Cell Biology · Unit 2: Membranes and compartments

Protein Sorting into Compartments

Almost every protein in a eukaryotic cell is made in the cytosol, yet thousands of them end up inside the ER, mitochondria, chloroplasts or peroxisomes. This lesson shows how short stretches of the protein's own sequence act as addresses, how the ER reads its address while the chain is still being made, and how the other organelles read theirs after the chain is finished. Afterwards you can predict where a protein goes from its signals, and what happens when a signal is moved, deleted or added to a passenger protein.

Watch first The Address Travels With The Protein · 2:32

Made in a test tube with no membranes, a secretory protein comes out longer than the one the cell secretes.

In 1971 Günter Blobel and David Sabatini proposed that proteins bound for the ER carry a signal that sends the ribosome making them to the membrane. In 1972 César Milstein’s group translated the messenger RNA for an antibody light chain in a cell-free system and got a product slightly larger than the light chain that myeloma cells secrete, which they proposed was a precursor. In 1975 Blobel and Bernhard Dobberstein tested the idea directly: the extra piece is an address, read while the chain is still on the ribosome, and cut off once the protein has arrived. This lesson is about such addresses: what they look like, who reads them, and what happens when you move one.

The idea: a zip code in the sequence

Almost every protein in a eukaryotic cell starts on a ribosome in the cytosol, the fluid that fills the cell outside the organelles. (A few are made inside mitochondria and chloroplasts from their own genes.) Yet a typical cell must deliver thousands of different proteins to the ER, mitochondria, chloroplasts in plants, peroxisomes and the nucleus. The inside of a membrane-bound organelle is its lumen (in mitochondria and chloroplasts the innermost space is called the matrix and the stroma).

The cell solves this with a signal sequence: a continuous stretch of amino acids that carries sorting information. Think of a zip code printed on the parcel. A protein with no signal stays in the cytosol by default, and a signal attached to a protein that normally stays in the cytosol can send it elsewhere. That second fact is the most useful test in this whole field: if moving a stretch of sequence moves the protein, the stretch is sufficient to sort it.

The signals differ by destination, and so does the timing:

  • ER: a hydrophobic stretch, usually near the amino terminus, read while the chain is being made (cotranslational).
  • Mitochondrial matrix: an amino-terminal presequence that tends to form an amphipathic helix, with positive charges on one face and hydrophobic residues on the other (see protein structure). Read after translation.
  • Chloroplast stroma: an amino-terminal transit peptide, read after translation.
  • Peroxisomal matrix: usually a carboxyl-terminal tripeptide such as serine-lysine-leucine (SKL), called PTS1, or less often an amino-terminal PTS2. Read after translation.

How a membrane protein ends up facing the right way, its topology, is set at insertion; membrane organization and topology covers that. Here the question is how a protein reaches the right membrane at all. Nuclear import, which goes through pores instead of across a membrane, has its own lesson.

Into the ER while being made

Ribosomes and translation belong to the genetic code and translation; you need only that the amino terminus leaves the ribosome first. The ER route runs as a cycle, and each step was pinned down in a cell-free experiment.

  1. Recognition. The signal recognition particle (SRP), a complex of six proteins and an RNA, binds the hydrophobic signal as it emerges from the ribosome. In 1981 Peter Walter and Blobel showed, translating preprolactin in a wheat germ cell-free system, that SRP made elongation stop at a chain of about 70 residues: the 30-residue signal plus roughly 40 residues hidden inside the ribosome. Those numbers belong to preprolactin in that system; in reticulocyte lysate SRP was later found to cause only a transient pause (Wolin & Walter, J Cell Biol 109:2617, 1989).
  2. Docking. The SRP-ribosome complex binds the SRP receptor on the ER membrane. Walter and Blobel found that washing dog-pancreas ER vesicles in salt removed their ability to bind ribosomes making prolactin, and purified SRP restored it. In 1982 Reid Gilmore, Walter and Blobel used SRP as bait to isolate a 72,000-dalton receptor protein.
  3. Handover. SRP and its receptor both carry GTP-binding subunits. With GTP bound they hold together and pass the ribosome to the Sec61 channel, the pore through which the chain enters the ER. GTP hydrolysis then separates SRP from its receptor, and elongation resumes with the chain threading through. A 2016 structure of an active Sec61 showed the signal sitting in the channel’s lateral gate after displacing one of its helices.
  4. Cleavage. Signal peptidase on the lumenal side cuts off the signal. It is a different enzyme from the SRP receptor: the 1982 affinity column separated the two activities.

