(Received for publication, September 19, 1996, and in revised form, December 30, 1996)
From the LSU Neuroscience Center and Department of
Ophthalmology, Louisiana State University Medical Center, School of
Medicine, New Orleans, Louisiana 70112, the § University
of Michigan, Departments of Ophthalmology and Anatomy and Cell Biology,
Ann Arbor, Michigan 48105, and
Univeristy of Texas, Health
Science Center, Department of Pathology,
San Antonio, Texas 78284-7750
Post-Golgi vesicles budding from the trans-Golgi network (TGN) are involved in the vectorial transport and delivery of rhodopsin to photoreceptor rod outer segments (ROS). We report here that newly synthesized docosahexaenoyl (DHA) phospholipids are sequestered and cotransported by rhodopsin-bearing post-Golgi vesicles to ROS. Frog retinas were pulse-labeled with [35S]methionine/cysteine and [3H]DHA prior to ROS isolation and subcellular fractionation. After a 1-h pulse, relatively uniform [3H]DHA-lipid labeling (DPM/µg protein) was observed in all fractions enriched in post-Golgi vesicles, TGN, Golgi, and endoplasmic reticulum (ER) membranes. During the subsequent 2-h chase translocation of free [3H]DHA from ROS to the photoreceptor inner segment contributed to an additional overall increase in labeling of lipids. The specific activity (dpm/nmol DHA) in ER-enriched fraction was similar or higher than in other subcellular fractions after both the pulse and the chase, indicating that the bulk of [3H]DHA-lipids was synthesized in the ER. After the chase a 2-fold increase in labeling of lipids in the ER and Golgi and a 2.6-fold in lighter TGN-enriched fractions was observed. The highest labeling was in the post-Golgi vesicle fraction (4-fold increase), with [3H]DHA-phosphatidylcholine and [3H]DHA-phosphatidylethanolamine showing the greatest increase. At the same time, newly synthesized [35S]rhodopsin shifted from the ER and Golgi toward TGN and post-Golgi fractions. Therefore, sequestration and association of [35S]rhodopsin and [3H]DHA-lipids in a TGN membrane domain occurs prior to their exit and subsequent vectorial cotransport on post-Golgi vesicles to ROS. Labeling of ROS lipids was very low, with phosphatidylinositol and diacylglycerols displaying the highest labeling. This indicates that other mechanisms by-passing Golgi, i.e. facilitated by lipid carrier proteins, may also contribute to molecular replacement of disc membrane DHA-phospholipids, particularly phosphatidylinositol.
Vertebrate photoreceptors are polarized and compartmentalized cells, with a photosensitive outer segment and a synaptic terminal domain at opposite ends of the cell. These domains are attached through a connecting cilium and a short axon, respectively, to a central region, the inner segment. The inner segment is the site where lipid and protein synthesis, polarized sorting of molecules, and initiation of membrane biogenesis for both the outer segment and synaptic terminals take place. Disc membranes in rod outer segments (ROS)1 display a unique lipid-protein composition. The visual pigment rhodopsin, which accounts for more than 85% of disc membrane proteins (1), is embedded within a highly fluid lipid bilayer comprised of phospholipids (PLs) highly enriched in docosahexaenoic acid (DHA, 22:6n-3) (2-6).
Amphibian photoreceptor cells are a useful experimental model to study
protein and lipid trafficking in a polarized cell. Photoreceptors
actively synthesize proteins (mainly rhodopsin) and DHA-PLs to support
the dynamic daily renewal of 50-80 large disc membranes in each rod
cell that results in the addition of membrane at 3
µm2/min (7, 8). ROS lack the capacity for de
novo synthesis of PLs (4, 9). Therefore, they depend entirely on
an external supply of PLs from the inner segment where they are
synthesized mainly in the rough endoplasmic reticulum (ER) (10). How
these highly unsaturated lipids become components of ROS membranes and at which stage of membrane biosynthesis and disc morphogenesis they
become associated with rhodopsin is not yet clear.
