Coupling Of The Dolichol-P-P-Oligosaccharide Pathway To Translation By Perturbation-Sensitive Regulation Of The Initiating Enzyme, GlcNAc-1-P Transferase

In mammalian cells, inhibition of translation interferes with synthesis of the lipid-linked oligosaccharide (LLO) Glc3Man9GlcNAc2-P-P-dolichol as measured with radioactive sugar precursors. Conflicting hypotheses have been proposed, and the fundamental basis for this regulation has remained elusive. Here, fluorophore-assisted carbohydrate electrophoresis (FACE) was used to measure LLO concentrations directly in cells treated with translation blockers. Further, LLO biosynthetic enzymes were assayed in vitro with endogenous acceptor substrates using either cells gently permeabilized with streptolysin-O (SLO) or microsomes from homogenized cells. In Chinese hamster ovary (CHO)-K1 cells treated with translation blockers, FACE did not detect changes in concentrations of Glc3Man9GlcNAc2-P-P-dolichol or early LLO intermediates. These results do not support earlier proposals for feedback repression of LLO initiation by accumulated Glc3Man9GlcNAc2-P-P-dolichol, or inhibition of a GDP-mannose dependent transferase. With microsomes from cells treated with translation blockers, there was no interference with LLO initiation by GlcNAc-1-P transferase (GPT), mannose-P-dolichol synthase, glucose-P-dolichol synthase, or LLO synthesis in vitro, as reported previously. Surprisingly, inhibition of all of these was detected with the SLO in vitro system. Additional experiments with the SLO system showed that the three transferases shared a limited pool of dolichol-P that was trapped as Glc3Man9GlcNAc2-P-P-dolichol by translation arrest. Overexpression of GPT was unable to reverse the effects of translation arrest on LLO initiation, and experiments with FACE and the SLO system showed that overexpressed GPT was not functional in vivo, although it was highly active in microsomal assays. Thus, the combined use of the SLO in vitro system and FACE showed that LLO biosynthesis depends upon a limited primary pool of dolichol-P. Physical perturbation associated with microsome preparation appears to make available a secondary pool of dolichol-P, masking inhibition by translation arrest, as well as activating a nonfunctional fraction of GPT. The implications of these results for the organization of the LLO pathway are discussed.


