A Quarter Century of EHD Protein Research: From Endosomal Recycling to Ciliopathies

ABSTRACT

Eps15 homology domain‐containing proteins comprise a conserved family of membrane‐remodeling ATPases that regulate endocytic trafficking, membrane fission, receptor recycling, primary ciliogenesis and membrane dynamics across eukaryotes. Since the initial identification of EHD1 and its Caenorhabditis elegans homolog RME‐1 as regulators of endocytic recycling, research over the past quarter century has expanded the functional scope of EHD proteins far beyond classical receptor return to the plasma membrane. In mammals, EHD1, EHD2, EHD3, and EHD4 occupy overlapping but distinct cellular locations and regulate diverse processes including tubular recycling endosome fission, caveolae stabilization, primary ciliogenesis, centrosome duplication, cytokinesis, mitochondrial homeostasis, lipid droplet biology, and lipophagy. These cellular functions are supported by extensive studies in cultured cells and animal models, including mice, zebrafish, flies, worms, and plants, highlighting both conserved and specialized roles for EHD orthologs. EHD dysfunction has also been associated with a broad range of human diseases, including metabolic and cardiovascular disorders, inflammatory and infectious disease, neurologic conditions, cancer, and ciliopathies. Although many disease links remain correlative or model‐based, the recent identification of an EHD1 founder mutation causing proteinuria, hearing loss, and polycystic kidney disease provides direct genetic evidence connecting EHD dysfunction to human pathology. This review summarizes 25 years of EHD research, emphasizing how EHD proteins coordinate membrane trafficking, organelle remodeling, and disease‐relevant cellular physiology.

Human EHD protein subcellular localization.

Abbreviations

BBS1

Bardet‐Biedl Syndrome 1

EHD

Eps15 Homology Domain

ERC

endocytic recycling compartment

MHC‐I

Major Histocompatibility Class 1

MICAL‐L1

MICAL‐like protein 1

PM

plasma membrane

RME‐1

receptor‐mediated endocytosis‐1

TRE

tubular recycling endosomes

1. Background and History

The Eps15 Homology Domain‐containing protein EHD1 was first identified as the product of a highly conserved gene present in invertebrates such as Drosophila melanogaster and Caenorhabditis elegans, as well as mammals including mice and humans. Based on its sequence, domain architecture, and predicted interaction partners, it was proposed to play a role in endocytosis [1]. Owing to its chromosomal location, EHD1 was initially proposed as a candidate gene for Bardet‐Biedl syndrome (BBS), reflecting early interest in its potential disease relevance [2]. Almost two decades later, it was demonstrated that EHD1 regulates primary ciliogenesis [3, 4]. Remarkably, nearly a quarter century after its discovery, a point mutation in EHD1 was directly linked to human disease, causing tubular proteinuria and hearing loss, consistent with a ciliopathy [5].

The primary function attributed to EHD1 and its C. elegans homolog RME‐1 in the early 2000s was the regulation of endocytic trafficking, particularly the recycling of internalized receptors back to the plasma membrane (PM) [6, 7, 8]. Over the past 25 years, numerous studies have confirmed that EHD1 controls the recycling of multiple receptors to the cell surface (see Table 1), with knockdown or depletion of EHD1 resulting in delayed recycling and intracellular retention of receptors [59, 60, 61]. While these observations have been repeatedly validated, the mechanistic understanding of how EHD1 regulates endocytic transport has evolved. Accumulating evidence now converges on a central role for EHD1 in mediating endosomal fission. Together, these findings established EHD1 as a central regulator of endocytic trafficking, paving the way for later work implicating it in other cellular processes.

TABLE 1.

Receptors regulated by EHD proteins.

EHD paralog	Plasma membrane protein	Model	References
EHD1	Yolk receptor	C. elegans	[6]
	Transferrin receptor	Chinese Hamster Ovary (CHO) cells, MEFs, HeLa	[7, 9, 10]
	MHC‐I	HeLa cells	[8]
	β1 integrin	HeLa cells	[11]
	CD59	HeLa cells	[12]
	MHC‐II	HeLa cells	[13]
	AMPA‐type glutamate receptors	Primary hippocampal neurons (rats)	[14]
	Cystic fibrosis transmembrane conductance regulator	HEK293‐CFTR	[15]
	Hyperpolarization‐activated cyclic nucleotide‐gated (HCN) ion channels HCN1, HCN2, and HCN4	Opossum kidney cells	[16]
	Potassium channel KCa2.3	HEK293 and HMEC‐1 cells	[17]
	G‐protein‐activated inwardly rectifying potassium channels (GIRK)	Primary hippocampal neurons (rats)	[18]
	LRP1	NIH3T3	[19]
	Smoothened receptor	NIH3T3	[20]
	TGFβR1	HTR8 cells	[21]
	IGF‐1R	TC71 NTC cells	[22]
	EGFR	MEFs, mice mammary epithelial cells	[23]
	Cx43	Cardiomyocytes (mouse)	[24]
	CD44	Lung adenocarcinoma and glioma cells	[25, 26]
	Colony‐Stimulating Factor‐1 (CSF‐1) receptor	Bone marrow–derived macrophages (BMDMs) (mouse)	[27]
	TNFR2	Bone marrow–derived macrophages (BMDMs) (mouse)	[28]
	Neuro‐glia cell adhesion molecule (NgCAM)	Hippocampal neurons (rat)	[29]
	Fer‐1‐like‐5 (Fer1L5)	Myoblasts (mouse)	[30]
	Src *non‐receptor tyrosine kinase	HeLa	[31]
	β2‐adrenergic receptors (β2AR)	Non‐small cell lung cancer cells	[32]
	LDLR	MEFs	[33]
EHD2	LeEix2 receptor	Nicotiana tabacum cv samsun and Nicotiana benthamiana leaves	[34, 35]
	GLUT4 transporter	3T3‐L1 adipocytes	[36]
	Transferrin receptor	3T3‐L1 adipocytes, Arabidopsis thaliana	[36, 37]
	Delta‐like ligand 4 (Dll4)	Human umbilical vein endothelial cell	[38]
	Myoferlin	Myoblasts (mouse)	[39, 40]
	KATP channel	Ventricular myocytes (mouse)	[41]
	eNOS	HUVECs	[42]
	ORAI1	Triple‐negative breast cancer cells (HS87 and MDA‐MB cells)	[43]
	E‐cadherin	Breast carcinoma, hepatocellular carcinoma, esophageal squamous carcinoma	[44, 45, 46]
EHD3	Transferrin receptor	HeLa cells	[47]
	CD59	HeLa cells	[12]
	CI‐M6PR	HeLa cells	[48]
	CaV3.1	Atrial myocytes (mouse)	[49]
	CaV3.2	Atrial myocytes (mouse)	[49]
	EGFR	U251 cells	[50]
	β3 integrin	HeLa cells	[51]
	Na/Ca exchanger	Cardiac myocytes (mouse)	[52]
	L‐type calcium channel type 1.2 (CaV1.2)	Cardiac myocytes (mouse)	[52]
EHD4	Nogo‐A	Primary hippocampal neurons (rats)	[53]
	TrkA	PC12 cells	[54]
	MHC‐I	HeLa cells	[55]
	Transferrin receptor	HeLa cells	[55]
	LDL receptor	HeLa cells	[55]
	VE‐cadherin	HUVECs	[56]
	Na+ transport channel Nav1.5	Ventricular cardiomyocytes (mouse)	[57]
	CD59	HeLa cells	[12]
— a	TCR‐CD3	Primary spleen cells (mice)	[58]

^a EHD1, EHD3, and EHD4 triple knockout cells.

In mammals, the EHD protein family (often referred to as the “C‐terminal EHD proteins,” because of the unusual positioning of their Eps15 homology (EH) domain at the carboxyl terminus) comprises four closely related members: EHD1, EHD2, EHD3, and EHD4. In this review, we examine the physiological roles of EHD1 and its paralogs in human disease. At the cellular level, we will highlight their involvement in diverse processes, including receptor and membrane recycling, caveolae stabilization, primary ciliogenesis, centrosome duplication and cell division, and mitochondrial homeostasis. We will also discuss the functions of EHD proteins expressed in animal models including mice, worms, flies, and zebrafish, as well as in plants. Potential mechanisms of EHD activity will be considered, with particular emphasis on structure and binding partners, lipid binding, ATP binding and hydrolysis, and membrane fission. Finally, we outline future directions for EHD research.

2. Physiological Functions of EHDs

The C‐terminal Eps15 homology domain‐containing protein (EHD) family comprises four highly homologous human paralogs: EHD1, EHD2, EHD3, and EHD4 (see Figure 1C). These ATPases localize to distinct subcellular regions and regulate key steps of receptor trafficking [59, 60, 62] (see Figure 2). Impaired ATP‐binding EHD mutants exhibit delayed receptor recycling, highlighting the role of ATP binding and hydrolysis for functional recycling endosomes [6, 7, 8, 47, 63]. Among the four paralogs, EHD1 is the most extensively studied, likely due to its strong evolutionary conservation in invertebrates [6]. In mammals, EHD1 also mediates recycling of various types of internalized PM receptors such as transferrin [7], Major Histocompatibility Complex Class I (MHC‐I) [8], and β1 integrins [11, 31] and others, transporting them from the endosomal recycling compartment (ERC) to the PM.

FIGURE 1.

EHD protein domain architecture, structure and homology between human paralogs. (A) Domain architecture of EHD1. (B) Proposed EHD1 structure from UniProt. (C) EHD protein paralog amino acid identity.

FIGURE 2.

Human EHD protein subcellular localization. EE, early or sorting endosome; TRE, tubular recycling endosome.

By coordinating receptor transport, EHD proteins influence diverse physiological processes including cholesterol homeostasis [33], muscle development [39], potassium channel regulation [17], angiogenesis [56], and trafficking of the cystic fibrosis transmembrane conductance regulator (CFTR) [15]. Consistent with these roles, altered EHD expression or function has been associated with a range of human diseases, including metabolic and cardiovascular disorders [16, 24, 64], ciliopathies [5], and multiple cancers [25, 44, 65, 66, 67, 68]. However, the nature of this evidence varies considerably across disease contexts, ranging from direct human genetic findings to correlative tissue‐expression studies and mechanistic insights from cell and animal models.