Newly synthesized rhodopsin is vectorially transported from its site of
synthesis in the rough ER to ROS by vesicles that bud from the
trans-Golgi network (TGN), cluster beneath the connecting cilium, and
fuse with the inner segment plasma membrane within the periciliary
ridge complex (11, 12). A very low buoyant density ( = 1.09 g/ml)
post-Golgi vesicular subcellular fraction carrying newly synthesized
rhodopsin has been isolated and characterized from frog retinal
photoreceptor cells (12-15). Whereas rhodopsin and the bulk of DHA-PLs
are synthesized in the rough ER and both can follow a vesicle-mediated
traffic through the biosynthetic pathway, other mechanisms can
contribute to trafficking and selective delivery of PLs to
intracellular organelles (16-18). For example, the rapid monomer
transport of PLs is facilitated by transfer proteins (TP) through the
cytosol (19). A transfer protein that with high affinity transfers PC
to ROS membranes has also been described (20), and immunohistochemical
analysis of chicken retinas at hatching revealed the presence of
phosphatidylinositol (PI)-TP in retinal cells including the inner
segment of photoreceptors (21).
Phospholipid renewal of ROS membranes involves both membrane replacement (as new disc membranes are assembled at the base of the ROS) and molecular replacement (i.e. PL transfer protein mediated and remodeling of disc PLs by turnover) (22, 23). Using various radiolabeled lipid precursors under experimental conditions that inhibit protein synthesis or vesicle-mediated transport, lipids can be transported to ROS by independent pathways by-passing the Golgi (24-28). However sorting of DHA-PLs, vectorial transport to ROS, and the contribution of alternative pathways to their trafficking has not been experimentally addressed. This question is especially intriguing because biochemical and autoradiographic studies of [3H]DHA trafficking in frog photoreceptors after in vitro (29, 30) and in vivo (31, 32) labeling have disclosed that newly synthesized [3H]DHA-PLs display a polarized delivery to ROS, where they are incorporated at the base as new discs are formed, in a pattern paralleling incorporation of radiolabeled amino acids into disk membrane proteins.
The aim of the current study was to investigate if newly synthesized DHA-lipids could be, at least in part, segregated and cotransported with rhodopsin in vesicles budding from the TGN and then enter the ROS as new membranes are formed (membrane replacement). To address this question, we pulse-labeled frog retinas for 1 h in the presence of [3H]DHA and [35S]methionine/cysteine, followed by a 2-h chase in a cold buffer prior to subcellular fractionation (12, 33). [3H]DHA was chosen because this precursor is actively esterified into PLs in the inner segment of photoreceptor (34-38) prior to their vectorial transport to the ellipsoid region at the base of ROS and also to synaptic terminals (29, 30). Our results reveal that newly synthesized [3H]DHA-PLs, especially the main components of disc membranes (i.e. DHA-phosphatidylcholine (PC) and DHA-phosphatidylethanolamine (PE)), are segregated and loaded together with newly synthesized [35S]rhodopsin in post-Golgi vesicles. Some lipids, i.e. [3H]DHA-PI and [3H]DHA-diacylglycerols (DAG), are very rapidly synthesized and delivered to ROS probably by alternative pathways that by-pass the Golgi and may be facilitated by carrier proteins.
Frogs, Rana berlandieri (100-250 g), were purchased from Rana Co. (Brownsville, TX), maintained in a 12-h light/12-h dark cycle, and fed crickets for a week prior to the experiment. [4, 5-3H]DHA (specific activity 17 Ci/mmol) and [35S]Express protein labeling mixture (1,000 Ci/mmol) were from DuPont NEN. High performance thin layer chromatographic plates (10 × 10 cm, 150 µm thickness) were from Analtech (Newark, DE). Lipids and fatty acid methyl ester standards and protease inhibitors were from Sigma. High performance liquid chromatography grade solvents were from EM Science (Gibbstown, NJ). All other reagents used were of the highest purity available.
In Vitro Labeling of Retinal Lipids and Proteins with [3H]DHA and [35S]Methionine/CysteineRetinas recovered from frog eyecups 2 h before the time of light offset were dissected and subsequently incubated under dim red light. Two sets of 21 retinas were incubated in 30 ml of an oxygenated medium as described (14, 35) at 22 °C in the presence of [3H]DHA (5.7 µCi/retina, final DHA concentration 0.24 µM), and [35S]Express protein labeling mixture (25 µCi/retina) for 1 h. One set of retinas was further incubated for 2 h in cold buffer containing unlabeled amino acids (pulse-chase samples) prior to subcellular fractionation. Since the in vitro retinal metabolism of [3H]DHA is altered at µM DHA concentrations (29, 30, 39), cold DHA was not added during the chase. Therefore, this chase period will hereafter be identified as "chase" to reflect this condition.
