Investigation of in Vivo Roles of the C-terminal Tails of the Small Subunit (ββ′) of Saccharomyces cerevisiae Ribonucleotide Reductase

Background: S. cerevisiae ribonucleotide reductase (RNR) comprises α and ββ′ subunits in an (α2)m(ββ′)n active holoenzyme. Results: C-terminal tail mutants of ββ′ prevent radical transfer across the α-β interface and consequently deoxynucleotide production in vivo. Conclusion: The ββ′ C-terminal tails interact only with the proximal α within each α/β(α/β′) pair. Significance: The predisposition of the ββ′ C-terminal tails in active RNR in vivo is established. The small subunit (β2) of class Ia ribonucleotide reductase (RNR) houses a diferric tyrosyl cofactor (Fe2III-Y•) that initiates nucleotide reduction in the large subunit (α2) via a long range radical transfer (RT) pathway in the holo-(α2)m(β2)n complex. The C-terminal tails of β2 are predominantly responsible for interaction with α2, with a conserved tyrosine residue in the tail (Tyr356 in Escherichia coli NrdB) proposed to participate in cofactor assembly/maintenance and in RT. In the absence of structure of any holo-RNR, the role of the β tail in cluster assembly/maintenance and its predisposition within the holo-complex have remained unknown. In this study, we have taken advantage of the unusual heterodimeric nature of the Saccharomyces cerevisiae RNR small subunit (ββ′), of which only β contains a cofactor, to address both of these issues. We demonstrate that neither β-Tyr376 nor β′-Tyr323 (Tyr356 equivalent in NrdB) is required for cofactor assembly in vivo, in contrast to the previously proposed mechanism for E. coli cofactor maintenance and assembly in vitro. Furthermore, studies with reconstituted-ββ′ and an in vivo viability assay show that β-Tyr376 is essential for RT, whereas Tyr323 in β′ is not. Although the C-terminal tail of β′ is dispensable for cofactor formation and RT, it is essential for interactions with β and α to form the active holo-RNR. Together the results provide the first evidence of a directed orientation of the β and β′ C-terminal tails relative to α within the holoenzyme consistent with a docking model of the two subunits and argue against RT across the β β′ interface.

The small subunit (␤ 2 ) of class Ia ribonucleotide reductase (RNR) houses a diferric tyrosyl cofactor (Fe 2 III -Y ⅐ ) that initiates nucleotide reduction in the large subunit (␣ 2 ) via a long range radical transfer (RT) pathway in the holo-(␣ 2 ) m (␤ 2 ) n complex. The C-terminal tails of ␤ 2 are predominantly responsible for interaction with ␣ 2 , with a conserved tyrosine residue in the tail (Tyr 356 in Escherichia coli NrdB) proposed to participate in cofactor assembly/maintenance and in RT. In the absence of structure of any holo-RNR, the role of the ␤ tail in cluster assembly/maintenance and its predisposition within the holo-complex have remained unknown. In this study, we have taken advantage of the unusual heterodimeric nature of the Saccharomyces cerevisiae RNR small subunit (␤␤), of which only ␤ contains a cofactor, to address both of these issues. We demonstrate that neither ␤-Tyr 376 nor ␤-Tyr 323 (Tyr 356 equivalent in NrdB) is required for cofactor assembly in vivo, in contrast to the previously proposed mechanism for E. coli cofactor maintenance and assembly in vitro. Furthermore, studies with reconstituted-␤␤ and an in vivo viability assay show that ␤-Tyr 376 is essential for RT, whereas Tyr 323 in ␤ is not. Although the C-terminal tail of ␤ is dispensable for cofactor formation and RT, it is essential for interactions with ␤ and ␣ to form the active holo-RNR. Together the results provide the first evidence of a directed orientation of the ␤ and ␤ C-terminal tails relative to ␣ within the holoenzyme consistent with a docking model of the two subunits and argue against RT across the ␤ ␤ interface.
Atomic resolution structures of E. coli ␣ 2 and of ␤ 2 from the Eklund group (12,13) led them to propose a docking model for the active ␣ 2 ␤ 2 complex, which in conjunction with biochemical studies (14 -18) demonstrated the importance of the C-terminal tail of each ␤ (15 and 8 amino acids in the prokaryotic and eukaryotic RNRs, respectively) for the interaction with ␣. Although their model had the tail from each ␤ associated with the corresponding ␣, an equally probable model could have the tail of ␤ associated with the adjacent ␣ (see Fig. 2, right panel). Unfortunately, the C termini (30 -35 amino acids) of all ␤ 2 structures are disordered (19), and thus, the molecular details of the tail with respect to both itself and ␣ remain unknown.