2.1. Metabolic Regulation and Membrane Homeostasis

EHD proteins contribute to several aspects of metabolic regulation. EHD1 overexpression attenuates the downstream signaling by insulin‐like growth factor 1 receptor (IGF‐R1), potentially by affecting receptor internalization through interactions with SNAP29 and the adaptor protein 2 (AP‐2) complex [69]. In adipocytes, both EHD1 and EHD2 are required for insulin‐stimulated recycling of the GLUT4 glucose transporter [36, 70], linking EHD‐dependent trafficking to glucose uptake and insulin responsiveness [71, 72].

EHD proteins also participate in lipid homeostasis. EHD1 is involved in cholesterol regulation, potentially through direct or indirect effects on low‐density lipoprotein (LDL) receptor internalization [33]. EHD2 plays an important role in regulating adipocyte function and fat storage [73, 74, 75, 76]. At the PM, EHD2 stabilizes caveolae [77, 78, 79], and Ehd2 knockout adipocytes show reduced fatty acid uptake [80]. EHD2 also associates with lipid droplets [73, 81, 82, 83, 84] promoting their degradation, with EHD2 depletion resulting in increased lipid droplet area [80, 83]. Together, these observations support a broader role for EHD proteins in coordinating membrane trafficking pathways related to glucose and lipid metabolism.

2.2. Immune Cell Trafficking and Host–Pathogen Interactions

Although EHD proteins have not been studied as extensively in immune cells, there is evidence that they function in both adaptive and innate immunity. By facilitating recycling of MHC‐I [8] and MHC‐II [13] receptors, EHD1 contributes to antigen presentation. Additionally, EHD1, EHD3, and EHD4 have been implicated in T‐cell responses and activation in CD4⁺ T‐cells, with the knockout of Ehd1, Ehd3, or Ehd4 resulting in impaired T‐cell receptor (TCR) recycling of the TCR‐CD3 complex [58].

In innate immune cells such as developing macrophages, EHD1 is required for endocytic trafficking of Colony‐Stimulating Factor‐1 (CSF‐1) receptor [27]. Ehd1 knockout bone marrow–derived macrophages (BMDMs) show reduced CSF‐1 receptor levels, defective downstream signaling, and impaired macrophage proliferation and migration [27].

EHD proteins also affect host–pathogen interactions. EHD4 depletion reduces Human Immunodeficiency Virus (HIV) infectivity [85], and in HIV‐infected macrophages, EHD3 is needed for macropinocytosis and its loss impairs phagophore‐lysosome fusion [86]. Listeria monocytogenes enters host intestinal cells at EHD2‐enriched PM sites [87], consistent with a role for EHD2 in bacterial uptake and spread. These findings show that EHD proteins regulate immune cell trafficking and are involved in the process of pathogen uptake during infection.

2.3. Neuronal and Synaptic Functions

EHD proteins regulate multiple aspects of neuronal trafficking and signaling. EHD1 facilitates recycling of AMPA‐type glutamate receptors from endosomes to the PM in post‐synaptic neurons, thereby strengthening synaptic transmission [14]. EHD1 also regulates neuronal excitability by recycling G protein‐activated inwardly rectifying K(+) (GIRK) channels [18]. EHD1 and EHD4 hetero‐oligomerize and regulate neuro‐glia cell adhesion molecule (NgCAM) internalization and axonal targeting, consistent with a role in transcytosis [29].

EHD proteins also support neurite growth and axonal transport. Nerve Growth Factor (NGF) stimulation recruits active Rab35 together with MICAL‐L1 and EHD1 to sorting endosomes [88], promoting transport of endosomes to growing axonal ends, thus facilitating neurite growth [88, 89]. EHD1 levels increase upon spinal cord injury [90], facilitating receptor transport for neurite growth and recovery [88, 89, 90], and EHD1 distribution on endosomes is maintained via interaction with the E3 ubiquitin ligase, Triad1, facilitating neurite growth during spinal cord injury [91]. In rat hippocampal neurons, EHD1 also negatively regulates exocytosis by competitively binding SNAPIN, preventing the latter from interacting with the SNARE protein SNAP‐25 [92].

Additional evidence implicates other EHD family members in neuronal function. EHD2 is upregulated in hemorrhagic rats [93], suggesting a mechano‐protective role in maintaining astrocyte PM turnover [94]. In the lamprey eel, the EHD1 and EHD3 ortholog (l‐EHD) is required for endocytic vesicle fission at neuronal synapses [95]. EHD4 levels also rise in rat neurons after NGF stimulation [54], and EHD4‐mediated macropinocytosis [96] is required for receptor transport, either supporting [96, 97] or inhibiting axonal growth [53]. At the neuromuscular junction (NMJ), both EHD1 and EHD4 localize to synaptic clefts, whereas the Drosophila ortholog Past1 is required for proper assembly of the subsynaptic membrane reticulum and for normal synaptic transmission [98]. Overall, these studies support broad roles for EHD proteins in synaptic trafficking, neuronal plasticity, and axonal remodeling.

2.4. Muscle Development and Repair

EHD1 and EHD2 play important roles in skeletal muscle development. Both proteins are required for myoblast fusion, a process essential for the formation of multinucleated muscle fibers [30, 39]. EHD2, which is highly expressed in developing muscle, interacts with myoferlin, a PM protein required for myoblast fusion, and mediates its recycling to the PM [39]. Accordingly, Ehd2 knockout myoblasts accumulate intracellular myoferlin and exhibit markedly impaired fusion [39]. Both EHD1 and EHD2 also interact with Fer‐1‐like‐5 (Fer1L5), another myoferlin‐family member required for myoblast fusion; knockdown of either protein significantly reduces fusion [30].

In mature muscle, EHD proteins contribute to membrane organization and repair. Heterozygous Ehd1 knockout mouse skeletal muscles show reduced muscle mass and poor contraction [99], partly due to faulty BIN1 accumulation on abnormal transverse‐tubule (T‐tubule) extensions [99, 100]. In skeletal muscle repair, EHD1 and EHD2, along with myoferlin and actin‐related proteins, localize to sites of tissue injury to promote membrane fusion and resealing of the damaged muscle fibers [40]. In primary human myotubes, however, EHD2 alone is required [101].

EHD proteins also regulate excitability in cardiac muscle. EHD1 recycles hyperpolarization‐activated cyclic nucleotide–gated (HCN) channels (HCN1–4), and its dysfunction contributes to cardiac arrhythmias [16]. EHD1 also interacts with phosphorylated Cx43, a gap junction protein, promoting its internalization, recycling, and retrograde transport from the Golgi to the Endoplasmic Reticulum (ER), and is essential for cardiac remodeling under ischemia [24]. EHD2 stabilizes K_ATP channels at the sarcolemma [41], whereas EHD3 maintains surface levels of T‐type Ca²⁺ channel (TTCC) and additional cardiac transporters [49] through interactions with ankyrin‐β [52, 102]. EHD4 may participate in the trafficking of the voltage‐gated Na⁺ transport channel in ventricular cardiomyocytes [57]. These findings highlight important roles for EHD proteins in both skeletal and cardiac muscle physiology.

2.5. Vascular and Endothelial Functions

EHD proteins also regulate endothelial trafficking and vascular function. EHD4 was first identified as a part of the extracellular matrix in human placenta [103]. It is highly expressed in human umbilical vein endothelial cells (HUVECs), where it facilitates endothelial cell migration via vascular endothelial (VE)–cadherin recycling, promoting angiogenesis [56], and is enriched in pancreatic islet vasculature [104]. Gene expression of EHD4 and EHD3 is also correlated with liver fibrosis, suggesting their potential role in angiogenesis [105, 106]. Additionally, EHD4 is required for the efficient recruitment of EHD1 to sorting endosomes [107], and EHD1 mediates recycling of the calcium‐activated potassium channel KCa2.3, crucial for endothelial polarity and blood pressure regulation [17].

EHD2 likewise contributes to vascular biology. In HUVECs, it promotes the internalization and transcytosis of delta‐like ligand 4, essential for Notch‐Delta signaling and angiogenesis [38]. At the PM, EHD2 stabilizes the eNOS receptor; its deletion causes patchy eNOS distribution and impaired blood vessel relaxation [42]. Taken together, these studies support important functions for EHD proteins in endothelial trafficking, angiogenesis, and vascular homeostasis.

2.6. Primary Ciliogenesis

Three of the four EHD proteins play key roles in primary ciliogenesis (see Figure 3). The primary cilium is a key signaling organelle that controls cellular maturation and development, and failure to generate a primary cilium is the cause of severe developmental disorders known as ciliopathies [108]. Both EHD1 and EHD3 localize to the ciliary base, and depletion of either protein disrupts ciliogenesis in retinal pigment epithelial (RPE) and medullary epithelial cells [3]. EHD4 oligomerizes with EHD1 [55], and is also required for ciliogenesis in mouse fibroblasts [107].

FIGURE 3.

Proposed role for EHD1 in primary ciliogenesis. EHD1 promotes centriolar satellite movement from the periphery to the centrosome, potentially through modulation of microtubule growth. The centriolar satellites deliver the E3 ubiquitin ligase HERC2 to the mother centriole, where it ubiquitinates CP110. Upon CP110 polyubiquitination, the AAA ATPase valosin‐containing protein/p97 (p97) extracts CP110 from the mother centriole and targets it for proteasomal degradation, allowing the formation of the ciliary vesicle and growth of the axoneme, and ultimately generation of the primary cilium.

These findings extend the role of EHD proteins beyond classical endocytic recycling and place them within the broader machinery that coordinates membrane remodeling at the centrosome and ciliary base. In the cochlea, EHD4 interacts with stereociliary membrane proteins involved in mechano‐transduction, although Ehd4 knockout mice compensate through increased EHD1 expression and do not develop overt hearing defects [109, 110].

3. Disease Associations of EHD Proteins

Altered EHD expression, posttranslational modification, or function is associated with a range of human diseases, including metabolic and cardiovascular disorders, ciliopathies, inflammatory conditions, and cancer [5, 111, 112, 113, 114, 115]. However, the strength of this evidence varies substantially across disease categories. In most cases, the link to disease is based on altered expression, methylation, prognostic association, or mechanistic studies in cultured cells and animal models. Direct human genetic evidence remains limited, with the strongest example being the recently identified EHD1 founder mutation associated with a ciliopathy‐like syndrome [5]. Accordingly, the sections below distinguish, where possible, among direct human genetic findings, patient‐derived expression or association data, and mechanistic inferences from model systems.

