A conserved C-terminal RXG motif in the NgBR subunit of cis-prenyltransferase is critical for prenyltransferase activity

cis-Prenyltransferases (cis-PTs) constitute a large family of enzymes conserved during evolution and present in all domains of life. In eukaryotes and archaea, cis-PT is the first enzyme committed to the synthesis of dolichyl phosphate, an obligate lipid carrier in protein glycosylation reactions. The homodimeric bacterial enzyme, undecaprenyl diphosphate synthase, generates 11 isoprene units and has been structurally and mechanistically characterized in great detail. Recently, we discovered that unlike undecaprenyl diphosphate synthase, mammalian cis-PT is a heteromer consisting of NgBR (Nus1) and hCIT (dehydrodolichol diphosphate synthase) subunits, and this composition has been confirmed in plants and fungal cis-PTs. Here, we establish the first purification system for heteromeric cis-PT and show that both NgBR and hCIT subunits function in catalysis and substrate binding. Finally, we identified a critical RXG sequence in the C-terminal tail of NgBR that is conserved and essential for enzyme activity across phyla. In summary, our findings show that eukaryotic cis-PT is composed of the NgBR and hCIT subunits. The strong conservation of the RXG motif among NgBR orthologs indicates that this subunit is critical for the synthesis of polyprenol diphosphates and cellular function.

Dolichyl phosphate is an obligate lipid carrier for protein N-glycosylation, O-mannosylation, C-mannosylation, and glycosylphosphatidylinositol-anchor synthesis in eukaryotic cells and undecaprenyl phosphate is essential for peptidoglycan biosynthesis in bacteria. cis-Prenyltransferase (cis-PT) 2 is the rate-limiting enzyme committed to dolichyl phosphate biosynthesis in Eukaryotes and Archaea, as well as undecaprenyl phosphate biosynthesis in Eubacteria (1,2). Both eukaryotic and prokaryotic cis-PTs belong to a large protein family, well conserved during evolution (1), and the cis-PT family was identified among the 355 protein families that trace to the last universal common ancestor of all cells (LUCA, or the progenote) by phylogenetic criteria (3).
The bacterial enzyme, undecaprenyl diphosphate synthase (UPPS) is a homodimeric enzyme that catalyzes chain elongation of farnesyl diphosphate (FPP) by sequential reactions with eight isopentenyl diphosphate (IPP) molecules. UPPS has been structurally and mechanistically characterized in great detail (4,5); however, the eukaryotic enzyme has not been purified to date. Our group discovered that unlike UPPS, mammalian and fungal cis-PT is heteromeric complex consisting of NgBR (Nus1) and hCIT (dehydrodolichol diphosphate synthase) subunits in human cells (6,7), and these findings were confirmed for a number of plant cis-PTs (8 -12), demonstrating a major difference in the composition of cis-PT activity in prokaryotes and eukaryotes. Moreover, loss of function mutations identified in patients via exome sequencing in either NgBR or hCIT causes a congenital glycosylation disorder and results in severe clinical manifestations including cognitive defects and retinitis pigmentosa (6,(13)(14)(15)(16), and microdeletions within NgBR locus are linked to pediatric epilepsy (17,18). Further phylogenetic analysis of NgBR and UPPS suggests that a C-terminal motif important for mammalian and fungal cis-PT is shared by the single-subunit cis-PT UPPS (1).
In the rubber plant (Hevea brasiliensis), a primary source of natural rubber, an NgBR ortholog is essential for rubber synthase activity (8), confirming previous work in rubber producing lettuce (10) and dandelions (11). In these papers, the enzymatic activity of the rubber synthase complex is inferred to be solely attributable to the hCIT component of complex. Moreover, the NgBR ortholog is proposed to serve as a scaffold necessary for targeting the hCIT ortholog to the rubber particle monolayer but not critical for catalysis. In the present study, we have purified and characterized for the first time a human heteromeric cis-PT complex composed of NgBR and hCIT subunits. Biochemical characterization of purified WT and various mutants of the NgBR-hCIT complex shows that both subunits contribute to catalytic activity. Furthermore, we provide evidence that a conserved C-terminal motif RXG motif is critical for enzyme catalysis in both two-component and single-subunit cis-PTs.
orthologs define the position of cis-PT homology domain among NgBR orthologs (Fig. 1). Analysis of the architecture of representative cis-PT homology domains revealed a number of characteristic features for each of the group (single-subunit orthologs, hCIT orthologs, and NgBR orthologs) separately or shared between two groups (1,4,7,10,19). Only four of the previously identified five conserved regions (4) are present in all three groups. The NgBR group lacks region II and has highly degenerate regions I and III. Among other conserved residues, the NgBR class is missing a highly conserved stretch of the six amino acid residues involved in FPP binding and catalysis in region I (boxed region) found in UPPS and hCIT. Regions IV and V, including the predicted dimer interface based on the structural data obtain for the UPPS of Micrococcus luteus and Escherichia coli (20,21), are well conserved among all three groups. Finally, the NgBR group shares with homomeric enzymes, but not with the hCIT group, an RXG C-terminal conserved motif (see red arrowheads) (1,6), with Gly being absolutely conserved in EcUPPS and NgBR classes. The Arg residue is substituted with Asn in SaUPPS and a majority of plant and fungal NgBR orthologs (6, 8 -12, 22). In most cases, X is any nonpolar amino acid however putative Hypocreales NgBR orthologs, as well as UPPS of some Trypanosomatidae have positively charged His or Arg at this position. Meta-analysis of available experimental data concerning cis-PTs suggests that both the catalytic motif of region I and the C-terminal RXG motif are indispensable for enzymatic activity being part of the same subunit or separated between hCIT (catalytic motif) and NgBR (RXG) class of cis-PTs proteins. To directly examine the contribution of each subunit to enzymatic activity of the heteromeric hCIT-NgBR complex, a purification scheme of the complex was undertaken.