Within this C-terminal tail resides a conserved tyrosine residue, Tyr 356 in E. coli ␤ 2 , that has been proposed to play an important role in electron transfer in Fe 2 III -Y ⅐ cluster assembly from both the Fe 2 II and the Fe 2 III (met) state in ␤ 2 (1) and in the long range RT to initiate nucleotide reduction in ␣ 2 (see Figs. 1A and Fig. 2, left and center panels). In the former case, efforts to regenerate Y ⅐ from the met state of the cofactor using a Y356A mutant gave only low levels of Y ⅐ relative to the wildtype (WT) control. Thus, electron transfer required for reducing the Fe 2 III state was prohibited by this mutation. Deletion of the C-terminal tail also gave poor recovery of Y ⅐ in efforts to assemble the cofactor from the Fe 2 II state (20). In the latter case, recent studies using unnatural tyrosine analogs site-specifically in place of Tyr 356 revealed that Tyr 356 ⅐ plays an essential role in Cys 439 oxidation in ␣ 2 (21) (see Fig. 1A).
The RNR of the budding yeast S. cerevisiae possesses several unique properties that have made it feasible to investigate the functions proposed for the C-terminal tail of ␤ 2 and more specifically the conserved tyrosine residue (Tyr 356 , see Fig. 1B) within the tail. In contrast with the small ␤ 2 subunit of most organisms, the yeast small subunit is a heterodimer in vivo, ␤␤Ј, encoded by the RNR2 and RNR4 genes, respectively (22)(23)(24)(25). ␤Ј is structurally homologous to ␤ but lacks three iron ligands, and as a consequence, contains no metallo-cofactor (18,22,(25)(26)(27). Although ␤Ј is catalytically inactive, it is required for converting ␤ into a conformation that is competent for iron loading and Y ⅐ formation in vitro and in vivo (23,24,28). Importantly, recombinant apo-␤ 2 and ␤Ј 2 rapidly form apo-␤␤Ј in vitro (24), and although cluster assembly from Fe II , O 2 , and reductant is inefficient (0.25 Y ⅐ /␤␤Ј), these properties allow us to study reconstitution in vitro with mutant small subunits.
Although ␤ is essential for cell viability, cells lacking ␤Ј (⌬rnr4) are viable, albeit with extremely low Y ⅐ content and RNR activity in some yeast strain backgrounds (e.g. S288C) (29). These strains are also hypersensitive to the Y ⅐ quenching reagent hydroxyurea (28). We have been able to take advantage of this viability and our ability to permeabilize WT and ⌬rnr4 cells to take up proteins (e.g. ␣,␤Ј) to provide a robust assay for in vivo Fe 2 III -Y ⅐ cofactor assembly by the addition of ␤Ј 2 and assay for RNR activity by the addition of ␤Ј 2 followed by ␣ 2 (28) herein.
In this study, we have also used EPR spectroscopy on whole cells and permeabilized cells to show in vivo that the C-terminal 8 amino acids of ␤Ј, rich in carboxylates, are not required for binding and delivery of iron to ␤ and that consequently this tail does not function as a molecular chaperone as we originally proposed from in vitro studies (18). We also demonstrate that ␤-Tyr 376 is not on an essential electron transfer pathway to deliver the reducing equivalent to generate the Fe 2 III -Y ⅐ cofactor from the Fe 2 II or met (Fe 2 III tyrosyl radical reduced) state either in vitro or in vivo. Moreover, studies using plasmid shuffle strains reveal that ␤-Tyr 376 is nevertheless essential for the long range RT and cell survival, whereas ␤Ј-Tyr 323 is not essential. Finally, although the ␤Ј mutant lacking the C-terminal 8 amino acids (␤Ј-⌬8aa) is capable of heterodimer formation with ␤ and supporting Fe 2 III -Y ⅐ cofactor assembly in ␤, the resulting ␤␤Ј is catalytically inactive (Ͻ 0.5% WT activity), suggesting an essential role of the ␤Ј tail for interactions with ␣ 2 to form an active holo-enzyme. Our results for the first time provide in vivo evidence for the predisposition of the C-terminal tails of ␤ and ␤Ј relative to ␣ in the active RNR; these tails interact only with the ␣ in the ␣/␤ (␣/␤Ј) pair in the holo-RNR and do not cross over to interact with the adjacent ␣ (see Fig. 2, right panel).

