The C-terminal domain of Hsp70 is responsible for paralog-specific regulation of ribonucleotide reductase

The Hsp70 family of molecular chaperones is well-conserved and expressed in all organisms. In budding yeast, cells express four highly similar cytosolic Hsp70s Ssa1, 2, 3 and 4 which arose from gene duplication. Ssa1 and 2 are constitutively expressed while Ssa3 and 4 are induced upon heat shock. Recent evidence suggests that despite their amino acid similarity, these Ssas have unique roles in the cell. Here we examine the relative importance of Ssa1-4 in the regulation of the enzyme ribonucleotide reductase (RNR). We demonstrate that cells expressing either Ssa3 or Ssa4 as their sole Ssa are compromised for their resistance to DNA damaging agents and activation of DNA damage response (DDR)-regulated transcription. In addition, we show that the steady state levels and stability of RNR small subunits Rnr2 and Rnr4 are reduced in Ssa3 or Ssa4-expressing cells, a result of decreased Ssa-RNR interaction. Interaction between the Hsp70 co-chaperone Ydj1 and RNR is correspondingly decreased in cells only expressing Ssa3 and 4. Through studies of Ssa2/4 domain swap chimeras, we determined that the C-terminal domain of Ssas are the source of this functional specificity. Taking together, our work suggests a distinct role for Ssa paralogs in regulating DNA replication mediated by C-terminus sequence variation.

