Mycobacterial surface-mediated mechanical priming initiates spacious to compact phagosome switching for egress

Summary

Mycobacterium tuberculosis (Mtb) establishes intracellular niches by remodeling host membranes into either spacious or compact phagosomes, yet how direct bacterial contact within these distinct compartments facilitates bacterial egress remains unknown. Using fixed cell imaging, reconstituted phagosome-like vesicles, and numerical simulations, we uncover that mycobacterial load-mediated contact determines phagosome fate by driving membrane-bending, lipid-wrapping, and phase-separation. Low-to-moderate bacterial loads promote membrane vesiculation and formation of compact phagosome-like structures, whereas high bacterial loads generate scaffold-like architectures that prime membranes for rupture by increasing deformability and inducing lipid demixing. Notably, direct bacterial contact further potentiates the membrane-disrupting activity of the virulence factor ESAT-6, synergistically compromising phagosomal integrity. We propose that mycobacteria actively control the dynamic balance between spacious and compact phagosomes by modulating host membrane mechanics through the membrane-to-contact area ratio and proximity to the lipid demixing threshold. Together, these findings identify a biophysical switch underlying pathogen-driven membrane remodeling and intracellular survival and escape.

Graphical abstract

Highlights

• •Mycobacteria remodel host membranes through direct physical contact
• •Bacterial load controls compact and spacious phagosome dynamics
• •Bacteria-membrane interactions amplify ESAT-6 activity and prime phagosomal deformation
• •A biophysical mechanism facilitates pathogen survival and escape

Microbiology

Introduction

The interaction of bacteria with the cellular membrane interfaces is essential for their host entry, intracellular life, as well as egress.^1,2 Such highly complex processes are driven by multiple coordinated protein- or lipid-dependent noncovalent interactions.³ Several invasive bacterial pathogens, such as Listeria monocytogenes and Pseudomonas aeruginosa, are known to hijack host membrane interfaces for actin polymerization, resulting in bacterial engulfment by the host cell membrane.^4,5 P. aeruginosa induces cell membrane invaginations via a lipid zipper formation by the interaction between the glycosphingolipid Gb3 and the bacterial surface lectin LecA, in addition to other cell surface receptors.⁶ Clathrin and Arp2/3 complexes have been identified as the key regulators of actin polymerization for zippering pathogens such as Listeria and Yersinia.^7,8 Likewise, Mycobacterium tuberculosis (Mtb) is known to gain entry into the host cell via passive internalization or engulfment mediated by several surface adhesion proteins.^9,10 Mtb are ejected from the host cell by generating actin-based structures called ejectosomes that are aided by autophagy machinery, enabling a non-lytic mechanism.^11,12 Our understanding of bacterial host entry has been reasonably refined in recent times and mediated by a diverse set of pathogen-specific biochemical interactions. However, the disruption of the pathogen-containing vacuolar compartment of the host cell remains an open question, particularly in the case of Mtb.¹³

Mtb survives intracellularly by hijacking phagosome maturation.^14,15 ESAT-6 (6 kDa early secretory antigen target), a highly immunogenic effector protein secreted by Mtb, is critical for the process of phagosome (also referred to as mycobacteria-containing vacuole (MCV)) rupture.^16,17 Purified recombinant ESAT-6 is known to induce lysis of cultured macrophages through permeabilization of both the phagosome membrane as well as the cell membrane upon gaining access to the cytoplasm.^18,19,20 Despite significant evidence that supports the direct role of ESAT-6 in phagosome lysis and virulence,^21,22,23 a recent study reported that ESAT-6 interaction alone is not sufficient for phagosome disruption. Instead, direct bacterial physical contact is important for gross phagosomal membrane disruption.²⁴ In vivo observations also suggest that ESAT-6 requires additional factors such as the lipid phthiocerol dimycocerosates (PDIMs), which are complex branched cell-wall-associated lipids known to induce sterol clustering in host membranes, eventually inducing phagosome disruption.^{25,26,27,28,29,30} Interestingly, Mycobacterium marinum transposon mutants induce phagosome damage despite their inability to secrete ESAT-6, suggesting that ESAT-6 alone may not be sufficient for the ESX-1 virulence function.³¹ Though the ability to damage the phagosomal membrane has emerged as central to Mtb virulence, the mechanism of phagosome membrane remodeling and disruption remains poorly understood.¹³ Mtb is known to be a non-motile bacteria; however, it exhibits sliding mobility upon encountering surfaces.³² Very recently, motile swimming bacteria such as E. coli and B. subtilis have been shown to induce non-equilibrium membrane fluctuations,³³ deformations,³⁴ as well as conferring coupled mobility in synthetic giant vesicles encapsulating the bacteria.³⁵ However, the dynamics and impact of direct physical contact of a non-motile bacteria with the vacuolar compartment remain unknown to date. Understanding the implications of direct physical contact of the bacteria is not only important for vacuolar disruption but also for a broad class of bacterial pathogens' interaction with host cell membranes, both during internalization as well as egress.