The GTP step works as a timer. A complex that commits only with GTP bound, and comes apart only when GTP is hydrolysed, can wait until the ribosome is safely on Sec61 before letting go.

Worked problem: reading a protection gel

Here is a made-up data set, patterned on the Blobel and Dobberstein experiments. A secretory protein’s messenger RNA is translated in reticulocyte lysate. You read band intensity in arbitrary units. The precursor runs at 26 kDa and a processed form at 23 kDa.

  • Lane 1, no microsomes. 26 kDa = 100. After protease: nothing.
  • Lane 2, microsomes present during translation. 26 kDa = 40, 23 kDa = 60. After protease: 23 kDa = 57, 26 kDa = 0. After protease plus detergent: nothing.
  • Lane 3, microsomes added after translation. 26 kDa = 100, no 23 kDa. After protease: nothing.

Assume the protease cannot cross an intact membrane and that the bands are proportional to the amount of protein.

Step 1: how much was processed, and was it protected? In lane 2, 60 of 100 units are processed, so 60 percent of chains lost their signal. Of those 60, 57 survive protease:

protected fraction = 57 / 60 = 0.95, or 95 percent

So nearly all processed chains are inside sealed vesicles, while every unprocessed chain is outside and digested.

Step 2: was it the membrane? Detergent dissolves the vesicles and the 23 kDa band vanishes. The protein is not resistant to protease on its own; the membrane protected it.

Step 3: does timing matter? In lane 3 there is no processing and no protection. In this assay, translocation needs ongoing translation.

Step 4: how big was the signal? The shift is 26 − 23 = 3 kDa. With an average residue mass of about 110 Da:

residues removed ≈ 3,000 Da / 110 Da per residue ≈ 27 residues

That fits a signal of the usual size; preprolactin’s is 30 residues.

Protection gel reader: was the protein inside the vesicles?⚙ original · interactive
Microsomes (ER vesicles) in the tube
Then treat with

    Protease cannot cross an intact membrane, so a band that survives protease but not protease plus detergent was inside a sealed vesicle; bands are proportional to the amount of protein. This is the lesson’s made-up data set, a sketch of the logic, not a real gel.

    SRP is not the only road. In budding yeast, rapid SRP depletion in 2018 left many proteins with cleavable signals still targeted while being made, although proteins with transmembrane domains generally needed SRP. Tail-anchored proteins, whose only hydrophobic segment is at the very end, are inserted after translation by other machinery.

    After translation: mitochondria, chloroplasts, peroxisomes

    The other organelles read finished, released chains. Each needs a receptor at the outer surface, a way across, and a source of energy, and they differ in whether the cargo must unfold.

    Mitochondria. A matrix protein’s presequence is recognised by the TOM complex of the outer membrane and passed to the TIM23 complex of the inner membrane. Two energy sources drive it: the membrane potential across the inner membrane, which pulls on the positively charged presequence, and ATP used by mitochondrial Hsp70 in the matrix motor. Many presequences are then cut off. The cargo must unfold. In 1986 Eilers and Schatz showed that methotrexate, which holds a presequence-DHFR fusion folded, blocked its import. Two cautions. Metabolite carriers of the inner membrane have internal signals and use a separate TIM22 route. And a 2023 study of the yeast complex placed the protein-conducting path in Tim17, not in Tim23 as older diagrams show.