Purification and biochemical characterization of human cis-PT
Optimal expression and tagging strategy was first established using a previously described expression system in a triple knock-out strain lacking nus1⌬, rer2⌬, srt1⌬ in Saccharomyces cerevisiae (6). Cell survival was ensured because of expression of Glcis-PT on the URA3 plasmid, and cells were co-transformed with the LEU2 and MET15 plasmids bearing WT or epitope-tagged versions of hCIT and NgBR, respectively. The yeast cells were streaked onto complete plates or synthetic complete medium containing 1% 5-fluoroorotic acid (FOA). Because Ura3 protein, which is expressed from the URA3 marker present in the plasmids, converts FOA to toxic 5-fluorouracil, the survival of the yeast cells on the FOA plate depends on the functionality of the expressed NgBR-hCIT complex. hCIT isoform 1 (UniProtKB no. Q86SQ9-1) and isoform 2, (UniProtKB no. Q86SQ9-2), but not isoform 3 (UniProtKB no. Q86SQ9-3), supported the growth of yeast when co-expressed

Heteromeric mammalian cis-prenyltransferase
with NgBR in the triple delete strain (data not shown). N-terminal tagging, but not C-terminal tagging, of hCIT generated a construct indistinguishable from WT, non-tagged hCIT in yeast. Both N-and C-terminal tagging of NgBR reduced its stability and activity when expressed in yeast; therefore a His 6 tag was placed internally after Gly 31 between the putative signal anchor and TM1 ( Fig. 2A). This tag did not affect the growth phenotype and cis-PT activity in yeast co-expressing N-terminally Strep-tagged hCIT.
Next, the cDNAs encoding His 6 -NgBR and Strep-tagged-hCIT were cloned into an internal ribosome entry site (IRES)containing bicistronic vector, allowing the simultaneous expression of two proteins. This was necessary based on our previous work demonstrating that co-translation of both subunits was required for activity in an in vitro translation system (6). To generate protein for purification, the construct was transiently transfected into Expi293F cells and cells collected 72 h later by centrifugation. The pellet was washed with PBS and lysed in detergent free buffer (Ex). To prefractionate cell extracts prior to affinity chromatography, crude membrane extracts were subjected to ultracentrifugation, pellet (P) was solubilized by Dounce homogenization in the presence with 0.5% Triton X-100, and solubilized protein was cleared by a second ultracentrifugation step (SUP). N-terminally Streptagged hCIT and internally tagged His 6 -NgBR complex were purified using a dual affinity purification scheme ( Fig. 2B; see Table 1 for enrichment). Each step of purification was monitored by Western blotting and measurement of cis-PT activity, and this scheme enriched the specific activity of the enzyme 1,700-fold over the starting material. As judged by SDS-PAGE and Coomassie staining of each fraction, the hCIT-NgBR complex was ϳ95% pure (Fig. 2C), and the purified heteromeric complex ran as monodispersed peak on size-exclusion chromatography (Fig. 2D, solid line) that tracked with cis-PT activity (Fig. 2D, dashed line).
It was previously reported that exogenous lipids stimulate the activity of semipurified M. luteus UPPS (30) and phospholipids may modulate the functional properties of several membrane proteins (31). To test the influence of lipids on human cis-PT activity, Triton X-100 was removed from the enzyme preparation by buffer exchange, and activity was measured upon the addition of Triton X-100 or different phospholipids at concentration above their critical micellar concentration. As seen in Fig. 3D, phosphatidic acid modestly increased enzymatic activity (1.8-fold), whereas other phospholipids had a marked effect in activating the enzyme. Cardiolipin increased cis-PT activity 7-fold, and phosphatidylcholine, phosphatidylethanolamine, phosphatidylinositol, and phosphatidylserine increased the activity by ϳ8 -12-fold, suggesting that the lipid environment strongly impacts cis-PT activity.