In this work, we developed an in vitro reconstitution method to dissect the mechanism of phagosome membrane remodeling induced by the direct physical contact of bacteria by encapsulating live Mycobacterium smegmatis (as a model opportunistic bacterium) inside a host phagosome-like compartment. Furthermore, it also enables us to delineate the role of direct physical contact in phagosome membrane remodeling from virulence factors. We capture the dynamics of bacterial direct contact as well as bacterial-load-mediated host membrane remodeling. We discover that the direct contact of an isolated M. smegmatis with the encapsulating Phagosome-mimicking membrane results in significant lipid wrapping, membrane bending, and local lipid phase separation at the contact site. In contrast, the spacious and compact phagosome mimics with varying bacterial load induce complete lipid phase separation of the compartment. The lipid wrapping is driven by the surface condensation of the compartmental lipids onto the bacterial surface in direct contact and lowers host lipid diffusion by half at the contact area. The direct physical contact under moderate load decreases the area stretch and bending modulus of the host membrane. On the contrary, higher bacterial load induces a scaffold-like architecture, priming the host compartment for rupture. The degree of bending and area expansion of the phagosome-like compartment is dictated by the ratio of the compartmental surface area available to the surface area of the bacterial contact site. Further direct contact of the bacteria enhances the membranolytic potential of ESAT-6, thus contributing to its virulence. Together, we propose that load-mediated direct contact of the bacteria drives the transition of the spacious and compact phagosome population.

Results

Internalization of Mycobacterium tuberculosis leads to heterogeneous compartmentalization

To capture the dynamics of progression of phagosomal architecture during mycobacterial infection, we monitored lysosomal marker acquisition in THP1 macrophages infected with live mCherry-expressing Mtb for different time periods ranging from 2 to 48 h post-infection (Figure 1A). Over the course of infection, the proportion of compact phago-lysosomes decreased by nearly 50%, while the number of spacious phagosomes stayed relatively constant (Figure 1B). Interestingly, from 2 to 6 hpi, there is a significant increase in the population of the compact phagosome while the spacious population decreases (Figure 1I). This may reflect a subpopulation of spacious phagosomes transitioning into compact forms, while the remaining fraction persists as spacious for the live bacteria, which shows a negative correlation. Over time, early formed compact phagosomes appear to undergo disruption, resulting in a decrease in their abundance from 6 to 48 hpi. In contrast, the spacious phagosome population continues to increase, likely due to the addition of newly formed compartments alongside those that do not convert into compact structures in live Mtb bacteria (Figures 1I and S3). Interestingly, a contrasting trend is observed in phagosomes containing dead bacteria. The number of compact phagosomes remains relatively constant over time from 2 hpi to 48hpi compared to live bacteria (Figures 1H–1J), suggesting that only passive surface properties are not enough to disrupt the phagosome; rather, biochemical effectors are important. However, the trend for spacious phagosomes remains similar (Figure 1J). Quantitative image analysis shows a progressive increase in LAMP1 intensity on LAMP1-positive vesicles over time (p < 0.01) (Figures S1B and S1C). This finding is consistent with a previous report highlighting that Mtb containing phagolysosome is increasingly marked by LAMP1 with time.³⁶ Morphometric analysis across time points revealed a persistent heterogeneity among LAMP1-positive vacuoles (Figures S1–S3), with both compact and spacious phago-lysosome populations present (p < 0.0001) (Figures S2A and S2B; Videos S1 and S2). Direct physical contact between Mtb and the phagolysosomal membrane in both compact and spacious states is suggested, regardless of bacterial viability (Figures 1C–1G). Compact phagosomes show significant positive correlations between MTB and LAMP1 fluorescence (For Live Bacteria Pearson’s r = 0.458 (p = 0.0057) and Dead Bacteria Pearson = 0.5515 (p = 0.0007)), whereas spacious phagosomes show a negative correlation (For Live Bacteria Pearson’s r = −0.5413 (p = 0.0004) and Dead Bacteria Pearson’s r = −0.362 (p = 0.0066)), across the Mtb-positive phagolysosome ROIs. The distinct populations of LAMP1-positive vesicles in both dead and live bacterial infections indicate that just the presence of mycobacterial surface in direct physical contact with the phagosomal membrane is sufficient to induce changes in vesicle architecture.³⁶ This data, thus, suggests that changes in lysosomal marker intensity and vesicular structure are not exclusively dependent on bacterial viability or active secretion of effector proteins. Instead, the physical association between Mtb and the host phagosomal membrane might serve as a sufficient trigger for these processes, setting the stage for further pathogen- or host-directed modulation and highlighting the role of the bacterial surface.³⁷

Figure 1.

