COMUNICACIONES

It has been shown morphologically that granule content proteins may aggregate into an electron dense core in the TGN in a region distinct from both constitutively secreted proteins and from budding immature dense core granules (Orci et at., 1987; Tooze et a l,

1987). These granule cores can be induced in the pancreatic ER of previously starved guinea pigs or guinea pigs treated with cobalt chloride (which prevents protein exit from the RER) (Tooze et a l, 1989). A rise in the concentration of the granule content proteins

in the ER is thought to cause this formation of granule cores or "intracistemal granules" (ICGs). These are homogenous aggregates that exclude soluble resident ER proteins, such as BiP (section 1.1.1.).

In addition to the high concentration of proteins in the TGN relative to those in the ER (Salpeter and Farquhar, 1981), the TGN in regulated secretory cells is reported to have a Ca^+ concentration greater than that of the RER (approximately lOmM) (Anderson and Orci, 1988; Chanat and Huttner, 1991 and references therein). Chanat and Huttner (1991) observed that a change in pH from neutrality to 6.4 and Ca^+ concentrations above ImM are sufficient to precipitate secretogranin II away from constitutively secreted proteins in the RER of PC 12 cells and GH4 cells. In vitro studies have also revealed granule proteins to bind Ca^+ and precipitate out of solution under conditions of low pH and high Ca^+ concentration (Gerdes et al., 1989; Reiffen and Gratzl, 1986; Yoo and Albanesi, 1990; Yoo and Albanesi, 1991). However, in these studies a pH of pH 5.2- 5.5 was used, which may not be applicable to the interpretation of events in the TGN lumen (which has a pH of approximately 6.4). A recent study on semi-intact PC 12 cells found that intra-cistemal Ca^+ concentration was not important for secretogranin II sorting into immature granules, but instead was critical to prohormone processing (Camell and Moore, 1994).

Early experiments using the pH disrupting agents, ammonium chloride and chloroquine, have indicated that the low pH of the TGN may be involved in sorting proteins into dense core granules: Treatment of AtT-20 cells with high concentrations of chloroquine resulted in the constitutive secretion of the endogenous regulated secretory product, ACTH

(Moore et al., 1983b) although a subsequent study with lower levels of chloroquine did not compromise granule targeting (Mains and May, 1988). lOmM ammonium chloride has been shown to block the sorting of secretogranin II to dense core granules in PC 12 cells (Gerdes et al., 1989) and 5mM ammonium chloride has been shown to block the correct sorting of vWf in endothelial cells (Wagner et al., 1986). However, in

chloride did not interfere with sorting to the regulated secretory pathway (von Zastrow et al.y 1989). In experiments on semi-intact PC 12 cells, the pH gradient was critical for exit from the TGN to both constitutive and regulated secretory pathways, implying that low pH provides no specificity to sorting at the TGN (Camell and Moore, 1994).

The role of low pH in TGN sorting therefore remains confused and may only have a contributary role as part of a balance of between high Ca^+ concentration and low pH (and perhaps other factors) that provide the correct environment for this event A TGN milieu- induced aggregation mechanism of sorting may account for the above discrepencies: In cases where pH was found unimportant for sorting to granules (Mains and May, 1988; von 2Iastrow et al., 1989) perhaps the combination of pH and Ca^+ concentration still remained near the optimal condensation conditions for that specific protein. Specific Ca^+ and pH requirements for the condensation of each protein type would account for spatial separation from constitutive secretory proteins and also from aggregates destined for other granules.

Chromogranin A is conform ationally dependent on pH, forming 30% a-helical tetramers at pH 5.5 and 40% a-helical dimers at pH 7.5. Chromogranin A undergoes aggregation at pH 5.5, a process that is reduced with a rise in pH and is sensitive to Ca?*

concentrations at the higher pH values (Yoo and Albanesi, 1990; Yoo and Albanesi, 1991). Secretogranin II also aggregates at pH 5.5, although the ability of Ca^+ concentration to promote aggregation is abolished at higher pH values (Gerdes et a l,

1989). Due to these well characterised aggregation properties and the fact that, as yet, they have no physiological role, it has been proposed that the granins may be

"aggregators" that ensure precipitation of other granule proteins and/or association of the granule core with the granule membrane (Huttner et al., 1991).

von Willebrand factor has been studied with respect to oligomerisation as a prerequisite for sorting to Weibal-Palade bodies: The fact that endothelial cells constitutively secrete dimers of both pro-vWf and mature vWf while retaining the large, biologically potent

multimers of vWf, originally indicated that covalent multimerisation of vWf may be required for its incorporation into Weibal-Palade bodies (Spom et al., 1986). However, (Wagner et al., 1991) found that storage of vWf occurs independently of its covalent multimerisation and instead suggest that the conditions in Weibal-Palade bodies favour multimerisation after sorting. vWf sorting to Weibal-Palade bodies is inhibited by pH disrupting agents (Wagner et al., 1986), which would indicate that although

multimerisation vWf is not involved in sorting, pH-dependent condensation of pro-vWf may an important step in Weibal-Palade body biogenesis.

The oligomerisation of insulin in pancreatic cells has also been studied in this context: Insulin oligomerises to form dimers and hexamers and the introduction of a point

mutation (His BIO to Asp BIO) that prevents hexamer formation results in the constitutive secretion of insulin (Carroll etal., 1988; Chan et al., 1987; Gross et al., 1989).

However, an alternative study on this mutation and a mutation that results in monomer formation (Ser B9 to Asp B9) indicates that subunit oligomerisation is not required for insulin incorporation into granules (Quinn et a i, 1991), although wild type insulin crystalisation was not successful in these experiments.

Insulin has been shown in in vitro studies to interact with a 25kD protein, an association disrupted when the pH is lowered to 5.0 (Chung et al., 1989). This insulin binding protein, or hormone binding protein (HBP 25), is present in the Golgi of the canine pancreatic acinar cells and AtT-20 cells, and was proposed to be a "sorting carrier" as part of the universal granule sorting mechanism. A non-specific regulated secretory protein- binding molecule of 31kD has been identified in porcine pancreatic cells that is similar to HBP 25 and binds to regulated but not constitutive secretory proteins. However, p31 could not be recovered from other tissues and was identified as the fully functional protein chymotrypsin (Gorr et al., 1992).

Finally, during studies on constitutive-like secretion from immature granules of isolated pancreatic islets, soluble C-peptide was observed exiting immature granules while

crystalline insulin was retained (Kuliawat and Arvan, 1992). In addition, insulin in these cells was shown to form from its precursors after exit from the TGN (Huang and Arvan,

1994; Orci et a l, 1994). These authors propose that absolute sorting to granules does not occur via precipitation in the TGN, but by a constitutive-like secretion of soluble material from immature granules as a passive consequence of the inability to form precipitates.

The sorting of soluble content proteins to dense core granules by selective aggregation (under specific conditions) would account for all available experimental data (reviewed by Huttner and Tooze, 1989; Kelly, 1991; Reaves and Dannies, 1991). Presumably, the relevant physical properties of granule content proteins for this aggregation would be conferred by a combination of particular primary sequences, perhaps as part of the propeptide sequences in the case of somatostatin and vWf. However, while mileu- induced aggregation of granule content proteins could account for their sorting and final retention in dense core granules, the targeting of dense core granule membrane proteins need not necessarily operate by such a mechanism.