3.1. Cardiometabolic Disease

Evidence linking EHD proteins to cardiometabolic disease is driven mainly by patient expression studies together with research on adipocytes and other model systems. In Type II diabetic patients, the EHD2 gene is methylated and suppressed [71, 116]. EHD2 protein levels in hypoxic adipocytes contribute to insulin resistance [72]. EHD2 is also suppressed in obesity [116], and both EHD2 and caveolin‐1 levels are reduced in insulin‐resistant adipocytes [72]. These observations implicate altered EHD2 expression in metabolic dysfunction, although direct causal human genetic evidence is lacking.

Mechanistically, both EHD1 and EHD2 are required for insulin‐stimulated recycling of GLUT4 [36, 70], providing a plausible link between EHD dysfunction and impaired glucose uptake. EHD2 additionally regulates caveolar stability, fatty acid uptake, and lipid droplet turnover [73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84], while EHD1 contributes to cholesterol homeostasis, potentially through its impact on LDL receptor trafficking [33]. Thus, although the majority of evidence is correlative, this supports a functional role of EHD‐dependent membrane trafficking in glucose and lipid homeostasis.

3.2. Cardiovascular Disease

Evidence for EHD involvement in cardiovascular disease similarly combines human association data with strong mechanistic support from cellular and animal models. EHD3 levels are elevated in heart failure [64], and EHD2 synthesis is dysregulated during hematopoiesis in patients with excessive platelet formation [117]. Elevated EHD2 levels have also been reported in hypertension drug‐treated rats [118]. These observations support cardiovascular relevance but lack direct causation in humans.

Mechanistic studies provide a stronger framework for interpretation. EHD1 regulates recycling of HCN channels and trafficking of phosphorylated Cx43 [16, 24], implicating it in arrhythmia and ischemic remodeling. EHD2 stabilizes KATP channels at the sarcolemma, and ATP‐deficient EHD2 mutants increase channel internalization and reduce conductance in ventricular myocytes [41]. EHD3 controls trafficking of T‐type Ca2+ channels, β1‐adrenergic receptors, Na/Ca exchangers, and l‐type calcium channels, and cardiac depletion causes abnormal conductance, bradycardia, and defective ventricular function [49, 52]. EHD1, EHD2, and EHD4 also participate in endothelial trafficking pathways relevant to vessel relaxation and angiogenesis [38, 42, 56]. Overall, these findings support a role for EHD proteins in cardiovascular physiology and disease, despite the lack of direct genetic evidence.

3.3. Immune, Inflammatory, and Infectious Disease Associations

Links between EHD proteins and immune or infectious disease are currently supported mainly by altered expression in disease settings and by mechanistic studies in immune cells. For example, macrophages associated with coronary heart disease show elevated EHD1, which is required for TNFR2 recycling and maintenance of IL1‐β and TNF production [28]. EHD2 expression increases in SARS‐CoV‐2‐infected lung cells [119], and EHD2 [120] and EHD4 [121] are elevated in obstructive pulmonary inflammation. EHD4 secretome levels are also increased in pediatric pneumonia [122]. These findings suggest that EHD proteins may be engaged in inflammatory remodeling, but they do not establish causation.

Functional studies nonetheless provide mechanistic context. EHD1 regulates antigen presentation through MHC‐I and MHC‐II recycling [8, 13] and is required for macrophage CSF‐1R trafficking [27]. EHD1, EHD3, and EHD4 also contribute to TCR recycling [58]. In viral settings, EHD4 depletion reduces HIV infectivity [85], EHD3 supports macropinocytosis and phagophore‐lysosome fusion in HIV‐infected macrophages [86], and EHD2‐enriched PM domains facilitate Listeria monocytogenes entry into cells [87]. Accordingly, evidence places EHD proteins at the intersection of immune trafficking and host–pathogen interactions.

3.4. Neurologic and Neuropsychiatric Associations

The neurologic disease literature is more heterogeneous and remains largely correlative to date. Elevated EHD3 levels have been connected to cognitive function in autism patients [123], genetic alterations in EHD3 have been reported in females with major depressive disorder [124, 125, 126], and EHD4 variation has been associated with Progressive Supranuclear Palsy, a primary taupathy [127]. EHD4 is also upregulated in aged post‐stroke rat brains [128], and elevated EHD4 protein levels have been reported in alcohol‐drinking males [129]. These studies suggest potential neurologic relevance.

Studies done in neurons provide biological context for these associations. EHD1 regulates recycling of AMPA receptors and GIRK channels [14, 18], EHD1 and EHD4 control NgCAM trafficking [29], and EHD1 is required for neurite growth after NGF stimulation and spinal cord injury [88, 89, 90, 91]. Additional model‐organism studies, including work on Past1 at the Drosophila NMJ [98] and l‐EHD at lamprey synapses [95], further support conserved roles in neuronal membrane trafficking. Overall, these findings are consistent with roles for EHD proteins as regulators of neuronal function, but the direct connection to specific human neurologic diseases remains provisional.

4. Cancer

Among non‐ciliopathy disease categories, the most extensive literature linking EHD proteins to pathology can be found in various cancers. However, this evidence is dominated by altered expression, prognostic association, and mechanistic studies in model systems rather than by human genetic causation. As a result, EHD proteins are best viewed at present as modulators of tumor‐associated membrane trafficking pathways whose effects are often context dependent.

4.1. EHD1 in Cancer

EHD1 is the best studied family member in cancer, consistent with the broader principle that cancer cells frequently reprogram membrane‐trafficking pathways [130], including those regulated by EHD proteins [60]. High levels of EHD1 expression correlate with reduced disease‐free survival, reduced overall survival, and poor response to chemotherapy in multiple tumor types [22, 25, 26, 65, 131, 132, 133, 134]. Mutations in EHD1 have also been linked to malignant papillary mesothelioma [135] and monoclonal plasma cells in POEMS syndrome [136], although these findings do not yet establish a broadly causal oncogenic role.

Mechanistically, EHD1 promotes trafficking of multiple receptors and signaling proteins with known roles in tumor progression. It mediates transport of IGF‐1R and EGFR from the Golgi to the PM, possibly through hybrid secretory‐endosomal routes [22, 23, 65, 66]. It also promotes recycling of CD44, β1‐integrin, and β2‐adrenergic receptors, thereby supporting stemness, motility, VEGF signaling, and angiogenesis [11, 25, 26, 32]. Elevated CD44 surface levels suppress Hippo signaling, promoting YAP nuclear translocation, stemness, and tumor growth [25], whereas EHD1 depletion reduces CD44 and stemness markers [26], impairing migration and invasion [25, 137]. EHD1 also promotes β1‐integrin recycling [11] and regulates Src transport to the PM [31], which is required for focal adhesion turnover and cell motility; accordingly, EHD1 depletion causes β1‐integrin accumulation on endosomes and Src aggregation in the ERC, resulting in impaired migration [11, 31]. In addition, EHD1 recycles β2‐adrenergic receptors, supporting VEGF signaling and angiogenesis in growing tumors [32]. EHD1 depletion attenuates Akt/mTOR signaling, lowers HIF2α and matrix metalloprotease output, reduces β‐catenin activation and glucose uptake, and decreases tumor proliferation and invasiveness [67, 138, 139, 140]. High EHD1 expression also correlates with TGFβ‐1 receptor expression, consistent with a role in TGFβ receptor recycling that could enhance tumor proliferation and survival [141]. The interaction of EHD1 and EHD4 with Phostensin at endosomes regulates transferrin receptor recycling [140, 142], providing an additional trafficking axis that may influence tumor progression. Moreover, EHD1 loss prevents 14‐3‐3ζ dimerization and enhances its phosphorylation, attenuating β‐catenin activation and reducing glucose uptake and tumor growth [67], while EHD1 depletion also reduces CDK2 phosphorylation, thereby suppressing proliferation [139]. EHD1 also contributes to chemoresistance and immune evasion. EHD1 and EHD2 have both been implicated in chemoresistance [44, 143], and EHD1 depletion increases intracellular cisplatin and restores drug sensitivity [134]. Conversely, EHD1 upregulation promotes PI3K/Akt signaling and gefitinib resistance, whereas its suppression restores gefitinib sensitivity [144, 145]. EHD1 also recycles Programmed Death‐Ligand 1 (PDL1), enabling immune evasion, and its depletion improves anti‐PD‐L1 immunotherapy efficacy [146]. Furthermore, interaction of N6‐methyladenosine‐modified EHD1 mRNA with YTH N6‐methyladenosine RNA‐binding protein 1 stabilizes EHD1 expression and promotes tumor progression [146]. Overall, the evidence strongly favors a pro‐tumorigenic role for EHD1 in most contexts.

4.2. EHD2 in Cancer

The role of EHD2 in cancer is more context dependent. High EHD2 expression correlates with poor survival in some cancers, including clear cell renal cell carcinoma and glioma, and may serve as a biomarker [44, 147, 148, 149, 150, 151, 152]. In contrast, tumor‐suppressive roles have been described in lung adenocarcinoma and hepatocellular carcinoma [45, 153, 154].

Functional studies reflect this duality. In some cases, EHD2 depletion impairs migration and invasion and reduces CD44 levels and stemness markers, suggesting a pro‐tumorigenic role [147, 148]. EHD2 also regulates ORAI1 surface abundance and calcium homeostasis [43], and its locus can generate a circular RNA that promotes glucose uptake and metabolic activity in lung adenocarcinoma [155]. EHD2 is also a HIF‐2α target that may serve as a biomarker for clear cell renal cell carcinoma [149], and knockout of HIF‐1α or EHD2 represses hypoxia‐induced macropinocytosis, thereby impeding hypoxic hepatocarcinoma cell growth [156]. Suppression of EHD2 by lncRNA TUSC8 inhibits migration [157], although gain‐of‐function mutations have also been reported to promote tumor spread [158]. In other settings, however, loss of EHD2 enhances migration, invasion, metastasis, and epithelial‐mesenchymal transition, whereas EHD2 upregulation inhibits tumor progression [153, 159]. Additional studies support tumor‐suppressive functions for EHD2: in resting cells EHD2 is bound to myoferlin, whereas IL‐6 stimulation promotes dissociation, freeing myoferlin to bind activated STAT3 and chaperone it to the nucleus, consistent with a negative regulatory role for EHD2 [160]. Ese3 transcription factor‐mediated EHD2 downregulation increases proliferation and tumor volume [161], EHD2 overexpression reduces migration, invasion, and epithelial‐mesenchymal transition while promoting cell‐cycle arrest and apoptosis in colon cancer [162], and EHD2 is negatively regulated at the transcriptional level by PIGX in proliferating breast cancer [163]. The positive relationship between EHD2 and E‐cadherin is notable in that it supports a tumor‐suppressive interpretation in several epithelial cancers [44, 45, 46, 164]; indeed, EHD2 depletion reduces E‐cadherin levels and promotes cancer cell migration and epithelial‐mesenchymal transition [44, 45, 46]. Thus, EHD2 may act either as an oncogenic or tumor‐suppressive factor, depending on cellular context.