Lamp1 co-localization with Mtb increases with time and forms heterogeneous populations of compact and spacious phagosomes

(A) Confocal imaging of THP-1 cells infected with Live Mtb tagged with mCherry (red) and immunostained for Lamp1 (green) and fixed with 4% paraformaldehyde/PBS at timepoints- 2, 6, 12, 24, and 48 h post-infection (hpi). Scale bars, 10 μm. Representative images of two technical replicates from one independent experiment. (No. of Lamp1 associated Mtb = 100).

(B) Comparison of percentage changes in phagosome counts obtained from manual segmentation shows a progressive decline in compact phagosomes up to 48 hpi, while spacious phagosomes remain relatively unchanged.

(C) Representative images for compact and spacious morphology of Lamp1 vesicles encapsulating Mtb in infected cells. Scale bars, 10 μm. ROI drawn across the structure in ImageJ to retrieve the intensity plot.

(D–H) Intensity plots of representative image for the Live/Dead Mtb (red) and Lamp1 (green) show the distribution of fluorescence in compact phagosomes, where Lamp1 closely colocalizes with Mtb, shown by the overlapping of signal peaks (Pearson’s r > 0.45) (no. of compact phagosomes = more than 100) (F, G). Intensity plots of representative image for the Live/Dead Mtb (red) and Lamp1 (green) show the distribution of fluorescence in spacious phagosomes where Lamp1 and Mtb signal peaks show no overlapping (Pearson’s r < −0.36) ((no. of spacious phagosomes = 50) (H). Schematic representation of compact and spacious phagosomes in vesicles encapsulating Live Mtb in infected cells.

(I and J) Percentages of compact and spacious Mtb-positive phagolysosomes in both live and dead bacteria infection, obtained from manual segmentation across time points 6, 12, 24, and 48hpi (∗p < 0.05 and ∗∗p < 0.01 in one-way ANOVA).

Direct contact of mycobacteria induces lipid membrane wrapping and creates local diffusion barriers in the phagosome-mimicking membrane

To dissect the role of direct contact of bacteria on host phagosomal membrane shaping and remodeling independent of virulence factors like ESAT-6, we used GFP - M. smegmatis as a model organism. We used a method to encapsulate live Mycobacterium smegmatis inside giant unilamellar vesicles composed of phagosomal membrane lipids (p < 0.05) (as described in STAR Methods and Figures S4A and S4B). We optimized electroformation to encapsulate live fluorescent bacteria. This method reduces the impact of external impurities, such as oil used in emulsion methods, and ensures reliable interaction with the Phagosome-mimicking membrane. Despite being avirulent, the fast-growing M. smegmatis is considered an ideal surrogate for mycobacterial research.³⁸ We first quantified the viability of the bacteria in the experimental conditions through a colony-forming unit (CFU) assay and found it to be ∼60% (Figure S4C). To capture the dynamics of bacterial interaction in real-time, live bacteria were encapsulated in phagosome membrane-mimicking GUVs (40% DOPC, 5.5% DOPE, 22% BSM, 32.5% cholesterol, 0.03% PEG-Biotin) labeled with 0.1% rhodamine-PE (Rh-PE), immobilized using 0.03% Biotinylated-PEG. Upon monitoring over 60 min, suspended bacteria were found to exhibit random Brownian motion inside the reconstituted phagosome. Interestingly, the area of contact between the bacteria and the phagosome-mimicking membrane showed an accumulation of host lipids (Figures 2A and 2B; Video S3). Upon visualizing the GUV in 3D at different planes (Figures 2C, S5A, and S5B), we observed lipid wrapping over the bacterial surface area, as also illustrated in the schematic diagram (Figure 2A). Such interaction induces membrane bending at the contact area in both live and dead bacterial surfaces, indicating surface interactions at play. Furthermore, fluorescence recovery of Rh-PE at the bacterial contact site is reduced by ∼ half following photobleaching in the presence of live bacteria. In contrast, a delayed but full recovery is observed in the case of dead bacteria (Figure 2D), with a Pearson’s r > 0.75 (p < 0.005) indicating a strong positive correlation in their fluorescence recovery dynamics, hinting that strong inter-surface interaction between host and live bacteria creates a diffusion barrier. This subsequently leads to lipid segregation within the phagosome-mimicking membrane. To track the change in lipid reorganization and packaging, we employed two dye model systems (Rh-PE in red and Top-fluor cholesterol in green) and monitored their fluorescence intensity over time (Figure 2E; Video S4; Video S5). We observed that the fluorescence intensity of Top-fluor cholesterol decreases in the vicinity of the contact area (p < 0.0001) (Figures 2F–2H). In contrast, the fluorescence intensity of Rh-PE increases near the contact area, as shown in zoomed images (Figures 2F, 2G, and 2I), where the inverse correlation (Pearson’s r = −0.6227 (p = 0.0002)) between Rh-PE and Mtb-GFP across time indicates that with increasing GFP signal, RFP goes down. In the control condition (without bacteria), there is no change for both red and green channels (Figure S5C). Together, live-bacterial direct contact induces membrane bending, lipid wrapping at the contact area, and creates a diffusion barrier, resulting in the reorganization of lipids and changes in lipid packing (increased disorderliness) as evident from the marked increase in the fluorescence intensity of Rh-PE in the region.³⁹