EXPERIMENTAL PROCEDURES
Strains, Plasmids, and Media-Yeast strains and plasmids used in this study are listed in Tables 1 and 2, respectively. Rich YPD medium contains 1% yeast extract, 2% peptone, and 2% glucose. Synthetic, defined 6.7 g/liter yeast nitrogen base without amino acids (Difco), 2% glucose, complete supplemental mixtures (CSM), or CSM with dropout of amino acid(s) (MP Biomedicals). Solid media contain 1.5% agar (Difco) as a solidifying agent. YP-and CSM-raffinose media contain 2% raffinose as the sole carbon source, which maintains the GAL1 promoter at residual activity level (uninduced state). YP-and CSM-Raff/Gal media contain 2% raffinose and 0.5% galactose that activate the GAL1 promoter (induced state).
Purification of Recombinant ␣, His 6 -␤, His 6 -␤Ј, His 6 -␤Ј⌬8aa, His 6 -␤-Y376F, and His 6 -␤Ј-Y323F-E. coli transformants harboring the desired expression vectors were grown in LB medium with antibiotics (ampicillin or kanamycin) at 37°C to an A 600 of 0.4 -0.8 before cooling the cultures to 16°C by adjusting the temperature setting of the incubator. Induction of subunit expression was carried out by the addition of 0.5 mM (for ␤ and ␤Ј) or 1 mM (for ␣) isopropyl ␤-D-1-thiogalactopyranoside to the medium, and growth was continued for 16 h. The proteins were purified as described previously (18,24,31). ␣ was purified using a DEAE Sephadex column and a dATP affinity column (31) and had a specific activity of 102 nmol min Ϫ1 mg Ϫ1 when assayed in the presence of a 10ϫ molar excess of FLAG-tagged ␤␤Ј (specific activity 3000 nmol min Ϫ1 mg Ϫ1 ) that was isolated from yeast ⌬crt1 MHY619 cells (29). Wildtype (WT) and mutant ␤ and ␤Ј were both N-terminally His 6tagged with the following N terminus: MGSSHHHHHHSS-GLVPRGSH-native protein, and were purified by using an nickel-nitrilotriacetic acid column as described (18). All purified proteins were homogeneous as judged by SDS-PAGE.
Reconstitution of Fe 2 III -Y ⅐ with ␤ 2 and ␤Ј 2 was carried out using previously reported protocols (32). Briefly, purified ␤ 2 and ␤Ј 2 were degassed on a Schlenk line and brought into an anaerobic box (MBraun) in a cold room. The ␤ 2 and ␤Ј 2 were mixed for 2 min before the addition of 3 eq of ferrous iron. The resulting reaction mixture (137 l) contained 44 M each of ␤ 2 and ␤Ј 2 , 50 mM HEPES at pH 7.6, 5% glycerol and was incubated in the glove box for 10 min. The sample was then removed from the box, and 165 l of O 2 -saturated buffer was added and mixed immediately at room temperature, resulting in 20 M ␤␤Ј. The sample (260 l) was transferred to EPR tubes, and the rest of the sample was frozen immediately and used for subsequent activity assays.
Permeabilized ⌬rnr4 Yeast Cells Reconstituted with RNR4 and Analyzed by EPR Spectroscopy-A suspension of ⌬rnr4 cell (60 l, 32 A 600 /ml) permeabilized as described previously (28) was added to 200 l of 0.1 M potassium P i buffer (pH 7.5) containing 0.6 M sorbitol, 50 mM DTT, and 6.5 M His 6 -␤Ј 2 or His 6 -(␤Ј⌬8aa) 2 . The reaction mixture was incubated at 30°C for 3 min before being transferred into an EPR tube and rapidly frozen in liquid nitrogen. Our previous study has found that rapid and efficient Fe 2 III -Y ⅐ cofactor assembly (0.5-0.6 Y ⅐ /␤␤Ј in contrast with 0.25-0.3 Y ⅐ /␤␤Ј in vitro) in permeabilized ⌬rnr4 cells requires high levels (50 mM) of DTT, although its specific role in the process is unclear (28).
