Altered Endoplasmic Reticulum Integrity and Organelle Interactions in Living Cells Expressing INF2 Variants

Abstract

The cytoskeleton mediates fundamental cellular processes by organizing inter-organelle interactions. Pathogenic variants of inverted formin 2 (INF2) CAAX isoform, an actin assembly factor that is predominantly expressed in the endoplasmic reticulum (ER), are linked to focal segmental glomerulosclerosis (FSGS) and Charcot-Marie-Tooth (CMT) neuropathy. To investigate how pathogenic INF2 variants alter ER integrity, we used high-resolution live imaging of HeLa cells. Cells expressing wild-type (WT) INF2 showed a predominant tubular ER with perinuclear clustering. Cells expressing INF2 FSGS variants that cause mild and intermediate disease induced more sheet-like ER, a pattern similar to that seen for cells expressing WT-INF2 that were treated with actin and microtubule (MT) inhibitors. Dual CMT-FSGS INF2 variants led to more severe ER dysmorphism, with a diffuse, fragmented ER and coarse INF2 aggregates. Proper organization of both F-actin and MT was needed to modulate the tubule vs. sheet conformation balance, while MT arrays regulated spatial expansion of tubular ER in the cell periphery. Pathogenic INF2 variants also induced mitochondria fragmentation and dysregulated mitochondria distribution. Such mitochondrial abnormalities were more prominent for cells expressing CMT-FSGS compared to those with FSGS variants, indicating that the severity of the dysfunction is linked to the degree of cytoskeletal disorganization. Our observations suggest that pathogenic INF2 variants disrupt ER continuity by altering interactions between the ER and the cytoskeleton that in turn impairs inter-organelle communication, especially at ER-mitochondria contact sites. ER continuity defects may be a common disease mechanism involved in both peripheral neuropathy and glomerulopathy.

Keywords: cytoskeleton; endoplasmic reticulum; glomerulosclerosis; mitochondria; podocyte.

Figure 1

Domain structure and locations of disease variants of human INF2. (A) Human INF2 is a multidomain and homo-dimeric protein that mainly consists of two major components: a regulatory domain that comprises DID and DAD and an actin-organizing unit that includes the FH1 and FH2 domains. The amino acid numbering is shown below each box. There are two isoforms of human INF2: INF2-CAAX (ER-resident) and INF2-non-CAAX (cytoplasmic form) [18,36]. INF2-CAAX predominates in the kidney [30]. INF2 variants exclusively cluster in the DID domain: CMT + FSGS variants are located in the N-terminal half of DID Leu57-Glu183, whereas FSGS variants are found in the C-terminal half of the DID Glu184-Leu245. DID: Diaphanous inhibitory domain. DAD: Diaphanous autoregulatory domain. FH1: Formin homology 1. FH2: Formin homology 2. (B) Schematic diagram showing the domain organization of the ER, with tubular and sheet morphologies. ER membranes are enriched in the perinuclear region (sheet ER), whereas the peripheral ER spreads throughout the cytoplasm as an interconnected tubular network. (C) Subdomains of the peripheral ER. Three components comprising the peripheral ER are shown. A tubule-to-sheet conformation is dynamically regulated by ER proteins, as well as actin-microtubule networks. A cluster of tubules forms the matrix [1,2,5].

Figure 2

Wild-type INF2 localizes in the peripheral ER in living HeLa cells. (A) Colocalization of wild-type INF2 with the ER compartment. HeLa cells were transiently co-transfected with eGFP tagged, wild-type INF2 (WT, CAAX isoform) (green) and the ER-marker mCherry-calreticulin (red). High-resolution imaging with DragonFly spinning-disk microscopy revealed that WT-INF2 resides in a dispersed reticular network having both tubule and sheet structures (asterisk) that were labeled with the ER marker calreticulin. These structures appear as lace-like, interconnected tubules with a three-way junction (TWJ) and perinuclear sheet. The ER pattern in WT-INF2-expressing HeLa cells is indistinguishable from the control cells expressing calreticulin alone. Bars = 10 µm and 1 µm. (B). Quantitative analysis of colocalization of WT-INF2 with the ER compartment. Colocalization of INF2 with the ER marker (calreticulin) and nuclei (Hoechst) was analyzed using Fiji (version 2.15.1, NIH). The proportion of the merged punctae was calculated using the Manders coefficient. INF2 punctae co-localize preferentially with the ER marker, and the distribution is distinct from the nuclear compartment (p ≤ 0.0001). Data were analyzed by ANOVA test (Prism 8, n = 10 images per subgroup); ns, not significant; ****, p ≤ 0.0001.