4.3. EHD3 and EHD4 in Cancer

EHD3 is abundantly expressed in small‐cell lung cancer but is markedly downregulated or epigenetically silenced in multiple other cancers [111, 112, 113, 114, 115]. EHD3 also has important effects on receptor signaling. For example, EHD3 upregulation can stabilize EGFR by reducing its ubiquitination, without affecting Akt/ERK activation [50]. In addition, EHD3 mediates β3‐integrin recycling, and its depletion impairs receptor recovery and adhesion [51]. In gastric cancer models, EHD3 depletion reduces proliferation, migration, invasion, epithelial‐mesenchymal transition, and tumor size [68]. However, in glioma, EHD3 overexpression suppresses proliferation, migration, and tumor growth [114]. Thus, as with other EHD family members, the effects of EHD3 in cancer appear strongly context dependent.

EHD4 also appears to have tumor‐type‐specific roles. It has been reported to be protective in colorectal cancer [165], but elevated in hepatocellular carcinoma [166], and it may also stabilize an anti‐apoptotic microRNA required for tumor progression. Overall, a complex pattern has emerged for EHD proteins in cancer, with family members exhibiting context‐dependent oncogenic or tumor‐suppressive functions.

4.4. Ciliopathies

Ciliopathies provide the strongest direct human genetic evidence linking EHD dysfunction to disease. The clearest example is a founder mutation in EHD1, R398W, which causes high‐frequency hearing loss, proteinuria, and polycystic kidney disease, a phenotype consistent with a ciliopathy [5]. In mice, the same pathway is linked to defects in spermatogenesis and male infertility [110], further supporting the physiological significance of EHD1‐dependent ciliary function.

This human genetic evidence is reinforced by mechanistic studies. EHD1 and EHD3 localize to the base of primary cilia, and depletion of either protein disrupts ciliogenesis in retinal pigment epithelial and medullary epithelial cells [3, 4, 167, 168]. EHD4 oligomerizes with EHD1 [55] and is also required for ciliogenesis in mouse fibroblasts [107]. Although altered EHD3 levels have been associated with diabetic retinopathy in a patient subpopulation [169], and EHD4 participates in cochlear membrane trafficking [109], these links remain less definitive than the EHD1 R398W syndrome [5]. Overall, the ciliopathy field currently provides the most compelling example of a direct disease connection for the EHD family.

5. EHDs in Animal Models

5.1. Mus musculus

Mus musculus encodes four EHD homologs, EHD1, EHD2, EHD3, and EHD4, that share over 70% amino acid sequence identity [1] (see Figure 1C). Mouse EHD1 displays 99% amino acid sequence identity with its human ortholog, making the mouse an ideal model for studying EHD1's physiological functions.

EHD1 is ubiquitously expressed in mouse tissues, with particularly high levels in developing male germ cells of the testis, consistent with its role in spermatogenesis [1, 170]. Mutant Ehd1 (R398W) knock‐in mice exhibit defective acrosome formation [110]. Consistent with a role for EHD proteins in the testis, Ehd4 knockout mice display abnormal spermatogenesis [171]. EHD1 levels are doubled in these knockout mice, indicating partial compensation for EHD function during male gonadal development [171]. EHD3 in Leydig cells is also required for testosterone secretion in mice testes [172]. EHD3 is highly expressed in kidney glomerular endothelial cells [173, 174], where it is essential for maintaining effective filtration. In diabetic (obese) mouse models, reduced EHD3 expression correlates with increased glomerular fenestrations and impaired filtration [175]. Ehd3 knockout mice have compensatory upregulation of EHD4 [176], while double Ehd3 and Ehd4 knockout mice develop severe proteinuria [176].

Mouse models have also been used to elicit the role of EHD2 in adipocytes for fat storage [80, 116]. EHD2 regulates lipid droplet size and is essential for insulin‐mediated signaling and glucose uptake [72, 177, 178, 179]. EHD2 is also required for the stability of eNOS receptors at the PM, with Ehd2 knockout mice displaying impaired blood vessel relaxation [42]. However, EHD2 is dispensable for eNOS‐dependent calcium release in vascular smooth muscle cells of aged mice [180]. Additionally, EHD1 and EHD3 regulate the dynamic transport of β‐secretase in hippocampal neurons, which is dysregulated during Alzheimer's disease progression [181].

During embryogenesis, EHD1 is expressed in the developing heart, occipital lobes, and limb buds [1], suggesting roles in early organogenesis. While some Ehd1 knockout studies report no overt phenotype in adults [9], others observed partial embryonic lethality [170]. These discrepancies likely reflect genetic background differences or compensation by other EHD proteins. Notably, transferrin receptor recycling is significantly delayed in Ehd1 knockout mice, mirroring the receptor recycling defects seen in human cells [9, 10]. Mutant Ehd1 knock‐in mice, with the same R398W founder mutation, exhibit impaired uptake of low molecular weight dextran in proximal convoluted tubules, leading to proteinuria [5]. This mislocalized and inactive EHD1 mutant causes hearing loss in the inner ear, underscoring EHD1's critical roles in endocytic trafficking and sensory function [5].

5.2. Danio rerio

Danio rerio expresses five EHD homologs: ehd1a, ehd1b, ehd2a, ehd2b, and ehd3, reflecting gene duplication events. Ehd1a/b paralogs are highly similar to each other (89% identity) and closely related to the human EHD1 ortholog, while ehd3 most closely resembles the human EHD3 ortholog (see Figure 4A,B). Knockdown of ehd1a/b causes synergistic embryonic lethality, exacerbated by concurrent ehd3 loss, indicating distinct but complementary roles in organism survival [3].

FIGURE 4.

C‐terminal EHD paralogs in humans and other species. (A) EHD proteins classified by species and related to their nearest human homolog. (B) A dendrogram was generated to show similarity of EHD homologs. Protein sequences were collected from NCBI and aligned using MUSCLE, the alignment results were entered into IQ‐Tree, and iTOL was used to visualize the results. This process was streamlined using the Phylogenetic_Analysis_Protein_Sequences pipeline created in Colab (https://github.com/Ash100/Alignment). Bootstrap values indicating branch confidence are shown.

Functional analyses show that ehd1 and ehd3 play cooperative but tissue‐specific roles in ciliogenesis and endocytic trafficking [3]. Both ehd1 and ehd3 are required for photoreceptor cilium formation, whereas ehd1 alone is critical for kinocilium formation in otic cells, consistent with its higher expression in the ear [3]. At neuromasts, ehd1 is essential, whereas ehd3 makes a partial contribution to cilia formation, and defects are rescued by human EHD1 or EHD3, but not by EHD4, demonstrating strong functional conservation of EHD proteins from mammals to zebrafish [3]. Additionally, ehd1b is highly expressed in kidneys, and ehd1a/b knockout embryos show reduced endocytic uptake in the pronephric tubules, demonstrating conserved renal trafficking functions [5]. Consistent with human cells and mouse models, loss of ehd2a/b in zebrafish embryos promotes dysmorphic sprouting in blood vessels due to impaired delta‐notch signaling [38].

5.3. Drosophila melanogaster

Drosophila melanogaster encodes a single EHD ortholog Past1 (putative achaete scute target 1), originally identified as one of the targets for the achaete‐scute gene complex. Past1 was initially cited as phylogenetically closest to the mammalian EHD2 ortholog [182], but displays greater amino acid identity with both EHD1 and EHD3 (see Figure 4A,B). It has two transcripts: the longer is expressed throughout development in both sexes, while the shorter is male‐specific, expressed in the third larval stage and enriched in testes, indicating germ‐line specific regulation [182].

Past1 deletion mutants show defective sperm individualization [182], consistent with mouse models [110, 171, 176], as well as abnormal ovary morphology, and ~20% slower development, partially due to interactions with the Notch receptor [182]. Past1 deletion mutants also exhibit lower survival at high temperatures [182], similar to temperature‐sensitive shibire mutants [183].

Past1 may regulate developmental processes, including eye development, potentially through effects on Numb‐associated endocytic trafficking [184], with Past1 deletion mutants showing defective photoreceptors and cone cell differentiation [185]. At the NMJ, Past1 interacts with Syndapin to stabilize tubulovesicular structures, and NMJ in Past1 knockout flies display collapsed synaptic boutons [98]. In cultured cells, Past1 localizes mainly to the PM and mediates endocytosis in garland cells and larval nephrocytes [182]. Additionally, Past1 regulates endosome‐to‐Golgi retrograde trafficking required for salivary granule maturation [186], highlighting its diverse roles in membrane trafficking during development.

5.4. Caenorhabditis elegans

Caenorhabditis elegans encodes a single EHD ortholog, RME‐1 (receptor‐mediated endocytosis‐1), sharing 67% identity with human EHD1 and similar identity with human EHD3 [6] (see Figure 4). RME‐1 was first identified in a genetic screen where mutant oocytes showed abnormal uptake of fluorescently labeled yolk proteins [187]. A recent study also indicates that RME‐1 is required for efficient secretion of vitellogenins, the lipoprotein precursors of yolk proteins, to the pseudocoelom [188]. The Rme‐1 gene in C. elegans has five splice isoforms: the shortest is expressed at all developmental stages, while the longest appears weakly in the second larval stage and increases with age, peaking in adults [6]. Rme‐1 deletion mutants or knockdown worms show no significant embryonic or larval lethality, although brood sizes of deletion mutants are smaller than wild‐type counterparts [6].

