The peroxisomal membrane protein PEX11β has key roles in many, if not all, of the steps of the growth and division pathway. PEX11β possesses two membrane-spanning domains with a very short C-terminus and a larger N-terminus, both facing the cytosol ^[1][2][41,42]. The N-terminal domain contains amphipathic helices, which allow interactions with membrane lipids ^[3][4][43,44]. Furthermore, the N-terminus is required for the oligomerization of PEX11β ^[1][5][6][41,45,46]. Expression of PEX11β initially results in the deformation of the peroxisomal membrane at a defined site on the “mother” peroxisome. Subsequently, a membrane protrusion forms, which further elongates, before it constricts at multiple sites ^[7][40]. It is suggested that both phospholipid binding via the amphipathic helices and oligomerization of PEX11β are the driving forces for membrane remodelling, deformation and elongation. In addition to a PEX11β protein scaffold, certain membrane lipids, which promote changes in membrane curvature, may be required ^[8][9]. The role of the PEX11 isoforms PEX11α and PEX11γ in peroxisome division is less clear.

Although the fission GTPase DRP1 (Dynamin-releated protein 1) can form ring-like structures around organelle membranes, the diameter of the organelles is too large to allow ring assembly and requires constriction prior to DRP1 assembly. As peroxisomes can still constrict after loss of DRP1 ^[9][47] (see Section 3.1), other factors besides DRP1 are required. How peroxisomes constrict prior to fission is still unknown, although PEX11β may play a role as it is found at constriction sites ^[10][48], its manipulation blocks constriction ^[7][40], and it can constrict liposomes in vitro ^[11][49]. Mitochondrial division is facilitated by contact with the endoplasmic reticulum (ER) to form constriction sites by wrapping extended ER tubules around mitochondria ^[12][2]. DRP1 and its adaptors assemble at the mitochondria–ER contacts. These interaction sites may be determined by replicating mtDNA, which is positioned at these sites ^[13][3]. Furthermore, the ER-bound inverted-formin 2 (INF2) and the mitochondrial-anchored formin-binding Spire1C proteins assemble actin filaments at the mitochondria–ER contact sites, which mediate constriction prior to DRP1 ring formation ^[14][15][1,5]. Recent studies suggest that both membrane bending (induced by constriction) and tension (e.g., by cytoskeletal forces) contribute to mitochondrial fission at constriction sites ^[16][17][50,51]. If cooperative processes between the ER and actin are also involved in peroxisome constriction is unclear, but cannot yet be excluded. Peroxisomes are in close contact with the ER and form membrane contacts ^[18][19][52,53]. However, the ER is mainly found at the mother peroxisomes, and not at peroxisome tubules, which would need to constrict ^[20][54]. A major role of the peroxisome–ER contacts is the supply of membrane lipids for peroxisomal membrane expansion ^[18][21][22][31,52,55]. Mitochondrial constriction may be more complex than peroxisome constriction and require additional forces, as mitochondria need to coordinate division of their outer and inner membranes and may link this to mtDNA replication. As peroxisomes do not contain DNA and have only a single limiting membrane, constriction and division processes may be less complex.

Constriction of the peroxisomal membrane goes along with the assembly of the division machinery. It is composed of PEX11β and the tail-anchored membrane adaptor proteins FIS1 (Fission 1) and MFF (Mitochondrial Fission factor) , which can recruit DRP1 to the peroxisomal (and mitochondrial) membrane. Both FIS1 and MFF have been shown to interact with PEX11β at peroxisomes ^{[2][6][23][24]}[42,46,56,57]. PEX11β is not believed to be an adaptor for the recruitment of DRP1, but can interact with DRP1 to promote DRP1 assembly and subsequent stimulation of its GTPase activity ^[25][58]. DRP1 activity depends on a supply of GTP, which may be locally generated by DYNAMO1 (dynamin-based ring motive-force organizer 1)/NME3 (nucleoside diphosphate kinase 3). The Cyanidioschyzon merolae DYNAMO1 protein localizes to both the peroxisomes and mitochondria with a role in fueling membrane fission through local GTP generation ^[26][59]. NME3, the mammalian orthologue, was also found to localize to peroxisomes. NME3-suppression results in elongated peroxisomes, suggesting a role in peroxisome division by efficiently loading DRP1 with GTP and thus facilitating its fission activity ^[27][60].

