Mammalian START-like phosphatidylinositol transfer proteins – Physiological perspectives and roles in cancer biology

Abstract

PtdIns and its phosphorylated derivatives, the phosphoinositides, are the biochemical components of a major pathway of intracellular signaling in all eukaryotic cells. These lipids are few in terms of cohort of unique positional isomers, and are quantitatively minor species of the bulk cellular lipidome. Nevertheless, phosphoinositides regulate an impressively diverse set of biological processes. It is from that perspective that perturbations in phosphoinositide-dependent signaling pathways are increasingly being recognized as causal foundations of many human diseases – including cancer. Although phosphatidylinositol transfer proteins (PITPs) are not enzymes, these proteins are physiologically significant regulators of phosphoinositide signaling. As such, PITPs are conserved throughout the eukaryotic kingdom. Their biological importance notwithstanding, PITPs remain understudied. Herein, we review current information regarding PITP biology primarily focusing on how derangements in PITP function disrupt key signaling/developmental pathways and are associated with a growing list of pathologies in mammals.

1. Introduction

Phosphatidylinositol (PtdIns) and its phosphorylated derivatives, the phosphoinositides (PIPs), represent the chemical foundation of a major intracellular signaling system that controls almost all aspects of cell biology [1]. Presently, our understanding is summarized as follows: PtdIns is synthesized at the endoplasmic reticulum (ER) by PtdIns synthase (PIS) that uses myo-inositol and cytidine diphosphate diacylglycerol (CDP-DAG) as substrates. PtdIns is subsequently modified by phosphorylation at one or more specific position on the inositol ring to produce as many as seven phosphorylated isomers of PtdIns (PtdIns-3-phosphate, PtdIns3P: PtdIns-4-phosphate, PtdIns4P; PtdIns-5-phosphate, PtdIns5P; PtdIns-4,5-bisphosphate, PtdIns(4,5)P₂; PtdIns-3,5-bisphosphate, PtdIns(3,5)P₂; PtdIns-3,4-bisphosphate, PtdIns(3,4) P₂: and PtdIns-3,4,5-trisphosphate, PtdIns(3,4,5)P₃) depending on the specific organism. Mammals produce all seven PIPs. The production, interconversion and degradation of these lipids is controlled by a large set of positionally-specific lipid kinases and lipid phosphatases that themselves are subject to various layers of regulation.

PtdIns is a major phospholipid in some organisms such as yeast where it constitutes 20–25 mol% of total phospholipid. More typically, PtdIns is a minor phospholipid in mammalian cells where it constitutes ~5–10 % of total phospholipid. The PIPs are all minor phospholipids. In mammals, PtdIns4P and PtdIns(4,5)P₂ each constitute ~2–3 % of total inositol phospholipid whereas the others are present at significantly lower levels still [2]. Yet, the biological importance of these lipids is truly impressive [1]. Definitive recognition of the signaling roles of these lipids was documented in the decades of the 1950s through the 1980s in the seminal work of Lowell and Mabel Hokin, Robert Michell, Rex Dawson, Robin Irvine, Michael Berridge, Yasutomi Nishizuka and others. These pioneers linked agonist signaling to activation of phospholipase C (PLC) and hydrolysis of PtdIns(4,5)P₂ to produce the soluble inositol phosphate Ins-1,4,5-P3 (IP3) that launches Ca²⁺ – signaling responses, and the signaling lipid diacylglycerol (DAG) that launches protein kinase C (PKC)-mediated signaling responses [3–10]. Subsequent studies in many laboratories around the world demonstrate this to only be the small tip of a very large iceberg. Indeed, discoveries of new versions of these lipids continued extending even into the 1990s [11,12]. Systems responsive to PIP signaling include the actin cytoskeleton, multiple membrane trafficking pathways, multiple lipid metabolic pathways, lipid distribution, organelle ‘identity’, local membrane structure, activities of multiple protein families from plasma membrane to the nuclear matrix and most (all?) points in between [1,13,255].

Consequently, even subtle derangements in phosphoinositide metabolism are root causes for a wide range of human diseases from rare genetic disorders such as Lowe’s syndrome and Charcot–Marie–Tooth disorder, to more common disorders such as obesity, diabetes and cancer [14]. In addition, multiple virulent bacteria (e.g. Listeria, Yersinia, Escherichia, Salmonella) and viruses (e.g. hepatitis C, HIV, influenza) hijack host PIP signaling as important components of their virulence programs. As a result, proteins involved in PIP regulation are garnering increasing interest in the context of human health as potential therapeutic targets [1,15,16,254].

1.1. The concept of lipid transfer proteins in the context of the PtdIns cycle

The pioneering work of Donald Zilversmit and Karel Wirtz identified soluble proteins that exhibit the capability of mobilizing specific lipid molecules between membrane systems in vitro [17–19]. These studies were motivated by the hypothesis that cells express lipid transfer proteins which mobilize lipid transfer between organelles in vivo. The concept of intermembrane ‘PtdIns-transfer’ was further crystallized by Robert Michell’s prescient synthesis that predicted a role for both PtdIns and phosphatidic acid (PtdOH) transfer proteins in linking PLC mediated PtdIns(4,5)P₂ hydrolysis at the plasma membrane with calcium signaling and PtdIns resynthesis in the endoplasmic reticulum (ER) [20,21] (Fig. 1). In this conjecture, plasma membrane PtdIns pools depleted by PLC signaling are replenished from the ER via soluble PtdIns transfer proteins. A second round of lipid transport is proposed to close the cycle where PtdOH produced at the plasma membrane from the DAG generated by stimulated PLC activity is transported by the lipid transfer protein(s) to the cytoplasmic face of the ER. There, PtdOH is converted to CDP-DAG to fuel synthesis of PtdIns pools in ER and complete the so-called PtdIns cycle. This attractive hypothesis carries considerable weight in the interpretation of PtdIns transfer protein (PITP) function even to this day. As described below, there are alternative views for how PITPs interface with phosphoinositide signaling. Deciphering which mechanisms operate where and when define major questions for future research.

Fig. 1.

The phosphatidylinositol cycle and its synthesis. Phosphatidylinositol (PtdIns) is synthesized at the endoplasmic reticulum (ER) via the sequential action of cytidine diphosphate diacylglycerol (CDP-DAG) synthase 1/2, which converts phosphatidic acid (PtdOH) to the CPD-DAG intermediate, and PtdIns synthase which produces PtdIns. Michell’s PtdIns cycle hypothesis envisions mobilization of PtdIns from ER to the plasma membrane (PM) via soluble PtdIns transfer protein (PITP). PtdIns is converted to PtdIns4P by Type III PtdIns 4-kinase (PI4K) while Type I PtdIns 4-phosphate 5-kinase (PI4P5K) generates PtdIns(4,5)P₂ from PtdIns4P at the plasma membrane. PtdIns(4,5)P₂ undergoes hormone (agonist)-stimulated hydrolysis via the action of phospholipase C (PLC) to produce 1,4,5-trisphosphate (IP₃) and DAG as soluble and membrane second messengers, respectively. Subsequently, DAG is converted to PtdOH by DAG kinase, and PtdOH is retrieved back to the ER via a soluble lipid transfer protein such as a PITP.

1.2. A broad overview of PITPs

Pioneering breakthroughs in the PITP field came from the cloning of the major yeast PITP Sec14 and recognition of its role in coordinating lipid metabolism with protein trafficking from the yeast Golgi/endosomal system [22–24], and subsequent cloning of the first mammalian PITP structural gene with its product designated PITPα. These studies led to the understanding that PITPs are an ancient family of proteins highly conserved throughout the eukaryotic kingdom. That is, from single-cell eukaryotes to humans. PITPs fall into two structurally unrelated subfamilies: the Sec14-like and the START-like (Steroidogenic acute regulatory-like, StAR-related lipid transfer domain containing, StART-like or StARkin-related) PITPs [25,26]. Sec14-like proteins form a large superfamily and even simple eukaryotes such as yeast express multiple single-Sec14 domain members that include Sec14 and five functionally nonredundant paralogs/homologs ([22,27,28]). Other organisms expand their cohort of single-domain Sec14 proteins as well as incorporating Sec14 domains into multi-modular structures. Of the 32 Sec14-like proteins expressed by Arabidopsis thaliana, 12 link N-terminal Sec14-domains to C-terminal nodulin domains. These Sec14-nodulin proteins exhibit PITP activities and are required for polarized membrane morphogenesis [29]. Another 6 position Sec14-domains upstream of GOLD domains. The remainder represent single-domain Sec14-like proteins [30–33]. Clearly, the most anatomically complex plants have powerfully diversified their cohorts of Sec14-like proteins.

In mammals, Sec14-like proteins include α-tocopherol transfer protein (αTTP), Caytaxin, and cellular retinaldehyde binding protein (CRALBP). Functional deficits in these proteins lead to human diseases – i.e. ataxia with vitamin E deficiency [34,35], Cayman-type cerebellar ataxia [36,37], and retinal degeneration [38,39] respectively.

Mammals also express multi-domain versions – some of which are also associated with important human diseases. These include the guanine nucleotide exchange factors (GEFs) Kalirin/ Duo (a genetic risk factor for ischemic stroke, [40]; coronary artery disease, [41]; schizophrenia, [42]; Alzheimer’s disease, [43]) and the Rho GEF Dbl. N-terminal deletions in Dbl that remove the Sec14-domain and multiple spectrin repeat motifs yield a pro-oncogenic form [44,45]. Sec14 domains are also present in GTPase activating proteins (GAPs) including the CDC42 GAP/p50 Rho GAP (linked to chronic myeloid leukemia) [46]; and the Waldenstrom macroglobulinemia form of non-Hodgkin lymphoma [47]; and protein tyrosine phosphatases [48]. Comprehensive discussions of the roles of mammalian Sec14-like and Sec14-domain containing proteins in disease are reviewed elsewhere [49].

START-like PITPs are related to the StAR-related lipid transfer superfamily. These are absent from fungi, including the yeasts Saccharomyces cerevisiae and Schizosaccharomyces pombe, and to date none have been identified in plants [50]. These are however present in protist slime molds (Dictyostelium discoideum), protozoa (Plasmodium falciparum, Toxoplasma gondii) and in metazoans (flies, worms, fish, mammals) (Fig. 2). As a result, some draw a distinction between the two groups, referring only to START-like PITPs as ‘PITPs’ and to the yeast proteins as ‘Sec14-like PITPs’ [51].

Fig. 2.