Introduction Cells in all organisms must be able to cope with stressors that trigger both protein unfolding and misfolding. The core machinery induced in response to challenges of proteostasis are the molecular chaperones or Heat Shock Proteins [1,2]. The essential chaperone Heat Shock Protein 70 (Hsp70) is a main player in proteostasis, binding nascent chains and responsible for stabilization, folding and degradation of a large majority of the proteome [1,3,4]. Structurally, Hsp70 chaperones comprise of three major functional domains; an N-terminal ATPase domain (NBD) connected by a flexible linker to a substrate binding domain (SBD) [2]. The SBD can be delineated further into the SBD beta "basket" into which substrates dock, the SBD beta "lid" which traps substrates for folding. The SBD is followed by an unstructured C-terminal domain (CTD) which is responsible for binding a variety of co-chaperone helper proteins [2]. The binding and hydrolysis of ATP in the NBD promotes a range of conformational changes which are transduced through the linker into the C-terminus of Hsp70 resulting in clamping of the lid over the SBD basket trapping clients and promoting protein folding [2]. Hsp70 requires the assistance of co-chaperone proteins comprised of J-proteins and nucleotide exchange factors (NEFs) that facilitate the stimulation of Hsp70 activity and folding of client proteins [3,5,6]. Although the exact mechanism by which Hsp70 folds clients remains unclear, recent evidence suggests that Hsp70 works like a molecular "hair straightener", pulling out kinks in non-optimal protein conformations, allowing correct folding to occur [7][8][9][10][11].
Given their role in protein folding, it is unsurprising that Hsp70 appears to exist in most organisms studied so far and is essential for their viability [2]. However, the cellular rationale for the large number of highly related paralogs of Hsp70 remains unclear. Budding yeast (S. cerevisiae) expresses 14 different Hsp70 paralogs, five of which are localized to specific organelles [12]. Out of the remaining 9 cytosolic Hsp70s, 4 are from the Stress Seventy sub-family A (SSA) comprised of Ssa1, 2, 3 and 4 [12][13][14][15][16]. Previously, Hsp70 paralogs were thought to be functionally indistinguishable apart from spatiotemporal expression patterns, however recent findings suggest unique functions for Ssa paralogs [12,13].
Ssa paralogs arose from genome duplication and are highly conserved, with Ssa1 sharing 99%, 84% and 85% amino acid identity with Ssa2, 3 and 4, respectively [12]. The most prominent difference between the Ssa1-4 paralogs is their expression levels; Ssa1/2 are expressed constitutively at high levels whereas Ssa3/4 are only expressed during cell stress [12,[17][18][19][20][21]. Although yeast can survive on the loss of any of 3 Ssas if the 4 th is expressed at high levels, the phenotypes of these cells vary in terms of heat resistance longevity, ability to fold certain clients [22][23][24][25].
In this study, in order to better understand functional differences between Ssa1-4, we utilized the model chaperone client Ribonucleotide Reductase (RNR). RNR is an enzyme that is important for the production of deoxyribonucleotides (dNTPs) which are used in DNA synthesis and repair [26]. RNR is comprised of two diverse subunits, the large subunit R1 (R1 in vertebrates, Rnr1/Rnr3 in yeast) which contains the allosteric regulatory sites [27] and the small subunit R2 (R2/R2B in vertebrates, Rnr2/Rnr4 in yeast) which consists of a cell cycle regulated binuclear iron center and a tyrosyl free radical [28][29][30][31][32]. Due to its crucial role in the maintenance of genome integrity and subsequently cell survival, RNR remains an attractive anticancer target [29,31,33]. Several RNR inhibitors have been developed and used in a clinical setting including hydroxyurea (HU), triapine and gemcitabine [29, [34][35][36].
Previous studies in both yeast and mammalian cells have identified Hsp70 as an important regulator of RNR with small molecule chaperone inhibitors such as 17-AAG promoting RNR subunit degradation [37][38][39]. Hsp70 inhibition sensitizes cancer cells to gemcitabine and the combination of Hsp70 and RNR inhibitors has the potential to form the basis of a novel anti-cancer therapeutic [37][38][39][40][41][42]. Recently, the Hsp70 co-chaperone Ydj1/DNAJA1 (yeast/mammalian) was identified to assist Hsp70 in the regulation of RNR. Lack of Ydj1 in S. cerevisiae results in reduced Rnr2 subunit expression and stability against degradation [38,39]. This interaction was found to be conserved in humans, where DNAJA1 and R2B assist in RNR complex stability and activity in mammalian cells [38,39]. Additionally, inhibition of DNAJA1 with 116-9e, a small molecule inhibitor that blocks Hsp40 binding to Hsp70 through the Jdomain resulted in disruption of R2B-DNAJA1 interaction and sensitized cells to HU and triapine [38,39].
Here we characterize the relative roles of Ssa1, 2, 3, and 4 in regulating ribonucleotide reductase in yeast. We reveal that yeast expressing single Ssas as their sole cytosolic Hsp70 on identical promoters display differing abilities to respond to DNA damaging agents. This can be explained by the loss of RNR subunit stability, occurring due to decreased Ssa-RNR interaction. Finally, we provide evidence that the C-terminal domain of Hsp70 is responsible for selectivity in activating RNR.

Ssa paralogs contribute differentially to the resistance to DNA-perturbing agents and the transcriptional response to DNA damage
Previous studies have demonstrated a critical role for Hsp70 and Hsp90 in supporting RNR activity in yeast and mammalian cells [37][38][39]. In order to dissect the unique roles of the yeast Hsp70 paralogs Ssa1-4 in supporting the DNA damage response (DDR), we screened cells expressing either Ssa1, 2, 3 or 4 (under the constitutive Ssa2 promoter) as the sole cytosolic Hsp70 for growth against various DNA damaging agents including hydroxyurea (HU), 5-fluorouracil (5-FU), hydrogen peroxide (H 2 O 2 ) and methyl methanesulfonate (MMS) (Fig 1A-1D). Yeast expressing Ssa1 or Ssa2 were markedly more resistant to all DNA damaging agents compared to Ssa3 or Ssa4 cells (Fig 1A). Cells expressing Ssa3 or Ssa4 as their sole Ssa displayed an increased sensitivity to HU, 5-FU and H 2 O 2 but not MMS (Fig 1A-1D). To determine whether this phenotypic difference was due to altered DNA damage response-regulated transcription, we compared induction of RNR3 in HU-treated Ssa1, 2, 3 and 4 cells (Fig 1E). Consistent with the phenotypes in Fig 1, Ssa3 and Ssa4 cells were unable to fully activate DDR transcription, displaying a significant decrease in HU-mediated RNR3 expression ( Fig 1E).