Phenotypic analysis and biochemical characterization of the purified proteins support importance of the RXG motif
To examine the role of the conserved RXG motif in hcis-PTenzymatic activity and to confirm the predicted position of the cis-PT homology domain in NgBR, we performed pheno- Incorporation of 14 C-labeled IPP into organic fraction was measured as a readout for cis-PT activity (right y axis, red line).

Heteromeric mammalian cis-prenyltransferase
typic analysis of a series of truncation and point mutants of NgBR or hCIT using the yeast triple deletion strain (rer2⌬, srt1⌬, and nus1⌬) mentioned above. As seen in Fig. 4A, no growth defects were observed with complexes containing the following mutations compared with WT hcis-PT-NgBR H100A (His is the first amino acid in region I and is highly conserved among proteins bearing cis-PT homology domain), NgBR R290H (a human mutation causing a glycosylation disorder), ⌬85-NgBR (a truncated protein lacking the first 84 amino acids including TM domain conserved across eukaryotic orthologs of NgBR), and hCIT K42E (a human mutation of hCIT that causes retinitis pigmentosa). Based on our previous work (6), this results suggest that at least 20% residual enzymatic activity of above mutants is necessary to generate sufficient amount of dolichol to support intact glycosylation and survival in yeast. In addition to the above variants, we sought to test the essentiality of the conserved Asp 34 in hCIT, which corresponds to the strictly conserved Asp 26 of EcUPPS involved in catalysis (20,23,32). As seen in Fig. 4A, the hCIT D34A -NgBR complex did not support the growth of the triple knock-out yeast cells, confirm-ing its predicted role in enzymatic activity of hcis-PT. Furthermore, cells expressing a truncated ⌬101-NgBR (lacking the first 100 amino acids) had a severe growth defect, suggesting that the first conserved region in NgBR, despite its degeneration, is important for function. This finding is in line with the recent structural analysis of the plant Z,Z-farnesyl diphosphate synthase (zFPPS), suggesting that the N terminus of the enzyme might extend to the active site of the neighboring monomer near the C terminus (33).
Finally, the conservation of the RXG motif was confirmed by a number of modifications at the C terminus of NgBR. Deletion of the C-terminal K 293 (NgBR 292 ), the addition of Ala to NgBR (NgBRϩA 294 ), HA tagging (NgBR-HA), or NgBR G292A substitutions all reduced growth documenting the importance of the last four residues in NgBR for function (Fig. 4A). Next, we compared the steady-state activities of purified mutants able to support the growth of yeast compared with the purified WT enzyme. We did not include the N-terminal truncation variants of NgBR because introducing the epitope tag after the predicted TM1 (35-56 amino acids) rendered the enzyme inactive  Heteromeric mammalian cis-prenyltransferase in yeast. As shown in Fig. 4B, all mutants analyzed displayed lower enzymatic activity compared with WT enzyme. In addition, mutations that strongly impacted growth in yeast (NgBR G292A , NgBR 292 , and NgBRϩA 294 ) had the lowest enzymatic activity under varying concentrations of co-substrates FPP and IPP ( Fig. 5 and Table 2). These data support the important role of C terminus, in particular Gly 292 in cis-PT activity.
To further characterize hcis-PT and to elucidate the impact of mutations on catalysis and substrate binding, steady-state kinetic parameters were measured for the complexes containing the WT subunits, hCIT-NgBR, hCIT K42E/ NgBR, hCIT-NgBR H100A , hCIT-NgBR R290H , and hCIT-NgBR G292A ( Table  2 and Fig. 5, A and B). The K42E substitution of hCIT reduced k cat and increased the K m for FPP but not the affinity for IPP. NgBR H100A and NgBR R290H decreased the k cat ϳ4-fold compared with WT enzyme. However, NgBR H100A did not exhibit marked changes in affinity for IPP or FPP, whereas NgBR R290H had moderately but significantly reduced affinities for IPP. Substitution of G292A in NgBR, which markedly reduced the growth phenotype in yeast, decreased the catalytic activity of the enzyme to an even greater extent with a k cat that is 10-fold lower than that of the WT enzyme. In this case, there were modest (5-fold) but significant differences in the K m for IPP and a slight decrease in the K m for FPP implying that C-terminal mutations in NgBR markedly affected enzymatic activity and IPP binding.   , the rer2⌬, srt1⌬, nus1⌬ triple deletion strain expressing G. lamblia cis-PT from URA3 plasmid was co-transformed with the LEU2 and MET15 plasmids bearing wild-type or mutated variant of hCIT and NgBR as indicated. The cells were streaked onto complete plates (YPD) or synthetic complete medium containing 1% FOA. The Ura3 protein, which is expressed from the URA3 marker present in the plasmids, converts FOA to toxic 5-fluorouracil. The growth of cells was monitored over time to assess phenotypic differences. The combination of alleles affecting the growth is indicated in italic, and the combination not supporting the growth is marked in red. B, cis-PT activity was measured using purified WT or mutant hcis-PT complex as described under "Experimental Procedures." The values are the means Ϯ S.D. of three to nine independent measurements from at least two independent isolations of enzyme except for hCIT-NgBRϩA 294 and hCIT-NgBR 292 performed on one batch.