RME‐1 studies revealed an evolutionarily conserved structural and functional role for EHD proteins in receptor trafficking [6]. Point mutants display defective endocytosis in coelomocytes, where the G81R mutation in the ATP‐binding loop hampers nucleotide binding [6, 63]. Deletion mutants accumulate intestinal vacuoles starting at the fourth larval stage, increasing in size and number with age, reflecting RME‐1's role in basolateral recycling of receptors [6]. Indeed, RME‐1 interacts with C. elegans ALX‐1 to control recycling of basolateral cargo in the worm intestine [189]. Additional studies further support a role for RME‐1 within the basolateral recycling pathway, showing that RAB‐10 functions upstream of RME‐1 in the intestine, whereas loss of the Numb homolog NUM‐1 can bypass the requirement for RME‐1, consistent with pathway‐level regulation of RME‐1‐dependent recycling [190, 191].

In worms, RME‐1 also interacts with AMPH‐1, a bar domain‐containing protein that may limit RME‐1 ring assembly on membrane tubules, potentially contributing to the regulation of endosome fission [192]. Moreover, RME‐1 is recruited to arrested endocytic intermediates induced upon loss of filamentous actin [193]. Overall, studies in C. elegans largely support a conserved role for RME‐1 in endocytic recycling that is consistent with the functions of its mammalian homologs.

6. Plants

Plant EHD orthologs have conserved roles in membrane trafficking and regulate distinct physiological processes. Arabidopsis thaliana encodes two EHD‐like proteins, AtEHD1 and AtEHD2, that share ~74% similarity with mammalian EHD1 and EHD2 orthologs, respectively, although it should be noted that the plant AtEHD proteins contain their EH domains at the N‐terminus of the protein, unlike all other EHD proteins [37]. Both proteins localize to the PM, suggesting roles in endocytic and vacuolar trafficking. AtEHD1 silencing accelerates flowering [37], likely via altered hormone or stress signaling, and impaired endocytic recycling. AtEHD2 overexpression inhibits internalization of receptors like transferrin and leucine‐rich repeat receptor‐like protein (LeEix2), revealing differential regulation of membrane trafficking by AtEHD proteins [37]. AtEHD2 plays a crucial role in pathogen defense by inhibiting LeEix2 internalization, following binding to the fungal ligand ethylene‐induced‐xylanase, thus preventing downstream ethylene biosynthesis [34]. The coiled‐coiled and nucleotide‐binding domain of AtEHD2 binds and prevents LeEix2 receptor internalization [194]; mutants lacking these domains exhibit altered actin organization at the PM, reduced ethylene production, and compromised pathogen defense [194]. Additionally, SUMOylation regulates AtEHD2 distribution and function in anti‐fungal defense [35]. Mutation of the SUMOylation site in AtEHD2 promotes LeEix2 endocytosis and thereby mitigates anti‐fungal defense response [35].

7. Physiological Functions of EHDs

7.1. Receptor Recycling

Upon internalization, PM receptors are delivered to the early or sorting endosome (EE) [195, 196]. The EE is a sorting station from which receptors and lipids are shunted to different trafficking routes, including late endosomes (LEs) and lysosomes for degradation, retrograde trafficking to the trans‐Golgi, or recycling back to the PM [61, 197]. These trafficking pathways are tightly regulated, and impaired receptor trafficking has been linked to Alzheimer's disease, cardiovascular disease, various cancer types, and other illnesses [198, 199]. Receptors destined for the PM may be recycled directly from the EE, a process termed fast recycling, or they can initially be trafficked to a perinuclear endocytic recycling compartment (ERC), in a process referred to as slow recycling [61]. All four EHD paralogs have been identified as regulators of various steps of these trafficking pathways [60] (Figure 2).

Studies with the C. elegans homolog RME‐1 provided the first evidence that the EHD proteins regulate endocytic trafficking [6, 7]. In Chinese Hamster Ovary cells, EHD1 mutants impaired transferrin (Tf) receptor recycling and caused accumulation of cargo in the ERC [7]. Subsequent studies with the human homolog showed that EHD1 regulates the recycling of MHC‐I receptors from tubular recycling endosomes (TREs) [8]. Although overexpression of a dominant‐negative EHD1 mutant did not affect the internalization of cargos, it disrupted recycling of both clathrin‐dependent (Tf) and clathrin‐independent (MHC‐I) receptors [7, 8]. Since these initial studies, many receptors have been identified as being regulated by EHD1, including the cystic fibrosis transmembrane conductance receptor [15], β1 integrin [11], GLUT4 transporters [36], CD59 [12], MHC class II molecules [13], AMPA‐type glutamate receptors [14], and others (Table 1).

Beyond its role at the ERC, EHD1 has also been localized to RAB35‐positive EEs, where it regulates cargo trafficking from EEs to recycling endosomes in complex with CRMP2 and MICAL‐L1 [200, 201, 202]. Moreover, loss of EHD1 function leads to enlarged EEs [19, 203], elongation of TREs [204], accumulation of cargos in endosomes, and impaired receptor recycling [7, 8]. These findings raised the possibility that EHD1 plays a role in fission at endosomes (see Figure 5).

FIGURE 5.

Proposed role for EHD1 in mediating membrane fission. After cargo sorting, endosome budding is facilitated by BAR domain proteins and actin remodeling. MICAL‐L1 binds to membranes and recruits FCHSD2, which subsequently activates ARP2/3 to generate branched actin. MICAL‐L1 also recruits CIN85 and CD2AP, which may support removal of actin capping protein (not shown) to further promote branched actin. Once the vesicle is generated, to undergo fission Coronin1C and Coronin2A are recruited to debranch actin and clear the membrane for fission. EHD1 is recruited by MICAL‐L1. EHD1 undergoes oligomerization and promotes fission of the budding vesicle through its ATP hydrolysis.

The four EHD paralogs localize to different intracellular compartments and regulate different points of the endocytic trafficking pathway [8, 205, 206] (Figure 2). EHD1 and EHD3 have similar subcellular distribution patterns [8, 205]. While both EHD1 and EHD3 localize to EEs and TREs, EHD1 is enriched at TREs, and EHD3 has increased EE localization [8, 47, 62, 200, 205]. EHD1 has also been implicated in retromer‐based trafficking from the EE to the Golgi [207, 208, 209]. In contrast with EHD1 at TREs, EHD3 stabilizes tubular recycling endosomes [210]. Unlike EHD1 depletion, EHD3 knockdown leads to cargo accumulation in EEs rather than the ERC, highlighting EHD3 as a regulator of cargo trafficking from EE to ERC [47, 206]. EHD4 localizes to EEs but can also be observed on TREs [55, 206]. EHD4 depletion leads to EE enlargement, consistent with impaired trafficking and/or endosomal maturation and fission, and functional assays support a role for EHD4 in recycling to the PM as well as transport of cargo from EEs to LEs [55].

Although EHD1, EHD3, and EHD4 perform both distinct and overlapping functions in endocytic trafficking, heterodimerization among these proteins adds a level of regulatory complexity [47, 203, 206]. Indeed, EHD4 depletion reduced EHD1 recruitment to EEs and TREs, suggesting a cooperative role in regulating cargo trafficking [55, 203]. EHD1 and EHD3 heterodimerization has been proposed to facilitate cargo transport from EEs to TREs [47]. EHD2, however, has not been implicated in receptor recycling, but is a regulator of caveolae‐mediated endocytosis.

7.2. Caveolae Stabilization

Caveolae are small invaginations in the PM that participate in myriad cellular processes. These include endocytosis [211, 212, 213, 214], fatty acid uptake [215], mechanoprotection [216, 217, 218, 219], lipid homeostasis [220, 221, 222, 223], regulation of membrane composition [224, 225], and cellular signaling [226, 227, 228]. Assembly of caveolae occurs in cholesterol, sphingomyelin, and ceramide‐dense microdomains of the PM and is driven by the oligomerization of caveolin and cavin proteins [229, 230, 231]. Caveolin proteins (caveolin‐1–3) comprise the caveolar membrane coat and promote caveolar bulb curvature [232, 233, 234, 235], whereas cavin proteins are recruited from the cytoplasm to generate caveolae and stabilize the invaginated bulbs [234, 236, 237, 238].

EHD2 was first discovered in caveolar fractions via ultracentrifugation and was later shown to colocalize at PM puncta with caveolin‐1 and cavin‐1 [84, 239, 240, 241]. Whereas caveolin‐1 is evenly distributed around the invaginated structure, EHD2 specifically localizes closer to the neck of the invagination at the PM [241, 242, 243]. PM association of EHD2 is regulated by phosphatidylinositol 4,5‐bisphosphate levels, which are enriched at caveolae [244, 245]. EHD2 binds and hydrolyzes ATP, which regulates its dissociation from caveolae. Expression of EHD2 mutants with impaired ATPase activity significantly reduced the number of caveolae at the membrane [77, 239]. However, EHD2 must oligomerize in order to regulate caveolae [77, 241].

EHD2 depletion does not affect the formation or phenotype of caveolae in cells, suggesting a role in caveolar dynamics rather than their generation [239, 240, 241]. Indeed, caveolae in EHD2‐depleted cells have increased caveolae internalization and motility, supporting a role for EHD2 in caveolar stabilization at the PM [78, 239, 241, 246, 247]. EHD2 likely regulates caveolae mobility in cooperation with the actin cytoskeleton [241], and the EHD2 interaction partner EHBP1 is an actin regulatory protein that localizes to the caveolar neck, thus providing a basis for caveolae‐actin tethering [38, 245, 248]. Another EHD2 interactor, Syndapin2/PACSIN2, also localizes to caveolae to cooperatively promote caveolae stability and to prevent scission [240, 247].

The mechanism for EHD2 stabilization of caveolae has not fully been elucidated. However, it has been hypothesized that EHD2 increases the diameter of the caveolar neck to prevent lipid phase separation or the assembly of proteins required for protein‐mediated scission [247, 249, 250, 251]. EHD2 has a slow nucleotide hydrolysis rate, which may contribute to slow caveolar dynamics by increasing EHD2‐caveolae retention time [239, 252]. Lastly, diffusion of glycosphingolipids and cholesterol into caveolae stimulates endocytosis. It has been proposed that EHD2 modulates caveolae stability by regulating the diffusion of these lipids into caveolae [233, 253, 254, 255, 256].

Caveolae are important mechanosensory structures on the PM that adapt different conformations according to changes in membrane tension [216, 236]. High membrane tension leads to the flattening of caveolae, which is reversed upon release of mechanical stress [216, 236]. EHD2 depletion reduced caveolae recovery from mechanical stress, indicating it is required for re‐stabilization of caveolae after stress [257], and EHD2 undergoes SUMOylation upon mechanical stress, releasing it from the PM and facilitating nuclear translocation. EHD2 is also involved in caveolae mechanotransducer function [257].