EPR spectra were acquired on a Bruker EMX X-band spectrometer at 30 K using an Oxford Instruments liquid helium cryostat. Acquisition parameters for reconstitution in permeabilized cells and whole cell EPR with intact cells at 9.4 GHz were 0.2-milliwatt power, 1.8 ϫ 10 5 gain, 2-G modulation amplitude, and 100-kHz modulation frequency. Spin quantitation was performed by double integration of the signal using E. coli NrdB as a standard in which the Y ⅐ concentration was determined by the dropline correction method (33). Analysis was carried out using WinEPR software (Bruker).
RNR Activity Assays-Activity of ␤␤Ј in ⌬rnr4 permeabilized cells was measured as described previously (28). Typically, a solution (180 l) containing 0.1 M potassium P i , pH 7.5, 0.6 M sorbitol, 3 mM ATP, 10 mM NaF, 50 mM DTT, 4.4 M ␣ (specific activity 102 nmol min Ϫ1 mg Ϫ1 ), and 7.7 M His 6 -␤Ј 2 or His 6 -(␤Ј⌬8aa) 2 was mixed with 1.5 A 600 ⌬rnr4 permeabilized cells (180 l) at 4°C and warmed at 30°C for 1 min. The reaction was started by the addition of 1 mM [5-3 H]CDP (ViTrax, 17 Ci/mmol, 5790 cpm/nmol). Aliquots (58 l) were removed at 0, 5, and 10 min. The reaction was stopped by placing the sample in a boiling water bath for 2 min. Each sample was then adjusted to pH 8.5 by the addition of 0.5 M of Tris-HCl (final concentration 50 mM) and incubated with 10 units of alkaline phosphatase (Roche Applied Science, from calf intestine) for 2 h, and the amount of dC was analyzed by the method of Steeper and Steuart (34) as revised in Ref. 35.
For activity determination of the reconstituted recombinant ␤␤Ј, a typical reaction mixture of 160 l contained 50 mM HEPES, pH 7.5, 15 mM MgSO 4 , 1 mM EDTA, 3 mM ATP, 50 mM DTT, 5 M ␣ 2 , 1 M ␤␤Ј, and 1 mM [5-3 H]CDP (4000 -8000 cpm/nmol). Aliquots (50 l) were removed at 0, 5, and 10 min and analyzed as described above. Western Blotting for ␤ Protein Levels in Yeast Cell Extracts-Yeast cells (2 ϫ 10 7 ) from log-phase cultures were treated with 200 l of 0.1 M NaOH for 5 min at room temperature (36). The cells were pelleted by a brief centrifugation, resuspended in 200 l of SDS loading buffer, and lysed by boiling in a sand bath for 5 min. Lysate from 1 ϫ 10 6 cells (10 l) was resolved by 10% SDS-PAGE, and the proteins were transferred to nitrocellulose membranes and probed with primary and secondary antibodies. The dilutions for antibodies were: polyclonal anti-␤ (31) at 1:200,000, monoclonal anti-Myc 9E10 (Covance) at 1:1,000, and goat-anti-mouse and goat-anti-rabbit (Jackson Immuno-Research Laboratories) at 1:10,000. Blots were developed by using SuperSignal West Femto maximum sensitivity substrate (Thermo Scientific). A CCD camera (ChemiDoc XRS Bio-Rad) was used to record the blotting signals. The proteins were quantified by analyses of the signal intensities with Quantity One (Bio-Rad).

RESULTS
The C-terminal Tail of ␤Ј Is Dispensable for Fe 2 III -Y ⅐ Formation in ␤-We have previously proposed that ␤Ј might use the aspartate and glutamate residues within its C-terminal tail (Figs. 1B and Fig. 2) to bind and deliver Fe II to ␤ in the ␤␤Ј heterodimer in a manner similar to that of copper loading of Sod1 by the chaperone protein Ccs1 (38,39). To test this hypothesis, we previously constructed the ␤Ј-⌬8aa mutant that lacks the last 8 amino acid residues of ␤Ј (18). In vitro cluster assembly experiments using apo ␤␤Ј and ␤␤Ј-⌬8aa, Fe 2ϩ , and O 2 gave the same levels of Y ⅐ /␤␤Ј, suggesting that the tail did not play an essential role in this process. However, in contrast to the self-assembly studies in E. coli and mouse ␤ 2 , the yield of active yeast ␤␤Ј is poor (18). Thus, we have used the same constructs (18) to introduce ␤Ј 2 into permeabilized ⌬rnr4 cells that contain very low activity (Ͻ1% of WT), high levels of ␤ 2 , and undetectable Y ⅐ determined by whole cell EPR analysis (28). The addition of ␤Ј 2 to permeabilized ⌬rnr4 cells resulted in rapid heterodimer formation and Fe 2 III -Y ⅐ formation within 3 min (Fig. 3A, reproduced from supplemental Fig. 2B in Ref. 28). The (␤Ј⌬8aa) 2 mutant gave the same amount of Fe 2 III -Y ⅐ as ␤Ј 2 , suggesting that inside the cell, the ␤Ј C-terminal tail is not involved in iron delivery and cofactor formation.
