Arp2/3 mediated dynamic lamellipodia of the hPSC colony edges promote liposome-based DNA delivery

Abstract

Cationic liposome-mediated delivery of drugs, DNA, or RNA plays a pivotal role in small molecule therapy, gene editing, and immunization. However, our current knowledge regarding the cellular structures that facilitate this process remains limited. Here, we used human pluripotent stem cells (hPSCs), which form compact colonies consisting of dynamically active cells at the periphery and epithelial-like cells at the core. We discovered that cells at the colony edges selectively got transfected by cationic liposomes through actin-related protein 2/3 (Arp2/3) dependent dynamic lamellipodia, which is augmented by myosin II inhibition. Conversely, cells at the core establish tight junctions at their apical surfaces, impeding liposomal access to the basal lamellipodia and thereby inhibiting transfection. In contrast, liposomes incorporating mannosylated lipids are internalized throughout the entire colony via receptor-mediated endocytosis. These findings contribute a novel mechanistic insight into enhancing therapeutic delivery via liposomes, particularly in cell types characterized by dynamic lamellipodia, such as immune cells or those comprising the epithelial layer.

Graphical Abstract

Mechanisms for cationic liposome transfection to the marginal cells with lamellipodial or epithelial-like central cells of hPSC colonies. Data shown here suggest cationic liposomes fuse with the negatively charged dynamic lamellipodia membrane in an Arp2/3 dependent manner and the process is enhanced by Myosin II inhibition, such as for marginal cells at the hPSC colony edges. However, cells more epithelial in nature such as those inside the hPSC colony do not possess dynamic lamellipodia at the apical surface, rather they form tight junctions that inhibit cationic liposome transfection. Epithelial cells rely on receptor-mediated endocytosis in both myosin II-dependent and independent manners to internalize liposomes with lipids that contain ligands for cell surface receptors such as mannose.

Significance statement.

Drug or gene delivery to human cells is essential for effective treatment. Non-viral cationic liposomes provide a safe delivery option. However, the cellular structures required for internalizing liposomes are not yet fully understood. Using human stem cells which grow in colonies with more dynamic cells at the periphery and epithelial-like cells at the center, we discovered that Arp2/3-dependent dynamic lamellipodia promotes cationic-liposome delivery in dynamic cells while receptor-mediated endocytosis is required for epithelial-like cells. This is significant as it provides a means for enhanced liposome-based therapeutic delivery to both lamellipodia-containing and epithelial-like cells.

Introduction

Liposomes serve as highly efficient and safe delivery vehicles for a wide spectrum of therapeutics, encompassing drugs and genetic materials such as genes and messenger RNAs (mRNAs). A compelling example lies in the development of anti-COVID-19 vaccines, which have played a pivotal role in combating the ongoing pandemic crisis. In this effort, mRNA encoding the spike protein of the coronavirus is encapsulated within cationic liposomes, facilitating their delivery into the bloodstream. Subsequently, these liposomes are internalized by immune cells, triggering the production of antibodies against the virus. Notably, cationic lipids significantly augment the fusion of these liposomes with the anionic surface of the plasma membrane lipid layer, culminating in the liberation of their payload into the cytosol. Moreover, cationic liposomes have the capacity to enter cells through endocytosis while escaping lysosomal degradation. When cationic liposomes are endocytosed, their uptake by cells leads to an intriguing series of events. The endosomal membrane, originating from the plasma membrane, interacts with the cationic liposomal lipids, inducing membrane destabilization. Consequently, a flip-flop phenomenon occurs, wherein the anionic lipids from the cytosolic surface of the endosomal membrane reorient to face its inner surface, while the liposomal cationic lipids project outward into the endosomal membrane. This interaction triggers the lateral diffusion of anionic lipids (from the endosome membrane) and cationic lipids (from the liposome), resulting in the formation of charge-neutralized lipid pairs. This, in turn, leads to the disruption of the lipid bilayer, facilitating the release of liposomal content into the cytoplasm. These intricate mechanisms allow liposomes to circumvent endo-lysosomal degradation, ensuring the efficient release of cargo into the cytoplasm.¹ While substantial progress has been made in elucidating the compositions and delivery mechanisms of liposomes via endocytosis, there remains a lack of comprehensive knowledge concerning the cell surface structures that are indispensable for cationic liposome internalization.

Numerous cell-surface mechanisms may contribute to the internalization of cationic liposomes. These mechanisms include fusion with the anionic plasma membrane surface,^2,3 endocytosis^1,4 or receptor-mediated endocytosis when liposomes contain modified lipids.⁵ These processes can occur either at the apical surface of epithelial cell layers or at membrane structures found at the leading edge of migratory cells, such as lamellipodia or filopodia. It is noteworthy that endocytic internalization is contingent upon actomyosin contractility,⁶ while the fusion of liposomes with the plasma membrane can occur independently of actomyosin processes.² The therapeutic delivery of nucleotides including DNA and mRNA, or drugs, frequently employs cationic liposomes as carriers.⁷ An in-depth comprehension of the cellular structures and associated molecular mechanisms necessary for the internalization of cationic liposomes holds significant promise for the advancement of liposome-based therapeutics. These potential benefits encompass the reduction of drug dosages, the enhancement of immunization efficacy, and the augmentation of CRISPR-based gene-editing efficiency. This knowledge has far-reaching implications for gene therapy and disease modeling research.