Mammalian START-like PITP structural architecture. Schematic representation of class I and -II PITPs with their specific domains and splice variants. PITPs are grouped into two major classes. Class I PITPs consists of soluble PtdIns/PtdCho exchange proteins while class II proteins are subdivided into two classes: class IIa consisting of insoluble multi-domain proteins and the class IIb soluble protein PITPnc1. The STAMP PITP of Toxoplama gondii is also included. Relevant amino acid numbers are indicated for each protein/domain and different domains are color coded. Three splice variants of PITPβ and two splice variants of human PITPnc1 are also shown. The proteins with different C-terminus splice variants are shown in separate colors. Domain abbreviations: FFAT, the diphenylalanine (FF) in an acidic tract motif that serves as binding site for VAP proteins [243]; DDHD, an ~180 residue domain that represents a potential metal binding unit [242]; LNS2, the PtdOH-binding Lipin/ Nde1 /Smp2 domain [208]; TM, putative membrane-spanning domain.

1.3. PITPs: transfer proteins or metabolic nanoreactors?

Despite structural differences the Sec14- and START-like PITPs not only share the ability to transfer PtdIns between membranes in vitro, but high-level mammalian START-like PITP expression rescues yeast growth, and Golgi secretory defects associated with Sec14 deficiency [52–54]. This clear phenotype is observed in yeast cells that are naturally rich in PtdIns (second most abundant phospholipid in yeast membranes comprising 20–25 mol% of total phospholipid), and this degree of functional conservation suggests convergent evolution. As such, Sec14 offers conceptual insights for interpreting mechanisms of function for both Sec14-like and START-like PITPs. In that regard, while mammalian PITPs are generally interpreted as lipid transporters, compelling evidence from studies of Sec14 PITPs suggest other mechanisms where the PITP is a metabolic sensor, or scaffold for PtdIns (i.e. a nanoreactor), that uses a heterotypic lipid exchange cycle to ‘present’ PtdIns to PtdIns 4-OH kinases [25,54–56] (Fig. 3A,–B). In this manner, PITPs render PtdIns a superior substrate for the lipid kinase and thereby potentiate its activity so that it overcomes the antagonists of PtdIns-4-phosphate signaling such as the Sac1 phosphatase [57–59]. Interestingly, the ‘presentation model’, which does not consider intermembrane transport as a functional PITP activity, is consistent with all data presented as evidence for intermembrane PtdIns transfer. But PtdInstransfer models are not supported by genetic, structural, biochemical and physiological data that are more consistent with PtdIns presentation models. These issues have been discussed at length elsewhere and present alternative mechanistic views to lipid transfer. These ideas all await further interrogation [25,55,60–63]. Herein, we focus on the START-like PITPs and what is known about their physiological functions – particularly as these relate to disease mechanisms.

Fig. 3.

Heterotypic lipid exchange cycle of PITP (The “presentation” model): START-like PITPs exchange a second ligand (PtdCho for class I proteins and PtdOH for class II proteins) for PtdIns, and present PtdIns to an otherwise biologically insufficient PI4K enzyme, thereby potentiating PI4K activity and stimulating production of PtdIns4P pools dedicated to specific signaling reactions. The PtdIns4P pool available for signaling is depleted by PtdIns4P antagonists such as Osh proteins and the phosphoinositide phosphatase Sac1.

1.4. Mammalian START-like PITPs classification

The mammalian START-like PITPs, are variably classified according to two systems based on: 1) physical characteristics and behavior, and 2) sequence homology and evolutionary descent. In both cases, the soluble ~32 kDa PITPα and PITPβ are classified as class I. These are additionally described as ‘classical’ or ‘canonical’ based on their lipid binding specificity due to historical reasons. PtdIns/PtdCho PITPs were the first PITP activities to be identified in cytosolic fractions and by various methods those activities were ultimately recognized to be associated with the ~32 kDa PITPα and PITPβ – both of which bind PtdIns and PtdCho at distinct sites with ≈ 20-fold greater apparent affinity for PtdIns than PtdCho [60,64,65]. PITPnc1 (Phosphatidylinositol Transfer Protein, Cytoplasmic 1, a homolog of the Drosophila melanogaster retinal degeneration B protein RdgBβ), is the least characterized soluble START-like PITP and, being ~32–38 kDa (depending on splice form) and soluble, is in some instances classified as a class I PITP. However, it differs significantly from the highly homologous PITPα and PITPβ. It shares only 39 % primary sequence identity with PITPα and PITPβ [66,67]. Rather, the soluble PITPnc1 exhibits greater sequence homology with class II PITPs and this distinction is reflected in its biochemical properties as PITPnc1 is a PtdIns/PtdOH exchange protein that shows poor PtdCho binding activity [67].

Class II PITPs are multi-domain PITPs that exhibit N-terminal PITP domains followed by large C-terminal extensions, and the founding member of this family is the large and tightly membrane-associated multidomain Drosophila RdgBα protein [53,68–70]. Fly rdgBα mutants suffer light-accelerated retinal degeneration and its PITP domain is both necessary and sufficient for biological function [53,71]. The soluble PITPnc1 shares greater sequence homology with the class II PITPs and this is reflected in shared biochemical properties as the class II proteins also exhibit PtdIns/PtdOH exchange activites. In this review, we subdivide the class II PITPs into the insoluble class IIa proteins represented by PITPnm1 (RdgBα1/Nir2), PITPnm2 (RdgBα2/Nir3), and PITnm3 (RdgBα3/Nir1) and class IIb comprised of the soluble PITPnc1 (RdgBβ) (Fig. 2).

1.5. Mammalian PITP structure

The START/ StART-like PITPs are members of a larger protein superfamily of proteins containing the evolutionarily conserved Bet v1 fold (Bet v 1-like superfamily in the Structural Classification of Proteins (SCOP) database [SCOP:d.129.3]/ Bet v 1-like clan in the Pfam protein family database [PfamC:CL0209]) [51,72]. This ancient lipid-binding scaffold was discovered in white birch (Betula verrucosa) major pollen allergen and is present in multiple important commercially and clinically important plant proteins [73]. The three-dimensional Bet v1 fold forms a structure that surrounds a large hydrophobic cavity, and this fold now describes a large protein superfamily primarily comprised of lipid-binding proteins [74]. The archetypical StART domain was described by the crystal structure of StAR (Steroidogenic acute regulatory protein), a mitochondrial protein of steroid-producing cells [75]. This domain is now recognized in thousands of proteins from organisms ranging from bacteria to vertebrates forming a superfamily known as the START/ StART, Bet v1-like or SRPBCC (START/ RHOalphaC/PITP/Bet v1/CoxG/CalC) domain proteins. While outside of the scope of this review, the wider StART family contains notable members including ceramide transfer protein (CERT), the cholesterol-binding protein GRAM, the StaR and MLN64 proteins, and PtdCho transfer protein PCTP [51,76,77].

Genetic screens in yeast expressing PITPα yielded PtdIns binding mutants that provided the initial information regarding what residues are candidates for specific interactions with the PtdIns headgroup [52]. Subsequent determination of both apo- and lipid-bound PITPα/PITPβ structures both confirmed and extended the results from the genetic screens [26,78,79]. Several other functional elements are inferred from these structures including a putative membrane insertion loop, a regulatory domain that is potentially phosphorylated by protein kinase C (PKC), and a C-terminal helix potentially involved in allosteric regulation of PtdIns-binding given that substitutions within this helix result in specific PtdIns-binding defects even though the residues involved are removed from the PtdIns-binding substructure [52,64].

1.6. Class I PITPs: function and pathology

PITPα and PITPβ are encoded by the PITPNA and PITPNB genes, respectively. These exhibit distinct expression and subcellular localization patterns despite their structural and biochemical similarities. PITPα and PITPβ are ubiquitously expressed, with highest levels of PITPα in brain, and PITPβ in liver and neutrophils [80–83,247]. In terms of subcellular localization, PITPα shows a predominantly cytoplasmic profile with a pool localized to the nuclear matrix [80,83]. Lipid binding appears to play some role in nuclear localization since a PtdIns-binding mutant fails to localize to nuclei [84]. A recent study reports PITPα/PITPβ translocate into the nucleus following cellular stress (genotoxic, oxidative, ferroptotic) to activate nuclear phosphoinositide signaling response(s) ([85] BioRxiv). While a PITPα role in regulating nuclear phosphoinositide signaling is an exciting possibility, the available pulse-chase data suggest it does not significantly contribute to bulk nuclear PtdIns import [84].

PITPβ localizes to ER, and cytoplasm and is variously reported at cis- and trans Golgi [80,83,86–89]. PITPβ has seven spliceoforms (PITPα only one), and two have been studied– PITPβsp-1 and PITPβsp-2 [88,90]. Phosphorylation of PITPβsp-1 at Ser262 (absent in PITPα) was previously proposed to support Golgi localization [91] but mutants altered at this residue efficiently target to this organelle [88]. Localization to TGN membranes is endowed by a four cooperative element targeting code comprised of three functionally redundant motifs residing in the C-terminal 28 residues plus a W₂₀₂W₂₀₃ motif [88].

1.7. Class I PITP signaling in transformed cells

Unfortunately, the literature regarding in vivo roles of PITPα or PITPβ in cell proliferation is confusing. Early studies suggested PITPα is required for receptor tyrosine kinase signaling at the plasma membrane. The experimental models used to make these conclusions were: (i) epidermal growth factor receptor (EGFR) signaling in permeabilized cells [92,93], (ii) pull-down experiments showing PITPα binds to the P3 motif of the netrin receptor ‘deleted in colorectal cancer’ (DCC) cytoplasmic tail, and (iii) neurite outgrowth experiments to suggest PITPα is essential for netrin signaling [94]. In these contexts, PITPα is postulated to transport PtdIns from the ER to the site of receptor signaling in the plasma membrane to generate a signaling pool of PtdIns(4,5)P₂. Subsequent studies indicating PITPα null mice or cells do not show phenotypes that are easily reconciled with defects in EGFR or DCC signaling ([95]; reviewed in [63]). As one example, the P3 motif is cleanly deleted in the DCC^kanga mouse [96]. Yet, the kanga mouse does not phenocopy the PITPα-deficient mouse. Moreover, PITPα-deficient mice form an anatomically normal corpus callosum whereas DCC mutants fail to form this structure. Given the PITPα/DCC connection has not witnessed any further progress in nearly 20 years, the link between PITPα function and netrin signaling is not compelling.

In other studies, reduced PITPα expression in rat WRK-1 mammary tumor cells is reported to diminish cell proliferation [97], while PITPα ablation in murine embryonic stem cells has no effect on growth or pluripotency [95]. PITPα over-expression is reported to significantly stimulate proliferation of murine NIH3T3 fibroblasts [98], and that such cells secrete a highly mitogenic arachidonic acid metabolite [99–101]. The mitogenic factor remains to be identified, but it is suspected to be an endocannabinoid that exerts various anti-apoptotic effects. As conditioned media from PITPα over-expressing cells protects primary spinal cord motor neuron cultures against serum deprivation-induced cell death – it is suggested that PITPα is a potential target for manipulation in managing neurological disorders [102]. By contrast, elevated PITPβ expression in NIH3T3 cells leads to a slowed cell cycle, reduced proliferative capacity, and increased sensitivity to UV- and TNFα-induced apoptosis [100,101]. These overexpression data are difficult to reconcile with the studies in more physiologically relevant model organisms (see below).

