170 | MARCH 2004 | VOLUME 5 www.nature.com/reviews/molcellbio

An ongoing debate about mammalian Golgi-membrane inheritance might now have beenresolved by a paper from Pecot and Malhotra inCell. The debate has been whether Golgi mem-branes remain independent during mitosis orfuse with the endoplasmic reticulum (ER).And anew assay, designed by the authors, now indicatesthat Golgi membranes keep their independence.

During mitosis, Golgi membranes are ini-tially fragmented into large blobs, and theseblobs are then fragmented further to producethe ‘Golgi haze’. It has been hard to resolvewhether the Golgi fragments in this hazeremain separate from the ER during mitosis, orwhether, at some point, they fuse with the ERand then re-emerge.

The new assay is based on the ability of twoproteins — FK506-binding protein (FKBP) andFKBP–rapamycin-associated protein (FRAP) —to interact in the presence of the ligandrapamycin. Pecot and Malhotra made two con-structs: one that that contained FKBP fused tothe Golgi-protein sialyltransferase (ST–FKBP)and another that contained FRAP fused to theER-protein human invariant-chain protein(Ii–FRAP).

If the Golgi fuses with the ER at any timeduring mitosis, ST–FKBP will bind to Ii–FRAP

in the presence of rapamycin and be retained inthe ER. However, if the Golgi retains its indepen-dence, ST–FKBP will be found in re-formingGolgi membranes in daughter cells after mitosis.Pecot and Malhotra found that, after mitosis,ST–FKBP was localized to re-forming Golgifragments in the presence and absence ofrapamycin, which indicates that the Golgi doesnot fuse with the ER during mitosis.

To confirm their conclusion, the authorsused brefeldin A (a drug that induces ER–Golgifusion) and rapamycin to show that ST–FKBPbecomes trapped in the ER during mitosis ifthe ER and Golgi membranes are artificiallyfused. This result indicates that if Golgi pro-teins did cycle through the ER during mitosis,ST–FKBP would have been retained in the ERin this assay.

The authors believe that this work hasresolved the debate regarding mammalianGolgi-membrane inheritance, and it now seemsthat “…the biogenesis of the Golgi apparatus indaughter cells is from preexisting Golgi elementsrather than de novo”.

Rachel SmallridgeReferences and links

ORIGINAL RESEARCH PAPER Pecot, M. Y. & Malhotra, V. Golgimembranes remain segregated from the endoplasmic reticulumduring mitosis in mammalian cells. Cell 116, 99–107 (2004)

H I G H L I G H T S

It’s well known that heat shock causes therepression of pre-messenger RNA splicing.Reporting in Nature, James Manley andcolleagues have now identified the repressorresponsible for this effect — a generalsplicing repressor that is known as SRp38.

SRp38 is activated by dephosphorylationand, indeed, the authors showed that SRp38became dephosphorylated when HeLa cellswere heat shocked in vivo. When the cellswere allowed to recover, SRp38 reverted to itsphosphorylated form. Heat treatment ofHeLa nuclear extracts in vitro also resulted inthe rapid dephosphorylation of SRp38 andinhibited splicing activity. When nuclearextracts were left to recover, splicing activitywas restored.

Direct evidence came from depletionstudies, in which the depletion of SRp38 fromthe nuclear extracts of heat-treated cells, orfrom heat-treated nuclear extracts, restoredsplicing activity. Subsequent addition ofdephosphorylated SRp38 to the depletednuclear extract restored splicing inhibition.

Glutathione-S-transferase (GST) ‘pull-down’ assays revealed a mechanistic clue. TheU1 small nuclear ribonucleoprotein (snRNP)bound strongly to the GST–SRp38 fusion

protein, but only when SRp38 was in itsdephosphorylated form. Previous studieshad shown that the interaction betweenanother splicing factor, ASF/SF2, and U1snRNP stabilizes the U1 snRNP recognitionof 5′ splice sites. Manley and colleagues nowshow that SRp38 in its dephosphorylatedform results in a strong decrease inASF/SF2–U1-snRNP complex formation at5′ splice sites. This suggests thatdephosphorylated SRp38 represses splicingby interacting with U1 snRNP, and therebyinterferes with U1 snRNP recognition of 5′splice sites.

So, is SRp38-mediated repression ofsplicing important for cell survival understress conditions? The answer is yes — eventhough SRp38-deficient cells are viable, theyare temperature sensitive and recover moreslowly from heat shock than wild-type cells.So, the authors conclude that SRp38“…represses splicing following heat shock, atleast in part, to prevent the possibleaccumulation of inaccurately spliced mRNA”.

Arianne HeinrichsReferences and links

ORIGINAL RESEARCH PAPER Shin, C. et al.Dephosphorylated SRp38 acts as a splicing repressor inresponse to heat shock. Nature 427, 553–558 (2004)

Some don’t like it hot

R N A S P L I C I N G

Keep yourindependence

O R G A N E L L E I N H E R I TA N C E

© 2004 Nature Publishing Group