Rod Outer Segments Isolation and Retinal Subcellular FractionationAll the procedures followed for the isolation of ROS and subcellular fractionation have been described in detail elsewhere (12, 33). Briefly, following the pulse and pulse-chase labeling, retinas were sheared through a 14-gauge needle, and ROS were separated by flotation on 34% sucrose. Retinal pellets were rehomogenized in 0.25 M sucrose in 10 mM Tris acetate, pH 7.4, containing 1 mM MgCl2 and centrifuged for 4 min at 4,000 rpm (1250 gav, JA20 rotor, Beckman Instruments, Inc., Palo Alto, CA). The postnuclear supernatant (3 ml) recovered after this centrifugation is enriched in photoreceptor biosynthetic membranes and organelles involved in rhodopsin transport (12, 33). To isolate post-Golgi vesicles from TGN, Golgi, and ER membranes, the postnuclear supernatant was overlaid on a 10-ml linear 20-39% (w/w) sucrose gradient in 10 mM Tris acetate buffer, pH 7.4, containing protease inhibitors and 1 mM MgCl2, above a 0.5-ml cushion of 49% (w/w) sucrose in the same buffer. Gradients were centrifuged for 13 h at 4 °C in a SW40 rotor (Beckman) at 28,000 rpm (100,000 gav). Fourteen fractions (0.9 ml each) were reproducibly collected from the top of the gradient. A Buchler Auto Densi-Flow fractionator was used to prepare the gradient and to collect the fractions. The subcellular fractions were diluted 4-fold with 10 mM Tris-HCl and pelleted at 50,000 rpm (240,000 gav) for 40 min in a SW50 rotor. The pellets were resuspended in 210 µl of Tris acetate, pH 7.4, and divided into two aliquots, one-third for protein analysis and two-thirds for lipid analysis.
Protein AnalysisSDS-polyacrylamide gel electrophoresis was performed as described previously (12). 35S-Labeled rhodopsin was determined in subcellular fractions by exposure of dried SDS gels to storage phosphor screens, and the intensity of luminescence associated with the rhodopsin band was measured and analyzed by a PhosphorImager densitometer (Molecular Dynamics). Total proteins were quantified according to Fanger (40), using bovine serum albumin as a standard.
Lipid Extraction and AnalysisLipids were extracted from the fractions by adding 3 ml of chloroform:methanol (2:1, v/v) following the Folch procedure (41). Individual phospholipids and neutral lipids were isolated in the same TLC plate following a two-dimensional, three-step TLC procedure (6) as follows: an aliquot of the labeled lipid extract containing phospholipid and neutral lipid standards as a carriers was applied on the lower right corner (1.5 cm from each border) of 10 × 10 cm high performance thin layer chromatography plates previously sprayed with 3% magnesium acetate and activated for at least 1 h at 100 °C. The plate was developed in the first dimension twice using the Rouser I chromatographic system (chloroform/methanol/ammonia, 65:25:5, v/v) until the solvent front reached 2 cm from the top of the plate. After drying with cold air, the plate was turned to the right 90°, and neutral lipid standards (cholesterol ester, triacylglycerol, diacylglycerol, and monoacylglycerol) were spotted 1.5 cm from the bottom and 0.5 cm from the right border. Plates were then developed in hexane/ether (60:40, v/v) to isolate individual neutral lipids that had accumulated at the front of the first chromatographic system. The silica gel was cut with a vertical line to isolate neutral lipid (right) from phospholipid (left) areas of the plate (approximately 3 cm from the right border), and the silica was scraped off from the bottom right corner prior to running the plates in the Rouser II system (chloroform/acetone/ methanol/acetic acid, 30:40:10:10:5, v/v). This third chromatographic step, run in the same direction as the second step, allows the isolation of individual phospholipid classes and free fatty acids that run with the solvent front above PE. Lipid spots were visualized by iodine staining, and the radioactivity was determined in a Beckman scintillation counter.