In this study, we harnessed the unique properties of human pluripotent stem cells (hPSCs), which form compact colonies characterized by the presence of epithelial cell junctions at the colony center, and more dynamic cells at the periphery.⁸ Our investigation revealed a distinctive pattern: marginal cells exhibit dynamic lamellipodia structures akin to migratory cells and display selective transfection by cationic liposomes compared to central cells. Conversely, central cells demonstrate epithelial-like morphology and behavior, characterized by honeycomb-shaped apical tight-junction formations, as evidenced by Zonula occludens (ZO-1) and actin localizations, and receptor-mediated endocytosis through the apical surface. Previous studies have also reported epithelial-like features in hPSCs.⁸ Remarkably, transfection of marginal cells was found to be independent of actomyosin contractility and notably enhanced under myosin II inhibition, but critically dependent on actin-related protein 2/3 (Arp2/3)-mediated dynamic actin meshwork-driven lamellipodial structures. The Arp2/3 complex, composed of 7 subunit proteins including Arp2, Arp3, and ARPC1-5, nucleates actin polymerization at the side of an existing actin filament upon activation by nucleation promoting factors such as the Wiskott-Aldrich Syndrome (WASP) family of proteins.^9-14 Positioned at the leading edge of cells, the Arp2/3 complex forms a dynamic actin meshwork, serving as a central structure for dynamic lamellipodia and is essential for cell migration.^15,16 The transfection of hPSC colony marginal cells by cationic liposomes is a unique phenomenon. Since, in contrast to the marginal cells, we observed that cells distributed throughout the hPSC colony were efficiently transfected by liposomes containing mannosylated lipids, achieved through receptor-mediated endocytosis. Given that endocytosis is typically dependent on actomyosin contractility, our findings suggest enhanced cationic liposome delivery to hPSCs under myosin II inhibition point toward an intriguing actomyosin-independent, passive membrane fusion mechanism. It is noteworthy that although hPSCs across the entire colony exhibited transfection with mannosylated liposomes, this process was somewhat influenced by myosin II inhibition, albeit not entirely. This observation implies the presence of both actomyosin contractility-dependent and -independent receptor-mediated endocytic delivery mechanisms.

Thus, our study offers a unique insight into mechanisms whereby Arp2/3-mediated lamellipodial structures, similar to those found in migratory cells^15,17 or the peripheral cells of the epithelial layer,¹⁸ can be harnessed to enhance cationic liposome-based delivery. Simultaneously, it underscores the potential for enhancing liposomal delivery in epithelial-like cells through receptor-mediated endocytosis, particularly via mannose receptors at the apical surface. This finding is significant as it holds the promise of optimizing treatments by reducing both dosage and treatment duration when targeting specific cell types for cationic liposome-based therapeutic delivery. The implications of this study extend to hPSC biology and its applications in disease modeling research, given the inefficiency of liposomal transfection in hPSCs, which currently limits gene-editing-based disease modeling research.

Experimental model and subject details

Quantification and statistical analysis

All data presented are mean ± SEM. Cells were treated with small molecules at different time points as independent biological samples in independent culture wells. Statistical tests between 2 independent datasets were done by unpaired Student’s t-test. Graphs were made using GraphPad Prism 9.0 software. Figures were made in Adobe Illustrator and cartoon illustrations were made using BioRender.com.

Methods

A reagent list with catalog numbers is available in the resource table (Supplementary Table S1).

Human pluripotent cell (hPSC) culture

H7-ESCs (WiCell, https://www.wicell.org/) and EP1-iPSCs¹⁹ were grown in mTeSR1 media (mT) in 5%CO₂, 37 ^oC incubator on matrigel (MG) coated plates. To obtain hPSC colonies, cells were passaged by clump passaging using a gentle cell dissociation reagent (GD) after reaching 80% confluency. GD was added to cells for 4 minutes at 37 ^oC, aspirated, then mT was used to resuspend colonies; cell suspensions were mixed by pipetting 3-4 times to break up the colonies into small clumps and then seeded into new MG-coated wells. Clump passaged colonies were cultured for an additional 2-3 days before experiments. For single-cell passaging, cells were incubated with accutase for 10 minutes and then quenched with double volume of mT with 5 μM blebbistatin (blebb). The cells were pelleted by centrifugation at 150 × g for 5 minutes, then resuspended in media containing blebb. Subsequently, the cells were counted and seeded at a density of 25 000 cells per well in a 24-well plate for transfection at the single-cell stage.

Plasmid amplification and purification

The plasmid of 100 ng (CAG-mCherry or GFP-ARP3) was added to 50 μL of Top10 Escherichia coli and kept in ice for 5 minutes. CAG-mCherry was a gift from Jordan Green (Addgene plasmid # 108685; http://n2t.net/addgene:108685; RRID:Addgene_108685) and pEGFP-N1-ARP3 from Matthew Welch (Addgene plasmid # 8462; http://n2t.net/addgene:8462; RRID:Addgene_8462). The bacteria were then heat shocked to promote uptake of the plasmid at 42 °C for 45 seconds before being placed back into ice for 2 minutes. SOC media of 250 μL was added to the bacteria and incubated in a 37 ^oC shaker for 1 hour before being added to 5 mL LB-broth with carbenicillin (50 μg/mL, for CAG-mCherry) or kanamycin (50 μg/mL, for GFP-ARP3) and incubated overnight at 37 °C in a shaking incubator. Plasmid was extracted following the kit (Zymo D4210) protocol, and concentration was measured using nanodrop.

hPSC transfection

Human pluripotent stem cells (hPSCs) were cultured as described above. Single cells after accutase passage were transfected 24 hours after seeding. The clump passaged colonies were added to a larger volume of media and equally split into the wells of 24-well plates. The cells were further cultured for 2-3 days until distinct colonies with well-defined edges and centers were established, reaching a size approximately 1/10th of the initial field of view size under a 10× objective at the onset of drug treatment or transfection. Twenty-four hours after transfection, images were taken by the EVOS fluorescence microscope (Thermo Fisher Scientific) and cells were collected for flow cytometry. Using ImageJ software, fluorescence intensity was quantified by drawing a “donut” containing the colony edge, measured as the edge; the “hole” of the donut was then measured as the center. Raw integrated density was divided by the total area to get the average intensity per area for both the edge and center of each colony.

Colonies were treated with 5 μM blebb, 100 μM CK666, both blebb and CK666, or the equivalent volume of DMSO in mT for the indicated time points and then transfected. Cell transfections (for 1 well of a 24well plate) were done by mixing 2 μL of lipofectamine stem (Invitrogen) and 600 ng of indicated plasmids in 50 μL optimem, vortexing, and letting the mixture sit at room temperature for 10 min. This mixture containing plasmid was then added to the cell culture and incubated for 24 hours.

The hPSC colonies were treated with Latrunculin A (Lat A) at 2 μM dose for a shorter timepoint of 1 hour and then transfected for another 1 hour as longer exposure (24 hours) detaches colonies. After this, the media with drug and liposomes was removed and the colonies incubated with fresh media for 24 hours.

