required dithiothreitol and was inhibited by HU, a specific inhibitor of ribonucleotide reductase. It was also inhibited allosterically by dATP or dTTP, to a degree similar to inhibition of ribonucleotide reductase by these compounds. The incorporation of rCDP into DNA was little inhibited by added dCTP, not an allosteric inhibitor of ribonucleotide reductase. Incorporation of label from ['*C]rCDP would be expected to greatly decrease if the pathway included passage of the labeled dCDP or dCTP through a free pool.
Incorporation was not as rapid for dNTPs as for rNDPs, even under suitable conditions. This result is not consistent with entry of rNDP products into a pool of free dNTPs. The dNTPs were even more slowly incorporated under conditions permitting rapid incorporation of rNDPs. This inhibition was shown not to be because of competition for incorporation between dNTPs added and dNTPs formed from rNTPs. The principal factor responsible for this inhibition of dNTP incorporation was dithiothreitol. This compound did not inhibit DNA polymerase itself, as demonstrated with a soluble enzyme preparation ( 7 ) , and also because dithiothreitol did not inhibit incorporation into DNA of label provided as rCDP. These results strongly infer that activation of ribonucleotide reductase with dithiothreitol allosterically prevents access of dNTPs to DNA polymerase. In support of this hypothesis, hydroxyurea reversed the effects of dithiothreitol; it both inhibited ribonucleotide reductase and permitted more rapid incorporation of dNTPs into DNA if dithiothreitol was present.
Data of two sorts show that DNA polymerase is sequestered from added dNTPs in permeabilized cells in the presence of dithiothreitol. First, the dNTPs are slowly incorporated. Second, dCTP did not decrease incorporation of label from [''C] rCDP. An alternative explanation to the above is that the nuclear membrane provides a permeability barrier to dCTP. This seems very unlikely, since this membrane is permeable to many molecules including proteins (29). Various nucleotides also readily enter the nucleus of permeabilized cells. These cells allow rapid entry and incorporation into nuclear DNA of dNTPs in the absence of dithiothreitol and slower but quite appreciable incorporation in the presence of dithiothreitol. dATP, dTTP, and dGTP are effective nuclear ribonucleotide reductase allosteric inhibitors in the presence of dithiothreitol. Also, rNDPs and rATP are rapidly utilized in the synt.hesis of DNA, in the presence of dithiothreitol. Finally, according to this model HU would have to increase nuclear membrane permeability in order to increase dNTP incorporation (including dCTP). All of these results argue strongly against involvement of the nuclear membrane as a barrier to dNTPs.
The channeling of rNDPs into DNA, rate limitation by ribonucleotide reductase, and allosteric effects between this enzyme and DNA polymerase are best explained by a physical connection between these and other enzymes of the DNA synthesis pathway as in prokaryotic cells (30). We have previously proposed that these enzymes are associated into a complex which we have named replitase (7). Work in progress on properties of this complex in extracts support this idea. 2 The present studies on channeling are strengthened by demonstration that ribonucleotide reductase behaves like other enzymes of the pathway, appearing in the nucleus at the beginning of S phase and being rapidly sedimented along with nascent DNA.' The sedimented ribonucleotide reductase is not nonspecifically bound to some high molecular weight material as shown by cycle specificity of the sedimentation pattern.2 Also a mutant that overproduces ribonucleotide reductase 10-fold has, in S phase, only a stoichiometric quantity of enzyme recovered in the nucleus; the great surplus remains in the cytoplasm (31). These results are consistent with close kinetic coupling of ribonucleotide reductase to the site of new DNA synthesis.