Ribonucleotide reductase subunit levels are compromised in cells solely expressing either Ssa3 or Ssa4
Our previous studies described a role for Hsp70 function in maintaining an active RNR complex in yeast and human cells. To determine whether the inability of Ssa3/4 cells to grow in the presence of DNA-damaging agents could be explained by loss of RNR function, we queried the steady-state levels of Rnr1, Rnr2 and Rnr4 protein in cells expressing single Ssa paralogs. Although Rnr1 levels remained independent of Ssa1 paralog (Fig 2A), clear differences in Rnr2 and Rnr4 levels were observed (determined by normalizing RNR levels to a loading control of PGK1). Rnr2 levels in Ssa3/4 cells were significantly lowered in both untreated and treated conditions compared to cells whose primary Hsp70 was Ssa1 or Ssa2 ( Fig 2B). In contrast, while Rnr4 levels in untreated conditions were independent of Ssa version, the levels of HU-induced Rnr4 expression were substantially decreased in Ssa3 or Ssa4 cells (Fig 2C).
The steady state level of proteins are carefully balanced by both rate of transcription and protein degradation. To determine whether the altered RNR subunit expression observed in Ssa3 and Ssa4 cells was a result of altered transcription, we quantified RNR1, RNR2 and RNR4 mRNA expression in Ssa-paralog specific yeast using real-time quantitative polymerase chain reaction (RT-qPCR). RNR1 and RNR2 transcription was independent of Ssa paralog, whereas HU-induced RNR4 induction was compromised in cells expressing only Ssa3 and Ssa4 ( Fig  3A). To determine whether the protein stability of Rnr1, Rnr2 and Rn4 had also been compromised in Ssa3 and Ssa4-expressing cells, we examined the half-life of RNR subunits by transcriptional shut-off experiments. While Rnr1 stability was independent of Ssa paralog, Rnr2 stability was substantially lowered in Ssa3/4 cells upon treatment with HU ( Fig 3B). In contrast, Rnr4 stability was only decreased in Ssa3 cells (Fig 3B).   and RNR4 mRNA levels in Ssa1, 2,3 or 4-expressing yeast. Levels of RNR1, RNR2 and RNR4 mRNAs in Ssa1, 2,3 or 4 cells were determined by reverse transcription and RT-qPCR. Signals of RNR1, RNR2 and RNR4 were normalized against that of ACT1 in each strain, and the resulting ratios in Ssa1 cells were arbitrarily defined as onefold. Data are the average and SD from three replicates � , P�0.05 �� , P�0.01; ��� , P�0.001; ���� , P�0.0001 as compared to indicated strains. (B) RNR subunit stability in yeast expressing single Ssa paralogs. Ssa1, 2,3 or 4-expressing cells transformed with either pGAL1-HA-Rnr1, 2 or 4 plasmids were grown to mid-log phase in YP Galactose medium. RNR expression was shut off by addition of 2% glucose to cultures. Cell lysates from these samples were analyzed over time by Western Blotting for the stability of HA-RNR subunit (HA antibody) and loading control (PGK1). The ratio of RNR/PGK was quantified and determined from three biological replicate experiments. https://doi.org/10.1371/journal.pgen.1010079.g003

RNR subunits display a binding preference for Ssa1 and Ssa2
Hsp90, Hsp70 and Hsp40 proteins in yeast and mammalian cells bind RNR subunit components [37][38][39]. In order to determine if the observed RNR instability in Ssa3/4 cells was a consequence of decreased RNR-chaperone interaction, we assessed the physical interaction of Rnr1, Rnr2 and Rnr4 proteins with Ssa1, 2, 3, and 4 by yeast two-hybrid analysis. In both experiments, Rnr1 interacted equally with all four Ssas (Fig 4A), whereas both Rnr2 and Rnr4 displayed a clear binding preference for Ssa1 and Ssa2 (Fig 4B and 4C).

