In many instances, the formation of membrane-delimited compartments from pre-existing organelles is regulated by small GTPases of the Rab and Arf family [22, 23, 39, 40]. A growing amount of evidence suggests that monomeric GTP-binding proteins may also play a role in peroxisome biogenesis. For example, mass spectrometry-based quantitative proteomic profiling studies have shown that in yeast Rho1 is recruited to peroxisomes upon their induction by oleate, and it has been suggested that this protein regulates the assembly state of actin on the peroxisome membrane, thereby controlling peroxisome membrane dynamics and biogenesis . Recently, it has also been shown that MYA2, an isoform of Arabidopsis thaliana myosin XI, is targeted to peroxisomes through an interaction with AtRab2CA . Furthermore, mammalian Arf1 and Arf6 have been shown to bind to isolated rat liver peroxisomes, and ScArf1 and ScArf3, the yeast orthologues of these proteins, have been implicated in the control of peroxisome proliferation . A similar function has been suggested for Trypanosoma brucei Arf1, a protein which shares characteristics with both Arf1 and Arf6 and has a vital role in the maintenance of the endocytic pathway . Interestingly, no Arf proteins have been identified in proteome studies of yeast and mammalian peroxisomes [41, 44–50]. A possible explanation might be that the relative abundance of these proteins in the analyzed samples is extremely low.
In this study, we initially investigated whether or not yeast strains deficient in Arf1, Arf2, Arf3, Arl1, Cin4, Vps21, Ypt6, Ypt7, Ypt10, Ypt11, Ypt31, Ypt32, Ypt52 or Ypt53 were capable of growing on oleate as a sole carbon source and/or displayed a peroxisomal protein sorting defect. The results of these experiments demonstrated that all these strains, which embody every viable yeast deletion mutant of the Rab and Sar1/Arf subfamilies of small GTPases, contain functional peroxisomes. Interestingly, the average number of peroxisomes in oleate-grown cells was significantly upregulated in the Δarf1 and Δarf3 strains compared to the wild-type strain, and this phenotype was even more pronounced in the double Δarf1Δarf3 deletion mutant. This observation indicates that ScArf1 and ScArf3 dampen peroxisome proliferation, either directly or indirectly. Although this result confirms previous observations that ScArf1 and ScArf3 are implicated in the control of peroxisome proliferation , it is not consistent with the authors' conclusion that ScArf1 is required for the oleate-induced peroxisome proliferation in S. cerevisiae. The reasons for the discrepancy between our results and the results previously presented might be several. For example, while we were making use of an Arf1 deletion mutant, Lay et al.  utilized an Arf1 temperature-sensitive mutant in an Δarf2 background. Also, while these authors analyzed the peroxisome number per cell after 6 hours of growth on oleate, we analyzed the cells after 5 or 9 days of growth on oleate. Note that after transfer of the cells to oleate-containing medium, the number of peroxisomes per cell continuously increased as a function of time for at least 14 days, in the Δarf1 and Δarf3 strains compared to the wild-type strain (data not shown). At this point, it is not clear how Arf1 and Arf3 suppress oleate-induced peroxisome proliferation in S. cerevisiae. One possibility might be that these proteins are involved in the generation of peroxisome-derived vesicles, which transport ER-derived factors from peroxisomes back to the ER [7, 8, 12, 16] or deliver metabolites, lipids, or (mistargeted) proteins to mitochondria . Disturbances in such pathways would result in an increase in peroxisome number.
We next investigated whether or not endogenous Arf proteins were associated with the membrane of rat liver peroxisomes. In a first series of experiments, we obtained evidence that 1D9, a mouse monoclonal antibody with specificity for mammalian Arf1, Arf2, Arf3, Arf5, Arf6, and – to a lesser extent – Arf4 , recognized a 21 kDa band in highly purified peroxisomal membrane preparations. We could show that this 21 kDa protein was associated with the outside face of the peroxisomal membrane (as determined by immuno-electron microscopy), and identify this protein as Arf6 (as determined with Arf-specific antibodies). This protein, which is considered to be the mammalian orthologue of yeast Arf3, has been reported to act in a wide range of processes, including endocytosis, cytokinesis, phagocytosis, and the organization of the actin cytoskeleton . Our observation that endogenous Arf6 is also associated with peroxisomes may seem surprising given that the subcellular localization of this protein has already been documented in several reports [30, 52–54]. However, as (hepatic) peroxisomes occupy less than 2% of the cell volume , a partial (and transient) association of Arf6 with these organelles may have been easily missed. Also, as the strength of the signal and its detection depend heavily on the amount of protein and may be easily masked by stronger signals, the methods employed might not be sensitive enough. No evidence could be obtained for the association of any other endogenous Arf protein with the peroxisomal membrane. However, these results should be considered with caution as such proteins may dissociate during the purification procedure and be present on the peroxisomal membrane in concentrations below the detection limits of the employed Arf-specific antibodies.