Flow cytometry

After transfection for 24 hours, the hPSC colonies were incubated in 200 μL accutase for 10 minutes, followed by quenching with 400 μL mT supplemented with 5 μM blebb. Subsequently, cells were pelleted by centrifugation at 100 × g for 5 minutes and resuspended in 200 μL mT containing 5 μM blebb. This 200 μL cell suspension was transferred into a 96-well round-bottom plate and read on the Attune NxT Acoustic Focusing Flow Cytometer (Thermo) equipped with Attune Auto Sampler (Thermo). Flow gating to separate live and dead cells was done by treating hPSCs with 2 μg/mL puromycin, which induces cell death, or an equivalent amount of DMSO control for 48 hours and then made into a single-cell suspension as described above. The cell population that disappeared under puromycin treatment was considered as the live cell population, while the remaining cells labeled with dead cell marker propidium iodide were considered as dead cells (Supplementary Figure S3A, S3B). These gates were then used for the subsequent transfection experiments. From the live cell population, the singlet population was then gated, and from the singlet population, the percentage of RFP or green positive cells or the average single-cell fluorescence intensities were measured using the Attune NxT Software. These gating conditions from each experiment are shown in Supplementary Figure 3. Data were exported to Excel or Prism for analysis and plotting. Each well was considered an independent biological sample, and 3 or more biological repeats were used for each condition.

Immunofluorescence imaging

H7-hESCs were seeded using GD passaging on MG-coated glass coverslips (1.5 thickness) and cultured for 2-3 days until colonies were established. Following the indicated treatments media was aspirated and cells were washed with 1X PBS, and then fixed with 4% paraformaldehyde for 30 minutes at 37 ^oC. At this step, if needed cells were washed once and then stored in PBS at 4 °C until immunostained. Fixed cells were permeabilized with 0.5% Triton-X100 in PBS for 5 minutes and then washed in washing buffer (1% donkey serum, 0.05% Triton-X100 in PBS) 3 times for 5 minutes each. Cells were then incubated in a blocking buffer (5% donkey serum, 0.2% Triton-X100 in PBS) for 1 hour at room temperature. After blocking, antibodies against Cortactin (Rabbit, Cell Signaling), ZO-1 (Mouse, Invitrogen), or Oct4 (Goat, R&D systems) were added (1:200 in blocking buffer) and the coverslips were incubated overnight at 4 °C. Next, coverslips were washed with washing buffer 3 times for 5 minutes each and incubated for 2 hours at room temperature in the dark with secondary antibody anti-rabbit Alexa-568 (Invitrogen), anti-mouse Alexa-647 (Invitrogen), or anti-goat Alexa-568 (1:500 in blocking buffer), and Alexa Fluor 488 conjugated Phalloidin (4 U/mL) for F-actin. The coverslips were washed with washing buffer 3 times for 5 minutes each, with 1.43 μM DAPI added to the second wash. Coverslips were then mounted using DAKO and sealed with clear nail polish. Confocal z-stack images were taken using Zeiss LSM700 with 63×/1.4 oil and 10×/0.3 air objectives.

Live cell imaging

H7-hESCs were seeded into glass-bottom MatTek dishes using GD passaging. After transfection, dishes were imaged using Tokai Hit live cell imaging chamber at 37 ^oC and 5%CO₂. Confocal z-stacks were taken using Zeiss LSM700 with 63×/1.4 oil objective and 10x/0.3 air objective. For lamellipodial measurements, colonies were transfected with cationic liposomes containing CAG-mCherry and GFP-ARP3 plasmids, using 600 ng of each plasmid, and incubated for 24 hours. Subsequently, the transfection media was replaced with fresh media containing small molecules and incubated for an additional 3 hours. Dishes were then placed into the Tokai Hit live cell chamber and a time series was captured with 15 seconds intervals for 5 minutes with simultaneous acquisition of red and green fluorescence signals. Lamellipodial dynamics were measured from kymographs using ImageJ software. A line was placed over the moving edge of GFP-ARP3 expressing cell and a kymograph was made using the Kymograph plugin. Edge protrusion and retraction rates were quantified from kymographs for edge movement rates.

For single-cell fluorescence intensity measurements, colonies were first treated with the indicated small molecules at the indicated timepoints, then transfected with cationic liposomes containing CAG-mCherry plasmid and incubated for another 24 hours in the presence of the small molecules before imaging. Confocal z-stacks of transfected cells were taken at the colony edges and intensity per cell area from the sum projections were measured using ImageJ software.

For partial digestion of the cell junctions (Figure 6), colonies were incubated with 200 μL accutase or trypsin for 1-3 minutes. Then trypsin or accutase was aspirated and cells were gently washed twice with normal mT media. Treated colonies were then transfected with cationic liposome containing CAG-mCherry plasmid and incubated for 24 hours. Colonies were washed and incubated in mT with live cell nuclear dye Hoechst (5 μg/mL) for 15 minutes, washed, and then imaged using both 63×/1.4 oil and 10×/0.3 objectives.

Figure 6.

Junctional integrity of the epithelial-like hPSC colony center inhibits cationic liposome transfection. (A) Experimental design to test if intact cell-cell junctions at the hPSC colony centers limit cationic liposome transfection. ZO-1 containing tight junctions are disrupted by mild digestion with trypsin/accutase. (B) H7-hESC colonies were exposed to trypsin or accutase for 1-3 minutes, washed, and then transfected with cationic liposomes containing CAG-mCherry plasmid. (C) Confocal images (10×/0.3) show untreated cells (UTC) with marginal cell transfection at the edges, while colonies briefly treated with accutase, or trypsin got transfected both at the edge and center. (D) Single-cell suspensions were collected 24 hours after transfection with indicated treatments and run through a flow cytometer. RFP-positive cell populations are indicated by arrows. (E) Quantification of the percentage of RFP-positive cells. Each point represents an independent biological repeat, n = 4. Error bars are SEM. ****P < .0001. Unpaired Student’s t-test between independent datasets. (F) Images of H7-hESC colonies (10×/0.3) treated with trypsin or accutase for 2 minutes and then fixed and stained for ZO-1. (G, H) Max projections of confocal z-stacks (63×/1.4 oil) of ZO-1 in H7-hESC colony centers after (G) 2 minutes trypsin or (H) 16h blebbistatin treatments.