Figure 2.

Direct surface contact of Mycobacterium smegmatis triggers lipid wrapping and creates a diffusion barrier in the phagosome membrane

(A) Schematic representation of the experimental design depicting the impact of direct contact on immobilized GUV mimicking the host phagosomal membrane.

(B) Time-lapse confocal imaging of GUVs labeled with 0.1% Rh-PE (red) and immobilized using 0.03% PEG-Biotin encapsulated with live M. smegmatis-GFP (green). Scale bars, 10 μm. Representative images of three independent experiments. (No. of GUVs per experiment = 30–35).

(C) Schematic representation and XYZ Projection of the mycobacteria-mediated curvature change from (B).

(D–G) Mean fluorescence recovery curve after photobleaching experiments of direct contact of bacteria (live and dead) on the membrane (Pearson’s r > 0.75, p < 0.005). Data points using mean values from three independent experiments. Scale bars, 5 μm (E). Time-lapse confocal imaging of GUVs of phagosomal membrane composition labeled with 0.05% Rh-PE (red) and 0.05% TopFluor cholesterol (green) and immobilized using 0.03% PEG-Biotin encapsulated with live M. Smegmatis (green). Scale bars, 10 μm. Images are the representation of three independent experiments. (No. of GUVs per experiment = 30–35). Zoomed image of confocal images of the region near the direct contact of bacteria at (F). 0 min, (G). 60 min (Scale bars, 5 μm). White box in e indicates the region for the line intensity profile at respective time points.

(H) Representative fluorescence intensity profile of Topfluor cholesterol (green shades) and Phagosome Rh-PE (Red shades) at two different time points, as stated in f and g (p < 0.0001).

(I) Representative mean fluorescence intensity profile of green channel (Topfluor cholesterol) and red channel (Phagosome Rh-PE) as stated in e. (No. of GUVs = 3) (Pearson’s r = −0.6227, p = 0.0002).

Increasing bacterial load induces membrane disorderliness and global phase separation in the phagosome-mimicking membrane

We next investigated the influence of a moderate bacterial load mimicking a spacious phagosome on the dynamics of the phagosomal membrane. Henceforth, low-to-moderate load and high load mycobacterial conditions are referred to as spacious and compact phagosome mimics, respectively. To capture this, we encapsulated an increased load of bacteria inside the GUV, as illustrated in the schematic (Figure 3A). We observed that without bacterial load, Rh-PE fluorescence intensity appears to be homogeneous throughout the equatorial plane of the membrane. In sharp contrast, in a spacious phagosome mimic, the live bacteria induce striking phase separation in phagosome-like compartments, as evident from the Rh-PE signal that is restricted to bacterial contact areas (Figure 3B). However, no significant phase separation was observed in dead bacteria (Figures S6A and S6B). We observed a change in the fluorescence intensity and thickness of the membrane upon monitoring the GUV mimicking a spacious phagosome for 60 min (Figure 3C; Video S6). Specifically, there is significant mobilization of Rh-PE toward the bottom of the GUV after 60 min, where the larger pool of bacteria settles down (Figure 3D, bottom right panel XZ plane). Also, the accumulation of lipids at the bottom of GUVs induces intraluminal budding at the bottom (Figures 3D and S7). This suggests that bacteria make the membrane more disordered at the contact areas, which increases membrane fluidity and promotes the redistribution of lipids, leading to the subsequent accumulation of lipids at the bottom, resulting in intraluminal budding. Likewise, in both live and dead bacterial conditions, the diffusion of Rh-PE on the membrane was partially slowed down, as evident from the fluorescence recovery after photo bleaching at the equatorial plane (Pearson’s r > 0.93 (p < 0.0001)) (Figure 3E). Super-resolution microscopy revealed that the interactions were caused by the bacteria hitting and retracting off the phagosome-mimicking membrane, which induced higher disorderliness at the point of contact in the GUVs (Figure 3F) as seen from the Rh-PE signal at the equatorial plane (Figure 3G). Interestingly, the membrane sustained the disorderliness despite the bacteria having retracted after hitting the surface and no longer being in contact. These observations collectively suggest that increasing bacterial load inside the GUV results in stronger interactions with the phagosome-mimicking membrane via Brownian motion, affecting the lipid packing. This subsequently leads to phase separation and the mobilization of Rh-PE at the bottom as more stable contacts get established as the bacteria settle to the bottom due to gravity. This might also impact the mobility of the lipids at other planes as a result of stretching and lipid reorganization in the membrane.