Aliquots of lipid extracts were taken for gas-liquid chromatography analysis of endogenous fatty acid content and composition. Fatty acid methyl esters were prepared in glass tubes by transmethylation with 2 ml of toluene/methanol/sulfuric acid (100:100:4, v/v) for 4 h at 65 °C, after flushing the tubes with nitrogen and capping with a Teflon-lined cap. The tubes were cooled at room temperature, and 1 ml of water, 3 ml of hexane, and a mixture of two internal standards (17:0 and 21:0 methyl esters) were added. Fatty acid methyl esters resuspended in hexane were separated onto a SP-2330 column (30 m, 0.25 mm inner diameter, 0.2-µm film thickness, Suppelco, Bellefonte, PA) by using helium as a carrier gas, in a Varian Vista 401 gas chromatograph (Palo Alto, CA). The injector and detector temperatures were 220 and 250 °C, respectively, and the column temperature was programmed from 70 to 230 °C (42). The peaks were detected by flame ionization, identified by comparison with the retention times of authentic fatty acid methyl esters standard, and quantified using the internal standards.
Statistical AnalysisValues for [3H]DHA lipid labeling and [35S]rhodopsin are presented as a mean ± S.E. for n = four individual experiments. Data were compared using Student's t test for pair samples. A p value of < 0.05 was considered statistically significant.
To
determine whether newly synthesized [3H]DHA-PLs and
[35S]rhodopsin are transported together in the same
population of post-Golgi vesicles recovered in fraction 5 of the
sucrose gradient, retinas were pulse-labeled for 1 h in the
presence of both precursors and further incubated for 2 h (chase)
in cold buffer prior to subcellular fractionation. This experimental
protocol gives sufficient labeling of newly synthesized rhodopsin
within 1-h pulse and a maximum labeling of the vesicles (fraction 5)
during the following 2-h chase (12, 43). While the total
[35S]rhodopsin labeling recovered from the combined 14 fractions was similar for pulse and chase samples (data not shown),
total esterified [3H]DHA was increased by 2.6 ± 0.2-fold: from 1.2 × 106 dpm/21 retinas after a 1-h
pulse to 3.2 × 106 dpm/21 retinas after a 2-h chase.
After 1 h of pulse labeling, 58% of total
[35S]rhodopsin was recovered in TGN and Golgi (fractions
7-11) (Fig. 1) as expected (12). After a 2-h
"chase," a shift toward post-Golgi fractions 4-6 was observed,
with fraction 5 displaying the greatest increase (ratio chase/pulse:
2.4 ± 0.5). The profile of total esterified [3H]DHA
among subcellular fractions was similar to that of
[35S]rhodopsin (Fig. 1), with the highest percent values
observed in those fractions that also accumulated the largest
proportion of membranes (Fig. 2D). Remarkably
after the 2-h chase only post-Golgi fraction 5 displayed significantly
higher [3H]DHA percent labeling (ratio chase/pulse,
1.7 ± 0.2) at the time when newly synthesized
[35S]rhodopsin accumulated in this fraction.
Simultaneously, percent labeling in fractions 12 and 13 was lower than
during the pulse.
Labeling recovered from individual subcellular fractions based upon protein content is shown in Fig. 2 and reveals four features. First, after a 1-h pulse, all fractions displayed similar labeling of esterified [3H]DHA (Fig. 2B). Second, after the subsequent 2-h "chase," labeling of DHA-lipids increased at least 2-fold in the heavy fractions that correspond to the density of ER (13-14), Golgi (11-12), and TGN (10), by 2.6-fold in lighter TGN fractions 7-9, and peaking at 4-fold higher labeling in post-Golgi vesicles recovered in fraction 5 as compared with pulse-labeled values (Fig. 2, B-C). Third, no significant difference between pulse and chase labeling was observed in the free [3H]DHA pool, indicating an equilibrium between the arrival of the precursor to these membrane compartments and its esterification into lipids either by de novo synthesis and/or turnover (Fig. 2A). The peak of labeling observed in fraction 8 may be the result of free [3H]DHA contributed by a small proportion of heavily labeled ROS cosedimenting between fractions 7 and 9 at a buoyant density of 1.13 g/ml (12). Fourth, most of the label recovered from ROS after pulse and chase labeling was found as free DHA (92 ± 1 and 76 ± 1%, respectively) (Fig. 2A), whereas ROS lipids labeling was the lowest among all subcellular fractions analyzed (Fig. 2B). A 2.5-fold increase in ROS [3H]DHA-lipids labeling after the "chase" (from 138 ± 25 to 350 ± 40 dpm/µg protein) accounted for by the concomitant loss of only 15% of free [3H]DHA (from 2600 to 1160 dpm/µg protein).