Transfection with mannosylated liposomes

The hPSC colonies in MatTek dishes were treated with the indicated small molecules or DMSO vehicle control at indicated time points, then 10 μL of liposomes (Encapsula Nanosciences) was added to the cell culture wells in the presence of small molecules and gently mixed by pipetting. After 24 hours incubation, media was aspirated, colonies were washed twice with mT, and then 2 mL of mT with 1 μL Hoescht (5 μg/mL) was added for live confocal imaging using both 63×/1.4 oil and 10×/0.3 objectives.

Lentivirus

The hPSC colonies at ~80% confluency were clump passaged using GD and seeded into 96-well MG-coated wells. The next day, cell counting was done from one well using accutase mediated single cell dissociation. Lentivirus (LV) (Life Technologies Cat # A32060) with a viral vector containing P_EFS-GFP was added to each well at a multiplicity of infection of 10 along with 8 μg/mL polybrene. The plate was then centrifuged at 800 × g at room temperature for 1 hour before placing it into a 37 ^oC, 5% CO₂ incubator overnight. The next day, media with LV was replaced with normal mT and GFP signal was observed over time with daily media exchange.

qPCR

The hPSC colonies were seeded by clump passaging in a 24-well plate, treated as indicated, washed, then cultured and passaged another 4 times over approximately 2 weeks. Cells were collected at 75%-80% confluency by incubating with 200 μL accutase for 10 minutes, then transferred into mT media supplemented with 5 μM Blebb. Cells were pelleted by centrifugation at 100 × g for 5 minutes. The supernatant was aspirated, and the cell pellet was used for RNA extraction following the kit protocol (Qiagen 74104). RNA concentration was measured by Nanodrop 2000c (Thermo) and 1 μg of RNA was used to create cDNA following the kit protocol (Abm #G592). Primer sequences are available in Supplementary Table S1. The qPCR experiments were performed using 100 ng of total cDNA with BlasTaq qPCR master mix (Abm #G892) in a 20 μL reaction on a QuantStudio6 Flex RT PCR system (Applied Biosystems). GAPDH was used as a housekeeping control on every plate to determine the ΔCt, and subsequently, the ΔΔCt was calculated relative to the average ΔCt of the control samples. All samples had 3 technical repeats for each biological repeat, with a total of 3-5 biological repeats for

Results

Marginal cells at the hPSC colony edges are selectively transfected by cationic liposomes

Human pluripotent stem cells grow in colonies with cells at the center forming tight junctions with apico-basal polarity similar to the epithelial cell layer, and with cells at the edge or margin being more dynamic in nature.⁸ This gives a unique advantage to study the cellular structures essential for the cationic liposome transfection as used for therapeutic delivery. We have formed cationic liposomes by mixing cationic lipids (Lipostem, Thermo) and negatively charged plasmid DNA containing mCherry under CAG promoter following the user manual. These liposomes are then added to the H7 human embryonic stem cells (H7-hESCs) grown in sparse culture by single cell accutase passaging or in small colonies by clump passaging. Next, 24-hour post--transfection fluorescence expression was checked by imaging as shown in Figure 1A. Much to our surprise, we observed that hPSCs at the colony margins got selectively transfected but not at the colony center (Figure 1B), while sparse cell culture got transfected randomly with no such pattern (Figure 1 C) as observed by the mCherry fluorescence. To measure if hPSCs at the colony center are not transfected and hence not expressing mCherry, we drew a line across the colony center through the margins and measured fluorescence intensity profile along that line (Figure 1D). Indeed, we observed specific fluorescence intensity peaks on the line corresponding to the margins but not at the center (Figure 1E). This observation was further verified by measuring fluorescence intensity around the colony edges and centers, which showed significantly higher expression at the edges (Figure 1F). The liposomal transfection to marginal cells is seen throughout the culture, as revealed by wider field of view using a 4× objective magnification image (Supplementary Figure S1). To test if liposome-mediated transfection to marginal cells is a cell type-specific phenomenon, we transfected induced pluripotent stem cell (EP1-iPSC)¹⁹ colonies with cationic liposomes containing CAG-mCherry plasmid. We observed a similar transfection pattern of the marginal cells of these iPSC colonies (Supplementary Figure S2A) establishing that the phenomenon is not specific to a particular stem cell line, rather it is a property of hPSCs. This observation is further verified by measuring fluorescence intensity profiles across the line through the colony which showed specific intensity peaks at the colony margins but not at the centers (Supplementary Figure S2A, S2B). Similarly, fluorescence intensity measurements showed significantly high fluorescence at the margins compared to centers (Supplementary Figure S2C). We occasionally observe some cells adjacent to the margin but inside the colony also express fluorescence protein which could arise from the dividing transfected marginal cells or through direct transfection. These data suggest hPSCs at the colony margins possess unique structures compared to the center cells and enhancing those structures could increase the cationic liposome transfection efficiency in the marginal cells.

Figure 1.

Marginal cells at hPSC colony edges selectively get transfected with cationic liposomes. (A) Illustration of hPSC transfections where H7-hESCs, after clump or single-cell passaging, were transfected with cationic liposomes containing CAG-mCherry plasmid and images were taken 24 hours after transfection. Representative 10× brightfield and fluorescence images of (B) clump-passaged colonies and (C) single-cell passaged cells. (D) Representative images of clump passaged H7-hESC colony 24 hours after transfection with CAG-mCherry plasmid. (E) Line trace through the center of the colony as shown in (D) reveals fluorescence intensity peaks at the edges but not at the center. (F) Quantification of the fluorescence intensity at colony edges and centers from 58 colonies from 5 independent experiments. Each data point represents a single colony. Error bars are SEM. Unpaired Student’s t-test, ***P < .001.