    Chloroplasts. Transit peptides are read by TOC in the outer envelope and TIC in the inner. In 1978 Nam-Hai Chua and Gregory Schmidt showed that finished precursors of the Rubisco small subunit from pea and spinach, 4,000 to 5,000 daltons larger than the mature protein, entered purified intact chloroplasts and were processed. In that paper the precursor from the alga Chlamydomonas did not enter those higher-plant chloroplasts, but a later study (Mishkind, Wessler and Schmidt, 1985) found that the algal precursor is imported efficiently by spinach and pea chloroplasts, so the negative result did not hold. Recent structures implicate an ATP-driven motor, Ycf2-FtsHi, at the inner envelope.

    Peroxisomes. PTS1 cargo is bound in the cytosol by the receptor PEX5; PTS2 cargo by PEX7. Peroxisomes break the unfolding rule: in 1995 Walton, Hill and Subramani found that mammalian peroxisomes imported proteins held folded by cross-links, and even gold particles coated with signal-bearing protein. In Xenopus egg extracts in 2022, PEX5 went all the way into the lumen with its cargo, and was unfolded when pulled back out. Insertion of loaded receptor has been described as ATP-independent, but recycling PEX5 needs ubiquitin and ATP, so sustained import is not free.

    Reading and rewriting signals

    Signals are modular, so you can predict and test. Put the first 53 residues of a yeast mitochondrial precursor on mouse DHFR, and the fusion enters the matrix of yeast mitochondria. Add SKL to the end of a bacterial enzyme, and it goes to peroxisomes in mammalian cells. Delete the short region, at most twelve residues, at luciferase’s tail, and it stays in the cytosol.

    Three warnings keep predictions honest. First, sufficiency for one passenger is not universality. Second, SKL is a specific signal; only a few conservative substitutions kept it working in luciferase. Third, position matters: a signal that works at one end may fail at the other or inside the chain. When a protein carries two signals, the outcome is empirical. Sorting continues past the ER, through vesicle traffic and the secretory pathway; lipid tails that pin proteins to membranes are in lipid anchors.

    Design problem

    Protein Y has an amino-terminal amphipathic helix and ends in SKL. Where does it go, and how would you test it?

    One way to do it

    Predict first. The amino-terminal helix looks like a matrix presequence and SKL is a PTS1, so Y could go to mitochondria, peroxisomes or both. The presequence emerges first and, if accessible, may win. Treat that as a hypothesis.

    Build four GFP fusions: full-length Y, Y without the helix, Y without SKL, and Y without both. Image each in cultured cells with a mitochondrial marker and a peroxisomal marker, and score colocalization. Expect the double deletion in the cytosol. If deleting the helix moves Y to peroxisomes and deleting SKL leaves it in mitochondria, each signal works and the helix dominates in the full protein.

    Then test import directly. Incubate radiolabelled Y with isolated mitochondria, with and without an uncoupler, and treat with protease. Protected, processed Y that disappears with the uncoupler means true matrix import, not sticking to the surface. A similar protease test on isolated peroxisomes checks the SKL route.

    How we measure it

    Cell-free translation with microsomes and protease protection

    Messenger RNA is translated in a cell extract with or without ER vesicles (microsomes), and the products are treated with protease, with or without detergent. A protein that survives protease only while the membranes are intact is inside the vesicles. Adding membranes during or after translation tells you whether entry needs the chain to be growing.

    Salt extraction and add-back

    Membranes are washed in high salt to strip loosely bound proteins, tested for a lost activity, then given back a purified candidate factor. If the factor restores the activity, it is what the wash removed. This is how SRP was shown to be the selective ribosome-binding activity.

    Import into isolated organelles

    A radiolabelled precursor made in vitro is incubated with purified mitochondria or chloroplasts, then treated with protease. Protected, processed protein has been imported. Uncouplers, ATP depletion or a drug that locks the cargo folded test what import needs.

    Signal transplantation and mutagenesis

    A candidate signal is fused to a passenger protein that normally stays in the cytosol, or the signal is mutated residue by residue, and the fusion's location is read by microscopy or import. Moving a passenger shows that a signal is sufficient; losing location after a deletion shows that it is necessary.