RNR-Ydj1 interaction is weaker in cells solely expressing Ssa3 or Ssa4
Recent studies have revealed that Hsp70 paralogs display differing affinities for their associated co-chaperones [25,43]. In light of our previous work indicating that Ydj1 directly regulates RNR activity [38], we sought to determine whether Ydj1 association with RNR was reduced in Ssa3/4-expressing cells. We immunoprecipitated HA-tagged Rnr1, Rnr2 or Rnr4 from Ssa1, Ssa2, Ssa3 or Ssa4 cells and assessed the association of Ydj1 via Western Blotting. Ydj1 interacted with Rnr1 equally in Ssa1-4 cells in both unstressed and HU-treated cells (Fig 5A). Interestingly, both Ydj1-Rnr2 and Ydj1-Rnr4 interaction decreased in cells solely expressing Ssa3 and Ssa4 (Fig 5B and 5C).

The SBDβ and CTD regions of Ssa2 (542-639) are required for full RNR activity
Hsp70 is comprised of 3 major domains, NBD, SBD and CTD, the first of which is connected to the last two via a flexible linker. Ssa2 and Ssa4 share 84% sequence similarity with the end of the SBD and entirety of the CTD being the least conserved (Fig 6A and 6B). To understand the origin of the functional difference between the Ssa2 and Ssa4 in regard to RNR function we created Ssa2-Ssa4 chimeras based on these regions previously delineated in [13] and assessed their resistance to media containing HU. The Ssa24 construct consists of amino acids (a.a.) 1-542 of Ssa2 fused to a.a. 543-639 of Ssa4 and vice-versa for the Ssa42 construct. Although yeast expressing the Ssa42 was as resistant to HU as Ssa4 cells, Ssa24 cells phenocopied Ssa4 cells ( Fig 6C). In an effort to determine whether the a.a. region 1-542 of Ssa2 controlled the transcriptional output of the DNA damage response, we compared expression of β-galactosidase driven by a DNA-damage responsive promoter (RNR3 promoter-lacZ) in HU-treated cells expressing Ssa2, Ssa4, Ssa24 and Ssa42. In correlation with the previous result, Ssa24 cells were unable to fully activate RNR3 transcription (Fig 6D). Taken together these results suggest the amino acids 542-639 of the Ssas are critical for full RNR activity and thus the cellular response to HU.