Figure 3

ER morphology in living HeLa cells expressing wild-type and pathogenic INF2 variants. eGFP tagged, wild-type INF2 (WT, CAAX isoform) and pathogenic variants R218W (FSGS), T161N (intermediate), and G73D (CMT + FSGS) (green) were transiently co-transfected with the ER-maker mCherry-calreticulin in HeLa cells. High-resolution live images were captured with a DragonFly microscope. Boxed areas in the peripheral ER are magnified. Cells expressing WT-INF2 show a disperse ER pattern composed of a lace-like, interconnected tubule meshwork. Cells expressing FSGS variants (R218W and T161N) exhibit a predominant sheet-like pattern with fine granular aggregates, while some residual tubular structures remain (asterisk). The clustering of tubules occurs frequently at the cell edges. In contrast, cells expressing the CMT + FSGS variant (G73D) accumulate more sheet-like materials with coarse granular aggregates, which often appear to be fragmented or swollen (double asterisk). Boxes indicate the magnified areas. Bars = 10 µm and 1 µm.

Figure 4

Comparison of ER phenotypes among live HeLa cells expressing wild-type or variant INF2. (A) Quantification of ER patterns. Representative images of three ER patterns in live HeLa cells expressing WT-INF2 and pathogenic variants are shown. The diffuse reticular ER morphology usually seen in normal control cells is classified as Tubular ER “Class 1”; “Class 2 Mixed tubule and sheet” is defined by a mixed pattern in which tubules coexist with sheets; “Class 3 Sheet” shows predominantly round or ellipsoid-shaped sheet-like materials and aggregation. Boxes present the magnification areas. Cells expressing WT-INF2 or pathogenic variants were categorized by visual inspection, and the proportion of the subtypes was compared. At least 50 cells from at least three experiments were analyzed for each INF2 variant. Statistical analysis was performed using Fisher’s exact test (R, version 4.3.0, 2023); *, p ≤ 0.05; ****, p ≤ 0.0001. (B) Comparison of INF2-ER colocalization among INF2 variants. Intensity correlation of INF2 wild-type and pathogenic variants (eGFP) with an ER marker (mCherry-calreticulin) was analyzed using the Manders coefficient (Fiji, version 2.15.1, NIH). The proportion of INF2 colocalization with the ER was compared between cells expressing WT-INF2 and cells expressing the pathogenic variants. Mild variants R218W and N202S co-localize preferentially with the ER to a similar degree as that for WT-INF2. In contrast, cells expressing the severe G73D variant showed much less ER colocalization (p < 0.05). The T161N variant showed an intermediate colocalization frequency. Data were analyzed by an ANOVA test (Prism 8, n = 10 images/group); ns, not significant; *, p ≤ 0.05. The schematic representation shows hypothetical models for INF2 colocalization with ER markers.

Figure 5

Effects of the actin inhibitor CytoD on the peripheral ER pattern in living HeLa cells expressing wild-type INF2 variants. Living HeLa cells were transiently transfected with eGFP-INF2 WT (green) and either mScarlet-LifeAct or mCherry-calreticulin (red). The effects of actin polymerization in the ER morphology were examined after treating the cells with cytochalasin D (CytoD; 1 mM for 30 min) [5]. Before CytoD treatment, cells expressing eGFP-WT-INF2 generated robust, central stress fibers (arrows), as well as peripheral bundles (arrowheads) to produce a disperse, reticular pattern of ER that was labeled with eGFP-INF2. After CytoD treatment, the ER pattern in the WT-INF2 cells acquired a sheet-like appearance (S) rather than tubular structures (T), as labeled with eGFP-INF2. The calreticulin labeling showed a predominant sheet-like pattern with scattered granular aggregates. The data indicate that actin depolymerization disrupts the tubule ER pattern in WT-INF2 expressing cells, leading to a tubule-to-sheet ER transformation. Boxes indicate the magnified areas. Bars = 10 µm and 1 µm.

Figure 6

Relationship of actin organization and peripheral ER structures in living HeLa cells expressing wild-type or pathogenic INF2 variants. Living HeLa cells were transiently transfected with eGFP-WT-INF2, intermediate (T161N), or severe (G73D) variants (green) and either an F-actin marker (mScarlet-LifeAct) or an ER marker (mCherry-calreticulin, red). T161N variant cells generated fewer central actin cables (arrowheads) and instead have more sheets or granular ER components, in addition to small punctate aggregation of INF2. G73D variant cells produced shorter and thinner F-actin filaments (arrows) and accumulated more coarse granular sheet-like ER components with fragmented INF2 aggregates in the peripheral ER. The predominant sheet-like pattern mirrored that of control WT-INF2 cells treated with CytoD to induce actin depolymerization (Figure 5). Boxes indicate the magnified areas. Bars = 10 µm and 1 µm.

Figure 7

Effects of a microtubule (MT) inhibitor on the peripheral ER structure in living HeLa cells expressing wild-type and pathogenic INF2 variants. HeLa cells were transiently cotransfected with eGFP-INF2 WT (green) and mScarlet EMTB (red). Cells expressing eGFP-WT-INF2 showed a reticular, tubular ER pattern, along with an MT array having appropriate spacing. The effects of MT depolymerization on ER morphology were examined by treating cells with nocodazole (Noc, 2.5 µg/mL) for 30 min. In eGFP-WT-INF2 cells, INF2-labeled ER structures form an expansive network reaching to the farthest edge of the cells. Noc-induced MT depolymerization altered the tubular structure (T) such that it was sparser in some areas. Sheet (S) or matrix structures formed in the remaining network. There was partial retraction of the ER network towards the cell center (asterisk), suggesting a role for MT in enabling the ER to expand throughout the cell. Boxes indicate the magnified areas. Dashed lines depict the cell contour. Bars = 10 µm and 2 µm.