    Finish line

    Test yourself

    8 questions

    Practice and MCAT-style, with explained answers.

    Next · Lesson 8

    The Nuclear Envelope and Nuclear Transport

    Why does a protein that shuttles in and out of the nucleus pile up inside when one drug is added?

    Continue →

    ← Lesson 6 · Organelles as Distinct Environments

    PEX5 TPR domain bound to a PTS1 peptide — how the peroxisomal receptor reads SKL
    PEX5 TPR domain bound to a PTS1 peptide — how the peroxisomal receptor reads SKL ● LIVE · RCSB PDB 1FCH
    open on RCSB PDB ↗
    Loading live structure 1FCH from RCSB PDB…

    The repeats curl into a bowl. Look for the short peptide lying in the groove, with its carboxyl-terminal leucine and free carboxylate tucked deepest: that is why the signal works only at the very end of a protein.

    Drag to rotate · scroll or pinch to zoom · right-drag to pan
    Tom20 receptor domain with a mitochondrial presequence — reading an amphipathic helix
    Tom20 receptor domain with a mitochondrial presequence — reading an amphipathic helix ● LIVE · RCSB PDB 1OM2
    open on RCSB PDB ↗
    Loading live structure 1OM2 from RCSB PDB…

    The presequence forms a helix whose hydrophobic face sits in a groove on Tom20, while its positive charges point away. The receptor reads a shape and a pattern, not an exact sequence.

    Drag to rotate · scroll or pinch to zoom · right-drag to pan
    SRP GTPase bound to its receptor GTPase — the GTP timer of ER targeting
    SRP GTPase bound to its receptor GTPase — the GTP timer of ER targeting ● LIVE · RCSB PDB 1RJ9
    open on RCSB PDB ↗
    Loading live structure 1RJ9 from RCSB PDB…

    These are the bacterial versions of SRP54 and the SRP receptor's alpha subunit. The two GTP sites face each other across the interface, so each GTPase helps the other hydrolyse GTP and the complex comes apart.

    Drag to rotate · scroll or pinch to zoom · right-drag to pan
    Methotrexate — the drug that locks DHFR folded and blocks mitochondrial import
    Methotrexate — the drug that locks DHFR folded and blocks mitochondrial import ● LIVE · PubChem CID 126941
    open on PubChem ↗
    Loading live structure 126941 from PubChem…

    Methotrexate binds dihydrofolate reductase very tightly and stabilizes its fold. Bound to a presequence-DHFR fusion, it stopped import into mitochondria in the Eilers and Schatz assay.

    Drag to rotate · scroll or pinch to zoom · right-drag to pan
    Günter Blobel

    Günter Blobel

    Proposed, with David Sabatini and then Bernhard Dobberstein, that a secretory protein carries an amino-terminal signal that sends its ribosome to the ER membrane, and with Dobberstein and Peter Walter built the cell-free systems that tested it.

    Blobel worked at the Rockefeller University, where George Palade's laboratory had traced secretory proteins from the rough ER outward. The question he took up was how a protein made on a ribosome gets across the ER membrane in the first place. With Bernhard Dobberstein he published two papers in the Journal of Cell Biology in 1975. In the second, ribosomal subunits from rabbit reticulocytes, which make globin and no secretory proteins, were mixed with ER vesicles from dog pancreas and messenger RNA for an antibody light chain from mouse myeloma cells. The light chain was made, trimmed and hidden inside the vesicles, but only if the vesicles were present while it was being made. Globin was never hidden. Parts from three species worked together, which argued that the address is in the protein, read by general machinery. The same style of mixing and testing later identified SRP and its receptor. Blobel received the 1999 Nobel Prize in Physiology or Medicine alone, for the discovery that proteins carry intrinsic signals that govern their transport and localization in the cell.

    ✓ verbatim Blobel & Dobberstein, J. Cell Biol. 67:852 (1975); Nobel Prize in Physiology or Medicine 1999, NobelPrize.org facts page.
    Peter Walter

    Peter Walter

    As a graduate student with Blobel, purified the signal recognition particle and showed, with Blobel, that it pauses the synthesis of a secretory protein until the ribosome reaches ER membrane, and later, with Reid Gilmore, identified its receptor, which Meyer, Krause and Dobberstein found independently as the docking protein.