Heteromeric mammalian cis-prenyltransferase The RXG motif in homomeric EcUPPS and Glcis-PT is critical for cis-PT function and enzymatic activity
Comparison of the primary amino acid sequences of singleand two-component enzymes reveals that both classes share a conserved RXG C-terminal motif (Fig. 1). Based on the crystal structure of UPPS, Arg 242 in the RXG motif is involved in the binding of the diphosphate group of IPP (32). Furthermore, the role of C terminus in IPP binding and catalysis is supported by structural information obtained for decaprenyl diphosphate synthase of Mycobacterium tuberculosis and recent studies on zFPPS of Solanum habrochaites (33,34), implicating the C terminus in IPP binding. To verify the importance of the RXG motif in homomeric cis-PTs, we assayed growth of the triple deletion yeast strain, transformed with WT or mutant forms of homomeric cis-PT, namely EcUPPS (E. coli UPPS) and Glcis-PT (Giardia lamblia UPPS). The cells were transformed with LEU2 plasmids bearing WT, R242H, and G244A of EcUPPS or WT, R236H, and G238A of Glcis-PT, and the growth was monitored over 7 days. As seen in Fig. 6A, WT E. coli and G. lamblia cis-PTs exhibited normal growth phenotypes on YPD and FOA plates and Arg to His mutants because each construct had similar phenotypes compared with WT transformed cells. In contrast, UPPS G244A and Glcis-PT G238A mutants failed to grow on FOA plates, demonstrating an essential role of G in the RXG motif in homomeric cis-PT function. In addition, cis-PT activity was measured using purified WT, R236H, and G238A mutants of Glcis-PT. Mutation of both residues reduced activity, with Glcis-PT R236H having ϳ5 times lower activity and Glcis-PT G238A being virtually inactive (Fig.  6B) compared with WT.

Presence of heteromeric cis-PT in Methanosarcina acetovorans supports the importance of the RXG motif
Heteromeric cis-PTs was predicted in subgroup of Euryarchaeota, based on the fact that Halomebacteria and Archaeoglobaceae have at least two UPPS orthologs: a putative cis-PT that is closely related to single subunit Archea enzymes but lacks the C-terminal RXG motif and a NgBR/Nus1-like protein (1, 35) (Fig. 1). To test the conservation of this motif, we cloned putative undecaprenyl diphosphate synthase of M. acetovorans consisting of MaUPPS-A (MA3723, hCIT group) and MaUPPS-B (MA4402, NgBR/Nus1 group). The triple deletion strain bearing Glcis-PT on the URA3 plasmid was co-transformed with the LEU2 and MET15 plasmids expressing wildtype or mutated variants of MaUPPS-A and MaUPPS-B, respectively. The growth of yeast cells on the FOA plates was monitored over 5 days. As seen in Fig. 7A, co-expression of

MaUPPS-A and MaUPPS-A is indispensable to support cell growth, but neither MaUPPS-A nor MaUPPS-B is sufficient.
Similarly, as it was observed in case of NgBR, mutation of His 29 (corresponding to His 100 of NgBR) and Arg 221 (corresponding to Arg 290 of NgBR) in MaUPPS-B does not impair growth, but MaUPPS-B G223A substitution caused a severe growth delay. Quantification of the cis-PT activity reveals that each of the analyzed mutation inhibits enzyme activity with most profound effect of MaUPPS-B G223A substitution (Fig. 7B).

Discussion
Here we demonstrate that two subunits of the human cis-PT, NgBR and hCIT, are required to form a functional enzyme. The purification of the cis-PT complex supports prior work showing that co-translation of both subunits is required for polyprenol synthesis in vitro and for survival in yeast lacking orthologs of each component (6). Moreover, the catalytic Asp 34 in hCIT and RXG motif in NgBR, both conserved in homomeric cis-PT such as UPPS, are critical for catalytic activity of the complex. These data highlight the evolutionary conservation of essential elements required for cis-PT function throughout all walks of life.
Eukaryotic cis-PTs were initially presumed to be homomeric based on detailed studies in undecaprenyl diphosphate synthases of E. coli and M. luteus (4, 5, 36 -38). Recent work by us (6,7,19) and others have shown the essential role of NgBR (and its orthologs including Nus1 in S. cerevisiae) and hCIT (and its orthologs including Rer2 and Srt1 in S. cerevisiae) as both being required for cis-PT activity and polyprenol synthesis (6, 8, 10 -12). The expression of only hCIT or only NgBR does not support growth in the nus1⌬, rer2⌬, srt1⌬ strain of S. cerevisiae, and in vitro translation of either subunit is not catalytically