The linkage of PITPα/ PITPβ activity to cell proliferation and plasma membrane signaling is more relevant in certain cancer contexts, however. The Hippo pathway offers a prime example. The Hippo pathway controls cell proliferation and organ development in mammals as it maintains tissue homeostasis and sets organ size during development. This signaling modality consists of a cascade of core MST1 (mammalian Ste20-like kinase 1 also called STK4)/ MST2 (STK3) kinases that target the LATS1/2 (Large tumor suppressor kinase 1/2) kinases that control a homologous pair of transcriptional regulators termed YAP (Yes-associated protein) and TAZ (WWTR1) [103]. At high cell densities, the Hippo pathway is “ON”. In the “ON” state, the MST kinases phosphorylate the LATS kinases that in turn phosphorylate YAP – leading to retention of YAP in the cytoplasm and its degradation. At low cell densities, the Hippo pathway is “OFF”. In the “OFF” state, YAP translocates to the nucleus and the YAP/TAZ complex binds to TEAD (Tea domain) 1–4 family of transcriptional factors to induce expression of pro-proliferative and survival enhancing factors [103–105]. Even subtle perturbations in this circuit result in disease – including cancer [106,107].

Recent findings show PITPα/ PITPβ potentiate Hippo signaling by enhancing plasma membrane PtdIns4P synthesis ([108]; Fig. 4A). This understanding derived from translational interests in identifying and developing novel therapeutic strategies targeting the pathway. Microcolin B (MCB) was identified in a chemical library screen of marine natural products as a Hippo signaling activator that preferentially kills YAP-dependent uveal melanoma (UM) 92.1 cells. Subsequent structure-activity relationship (SAR) studies identified VT01454 (an N-terminally acylated peptide) as a more potent MCB analog that displayed selective cytotoxicity towards YAP-dependent cancer cells [108], and this toxicity is executed via activation of upstream MST1/2 kinases that initiate the Hippo phosphorylation cascade and inactivate the pathway [108]. The cellular targets of VT0145 are PITPα and PITPβ and it acts as a covalent inhibitor that adducts to the Cys residue of the headgroup binding region of the PITP lipid binding pocket. The pharmacological effects are recapitulated in double PITPα/ PITPβ knockdown experiments in a range of cancer cell lines, and VT01454-induced stimulation of both YAP phosphorylation and cytoplasmic localization is suppressed by elevated expression of PITPβ (the effect of PITPα was not reported) [108]. Experiments targeting specific organelles with the Sac1 PtdIns4P phosphatase indicate that it is the plasma membrane PtdIns4P pool that activates Hippo signaling [108]. Moreover, lipid sensor-based experiments indicate VT01454 treatment significantly reduces plasma membrane PtdIns4P with a minor effect on Golgi PtdIns4P and no major effect on overall cellular PtdIns(4,5)P₂. The former findings are in interesting contrast to the in-situ results obtained in primary neural stem cells and murine and human pancreatic β-cells [109–111]. One possibility is there are significant cell-type differences in PITPα/ PITPβ function. Alternatively, the plasma membrane might be a more effective membrane sink for VT01454 relative to membranes of the Golgi system. Unfortunately, given the challenges associated with its synthesis, VT01454 production has apparently been discontinued.

Fig. 4.

Role of Class I PITPs in human disease. (A) Class I PITPα, and PITPβ are identified as upstream regulators of the Hippo signaling pathway that enhance PtdIns4P synthesis in the plasma membrane. The absence of PITPα and PITPβ promotes YAP phosphorylation and cytoplasmic retention. PITPα and PITPβ are identified as the molecular targets of Microcolin B (MCB) and its derivative VT01454, which upregulate the Hippo pathway by preferentially depleting plasma membrane PtdIns4P pools [108]. (B) Class I START PITPα and PITPβ are involved in the polarized distribution of Golgi in neuronal stem cells. PITPα and PITPβ exchange PtdCho and PtdIns, stimulating PI4K activity to produce PtdIns4P on trans-Golgi network (TGN) membranes. That PtdIns4P pool recruits GOLPH3 and CERT and likely other effectors to TGN membranes where GOLPH3 interacts with the non-conventional actin protein myosin, Myo18A. The recruitment promotes apical enrichment of the Golgi system in the apical process of neural stem cells [110]. (C) PITPα is downregulated in human T2D pancreatic beta cells and its ablation in both murine and human beta cells results in impaired glucose stimulated insulin secretion and granule maturation leading to chronic ER stress. PITPα executes heterotypic PtdCho/PtdIns exchange on beta cell TGN membranes and thereby stimulates PI4K activity and PtdIns4P synthesis. PITPα re-expression in islets donated by type 2 diabetic (T2D) patients rescue pancreatic beta cell dysfunction by restoring insulin granule biogenesis and glucose stimulated insulin secretion to those islets [111]. (D) PITPα/PITPβ regulate non-canonical Planar Cell Polarity (ncPCP) signaling in neural stem cells (NSCs) by potentiating the trafficking of specific cohorts of receptors of the ncPCP pathway such as Vangl2 from the TGN to the plasma membrane [109]. This in turn controls actomyosin dynamics at the nuclear periphery thereby regulating interkinetic nuclear migration (IKNM). PITPs promote PtdIns4P signaling in the NSC TGN/endosomal system to support efficient membrane trafficking of a subset of ncPCP receptors (e,g, VANGL 2) to the plasma membrane. ncPCP receptor signaling activates the cytoplasmic actomyosin system in the nuclear periphery which, in turn, provides the intracellular force for IKNM in proliferating NSCs. IKNM is a distinctive feature of the NSC cell cycle and facilitates asymmetric expansion of newborn NSCs in the lateral, as opposed to the radial, dimension. Actomyosin dynamics at the apical tip of the NSCs might also contribute to IKNM by altering fluidity/plasticity of the apical plasma membrane. IKNM shapes morphogenesis of the neocortex as a thin tissue with an expansive surface area [109]. Defects in tangential expansion of the neocortex are associated with diseases of intellectual disability and autism.

1.8. PITP function in model organisms and primary cells – examples of functional redundancy

Functional data make a much stronger case that PITPα/β resembles the yeast Sec14 PITP [22,27] in its involvement in membrane trafficking from the TGN secretory vesicle biogenesis in vitro [112–115] and in vivo [109,110]. The in vivo evidence comes from in situ experiments inter-rogating PITP function in neural stems cells of embryonic mouse fore-brain. In this physiologically relevant system, PITPα and PITPβ share functional redundancy. Moreover, both PtdIns and PtdCho-binding/exchange are required to reside in the same PITP molecule to produce a functional PITP [110]. These data recapitulate the results obtained using specific lipid binding mutants of the yeast Sec14 as reported by Schaaf et al. [25]. Interestingly, whereas the neural stem cell work describe PITPα/β function in terms analogous to that of yeast Sec14, and Sec14 is an effective surrogate for PITPα/β in the in vitro systems mentioned above, Sec14 is unable to substitute for PITPα/β function in neural stem cells [110].

A wealth of animal studies indicate important roles for PITPα/β in normal physiology and disease. Knockout of murine PITPα results in mice born at normal Mendelian ratios but the homozygotes die within a few days. Mortality results from a complex phenotype: spinocerebellar neurodegeneration, hypoglycemia, and intestinal and hepatic steatosis [116]. The intestinal steatosis of PITPα-deficient mice is feeding related and involves accumulation of neutral lipid in the lumen of the ER – suggesting a defect in chylomicron biogenesis [116]. Such a defect is by itself a critical failure for a nursing neonate. Neurodegeneration occurs with very rapid onset and is characterized by reactive gliosis of brain, and brain stem, and widespread demyelination (or incomplete myelination) of spinal cord motor neurons [116–118].

The naturally occurring murine vibrator (vb) mutation is a hypomorphic allele resulting from insertion of an endogenous transposable element (intracisternal A particle) into intron four of the PITPNA gene. This insertion mutation results in an 80 % reduction in PITPα mRNA and protein levels [118–120]. Mice homozygous for the vb allele suffer a neurodegenerative disease characterized by progressive whole-body tremor, spinocerebellar neurodegeneration and ascending motor paralysis despite unimpaired PITPβ expression [116–121]. Whereas vibrator (vb) mice in the C57BL/6 genetic background produce 20 % of WT PITPα protein levels and die 33–42 days after birth, vb/null heterozygotes express PITPα at only 10 % of WT levels and die 17–20 days after birth with more severe and more rapid onset of neurodegenerative symptoms [117]. Strikingly, vb/vb homozygotes in other genetic backgrounds show extended lifespan. One such modifier is identified as nuclear export factor-1 [Nxf1/ TAP]. The modifier allele is mechanistically trivial as it improves the splicing efficiency of the offending IAP-containing intron with a resultant increase in PITPα expression [119]. An extreme example is provided by the A/J background where vb/vb homozygotes live to nearly 2 years – indicating the existence of several genetic loci that combine to powerfully modify PITPα deficiency [120,121]. An exciting aspect of these A/J vb/vb mice is that they do not exhibit elevated PITPα expression, these animals suffer the same neuropathologies, and the severity and rate of onset of tremors and neurodegenerative disease resemble those that afflict vibrator mice in the C57BL/6 background. Interestingly, as A/J mice are considered to be ‘aggressively sedentary’, this modifier effect might have a behavioral basis that derives from reduced motor activity. This speculation is based on our anecdotal observation that vb/vb homozygote life-span is increased if chow is placed on the floor of the cage rather than in an elevated basket. Additional modifier studies promise valuable insights into PITPα function and pathologies associated with PITPα -deficiency.

Early attempts at generating PITPNB null embryonic stem cells or mice failed – leading to the suggestion that PITPβ is a mammalian counterpart to yeast Sec14 essential for housekeeping Golgi function [95]. Other studies proposed PITPβ acts at the Golgi-ER interface where it potentiates COP1- and PtdIns4-dependent retrograde Golgi to ER trafficking with no effect on anterograde membrane trafficking [86]. However, it is now clear that PITPNB null mice in the C57BL/6 background are born alive at Mendelian ratios, are fertile and have no discernable phenotype past one year of age [110]. Several lines of evidence indicate some level of PITPα/ PITPβ functional redundancy in these animals. First, PITPα/ PITPβ double null embryos suffer preimplantation lethality (i.e fail prior to E9.5) and in utero electroporation experiments indicate PITPα/PITPβ are functionally redundant in regulating TGN/endosomal PtdIns4P pools in NSCs [109,110]. PITPα/PITPβ -mediated maintenance of TGN/endosomal PtdIns4P pools in NSCs promotes recruitment of phosphoinositide effectors (e,g, GOLPH3, CERT) to these membranes (Fig. 4B). This activity plays a critical role in linking the TGN to the actin cytoskeleton and in maintaining a highly polarized neuroepithelium and functional neurological niche [110]. Evidence for evolutionary conservation of vertebrate PITP function in NSC biology is provided by the demonstration that deficiencies in the single Drosophila PITPα/β ortholog (Vib) disrupt regulation of neuroblast homeostasis – i.e. the fly equivalent of NSCs. In flies, Vib regulates plasma membrane PtdIns4P to promote asymmetric partitioning of cell fate determination factors into mother and daughter cells during cytokinesis [122].