While EHD2 has been the most extensively studied EHD protein at caveolae, other EHDs might have compensatory functions. Upon EHD2 knockdown, EHD1 and EHD4 localized to caveolae [251]. Depletion of EHD1, 2, and 4 decreased caveolae clustering on the membrane, increased caveolae mobility, significantly reduced neck diameter, and increased cellular sensitivity to PM rupture [251]. Elucidating the roles of EHD1 and 4 on caveolae dynamics requires further investigation.

7.3. EHD Proteins in Primary Ciliogenesis

Over the past decade, EHD proteins (primarily EHD1) have been implicated in the regulation of primary ciliogenesis (see Figure 3). The primary cilium (PC) is a single, immotile, antenna‐like organelle that is generated under conditions of stress to equip the cell to navigate various environmental stressors [168]. Structurally, the PC is a microtubule axoneme that extends from the mother centriole (m‐centriole), also known as the basal body [258, 259]. The axoneme is surrounded by a ciliary membrane likely derived from endocytic vesicles. This membrane fuses with the lipid bilayer at the PM to form a distinct lipid environment that is contiguous with, yet compositionally distinct from, the PM. Protruding into the extracellular environment, the PC is enriched in ion channels, G protein–coupled receptors, and receptor tyrosine kinases, which contribute to higher‐order processes including signal transduction, mechano‐transduction, and nutrient sensing [258, 259]. Impaired ciliogenesis manifests in over 30 different syndromes with affected eyes, kidneys, brains, and other organs, collectively referred to as ciliopathies [260].

Endocytic regulators have been directly connected to primary ciliogenesis, including Rab8 and Rab11 [261, 262, 263, 264]. Rabin8, a Rab8 guanine nucleotide exchange factor (GEF), is delivered by Rab11 to the centrosome to activate Rab8, which is required for ciliation [261, 262]. Based on interactions with the Rab8 effector MICAL‐L1, significant roles for EHD1 and EHD3 in ciliogenesis were identified [3]. EHD1 knockdown impaired PC formation in RPE‐1 cells, and both EHD1 and EHD3 knockdown impaired PC biogenesis in IMCD3 cells [3]. Moreover, EHD1 and EHD3 are both localized to the ciliary pocket membrane (CPM) [3]. EHD1 also colocalized with Rabin8‐positive preciliary vesicles, but since it associates with the m‐centriole even after Rab8 depletion, it likely functions upstream of the Rab8‐Rab11 cascade during ciliogenesis [3].

Early in ciliogenesis, Myo‐Va–associated preciliary vesicles dock at the distal appendages of the m‐centriole, where they later fuse to generate the ciliary vesicle [262, 265]. For fusion to occur, the centriolar capping protein CP110 must be removed from the m‐centriole [266], allowing axoneme extension and formation of the ciliary membrane [267, 268]. In the absence of EHD1, CP110 was retained on the m‐centriole during serum starvation, thus preventing ciliogenesis [3, 266].

The endosomal scaffold protein, MICAL‐L1, is also required for CP110 removal from the m‐centriole during ciliogenesis; intriguingly, it is required both for the recruitment of EHD1 to tubular recycling endosomes and to the m‐centriole [4]. One mechanism for CP110 removal is that EHD1 is required for the delivery of the E3 ligase HECT domain and RCC1‐like domain 2 (HERC2) on centriolar satellites to the m‐centriole, where the latter ubiquitinates CP110 and promotes its degradation. Indeed, depletion of HERC2 significantly reduced CP110 ubiquitination [269, 270], and EHD1 knockdown significantly decreased the interaction between HERC2 and CP110 [269]. How EHD1 regulates the movement of centriolar satellites remains an outstanding question, but may stem from its ability to regulate microtubule growth [271].

EHD1 may play additional roles in primary ciliogenesis. The EHD1‐interaction protein SNAP29 is a SNARE protein that mediates the fusion of membranes [69]. SNAP29 co‐localized with EHD1 to the ciliary pocket and DAVs, and its recruitment was EHD1‐dependent [3]. Another study showed that the EHD1‐SNAP29 interaction was dependent on EHD1 ATP binding and that expression of a GFP‐EHD1 mutant incapable of binding ATP or SNAP29 was unable to rescue ciliogenesis in EHD1‐depleted cells [107]. These findings support a model that EHD1 mediates SNAP29 recruitment to the mother centriole to promote the fusion of DAVs to form the ciliary vesicle during ciliogenesis [3].

Another EHD1‐interaction partner involved in PC regulation is Syndapin1, as knockdown of the latter impairs ciliogenesis [272]. Syndapin1 localizes with EHD1 on preciliary vesicles at the CPM, and both are present on membrane tubules extending from the CPM and m‐centriole [272]. Knockdown of either Syndapin1 or EHD1 significantly reduced the number of tubules extending from the m‐centriole, suggesting an additional role for EHD1 in ciliogenesis post CP110 cap removal [272]. Taken together, at least three distinct steps have been proposed for EHD1 in ciliogenesis: (1) interaction with MICAL‐L1 to promote HERC2 delivery for CP110 ubiquitination and degradation, (2) SNAP29‐mediated DAV fusion to form the ciliary vesicle, and (3) cooperation with Syndapin1 to promote tubulation of the ciliary vesicle and facilitate PM fusion.

EHD4 has also been implicated in primary ciliogenesis, potentially through its hetero‐oligomerization with EHD1 and EHD3 [55, 63, 107, 205]. To date, EHD2 is the only paralog that has not been implicated in ciliogenesis.

Intriguingly, EHD1 was first described as a candidate gene for Bardet‐Biedl syndrome (BBS1) [2], suggesting its potential involvement in ciliogenesis and ciliopathies. Now, over a quarter century later, the first human disease implicating a single amino acid substitution in the EHD1 protein has been identified in a polycystic kidney disease with features consistent with a ciliopathy [5]. Overall, these findings and the direct connection to ciliopathies support the notion that EHD proteins are not only key regulators of primary ciliogenesis but also play a wider role in cellular function than originally envisioned.

7.4. Centrosome Duplication/Cell Division

The involvement of EHD proteins in primary ciliogenesis suggests that they may also regulate additional cellular events. PC appears during the G1‐phase of the cell cycle, and ciliated cells are generally unable to progress through division. However, centrosome duplication occurs during S‐phase, which is required for cell division. Centriole engagement refers to the tight, orthogonal association of parent and progeny centrioles, allowing duplication only once [273, 274]. Upon exit from M‐phase, the centrioles in the newly segregated centrosomes lose their tight association in a process termed “disengagement” [274, 275]. It was previously shown that pericentrin (PCNT) cleavage and CEP215 removal are required for centriole disengagement during cell division [276, 277].

The function of endocytic proteins in the cell cycle has not been extensively studied [278]. However, several studies have shown a role for endocytic proteins in centrosome regulation [265, 279, 280, 281]. EHD1 depletion impaired centriole duplication in U2OS cells and reduced the percentage of disengaged centrioles from 65% in mock‐treated cells to less than 20% [282]. Upon EHD1 depletion, PCNT and CEP215 remained on the centrosome, with CEP215 absent from the spindle midbody [282, 283]. This is likely because EHD1 is required for the transport of CEP215‐containing vesicles away from the centrosome, a step necessary for centriole disengagement [282].

Consistent with a role in the earlier stages of centrosome duplication, EHD1 has also been implicated in cytokinesis. Indeed, EHD1 knockdown led to a significant increase in bi‐ and multi‐nucleated HeLa cells, reaffirming a dysregulated cell cycle [283]. EHD1‐depleted cells displayed both asymmetric cell spreading and cytokinesis failure [283]. EHD1 knockdown also impaired the trafficking of recycling vesicles to the intercellular bridge during cytokinesis, suggesting a role for EHD1 in both cytokinesis and pre‐cytokinetic events [282, 283].

7.5. Mitochondrial Homeostasis

EHD proteins have also been linked to mitochondrial function. Mitochondrial homeostasis, key to proper mitochondrial function, relies on continuous cycles of fission and fusion [284, 285]. Dysregulated mitochondrial dynamics manifest in neurological and neurodegenerative diseases, including Parkinson's and Huntington's disease [286, 287, 288]. A role for endosomal proteins in mitochondrial dynamics was demonstrated by overexpression of the retromer component VPS35, which led to shorter, fragmented mitochondria, whereas inhibition of VPS35 elicited longer, more stable mitochondria [289, 290]. The mechanism of VPS35‐mediated mitochondrial fission likely involves lysosomal transport of inactive dynamin‐related protein 1 (Drp1), a GTPase implicated in mitochondrial fission [289, 291, 292].

VPS35 is found in a complex with EHD1 [208]. Similar to VPS35 inhibition, EHD1 depletion led to elongated and more stable mitochondria [293]. EHD1's impact on mitochondrial fission is likely via VPS35; EHD1 depletion reduces VPS35 expression, favoring a model by which it cooperates with VPS35 and the retromer to facilitate Drp1 recycling from the mitochondrial membrane to promote inactive Drp1 turnover and mitochondrial fission [289, 293].

EHD2 has also been connected indirectly with mitochondrial function. Prohibitin (PHB‐1), a mitochondrial protein involved in mitochondrial biogenesis, mitophagy, and regulation of aerobic respiration, was identified as an EHD2 interaction partner [76, 294, 295]. Both EHD2 and PHB localize to lipid droplets (LDs) upon stimulation of lipolysis [81], and it was proposed that EHD2 might regulate PHB‐1 trafficking to mitochondria [76].

7.6. Lipid Droplet Dynamics

Lipid droplets (LDs) store fatty acids and neutral lipids and are important for maintenance of energy homeostasis. EHD2 was the first EHD paralog identified on LDs under both steady‐state conditions and upon lipolytic stimulation, suggesting that its localization is independent of lipid droplet structural reorganization during lipolysis [81]. It was later demonstrated that EHD2 localization to LDs relies on the presence of caveolin 1 (Cav1), highlighting an association between LDs and caveolae [74, 296]. Upon stimulation of lipolysis, the density of caveolae on the PM decreases [75], and EHD2 mRNA and protein levels decrease and coincide with reduced LD size [75]. Another study found that during mechano‐adaptation of adipocytes upon lipid loading, EHD2 transfer from caveolae to LDs is dependent on Cav1 phosphorylation [74]. This work highlights mechanical crosstalk between caveolae and LDs, allowing adipocytes to adapt to highly dynamic cellular energy states.