The C-terminal Tail of ␤Ј Is Essential for RNR Enzyme Activity, Indicating a Role in Interaction with ␣ 2 -In the experiments described above, the permeabilized ⌬rnr4 cells, subsequent to treatment with ␤Ј 2 for 10 -20 s, were then incubated with 4.4 M ␣ 2 and 1 mM [ 3 H] CDP to assay for ␤␤Ј activity. Deoxy-CDP was produced at 3 nmol min Ϫ1 per A 600 cells (Fig. 3B, filled  squares). In contrast, with ␤Ј 2 -⌬8aa under otherwise identical conditions, the activity was Ͻ0.5% of that of the WT ␤Ј 2 (Fig.  3B, open squares). Thus, although the C-terminal tail of ␤Ј 2 is dispensable for iron delivery and subsequent Fe 2 III -Y ⅐ assembly in ␤, it plays a crucial role in nucleotide reduction within the active holo-enzyme. The results indicate that an active complex is not formed at the concentrations of (␤Ј⌬8aa) 2 (ϳ6.5 M) examined. Our previous studies have established that endogenous concentrations of ␤ and ␤Ј in WT cells are 0.5-1.0 M and that concentration of ␤ in ⌬rnr4 cells is 5-10 M (24). Thus, the C-terminal 8 amino acids of ␤Ј are critical for association with ␣ 2 and formation of holo-enzyme.  Fig.  2) within the C-terminal tail of the small subunit is conserved in all class I RNRs. One proposed function for this residue is its involvement in the delivery of the required reducing equivalent for assembly and maintenance of the Fe 2 III -Y ⅐ cofactor (20,40). A second proposed function, established in the E. coli Ia RNR, is its involvement in the pathway of RT that is essential for nucleotide reduction (14,15,41,42) (Fig. 1A).
To determine the functional importance of this tyrosine residue in ␤ or ␤Ј, we constructed rnr2(Y376F) and rnr4(Y323F) mutant-containing plasmids and transformed them into the RNR2 and RNR4 shuffle strains, respectively. Each shuffle strain contains a chromosomal deletion of the gene being tested and is kept alive by a WT copy of the cognate gene on an URA3marked plasmid. Survival of the transformants on media containing 5-fluoroorotic acid (5-FOA), a reagent that is converted by the URA3 gene product to the cytotoxic chemical 5-fluorou-racil that kills the Ura ϩ cells, indicates that the mutant-bearing plasmids can provide the essential function in the absence of the WT copy on the URA3 plasmid. The rnr2(Y376F) mutant failed to grow on a 5-FOA-containing plate (Fig. 4A). Western blotting confirmed the expression of the rnr2(Y376F) protein in the shuffle strain transformant prior to 5-FOA selection (Fig.  4B), indicating that the mutant protein was expressed but failed to replace the WT protein function. In contrast, the rnr4(Y323F) mutant was viable and exhibited no obvious difference in growth from the WT control strain on a 5-FOA plate (Fig. 4C). These results are consistent with the proposed essential role of ␤-Tyr 376 either in Fe 2 III -Y ⅐ assembly/maintenance or in the RT pathway, or both, and also indicate that ␤Ј-Tyr 323 is not required for either event.
Measurement of in Vitro Fe 2 III -Y ⅐ Cofactor Formation and Enzyme Activity of the C-terminal Tail Tyrosine Mutants of ␤␤Ј-Vegetative growth of yeast cells bearing the ␤(Y376F) and ␤Ј(Y323F) mutants offers a qualitative rather than quantitative measurement of their in vivo activity as the threshold of RNR activity to maintain viability or optimal growth in different yeast strains may vary. For instance, the ⌬rnr4 mutation is lethal in the W303 background (26) but viable in the S288C background despite very low Y ⅐ content (not detectable by whole cell EPR) (28,29). To gain mechanistic insight into the lethality resulting from the ␤(Y376F) mutant and to further rule out a role of ␤Ј-Tyr 323 in RT, we have directly monitored in vitro the Fe 2 III -Y ⅐ cluster assembly and activity of the ␤␤Ј complexes formed from WT ␤, ␤Ј, and the ␤(Y376F) and ␤Ј(Y323F) mutants.