The specific activity of total [3H]DHA-lipids (dpm/nmol of endogenous DHA content) showed that fraction 14 (ER) displayed, after both the pulse (2790 ± 245 dpm/nmol DHA) and the "chase" (5060 ± 840 dpm/nmol DHA) labeling, similar or higher values than other fractions. After the pulse labeling, the specific activity of fraction 5 (1990 ± 115 dpm/nmol DHA) was significantly lower than that of fraction 14 (p < 0.05). After the chase, the specific activity of fraction 5 increased 3-fold (6580 ± 550 dpm/nmol DHA) but was not significantly different from the specific activity of fraction 14 (p > 0.19).
Phosphatidylcholine, Phosphatidylethanolamine, and Phosphatidylinositol Are Highly Labeled in All Subcellular Fractions; [3H]DHA-PC and [3H]DHA-PE Preferentially Shift toward Post-Golgi Vesicles (Fraction 5) after the ChaseThe
two most abundant PLs in retinal membranes, PC and PE, reveal a similar
labeling profile after a 2-h "chase" (Fig. 3). [3H]DHA-PC and [3H]DHA-PE gradually
increased from fractions enriched in ER to heavy fractions of the TGN,
with a sharp peak in post-Golgi fraction 5 reaching a 5.2- and 4.6-fold
increase, respectively, above pulse labeling. The profile of
[3H]DHA-PI was very similar to that of
[3H]diacylglycerol (DAG), and their labeling was
significantly increased only in TGN fractions 7-9, with the highest
increase in post-Golgi fraction 5 (3.9- and 2.5-fold, respectively). PS
was the only phospholipid that did not show a peak of labeling in
fraction 5 but displayed 3-fold increase between fractions 5 and 7. Phosphatidic acid (PA) labeling gradually increased from ER fractions
to post-Golgi fraction 5 with no significant differences from the 1-h
pulse labeling.
[3H]DHA-PL labeling in ROS was very low (Fig. 3,
insets) with a different pattern of distribution than all
other subcellular fractions (Fig. 4). Although the small
amount of ROS that is recovered in the gradient between fraction 7 and
9 may contribute to their highly free [3H]DHA labeling
(Fig. 2), it cannot contribute to but rather results in an
underestimation of lipid labeling in these fractions that arise from
inner segment membranes. [3H]DHA-PI displayed, by far,
the highest labeling in ROS, with a 7.9-fold increase after a 2-h
"chase," followed by DAG (4.8-fold), PC, and PS (3-fold each). No
differences were observed between a 1-h pulse and 2-h "chase"
labeling in [3H]DHA-PE and [3H]DHA-PA (Fig.
3, insets). This short-term incubation may reflect the
labeling of disc membrane lipids by molecular replacement including
(a) [3H]DHA incorporation by turnover,
(b) the fast transport from the inner segment of a portion
of newly synthesized [3H]DHA-lipids (i.e.
[3H]DHA-PI), and/or (c) further metabolism of
newly incorporated [3H]DHA-PLs into ROS such as
N-methylation of [3H]DHA-PE to
[3H]DHA-PC (4, 44).
The percent distribution of esterified [3H]DHA displayed
high values for PI, PC, and PE in all fractions except for ROS (Fig. 4). In fraction 5 they reached a similar value (28%), although in
other fractions, PI labeling alone prevailed. The highest percent labeling of PI was observed in fraction 1 (48%) and to a lesser extent
in fraction 2 (34%), probably associated with cytosolic proteins
recovered at the top of the gradient that may sediment after the 40,000 centrifugation (12). The ratio PI to PC labeling was higher in TGN
fractions as compared with post-Golgi fractions 4-6, showing the
highest value in fraction 8 of the TGN after both pulse and chase
labeling (Fig. 5).