1.9. PITP function in model organisms and primary cells – examples of functional diversification

While PITPα/ PITPβ share functional redundancy in some contexts, this relationship is not absolute as there are cases where expression of one does not substitute for functional ablation of the other. The role of PITPα in insulin secretion and type 2 diabetes (T2D) is one of several important examples of such (Fig. 4C). Reduction in insulin-producing pancreatic β cells underlies T2D progression. The subsequent chronic increase in insulin demand ultimately results in a destructive feedback loop of ever-increasing insulin production in the face of defective insulin secretion. This failed loop results in β-cell failure and ultimately to cell death [123]. PITPα is required for β-cell viability in mice [116]. Recent work powerfully extends those initial results by demonstrating PITPα promotes PtdIns4P signaling and insulin trafficking from the TGN of both human and murine pancreatic β-cells (Fig. 4C). A signature phenotype observed in both whole body and β-cell-specific PITPα KO mice are reduced numbers of pancreatic islets that show shrunken and vacuolized islet morphologies, reduced numbers of insulin-producing β-cells, decreased levels of insulin secretion, accumulation of proinsulin, and activation of ER stress responses [111,116]. Moreover, these mice present random-fed hyperglycemia due to impaired glucose-stimulated insulin secretion [111]. All of these phenotypes are apparent in the face of robust expression of PITPβ. The activation of ER stress responses is proposed to be an effect of the late secretory block that feeds back to prevent further proinsulin export from the ER.

From the human patient perspective, the 3′ untranslated region (UTR) of PITPNA is a target for microRNA miR-375 – a potent negative regulator of insulin secretion in humans whose inhibition elevates insulin production [111,124]. Human T2D patients display markedly reduced PITPα abundance (with unaltered PITPβ levels) in pancreatic islets compared with non-diabetics. This reduction inversely correlates with glycemic status and body mass index (BMI) [111,125]. Strikingly, ex vivo restoration of PITPα expression in human pancreatic islets of T2D donors is sufficient to rescue β-cell function as evidenced by restoration of PtdIns4P signaling, improvements in insulin granule biogenesis/maturation and insulin secretion, restoration of glucose responsiveness, and alleviation of ER stress [111]. These findings raise the exciting possibility that restoration of PITPα in T2D patients will be an effective treatment for T2D.

In terms of the cellular defects, both murine and human PITPα-deficient β-cells yield consistent results. PITPα deficits result in accumulation of immature- and empty insulin secretory granules, decreased number and size of docked- and mature insulin secretory granules at the plasma membrane, and distended Golgi and ER membranes. The insulin trafficking defect is accompanied by induction of an ER stress response (the result of proinsulin accumulation in the ER), deranged mitochondrial dynamics and performance, and subsequent apoptosis [111]. As expected, PITPα deficiencies in pancreatic β cell islets were accompanied by significant reductions in total cellular PtdIns4P and in TGN membrane pools of PtdIns4P ([111]; Fig. 4C). These PITPα LOF phenotypes are on display even though β-cells produce PITPβ.

Another example of distinctions in PITPα/PITPβ functional mechanisms is provided by the megakaryocyte (MK) and platelet systems. MK-specific knockout of PITPα or both PITPα and PITPβ result in decreased numbers of hematopoietic stem cells (HSCs), MK-erythrocyte progenitors, and cycling hematopoietic progenitor cells (HPCs) [126]. Megakaryocytes are large platelet producing cells, derived from pluripotent hematopoietic stem cells (HSCs). Platelets and megakaryotes possess α-granules and dense granules. The α-granules contain proteins involved in platelet adhesion during vascular repair including fibrinogen, Von Willebrand Factor (VWF), P-selectin, and angiogenesis regulatory proteins. Dense granules function predominantly in recruiting platelets to sites of vascular damage and secrete ADP and ATP. ADP acts as an agonist triggering platelet morphological change, granule release and aggregation. Megakaryotes lacking PITPα/ PITPβ possess morphologically normal dense granules but exhibit ultrastructural defects with depleted α-granules and large, misshaped multivesicular bodies, suggestive of defects in biogenesis, storage, and/or regulated secretion of α-granule contents [126]. PITPα-deficient platelets exhibit reduced PtdIns4P and PtdIns(4,5)P₂ pools, reduced cytosolic Ca²⁺ and an 80 % reduction in IP₃, compared to wild type controls following thrombin-stimulated platelet activation [127]. The data are interpreted as PITPα-deficient platelets being defective in the ability to initiate and/or sustain PLC signaling [51,127]. However, mice lacking PITPα in platelets display almost normal levels of platelet aggregation and α-granule release following stimulation with thrombin. Moreover, PITPα-deficient platelets form thrombi normally at sites of intravascular injury. One phenotype of note is that PITPα-depleted platelets are defective in fibrin deposition on injected melanoma cells with the result that fewer lung metastases are formed [127].

Interestingly, whereas PITPβ is not expressed quite as highly as is PITPα in platelets, PITPβ-deficient platelets show phenotypes that resemble those of PITPα-deficient cells. While it is concluded that PITPα/PITPβ share functional redundancy in platelet phosphoinositide signaling, the striking result is that PITPα-deficient platelets show very little cytosolic PtdIns transfer protein activity whereas PITPβ-deficient platelet cytosol shows essentially no reduction in PtdIns transfer activity at all relative to wild-type platelets [253]. On the basis of these data, Zhao and coworkers conclude that the signaling function of platelet PITPα/PITPβ is independent of simple PtdIns transfer between membranes.

A third potential context is retinal signaling. Zebrafish PITPβ shows highest expression in double cone photoreceptor cells, and PITPβ knockdown compromises development and maintenance of retinal double cone photoreceptor cell outer segments [87]. It is not clear whether this specificity of phenotype might be a simple consequence of differential expression of PITPβ and PITPα in double cone cells. In humans, transcriptome analyses of retinal pigment epithelium (RPE) identify PITPNA as a gene of interest in age-related macular degeneration (AMD). Global expression profiling of human fetal and adult RPE cells identified a highly expressed ‘signature’ RPE gene set. As RPE dysfunction is a signature observed in early-stage AMD, the ‘signature’ genes cross-referenced against single nucleotide polymorphisms (SNPs) from AMD and control patients. PITPα was highlighted as one of the four most significant associations – suggesting PITPα deficiencies in RPE might contribute to AMD [128]. The retinal context will be revisited in discussion of the type 2 PITPs of which the Drosophila retinal degeneration Bα (RdgBα) protein is the prototype.

1.10. PITPα and PITPβ in other syndromes

Both PITPα and PITPβ are linked to a number of syndromes based on transcriptomic or regional gene deletion data with no direct evidence that these proteins directly contribute to the disease state. We do not consider these associations in this review. Rather we invest focus on those contexts where a convincing functional case can be made.

The involvement of PITPα and PITPβ development of the embryonic neocortex is a particularly interesting one. This tissue expands dramatically during embryonic development and the contour of this expansion, and subsequently of the adult neocortex, is specified by formation of a thin pseudostratified neural stem cell (NSC) epithelium. This expansion is asymmetric as it favors the tangential (lateral) vs the radial dimension – resulting in a thinning of the tissue. PITPα/PITPβ regulate this process via: (i) control of efficient trafficking from the TGN to the plasma membrane of specific receptors of the non-canonical Wnt planar cell polarity (ncPCP) signaling pathway, and (ii) interkinetic nuclear migration (IKNM) during the NSC cell cycle that drives a specific form of convergent extension that promotes formation of a thin neocortex with a large surface area ([109]; Fig. 4D). In terms of human pathologies, defects in neocortical thinning during development are associated with various diseases of intellectual capacity.

Screening of ethyl-N-nitrosourea (ENU) mutagenesis of mouse germline stem cells identify an involvement of PITPα deficiencies in erectile dysfunction (ED) [129]. Mice homozygous for the recessive PITPα substitution (Gly172Arg) are viable mice and display neurological and liver dysfunction, hypoglycemia, gait abnormality and weight loss. These phenotypes broadly resemble milder versions of those documented for null mice. Unlike PITPNA null mice however, the ENU-induced PITPα mutants were still alive at 12 months and exhibited a protruded genitalia phenotype (PGP) that included inducible priapism – a disorder defined by prolonged and involuntary erection in the absence of sexual stimulation that likely involves activated nitric oxide signaling [129].

Strong linkage of PITPα deficiency to Duchenne muscular dystrophy (DMD) is also reported. DMD is a severe genetic disorder described by muscular dystrophin deficiency associated with progressive muscle wasting disease, degradation of skeletal- and cardiac function, cardio-respiratory failure, and premature death. PITPα is reported as a protective genetic modifier of DMD in studies employing the Brazilian Golden retriever muscular dystrophy (GRMD) model [130]. No such genetic interaction is observed in a Labrador retriever muscular dystrophy model, however [131]. Consistent with results of the GRMD model, PITPα deficiencies alleviate DMD in the dystrophin-deficient sapje zebrafish DMD model where partial PITPα knockdown improved overall survival, muscle structure and swim test performance [130]. Furthermore, phosphodiesterase 10 A (PDE10 A) inhibition reduced the DMD phenotype in dystrophin-deficient sapje-like zebrafish larvae, with concomitant reduction of PITPα expression [132]. PDE10 A inhibition in DMD patient-derived myoblasts (myocyte/muscle cell precursors) is also associated with reduced PITPα expression [132], and reduced PITPα is associated with myoblast differentiation and improved myoblast fusion in patient derived DMD human muscle cells [130]. In all models, disease severity correlated with reduced Akt Ser473 phosphorylation, and increased PTEN (Phosphatase and tensin homolog) levels. Since loss of PITPα expression consistently increased Ser473 phosphorylation of Akt and reduced levels of the PTEN phosphatase in DMD models, PITPα is speculated to act as a negative regulator of PtdIns (3,4,5)P₃ and Akt signaling via modulation of PTEN-mediated degradation of PtdIns(3,4,5)P₃ and reduced Akt signaling.