Figure 3.

Increasing mycobacterial load induces global phase separation in the phagosome membrane

(A) Schematic representation of the experimental design depicts the impact of a spacious phagosome mimic on an immobilized GUV mimicking the host phagosomal membrane.

(B) 3D projection of a vesicle without and with bacteria exhibiting the bacteria-induced phase separation. Images are the representation of three independent experiments. (No. of GUVs per experiment = 30–35).

(C) Time-lapse confocal imaging of GUVs labeled with 0.1% Rh-PE (red) and immobilized using 0.03% PEG-Biotin, mimicking a spacious phagosome encapsulated with live M. smegmatis (green). Scale bars, 10 μm. Representative images of three independent experiments. (No. of GUVs per experiment = 30–35).

(D) XYZ projection of the mycobacteria load-mediated changes from c representing the budding event over time.

(E) Mean fluorescence recovery curve after photobleaching experiments of spacious phagosome mimic GUVs on the membrane (Pearson’s r > 0.93, p < 0.0001). Data points using mean values from three independent experiments. Scale bars, 5 μm (F) Structured illumination microscopy (SIM) images of GUVs labeled with 0.1% Rh-PE (red) and immobilized using 0.03% PEG-Biotin encapsulated with a moderate load of live M. smegmatis (green), 5 μm. Images are the representation of three independent experiments. (No. of GUVs per experiment = 30–35). Line intensity plots in (G) of lines 1 and 2, respectively.

Compact phagosome mimic scaffolds membrane remodeling through bacterial surface-driven condensation

