The Secretory Pathway

Inject a guinea pig with radioactive leucine, and in its pancreas the newly made digestive enzymes sit over the rough endoplasmic reticulum within about five minutes. By about twenty minutes the label lies over the Golgi, and by an hour it is in the granules the cell will later release.

That timetable comes from Lucien Caro and George Palade, who published it in 1964. It answers one question, the order of the stops. This lesson asks the rest. What happens to a protein at each stop, how fast does it move, how does the cell keep its own ER proteins from leaving with the cargo, and how does it decide whether to release a protein at once or store it for later?

Following a cohort: pulse-chase in the pancreas

The pancreatic acinar cell is a good place to start because most of the protein it makes is for export. Palade’s method was pulse-chase: give radioactive amino acid for a short time (the pulse), then flood the tissue with unlabelled amino acid (the chase), so that only the protein made during the pulse is labelled. That labelled group is a cohort, and you follow it as it moves. The method is set out in the lesson on distinguishing experiments.

Caro and Palade did this in living guinea pigs and read the result by electron-microscope autoradiography: photographic emulsion laid over thin sections records each radioactive decay as a silver grain over a compartment. Grains moved from the rough ER to Golgi elements to zymogen granules. They read their data as suggesting two steps in the Golgi region, small peripheral vesicles first and then large central condensing vacuoles, sacs in which secretory protein is concentrated before it becomes a granule.

James Jamieson and Palade then moved to slices of pancreas kept alive in a dish, which allowed a precise three-minute pulse and chases of 7, 17 and 57 minutes. They broke the cells open and separated microsomes, small vesicles that form from broken membranes. Label was in rough microsomes (from the ER) at the end of the pulse, peaked in smooth microsomes (from the Golgi periphery) after 7 minutes of chase, and arrived in the zymogen granule fraction later. That is a precursor-product relation: each compartment fills as the one before it empties.

The two methods disagreed at one point. The granule fraction was half-maximally labelled at 17 minutes, yet in intact cells the granules carried label no earlier than 57 minutes. The cause was contamination: labelled condensing vacuoles were sedimenting with the granules. A year later the same slices showed that the step between the ER exit region and the peripheral Golgi vesicles needs energy, and is blocked by inhibitors of respiration but not of glycolysis.

Keep the scope exact. These experiments, in one tissue of one species, showed the order of compartments and the time spent in each. Jamieson and Palade argued that the kinetics were direct evidence for passage through small Golgi vesicles, but no one watched cargo ride inside a vesicle. Order and timing are what pulse-chase gives you.

Processing along the way: glycans as a location stamp

A secretory protein enters the ER as it is made, guided by its signal sequence and SRP, as protein sorting explains for the ribosome and channel and translation explains for protein synthesis. Signal peptidase removes the signal in the lumen, the space inside the ER. Very soon after the chain enters, oligosaccharyltransferase attaches a preformed core of 14 sugars (Glc3Man9GlcNAc2) to asparagine in the sequence Asn-X-Ser/Thr. So N-glycosylation begins in the ER, not the Golgi. The sugar chemistry is in the carbohydrates lesson.

In the ER, glucoses are trimmed while the protein folds, under the controls described in quality control. Folded cargo leaves at ER exit sites in COPII-coated vesicles (the coats and SNAREs are in vesicle traffic). In mammalian cells it then passes the ERGIC, the ER-Golgi intermediate compartment marked by the protein ERGIC-53, before reaching the cis face of the Golgi, the side facing the ER. It crosses medial and trans cisternae (the flattened sacs of the stack) and leaves from the trans-Golgi network (TGN), a tubular sorting station on the far side.

The Golgi rebuilds the glycan step by step. Mannosidase I, GlcNAc transferase I, mannosidase II and GlcNAc transferase II act early; galactose and sialic acid are added later. In Chinese hamster ovary cell membranes the early enzymes partly separated from the late ones on density gradients, and the authors suggested that they work in some or all of the cis-side cisternae. Because each enzyme sits in a particular region, the glycan carries a stamp of how far the protein has come.

The tool that reads the stamp is endoglycosidase H (Endo H). It cuts high-mannose glycans but not the complex glycans made after GlcNAc transferase I and mannosidase II have acted. Cut protein runs faster on a gel. Endo H resistance is a threshold in a sequence, not a switch between two organelles: in normal rat hepatocytes about 10% of intracellular haptoglobin is resistant at any moment.

Some processing comes last. In human and rat pancreatic beta cells, proinsulin is cut to insulin mainly in clathrin-coated immature secretory granules, not in the Golgi stack.

Worked problem: reading an Endo H time course

The data here are illustrative. A glycoprotein in a cultured cell line is pulse-labelled for 5 minutes and chased. At each time you immunoprecipitate it, digest with Endo H, and measure the two bands. Assume nothing is secreted or degraded in 60 minutes, so each pair sums to 1,000.

Chase (min)	Resistant	Sensitive
0	0	1,000
15	330	670
30	550	450
60	800	200

Step 1, fractions resistant: 0, 33%, 55% and 80%.