Figure 6. The RXG motif of EcUPPS and Glcis-PT is essential for growth and activity.
A, the rer2⌬, srt1⌬, nus1⌬ triple deletion strain expressing G. lamblia cis-PT from URA3 plasmid was transformed with the LEU2 plasmid bearing WT or mutants ORFs for Glcis-PT or EcUPPS. As a negative control, cells were transformed with empty vector. The mutated genes G244A and G238A did not supporting the growth and are underlined. B, cis-PT activity was measured using purified wild-type or mutant Glcis-PT. The data are means Ϯ S.D. values of three technical replicates.

Heteromeric mammalian cis-prenyltransferase
active (6). At first glance, this may appear to be in conflict with previous reports showing that eukaryotic hCIT orthologs heterologously expressed in S. cerevisiae were active without coexpression of NgBR/Nus1 orthologs. However, this can be explained by the presence of the endogenous Nus1 in S. cerevisiae. Furthermore, a number of hCIT orthologs in plants and protists that were studied in yeast are single-subunit enzymes that have been acquired through horizontal gene transfer and are more closely related to EcUPPS than to hCIT (1, 28, 39 -43). This complexity can be overcome and simplified by complementation experiments in the nus1⌬, rer2⌬, srt1⌬ strain as performed previously (6,11), where it is clear that both NgBR and hCIT or a single-subunit enzyme is required. Previous experiments have shown that co-expression of NgBR with hCIT results in their interaction and stabilization based on co-immunoprecipitation experiments (6,7). Whether NgBR/Nus1 serves as an accessory subunit for docking the complex to the ER membrane or is a structural component contributing to catalysis and substrate binding is still a matter of debate. Experiments showing that 1) the loss of NgBR does not lead to changes in the relative amount of hCIT associated with ER membrane fractions (7) and 2) a mutation in the C terminus of NgBR causes a congenital disorder of glycosylation syndrome because of defective dolichol synthesis (6) support the role of NgBR as a structural component of hcis-PT involved in catalysis. The idea of NgBR being solely a docking component was recently reported by Yamashita et al. (8), who identi-fied a three-component system composed of HRBP (a NgBR ortholog), HRT (a hCIT ortholog) and rubber elongation factor (44), all of which were required for long-chain prenol or natural rubber synthesis when reconstituted into washed rubber particles. Sequential alignment of nonredundant proteins bearing cis-PT homology domains presented in this paper enabled the observation of a highly conserved C-terminal RXG motif shared between NgBR/Nus1 and the UPPS group of enzymes. Furthermore, meta-analysis of available experimental data concerning cis-PT suggests that the C-terminal RXG motif is indispensable for enzymatic activity. To further elucidate the role of NgBR in hcis-PT activity, we successfully purified the WT NgBR-hCIT complex and its mutated versions to homogeneity and biochemically characterized the proteins. Mutations in either subunit of hcis-PT reduces enzymatic activity in comparison to the WT enzyme, and these results are in agreement with the predicted function of the analyzed residue in homomeric cis-PTs.
The first observation supporting a functional role of the extreme C terminus of NgBR regulating enzyme activity is derived from patients harboring an R290H mutation in NgBR, which reduces activity in vivo and in vitro (6) and now seen with the purified complex (Table 2) by 70 -80%. In addition, epitope tagging the C terminus of NgBR reduces activity (Fig. 4), and this has been previously shown in the cis-prenyltransferase (CPT)-like NgBR ortholog of Lactuca sativa (10). Crystallographic data obtained for homomeric EcUPPS show that Arg 242 in the RXG motif is involved in Mg ϩ2 -IPP binding (32). In addi-

Figure 7. Phenotypical and biochemical analysis of heteromeric MaUPPS supports predicted length of cis-PT domain of NgBR and importance of C-terminal RXG motif.
A, the rer2⌬, srt1⌬, nus1⌬ triple deletion strain expressing G. lamblia cis-PT from URA3 plasmid was co-transformed with the LEU2 and MET15 plasmids bearing wild-type or mutated variant of MaUPPS-A (hCIT/Rer2/Srt1 ortholog) and MaUPPS-B (NgBR/Nus1 ortholog) as indicated. Cells transformed with empty plasmid were used as negative control, and cells transformed with LEU2 Glcis-PT or co-transformed with hCIT (LEU2) and NgBR (MET15) were used as positive control. The cells were streaked onto YPD or synthetic complete medium containing 1% FOA. The growth of cells was monitored overtime to assess phenotypic differences. The combination of alleles affecting the growth is indicated in italics, and the combination not supporting the growth is underlined. B, cis-PT activity was measured using S. cerevisiae rer2⌬, srt1⌬, nus1⌬ membrane fractions expressing WT or mutated MaUPPs subunits as indicated. The data are means Ϯ S.D. values of four or five replicates in two independent experiments.