N-terminally His 6 -tagged ␤ 2 , ␤Ј 2 , ␤ 2 (Y376F), and ␤Ј 2 (Y323F) were expressed in E. coli, purified to Ͼ95% homogeneity, and subjected to complex formation and cofactor reconstitution in four different combinations: ␤␤Ј, ␤␤Ј(Y323F), ␤(Y376F)␤Ј, and ␤(Y376F)␤Ј(Y323F) as described under "Experimental Procedures." EPR analyses of the resulting products show Y ⅐ formation in all four experiments (Fig. 5A). Spin quantitation gave FIGURE 2. Models depicting the roles of ␤ and ␤ C-terminal tails containing the conserved tyrosine residue (␤-Tyr 376 and ␤-Tyr 323 ) in cluster biosynthesis/maintenance and association between ␤␤ and ␣ 2 to perform RT within the RNR holoenzyme. Biosynthesis of the Fe 2 III -Y ⅐ cofactor in ␤␤Ј involves delivery of two Fe II ions into ␤ that in the presence of O 2 and a reducing equivalent provide an active Fe III Fe IV species, which oxidizes ␤-Tyr 183 in ␤␤Ј to the Tyr 183⅐ (left panel, pathway A). Cofactor can self-assemble in vitro when apo-␤ 2 and apo-␤Ј 2 are mixed, which spontaneously form a heterodimer that then binds Fe II and self-assembles with O 2 and reductant. Substoichiometric Y ⅐ /␤␤Ј levels with stoichiometric Fe 2 III clusters are obtained both from in vitro selfassembly and from ␤␤Ј isolated from yeast cells, indicating that some ␤␤Ј is in an inactive Fe 2 III (met) state (left panel, pathway B). Conversion of the Fe 2 III to active cofactor can occur by a maintenance pathway that requires a reductant to generate the Fe 2 II state, which can then assemble to form Fe 2 III -Y ⅐ (middle panel). The conserved tyrosine residues at the C-terminal tails, ␤-Tyr 376 and ␤Ј-Tyr 323 , have been proposed to be involved in Fe II delivery, in electron transfer required for cofactor biosynthesis and maintenance, and in RT between ␤ and ␣ to generate a thiyl radical in ␣. The C-terminal tails of ␤ and ␤Ј could cross over to interact with the distant ␣ of the neighboring ␣␤ pair (right panel, I and II) or interact exclusively with the proximal ␣ of the same ␣␤ pair (right panel, III and IV). Red lines indicate the putative RT pathway between ␣ and ␤, which can occur between proximal (I and III) or distant (II and IV) ␣ ␤ subunits. The latter case requires a RT pathway across the ␤␤Ј interface. Only model III is supported by experimental results in this study. 0.25 Y ⅐ in WT ␤␤Ј and ranged from 0.35 to 0.43 Y ⅐ with the mutant combinations. Thus, neither ␤-Tyr 376 nor ␤Ј-Tyr 323 is required for cofactor assembly in vitro. These reconstituted small subunits were then used to evaluate whether the C-terminal tyrosine residue is directly involved in the RT pathway by enzyme activity assays in the presence of an excess of ␣ 2 . Both ␤␤Ј and ␤␤Ј(Y323F) were capable of converting CDP to dCDP, with specific activities proportional to the Y ⅐ contents (Fig. 4, B and C). By contrast, neither ␤(Y376F)␤Ј nor ␤(Y376F)␤Ј(Y323F) exhibited any detectable catalytic activity despite comparable Y ⅐ levels in ␤␤Ј and ␤␤Ј(Y323F) (Fig. 5, B  and 5C). These results demonstrate that ␤-Tyr 376 plays an essential role in the RT pathway, whereas ␤Ј-Tyr 323 is dispensable in this process.