The behavior of Mmycobacteria within the confined space of the mycobacteria-containing vacuole, whether in a compact phagosome or during replication, remains poorly understood, particularly with respect to the establishment of direct contact with the host membrane.¹⁶ We hypothesized that a substantial increase in mycobacterial load resembling the formation of bacterial clumps within the phagosome⁴⁰ would induce a higher probability of establishing direct contact with the phagosomal membrane. To examine the impact of clump formation on the phagosome membrane, we encapsulated a significantly higher load of mycobacteria to mimic the compact phagosome, thereby increasing the contact area with the membrane, as illustrated in the schematic (Figure 4A). To capture the interactions in real-time, we monitored the compact phagosome mimic for 60 min. Striking membrane tampering is induced by the random attachment of the bacteria (Figure 4B; Video S7). Interestingly, the phagosome-mimicking membrane remains permanently bent even in the case where the bacteria tend to retract away from their initial contact area. Due to the internal organization of bacteria and the membrane remodeling, they together keep the membrane permanently bent. The bacteria exhibit collective mobility as clumps in a compact phagosome mimic compared to that of the spacious phagosome mimic. In line with previous observations, we noted the accumulation of lipids at the bottom of the membrane after 60 min (Figures S9A and S9B), and subsequent budding at the bottom of the GUV. The budding phenomenon results from phase separation, where the reorganization and mobilization of liquid-disordered lipids induce localized membrane curvature at contact areas, making the membrane more deformable (Figures S9A and S9B). To establish phase separation, we used two dye systems as previously described and observed intense phase separation of the phagosome-mimicking membrane by the bacteria (Figure 4C). Next, to visualize the changes in smaller-sized GUVs similar to phagosome size, we encapsulated 4–6 bacteria within smaller GUVs around a size of 8-10 μm. Remarkably, we found that the bacteria caused complete deformation of the membrane (Figures S9C and S9F). This observation indicates that, for a given bacterial load, a higher contact area in a smaller GUV leads to a significantly greater degree of phagosome-mimicking membrane deformation compared to larger GUVs. This suggests a size-dependent effect, where the smaller volume and surface area of the GUVs amplify the impact of bacterial interactions, resulting in more pronounced membrane deformations. Super-resolution microscopy reveals that the phagosome-mimicking membrane becomes more disordered at the contact area of the bacteria. The white spots indicate the colocalization point, where the bacteria and phagosome-mimicking membrane are in contact (Figure 4D), as evident from the intensity of the overlapping regions of the host and the bacteria, which was further validated by performing a Pearson’s correlation analysis (Pearson’s r < -0.3 (p < 0.05)) (Figure 4E). Further, the shape of the bacterial clump appears to drive the shape of the phagosome-mimicking membrane via stretching (Figure 4F). To investigate the role of the bacterial surface on the phase separation of the host lipid membrane, we used a two-dimensional Ising model, where the two states (−1 and +1) represent the lipid ordered and disordered states. Bacterial contacts on the membrane were considered as patches where one kind of lipid has a high probability of accumulating (see supplemental information for details). We find that when the contacts are few (low bacterial load), global phase separation does not occur. However, the membrane can exhibit local phase separation in the vicinity of the contact site (Figure 4G). Interestingly, increasing the number of contacts can drive the entire membrane to phase separate (Figure 4H). The phase separation intensifies as the bacterial contact area on the membrane increases, becoming significantly more pronounced (Figure 4I). However, increasing the size of bacterial contact on the membrane, keeping the fraction of total area in contact constant, results in a decrease in the degree of phase separation (Figures 4J–4L). This is a consequence of the bacteria-inducing local phase separation. As more bacteria come closer together, there is a significant overlap in the locally phase-separated regions, thereby diminishing the effect on the entire membrane. Bacterial contact-induced phase separation is more pronounced when the lipid interactions are such that the membrane is near the lipid demixing point. (Figures S8A and S8B, Supplementary text). This suggests that the probability of interaction between bacterial surfaces and the Phagosome-mimicking membrane is influenced by the size of the phagosomal compartment. The disorderliness and degree of phase separation within the membrane are driven by the percentage of bacteria in contact, leading to increased membrane fluidity. As the number of bacteria within a confined volume increases, the probability of interaction and the formation of bacterial clumps also rises. The morphology of the vesicle is determined by the shape of these bacterial clumps (Figures 3B–3F, S9C, and S9D). Further, the fluidized membrane conforms to and wraps around the clumps, resulting in a more rigid scaffold-like structure. Collectively, these concurrent processes are likely to significantly impact membrane parameters.

Figure 4.

Compact phagosome mimic induces phagosome-mimicking membrane remodeling through condensation driven by bacterial surface interactions

(A) Schematic representation of the experimental design depicting the impact of a compact phagosome mimic on an immobilized GUV mimicking the phagosomal membrane.

(B) Time-lapse confocal imaging of GUVs labeled with 0.1% Rh-PE (red) and immobilized using 0.03% PEG-Biotin mimicking a compact phagosome encapsulated with live M. smegmatis (green). Scale bars, 10 μm. Representative images of three independent experiments (no. of GUVs per experiment = 30–35).

(C) Schematic representation and XY plane of the GUV of phagosomal membrane composition labeled with 0.05% Rh-PE (red) and 0.05% Top-Fluor cholesterol (green), and immobilized using 0.03% PEG-Biotin, mimicking a compact phagosome encapsulated with live M. Smegmatis (green) representing different planes. Scale bars, 10 μm. Images are representatives of three independent experiments. (No. of GUVs per experiment = 30–35).

(D) Structured illumination microscopy (SIM) images of GUVs, with a white spot showing the degree of colocalization.

(E) Showing line intensity graph plot for Rh-PE (red channel) and GFP (green channel), and the marked area with a black line denotes the overlapping region of red and green channels (Pearson’s r < - 0.30, p < 0.05). For the colocalization, the points are mentioned in d.

(F) Structured illumination microscopy (SIM) images of GUVs with high bacterial load show distortion of shape. Images are the representation of three independent experiments. (No. of GUVs per experiment = 30–35). Time-averaged state of the membrane.

(G) With a low load, phase separation is observed only near the contacts. Degree of phase separation 0.137 (N_c = 10, p = 2).

(H) With a high load, global phase separation is seen. Degree of phase separation 0.675 (N_c = 300, p = 2).