The Content of Endogenous Fatty Acyl Chains of Lipids from Post-Golgi Vesicular Fractions
Total fatty acyl group content,
reflecting mainly membrane phospholipids, increased gradually from the
heaviest, ER-enriched fractions (2 nmol/µg protein) to the post-Golgi
light vesicular fractions 4-5 (4 nmol/µg protein) (Fig.
6). The endogenous DHA content was very similar for all
fractions (approximately 20% of total acyl groups) except for TGN
fractions 7-9 where ROS, not completely removed prior to subcellular
fractionation, cosedimented. In ROS, DHA accounted for 50% of total
acyl groups. The lower % DHA content in fraction 5 as compared with
ROS suggests that either the lipids from the vesicles bearing rhodopsin
are less enriched in DHA-lipid and/or that lipids contributed by other vesicles with a lower degree of unsaturation are recovered in this
fraction. The latter possibility is unlikely since immunoisolation of
rhodopsin-bearing post-Golgi vesicles with anti-rhodopsin antibody indicated that they constitute >85% of the vesicles sedimenting in
fraction 5 of the gradient (12). The net amount of DHA per protein in
ROS (1.93 nmol/µg) was twice that of lipids from fraction 5 vesicles
(0.85 nmol/µg protein). Although rhodopsin is the most abundant
protein recovered in fraction 5, its contribution to the total proteins
in the fraction is less than 50%, whereas rhodopsin accounts for
85-90% of total ROS proteins (12). Therefore, the net amount of
DHA-PLs with respect to rhodopsin protein in these post-Golgi vesicles
may reach values similar to that of ROS.
This study provides the first available information about the closely coordinated trafficking, sorting, and association of newly synthesized [3H]DHA-PLs with [35S]rhodopsin in frog photoreceptors as the two major membrane components initiate their journey from the rough ER, where they are synthesized, move through the Golgi, and leave the TGN on post-Golgi vesicles vectorially driven to ROS for the assembly of new disc membranes. Our results also yield several other important findings: (a) free [3H]DHA may be incorporated by turnover in disc PLs although much less efficiently than when utilized in the inner segment for [3H]DHA-PLs synthesis; (b) some [3H]DHA-PLs, mainly [3H]DHA-PI, are actively synthesized in the inner segment, rapidly transported to ROS, and incorporated into disc membranes; (c) during the short period of in vitro labeling (1-3 h), the fast labeling of all disc membranes by molecular replacement (i.e. protein-mediated transport of [3H]DHA-PLs and/or [3H]DHA incorporation by turnover) as compared with the labeling of a few discs at the base by membrane replacement, makes it difficult to assess the contribution of the latter mechanism to the overall labeling of ROS lipids.
The time course of [35S]rhodopsin labeling through the different compartments of the secretory pathway reflects its vesicle-mediated vectorial traffic from the site of synthesis at the rough ER to only one destination, ROS. The well defined early accumulation of [35S]rhodopsin in Golgi fractions by 1 h of labeling followed by its displacement during the subsequent 2-h chase toward TGN and post-Golgi vesicles recovered in fraction 5 (14 and Fig. 1) was used as a marker of membrane flow to follow the fate of newly synthesized [3H]DHA-lipids. Labeled lipids recovered in Golgi and TGN-enriched fractions at any time represent newly synthesized lipids that become constitutive components of the membranes as well as the different [3H]DHA-lipid pools that are in transit through these compartments to different destinations. [3H]DHA-lipids labeling of post-Golgi vesicles (fraction 5), however, is a clear indication of those newly synthesized [3H]DHA-lipids cotransported with [35S]rhodopsin to ROS.
Within the first hour of pulse labeling, [3H]DHA incorporation into lipids reached similar values in all subcellular fractions. This rapid equilibrium [3H]DHA-lipids among all fractions was also reflected in a similar mol % content of endogenous DHA (20%). Only fractions 7-9, contaminated with ROS membranes, displayed higher mol % values. Thus, newly synthesized [3H]DHA-lipids were rapidly transported throughout the multiple compartments of the biosynthetic pathway either by vesicle budding and fusion, by carrier proteins, and/or by lateral diffusion through intermembrane bridges (17, 18). Incorporation of [3H]DHA by turnover in lipids trafficking along the transport pathway could also contribute to the uniform labeling distribution observed among subcellular fractions.