Figure 8

Effects of a microtubule inhibitor on the peripheral ER in living HeLa cells expressing wild-type or pathogenic INF2 variants. HeLa cells were transiently cotransfected with eGFP-INF2 (WT, T161N, G73D) (green) and mCherry-calreticulin (red). The effects of inhibiting MT polymerization on ER morphology were examined by treating cells with nocodazole (Noc, 2.5 µg/mL) for 30 min. In eGFP-WT-INF2 cells, Noc decreased peripheral tubular components with concomitant mild and focal enrichment of sheet or matrix structures, while leaving tubule structures intact in some areas (asterisk). T161N cells with Noc treatment show more tiny granular sheet-like ER components (mCherry-calreticulin) and diffusely disperse INF2 distribution (eGFP). G73D cells accumulated more coarse, sheet-like ER materials and punctate INF2 aggregates throughout the cytoplasm than T161N cells. Boxes indicate the magnified areas. Bars = 10 µm and 1 µm.

Figure 9

ER–mitochondria interaction in living HeLa cells expressing WT-INF2 or pathogenic variants. (A). ER and mitochondria colocalization. ER and mitochondria were labeled by eGFP-INF2 and MitoTracker, respectively. Cells expressing WT-INF2 had lace-like polygonal ER tubules spread towards the periphery, with mitochondria aligned in the perinuclear region, forming ample ERMCSs (ER–mitochondria contact sites). Cells expressing FSGS variants (R218W, N202S, T161N) exhibited a predominant aberrant mitochondria cluster (C) at the periphery, with fragmentation and disrupted ERMCSs. T161N variant cells showed more pronounced misdistribution and fragmentation than WT. CMT + FSGS variant (G73D) cells showed even more marked dysmorphic ER changes, with a diffuse, coarse granular ER appearance, leading to severe dissociation (D) of ERMCS. Bars = 10 µm and 2 µm. (B). Three subclasses of mitochondrial shape were observed: Class 1: predominant tubular-shape, Class 2: mixture of tubular and fragmented mitochondria, and Class 3: fragmented. Mitochondria were more fragmented in cells expressing INF2 variants (R218W, N202S, T161N, G73D) than those expressing WT-INF2. (C) Mitochondrial distribution is subclassified as either perinuclear (normal distribution) or peripheral (misdistribution). Data are from three independent experiments (n ≥ 30 cells per each variant); ns, not significant; *, p ≤ 0.05.

Figure 10

Impaired mitochondrial respiratory function in HeLa cells expressing INF2 variants. Living HeLa cells were co-transfected with eGFP-tagged WT-INF2, T161N, or G73D variants. After 12 h, cells were analyzed for mitochondrial function by measuring the OCR in response to the indicated reagents. (A) Diagram of mitochondrial respiration. Cells were treated consecutively with a series of complex inhibitor or coupling reagent (oligomycin, carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone (FCCP), and rotenone + antimycin A) and measured with a Flux analyzer (XFp Agilent). Data are normalized by transfection efficiency, and the mean ± SE from three independent experiments is shown. (B) Comparison of mitochondrial respiration for INF2 variants. WT-INF2 cells exhibited high basal respiration and respiratory capacity during the treatments, while G73D variant cells showed much lower respiratory parameters, suggesting more severely compromised mitochondrial function (p ≤ 0.05). Cells expressing the T161N variant had an intermediate performance. Data were analyzed by ANOVA test (Prism 8, n = 3 experiments); ns, not significant; *, p ≤ 0.05.

Figure 11

Effects of INF2 variants on ER morphology and organelle contacts. (A) ER integrity and ER–organelle interactions. Under physiological conditions, the ER is canonically classified as two simple structures: tubule (peripheral ER) and sheet (perinuclear cisterna ER). Tubular ER exhibits a reticular pattern with polygons connected by a three-way junction (TWJ). The ER network is mainly shaped and maintained by interplay with actin and MT. The ER networks contact organelles in close opposition and contribute to a variety of processes, including (1) mitochondria dynamics, (2) vesicle trafficking (with arrow shows the movement direction), and (3) plasma membrane Ca exchange. (B) Schematic of representative ER pattern in cells expressing INF2 variants. WT-INF2 cells showed a disperse reticular network pattern, appearing as lace-like interconnected tubules with a TWJ and occasional sheet. FSGS variant cells have strikingly altered ER patterns, evidenced by more abundant sheet-like structures with polygonal structures left intact in some areas. Cells expressing CMT + FSGS variants show a diffuse dysmorphic ER appearance, with a loss of continuity marked by irregular dilatation and fragmentation.