The ␤-Tyr 376 Residue Is Not Required for Fe 2 III -Y ⅐ Cluster Formation in Vivo, Ruling out a Role in Electron Delivery in Cofactor-The ␤(Y376F) mutant is catalytically inactive and cannot support mitotic growth of yeast cells, although it is capable of forming the Fe 2 III -Y ⅐ cofactor in vitro. The self-assembly of the cofactor in vitro requires Fe 2ϩ and O 2 with the reducing equivalent supplied by the Fe 2ϩ . In vivo, we have proposed that this reducing equivalent is supplied by YfaE, a 2Fe2S cluster ferredoxin in E. coli (40,43), and Dre2, an FeS requiring protein in S. cerevisiae (28). We wanted to test the proposal that ␤-Tyr 376 is part of the electron transfer pathway to deliver the reducing equivalent for cluster assembly inside the cell. Because cells harboring ␤(Y376F) as the only source of ␤ are unviable, we chose to examine a GalRNR2 strain in which ␤ expression from the chromosomal RNR2 locus is controlled by the GAL1 promoter. The GalRNR2 cells propagate normally in galactosecontaining media (GAL induced state) but are unable to grow in glucose-containing media (GAL repressed). When the trisaccharide raffinose is used as the sole carbon source for yeast cells, the GAL promoter is neither induced nor repressed, allowing basal levels of transcription (i.e. GAL uninduced or derepressed  state). Consistent with low GAL promoter activity, GalRNR2 cells cultured in raffinose media grow slowly with very low levels of endogenous ␤ protein (Fig. 6A, lanes 2-4) and Y ⅐ radical content (Fig. 6B, blue).
To study cluster assembly in ␤-mutants, we then transformed CENLEU2 (low copy number, 1-3 copies/cell) (37) plasmids with the ability to express N-terminally Myc-tagged ␤ WT and ␤(Y376F) from the native RNR2 promoter into the GalRNR2 cells. GalRNR2 transformants harboring the empty CENLEU2 vector were used as a negative control. All sets of transformants were inoculated from CSM-Leu-Raff/Gal plate (GAL induced) to 100 of ml CSM-Leu-Raff liquid medium (GAL uninduced) and grown to a density of A 600 ϳ1 over 16 -20 h. At this time, the endogenous, untagged ␤ expressed from GalRNR2 was depleted to residual levels (Fig. 6A, lanes 2-4), and the ectopically expressed Myc-␤ and Myc-␤(Y376F) are present at levels that are 7-14-fold higher (Fig. 6A, lanes 3 and  4). These levels are comparable with that of the endogenous ␤ in a WT control strain (Fig. 6A, compare lane 1 with lanes 3 and  4). The moderately (ϳ60%) higher level of Myc-␤(Y376F) is likely due to mild checkpoint activation resulting from the absence of a functional ␤.
The elevated levels of ectopically expressed Myc-␤ and Myc-␤(Y376F) over the endogenous ␤ under the experimental conditions allowed us to determine and compare Fe 2 III -Y ⅐ cluster content between the two strains using whole cell EPR spectros-copy (Fig. 6B). The Y ⅐ was detectable in both the Myc-␤-expressing cells (green) and the Myc-␤(Y376F)-expressing cells (red) but was not detectable in cells carrying the control vector (blue).
The EPR spectrum from the vector control was used as a background signal and was subtracted from both the Myc-␤ and the Myc-␤(Y376F) samples. The analysis shown in Fig. 5C revealed that Myc-␤(Y376F)-expressing cells have a Y ⅐ content that is 60% higher than that from the Myc-␤-expressing cells. When the Y ⅐ signals were normalized for levels of the Myc-␤ and Myc-␤(Y376F) proteins, respectively (Fig. 6A), there was no difference in steady-state Fe 2 III -Y ⅐ cluster content between wild-type ␤ and the ␤(Y376F) mutant. The results together demonstrate that ␤-Tyr 376 is not required for Fe 2 III -Y ⅐ cluster formation in vivo. It should be noted, however, that the in vivo rates of cluster assembly cannot be analyzed under the experimental conditions.