1.11. PITPβ in viral infection

An interesting aspect of PITPβ biology concerns replication of positive strand RNA viruses of the Picornaviridae family [133]. PtdIns4P synthesis occurs at specialized viral RNA replication organelles formed from host Golgi and ER membranes. Aichivirus A (AiV) is a causative agent of acute gastroenteritis in humans. This virus commandeers the host oxysterol binding protein (OSBP) to hijack cholesterol transport from the ER to the replication organelle. Deletion of host VAPA/ VAPB (Vesicle-associated membrane protein (VAMP)-associated proteins A/B), the Sac1 PtdIns4P phosphatase, or PITPβ resulted in loss of viral replication. These proteins are recruited to the replication organelle and are posited to participate in a cholesterol transport pathway [133]. Similarly, host OSBP1, Sac1 and PITPβ are essential for rhinovirus replication [134]. Current models interpret the data to indicate the virus hijacks counter-current flow of lipids between the Golgi and ER to move lipid to the replication organelle and thereby hijack the cholesterol required for viral replication [51,134]. This speculative model involves: 1) transport of PtdIns to the Golgi by PITPβ, 2) conversion of PtdIns to PtdIns4P, 3) OSBP mediated cholesterol transport to the Golgi driven by PtdIns4P transfer to the ER, 4) Conversion of PtdIns4P at the ER back to PtdIns by Sac1.

1.12. PITPβ in Alzheimer’s disease

Deep proteomic analyses stratifying patient risk identify increased PITPβ expression as a predictive marker for asymptomatic Alzheimer’s disease (AsymAD). The search for predictive markers is motivated by the fact that AsymAD individuals can display brain changes characteristic of Alzheimer’s disease for up to 20 years before displaying perturbed cognitive behavior. These analyses combined proteomic data from 6 large-scale brain tissue proteomic studies (5 from the dorsolateral pre-frontal cortex and 1 from the temporal cortex) that involved 620 subjects and queried >3000 proteins with label-free quantification and machine learning [135].

1.13. Cancer biology of PITPα and PITPβ

Although we are not aware of any studies to date that identify cancer driver mutations in PITPNA or PITPNB, there is evidence to implicate involvements of both PITPα and PITPβ in issues related to cancer. One example was discussed above where a PITPα/β natural product inhibitor microcolin B, or its derivative VT01454 killed Yap-addicted uveal melanoma (UM) 92.1 cells by activating the Hippo pathway. As another example, PITPα is identified as one of 28 genes highly predictive for human EGFR (HER2)-positive metastatic breast cancer resistance to Trastuzumab-docetaxel therapy. Trastuzumab (trade name Herceptin) is a humanized monoclonal antibody that binds a HER2 extracellular domain and inhibits signaling by promoting receptor dephosphorylation and/or blocking dimerization [136]. The result of this inhibition is diminished PI3K/Akt pathway signaling [137,138], and induction of the cyclin-dependent kinase (CDK) inhibitor p27kip1 [139]. How direct a role upregulated PITPα plays in Trastuzumab resistance remains to be established. Also, a proteomic study of pre-B2 lymphoblasts identifies elevated PITPβ as an indicator of good prognosis in children with acute lymphoblastic leukemia (ALL) associated with chromosomal translocation (12; 21) [140]. What this means mechanistically is unclear.

The evidence for an involvement for PITPα/ PITPβ in gastric cancer is more convincingly established. Defects in both PITPs result in elevated cytotoxic/nonprotective autophagic activity and much of the evidence comes from use of microcolin H as an antiproliferative compound in multiple tumor cell lines. Microcolin H is a cytotoxic lipopeptide from the marine cyanobacterium Moorea producens that binds and inhibits PITPα/β. Moreover, microcolin H displays antitumor efficacy in a nude mouse xenograft tumor model with low toxicity – suggesting promise as cancer therapeutic [141]. In that regard, Microcolin H is a deacetylated version of the Hippo pathway inhibitor VT01454 discussed above. Whether its production at scale will prove problematic (as for VT01454), and whether chronic treatment with this compound will exert undesired toxicities, remains to be determined.

Yet another PITPα involvement in cancer biology is demonstrated by the observation that mice lacking PITPα in platelets are protected from lung metastasis [127]. Platelets contribute to cancer outcome by adhering to tumor cells and shielding them from detection by immune cells. PITPα deficient platelets are defective in fibrin deposition and thrombin generation [127]. This is postulated to be a consequence of an inability of the platelets to expose phosphatidylserine (PtdSer) on the cell surface through the action of TMEM16F (Transmembrane protein 16F) [51], a Ca²⁺-activated ion channel / phospholipid “scramblase” located on the extracellular surface [142].

A topic beyond the scope of this review is the biological role of the long non-coding RNA PITPNA-AS1 that is transcribed from the opposite strand as PITPNA. What relationship, if any, is there between PITPα function and PITPNA-AS1 expression is unknown. But the fact that PITPNA-AS1 contributions are strongly implicated in a number of cancers identifies this as an important question in contemporary cancer research [143–149,252].

1.14. The class IIb PITPnc1

The class IIb PITPnc1 (retinal degeneration type B/RdgBβ) resembles PITPα and PITPβ in that it is a soluble PITP of ~32–38 kDa MW. It is distinguished from the other two PITPs by the fact that it is not a PtdIns/PtdCho exchange protein in vitro. Rather, PITPnc1 is reported a PtdIns/PtdOH exchange protein. That is, exactly the type of lipid transfer activity that Robert Michell’s PtdIns-cycle hypothesis (described above) posited to be an essential component of phosphoinositide signaling. PITPnc1 is expressed as two spliceoforms in mammals that differ in the length of the C-terminal disordered tail that follows an otherwise canonical StART-domain. Splice variant 1 (PITPnc-sp1) represents the longer 332 amino acid variant with a C-terminal 60 amino acid extension following the PITP-domain whereas variant 2 (PITPnc-sp2) is only 268 amino acids long [150]. PITPnc1 is ubiquitously expressed in mammals with elevated expression in heart, muscle, kidney, liver and peripheral blood, and lower levels in brain, eye, spleen, small intestine, lungs, and leukocytes [66]. The Human Protein Atlas GTEx (Genotype-Tissue Expression) data (RNA-seq data on RSEMv1.2.22 (v7)) extend the expression profiling to indicate PITPnc1 is most highly expressed in adipose tissue followed by basal ganglia, breast, lung, and cerebral cortex [151]. The splice variants likely execute specific functions as these display distinct expression patterns. PITPNC-sp1 is expressed in murine heart and developing brain, being particularly enriched in dentate gyrus, thalamus and cerebellar Purkinje cells [150,152]. The variants also display differential protein interactions. PITPNC-sp1 binds 14–3–3 proteins following C-terminal phosphorylation at Ser²⁷⁴/ Ser²⁹⁹, and this interaction stabilizes PITPnc-sp1 by reducing degradation that is otherwise promoted by PEST sequences [150]. The PITP-domain itself is also reported to interact with angiotensin II type I receptor-associated protein (ATRAP) in a manner that promotes PITPnc1 membrane-recruitment [150]. This gives rise to speculation for a PITPnc1 role in heart tissue during angiotensin II signaling and blood pressure control [67,153], but there is no evidence to support those ideas at present.

From an organismal perspective, both PITPnc1 null mice ([154]; Bankaitis lab unpublished data) and zebrafish are fully viable animals and the zebrafish studies link PITPnc1 to circadian rhythm regulation [155]. Zebrafish also express two PITPnc1 isoforms, a longer 331 amino acid PITPnc1a that exhibits 81 % identity and 90 % similarity to human PITPnc-sp1, and a shorter 305aa PITPnc1b that is likely orthologous to human PITPNC1-sp2 [155]. PITPNC1 null zebrafish display aberrant increased neuronal activity in arousal related circuits and increased wakefulness during the day/night cycle. The neuronal and behavioral hyperactivities are rescued by inhibition of insulin growth factor signaling [155]. Furthermore, PITPnc1 deficits increase both brain insulin-like growth factor receptor (IGFR) signaling, and brain growth [155]. Thus, contrary to how PITPs are often claimed to function in stimulating receptor signaling [92–94], PITPnc1 deficits enhance such a process.

High level expression of PITPNC1 in mouse adipose tissue reflects its role in promoting non-shivering thermogenesis [151,154]. Consistent with this phenotype, brown adipocytes derived from PITPnc1-deficient adipocytes are defective in fatty acid oxidation. In that regard, PITPNC1 expression is repressed by the long noncoding RNA (lncRNA) HOTAIR (HOX Transcript Antisense Intergenic RNA) – an important epigenetic regulator with fat depot-specific expression [151].

1.15. PITPnc1 and cancer

A growing body of evidence links PITPnc1 to cancer. The first indication came from studies indicating it is encoded within a chromosomal region amplified in breast cancer, associated with poor prognosis, lymph node metastases, larger tumor size, and higher histological grade [156]. The correlation of elevated tumor expression of PITPNC1 with reduced patient outcomes is apparent in gastric cancer [157] and in lung-, and pancreatic ductal adenocarcinomas (LUAD and PDAC) [158]. More direct association of PITPnc1 with cancer derives from the finding that PITPNC1 is the target of miR-126, an anti-oncogenic microRNA implicated in inhibition of breast cancer cell angiogenesis and lung metastasis, which downregulates PITPNC1 in some but not all tissue types [159,160]. Moreover, epigenetic aberrations associated with PITPNC1 are linked to sporadic breast cancer, and hypermethylation of the PITPNC1 enhancer (a distant regulatory element) is promoted by pterostilbene (PTS). This dietary polyphenol, and natural resveratrol analog, is abundant in foods with reported anti-cancer effects. PTS-binding blocks binding of the PITPNC1 enhancer by the pro-oncogenic transcription factor OCT1 with resulting downregulation of PITPNC1 expression [161].

Elevated PITPNC1 levels are observed in primary tumors that undergo metastasis [156] and in the secondary metastases [156,160]. Cell studies provide data consistent with these findings. PITPNC1 expression is elevated in highly metastatic cell lines and murine tail vein injection experiments in nude mice indicate lung colonization is inhibited following PITPNC1 knockdown and enhanced upon over-expression [156,158,160,162]. Pro-oncogenic activities for PITPNC1 are also observed in xenograft studies with LUAD and PDAC cells [158]. The relevance of such studies to true cancer metastases in immunocompetent mice and patients remains to be seen,

The precise mechanisms through which PITPNC1 promotes cancer remains unknown as PITPnc1-deficient cells present complex phenotypes in culture. Such studies associate PITPNC1 with enhanced trafficking from the Golgi complex of matrix metalloproteases and other factors that enhance cell proliferation, cell motility, extracellular matrix remodeling, angiogenesis, and metastasis [156] (Fig. 5A). Levels of both the GOLPH3 oncoprotein and of PITPnc1 are elevated in aggressive metastatic tumors, and PITPnc1-associated membrane trafficking correlates with GOLPH3 recruitment to the TGN. A model is proposed where PITPnc1 binds PtdIns4P on TGN membranes, recruits the Rab1 small GTPase and 14–3–3 proteins to the TGN, and thereby drives enhanced PtdIns4P production. This results in recruitment of GOLPH3 to those membranes where GOLPH3 in turn recruits the noncanonical actomyosin MYO18A. This recruitment is envisioned to impose a tensile force onto TGN membranes required to promote vesicle release [156,163,164] (Fig. 5A). The observation that acute myeloid leukemia (AML) patient data associate elevated serum levels of PITPnc1, GOLPH3, Rab1B, and Myo18A with reduced patient survival are consistent with this rather detailed model [165].