Marginal cells of hPSC colonies possess Arp2/3 dependent dynamic lamellipodia structures

It has been reported that hPSCs form tight junctions at the center of the colony while cells at the margins show more dynamic behavior.⁸ Arp2/3 mediated polymerization of the dendritic actin network at the cell edge forms dynamic lamellipodium critical for directional cell migration.¹⁷ We examined whether hPSCs at the colony margins form lamellipodia structures by confocal immunofluorescence imaging of actin cytoskeleton and the lamellipodium marker cortactin, at the edge and center as shown in Figure 2A. Cortactin localizes to the cortical F-actin sites required for forming dynamic lamellipodia membranes.²⁰ Indeed, we observed in DMSO-treated control conditions the basal surface of marginal cells possess cortactin-decorated actin meshwork structures at the leading edge (Figure 2B, white arrows), which are absent at the corresponding apical surface (Figure 2C). Actomyosin contractility promotes thick actin stress fiber (SF) formation²¹ but reduced actomyosin contractility results in more non-tensile thin actin filaments with enhanced lamellipodia structures.^22,23 We asked if myosin II inhibition by a potent inhibitor blebbistatin²⁴ dissolves actin SF and enriches lamellipodia at the marginal cells. Indeed, we observed a loss of SFs but more thin actin meshwork structures at the basal surface of edge cells (Figure 2B). Interestingly, we also observed more cells at the colony edges as well as immediately inside of the edges with cortactin-decorated actin cortex (Figure 2B, white arrows) indicating more lamellipodia formation under myosin II inhibition. CK666 is a potent inhibitor of Arp2/3, and hence lamellipodia formation,¹⁵ and we saw that Arp2/3 inhibition led to more actin SF formation with reduced cortactin decorated actin cortex in the marginal cells (Figure 2B) indicating loss of lamellipodia. We further observed myosin II inhibition by blebbistatin in the presence of Arp2/3 inhibition by CK666 reduced actin SF formation to some extent, but it did not increase cortactin localization at the cell edge (Figure 2B). Thus, our data indicates Arp2/3 activity can promote lamellipodia in the marginal cells independent of myosin II activity. The apical surface of the colony edges did not show actin SF, rather formed honeycomb-like filaments characteristic of epithelial apical surface with no cortactin localization to actin (Figure 2C). No noticeable changes in actin or cortactin assembly were observed upon Arp2/3 and/or myosin II inhibition (Figure 2C). At the basal surface of the colony center, cells possess an extensive amount of actin SFs which did not change under Arp2/3 inhibition but were significantly reduced by myosin II inhibition alone or in combination with Arp2/3 inhibition (Figure 2D). However, at the basal or apical surfaces of colony centers, cells did not show any cortactin localization on the actin filaments indicating the absence of lamellipodia in the center cells (Figure 2D and 2E). Interestingly, myosin II inhibition occasionally increased cortactin localization at the apical actin networks (Figure 2E, white arrows) indicating an increase in membrane dynamics at the apical cortex. Arp2/3 inhibition alone or in combination with myosin II inhibition did not alter actin or cortactin localizations in the apical surface (Figure 2E).

Figure 2.

Marginal cells at hPSC colony edges exhibit lamellipodia decorated with cortactin. (A) Illustration for hPSC colony imaging at the edge and centers. (B-E) Confocal z-stacks (63×/1.4 oil) were taken at H7-hESC colony edges and centers. Representative confocal immunofluorescence images of F-actin and cortactin after 3 hours DMSO, Blebbistatin (Blebb), CK666, or Blebb + CK666 treatments. (B, D) Shown are images at the bottom of the cells (basal) of the colony edge or center. (C, E) Images at the apical surface of a colony at the edge or center. Arrows indicate cortactin-decorated lamellipodia.

Since lamellipodia is a dynamic membrane structure regulated by Arp2/3 mediated dynamic actin meshwork, we asked if marginal cells of the hPSC colonies possess dynamic lamellipodia with Arp2/3 at the leading edge. To test this, we co-transfected H7-hESC colonies with cationic liposomes containing GFP-ARP3 (Addgene #8462)²⁵ and CAG-mCherry (Addgene #108685),²⁶ and performed live cell confocal imaging under control or inhibitory small molecules as explained in Figure 3A. Under the untreated condition, as well as in blebbistatin-treated myosin II inhibition, we found marginal cells localize GFP-ARP3 at the leading-edge with dynamic lamellipodia (Figure 3B, 3C; video 1) and (Figure 3E, 3F; video 2) respectively. Lamellipodia of the edge cells remained highly dynamic under control and myosin II inhibition as shown by the kymographs (Figure 3D, 3G) with a moderate increase in the membrane protrusion/retraction rates under myosin II inhibition (Figure 3K). Furthermore, Arp2/3 inhibition by CK666 led to the loss of GFP-ARP3 localization at the leading-edge of marginal cells and corresponding membrane dynamics (Figure 3H, 3I; video 3). This observation was validated by the kymograph analysis of cell edges (Figure 3J) and corresponding membrane protrusion/retraction rate measurements (Figure 3K). Thus, our data shows hPSCs at the colony edges but not at centers possess dynamic lamellipodia structures which are dependent on Arp2/3, but independent of actomyosin contractility.

Figure 3.

Lamellipodial dynamics of the marginal cells is Arp2/3 dependent but independent of myosin II activity. (A) Illustration of live cell imaging (63×/1.4 oil) of H7-hESCs 24 hours after dual transfection with cationic liposomes containing CAG-mCherry and GFP-ARP3 plasmids under indicated treatments. (B, E, H) Composite images show cell location relative to colony edge and center. (C, F, I) Montages of the time series taken with 15-second intervals for 5 minutes, showing edge dynamics of cells expressing ARP3 and mCherry from videos 1-3. (D, G, J) Kymographs correspond to the numbered lines in the first frame of the montages. (K) Quantification of the movement rate (μm/seconds) measured from the kymographs. Each data point represents a protrusion or retraction rate from the kymograph, n = 17-23 kymographs from 3-5 movies per treatment. Error bars are SEM. ***P < .001. Unpaired Student’s t-test between independent datasets.

Loss of Arp2/3-dependent lamellipodia limits marginal cell transfection of hPSC colonies