DISCUSSION
In this study, we provide evidence that the C-terminal tail of ␤Ј does not cross over to the neighboring ␣␤ pair thus providing, insights into the geometry of ␤␤Ј C-terminal tails relative to ␣ 2 within the holoenzyme. The unusual heterodimeric structure of yeast small subunit ␤␤Ј with only a single Fe 2 III -Y ⅐ cluster per ␤␤Ј have provided a unique system to investigate the RT pathway and the quaternary structure of the RNR holoenzyme Cells were inoculated from a CSM-Leu-Raff/Gal plate to 100 ml of CSM-Leu-Raff liquid medium and grown to a density of A 600 ϳ1 over 16 -20 h before being harvested, which depleted the endogenous, untagged ␤ expressed from chromosomal RNR2 locus. 1 ϫ 10 6 cells were used for Western blotting, and the rest of cells (2 ϫ 10 9 ) were used for EPR. Amido Black staining of the nitrocellulose membrane prior to Western blot shows equal loading of cell lysate (upper panel); Western blot using anti-␤ antibody detects both the untagged and 3ϫMyc-tagged ␤ proteins (lower panel). Lane 2 contains protein extracts from an untagged wild-type strain showing ␤ expression level from the native RNR2 promoter. The band between Myc ␤ and ␤ (marked by *) is a cross-reacting protein. B, whole cell EPR spectra of the GalRNR2 cells harboring the empty vector (blue), Myc ␤ (green), and Myc ␤(Y376F) (red) in the GalRNR2 cells grown in a raffinose-containing medium. C, EPR spectra after subtraction of the spectrum of vector control (blue in B) from Myc ␤ (green) and Myc ␤(Y376F) (red).
in an active complex inside the cell. Although ␤Ј on its own does not form a Fe 2 III -Y ⅐ cluster (18,24,26,31), it is conceivable that Tyr 323 of ␤Ј could participate in the RT process as a stepping stone between the ␣ 2 and ␤␤Ј during the turnover cycle (18). In this proposal, the C terminus of ␤Ј would not interact with the ␣ of its own ␣/␤Ј pair, but instead would interact with the ␣ of the neighboring ␣/␤ pair, which contains the Fe 2 III -Y ⅐ cofactor. Thus, this proposal posits that the C-terminal tail of ␤Ј would cross over at the subunit interface (Fig. 2, right panel, I and II). However, our observation that the ␤(Y376F) mutant is catalytically inactive and causes cell lethality, whereas the ␤Ј(Y323F) mutant has no effect on either yeast vegetative growth or enzyme activity, strongly suggests that the tails of ␤ and ␤Ј do not cross over to interact with the adjacent ␣ within the holocomplex. Instead, each ␤ and ␤Ј tail interacts with the proximal ␣ in the same ␣/␤ (␣/␤Ј) pair (Fig. 2, right panel, III and IV).
In the studies of Ge et al. (35) in 2003, single turnover experiments using the E. coli ␣ 2 ␤ 2 with one Y ⅐ /␤ 2 observed formation of two dCDPs with the same rate constant (35). At the time, two hypotheses were proposed for this unusual observation that a single Y ⅐ could service both active sites in ␣ 2 . One hypothesis was that in addition to the established RT pathway between the proximal ␣ and ␤ (Fig. 2, right panel, I and III), there was a second RT pathway involving both Trp 48 and Tyr 356 in ␤ with the electron transfer occurring across the ␤-␤ monomer interface via the two Trp 48 residues to generate a thiyl radical in the second ␣ (Fig. 2, right panel, II and IV). A second hypothesis involved rapid dissociation of ␤ 2 from ␣ 2 and their subsequent rapid reassociation, both faster than turnover, to allow the second dCDP to be generated quickly in the second ␣ of the ␣ 2 . Our results with the ␤(Y376F) and ␤Ј(Y323F) mutants suggest that the former hypothesis is unlikely for the yeast RNR. If RT could cross the ␤-␤Ј interface and generate thiyl radical in the active site of the adjacent ␣ according to the first model, then the ␤-Y376F mutant would not have completely inactivated the enzyme as observed. Thus, our studies support the model in which RT occurs only within one ␣/␤ pair (Fig. 2, right panel,  III).
In vitro reconstitution of the Fe 2 III -Y ⅐ cofactor uses excess ferrous iron to provide reducing equivalent required for its formation (11). However, in vivo, a ferredoxin-like protein YfaE in E. coli (43) and an FeS cluster-containing protein Dre2 (28) in S. cerevisiae have been proposed to provide electrons for the Fe 2 III -Y ⅐ formation and maintenance pathways. Before the identification of YfaE, Coves et al. (20) showed in vitro that a reduced met-␤ could regenerate the active Fe 2 III -Y ⅐ -containing ␤ and that the C-terminal tail Tyr 356 was important for this process (20). The C-terminal Tyr 356 was proposed to mediate electron transfer for the maintenance of the cofactor and potentially for cofactor assembly. Our whole cell EPR results with the ␤(Y376F) mutant in the GalRNR2 cells at uninduced state clearly demonstrate for the first time that ␤-Tyr 376 is not required for Fe 2 III -Y ⅐ formation in vivo, and thus, its essential function is likely due to its role in the RT pathway.