After the 2-h "chase" labeling in cold buffer, it became apparent that the high labeling of free [3H]DHA in ROS was not paralleled by an efficient esterification into disc membrane phospholipids but rather by a translocation to the inner segment where it was actively esterified. This could be accomplished by the presence of (a) DHA-fatty acid binding proteins in ROS (45) and in the cytosolic fraction of retinas (46, 47) and (b) DHA-CoA synthetase in microsomes. This enzyme that activates DHA prior to its esterification into lipids displays the highest activity in microsomes from frog retinas and very low activity in ROS (48). Although free DHA can be incorporated in disc membrane PLs by turnover of their acyl groups (49-51), our present results indicate that the bulk of DHA is incorporated into lipids in the inner segment prior to their delivery to ROS (52). Indeed, the similar or higher specific activity of total [3H]DHA-lipids observed in the ER-enriched fractions as compared with other fractions enriched in membranes of the Golgi and post-Golgi supports this notion.
An interesting observation after the 2-h cold chase incubation was that the increase in [3H]DHA-lipids labeling was not of the same magnitude for all subcellular fractions (Figs. 2 and 3). It showed a clear trend from a 2-fold increase in ER and Golgi, to 2.5-fold in TGN, and the highest 4-fold increase in post-Golgi vesicles (gradient fraction 5). Because these fractions also became heavily labeled with newly synthesized [35S]rhodopsin after the chase, it appears that some newly synthesized [3H]DHA-PLs are sorted, along with newly synthesized [35S]rhodopsin in transit toward the TGN exit. The highest [3H]DHA-PLs labeling observed in [35S]rhodopsin-bearing post-Golgi vesicles suggests that they budded from microdomains in the TGN enriched in both [3H]DHA-PLs and [35S]rhodopsin. In fact, rhodopsin shows a preference for association with more fluid lipids (53), and in ROS PLs with high DHA content are in closer association with rhodopsin than less unsaturated ones (54). That the highest [3H]DHA-PLs labeling in post-Golgi vesicles observed after the 2-h "chase" could be the result of differences in lipid turnover in this fraction is unlikely since (a) labeling gradually increased from Golgi to post-Golgi vesicles, (b) no differences in the specific activity (total [3H]DHA-PLs/endogenous DHA content) between post-Golgi vesicles and ER fraction was observed, and (c) after 1-h pulse labeling post-Golgi vesicles did not show higher labeling than other fractions. Taken together these data strongly argue in favor of a progression of label through a series of compartments. Moreover, the similar [3H]DHA-lipid labeling observed after 1 h among all subcellular fractions also suggests an early association between newly synthesized DHA-lipids and newly synthesized rhodopsin rather than with older rhodopsin molecules already moving ahead in transit through the Golgi.
The profile of individual [3H]DHA-lipids labeling in subcellular fractions after the 2-h "chase" was very similar for the two main membrane components PE and PC, which also displayed the highest increase of labeling in post-Golgi vesicles (Fig. 3). This observation and the very low labeling of [3H]DHA-PC and [3H]DHA-PE recovered in ROS (Figs. 3 and 4) suggests that their incorporation into disc membranes mainly occurs by membrane replacement. Our preliminary studies using brefeldin A, which perturbs rhodopsin trafficking, show that [3H]DHA-PL and [35S]rhodopsin transfer into fraction 5, PE and PC, in particular, were successfully blocked, and also that ROS lipid labeling was reduced.2 At difference with PE and PC, [3H]DHA-PI displayed a more sustained increase throughout TGN peaking in post-Golgi fracton 5. Although the labeling of [3H]DHA-DAG was much lower than that of [3H]DHA-PI (Fig. 4), both followed a very similar profile (Fig. 3), probably reflecting an active phosphodiesteratic catabolism of PI with the consequent generation of labeled DAG along the TGN compartment. In ROS, [3H]DHA-DAG and [3H]DHA-PI displayed the highest labeling (Fig. 4), suggestive of their active translocation and incorporation in disc membranes by molecular replacement. Several lines of experimental evidence appear to indicate that PI, synthesized de novo in the inner segment of photoreceptors, can actively be transferred to ROS by-passing the Golgi (24, 28). In ROS, PI can be further phosphorylated to phosphatidylinositol 4,5-bisphosphate (9, 55). Because ROS contains a light-stimulated phosphoinositide-specific phospholipase C (56, 57), the presence of a photoreceptor cytosolic PI-TP, possibly similar to the one found in rat brain cytosol (58), could contribute to sustain and modulate the inositol lipid-derived signals triggered by light.