Fig. 5.

Role of class IIb PITPnc1 in human diseases. (A) PITPnc1 is proposed to enhance secretion of metastasis factors in breast cancer through recruitment of GOLPH3 and Rab1b to trans Golgi network (TGN). PITPnc1 expression is silenced by ‘anti-metastatic’ micro-RNA miR-126 resulting in angiogenesis and tumor metastasis. PITPnc1 exchanges PtdIns for PtdOH and stimulates PI4K activity to produce an expanded TGN PtdIns4P pool. PtdIns4P mediates GOLPH3 and Rab1b recruitment to the cancer cell TGN, resulting in enhanced secretion of pro-metastatic factors. These include growth factors and matrix metalloproteases [156]. (B) PITPnc1 is identified as a downstream effector of KRAS in lung and pancreatic ductal adenocarcinoma (LUAD and PDAC). A PITPnc1 regulated signature links KRAS to MYC stability thereby regulating autophagy. PITPnc1 depletion inhibits lung and pancreatic cancer cell proliferation in vitro and tumor proliferation in vivo (i.e. lung colonization and liver metastasis) [158]. PITPnc1 inhibition results in Myc downregulation which in turn perturbs mTORC1 localization and promotes autophagy. JAK2 and KRASG12C inhibitors identified as surrogate PITPnc1 inhibitors show anti-tumor effects in LUAD and PDAC.

Recent work challenges the idea that a GOLPH3/Myo18A interaction is required for fission of TGN vesicles involved in bulk membrane trafficking, and it does so on multiple grounds [166]. First, Myo18A exists as two splice variants with Myo18Aα containing the N-terminal PDZ domain that is postulated to mediate binding to GOLPH3 [251]. In contrast to a previous report [163], detailed imaging analyses of both endogenous- and over-expressed Myo18Aα failed to detect any localization of this motor protein to the Golgi system in multiple mammalian cell lines. Second, loss of Myo18Aα had no significant effect on Golgi-morphology or dynamics in transformed cell lines [166]. Third, multiple lines of evidence indicate Myo18Aα is not an active myosin motor and it does not exhibit a GOLPH3-stimulated ATPase activity [166–169]. Fourth, whereas GOLPH3 and Myo18A show phenotypes consistent with these proteins being downstream effectors of PITPα/β-dependent PtdIns4P signaling with regard to Golgi organization in NSCs, these phenotypes are not associated with defects in bulk membrane trafficking from the Golgi complex ([109,110]; see above).

PtdIns is an upstream precursor for PtdIns (3,4,5)-trisphosphate (PIP3) synthesis, and PI3K-mediated oncogenesis. Studies of PITPNC1 null mutants in zebrafish identify some role for PITPnc1 in regulation of PI3K/Akt signaling [155]. Links between this pathway and PITPnc1 also emerge from mammalian LUAD, and PDAC cancer cell studies. Pharmacological inhibition of the MAP kinase components MEK1/2 and JNK1/2 downregulates PITPnc1 – suggestive of regulation by the K-Ras pathway given the crosstalk between K-Ras and PI3K/Akt signaling pathways [158,170–172]. The observation that PITPnc1 activity suppresses autophagy in LUAD and PDAC cells via c-Myc regulated localization of mTOR suggests active PITPnc1 inhibits mTOR regulated autophagy and promotes cell proliferation. PITPnc1 deficiencies downregulate Myc and mTOR with the result that cell cycle progression is inhibited (G1 arrest with decreased S phase) and autophagy is promoted [158] (Fig. 5B).

PITPnc1 exhibits the biochemical properties of the PITP activity postulated by Robert Michell to chaperone the intermembrane lipid trafficking required for function of the PtdIns-cycle [21]. Although the in vivo data do not support such a role in primary cells, the question of whether PITPnc1 regulates PtdOH signaling remains an open one. PtdOH is a bioactive glycerophospholipid with roles in cell growth, differentiation, and migration, and pathologic processes including cancer (reviewed [173]). This phospholipid is generated by multiple pathways including de novo synthesis, the phospholipase C (PLC)/diacylglycerol (DAG) kinase pathway, and the action of phospholipase D (PLD). Attempts to establish which pathway provides the PtdOH destined for PITPnc1 interaction remains unclear. The fact that PITPnc1 preferentially binds the C16:0/16:1 and C16:1/18:1 molecular species of PtdOH [67], suggests PtdOH pools produced from phosphatidylcholine (PtdCho) via activity of the PLD pathway [174,175]. In support, aberrant PLD/PtdOH signaling is observed in numerous human cancers (reviewed [173]). Speculations that PITPnc1 transfers PtdOH from plasma membrane to the ER to drive de novo PtdIns is not easily reconciled with in vitro studies suggesting that PtdOH produced by the PLD pathway (unlike that produced by PLC) is not converted to PtdIns but is cycled between PtdOH and DAG [176].

PtdOH influences membrane fusion and fission events and could potentially represent a foundation for enhanced secretory function in PITPnc1 overexpressing cells. As a lipid with a headgroup that exhibits a small axial diameter, PtdOH induces negative membrane curvature. It also is a metabolic precursor of lipids (lysophosphatidic acid and diacylglycerol) that bind proteins required for membrane fusion and fission (reviewed [177]). Conversion into lysophosphatidic acid (LPA) via the action of phospholipases A1 and A2 is of particular interest as pro-oncogenic roles for LPA are linked to virtually every aspect of cancer biology (reviewed [178]). PtdOH also functions as a docking site for recruitment of pro-oncogenic effector proteins. These include the guanine nucleotide-exchange factor Son of sevenless (Sos), the rate-limiting step in linkage of receptor tyrosine kinases to Ras-triggered intracellular signaling pathways [179], the serine–threonine kinase Raf-1 [180], and the NADPH oxidase component p47^phox [181]. The latter is of interest given that it is a ROS producing enzyme and PITPnc1 confers radiotherapy resistance in rectal cancer via inhibition of ROS generation [182].

The PITPnc1-associated radiotherapy resistance phenotype is complex. Recent data indicate PITPnc1 suppresses CD8⁺ T cell immune function and promotes radiotherapy resistance in rectal cancer by modulating fatty acid synthase (FASN). FASN expression correlates with tumor promotion and radiotherapy resistance and is considered a therapeutic target for many cancers [183]. A link to fatty acid metabolism is also observed in brown adipocytes harvested from PITPnc1 null mice where acute cold exposures report defects in non-shivering thermogenesis [154]. In support, PITPnc1 is associated with increased fatty acid uptake, oxidation and ATP-production and elicits anoikis resistance in gastric cancer cells. These features correlate with upregulation of PPARγ (peroxisome proliferator-activated receptor), of CPT1B (carnitine palmitoyltransferase 1B a protein involved in fatty acid transport into mitochondria), of CD36 (a widely expressed type 2 cell surface scavenger receptor/ fatty acid transporter), and of the sterol regulatory element-binding protein 1 (SREBP1) transcription factor. Moreover, metastatic nodule formation following orthotopic injection of PITPNC1 over-expressing cells in an omental metastatic nude mice model was decreased by pharmacological inhibition of fatty acid oxidation [157].

1.16. Multidomain PITPs – focus on Drosophila RdgBα

In addition to small soluble START-like PITPs, multidomain versions of these PITPs also exist and are classified as the class IIa START-like PITPs. The founding member of this family, and in many respects the most thoroughly studied one, is the Drosophila PITP retinal degeneration B protein (RdgBα). Mammalian PITPnm1 (Nir2), PITPnm2 (Nir3), and PITPnm3 (Nir1) are homologs of the fly RdgBα. Other versions include the class IIa C. elegans PITP-1 and the as yet uncharacterized STAMP protein of Toxoplasma gondii. RdgBα was identified in a Drosophila screen identifying genes required for photoreceptor function [250]. Mutant rdgBα flies display photoreceptor cell degeneration several days after eclosion/ hatching [184,185], and also exhibit olfaction defects [186]. While the retinal phenotype manifests itself slowly in dark-reared flies, its appearance is highly accelerated in the presence of light. The mutant flies display abnormal electroretinogram (ERG) responses that rapidly deteriorate following light exposure. As loss of function occurs prior to obvious physical signs of retinal degeneration, it was suggested that the light response precipitates degeneration [184,185]. Light dependent retinal degeneration is prevented by calcium depletion or inactivation of IP₃ signaling pathways downstream of rhodopsin activation, including PLC and PKC. Further support for PLC pathway involvement is provided by the identification of a mutant in the fly DAG kinase (rdgA), in the original genetic screen that identified rdgBα [187,188]. Since phosphoinositide signaling pathways transmit light signals following stimulation of rhabdomere G protein-coupled receptors, fly phototransduction represents a high-demand phosphoinositide turn-over/resynthesis circuit. As such, it provides a facile physiological model from which to study multidomain PITP function.

The multidomain RdgBα is a large 116 kDa peripheral membrane protein [69,189]. It localizes to both the axon and to a subcompartment of the ER termed the subrhabdomeric cisternae (SRC) or the submicrovillar cisternae (SMC) which lie adjacent to the microvillar plasma membrane [70,190]. Since this subcellular region resembles ER–PM contact sites, it is proposed to participate in PtdIns transfer from the ER to the plasma membrane and fuel the PIP cycle as envisioned in the Michell hypothesis [70,191] (Fig. 6). Furthermore, characterization of RdgBα as a PtdOH-binding protein suggests the possibility that it could also be involved in PtdOH transfer back to the ER [192]. Several lines of evidence are consistent with such roles. For example, rdgBα flies are deficient in restoration of PtdIns(4,5)P₂ levels in photoreceptor cells following light stimulation. In addition, basal PtdIns(4,5)P₂ levels are reduced in rdgBα mutant retina and retinal PtdOH levels are elevated following light stimulation – suggesting downstream roles in PtdOH metabolism [192]. However, other evidence indicates greater complexities (reviewed [61]). For example, rdgBα mutants respond competently to the first light pulse but are unable to terminate signaling and repolarize after this first stimulation [53,193]. Moreover, rhodopsin levels are downregulated in the mutant – perhaps as an adaptive response to elevated PLC signaling at the plasma membrane.

Fig. 6.