Discussion
A fundamental mystery in molecular chaperone research is why cells express so many apparently similar and functionally redundant chaperone paralogs. Historically, it was generally thought that the main differences were in their expression across cells and tissues where the constitutive Hsp70 performed general housekeeping duties and the inducible form protected cells against environmental stress. However, several recent studies have shown that even when chaperone paralogs are expressed in yeast at equivalent levels as the sole cytosolic Hsp70, these cells display dramatically different phenotypes [23-25, [44][45][46].
A challenge in understanding the differential role of the Ssa paralogs is a lack of verified client proteins. While several Hsp70 interactomes have been published under differing conditions and phosphorylation site mutations, validation of these and their cellular effect is still under investigation [37,[47][48][49][50][51]. Excellent attempts to dissect the roles of Ssa paralogs include the well-established Hsp90 client (Ste11) and a non-yeast client, v-Src in addition to the yeast prions [URE3] and [PSI + ] [23,25, 52,53]. In our previous studies, we managed to identify Hsp90, Hsp70 and Hsp40 as key regulators of RNR in yeast and humans [39], providing an ideal system in which to further probe SSA isoform-specific differences.
In this study, we observed distinct differences in the ability of yeast expressing single SSA paralogs to survive insults to their genome integrity. Ssa1 and Ssa2-expressing cells were more resistant to HU, 5-FU and H 2 O 2 than Ssa3 or Ssa4 cells. It is interesting to note that despite the above differences, there appeared to be no difference in the response to MMS between paralogs. This result may reflect the different kinds of DNA damage that these agents inflict on DNA. HU, 5-FU and H 2 O 2 act primarily by causing single-strand damage and replication stress as opposed to MMS which acts to cause double-strand breaks.
Hsp70 and its corresponding co-chaperone Ydj1 have been shown to play a role in the stabilization of the RNR subunits in both yeast and human cells [38,39]. In this study, we observed decreased Rnr2 levels and a lack of HU-inducibility of Rnr4 in Ssa3 and Ssa4-expressing cells. Further dissection of this phenomenon revealed that the lowered levels of Rnr2 were primarily due to increased subunit instability as determined by the promoter shut-off experiments. In contrast, the altered levels of Rnr4 in Ssa3/4 cells were a combination of transcriptional and protein stability effects. The lowered transcription of Rnr4 may be a consequence of altered DDR signaling, especially as RNR levels are directly controlled by the activity of Mec1, Tel1, Rad53 and Rad9, the latter of which is an Ssa1/2 client [54]. It is thus possible that in cells lacking Ssa1 and 2, Rad9 is destabilized leading to an inability to activate DDR and induce Rnr4 expression. While future studies on Ssa paralog interaction with main components of DDR signaling may be informative, our yeast two-hybrid experiments clearly show that Rnr2 and Rnr4 have a binding preference for Ssa1 and Ssa2 compared to their inducible counterparts. Given the amino acid conservation between the four paralogs, such a binding difference is rather striking. Clients of chaperones are processed via their co-chaperones and given that Ydj1 is key for RNR activity, we considered the possibility that paralog-specific binding of RNR subunits may be mediated via Ydj1. Our data in Fig 5 clearly shows this to be the case as Ydj1 interaction with Rnr2 and Rnr4 is decreased in Ssa3/4 cells.
Identifying regions of Hsp70 that determine client specificity remains challenging considering the essential nature of the protein and sequence similarity. Hsp70 is comprised of a nucleotide binding domain (NBD) which is important for co-chaperone binding and ATPase activity, a substrate binding domain (SBD) which is important for client interaction and a Cterminal domain (CTD) that binds co-chaperones [2]. Out of the four Ssa paralogs, Ssa2 cells are the most resistant to DNA damaging agents including HU, while Ssa4 cells are the most sensitive. In order to further dissect the of the sequence determinants for HU resistance, we used Ssa2-Ssa4 chimeras. Interestingly yeast expressing a chimera consisting of the a.a. 1-542 of Ssa2 and the a.a. 542-639 of Ssa4 (Ssa24) was sensitive to HU, pinpointing the CTD domain of Ssa2 as being key for RNR function and resistance to genome perturbing agents. The highest sequence variation between Ssa2 and Ssa4 occurs towards the end of the SBD (specifically the outer-facing