Dynamic lamellipodia could provide a negatively charged membrane surface for cationic liposome fusion and this process may be enhanced with more lamellipodia surface. To test this hypothesis, we first inhibited myosin II by blebbistatin which is known to increase lamellipodia membrane surface²² and then measured transfection efficiency by imaging and flow cytometry as explained in Figure 4A. Flow gating strategy to count the percentage of a live singlet cell population expressing mCherry and single-cell fluorescence intensity are described in Supplementary Figure S3. We observed transfection of H7-hESC colony marginal cells (blue arrows) as well as some cells inside the colony (yellow arrows) under myosin II inhibition (Figure 4B). Cells inside the colony presumably got transfected through the lamellipodia-like cortactin positive membrane that appeared at the edge and at the inside cells (Figure 2B) as well as at the apical surface of colony center under myosin II inhibition (Figure 2E). Remarkably, single-cell analysis by flow showed a significant increase in the percentage of cells that got transfected with blebbistatin (Figure 4C, 4D). If more lamellipodia surface under myosin II inhibition allows cells to internalize more liposomes, this will lead to more CAG-mCherry plasmid delivery and increased mCherry expression. Indeed, we observed a significant increase in fluorescence intensity per cell measured by flow (Figure 4E) or by confocal imaging (Figure 4F, 4G). Conversely, we tested if reducing lamellipodia by potent Arp2/3 inhibitor CK666 lowers liposome transfection efficiency. In agreement, we observed a strong reduction of cellular transfection efficiency measured by imaging and flow (Figure 4B-4D). Furthermore, Arp2/3 inhibition reduced the amount of liposome delivery into cells as reveled by reduced fluorescence intensity per cell by flow (Figure 4E) and by confocal imaging (Figure 4F, 4G). The Arp2/3 inhibitory effect was to some extent rescued by myosin II inhibition, as increased transfected cell population (Figure 4B-4D) and cellular fluorescence intensity by flow (Figure 4E) and by imaging (Figure 4F and 4G) were observed when hPSC colonies were treated both by blebbistatin and CK666 compared to CK666 alone. Hence, this data suggests actomyosin contractility as an inhibitory factor, with Arp2/3-mediated dynamic lamellipodia facilitating the cationic liposome delivery into the cells.

Figure 4.

Transfection of the hPSC colony marginal cells is dependent on Arp2/3 and augmented by myosin II inhibition. (A) Illustration for hPSC colony treatment with small molecules and then transfection with cationic liposomes containing CAG-mCherry plasmid to measure transfection efficiency. (B) Representative images of H7-hESC colonies 24 hours post-transfection in the presence of small molecules. Blue arrows indicate transfected cells at the colony edges, and yellow arrows indicate transfected cells inside the edges. (C) Single-cell suspensions were collected 24 hours after transfection with indicated treatments and run through a flow cytometer. RFP-positive cell populations indicated by arrows quantified for (D) percentage of RFP-positive cells and (E) average mCherry intensity, normalized to DMSO. Each point represents an independent biological repeat. n = 3-6. (F) Representative confocal images of H7-hESCs treated for 16 hours and then transfected with CAG-mCherry for 24 hours with continued treatment (63×/1.4 oil). (G) Quantification of mCherry intensity per cell area from sum projections of confocal z-stacks. Each point represents one cell, n = 40-47 cells. Error bars are SEM. *P < .05, **P < .01, ***P < .001, ****P < .0001. Unpaired Student’s t-test between independent datasets.

Transfection of hPSC colony marginal cells is independent of actin cytoskeleton

Given that the loss of myosin II contractility enhanced marginal cell transfection, we next asked whether the process is actin cytoskeleton dependent. To test this, we first checked the minimum liposome exposure time needed to have successful transfection and how long cells can remain attached under actin cytoskeleton disruption. Based on a time course experiment of liposome transfection, we found that 1h exposure is sufficient for hPSC transfection. H7-hESC colonies exposed to cationic liposomes containing mCherry plasmid for 1h to 24h as explained in Supplementary Figure S4A showed a similar pattern of marginal cell transfection (Supplementary Figure S4B) with similar levels of percentage of cell transfection as measured by flow (Supplementary Figure S4C-S4D). Cellular fluorescence intensity showed similar levels for all the different transection durations except for 24 hours which showed a moderate increase, suggesting that the number of liposomes entering into an individual cell is independent of liposome exposure time (Supplementary Figure S4E). Furthermore, we found when actin cytoskeleton is disrupted by a potent actin depolymerizing drug Latrunculin A (Lat A)²⁷ for 24 hours (Supplementary Figure S5A), it led to a reduced number of live cells for H7-hESCs (Supplementary Figure S5B, S5C, yellow gate), with a corresponding reduced number of mCherry expressing cells (Supplementary Figure S5D, arrow). Prolonged Lat A treatment led to the colony detachment from the dish as shown by reduced total cell count by flow (Supplementary Figure S5E). So, we had to disrupt actin filaments for a small period of time followed by transfection to assess the role of actin in cationic liposome transfection. Based on these findings we disrupted actin cytoskeleton by Lat A for 1 hour and then exposed hPSC colonies for one more hour with liposomes containing CAG-mCherry plasmid along with Lat A, followed by exchanging to media without Lat A and liposomes, and measuring transfection after 24 hours in hPSC media as explained in Figure 5A. We saw actin cytoskeleton disruption did not inhibit cell transfection at the edge cells (Figure 5B). Interestingly, we saw a moderate increase in the percentage of transfected population measured by flow (Figure 5C, 5D) but with a mild reduction in cellular fluorescence intensity (Figure 5 E). These data support the prior findings where disruption of actin filaments showed increased direct penetration of cationic liposomes through plasma membrane fusion.²⁸ The reduction of cellular fluorescence intensity for Lat A treated colonies could be due to the uptake of a smaller number of liposomes, limited by reduced lamellipodia surface at the marginal cells. Lat A treatment for 2 hours followed by a wash did not detach cells as shown by imaging (Figure 5B) and by total cell count by flow after 24 hours (Figure 5F). Lat A treatment was effective in disrupting the actin cytoskeleton within 3 hours as shown by confocal imaging of phalloidin labeled F-actin filaments (Figure 5G). These data support the notion that cationic liposome transfection to cells depends on the availability of dynamic lamellipodial plasma membrane but is independent of the actin cytoskeleton and myosin II activity.

Figure 5.

Marginal cell transfection of hPSC colony edges by cationic liposomes is unaffected by actin cytoskeleton integrity. (A) Illustration of hPSC colony transfection under Lat A treatments. H7-hESC colonies were pretreated with Lat A (2 μM) for 1 hour and then transfected for 1 hour with cationic liposomes containing CAG-mCherry plasmid along with Lat A. Media was then changed to remove plasmids and Lat A. (B) Images show marginal cell transfection at the edges 24 hours after Lat A washout. (C) Flow cytometry analysis on RFP positive cells (arrow) 24 hours after Lat A washout for (D) percentage of RFP positive cells, (E) mean RFP intensity per cell, and (F) total cell count. (G) Colonies were treated with Lat A for 3 hours, then fixed and stained for phalloidin to visualize the actin cytoskeleton changes (63×/1.4 oil). Each point represents an independent biological repeat, n = 3. Error bars are SEM. *P < .05, **P < .01. Unpaired Student’s t-test between independent datasets.