The role of FIS1 in peroxisome division has been controversial ^[23][28][29][56,61,62]. Whereas the loss of MFF results in highly elongated peroxisomes (and mitochondria) due to a block in peroxisome division, the loss of FIS1 does not appear to change peroxisome morphology. This has led to the assumption that MFF is the major adaptor protein for peroxisomal (and mitochondrial) fission, whereas FIS1 may fulfill more specialized functions ^[29][62]. However, hwere have recently shown that overexpression of PEX11β can promote peroxisome division in MFF-deficient fibroblasts dependent on FIS1 ^[30][63]. Furthermore, overexpression of MFF in PEX11β-deficient fibroblasts restored the normal, spherical peroxisome morphology. These findings indicate that two independent mechanisms for peroxisome division may exist, one via MFF and another via PEX11β/FIS1.

After division, newly formed peroxisomes need to be distributed within the cell. Peroxisomes move along microtubules in mammalian cells and can recruit microtubule-dependent motor proteins, such as kinesin and dynein ^[31][32][33][64,65,66]. The Rho GTPase MIRO1 has been shown to localize to peroxisomes and mitochondria ^[21][34][35][31,32,33]. Like FIS1 and MFF, MIRO1 is also a C-tail anchored membrane protein with dual peroxisomal and mitochondrial localization ^[35][36][33,36]. It can serve as an adaptor protein for the recruitment of microtubule motor proteins ^[37][67]. The MIRO1/motor complex can exert pulling forces at peroxisomes, which can lead either to peroxisome motility along microtubules and re-positioning, or to the formation of membrane protrusions/membrane expansion ^[21][31]. The latter process requires attachment of peroxisomes to a fixed point, e.g., to the cytoskeleton or to the ER. Although the depolymerization of microtubules curiously promotes peroxisomal elongation and does not inhibit fission ^[38][39][68,69], microtubule-dependent pulling forces via MIRO1/motor proteins may facilitate peroxisomal membrane expansion and division ^[21][31].

Expression of a peroxisome-targeted MIRO1 (to avoid mitochondrial alterations) promoted the formation of peroxisomal membrane protrusions in PEX5-deficient patient fibroblasts ^[21][31]. PEX5 is a cytoplasmic import receptor for peroxisomal matrix proteins, which can interact with the peroxisomal membrane to deliver cargo proteins ^[40][41][70,71]. Loss of PEX5 function (or of other peroxins of the peroxisomal matrix protein import machinery) causes Zellweger syndrome (ZS), a spectrum of peroxisome biogenesis disorders with severe developmental and neurological defects ^[42][72]. It results in import-deficient peroxisomes, which lack peroxisomal enzymes within the matrix and are metabolically inactive ^[43][73]. However, peroxisomal membranes can still be formed, but peroxisomes are reduced in number and enlarged, presenting as “ghosts” (empty peroxisomal membrane structures) ^[44][74]. Interestingly, these peroxisomal “ghosts” are still dynamic, as they can form membrane protrusions promoted by MIRO1/motor proteins and can also elongate and divide, e.g., after PEX11β expression ^[21][31]. This indicates that these peroxisomal membrane dynamics are mechanistically independent of peroxisomal metabolism. Patients with a defect in peroxisomal β-oxidation, e.g., in acyl-CoA oxidase 1 (ACOX1), also exhibit enlarged peroxisomes in skin fibroblasts. Remarkably, addition of docosahexaenoic acid (DHA, C22: 6n-3) restored the normal peroxisome morphology in those cells ^[5][45]. The synthesis of DHA requires the cooperation of peroxisomes and the ER; the precursor undergoes one round of β-oxidation in peroxisomes before it is routed to the ER and may become incorporated in phospholipids. This implies that peroxisomes contribute to the synthesis of lipids, which are in turn required for their own biogenesis/membrane plasticity ^[45][8].