Frog and primate retinas labeled in vitro and/or in vivo with [3H]DHA display an early high level of labeling of PI, reaching values similar to that of PC and PE (29, 30, 59). In the present study, we confirm and further extend our previous observation to show that [3H]DHA-PI preferentially accumulates in the lightest fractions 1-3 of the gradient where cytosolic proteins are recovered (12), and also in ROS and in TGN fractions. The high ratio [3H]DHA-PI to [3H]DHA-PC found in TGN fractions 7-9 of frog retinas is the first evidence indicating a relative enrichment with newly synthesized [3H]DHA-PI of membranes located at the exit from the TGN compartment. As previously shown in yeast (19) and on PC12 cells (60), the high PI/PC ratio may also be essential for budding of rhodopsin-bearing vesicles from TGN and further suggests the involvement of PI/PC-TP in the dynamics of Golgi function in photoreceptors. Moreover, PI-TP has been identified as a cytosolic factor that stimulates the formation of secretory vesicles in PC12 cells (60). Since membranes recovered in TGN-enriched fractions 6-11 also contain synaptophysin, a synaptic membrane protein (12), further studies are necessary to evaluate the possible contribution of [3H]DHA-PI, in transit together with synaptophysin toward synaptic terminals, to the high [3H]DHA-PI in TGN fractions 7-9. The overall contribution of synaptic protein biosynthesis in this fraction must be relatively minor, however, since rhodopsin synthesis greatly exceeds the rate of synthesis of all other retinal membrane proteins (61).
Post-Golgi vesicles recovered from the gradient fraction 5 ( = 1.09 g/ml) display lower density than ROS which sediments in fractions 7-8
(
= 1.12-1.13) and therefore must have a higher lipid to
protein ratio (12). This is also supported by freeze-fracture EM
studies (62) showing that vesicles clustered around the connecting cilium display half the density of the intramembranous particle of ROS
disks. Since in post-Golgi vesicles the total acyl group content,
derived mainly from PLs (4.2 nmol/µg protein), was similar to that of
the ROS (3.9 nmol/µg protein), other lipids such as sterols
presumably contribute to their lower density. Cholesterol delivery to
ROS may be accomplished either by a pathway(s) independent from that
followed by integral plasma membrane proteins (18, 63) and/or together
with rhodopsin and DHA-lipid-containing post-Golgi vesicles. As
vesicles fuse with the plasma membrane adjacent to the base of the
connecting cilium, they could generate the cholesterol-enriched domains
observed in frog photoreceptors surrounding the periciliary ridge
complex (64, 65) and in nascent discs at the base of the ROS (66).
In summary, this study shows that newly synthesized [3H]DHA-PLs, mainly [3H]DHA-PC and [3H]DHA-PE, are vectorially cotransported to ROS by rhodopsin-bearing post-Golgi vesicles, and other PLs such as [3H]DHA-PI may also reach ROS and the TGN by-passing the Golgi carried by transfer proteins. Moreover, in the complex process of membrane biogenesis, addition of rhodopsin and DHA-PLs at the base of ROS could be "the driving force" for the incorporation of other PLs that do not contain DHA (approximately 40-50% of total PLs in disc membranes) possibly reaching the periciliary region by independent pathways. Current studies aim to delineate the mechanism(s) that contribute to the complex polarized trafficking of DHA-PLs either by vesicular and/or by transfer protein-mediated transport to ROS.
We thank R. E. Anderson for thoughtful discussions of the results and Nilda Parkins, Belen Puleo-Scheppke, and Claudia Trippe for their excellent technical assistance.