Heteromeric mammalian cis-prenyltransferase
tion, the role of the C terminus in IPP binding and catalysis is supported by structural information obtained for decaprenyl diphosphate synthase of M. tuberculosis (34) and zFPPS (33). In the structure of diphosphate synthase, the extreme C terminus of one monomer interacts with the active site of the other subunit. Based on our biochemical data and the available structural information for homomeric cis-PTs, it is likely that Arg 290 in the RXG motif of NgBR is involved in IPP binding, and the flexible nature of glycine in the motif may permit conformational changes to seal the active site cavity. Collectively, these data imply that both subunits contribute to enzymatic activity, and the C-terminal tail of NgBR regulates aspects of cis-PT catalytic activity.
The general function of NgBR orthologs as a cis-PT subunit is further supported by the heteromeric cis-PT from M. acetovorans. MaUPPS-B (NgBR ortholog) is missing any predicted TM domains, and yet the MaUPPS-A-MaUPPS-B complex is indispensable for supporting the growth of triple deletion strain of S. cerevisiae. Because the mutations in the conserved region of MaUPPS-B have similar impact on MaUPPS activity as those in distantly related hcis-PT, we postulate a common mechanism for both enzymes.
In summary, this study strongly advances the concept that eukaryotic cis-PT is composed of two subunits, NgBR and hCIT, and this two component system is conserved in M. acetovorans. Interestingly, conservation of the RXG motif across phyla implicates a catalytic role of NgBR orthologs in the synthesis of polyprenol diphosphates critical for cellular function. Future structural studies on the heteromeric NgBR-hCIT complex will permit a deeper mechanistic understanding of how the C terminus regulates substrate binding and catalysis.

Materials
Unless otherwise stated, all reagents were of analytical grade and purchased from Sigma-Aldrich, Thermo Fisher Scientific, and Zymo Research (Irvine, CA). Restriction enzymes were from New England Biolabs (Ipswich, MA). [1-14 C]IPP (50 mCi/ mmol) was purchased from American Radiolabeled Chemicals (St. Luis, MO). Reverse phase thin layer chromatography (RP18-HTLC) plates were from Merck. Undecaprenol was obtained from the isoprenoids collection from the Institute of Biochemistry and Biophysics (Polish Academy of Science). Primary antibodies used in this study include ␣-NgBR (Abcam, ab168351), ␣-dehydrodolichol diphosphate synthase (Sigma, SAB2100572), and ␣-Strep-tag (IBA Solutions for Life Sciences, StrepMAB-Classic). Site-directed mutagenesis was performed using TagMaster site-directed mutagenesis kit (GM Biosciences). The E. coli UPPS was amplified from genomic DNA.

M. acetovorans ORFs encoding MaUPPS-A (MA3723) and
MaUPPS-B (MA4402) were amplified from synthetically synthetized codon optimized for expression in S. cerevisiae gBlocks Gene Fragments (ITD Integrated Technologies, Coralville, IA). Glcis-PT, NgBR, and hCIT isoform 1 were amplified from previous plasmids. The cassette containing synthetic intron and IRES was amplified from pIRESneo1 (Clontech). The Invitrogen Gateway cloning strategy was used to insert cDNA into yeast expression vectors. All PCR products were cloned into pCR8/GW/TOPO TA (Invitrogen), sequenced, and subcloned into final yeast expression vector pKG-GW1 or pKG-GW2. To express His-Strep tag II-Glcis-PT in bacteria, the Strep tag II-Glcis-PT was amplified with restriction enzyme recognition sites, and the Strep tag sequence was inserted in the primers. The PCR product was ligated in frame with the internal His 6 tag of the pRSF-DUET1 plasmid.
To express both subunits of hcis-PT from single mRNA in mammalian cells, the IRES surrounded by two multicloning sites was introduced into pCEP4 (Invitrogen) using NEBUilder HiFi DNA Assembly (NEB) to obtain the pKGmDUET vector. His 6 internally tagged NgBR was subcloned into BamHI/NotI sites of pKGmDUET from pKG-GW2-(31-HIS)NgBR plasmid, and SmaI/EcoRI/Klenow-treated Strep-hCIT was subcloned into PmeI site from pKG-GW1-Strep-hCIT plasmid.