region of the "lid") and the unstructured CTD (see Fig 6). Previous studies have identified this region as being important for the binding of co-chaperone proteins. The VEEVD sequence at the end of Hsp70/Ssa1 is critical for interaction with DNAJB-type co-chaperones including Sis1 [55][56][57][58]. However, there is also substantial evidence that the CTD is also important for the binding of Ydj1. Loss of the last 8 amino acids of Ssa1 substantially reduces the Ssa1-Ydj1 interaction [59] and a 20-amino acid motif in Ssa1 containing GGAP repeats was recently revealed to be necessary for Ssa1 to bind to Ydj1 and activate both the cell integrity and heat shock responses [60]. These results parallel those seen with mammalian Hsp70 paralogs, where each paralog displays a clear binding preference for certain co-chaperones [43]. Taken together, our data suggest that the sequence variation in the CTD is primarily responsible for differential recruitment and folding of RNR small subunits. It is worth noting that recent studies have uncovered a non-canonical binding site in Hsp70 required for binding and folding of alpha-synuclein [61]. It is possible that this region and the area that determines RNR subunit binding heavily overlap. It is interesting to note that co-chaperones are thought to bind select clients first and present them to Hsp70 for folding. This provides rationale for why there are so many diverse but related co-chaperones in both yeast and mammalian cells. Our data clearly shows that Ydj1 interaction with RNR is altered in Ssa3 and 4 expressing cells. This suggests that Ydj1 may actually form a complex with Ssa1/2 prior to binding RNR. The reduced binding of Ssa3 and Ssa4 to RNR then would naturally result in reduced binding to Ydj1. Future studies to delineate the structure of the RNR-chaperone complex should resolve some of the fascinating questions. Studies over the past decade have identified numerous posttranslational modifications (PTMs) on chaperones which are collectively known as the "Chaperone Code" [62][63][64]. This code modifies a variety of chaperone properties including localization, stability, and most importantly client and co-chaperone folding. Several PTMs have been identified in the C-terminal region of Hsp70. In future studies we hope to clarify the role of Hsp70 PTMs on interaction with RNR. Similarly, the data presented here and in our previous study clearly show that interaction between Ydj1 and RNR subunits are stress-induced [38]. It is conceivable that this interaction is also mediated by PTMs on either or both co-chaperone/ RNR proteins. Taken together, our data demonstrates a clear role for Ssa1 and Ssa2 (but not Ssa3 or Ssa4) in the maturation and stability of RNR subunits. This makes cellular sense given that Ssa3 and 4 are present at very low levels in cells except in response to proteotoxic stress. RNR complex proteins in yeast exposed to any form of DNA damage at standard temperatures would thus only have access to Ssa1 and 2. It is possible then that the lack of selective pressure on Ssa3 and 4 to be required to bind housekeeping proteins like RNR for cell viability may have contributed to their sequence divergence from Ssa1 and Ssa2 over time. Molecular chaperones (and often their clients) are well-conserved throughout nature. We have previously shown that Hsp70 and Hsp40 also bind RNR subunits in mammalian cells [38,39]. We envisage future studies that probe this complex interaction in mammalian cells, possibly in the hope of identifying novel ways to inhibit RNR in cancer, feasible given the role of DNAJA1 and other co-chaperones in anticancer drug resistance [65,66]
For tagging the genomic copy of RNR1, RNR2 and RNR4 with a HA epitope at the carboxyterminus, the pFA6a-HA-His3MX6 plasmid was used in a manner similar to [38]. A full table of yeast strains and plasmids that were used can be found in Tables A and B in S1 File. For serial dilutions, cells were grown to mid-log phase, 10-fold serially diluted and then plated onto appropriate media using a 48-pin replica-plating tool. Images of plates were taken after 3 days at 30˚C. 200mM HU was used for serial dilutions and to stress yeast cells, a concentration established in Tkach et al. [69].
For IC 50 calculations, cells were grown to mid-log phase, diluted in a sterile 96 well plate in media containing HU, 5-FU, H 2 O 2 , MMS and were 10-fold serially diluted at indicated concentrations. Cells were continuously shaken for 24 hours at 30˚C and the optical density of the reaction was measured at 600nm. The mean and standard deviation from three independent transformants were calculated.