Epithelial-like cell junctions limit cationic liposome transfection to hPSC colony central cells

While our data suggest that dynamic lamellipodia at hPSC colony marginal cells promote cationic liposome transfection, it is still not clear why the epithelial-like cells at the colony center are not transfected. It has been reported that the basement membrane of epithelial cells forms dynamic membrane structures with some similarity to lamellipodia.²⁹ We asked if the cell junctions at the apical side limit liposome access to the basal membrane, and hence their transfection. So, we hypothesized that if we break these cell junctions that should promote transfection of the center cells. To test this, we enzymatically digested the cell surface proteins with either accutase or trypsin for a brief 1 minute or 3 minutes and then transfected with cationic liposomes containing CAG-mCherry plasmid as explained in Figure 6A, 6B. Remarkably, we saw a significant increase in transfection of the hPSC colony central cells (Figure 6C) as revealed by confocal imaging. In agreement, when measured by flow, we saw strong increase in the percentage of transfected cells under brief digestion by accutase or trypsin (Figure 6D, 6E). Short exposure to these enzymes maintained colony morphology with cells still attached to the coverslip (Figure 6C, 6F). But when we checked junctional integrity by staining against actin and tight-junction protein Zonula occludens (ZO-1), we saw the opening of junctions inside the colony (Figure 6G, arrows). This further supports the notion that cell junction integrity limits cationic liposomes from accessing the basement membrane of the epithelial-like cells required for transfection. Since we observed some transfection inside the colony under myosin II inhibition by blebbistatin, we asked if this is due to the disrupted cell junctions. However, actin filament and ZO-1 staining revealed myosin II inhibition by blebbistatin does not compromise junctional integrity of the hPSC colony (Figure 6H). Thus, our data suggest cationic liposome delivery into the epithelial-like hPSC central cells requires destabilization of the cellular junctions.

Liposome delivery to the epithelial-like hPSC colony central cells require receptor-mediated endocytosis

Liposome-mediated therapeutic delivery to the epithelial cells through disrupting epithelial integrity is not ideal, as that would be associated with many health complications. The central cells of hPSC colonies possessing epithelial-like morphology provide a unique opportunity to investigate the properties of these cells required for liposome transfection without the need for junctional disruption. Hence, we asked if hPSC colony central cells could take up liposomes through receptor-mediated endocytosis without the need for disrupting epithelial integrity. To this end, we have used liposomes (m-fluoroliposome-DiO, encapsula) containing 4-aminophenylα-d-mannopyranoside conjugated lipids and lipophilic water-insoluble DiO dye whose excitation/emission peaks fall at 484/501 nm. 4-aminophenylα−d-mannopyranoside is a hydrophobic derivative of mannose and can be internalized through the mannose receptor for drug delivery as illustrated in Figure 7A. We used these liposomes to transfect hPSC colonies under control or small molecule treatments for 24 hours and measured transfection efficiency by imaging as explained in Figure 7B. Much to our surprise, we see liposome transfection throughout the colony (Figure 7C). The colony center shows less fluorescence signal than the edge cells (Figure 7C), indicating that edge cells uptake liposomes through both membrane fusion and receptor-mediated endocytosis. When we checked cells at the edge and center as illustrated in Figure 7D using high-magnification (63×) confocal imaging, we saw robust DiO fluorescence signals, notably on the internal membranes for these cells (Figure 7E, 7F). As receptor-mediated endocytosis depends on acto-myosin contractility,^30,31 we checked if the transfection efficiency is reduced under myosin II and Arp-2/3 inhibition by flow analysis as illustrated in Figure 7G. Indeed, we observed nearly 100% of cells got transfected under DMSO control treatment but a significant reduction in transfection under myosin II inhibition by blebbistatin treatment. However, the effect was less under Arp2/3 inhibition by CK666, but in the presence of both blebbistatin and CK666 we saw maximum reduction (Figure 7H, 7I). Although we see a reduction in transfected cell populations in the presence of CK666 and blebbistatin, cells were still able to uptake some liposomes as revealed by DiO signals on images (Figure 7E, 7F) and by flow analysis (Figure 7H, 7I). This suggests that myosin II contractility on Arp2/3 nucleated actin filaments aids in receptor-mediated endocytosis of liposomes on the apical surface of the epithelial-like hPSCs, with the existence of alternative mechanisms. It has been reported that viral transduction to cells occurs through receptor-mediated endocytosis.³² To test if viral transduction would occur throughout the hPSC colonies, we transduced H7-hESC colonies with lentivirus containing GFP vector under EF1α-short (EFS) promoter. Indeed, similar to the mannosylated liposomes, we observed hPSCs throughout the colony got transduced with a virus as shown by GFP expressions (Supplementary Figure S6).

Figure 7.

Epithelial-like hPSC colony center internalizes liposomes through receptor-mediated endocytosis. (A) Potential mechanisms for receptor-mediated liposome endocytosis where liposomes contain mannosylated lipids and DiO lipophilic dye. (B) H7-hESC colonies were pre-treated with small molecules for 16 hours and then liposomes were added for another 24 hours in the presence of small molecules before imaging. (C) Representative low magnification (10×/0.3) images of drug-treated H7-hESC colonies 24 hours after adding mannosylated liposomes. (D) Illustration for image acquisition at the colony edge and center. High magnification (63×/1.4 oil) images were taken at the colony (E) edge and (F) center showing internalization of mannosylated liposomes. (G) Colonies were collected as single-cell suspensions and run through flow cytometry after indicated treatments. (H) Flow plots and quantification of the (I) percentage of positive cells. Each point represents an independent biological repeat, n = 3. Error bars are SEM. ****P < .0001. Unpaired Student’s t-test between independent datasets.