Summary
In mammalian cells, inhibition of translation interferes with synthesis of the lipid-linked oligosaccharide (LLO) Glc 3 Man 9 GlcNAc 2 -P-P-dolichol as measured with radioactive sugar precursors. Conflicting hypotheses have been proposed, and the fundamental basis for this regulation has remained elusive. Here, fluorophore-assisted carbohydrate electrophoresis (FACE) was used to measure LLO concentrations directly in cells treated with translation blockers. Further, LLO biosynthetic enzymes were assayed in vitro with endogenous acceptor substrates using either cells gently permeabilized with streptolysin-O (SLO) or microsomes from homogenized cells.
In CHO-K1 cells treated with translation blockers, FACE did not detect changes in concentrations of Glc 3 Man 9 GlcNAc 2 -P-P-dolichol or early LLO intermediates. These results do not support earlier proposals for feedback repression of LLO initiation by accumulated Glc 3 Man 9 GlcNAc 2 -P-P-dolichol, or inhibition of a GDP-mannose dependent transferase. With microsomes from cells treated with translation blockers there was no interference with LLO initiation by GlcNAc-1-P transferase (GPT), mannose-P-dolichol synthase, glucose-P-dolichol synthase, or LLO synthesis in vitro, as reported previously. Surprisingly, inhibition of all of these was detected with the SLO in vitro system. Additional experiments with the SLO system showed Introduction In eukaryotes, the lipid-linked oligosaccharide (LLO) 1 Glc 3 Man 9 GlcNAc 2 -P-P-dolichol serves as the donor of oligosaccharide units that are transferred by oligosaccharyltransferase (OT) to appropriate asparaginyl residues in nascent polypeptides within the lumen of the endoplasmic reticulum (ER), forming glycoproteins with asparagine-linked (N-linked) Glc 3 Man 9 GlcNAc 2 glycans (1). The pathway for LLO synthesis has been elucidated by a combination of biochemical and genetic methods, and to date mutations in seven genes essential for LLO synthesis have been identified as the causes of Congenital Disorders of Glycosylation (CDG) Types Ia-g (2,3). LLO synthesis is initiated by the transfer of GlcNAc-1-P from UDP-GlcNAc to dolichol-P by a specific tunicamycin (TN)-sensitive GlcNAc-1-P transferase (GPT) (4). This reaction occurs on the cytoplasmic face of the ER membrane (5). GlcNAc-P-P-dolichol is then extended to Man 5 GlcNAc 2 -P-P-dolichol, a key LLO intermediate, by a series of cytoplasmically-oriented reactions that catalyze the transfer of one residue of GlcNAc from UDP-GlcNAc and five residues of mannose from GDP-mannose. Cytoplasmically-oriented Man 5 GlcNAc 2 -P-P-dolichol then flips to the lumenal leaflet (6,7) in a process involving the Rft1 protein (8). This lumenally-oriented 7 Unfortunately, the radiolabeling methods used with intact cells to formulate these hypotheses raised several complications. For example, the ability to monitor early LLO intermediates was limited since, compared with Glc 3 Man 9 GlcNAc 2 -P-P-dolichol, these molecules contain fewer sugar residues that can be labeled. These types of problems prevented accurate measurements of the actual amounts of various LLOs. Further, because only labeled LLOs could be assessed, the fates of unlabeled pre-existing LLOs were not known.
In this study the control of LLO synthesis by translation was re-examined by employing the recently developed fluorophore-assisted carbohydrate electrophoresis (FACE) approach for direct measurement of LLO compositions in intact cells (21). Moreover, measurements of LLO biosynthetic activities in vitro were made either with cells gently permeabilized with streptolysin-O (SLO), or with microsomes prepared by conventional homogenization techniques. Our results show that translation blockers interfere with LLO synthesis by trapping the dolichol-P available for LLO initiation by GPT as Glc 3 Man 9 GlcNAc 2 -P-P-dolichol, and reveal that the function and regulation of GPT, MPDS, and GPD synthase (GPDS) require properties of the ER that are retained in SLO-permeabilized cells but lost in microsomal preparations. The results directly support the hypothesis of Hubbard and Robbins for a limited pool of dolichol-P that is restricted for LLO synthesis. Upon re-evaluation many of the apparently contradictory results from other studies were found to be consistent with this hypothesis, and differences in experimental conditions appear to account for remaining discrepancies.
FACE analyses of unlabeled LLOs were performed as described (21). Briefly, oligosaccharide profiling gels (Glyko) were used to analyze oligosaccharides modified with 8-aminonaphthalene-1,3,6-trisulfonate (ANTS). Monosaccharide composition gels were prepared (N.G., unpublished) similar to those offered by Glyko, but with less interfering background material in the region used to measure chitobiose, and used for saccharides modified with 2-aminoacridone (AMAC). The gel was imaged with a Biorad Fluor-S MultiImager using a 530DF60 filter. Electronic gel images were generated, and individual fluorescent species were quantified with Quantity One software supplied with the scanner.
In vitro systems--Microsomal system: Cells were swollen and homogenized, and microsomes were recovered by centrifugation as described (26). Approximately 80 µg microsomal membrane protein was recovered per 10 7 cells. Streptolysin-O (SLO) permeabilization: Cells were treated on wet ice with SLO (Murex brand, distributed by Corgenix, United Kingdom; supplied as a lyophilized powder containing PBS and reconstituted with ice-cold water), then incubated with 37 degree transport buffer to allow SLO pores to form, exactly as described (27). Experiments directly comparing results with both in vitro systems used the same cell equivalents.
In vitro assays for GPT activity and LLO synthesis--Transport buffer (27) was used in 2 ml assays with both microsomes and SLO-cells. Nucleotide sugar donors were supplied exogenously, but only endogenous dolichol-P was used. For GPT, unless indicated otherwise, assays included 0.1 µCi/ml of UDP-[ 3 H]GlcNAc and were performed at 37 degrees for 30 min. Lipid products from microsomal reactions were recovered by organic extraction as described (28), and those from SLOcells were recovered by extraction of cells with chloroform-methanol (2:1) and back-washing with chloroform-methanol-water (3:48:47). MPDS and GPDS were assayed by the same procedure, except that UDP-[ 3 H]GlcNAc was replaced by 0.1 µCi/ml GDP-[ 3 H]mannose and 0.1 µM UDP-[ 3 H]glucose, respectively. For LLO synthesis, assays included 1 µM UDP-GlcNAc and 0.2 µCi/ml of GDP-[2-3 H]mannose and were performed at 37 degrees for 10 min, at which point 1 µM GDPmannose was added and the assay continued (chase) for another 5 min. LLOs were recovered by extraction into chloroform-methanol-water (10:10:3) as described (29).
Oligosaccharyltransferase acceptor peptides--Both the acceptor peptide Ac-Asn-Tyr-Thr-CONH 2 , described earlier (20), and the non-acceptor peptide Ac-Gln-Tyr-Thr-CONH 2 were obtained by custom synthesis (Synpep Corp.). The supplier reported that the purity of each peptide assessed by HPLC was at least 97%, and the syntheses were confirmed by mass spectroscopy. Peptides were dissolved in pure water.