Step 2, test for first order. If the fraction still sensitive, S, falls as S = e^(-kt), then k = -ln S / t at every time.

k = -ln(0.67) / 15 = 0.0267 per min; -ln(0.45) / 30 = 0.0266 per min; -ln(0.20) / 60 = 0.0268 per min

The three agree, so the data fit a single first-order step with k = 0.0267 per minute.

Chase time 15 min

0153060

Resistant : sensitive

Fraction resistant

Fraction still sensitive, S

k = −ln S / t

Half-time, t½ = ln 2 / k

Endo H cuts high-mannose glycans but not the complex glycans made after GlcNAc transferase I and mannosidase II have acted, so the resistant band times arrival at those Golgi enzymes. These data are illustrative, each pair sums to 1,000 if nothing is secreted or degraded, and the picture is a sketch of the logic.

Step 3, half-time.

t½ = ln 2 / k = 0.693 / 0.0267 = 26 min

Step 4, interpret. Resistance appears only after Golgi enzymes act, so 26 minutes covers ER exit plus the trip to those enzymes. It is an upper bound for ER exit alone.

Real cargoes differ. Four secretory proteins in isolated rat hepatocytes left the ER with half-times from 14 to 137 minutes, and in human HepG2 cells five secreted proteins matured at rates differing at least threefold. That spread is hard to explain if cargo simply flows out with the fluid, and Lodish and colleagues suggested a receptor that selects cargo. For comparison, VSV G protein fused to green fluorescent protein moved from ER to Golgi in living COS cells at about 2.8% per minute (a first-order rate constant), a half-time of 0.693 / 0.028 = 24.8 minutes.

Through the Golgi: vesicles or maturing cisternae

Cargo is not the only thing that flows forward. ER chaperones leak out too. In 1987 Sean Munro and Hugh Pelham found that many soluble ER residents end in KDEL (lysine, aspartate, glutamate, leucine). In COS cells, deleting KDEL from the chaperone grp78 made it secreted, and adding grp78’s last six residues to lysozyme kept lysozyme in the ER. Think of it as a return-to-sender label. A KDEL receptor in the ERGIC and Golgi binds the tag in the slightly acidic lumen there, rides back to the ER in COPI-coated vesicles, and releases its passenger at the near-neutral pH of the ER. (Yeast uses HDEL.)

How cargo crosses the stack has two pictures. In the stable-compartment model, each cisterna keeps its enzymes and small vesicles carry cargo from one to the next; James Rothman’s laboratory reconstituted transport between Golgi compartments in a cell-free system from CHO cells, needing ATP and cytosol. In the cisternal maturation model, the cisterna itself carries the cargo. It forms at the cis face, its resident enzymes are swapped by COPI vesicles moving backward, and it ends as trans Golgi and TGN.

Budding yeast made the question answerable for one organism, because its cisternae are scattered rather than stacked. In 2006 the laboratories of Benjamin Glick and Akihiko Nakano each filmed single cisternae carrying an early and a late fluorescent Golgi marker. A cisterna lost the early marker and gained the late one, and Glick’s group found the rate matched cargo transit. Later imaging showed cargo staying within a cisterna as it matured. In yeast, then, maturation is seen directly. For the stacked mammalian Golgi, Glick and Alberto Luini judged that no single model explains every observation, and proposed maturation as the best candidate for a shared core.

Two exits: constitutive and regulated secretion

At the TGN the routes split. In constitutive secretion, protein is released as fast as it is made. In regulated secretion, protein is stored at high concentration in granules until a stimulus arrives. Every cell has the first; endocrine, exocrine and nerve cells add the second.

In a cell-free system from PC12 neuroendocrine cells, a regulated protein and a constitutive one were sorted into different vesicles as they left the TGN. Granins, typical regulated cargo, stay aggregated at pH 6.4 with 1 to 10 mM calcium, conditions believed to hold in the TGN, so clumping may help pull regulated cargo apart from the rest. How the selection is made is still argued.

A stimulus acts at the end of this route. In guinea-pig pancreatic slices, a secretagogue did not speed synthesis or transit; it mainly increased discharge. Not every secreted protein follows this whole path either: some bypass the Golgi after entering the ER, and some proteins without a signal sequence leave without entering the ER at all.

Design problem

AtT-20 cells, a mouse pituitary tumour line, make the hormone ACTH from a larger precursor. They also carry an endogenous leukaemia virus and make its viral envelope glycoprotein. Design an experiment to show that one cell has two exit routes.

One way to do it

Pulse-label the infected cells, then chase, collecting the medium at intervals. Split the chase into two arms, with and without 8-bromo-cyclic AMP, a stimulus for these cells. Immunoprecipitate the viral glycoprotein, precursor ACTH and mature ACTH from medium and cells.

Expected result, as Gumbiner and Kelly reported in 1982: the viral glycoprotein and some precursor ACTH leave steadily, with little change from the stimulus, while mature ACTH is held in the cells and released four to five times more when stimulated. As a further test, Moore, Gumbiner and Kelly found that 200 micromolar chloroquine diverted new ACTH into the steady route.

Controls: equal synthesis in both arms (labelled protein in cells at the end of the pulse), viability under the stimulus, and a cell-associated count to show stored hormone falls when medium hormone rises. Limits: this is a tumour line, and the experiment does not identify the sorting signal.