The role of RdgBα protein in Drosophila melanogaster. Schematic representation of the phospholipase C (PLC) signaling pathway and the PtdIns(4,5)P₂ cycle between microvillar plasma membrane and submicrovillar cisternae (SMC) in fly photoreceptor cells. RdgBα is posited to function in transfer of PtdIns from SMC to plasma membrane and PtdOH from plasma membrane back to the SMC. The lipid transfer cycle is proposed to occur at ER-PM contact sites. The PITP domain is both necessary and sufficient to support the cycle.

A particularly interesting aspect of RdgBα activity rests with what protein domains play critical roles in physiological function. The popular interpretation of class IIa PITPs is that these proteins serve as bridges at ER-plasma membrane (PM) contact sites. The construction of the protein itself conforms to this general idea as: (i) the PITP domains of RdgBα represents only ~25 % of the total primary sequence, (ii) RdgBα contains an FFAT (diphenylalanine (FF) in an acidic tract) motif that binds the ER-localized VapA protein that serves as the fly FFAT receptor, (iii) a potential metal binding DHDD domain, and (iv) a Lipin/ Nde1 /Smp2 (LNS2) PtdOH binding domain. Further support for the concept is provided by the demonstration that VapA mutants phenocopy rdgBα mutants, and mutations in the rdgBα FFAT motif inactivate RdgBα in vivo [192]. However, VapA binds numerous proteins so its deficiency phenotype cannot be strictly attributed to loss of RdgBα membrane bridging activity. It is also important to note that the rdgBα FFAT mutants are somehow inactivated for PITP activity – potentially via a trivial mechanism involving protein folding deficits – and show only partial localization defects [192,194]. Thus, rdgBα FFAT mutant phenotypes are not subject to simple interpretation given their effects on PITP domain activity.

The most striking results are observed upon germline expression of a soluble version of the RdgBα PITP domain alone (i.e. comprising only 27 % of the protein and lacking the other domains proposed to be involved in functions such as PtdIns4P-binding, membrane-anchoring, etc.) in rdgBα mutant flies. Expression of the PITP domain: (i) restores the light response and (ii) protects flies from retinal degeneration in ambient light even after 30 days post-eclosion [53]. Although this unambiguous result is largely ignored in discussions of class IIa PITP function, it has been reproduced [192]. Moreover, this result is consistent with findings from other class IIa PITP models where such experiments have been performed (C. elegans PITP-1; see below). While a recent study reports that expression of the isolated PITP domain does not fully restore the light response in rdgB flies, this deficit is observed only under conditions of very high intensity light. Moreover, a modest 2-fold enhancement in PITP-domain expression relative to full-length RdgBα expression rescues the light response even under those conditions [191,194]. Mammalian PITPα expression fails to rescue the rdgBα fly phenotype [53,192]. This failure likely reflects the fact that PITPα is a PtdIns/PtdCho exchange protein, whereas the RdgBα PITP domain exhibits PtdIns/PtdOH exchange activity.

With regard to new insights into RdgB function in fly phototransduction, suppressor genetic approaches hold promise. A recent such screen reveals multiple mechanisms by which the phototransduction defect of rdgB flies can be alleviated. Interestingly, one suppressor class results from inactivation of the CERT ceramide transfer protein [195]. The mammalian CERT is required for sphingomyelin synthesis in trans-Golgi membranes and, in those systems, transfers ceramide from ER to the trans-Golgi network in a manner directed by CERT binding to a pool of trans-Golgi PtdIns4P pools [196].

1.17. Multidomain PITPs – focus on the mammalian class IIa proteins

As indicated above, mammals express three proteins that share homologies with the fly RdgBα – PITPnm1 (Nir2), PITPnm2 (Nir3), and PITPnm3 (Nir1). PITPnm1 and PITPnm2 have N-terminal PITP domains and share ~40 % amino acid sequence identity with PITPα/β [66,189]. PITPNM1/ 2 are products of a gene duplication that occurred early in the vertebrate ancestry with PITPNM1 lost in fish and many amphibians [197]. While both proteins appear to be ubiquitously expressed, PITPnm2 is generally less highly expressed relative to PITPnm1. The exception is thymus where PITPnm2 expression is 15-fold higher than expression in PITPnm1 and 5-fold higher than the highest levels of PITPnm1 in any tissue [249]. PITPnm2 localizes to cytosol and, like PITPnm1, translocates to ER-PM [198] and ER-phagosome contact sites [199]. PITPnm3, while lacking a PITP domain, is proposed to have a scaffold function that promotes PITPnm1 recruitment to ER-PM junctions [200].

PITPnm1 is implicated in TGN membrane fission and vesicle biogenesis as its functional ablation results in cargo retention in Golgi tubules accompanied by swelling and dispersal of Golgi cisternae. Loss of PITPnm1 activity is associated with decreased DAG levels – possibly due to increased consumption through the CDP-choline pathway for PtdCho biosynthesis [201]. This proposed signaling circuity recapitulates the metabolic crosstalk between the yeast Sec14 PITP and the CDP-choline pathway [23,24,202,248]. However, while generation of localized PtdIns4P pools is key to Sec14 function, the role of PtdIns4P in PITPnm1-mediated Golgi trafficking is suggested to be secondary, and perhaps nonessential. The evidence to that effect is based on the inability of PITPnm1 knockdown to displace the TGN-localization of a single PtdIns4P probe (PHOSBP) [201].

In terms of physiological function, PITPnm1/−2 are implicated in regulation of voltage-gated potassium (Kv2) channels and to interact directly with Kv2.1 complexes [203]. Kv2.1, PITPnm1, and VAP colocalize at ER-PM junctions in cultured neurons, and the brains of Kv2.1-null mice display altered PtdIns composition. On this basis, it is suggested that ER-plasma membrane junctions formed by Kv2 channel-VAP-PITPnm1 interactions regulate PtdIns homeostasis [203]. PITPNM1–3 are also under consideration as candidate genes for human retinal disease [204,205]. Both PITPnm1 and Dm-RdgBα exhibit PtdIns/PtdOH exchange activity and expression of mammalian PITPnm1 (but not PITPnm2) partially rescues the phenotype of RdgBα mutant flies. Moreover, metabolic chase and live cell imaging experiments demonstrate that PtdOH accumulates at the PM and fails to be converted to CDP-DAG in the ER in PITPnm1-deficient cells [206]. PITPnm1 is posited to drive counter-transport of PtdIns and PtdOH between ER and plasma membrane [198,206–209].

The model invokes coordinated action of multiple protein domains as follows: PITPnm1 localizes to ER through FFAT motif-mediated association with ER-resident VAP proteins [210]. PLC-mediated PtdIns(4,5)P₂ hydrolysis results in PM accumulation of DAG and PtdOH that is envisioned to drive PtdIns-loaded PITPnm1 translocation to ER-PM contacts via interaction of the PtdOH-binding LNS2 domain. The PM-interaction is supported by DGBL domain binding to DAG [206,208,209]. It is at this stage that PITPnm1-mediated PtdOH/PtdIns exchange resupplies the PM with PtdIns for replenishment of the PM PtdIns(4,5)P₂ pool. PtdOH-loaded PITPnm1 returns the phospholipid back to the ER – thereby supporting PtdIns resynthesis [198,206,211].

1.18. PITPnm1 and PITPnm2 function in the mammal

The current understanding of PITPnm1 function is derived from the cell line studies described above, which are typically performed under conditions of intense stimulation of PLC signaling. Functional studies in mice provide an interesting comparison. Broadly consistent with the important functions attributed to PITPnm1 in studies with cell lines, PITPNM1 null mutations were initially described as embryonic lethal [212]. However, a later study reported PITPNM1 null mice to be viable, fertile and effectively normal – even though PITPnm1 is ubiquitously expressed and represents the most highly expressed member of the multidomain PITPs. Only minor reductions in plasma cholesterol and sex-dependent alterations in circulating leukocyte counts were recorded in PITPNM1 null animals [213].

Mouse model studies do suggest class IIa proteins will have some involvement in sensory perception. PITPnm1 is expressed from late embryonic stages until adulthood within the inner hair cells of the organ of Corti, and is transiently expressed during early postnatal stages within outer hair cells. Moreover, PITPNM1 is dysregulated in murine progressive hearing-loss models caused by mutation of the microRNA (miRNA) miR-96 [214], and in conditional knockouts of Otx2 – a protein involved in head and sense organ developmental patterning [215]. PITPNM1 is also identified as a target for the miRNA miR-490–5p during cartilage development and osteoarthritis (OA) [216]. Loss- and gain-of-function experiments show miR-490–5p targets PITPNM1 with the result of attenuating chondrogenesis and accelerating cartilage degradation via activation of PI3K/Akt signaling. This may be of medical relevance as miR-490–5p inhibition alleviates cartilage injury in an in vivo model of destabilization of the medial meniscus [216]. However, as miRNAs exhibit multiple targets, it remains an open question whether PITPNM1 dysregulation in these contexts is causal or a more peripheral factor. With regard to hearing loss, it is likely the latter case as PITPNM1 null mice do not experience hearing deficits [213].

PITPNM2 null mice are also viable, fertile and rather normal although deficits in dim light responses and failures to transduce light input from rods to bipolar retinal cells are described [217]. The PITPnm2 LSN2 domain is reported to be more sensitive to plasma membrane PtdOH levels than the PITPnm1 LSN2 domain – leading to the idea that PITPnm2 binds membrane contact sites under basal conditions to sustain PtdIns(4,5)P₂ in resting cells, while PITPnm1 is mobilized following PLC activation [198].

More is known about PITPnm2 association with immune system function. PITPNM2 single nucleotide polymorphisms show significant gene associations with immune-related diseases such as asthma, eczema, multiple sclerosis and type II diabetes [197]. PITPnm2 is also linked to T-cell development and survival and might contribute to differential T-cell receptor signaling between preselection thymocytes and mature T-cells [197]. PITPNM2 transcripts are highly enriched in thymus, the site of T-cell development, and mice lacking PITPNM2 display reduced populations of CD4 single positive thymocytes with fewer cells passing developmental checkpoints. T-cell lineage is also affected as the mutants exhibit fewer regulatory and natural killer T-cells. PITPNM2 null mice suffer lymphopenia, characterized by reduced lymphocyte numbers in lymph nodes and spleen, a reduced T-cell complement, and fewer effector- and central memory-like cells. The T-cell defects are attributed to defective PtdIns supply from ER to ER-PM junctions based on slower PM PtdIns(4,5)P₂ replenishment as estimated by Tubby-mScarlet localization [197].