Results
Experimental model--To aid the reader, Figure 1

Use of FACE to test the effects of translation inhibitors on LLO synthesis--As expected
from prior studies (16), treatments of CHO-K1 cells with cycloheximide or puromycin for 1 hour inhibited protein synthesis by 99% (Figure 1, panel A) and incorporation of [ 3 H]mannose into LLO (CMW 10:10:3 extract) by 98% (panel B). When total LLO compositions from similarly treated cells were determined by FACE (panels C and D) no LLO accumulation, from GlcNAc-P-P-dolichol to Glc 3 Man 9 GlcNAc 2 -P-P-dolichol, was detected. Therefore, these data do not support hypotheses that propose interference with one or more LLO mannosyltransferases (which would cause accumulation of LLO intermediates) (18) or feedback inhibition of LLO initiation by accumulated Glc 3 Man 9 GlcNAc 2 -P-P-dolichol (17,20). Note that GlcNAc 1-2 -P-P-dolichol appeared to be barely detectable by FACE (no more than 1 pmol/10 7 cells), and was therefore at least 20-40 fold less abundant than Glc 3 Man 9 GlcNAc 2 -P-P-dolichol (20-40 pmol/10 7 cells).
To further test the possible role of Glc 3 Man 9 GlcNAc 2 -P-P-dolichol as a feedback inhibitor, Lec15.2 and Lec35.1 cells were examined since these accumulate Man 5 GlcNAc 2 -P-P-dolichol and produce no detectable Glc 3 Man 9 GlcNAc 2 -P-P-dolichol. Lec35 cells, in contrast with Lec15, do not glucosylate their Man 5 GlcNAc 2 -P-P-dolichol (9). FACE did not detect any net LLO accumulation in Lec15 or Lec35 cells treated with translation blockers. These mutant lines were both sensitive to the effects of translation inhibitors on LLO synthesis detected by metabolic labeling (data not shown) as reported earlier (20). Thus, there is no evidence from these experiments that accumulation of Glc 3 Man 9 GlcNAc 2 -P-P-dolichol, or any other glucosylated LLO, is necessary for inhibition of LLO synthesis.