(I) Degree of phase separation in the membrane as a function of the percentage of membrane area in contact with the membrane.

(J–L) Varying the contact patch size, keeping the fraction of membrane area in contact fixed. (J and K) show representative time-averaged states for p = 6 and p = 15 respectively. (Parameters in all figures: N = 100, ϕ = 0.5, J/k_BT = 0.4, h/k_BT = 5.0).

Mycobacterial load-mediated direct contact and ESAT-6 virulence facilitate a time-dependent shift in spacious and compact phagosome population

We next questioned if there is any correlation between the bacterial load inside the phagosome-mimicking membrane and the change in the membrane parameters (Figure 5). To test this, we used micropipette aspiration to aspirate GUVs encapsulating varying loads of the bacteria. The presence of Mycobacteria inside the GUVs was confirmed through bright-field microscopy. Depending on the mobility of the bacteria and shape fluctuation of the GUV, we broadly classified GUVs into Spacious phagosome mimic and Compact phagosome mimic (Figures 5A–5C; Video S8; Video S9; Video S10). Spacious phagosome mimics seem to exhibit higher membrane fluctuations and bacterial motion compared to Compact phagosome mimics and phagosome mimics (i.e., without bacterial load) (Figures 5A–5C; Video S8; Video S9; Video S10). Simultaneously, the geometry (aspiration length L_p and vesicle radius R_v, see Figures 5A–5C) of the GUV was monitored to document changes in membrane parameters by the interaction of bacteria with the host membrane. Increasing the bacterial load inside the GUV results in a higher probability of establishing bacterial contact with the phagosome-mimicking membrane, as seen in Figures 3 and 4. To quantify these changes in membrane properties, we aspirated the Spacious phagosome mimic with a constant change in pressure (i.e., ΔP= 50 Pa). We observed that the aspiration length Lp increases significantly (Pearson’s r = 0.9734, p = 0.0052) in spacious phagosome mimics as compared to phagosome mimics without bacteria (Figures 5A, 5B, and 5D). A spacious phagosome mimic is found to reduce the area stretch modulus by 5-fold as compared to the phagosome mimic (Pearson’s r = 0.9871, p = 0.0018) (Figure 5E). Likewise, the bending modulus of the phagosome-mimicking membrane is reduced by 6-fold as compared to the phagosome mimic (Pearson’s r = 0.9871, p = 0.0018) (Figure 5F). Surprisingly, no change in L_p is observed for the Compact phagosome (Figures 5C and 5D). The aspiration length (L_p = 0) suggests an absence of direct area expansion, precluding the calculation of both area stretch modulus and bending modulus. Interestingly, significant deposition of phagosome-mimicking membrane lipid is observed over the surface of the pipette (Figure 5C; Video S10). With increasing applied load, the membrane exhibited enhanced fluidization, which promoted membrane wetting on the pipette surface and led to lipid accumulation and clump formation at the contact site. This observation suggests that as the bacterial load increases, the fluidity of the phagosome-mimicking membrane also increases, leading to greater membrane deformability. In contrast, under high load conditions, bacteria aggregate to form a scaffold-like surface. The fluidized host membrane, influenced by bacterial activity, then wraps around this clump. This interaction collectively results in increased membrane rigidity. These bacterial interactions are efficient in remodeling the membrane properties, but are not sufficient for the bacteria to escape from the phagosome membrane.

Figure 5.

Direct surface contact of mycobacteria enhances the virulence of ESAT-6 and facilitates

(A) Schematic representation and time-lapse Epi-fluorescence imaging of GUVs labeled with 0.1% Rh-PE (red). Scale bars, 10 μm. Aspirated with Pressure (Δp = 50Pa) and monitored over time. Representative images from three independent experiments.

(B) Schematic representation and time-lapse Epi-fluorescence imaging of GUVs labeled with 0.1% Rh-PE (red) mimicking a spacious phagosome with live M. smegmatis (green). Scale bars, 10 μm. Aspirated with Pressure (Δp = 50Pa) and monitored over time. Representative images from three independent experiments.

(C) Schematic representation and Time-lapse Epi-fluorescence imaging of GUVs labeled with 0.1% Rh-PE (red) mimicking a compact phagosome with live M. smegmatis (green). (green). Scale bars, 10 μm. Aspirated with Pressure (Δp = 50Pa) and monitored over time. Representative images from three biologically independent experiments.