    Walter did his doctoral work with Blobel at Rockefeller, according to his laboratory's biography from 1977 to 1981. The 1975 experiments showed that ER vesicles could take up a protein while it was made, but not what in the vesicles did the work. Walter and Blobel found that washing dog-pancreas ER vesicles in high salt removed their ability to bind ribosomes that were making bovine prolactin, while ribosomes making rabbit globin were never bound. Adding back a purified particle restored the selective binding. In the companion paper the same factor, then called signal recognition protein, in a wheat germ system stopped the growth of preprolactin chains at about 70 residues, and ER vesicles released the stop. In 1982 they reported that the particle holds six proteins and an RNA it cannot work without. With Reid Gilmore they used the particle as bait to fish out a 72,000-dalton receptor from the membrane. Each step made one guess testable: a soluble factor reads the signal, and a membrane receptor takes the ribosome from it. Walter joined the faculty of the University of California, San Francisco in 1983.

    ✓ verbatim Walter & Blobel, J. Cell Biol. 91:551 and 91:557 (1981); Gilmore, Walter & Blobel, J. Cell Biol. 95:470 (1982).
    Gottfried Schatz

    Gottfried Schatz

    His laboratory showed that the amino-terminal part of a mitochondrial precursor is enough to carry an unrelated mouse protein into the matrix of yeast mitochondria, and that locking that passenger in its folded state blocks import.

    Most mitochondrial proteins are encoded in the nucleus and made in the cytosol, so a mitochondrion has to recognise its own proteins among thousands of others. In 1984 Ed Hurt, Beate Pesold-Hurt and Schatz joined the first 53 residues of the precursor of yeast cytochrome oxidase subunit IV to mouse dihydrofolate reductase (DHFR), an enzyme that normally stays in the cytosol. Incubated with energized mitochondria isolated from yeast, the fusion protein reached the matrix. The experiment showed that this stretch was sufficient to send a passenger there; it did not show that all 53 residues were needed. In 1986 Eilers and Schatz added methotrexate, a drug that binds DHFR tightly and holds it folded. The fusion protein no longer entered, while the authentic precursor still did, and the authors concluded that the passenger has to unfold at least partly to cross. Together the two papers turned DHFR fusions into a standard tool for asking what a mitochondrial signal does and what state the cargo must be in.

    ✓ verbatim Hurt, Pesold-Hurt & Schatz, EMBO J. (1984), doi 10.1002/j.1460-2075.1984.tb02272.x; Eilers & Schatz, Nature 322:228 (1986).
    Suresh Subramani

    Suresh Subramani

    His laboratory mapped the peroxisomal targeting signal of firefly luciferase to its last three residues, serine-lysine-leucine (SKL), showed that SKL sends a bacterial enzyme into mammalian peroxisomes, and later showed that peroxisomes import folded proteins.

    Firefly luciferase is made in the cytosol of the lantern cells and ends up in peroxisomes. Expressed in cultured mammalian cells, it went to peroxisomes there too, which made it a handy probe. In 1987 Stephen Gould, G. A. Keller and Subramani narrowed the needed region to no more than twelve residues at the carboxyl end, and showed that the region was necessary and sufficient in their reporter constructs. In 1989, with N. Hosken and J. Wilkinson, they mutated the carboxyl end by site-directed mutagenesis and found that the last three, SKL, carried the signal. Only a few conservative substitutions kept it working. Adding SKL to the end of chloramphenicol acetyltransferase, a bacterial enzyme, sent it into peroxisomes. In 1995 P. A. Walton, P. E. Hill and Subramani showed that peroxisomes import proteins held folded by disulfide bonds or chemical cross-links, and even gold particles coated with signal-bearing protein. That result set peroxisomes apart from mitochondria and opened the question of how a membrane lets something so large through.