Assay of GPT Activity In SLO-permeabilized cells reveals attenuation by translation
inhibitors--Consistent with earlier reports (17,18), we detected no effects for GPT ( Figure 3),

MPDS, GPDS (see below), and [ 3 H]-LLO synthesis (not shown) in microsomal membranes
prepared from CHO-K1 cells treated with translation inhibitors. In the course of these studies, it was noticed that the inhibition of LLO labeling with [ 3 H]mannose in adherent cells was partially relieved by detachment of the cells from culture dishes just prior to labeling. Thus, we considered the possibility that cellular perturbation might counteract the effect of translation inhibition on a key step in the dolichol pathway, in a manner highly reminiscent of the effects of perturbation on cells lacking the Lec35 gene product (9,29). Although Lec35p has an essential role in reactions requiring MPD or GPD in intact cells, the glycosylation defective phenotype in Lec35 mutants is lost by various forms of physical perturbation including preparation of microsomes. Gentle permeabilization of the plasma membrane of Lec35 cells with SLO does not affect the Lec35 phenotype, however, permitting the analysis of the Lec35 defect under in vitro conditions (9).
As shown in Figure 3  Even scraping the SLO-permabilized cells from dishes increased the remaining GPT activity resulting from translation arrest by approximately 50% (data not shown). Second, in cells that are not continuously synthesizing protein, GPT activity measured after permeabilization with SLO is highly reduced in a manner that is consistent with the loss of [ 3 H]-LLO synthesis in vivo. The residual GPT activity (10-20 %) may be due to minor perturbation that could not be controlled during permeabization with SLO.

LLO synthesis is diminished in cells treated with translation blockers and permeabilized
with SLO--If the diminished GPT activity in SLO-permeabilized cells represents a preservation of the effects of translation inhibitors on LLO synthesis in vivo, LLO synthesis in vitro with permeabilized cells should be similarly affected. Cells were treated with cycloheximide for various periods up to 1 hour, permeabilized, and assayed either for GPT activity with UDP-[ 3 H]GlcNAc, or total LLO synthesis with a mixture of UDP-GlcNAc and GDP-[ 3 H]mannose. In the latter case, the reactions included a chase with unlabeled GDP-mannose to extend partially mannosylated LLO intermediates. As shown in Figure 4 (panel A), GPT activity and LLO synthesis declined steadily during the 1 hour series of treatments with cycloheximide. As discussed above, assays performed with microsomes from similarly treated cells did not reveal significant losses of either activity.
Loss of LLO synthesis lagged behind the loss of GPT activity. One possible factor is that  Since the overexpression of GPT apparently resulted in functionally silent enzyme in intact cells, it was reasoned that this overexpression should not alter the effects of translation inhibitors on LLO synthesis. As shown in Figure 6, cycloheximide inhibited the synthesis of Translation arrest depletes dolichol-P used by MPDS and GPDS--Since the UDP-GlcNAc donor substrate was supplied exogenously for SLO-permeabilized cells, and the catalytic site of GPT is located at the cytoplasmic face of the ER membrane (5), it was unlikely that the decrease of GPT activity caused by translation inhibition was due to an effect on the supply of UDP-GlcNAc.

Exogenously expressed GPT has no apparent functional activity in SLO-treated cells and does not compensate for effects of translation inhibitors--
Examination of water-soluble [ 3 H]-labeled products by ion-exchange gave no indication of enhanced degradation of UDP-GlcNAc by translation-arrested cells (data not shown). Conversely, since both the dolichol-P and GPT in the SLO system are endogenous, effects on these components seemed more plausible.
If the inhibition of GPT activity after cycloheximide treatment was due to reduced availability of dolichol-P, we reasoned that other dolichol-P dependent reactions might also be inhibited. As shown in Table I Table I raised the question of whether GPT and MPDS shared the same limited pool of dolichol-P, or used separate limited pools. Evidence for shared pools in vivo was provided earlier by analyses of [ 3 H]mevalonate labeled dolichol conjugates in cells genetically or pharmacologically modified to increased or decrease GPT and MPDS activities (32). Enzymatic evidence for shared pools was reported in two earlier studies (33,34) but these used microsomal systems. To establish directly whether these two enzyme systems shared the same limited pool of dolichol-P, competition experiments were performed with cells permeabilized with SLO but not treated with translation blockers.
In preliminary experiments with nucleotide [ 3 H]-sugars of similar concentration and specific activity, the MPDS reaction in SLO-treated CHO-K1 cells transferred 5 to 10 times more sugar to endogenous acceptor than the GPT reaction. Thus, if these enzymes shared the same pool, it was expected that MPDS would consume a greater proportion of dolichol-P and compete more strongly than GPT. Pre-incubation of SLO-treated cells with 10 mM UDP-GlcNAc had no effect on MPDS activity, in the absence or presence of TN (data not shown). However, as shown in