Our studies here have identified hPSC colony marginal cell transfection is independent of actin cytoskeleton, and inhibiting myosin II by blebbistatin increased cationic liposome transfection which is mitigated by Arp2/3 inhibition with CK666. While colony central cells are resistant to cationic liposome transfection, partial digestion of cell-cell junction or use of mannosylated liposomes significantly increased central cell transfection. For practical application of these strategies to enhance hPSC transfection, this is important to study if these treatments affect pluripotency or induce differentiation in the subsequent passages. To test this, we have treated hPSCs with Blebb, CK666, Lat A, accutase, trypsin, and mannosylated liposomes as done in this study (Supplementary Figure S7A) and subsequently maintained cells in normal media for 4 passages for a duration of about 2 weeks. Then we measured Oct4 expression level in cells which is a bona fide marker for stem cell pluripotency.³³ Remarkably, we observed a similar level of Oct4 in the treated hPSC compared to control conditions both by confocal immunofluorescence imaging (Supplementary Figure S7B) and by qPCR (Supplementary Figure S7C). This indicates our strategies can be applied to enhance liposome-based delivery to hPSCs without affecting their pluripotency.

Thus, our study here reveals unique mechanisms where Arp2/3-mediated dynamic lamellipodia could be enhanced for increasing cationic liposome delivery while receptor-mediated endocytosis could be leveraged to promote liposome delivery to the epithelial-like cells.

Discussion

We have shown here that hPSC colonies possess unique morphological features with cells at the colony margin showing dynamic lamellipodia but with cells at the center forming epithelial-like junctions with characteristic honeycomb-shaped actin filament and junctional protein ZO-1 localizations at the apical side. We found cationic liposomes, which are a preferred vehicle for DNA/RNA delivery, preferentially delivered plasmids to the hPSC colony marginal cells but not to the center. We further discovered cell junctions at the colony centers block such transfection, but mild enzymatic digestion of the cell surface proteins led to the transfection of an entire colony. However, liposomes with mannosylated lipids are capable of transfecting the entire colony through receptor-mediated endocytosis without the need for disrupting cell-cell junctions. These results are very important as they provide dual mechanisms for delivering liposomes to cells with dynamic lamellipodia as well as to epithelial-like cells as illustrated in our graphical abstract. Studies here are done on hPSCs and the differential properties of cells at the hPSC colony margin possessing lamellipodia versus central cells with epithelial-like morphology are also supported by others.^8,34 However, there are several other stem cell types that also grow by colony formation such as mesenchymal stem cells,³⁵ cancer stem cells,³⁶ and mouse pluripotent stem cells.^37,38 This would be important to test if other colony-forming cells show similar properties between marginal and central cells and if that could be exploited using our discovered strategies to enhance cationic liposome-mediated gene delivery success.

Our discovery that myosin II inhibition resulted in more cortactin-decorated lamellipodia in the marginal cells with enhanced liposome delivery to individual cells as well as to a greater number of cells within the colony is supported by existing literature. It has been reported that acto-myosin contractility negatively regulates lamellipodia formation²² and inhibition of myosin II contractility increases liposome transfection to stem cells.³⁹ Hence, under myosin II inhibition more lamellipodia surfaces at the marginal cells are likely to promote more liposome uptake through membrane fusion. Cationic liposomes could be internalized either through membrane fusion⁴⁰ or through endocytosis.^1,4 It is unlikely that the transfection of colony marginal cells that we observed is through endocytosis which is largely dependent on myosin II activity,^30,31 but we saw enhanced marginal cell transfection under myosin II inhibition. However, at this point, we cannot rule out a myosin II-independent but other molecular motor-dependent endocytic uptake mechanism in the marginal cells.

Our observation that the cationic liposomes are restricted from transfecting the epithelial-like cells inside the hPSC colony is intriguing. This further supports a non-endocytic delivery mechanism for cationic liposomes, as otherwise, they would have transfected the entire colony. On the contrary, the dissociation of cell-cell junctions by mild enzymatic digestions led to robust center cell transfection by cationic liposomes, supporting the notion that liposomes need access to the dynamic basement membrane for membrane fusion and internalization. However, dynamic cells at the colony margins and epithelial-like cells at the centers are competent for receptor-mediated endocytic uptake as we observed robust transfection of both cell types by the mannosylated liposomes, which is dependent on the mannose receptor-mediated endocytosis. Mannosylated liposomes have also been shown as a robust delivery vehicle for mouse brain tissue.⁴¹

Thus, our discovery here provides many fold applications for developing novel therapy delivery and disease modeling research. The identification of receptor-mediated mannosylated liposome delivery throughout the hPSC colony is very important. Gene editing such as by CRISPR/Cas9 in hPSCs provides a unique opportunity for disease modeling research. Gene-edited stem cells can be differentiated to the target cell types for understanding disease mechanisms for developing therapy. Conversely, patient-derived induced pluripotent stem cells (iPSCs) with gene mutations can be corrected by CRISPR/Cas9 followed by differentiation to the effected cell type and injection for regenerative medicine.⁴² While these strategies have tremendous clinical benefits and are currently under active research, gene-editing in hPSCs is an extremely inefficient process owing to their poor transfection property.⁴³ Our discovery that accessing mannose receptor-mediated endocytosis through mannosylated liposomes provided nearly 100% hPSC transfection throughout the colony can significantly increase stem cell editing rate by increasing the delivery efficiency of CRISPR/Cas9 nucleotides. In addition to the direct application for stem cell research, our findings that Arp2/3 mediated lamellipodia promote cationic liposome delivery provides an opportunity to specifically enhance DNA or RNA delivery to cells with dynamic lamellipodia such as immune cells for vaccination or to cancer cells for inducing apoptosis and selective elimination. Our discovery that receptor-mediated endocytosis provides high-efficiency liposome delivery to epithelial-like hPSCs at the colony centers is also clinically important as a majority of our tissues are epithelial in nature and could be targeted through these mechanisms each condition.

Author contributions

Michelle Surma: collection and/or assembly of data, data analysis and interpretation, administrative support, manuscript writing. Kavitha Anbarasu: collection and/or assembly of data. Arupratan Das: conception and design, financial support, administrative support, collection and/or assembly data, data analysis and interpretation, manuscript writing, and final approval of manuscript.

Funding

This work was supported by a grant from the NIH, United States (R00EY028223) and a startup package to AD from the Indiana University School of Medicine.

Conflicts of interest

The authors declare no potential conflicts of interest.

Data availability

Further information and requests for resources and reagents should be directed to Arupratan Das (arupdas@iu.edu). Stem cells and plasmids are available from the lead contact’s laboratory upon request and completion of the Material Transfer Agreement. Data are available in the article itself and its supplementary materials.