    ✓ verbatim Gould, Keller, Hosken, Wilkinson & Subramani, J. Cell Biol. 108:1657 (1989); Walton, Hill & Subramani, Mol. Biol. Cell 6:675 (1995).
    Test yourself
    Question 1 of 8 Score 0 / 0

    Which feature most reliably marks a protein for cotranslational delivery to the ER?
    Sources & Integrity

    Every structure, slide, simulation, and clip on this page is listed below with the source it came from. None of the 3D structures is drawn by hand or by AI: each one loads by accession ID from RCSB PDB or PubChem. This panel is generated from the lesson's own data file, so what it lists is what the page loads.

    Molecular structures live data

    • PEX5 TPR domain bound to a PTS1 peptide — how the peroxisomal receptor reads SKL — RCSB PDB 1FCH · coordinates CC0 · Human PEX5 tetratricopeptide-repeat domain with a pentapeptide ending in SKL, X-ray crystal structure (Gatto et al., Nat. Struct. Biol., 2000)
    • Tom20 receptor domain with a mitochondrial presequence — reading an amphipathic helix — RCSB PDB 1OM2 · coordinates CC0 · Rat Tom20 cytosolic domain with a presequence peptide from aldehyde dehydrogenase, NMR structure (Abe et al., Cell, 2000)
    • SRP GTPase bound to its receptor GTPase — the GTP timer of ER targeting — RCSB PDB 1RJ9 · coordinates CC0 · Thermus aquaticus Ffh NG domain with FtsY NG domain and a GTP analogue, X-ray crystal structure (Egea et al., Nature, 2004)
    • Methotrexate — the drug that locks DHFR folded and blocks mitochondrial import — PubChem CID 126941 · PubChem compound record for methotrexate

    Simulations

    • Protection gel: inside or outside? · Original interactive — Biochemistrypedia (Deppmann Lab).

    Scientist profiles & quotes

    • Günter Blobel — Blobel & Dobberstein, J. Cell Biol. 67:852 (1975); Nobel Prize in Physiology or Medicine 1999, NobelPrize.org facts page.
    • Peter Walter — Walter & Blobel, J. Cell Biol. 91:551 and 91:557 (1981); Gilmore, Walter & Blobel, J. Cell Biol. 95:470 (1982).
    • Gottfried Schatz — Hurt, Pesold-Hurt & Schatz, EMBO J. (1984), doi 10.1002/j.1460-2075.1984.tb02272.x; Eilers & Schatz, Nature 322:228 (1986).
    • Suresh Subramani — Gould, Keller, Hosken, Wilkinson & Subramani, J. Cell Biol. 108:1657 (1989); Walton, Hill & Subramani, Mol. Biol. Cell 6:675 (1995).

    Quotes are verbatim from the cited source. Profile text is original to this site and was checked against the sources cited.

    Animated explainer

    The Address Travels With The Protein · narration: an AI narrator voice (Gemini, Charon) reading narration written from this lesson's text · animation: AI-generated in code (Manim) by Claude, a schematic with no drawn molecular structures, plots computed from their equations · captions from the narration script.

    Images & figures

    This lesson carries no raster figures. Every structure is rendered live from RCSB PDB or PubChem, and every plot is drawn from its equation in the browser.

    integrity Machines check the mechanics. The build confirms that every local file exists, that every question's answer key points at a real choice, and that any AI-generated scientific figure names the person who checked it; a weekly job re-resolves every accession ID against its database. Questions are original; MCAT-style items are written for this site and are not real AAMC questions. Banned sources (the Tymoczko textbook, publisher decks, copyrighted figures) are not used.