Yeast complementation assay
For yeast complementation analysis of cis-PTs, S. cerevisiae strains KG405 (nus1⌬ rer2⌬ srt1⌬), carrying the Glcis-PT gene on a plasmid with a URA3 marker was used (45). To analyze homomeric Glcis-PT and EcUPPS mutants, strain KG405 was transformed with vector pKG-GW1 (leucine selection) carrying WT or mutated versions of corresponding genes or empty vector as negative control. To phenotypically analyze hcis-PT, strain KG405 was transformed with vectors pKG-GW1 carrying hCIT variants (leucine selection) and pKG-GW2 carrying NgBR variants (methionine selection) in combination or empty vectors as negative control. To analyze putative heteromeric UPPS of M. acetovorans (MaUPPS), strain KG405 was transformed with vector pKG-GW1 carrying MaUPPS-A (MA3723, hCIT ortholog) and vector pKG-GW2 carrying WT and mutated variants of MaUPPS-B (MA4402, NgBR ortholog) in combination or with the corresponding empty vectors as negative control. Transformed yeast cells were grown overnight at 30°C in synthetic defined medium lacking uracil and leucine or lacking uracil, methionine, and leucine were streaked onto synthetic defined medium containing all amino acids, nucleotide supplements, and 1% (w/v) 5-FOA (Zymo Research) and onto YPD plates. The plates were incubated for up to 7 days at 30°C. Colonies growing on the 5-FOA plates were streaked on synthetic defined medium lacking uracil and incubated at 30°C for 3 days to verify the loss of the pNEV-Glcis-PT plasmid. Yeast strain KG405 and its derivative carrying MaUPPS complex, hcis-PT complex or single subunits enzymes expressed from pKG-GW1 plasmid instead of pNEV-Glcis-PT were cultured in 2% (w/v) Bacto peptone and 1% (w/v) yeast extract supplemented with 2% glucose (w/v) (YPD). Synthetic minimal media were made of 0.67% (w/v) yeast nitrogen base and 2% (w/v) supplemented with auxotrophic requirements. For solid medium, agar (BD Biosciences, Sparks, MD) was added at a 2% (w/v) final concentration. Yeast cells were transformed using the Frozen-EZ yeast transformation II kit (Zymo Research).

Purification of human cis-PT
To purify the human hcis-PT complex, constructs containing internally tagged His 6 -NgBR and N-terminally tagged Strep-hCIT(pKGmDUET-hCIT-NgBR) were transiently trans-Heteromeric mammalian cis-prenyltransferase fected in 200 ml of culture of Expi293F cells according to the manufacturer's protocol (Invitrogen). The cells were harvested 72 h post-transfection by centrifugation and washed with PBS. Each gram of the cells was resuspended in S-Tactin buffer (100 mM Tris-HCl, pH 8, 300 mM NaCl, 2 mM 2-mercaptoethanol, 1 mM MgCl 2 ), and cells were disrupted by sonication on ice. Unbroken material was cleared with a 1,000 ϫ g centrifugation, and the supernatant from this spin was recentrifuged at 200,000 ϫ g for 30 min at 4°C to obtain total membrane fractions. Total membrane fraction was homogenized in S-Tactin buffer supplemented with 0.5% Triton X-100 to solubilize membrane proteins. Solubilization was followed by additional 200,000 ϫ g centrifugation. Strep-hCIT was purified from 200,000 ϫ g supernatant using the Strep-Tactin XT system (IBA GmbH). The Strep-hCIT-His 6 -NgBR complex was eluted from Strep-Tactin XT resin with 100 mM Tris-HCl, pH 8, 300 mM NaCl, 2 mM 2-mercaptoethanol, 1 mM MgCl 2 , 0.1% Triton X-100, 5% glycerol, 50 mM biotin elution buffer. The Strep-Tactin purification of Strep-hCIT was followed by nickel column purification of His 6 -NgBR using HisPur nickel-nitrilotriacetic acid resin (Thermo Fisher Scientific). Sephadex G-25 in PD-10 desalting columns (GE Healthcare Life Sciences) were used to exchange buffer for 20 mM NaP i , pH 8, 300 mM NaCl, 2 mM 2-mercaptoethanol, 1 mM MgCl 2 , 0.1% Triton X-100, 20% glycerol. Purification efficiency was tracked by Western blot analysis of each fraction, Coomassie staining of SDS-PAGE of final eluate, and measurement of specific cis-PT activity ( Fig. 1 and Table 1).