Regarding functional redundancy, cell line studies show PITPnm1/−2 maintain PtdIns(4,5)P₂ homeostasis at phagocytic cups and support the actin contractility that drives phagosome closure. PITPnm2, and to a lesser extent PITPnm1, accumulate at phagosome-ER contact sites and PITPnm1/−2 knockdown is reported to decrease PM PtdIns(4,5)P₂ and receptor-mediated phagocytosis with the result that particle capture is stalled, contractile actin ring density is reduced and phagosome closure is aborted [199]. To our knowledge, PITPnm1/PITPnm2 functional redundancy has not been tested in the mouse.

1.19. PITPnm1 and viral infection

With regard to infectious disease, PITPnm1 and the ER-resident proteins VAP-A/-B are associated with efficient replication of hepatitis C virus (HCV; [218]). VAPs are proposed to recruit proteins involved in lipid exchange to the ER. This recruitment includes the oxysterol binding protein (OSBP) proposed to deliver cholesterol to the HCV replication organelle in exchange for PtdIns4P. The hypothesized pathway involves PITPnm1-mediated resupply of PtdIns to the replication organelle to maintain elevated levels of PtdIns4P. In this model, PITPnm1 acts in opposition to OSBP with PITPnm1, VAPs, OSBP, and PtdIns 4-OH kinase comprising a cycle of phosphoinositide flow between the ER and viral replication organelles [218]. A similar role is posited for PITPnm1 in replication of SARS-CoV-2 [218,219]. Recent data suggest that the SARS-CoV-2 capsid engages in functional mimicry of proteins found within the central nervous system including PITPnm1. Autoimmune responses linked to viral mimicry are proposed to induce multiple sclerosis (MS) and, as a protein linked to MS, it is speculated that PITPnm1 contributes to MS pathology [220].

1.20. Mammalian class IIa PITPs and cancer

National Cancer Institute Genomic Data Commons portal data (https://portal.gdc.cancer.gov) suggest PITPnm proteins are more frequently mutated in cancers than other PITPs. PITPnm1 is a prognostic indicator in breast cancer as elevation in cancer tissue is associated with poor prognosis. PITPnm1 promotes cancer cell migration and invasion [221], and PITPNM1 silencing reduces both cell proliferation and colony formation [222]. This PITP operates through modulation of PI3-kinase/Akt and ERK1/ERK2 pathways to shift the epithelial–mesenchymal-transition into motile and invasive states with enhanced metastatic potential [223]. Also, PITPnm proteins might impact cancer biology via immune system regulation. Elevated PITPnm1 is associated with enriched T-cell activation, differentiation, and infiltration. As regulatory T-cells influence the tumor micro-environment by migrating into tumors and suppressing lymphocyte antitumor activity, it is an interesting proposition that PITPnm1 potentiates this capacity [222,224,225]. Again, while there are a number of associations that implicate potential roles for PITPnm1 in cancer, murine cancer models have not been interrogated to address this idea in sufficient depth.

1.21. Multidomain PITPs of worms and apicomplexans

At present, little is known about the multidomain PITPs of other organisms. C. elegans express PITP-1 a non-essential class IIa PITP identified as a loss of function allele in a genetic screen for mutants defective in a sensory ASE right (ASER) gustatory neuron plasticity [226]. ASER is responsible for attraction and avoidance sensing of salt, and chemotaxis as a function of prior experience depends on ASER plasticity. That is, worms are attracted to NaCl if they have not previously encountered it, but being repelled if they have [227]. This reversible behavioral conditioning depends on two phosphoinositide signaling cascades with PITP-1 mediating the interface between them. The first employs PLC/DAG/PKC signaling for attractive cues, with high DAG being associated with attraction and low levels with avoidance. The second utilizes PtdIns(3,4,5)P₃/PI3K signaling for plasticity and repulsive chemotaxis [228,229]. A thorough treatment of this topic is beyond the scope of this review and has been discussed elsewhere [64]. While current models suggest PITP-1 traffics PtdIns from the ER to the synaptic membrane during signaling, the data are perhaps more consistent with yeast Sec14 nanoreactor/ presentation models in which PITP-1 modulates PtdIns availability to opposing signaling pathways e. g. PLC and PI3K. Whether PITP-1 has intrinsic PtdIns-transfer activity remains to be determined and the nature of its second ligand is unknown. However, while exposure to sensory cues does not evoke degenerative phenotypes as in flies, the PITP-1 PITP domain is yet again both necessary and sufficient to rescue the chemosensory attraction and plasticity defects of pitp-1 mutants [226].

The phylum Apicomplexa includes the medically important pathogens Plasmodium falciparum and Toxoplasma gondii. These organisms express both Sec14- and StART-like proteins [230]. T. gondii encodes a single- StART-like PITP candidate (TgME49_289570). This remarkable protein is predicted to represent a multi-domain protein that contains an N-terminal PITP domain, a PH domain, and a C-terminal OSBP domain. In yeast, Sec14-dependent PtdIns4P signaling is downregulated by the oxysterol-binding protein homolog Kes1/Osh4 [231–235]. Kes1/Osh4 is alternatively proposed to act as a sterol-regulated ‘brake’ clamping Pik1-generated Golgi PtdIns4P and sequestering it from its effectors of TGN/endosomal trafficking [235], or a lipid transfer protein that transports PtdIns4P from the Golgi system to the ER for degradation by the Sac1 phosphoinositide phosphatase (reviewed [236]). While neither the T. gondii PITP nor OSBP activities have yet been demonstrated, this striking architecture is predicted to couple two domains involved in promoting PtdIns4P signaling (the PITP and PH domains) with one that functionally antagonizes PtdIns4P signaling (the OSBP domain).

2. Conclusions

After their initial discovery, PITPs and lipid transfer proteins in general, retreated into a scientific backwater in the 1970s and 1980s because it was unclear how cells would gainfully use the curious ATP-independent lipid exchange activities by which these proteins are defined. The entire field was revived by the discovery that Sec14 is a PITP with essential roles in coordinating lipid metabolism with membrane trafficking functions. New ideas as to how PITPs can function in cells further stimulate interest in these unusual proteins. Today, PITPs and other lipid transfer proteins represent an area of intense activity in contemporary cell biology. This interest is only growing with the recognition that deranged PITP function is associated with an ever-growing list of human diseases. It is now abundantly clear that the diversity of PITP-deficiency phenotypes reflects the roles of PITPs in diversifying the biological outcomes of phosphoinositide signaling. As a result, the next stage of progress will require understanding the mechanics of PITP lipid exchange reactions in atomistic detail, and understanding the biochemical features of PITP signaling environments. Translating that knowledge to an understanding of disease mechanisms represents the next stage of evolution in PITP science.

Abbreviations:

ADP

Adenosine diphosphate

Akt

Protein kinase B

ALL

Acute lymphoblastic leukemia

AMD

Age-related macular degeneration

ASER

ASE right

AsymAD

Asymptomatic Alzheimer’s disease

ATP

Adenosine tri-phosphate

ATRAP

Angiotensin II receptor-associated protein

αTTP

α-tocopherol transfer protein

BMI

Body mass index

CDK

Cyclin dependent kinase

CDP-DAG

Cytidine diphosphate diacylglycerol

CERT

Ceramide transfer protein

COP1

Coat protein complex

CPT1B

Carnitine palmitoyltransferase 1B

CRALBP

Cellular retinaldehyde binding protein

DAG

Diacylglycerol

DCC

Deleted in colorectal cancer

DMD

Duchenne muscular dystrophy

ED

Erectile dysfunction

ENU

Ethyl-N-nitrosourea

ER

Endoplasmic reticulum

ERG

Electroretinogram

FASN

Fatty acid synthase

FFAT

Diphenylalanine in an acidic tract motif

GAP

GTPase activating protein

GEF

Guanine nucleotide exchange factor

GRMD

Golden retriever muscular dystrophy (model)

Gold domain

Golgi dynamics domain

GOLPH3

Golgi phosphoprotein 3

HCV

Hepatitis C virus

HER

Human EGFR

HOTAIR

Hox transcript antisense intergenic RNA

HPC

Hematopoietic progenitor cell

HSC

Hematopoietic stem cell

IKNM

Interkinetic nuclear migration

IP₃

Inositol triphosphate

LATS1/2

Large tumor suppressor kinase 1/2

licRNA

Long noncoding RNA

LNS2

Lipin/Nde1/Smp2 domain

LOF

Loss of function

LPA

Lysophosphatidic acid

LUAD

Lung adenocarcinoma

MCB

Microcolin B

miRNA

microRNA

MK

Megakaryocyte

MST1/2

mammalian Ste20-like kinase ½

MYO18

Myosin 18

NADPH

Nicotinamide adenine dinucleotide phosphate

ncPCP

Non-canonical Wnt Planar Cell Polarity

NSC

Neural stem cell

OA

Osteoarthritis

OSBP

Oxysterol binding protein

OSH

Oxysterol-binding homology/ homolog/ homologous proteins

PDAC

Pancreatic ductal adenocarcinoma

PH domain

Pleckstrin homology domain

PDE10A

Phosphodiesterase 10A

PGP

Protruded genitalia phenotype

PIK

Phosphoinositide kinase

PI3K

Phosphatidylinositol 3-OH kinase

PI4K

Phosphatidylinositol 4-OH kinase

PIP

Phosphoinositide

PITP

Phosphatidylinositol transfer protein

PLC

Phospholipase C

PLD

Phospholipase D

PKC

Protein kinase C

PM

Plasma membrane

PtdCho

Phosphatidylcholine

PtdOH

Phosphatidic acid

PTEN

Phosphatase and tensin homog

PtdIns

Phosphatidylinositol

PIS

Phosphatidylinositol synthase

PtdIns3P

Phosphatidylinositol 3-phosphate

PtdIns4P

Phosphatidylinositol 4-phosphate

PtdIns5P

Phosphatidylinositol 5-phosphate

PtdIns(3,5)P₂

Phosphatidylinositol 3,5-bisphosphate

PtdIns(4,5)P₂

Phosphatidylinositol 4,5-bisphosphate

PtdIns(3,4)P₂

Phosphatidylinositol 3,4-bisphosphate

PtdIns(3,4,5)P₃

Phosphatidylinositol 3,4,5-triphosphate

RdgB

Retinal degeneration B

ROS

Reactive oxygen species

RPE

Retinal pigment epithelium

SAR

Structure activity relationship

SOS

Son of sevenless

SREBP1

Sterol regulatory elemen-binding protein 1 transcription factor

START-like

StAR related lipid transfer domain

SMC

Submicrovillar cisternae

SNP

Small nucleotide polymorphism

SRC

Subrhabdomeric cisternae

TAZ

Transcriptional coactivator with PDZ-binding motif

TEAD

Tea domain

TGN

Trans-Golgi network

TEME16F

Transmembrane protein 16F

T2D

Type 2 diabetes

UTR

Untranslated region

VAP

Vesicle-associated membrane protein (VAMP)-associated proteins

Vib

Vibrator

VWF

Von Willebrand Factor

YAP

Yes associated protein

Data availability

No data was used for the research described in the article.