    Like I'm five

    A cell is like a city with many buildings, and almost every protein is built in the open street. Each protein that belongs inside a building carries a short address written in its own chain. Workers at each building read the address and let the right proteins in. A protein with no address stays out in the street.
    High school

    Proteins are made by ribosomes in the cytosol, the watery inside of the cell. Some must end up inside the endoplasmic reticulum, the mitochondria, the chloroplasts or the peroxisomes. Each of those proteins carries a short signal in its amino-acid sequence. For the ER, a particle called SRP grabs the signal while the protein is still being made and steers the ribosome to a channel in the ER membrane. For mitochondria, chloroplasts and peroxisomes, the protein is finished first, and receptors on the organelle recognise its signal and pull it in. If you cut off a signal, the protein stays in the cytosol; if you add a signal to another protein, it goes where the signal says.
    College

    Sorting signals are continuous stretches of sequence. An amino-terminal hydrophobic signal is bound by the signal recognition particle as it emerges from the ribosome; SRP slows elongation, docks on the SRP receptor, and in a GTP-dependent step transfers the ribosome to the Sec61 translocon, where the chain is threaded into the lumen and signal peptidase removes the signal. Mitochondrial matrix proteins carry amphipathic presequences read by TOM and TIM23; import needs the inner-membrane potential and ATP for matrix Hsp70, and the cargo must unfold. Chloroplast proteins carry transit peptides read by TOC and TIC. Peroxisomal matrix proteins carry a carboxyl-terminal PTS1 such as SKL, read by PEX5, or an amino-terminal PTS2 read by PEX7, and can be imported folded. Signal transplantation and deletion test sufficiency and necessity.
    Graduate

    The classic cell-free work defined ER targeting as a sequence of separable activities: signal-dependent elongation arrest by SRP (about 70 residues for preprolactin in the wheat-germ cell-free system), arrest release by a membrane receptor separable from signal peptidase, and reciprocal GTPase activation between SRP54 and SR alpha that commits and then dissolves the targeting complex. Structures of Sec61 with an engaged signal show the signal opening the lateral gate. Yet SRP dependence is client-specific: rapid depletion in yeast left many cleavable-signal clients targeted cotranslationally, and tail-anchored proteins use post-translational routes. Post-translational systems differ in what they accept. Matrix import needs unfolding and is driven by the membrane potential on the presequence and an Hsp70 motor; a 2023 yeast structure placed the conducting path in Tim17, not Tim23. Peroxisomes import folded and oligomeric cargo, and in Xenopus egg extracts PEX5 enters the lumen with its cargo and is unfolded on extraction. Open question: how does the peroxisomal membrane pass large folded cargo without leaking small solutes, and is there a stable pore at all?
    Three species in one tube

    The decisive 1975 reconstitution mixed parts that had never met in nature. Ribosomal subunits came from rabbit reticulocytes, cells that make mostly globin and no secretory protein. The membranes were ER vesicles from dog pancreas, stripped of their own ribosomes. The messenger RNA encoded a mouse antibody light chain. The light chain was still processed and hidden inside the vesicles, while globin made in the same kind of reaction was not. No special secretory ribosome was needed: the information lay in the protein being made.

    — Blobel & Dobberstein, J. Cell Biol. 67:852 (1975).
    A signal that works only at the very end

    When Subramani's laboratory added serine-lysine-leucine to the carboxyl end of chloramphenicol acetyltransferase, a bacterial enzyme with no business in a mammalian organelle, the enzyme went into peroxisomes of cultured mammalian cells. Three residues were enough for that passenger. The same work showed how picky the signal is: in luciferase only a limited set of conservative substitutions kept it working. Later structures of the receptor, PEX5, showed why the free carboxyl end matters, since it sits deep in the binding groove.

    — Gould, Keller, Hosken, Wilkinson & Subramani, J. Cell Biol. 108:1657 (1989).
    A cotranslational exception in yeast

    For decades the textbook rule was that cotranslational delivery to the ER means SRP. In 2018 Costa and colleagues depleted SRP from budding yeast rapidly and mapped which ribosomes stayed near the ER. Proteins with transmembrane domains generally lost ER targeting, wherever the domain lay in the chain. Many proteins with cleavable signal peptides did not: they were still targeted while being made. The result keeps SRP central but shows, at least in yeast, that other routes carry part of the load.

    — Costa, Subramanian, Nunnari & Weissman, Science (2018), doi 10.1126/science.aar3607.

     ↑ Top