Addition of oligosaccharyltransferase acceptor peptide to cyclohexmide-treated cells reverses inhibition of LLO synthesis in vivo and inhibition of GPT activity in vitro--If the inhibitory effects
of cycloheximide on LLO synthesis and GPT activity are due to trapping of the limited pool of dolichol-P as Glc 3 Man 9 GlcNAc 2 -P-P-dolichol, inhibition should be prevented by inclusion of an appropriate peptide acceptor for oligosaccharyltransferase. Such a peptide should discharge the Glc 3 Man 9 GlcNAc 2 -P-P-dolichol and generate a pool of dolichol-P-P to be recycled to dolichol-P (see Figure 1). As listed in Table II inclusion of the peptide Ac-Asn-Tyr-Thr-CONH 2 , an acceptor for oligosaccharyltransferase, lessened the inhibitory effects of cycloheximide while a non-acceptor control peptide, Ac-Gln-Tyr-Thr-CONH 2 , had no appreciable effect. The net effect of 500 µM acceptor peptide was restoration of approximately 9 % of control LLO synthesis, suggesting that while the effect of the peptide was specific, it discharged only a fraction of the Glc 3 Man 9 GlcNAc 2 -P-P-dolichol that is normally used during protein synthesis.
This level of improvement of LLO synthesis with acceptor peptide is consistent with the much stronger effect on GPT activity (approximately 250 % of untreated controls). GlcNAc 1-2 -P-P-  (Table I); that GPT and total LLO synthesis depend upon a limited pool of dolichol-P that can be trapped as Glc 3 Man 9 GlcNAc 2 -P-P-dolichol (Table II); and that at least two of the enzymes, GPT and MPDS, share the same limited pool of dolichol-P (Figure 7). To test these conclusions jointly, acceptor and non-acceptor peptides were added directly to SLO-permeabilized cells, both with and without prior treatments with cycloheximide, and each enzyme activity was measured. In all experiments ( Figure 8) the non-acceptor (control) peptide was virtually without effect (compare groups 1 and 2, 4 and 5).
However, the acceptor peptide markedly increased the activity of each enzyme in cycloheximidetreated cells (group 6). This indicates that all three enzymes share a limited pool of dolichol-P that can be trapped as Glc 3 Man 9 GlcNAc 2 -P-P-dolichol. Interestingly, acceptor peptide also increased the activities in cells not treated with cycloheximide (group 3). This suggests that the free dolichol-P in these cells was supplemented by dolichol-P resulting from discharge of Glc 3 Man 9 GlcNAc 2 -P-P-dolichol.
As shown with CHO-K1 cells not treated with cycloheximide in Figure 9, panel A, the enhancement due to acceptor peptide was substantial for the first 15 minutes, and then followed by a more modest rate of product accumulation. On the other hand, in the presence of non-acceptor peptide GPT product accumulated at a modest rate throughout the entire experiment. Calculation of the difference between the two graphs shows that net effect of the acceptor peptide was limited to the first 15 minutes of the experiment. As determined by FACE analysis, acceptor peptide discharged approximately 60% of the LLO. Discharge was nearly complete after 5 minutes, and preceded enhancement of GPT activity. This may reflect the time necessary for dephosphorylation of dolichol-P-P and return of dolichol-P to the cytoplasmic leaflet. these data indicate that ER glucosidases I and II acted upon the Glc 3 Man 9 GlcNAc 2 -P-P-dolichol during the incubation with transport buffer, in a manner previously reported (35).