EHD2 mediates the size of LDs by regulating fatty acid uptake through caveolae. While EHD2 knockout mice display fewer LDs per cell in brown adipocytes, they have significantly larger LDs in these cells, likely due to enhanced fatty acid uptake [80]. EHD2 regulation of fatty acid uptake and LD size is dependent on membrane association, oligomerization, and ATPase activity [80], which also controls surface‐associated LDs and their lipolytic activity [73]. EHD1 has also been localized to LDs in mouse embryonic fibroblasts. Indeed, cells derived from EHD1 knockout mice had reduced levels of both total and free cholesterol, potentially as a result of decreased uptake of LDL particles resulting from EHD1 regulation of LDL receptor recycling [33].

Lipophagy is a macroautophagic process for LD degradation; an LD is enveloped in an autophagophore, leading to free fatty acid release [297]. During stress‐induced lipophagy, Rab10 is activated and localized to LC3‐positive autophagic membranes. Activated Rab10 recruits EHBP1 and EHD2 to autophagic membranes, and this trimeric complex promotes autophagic membrane extension during lipophagy, required for LD engulfment [83]. Depletion of Rab10, EHBP1, or EHD2 significantly impaired LD catabolism [83]. EHD2 likely participates by deforming and remodeling the autophagic membrane, but further studies into its mechanism of action during lipophagy are required.

8. Mechanisms of EHD Function

8.1. EHD Structure and Homology

The N‐terminus of Eps15 has three non‐identical repeats of ~100 residues, which were termed EH domains [298, 299]. EH domains are highly homologous and contain two EF‐hand helix‐loop‐helix motifs connected by an antiparallel β‐sheet [300]. These domains tightly bind asparagine–proline–phenylalanine (NPF) motifs [301, 302]. Using nuclear magnetic resonance (NMR) studies, it was shown that EH domain‐NPF binding is mediated by a conserved hydrophobic pocket of the EH domain and binding is stabilized by hydrophobic and electrostatic interactions [60, 303, 304] (see Figure 1B). Since the initial characterization of the EH domain, over 50 other proteins containing EH‐domains have been identified, including Eps15R, intersectin‐1 and intersectin‐2, REPS1, and others [305]. Many of the EH domain‐containing proteins are involved in endocytosis and endocytic trafficking pathways [306].

Of the EH‐domain‐containing proteins, there are only four highly homologous mammalian paralogs in which the EH domain is localized to the C‐terminus [1, 307] (Figure 1A). Among these paralogs, EHD1 and EHD3 display 86.5% identity at the amino acid level, and the weakest identity is between EHD2 and EHD4, at 67.9% [60, 307] (Figure 1C). These proteins are also characterized by an ATP‐binding G‐domain (Figure 1B) [47, 60, 63, 252]. The two helical regions come together to form a helical domain, which mediates lipid binding [252]. The ATP‐binding G‐domain mediates nucleotide binding and hydrolysis and is homologous to the GTP‐binding domain of Dynamin [192, 252]. In addition to nucleotide binding, the G‐domain also mediates oligomerization. While all EHDs can homo‐oligomerize, EHD1 hetero‐oligomerizes with EHD3 and EHD4; EHD3 preferentially hetero‐oligomerizes with EHD1 but can also interact with EHD4; and EHD4 shows a higher propensity to hetero‐oligomerize with EHD1 [47, 203, 206]. EHD2, the least homologous paralog, favors homo‐oligomerization via a hydrophobic interface of the G‐domain, and mutations within the G‐domain impair dimerization [252, 308]. Mutations within the EHD1 G‐domain impair its ability to both homo‐ and hetero‐oligomerize with both EHD3 and EHD4, and loss of nucleotide binding abrogates EHD1/EHD3 hetero‐ and homo‐oligomerization [47, 203]. While EHD oligomerization is EH domain‐independent, the EH domain mediates interactions with other proteins (Table 2).

TABLE 2.

EHD protein interaction partners.

EHD paralog	Interacting protein	Interaction region	Interaction region (EHD1)	Biological function	Experimental method	References
EHD1	MICAL‐L1	NPF motif (1)	EH domain	Receptor recycling—endosome fission; EHD1 recruitment to primary cilium	LC‐MS/MS; Y2H; GST pull‐down	[309, 310]
	Rabenosyn‐5	NPF motif (1+2)	EH domain	Receptor recycling	LC‐MS/MS; GST pull‐down	[10]
	Rab11‐FIP2	NPF motif (2)	EH domain	Receptor recycling from the perinuclear endocytic recycling compartment	Y2H	[47]
	EHD1	Coiled‐coil region	Coiled‐coil region	Protein localization and receptor recycling	Y2H	[63, 205]
	EHD3	Coiled‐coil region	Coiled‐coil region	Protein localization and receptor recycling	Y2H; Co‐IP	[205]
	SNAP29	NPF motif	EH domain	Receptor recycling, IGF‐1R receptor internalization	Co‐IP; GST pull‐down	[69, 311]
	Syndapin I	NPF motif	EH domain	Receptor recycling	Co‐IP; GST pull‐down	[312]
	Syndapin II	NPF motif	EH domain	Receptor recycling	Co‐IP; GST pull‐down	[311, 312, 313]
	Snapin	n/a	EH domain	Negatively regulates exocytosis by preventing binding to SNAP25	Co‐IP	[92]
	Fer‐1‐like‐5 (Fer1L5)	NPF motif	EH domain	Myoblast fusion + recycling + muscle repair	Co‐IP; GST pull‐down and radioactive labeling	[30]
	Rabankyrin‐5	NPF motif	EH domain	Retrograde trafficking	Y2H; GST pull‐down; Co‐IP	[208]
	Epsin 1	NPF motif	EH domain	Endocytosis	GST pull‐down	[314]
	Epsin 3	NPF motif	EH domain	Endocytosis	GST pull‐down	[314]
	Cep215	n/a	n/a	Centrosome duplication	Co‐IP	[282]
	Phostensin	64‐ILV(XXXX)LRLS‐75	n/a (EH domain‐independent)	Connects EHD1 and actin during recycling	GST pull‐down	[142]
	Cx43	n/a	n/a	Cardiac remodeling in ischemia	Co‐IP	[24]
	CD44	n/a	n/a	Promotes cancer stemness	Co‐IP	[25, 26]
	Triad1	n/a	EH domain	Neurite growth during spinal cord injury	Co‐IP	[91]
	Insulin‐like growth factor 1 receptor	n/a	n/a	Localization on endocytic vesicles	EHD1 affinity column purification	[69]
	AP‐2 α‐adaptin	n/a	n/a	Localization on endocytic vesicles	EHD1 affinity column purification	[69]
	Clathrin heavy chain	n/a	n/a	Localization on endocytic vesicles	EHD1 affinity column purification	[69]
	Vps35/Vps26/Vps29 complex	n/a	n/a (EH domain‐independent)	Retrograde trafficking	Co‐IP	[315]
	KCa2.3	n/a	n/a	Receptor recycling	Co‐IP	[17]
	Kazrin C	C‐terminal	n/a	Receptor recycling (EE‐ERC)	GST pull‐down	[316]
	cPLA2	n/a	n/a	Receptor recycling (endosome fission)	Co‐IP	[12]
	Amphiphysin	NPF (1) and (2)	EH domain	Receptor recycling	Y2H; GST pull‐down	[192]
	Numb	NPF motif (C‐terminal domain)	EH domain	Clathrin‐independent receptor, Tac recycling impaired in Numb siRNA KD cells (CHO)	Co‐IP	[184]
	Sortilin	n/a	n/a	Sortilin stabilized in macrophages by EHD1 interaction	Co‐IP	[28]
	Programmed death‐ligand 1 (PDL1)	n/a	n/a	Immune evasion in lung adenocarcinoma (A549) cells	Co‐IP	[146]
EHD2	Myoferlin	NPF motif	EH domain	Myoblast fusion + recycling + muscle repair	Co‐IP; GST pull‐down + radioactive labeling	[39, 40]
	Fer‐1‐like‐5 (Fer1L5)	NPF motif	EH domain	Myoblast fusion + recycling + muscle repair	Co‐IP; GST pull‐down + radioactive labeling	[30]
	Prohibitin	n/a	n/a	Regulation of mitochondrial metabolism	Protein crosslink and MALDI‐TOF	[76]
	Rabenosyn‐5	NPF motif	EH domain	Not explored	Y2H	[10]
	EHBP1	NPF motif	EH domain	Actin regulation during endocytosis	GST pull‐down; MALDI‐TOF	[36]
	Epsin 1	NPF motif	EH domain	Endocytosis	GST pull‐down	[314]
	Epsin 3	NPF motif	EH domain	Endocytosis	GST pull‐down	[314]
	Nek3	Coiled‐coil domain	n/a	Modulation of Rac1 activity	Y2H; Co‐IP	[317]
	GLUT4	n/a	n/a	GLUT4 PM trafficking	Co‐IP	[70]
	AP‐1 μ1	n/a	n/a	Not explored	Co‐IP	[70]
	AP‐2 μ2	n/a	n/a	Not explored	Co‐IP	[70]
	CALM	n/a	n/a	Not explored	Co‐IP	[70]
	Caveolin‐1	n/a	n/a	Caveolae stability	Co‐IP	[239]
	Cavin‐1	n/a	n/a	EHD2‐caveolae assembly	GST pull‐down	[239]
EHD3	Rabenosyn‐5	NPF motif	EH domain	Receptor recycling	Y2H	[10]
	Ankyrin‐B	Membrane‐binding domain	Coiled‐coil domain	Stabilizes Na/Ca channels at PM in cardiomyocytes for conductance	GST pull‐downs	[52, 102]
	MICAL‐L1	NPF motif	EH domain	Receptor recycling	Y2H	[309]
	Rab11‐FIP2	NPF motif (2)	EH domain	Receptor recycling from the perinuclear endocytic recycling compartment	Y2H	[47]
	Syndapin I	NPF motif	EH domain	Receptor recycling	GST pull‐downs	[312]
	Syndapin II	NPF motif	EH domain	Receptor recycling	GST pull‐downs	[312]
	Kazrin C	C‐terminal	n/a	Receptor recycling (EE‐ERC)	GST pull‐down	[316]
EHD4	Numb	NPF motif (C‐terminal domain)	EH domain	Clathrin‐independent receptor, Tac recycling impaired in Numb siRNA KD cells (CHO)	Co‐IP; GST pull‐down	[184]
a	Cadherin 23	n/a (non‐NPF motif)	EH domain	EHD4 and CDH23 (protein involved mechanotransduction in Cochlear hair cells) interact	Y2H; Co‐IP	[109]
	Phostensin	64‐ILV(XXXX)LRLS‐75	n/a (EH domain‐independent)	Connects EHD4 and actin during recycling	GST pull‐down	[142]
	EHD1	Helical domain	Helical domain	Endosome fission	Y2H	[203]
	Rabenosyn 5	NPF	EH domain	EHD‐endosome localization; receptor recycling	Y2H	[203]
	Syndapin II	NPF	EH domain	EHD‐endosome localization	Y2H	[203]
	Syndapin I	NPF	EH domain	Receptor recycling	GST pull‐down	[312]
	Type VI collagen	n/a	n/a	Not explored	Type VI collagen affinity blot	[103]