To study the function of Arf6 in peroxisome biogenesis, we first examined the number of peroxisomes, peroxisomal morphology, and the location of (newly-synthesized) peroxisomal membrane and matrix proteins in (i) Ptk2 cells transiently overexpressing either wild-type Arf6 or a mutant defective in GTP-binding (Arf6T27N) or GTP-hydrolysis (Arf6Q67L) and (ii) Arf6-/- fetal mouse hepatocytes . Neither the overexpression of Arf6 variants nor the absence of Arf6 resulted in a detectable peroxisomal phenotype. Together, these observations suggest that peroxisome-bound Arf6 alone is not essential for peroxisome biogenesis in mammalian cells. The lack of a phenotypic defect is most probably the result of functional redundancy among individual Arf proteins , and since it has been reported that exogenously added Arf1 can bind to isolated rat liver peroxisomes , we decided to carry out another series of experiments in which Arf6WT, Arf6T27N, or Arf6Q67L were transiently co-overexpressed with either wild-type Arf1 or a mutant defective in GDP-binding (Arf1T31N,) or GTP-hydrolysis (Arf1Q71L). These experiments showed that a combined overexpression of Arf1 and Arf6 variants resulted in a (partial) mislocalization of newly-synthesized peroxisomal reporter proteins and an altered peroxisome morphology. Importantly, the intracellular transport pathway of HsLK2-Myc-His, a human mitochondrial multisubstrate lipid kinase , was not disturbed. Interestingly, combinations with Arf1T31N and Arf6T27N consistently displayed the strongest phenotypes. Similar observations were observed in CHO and HepG2 cells (data not shown). These mutants are thought to render endogenous Arf1 and Arf6 inactive, presumably by sequestering a select set of guanine nucleotide exchange factors [35, 52]. Since a similar, but less drastic, phenotype was observed when other Arf6 and Arf1 variants were expressed, and Arf3, Arf4, and Arf5 could not substitute for Arf1, it is unlikely that the (partial) mislocalization of the peroxisomal reporter proteins is the result of indirect changes in the level of activation of other Arf subtypes. In addition, our observation that a similar phenotype was obtained with all Arf1 and Arf6 variants also suggests the necessity for a completion of the GDP/GTP cycle of these proteins to achieve their potential function in peroxisome biogenesis [, and references therein]. In summary, these findings – which suggest that Arf1 and Arf6 act in tandem to regulate a peroxisomal trafficking pathway – lend strong support to and extend the hypothesis that the cooperation of two Arf proteins at the same subcellular location is a general feature of Arf signaling .
To date, only one group has studied the potential relationship between ADP-ribosylation factors and peroxisome formation in mammalian cells, and shown that exogenously added Arf1 and Arf6 can bind to isolated rat liver peroxisomes [17, 19]. In this work, we have extended these findings by showing that (i) also endogenous Arf6 associates with the cytoplasmic side of the peroxisomal membrane, and (ii) a combined overexpression of Arf6 and Arf1 variants selectively impairs peroxisomal protein import and alters peroxisome morphology. The molecular mechanisms underlying these phenotypic alterations remain to be clarified, since these Arf proteins do not interact directly with any of the currently identified human peroxins as tested with yeast- and bacterial two-hybrid assays (data not shown). Nevertheless, based on (i) a recent report showing that active Arf6 can recruit ARNO – a soluble Arf1 GTPase-activating guanine nucleotide exchange factor – to the plasma membrane , (ii) a growing body of evidence supporting the existence of vesicular trafficking pathways between peroxisomes and other subcellular compartments [8, 51], and (iii) the observation that Arf1 is involved in the transport pathway of tomato bushy stunt virus 33 kDa replication protein from peroxisomes to the ER in Nicotiana tabacum cells , it is tempting to speculate that activated peroxisomal Arf6 is involved in the formation of peroxisome-derived vesicles by serving as an adaptor for the recruitment of not yet identified guanine nucleotide exchange factor(s) to sites at the peroxisomal membrane which in turn may lead to Arf1 activation and cytosolic coat protein recruitment. Depending on the functional redundancy among the Arf proteins and the role of the peroxisome-derived vesicles in different organisms, the inability of a cell to form such vesicles may result in an increase in peroxisome number or a block in the formation of new organelles.