Purification of Glcis-PT
To purify Glcis-PT, the pRSF-DUET1-Glcis-PT plasmid was transformed into E. coli Rosetta 2 cells (Novagen). E. coli was grown in autoinduction medium (46) until the logarithmic growth phase at 37°C flowed 24 h of incubation at 17°C to express heterologous protein. The cells were harvested by centrifugation, washed with PBS, and stored Ϫ80°C. The proteins were extracted using B-PER reagent (Thermo Fisher Scientific) supplemented with 1 mM MgCl 2 and 2 mM 2-mercaptoethanol. The Glcis-PT was first purified using nickel column HisPur nickel-nitrilotriacetic acid resin (Thermo Fisher Scientific) followed by the Strep-Tactin XF (IBA GmbH) purification. Sephadex G-25 in PD-10 desalting columns (GE Healthcare) were used to exchange buffer for 20 mM NaP i , pH 8, 300 mM NaCl, 2 mM 2-mercaptoethanol, 1 mM MgCl 2 , 0.1% Triton X-100, 20% glycerol. Purification efficiency was tracked by Western blot analysis and Coomassie staining of SDS-PAGE of final eluate.

Size-exclusion chromatography
The size-exclusion chromatography was carried out using an ÄKTA purifier (GE Healthcare) with the size-exclusion column Superdex 200 10/300 GL (GE Healthcare) at flow rate of 0.5 ml/min. The UV absorption at 254 nm was measured as a readout for protein elution. The used buffer contained 50 mM Tris-HCl buffer, pH 8.0, and 300 mM NaCl, 2 mM 2-mercaptoethanol, 1 mM MgCl 2 , 0.1% Triton X-100, 5% (v/v) glycerol. The column was calibrated by using the low molecular weight and high molecular weight calibration kit (GE Healthcare). For determination of the void, volume blue dextran (2,000 kDa; GE Healthcare) was used. Size determination was calculated based on the standard linear equation based on the calibration of the column.

cis-PT enzymatic activity of hcis-PT
Standard incubation mixture contained, in a final volume of 50 l, 50 M FPP, 100 M [1-14 C]IPP (55 mCi/mmol), 50 mM Tris-HCl, pH 8, 1 mM MgCI 2 , 20 mM 2-mercaptoethanol, 1 mg/ml BSA, 10 mM KF, and 1% (w/v) phosphatidylinositol. Membrane or crude protein (100 g) or 20 -100 ng of purified protein was used for activity assays. In some experiments with crude membranes, zaragozic acid A (10 M) was added. Reactions were incubated for 60 min at 37°C and terminated by the addition of 1 ml of chloroform-methanol 3:2. The protein pellet was removed by centrifugation, and the supernatant was washed three times with 1/5 volume of 10 mM EDTA in 0.9% NaCl. The incorporation of [ 14 C]IPP into organic fractions containing polyprenyl diphosphate was measured by scintillation counting.

Kinetic parameters
Standard 25 or 100 l (hCIT-NgBR G292 mutant) reaction mixture containing 9 -25 nM hcis-PT was used. To measure kinetic parameters for FPP, 0.1-50 M FPP was use along with 100 M IPP for wild-type enzyme, hCIT K42E and NgBR H100A mutants, 200 M IPP for NgBR R292H mutant, and 400 M IPP for NgBR G292 mutant. To measure kinetic parameters for IPP, 0.1-400 M IPP was used along with 50 M FPP. The initial velocity data were fitted to the Michaelis-Menten equation using the GraphPad Prism 7.02 computer program (GraphPad Software, Inc.) to obtain K m values. k cat values were obtained from the Michaelis-Menten equation for IPP.

cis-PT enzymatic activity of MaUPPS
For S. cerevisiae expressing MaUPPS complex membrane fractions were prepared as described (47), and cis-PT activity measured (47,48)  Heteromeric mammalian cis-prenyltransferase washed three times with 1 ⁄ 5 volume of 10 mM EDTA in 0.9% NaCl. The organic phase was concentrated under a stream of nitrogen. Then organic solvent was evaporated, and lipids were loaded onto HPTLC RP-18 precoated plates with a concentrating zone and run in acetone containing 50 mM H 3 PO 4 . The plates were exposed to film to visualize the products of IPP incorporation. To measure incorporation of radioactive IPP into polyprenol fraction, the gel from the zone containing radiolabeled polyprenols was scraped and subjected to liquid scintillation counting. To analyze the length of polyprenol diphosphates before running the TLC analysis lipids were subjected to mild acid hydrolysis. Lipids were loaded onto HPTLC RP-18 together with undecaprenol as internal standard and subjected to radioautography as was described above.
Author contributions-K. A. G. designed all the expression systems for NgBR-hCIT, generated yeast strains, performed enzymatic assays on the NgBR-hCIT complex, and wrote the manuscript; B. H. E. cloned, expressed, purified, and characterized Glcis-PT WT and mutant enzymes and edited the manuscript; E. J. P. cloned and characterized NgBR constructs and participated in experimental design and writing; J. R. K. assisted with purification strategies and gel filtration chromatography; and W. C. S. supervised overall project and wrote manuscript.