The C-tail anchored peroxisomal membrane protein ACBD5, a member of the acyl-CoA binding domain protein family, is involved in the formation of peroxisome–ER contacts (see Section 3.3). ACBD5 interacts with the C-tail anchored ER-resident membrane protein VAPB (vesicle-associated membrane protein-associated protein B) to tether both organelles ^[18][19][52,53]. Protein interaction is mediated by the ACBD5 FFAT (two phenylalanines in an acidic tract)-like motif, which binds to the MSP (major sperm protein) domain of VAPB. It has been shown that the peroxisome–ER contact sites play a role in peroxisome positioning/motility, in cooperative metabolic processes (e.g., ether lipid synthesis) as well as in peroxisome membrane expansion ^[46][75]. Loss of ACBD5 or VAPB results in a shortening of the highly elongated peroxisomes in MFF-deficient cells, likely due to the interrupted membrane lipid transfer from the ER ^[18][52]. A transfer of ER lipids is also suggested by observations in MIRO1-expressing cells (see above). MIRO1/motor protein-mediated pulling forces generate membrane protrusions, which have a much higher surface area than the mother peroxisome they are generated from. It is likely that the mother peroxisome does not possess sufficient amounts of membrane lipids to allow for the generation of such elongated tubules ^[21][31]. How membrane lipids are transferred from the ER to peroxisomes is unclear. Although ACBD5 has acyl-CoA binding activity, it is suggested that this function is used to sequester and deliver very long-chain fatty acids (VLCFA) to the peroxisomal ABCD1 (ATP-binding cassette D1) transporter for uptake into peroxisomes and subsequent β-oxidation ^[47][76]. Specific lipid transfer proteins at peroxisome–ER contact sites may be involved in lipid transfer, such as oxysterol-binding proteins (ORPs), which can shuttle individual phospholipids. Recently, a role for VPS13D (vacuolar protein sorting-associated protein 13D) in peroxisome biogenesis was reported ^[48][77]. VPS13D can interact with MIRO1 and VAPB to form a bridge between the ER and peroxisomes ^[49][78]. VPS13D is a large protein with a hydrophobic groove, which could allow lipid channelling between organelles ^[50][79]. Such a “bulk flow” of lipids would be consistent with the observed rapid elongation processes of the peroxisomal membrane ^[21][31].

Peroxisome membrane expansion results in membrane growth, which is linked to the multiplication/proliferation of peroxisomes by fission. However, the formation of peroxisomal membrane protrusions has also been linked to organelle interaction and communication. Interestingly, PEX11β was found to be co-regulated with proteins of the mitochondrial ATP synthase complex in a large-scale mapping approach, suggesting coordination of peroxisomal and mitochondrial functions ^[51][27]. Expression of PEX11β promotes peroxisome protrusions and is required for their formation ^[21][51][27,31]. These protrusions were observed to interact with mitochondria in mammalian cells and may facilitate metabolite exchange (e.g., for cooperative fatty acid β-oxidation and exchange of cofactors) and/or contribute to redox homeostasis ^[51][27]. Similar observations were made in plant cells, where light stress induced long peroxisomal membrane extensions (peroxules) which interacted with mitochondria ^[52][53][80,81]. The formation of those protrusions was dependent on plant AtPEX11a ^[54][82]. As peroxisomes in mammalian cells are often in close contact with and tethered to the ER, such protrusions may support simultaneous interaction and communication with a third organelle, e.g., mitochondria.