β-Galactosidase assays
For RNR3-lacZ fusion expression experiments, ssa1-4Δ yeast cells expressing single Ssa constructs or Ssa2/4 fusions were grown overnight in SD-URA media at 30˚C and then re-inoculated at OD600 of 0.2-0.4 and then grown for a further 4 hours. Cells were treated with 150 mM or 200 mM HU for 3 hours and then RNR3-lacZ fusion assays were carried out as described previously in Truman et al. [70]. Briefly, protein was extracted through bead beating and protein was quantitated via Bradford assay. The beta-Galactosidase reaction containing 50 μg of protein extract in 1 ml Z-Buffer (30) was initiated by addition of 200 μl ONPG (4 mg/ ml) and incubated at 28˚C until the appearance of a pale-yellow color was noted. The reaction was quenched via the addition of 500 μl Na2CO3 (1M) solution. The optical density of the reaction was measured at 420nm. β-Gal activity was calculated using ((OD420 x 1.7)/(0.0045 x protein x reaction time)), where protein is measured in mg, and time is in minutes. The mean and standard deviation from three independent transformants were calculated.

Galactose promoter shut-off experiments
Ssa1-4Δ yeast cells expressing either Ssa1, 2, 3 or 4 as the sole Hsp70 isoform were transformed with either pGAL1-HA-Rnr1, 2 or 4 plasmids were grown to mid-log phase in YP Gal medium (1% yeast extract, 2% galactose, 2% peptone). Transcription of pGAL1-HA-Rnr1, 2 or 4 was shut off by the addition of 2% glucose to cultures. Aliquots of cells were collected at 0 and 4 hours after the addition of glucose. Cell lysates from these samples were analyzed by Western Blotting for stability of RNR subunit (HA antibody) and loading control (PGK1).

Purification of HA-tagged Rnr1, 2 and 4 from yeast
Ssa1-4Δ yeast cells expressing genomically-tagged HA-Rnr1, Rnr2 and Rnr4 were transformed with Ssa1-4 pRS315 plasmids were grown overnight in SD-LEU media, and then reinoculated into a larger culture of selectable media and grown to an OD600 of 0.800. The cells were then either unstressed or stressed with 200 mM HU for four hours. Cells were harvested and HA-tagged proteins were isolated as follows: Protein was extracted via bead beating in 500 μl binding buffer (50 mM Na-phosphate pH 8.0, 300 mM NaCl, 0.01% Tween-20). 200 μg of protein extract was incubated with 30 μl anti-HA magnetic beads (Sigma) at 4˚C overnight. Anti-HA beads were collected by magnet then washed 5 times with 500 μl binding buffer. After the final wash, the buffer was aspirated and beads were incubated with 65 μl Elution buffer (binding buffer supplemented with 10 μg/ml 3X HA peptide (Apex Bio)) for 1 hour at 4˚C, then beads were collected via magnet. The supernatant containing purified HA-RNR1, 2, and 4 were transferred to a fresh tube, 25 μl of 5x SDS-PAGE sample buffer was added and the sample was denatured for 5 min at 95˚C. 20 μl of sample was analyzed by SDS-PAGE.

Quantitation of yeast RNR subunit transcription
Quantitation of yeast RNR transcription was carried out as in Zhang et al. [71]. Briefly, Ssa1-4Δ yeast cells expressing unique Ssas were grown overnight in YPD media at 30˚C, re-inoculated at OD600 of 0.2-0.4 and then grown for a further 4 hours. Cells were treated with 200 mM for 2 hours and total RNA was extracted from cells using a GeneJet RNA extraction kit. Total RNA (1 μg) was treated with 10 units of RNase-free DNase I (Thermo) for 30 min at 37˚C to remove contaminating DNA. DNAse I activity was stopped by adding 1 μL of 50 mM EDTA and incubating at 65˚C for 10 minutes. cDNA synthesis was carried out by iScript reverse transcriptase (BioRad) on aliquots of 1 μg RNA. The single-stranded cDNA products were used in qPCR on an ABI Fast 2000 real-time PCR detection system based on SYBR Green fluorescence. Sequences of oligo pairs (same as used in [71]) are listed in Table C in S1 File. Signals of RNR1, RNR2 and RNR4 were normalized against that of ACT1 in each strain and the resulting ratios in WT cells were defined as onefold.

Yeast two hybrid analysis
Ssa1-4Δ cells expressing unique Ssas were transformed with the appropriate GAL4 AD and BD fusion proteins. Interaction between Ssa paralogs and RNR subunits was measured via β-galactosidase assays as in [70].
Supporting information S1 File. List of all yeast strains, plasmids and primers used in this study.