Discussion
These results provide strong support for the 1980 proposal of Hubbard and Robbins (16) for a limited pool of dolichol-P (Figure 1), and show that GPT, MPDS, and GPDS are all controlled by a common mechanism. Further, the limited pool is a primary source of dolichol-P. The SLO system allows the regulation of these key reactions in the dolichol cycle to be studied fastidiously in vitro, apparently by preventing intermingling of the limited primary pool with a secondary pool of dolichol-P. (18) found that translation inhibitors did not interfere with the syntheses of GlcNAc-P-P-dolichol, GlcNAc 2 -P-P-dolichol, or Man 1 GlcNAc 2 -P-P-dolichol in intact cells using labeled sugar precursors. Pan and Elbein (20) noted some reduction of GlcNAc-P-P-dolichol synthesis by translation inhibitors, but only with treatment times and concentrations above those necessary to block both translation and LLO synthesis. In part, these prior assessments of GlcNAc-P-P-dolichol synthetic rates in vivo may have been complicated by factors such as the reversibility of the GPT reaction and the ability of MPD to stimulate GPT (10).

Discrepancies with prior studies--Grant and Lennarz
Two groups found that dolichol-P added to the culture medium stimulated [ 3 H]Glc 3 Man 9 GlcNAc 2 -P-P-dolichol synthesis in untreated cells, but not in cells treated with translation inhibitors (18,20). The interpretation was that the inhibition of LLO synthesis was not due to a lack of dolichol-P. In both experiments, cells were incubated with [2-3 H]mannose.
However, it was also found that exogenous dolichol-P did not stimulate [ 3 H]Glc 3 Man 9 GlcNAc 2 -P- (20). Further, while exogenous (20). Such results suggested the existence of two functional pools of dolichol-P: one used for LLO initiation by GPT that is not increased with exogenous dolichol-P, and a second pool used for LLO extension with MPD that is increased by exogenous dolichol-P. However, in the present study, GPT and MPDS were found to use the same limited pool of dolichol-P.

dolichol-P stimulated [ 3 H]MPD synthesis in untreated cells incubated with [ 3 H]mannose, [ 3 H]GlcNAc 2 -P-P-dolichol synthesis from [ 3 H]glucosamine was not stimulated
It is therefore important to realize that in the study reported here and in that of Hubbard and Robbins (16), LLO synthesis inhibition was at least 98%, and effects of inhibitors were observed within 15 min of addition and lasted for at least 1 hr. In the other two series of studies, inhibition of LLO synthesis was generally 70-85%. Further, in one case effects were reported within 5 minutes, but significant resumption of LLO synthesis occurred after 40 min (18). In the other case, effects were not apparent until 2 hours after addition of inhibitors, and were then stable for at least an additional 6 hours (17). The model in Figure 1

The SLO system is highly preferable for LLO in vitro studies--Since microsomal systems
are generally considered to be reliable for studying LLO biosynthetic enzymes, it was surprising to find that the effects of translation blockers on the activities of GPT, MPDS, and GPDS, as well as the limitation on activity of overexpressed GPT, were obscured in microsomes. By selective permeabilization of the plasma membrane with SLO, effects of translation blockers on LLO synthesis previously reported in vivo were observed in vitro. Interestingly, the possibility was raised before that cell disruption might affect the detection of a regulated state important for coupling of glycosylation with translation (20). In addition, the inactivity of overexpressed GPT in the SLO system compared with microsomes ( Figure 5) provided an explanation for its inability to counteract the effects of cycloheximide on LLO initiation in vivo (Figure 6), and for the absence of increased of LLO quantities in Tn-10 cells compared with other Man 5 GlcNAc 2 -P-P-dolichol accumulating mutants ( Figure 6). Cycloheximide-treated Tn-10 cells assayed for GPT activity in the SLO system showed somewhat higher residual activity (approximately 30%) compared with similarly treated CHO-K1 cells (10-20%). Though the reason is unclear, it is possible that a small fraction of the overexpressed GPT is functional, or that this multi-transmembrane span enzyme actually disrupts the ER membrane and emulates the situation with microsomes. Since the overexpressed GPT in Tn-10 cells is catalytically nonfunctional, it is unlikely that it causes accumulation of Man 5,9 GlcNAc 2 -P-P-dolichol by consuming the majority of the available dolichol-  (32). Thus, the actual basis for the LLO defect in Tn-10 and 3E11 cells remains unclear. These results also indicate that overexpression of GPT mediates resistance to TN by buffering, rather than a compensatory increase in catalytic activity, consistent with prior reports of catalytically inactive forms of GPT that were still able to confer resistance to TN (5,36).