8.2. C‐Terminal EH‐Domain Binding

All EH domains interact with proteins containing the tripeptide NPF motif. However, the C‐terminal EHDs display selectivity compared to the rest of the EH‐domain‐containing proteins due to the highly positively charged surface area of the C‐terminal EHDs [60]. For example, the EH domain of EHD1 has a more positively charged surface potential than the EH‐2 domain of Eps15, highlighting a preference for binding to NPF motifs flanked by acidic residues, which form salt‐bridges with the EH domain [309, 310, 318, 319].

For example, the endosomal scaffold protein MICAL‐L1 is a crucial EHD1 interaction partner that contains two NPF motifs, with only the first flanked by acidic residues (NPFEEEEED) and required for EHD1 binding [309, 310]. Additional EHD1‐binding partners that are Rab effectors and have acidic clusters following their NPF motifs include Rabenosyn‐5, Rab11‐FIP2, and Rabankyrin‐5 [10, 47, 208]. In addition to NPF motifs, the EH domain of EHD1 can bind DPF and GPF motifs, and the EH domain of EHD2 can bind GPF motifs, but both bind at a lower affinity than NPF motifs [252, 320].

EHD protein–protein interactions are also mediated through EH domain‐independent mechanisms. For example, EHD1 and EHD4 bind phostensin independently of their EH domains [140, 293], and the worm EHD paralog Rme‐1 can bind to ALX‐1 both via its NPF motif and independently of EH‐NPF interactions [189].

8.3. Intracellular Localization and Membrane Association

Despite the sequence homology, the four mammalian EHD paralogs localize to distinct intracellular locations (Figure 2). EHD1 and EHD3 localize to tubular recycling endosomes and Rab11‐positive endosomes containing internalized transferrin and MHC‐I proteins [1, 7, 8, 321]. EHD2 is found at the PM with caveolae, and EHD4 is primarily localized to EEA1‐ and Rab5‐positive early endosomes [55, 321]. EHD1, EHD3, and EHD4 also localize to the primary cilium [3, 4, 107].

A major feature of the endocytic trafficking pathway is the establishment of compartmental identity, which is regulated by local phosphoinositide levels [322]. EHD proteins are capable of directly binding to a variety of phosphatidylinositol moieties as well as phosphatidylserine in a mechanism more dependent on phospholipid charge than structure [192, 252, 321, 323, 324, 325, 326].

The helical domains of EHDs mediate phospholipid binding, and mutation of residues in the central helical region significantly reduces EHD2 association with liposomes in vitro and causes a cytosolic redistribution in cells [252]. Expression of EHD1–4 mutants with abrogated nucleotide binding in cells impairs their membrane association and leads to cytosolic distribution, demonstrating that EHD membrane association is mediated through nucleotide binding [7, 8, 47, 252, 321]. Indeed, EHD2 binding to a lipid monolayer was enhanced when EHD2 was preincubated with the nonhydrolyzable ATP analog AMP‐PNP, supporting the role of ATP binding for membrane association [77]. EHD2 association with membranes is regulated by its ATP cycle: ATP‐bound EHD2 binds to membranes and oligomerizes, and upon ATP hydrolysis there is oligomer disassembly and membrane release [77]. Interestingly, EHD2‐membrane binding stimulates an eightfold increase in its ATPase activity, and EHD1‐membrane binding increases its ATPase activity almost fourfold [252, 326].

Membrane curvature is another important aspect of EHD‐membrane binding. EHD2 selectively binds to smaller liposomes with higher curvature, and EHD1 is preferentially retained on membrane tubules over flat, lipid bilayers [252, 326]. This is in line with the cellular localization of EHD2 to the curved caveolar necks and EHD1 localization to tubular recycling endosomes. Hetero‐oligomerization also regulates membrane binding, as shown by a redistribution of EHD1 in cells upon either EHD4 or EHD1 knockdown [55]. Finally, NMR studies demonstrated that the EH domains of EHD1 and EHD4 directly bind phosphoinositides, suggesting the EH domain is also involved in EHD localization and membrane association [323]. Consistent with these in vitro data, introduction of EHD1‐4 with EH domain deletions into cells disrupted localization to membranes [47, 206, 252, 321, 323]. These findings suggest that EHD protein localization within cells is regulated, at least in part, by their EH domains.

8.4. Membrane Fission

EHD1 contains a dynamin‐like nucleotide‐binding domain that binds and hydrolyzes ATP (242, 301). Although early work suggested possible GTP binding based on limited homology to H‐Ras [8], subsequent structural and biochemical studies established that the EHD G‐domain more closely resembles that of dynamin‐like ATPases [192, 252]. From the crystal structure of the EHD2 dimer, the dimerized helical domains adopt a “scissor shape” that binds membranes [252]. EHD‐induced membrane remodeling was first seen when EHD2 was incubated with liposomes, causing significant tubulation [252, 327]. Both the worm RME‐1 protein and the EHD1 Lampetra fluviatilis homolog tubulated liposomes in vitro [95, 192]. Moreover, EHD1 incubation with liposomes in the presence of ATP induced liposome vesiculation [204]. In addition, studies in cells demonstrated that EHD1 vesiculates tubular recycling endosomes [12, 328]. Consistent with the delayed recycling of a variety of cargo in EHD1‐depleted cells [7, 8, 9, 11], EHD1 is required for the fission of both canonical and tubular recycling endosomes.

Recent in vitro studies have shed new light on the mechanism of EHD1 involvement in membrane fission. When purified EHD1 was incubated with lipid tubules that extended from a flat, supported lipid bilayer, ATP‐bound EHD1 oligomerized to negatively charged regions, causing membrane sequestration and bulging, leaving the intervening regions of membrane with severe thinning. Ultimately, the membrane tubules thinned to a ~5 nm radius (a critical thickness for spontaneous membrane fission) and underwent fission [326, 329, 330, 331]. This highlights a key mechanistic difference between EHD1‐ and Dynamin1‐mediated membrane fission, as Dynamin1 clusters on membranes, causing membrane thinning in the regions to which it is bound [326]. There was also limited scission of tubules incubated with EHD2 and ATP, but the majority of tubules did not undergo scission, and the ones that did displayed significantly delayed scission, owing to a 40‐fold lower ATPase activity of EHD2 compared to EHD1 [326].

Unlike in vitro studies with purified proteins and liposomes, endosome fission is a complex process in cells that requires the coordination of sequential events by multiple proteins. A current model for endosome fission suggests that the WASH complex is recruited to the neck of budding vesicles on endosomes via the retromer, where it promotes ARP2/3‐mediated branched actin polymerization [332, 333, 334]. During the early stages of endosome fission, actin provides a physical barrier to facilitate cargo sorting and provides a necessary pushing force to generate the appropriate membrane tension required during fission [332, 333, 335, 336, 337]. Negative regulators of actin polymerization, including Coronin 1C [338, 339] and Coronin 2A [340], are recruited during later stages to promote disassembly of the actin network. Accordingly, this facilitates EHD1 recruitment by MICAL‐L1 to the neck of the budding vesicle where it promotes fission (Figure 5) [309, 326, 335, 337, 338, 339, 340, 341, 342].

8.5. The Next Quarter Century of EHDs and Outstanding Questions

Despite significant progress in defining the cellular functions of EHD proteins, several fundamental questions remain unresolved. At a mechanistic level, it is still unclear how EHD‐mediated membrane fission is spatially and temporally regulated within cells, and how ATP binding and hydrolysis are coordinated with membrane remodeling. In particular, how EHD proteins coordinate fission events with the actin and microtubule cytoskeleton is largely unknown, although EHD1 binds tubulin and regulates microtubules [271]. Furthermore, whether EHD proteins act autonomously or in concert with additional fission machinery proteins, and how these activities are modulated by lipid composition and post‐translational modifications such as phosphorylation and ubiquitination, remain important areas for investigation.

Another major challenge is understanding functional specificity and redundancy among mammalian EHD paralogs. Although EHD1 and EHD3 share high sequence similarity, they appear to fulfill non‐overlapping roles in certain contexts. How paralog‐specific interactions, expression patterns, and regulatory mechanisms contribute to these differences is not yet well defined. In parallel, the extent to which findings from invertebrate models with a single EHD ortholog can be extrapolated to mammalian systems requires further clarification.

From a physiological and clinical perspective, the recent identification of human disease linked to EHD1 mutations highlights the need to systematically explore the contribution of EHD proteins to tissue‐specific functions and pathologies. Whether additional EHD variants underlie unexplained ciliopathies, kidney disorders, or neurodevelopmental phenotypes remains an open question. Finally, emerging links between EHD proteins and processes such as centrosome dynamics, mitochondrial homeostasis, and cytoskeletal regulation suggest that their functional repertoire may extend well beyond endocytic trafficking, warranting broader investigation.

Addressing these questions will require integrated approaches combining classic biochemistry, structural biology, advanced imaging, genetic models, and human genomics. Together, such efforts promise to refine our understanding of EHD protein function and reveal new principles governing membrane dynamics in health and disease.

Funding

This work was supported by National Institute of General Medical Sciences (Grant R35GM144102) and the NIH Cancer Biology Training Program (Grant CA009476).

Ethics Statement

This article is a review article and does not contain any new studies with human participants or animals performed by any of the authors.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.