A possible role for potential dolichol recognition sequences (PDRS)--How might a limited
primary pool of dolichol-P be formed? One scenario is a membrane domain or "raft" containing both dolichol-P and LLO transferases. Such domains might be easily disturbed upon disruption of cells, allowing the transferases access to dolichol-P not in rafts. A second possibility involves the previously reported potential dolichol recognition sequence, or PDRS. As reviewed (4), one or two copies of PDRSs are found in transmembrane segments of many transferases required for the early stages of LLO synthesis, and were originally suggested to facilitate binding of the dolicholconjugate acceptor substrates to the enzyme catalytic sites. Mutations affecting this sequence in S.
cerevisiae MPDS have widely variable effects on activity of MPDS in vitro (37,38), from essentially no effect to 99% inhibition depending upon the exact mutation and the assay procedure.
Both PDRSs of hamster GPT are required for activity in vivo and in vitro (microsomes) (36), but the exact functions of these PDRSs were not clear. The discussions in these various reports all suggested that PDRSs might be required for folding, stability, or sorting of MPDS or GPT as opposed to substrate binding.
Given the results reported here, the function of the PDRS might be to allow each dolichol chain to act as an anchor around which LLO transferases might cluster and assemble the oligosaccharide. The limited pool would consist of only those dolichol-P molecules in such complexes, which would dissociate with perturbation. Some transferases might bind directly to the dolichol chain, and others might bind secondarily to the PDRS-containing enzymes. In this scheme the PDRSs would not be required for binding of acceptor substrate at the catalytic site, but instead facilitate interactions between the enzyme and substrate, and could easily reside in a separate domain of the enzyme. This idea is attractive because it provides an explanation for the great variability in the activities of the various PDRS mutants with the different systems used. It also suggests a basis for nature's use of C 55-95 polyisoprenol carrier lipids (with many potential anchor sites) as opposed to lipids with shorter chains that are theoretically also capable of tight membrane association. In this hypothesis, overexpressed GPT would not be functional because the endogenous GPT would be sufficient to occupy all available docking sites.
Potential role of control by translation arrest--While these results prompt a re-evaluation of our understanding of the organization of the components of the LLO pathway, do the effects of translation arrest have physiological relevance? There is good evidence for a role (39) for translation arrest resulting from ER stress due to unfolded protein response (UPR) activation of the PKR-like ER kinase "PERK" (40). Stress-induced translation arrest might prevent new synthesis of LLOs, which would not be needed in the absence of nascent proteins. To achieve sufficient translation arrest to inhibit LLO synthesis, potent ER stress would be required (40). In contrast, the stimulatory effects of ER stress on LLO extension reported earlier (11) occur with concentrations of UPR inducers that cause little or no translation arrest (12), that do not prevent protein Nglycosylation, and that do not inhibit LLO synthesis (11). Taken together, while mild ER stress enhances LLO extension, it may be that strong ER stress inhibits LLO initiation.
Summary --The dolichol-P-P-oligosaccharide pathway is linked to protein synthesis by regulation of the step catalyzed by the initiating enzyme, GPT, in a manner that is obscured by microsome preparation. In this regard, the SLO system presents considerable advantages for studying regulation of LLO synthesis.           However, for MPDS and GPDS, activities were determined after organic extraction and enrichment of phosphomonoester products by DEAE-cellulose chromatography (2 mM NaOAc eluate) to eliminate neutral mannosyl and glucosyl lipids.