(D–F) Plot for the length of protrusion(ΔL_P) vs. pressure (P) from a, b, and c (Pearson’s r = 0.9734, p = 0.0052). Data points shown are the Means ± SD of three independent experiments. (No. of GUVs = 3) (E). Quantifying the area stretch modulus(k_a) from the change in tension(Δτ) vs. direct expansion of the area plot from a, b, and c. Orange and blue spots with bars: experimental points without bacteria and Spacious phagosome, respectively, average ±SD. Dotted line: Linear fit, y = a∗x+b, for without bacteria a = 13.5 ± 1.09, b = (−1.25 ± −0.49)∗10⁻⁶, R² = 0.125 and for Spacious phagosome a = 2.81 ± 0.10 b.(-1.81 ± 0.727)∗10⁻⁷, R² = 0.056455, K_{a(without bacteria)} = 13.5 ± 1.09 mN/m K_{a(Spacious phagosome)} = 2.81 ± 0.10 mN/m (Pearson’s r = 0.9871, (p = 0.0018)). (No. of GUVs = 3) (F). Quantification of the bending modulus (k_b) from the semi-log plot of tension versus direct expansion of area from a, b, and c. Orange and blue spots with bars: experimental points for without bacteria and Spacious phagosome, respectively, average ±SD. Dotted line: Linear fit y = a∗x+b, for without bacteria a = 81.52 ± 1.489, b = −2.45 ± (6.74∗10⁻⁶), R² = 0.605 and for Spacious phagosome a = 13.021 ± 0.9696, b = −2.27 ± (4.28∗10⁻⁷), R² = - 0.097, K_{b(without bacteria)} =(1.33 ± 0.022)∗10⁻²⁰ J, K_{b(Spacious phagosome)} = (0.21 ± 0.02)∗10⁻²⁰ J (Pearson’s r = 0.9871, (p = 0.0018)). (No. of GUVs = 3).

(G) Time-lapse confocal imaging of GUVs labeled with 0.1% Rh-PE (red) and immobilized using 0.03% PEG-Biotin, incubated with 2.5 μM doped FITC-ESAT-6 (green). Scale bars, 10 μm. Images are representative of 3 independent experiments (no. of GUVs per experiment = 30–35).

(H) Time-lapse confocal imaging of GUVs labeled with 0.1% Rh-PE (red) and immobilized using 0.03% PEG-Biotin encapsulated with a moderate load of live M. smegmatis (green) and incubated with 2.5 μM doped FITC-ESAT-6 (green). Scale bars, 10 μm. Images are representative of 3 independent experiments (no. of GUVs per experiment = 30–35).

(I) Time-lapse confocal imaging of GUVs labeled with 0.1% Rh-PE (red) and immobilized using 0.03% PEG-Biotin encapsulated with a high load of live M. smegmatis (green) and incubated with 2.5 μM doped FITC-ESAT-6 (green). Scale bars, 10 μm. Images are the representation of three independent experiments. (No. of GUVs per experiment = 10–15).

Furthermore, to investigate the role of virulent secretory proteins in bacterial escape, we treated these three conditions with a low concentration (2.5 μM) of monomeric ESAT-6 protein and monitored them for 2 h. Interestingly, we observed that in this low concentration and in the absence of bacteria, there was no significant change in the membrane shape (Figure 5G; Video S11). However, a spacious phagosome with significant budding in the membrane was observed after 60 min (0.0667 FPS) (Figure 5H; Video S12). As observed in the previous experiment, under a Compact phagosome, the GUV membrane became more rigid compared to the spacious phagosome (Figure 5C), creating favorable conditions for membrane rupture and bacterial escape from the phagosome. Subsequent treatment with ESAT-6 induces phagosome membrane rupture, resulting in bacterial escape (Figures 5I, S11A, and S11B, Video S13). This indicates that bacterial load enhances the membranolytic activity of the virulence factors. To investigate how mycobacterial lipid composition influences membrane mechanics, we compared the biophysical properties of lipid extracts derived from Mtb H37Ra and Mycobacterium smegmatis. Surface pressure-area isotherm analysis revealed that H37Ra-derived lipids display significantly higher compressibility relative to M. smegmatis lipids, indicating a greater capacity for lateral packing rearrangement under applied stress. This increased compressibility is likely attributable to the enrichment of complex virulence-associated lipids such as PDIM and SL-1, which are known to disrupt ordered lipid packing and promote membrane fluidity and deformability. In contrast, M. smegmatis lacks these specialized lipids, resulting in comparatively rigid membrane behavior. Functionally, enhanced lipid compressibility may facilitate membrane deformation processes such as bending, wrapping, and vesiculation, thereby lowering the energetic barrier for host membrane remodeling during infection. Such properties could potentiate the ability of pathogenic mycobacteria to manipulate phagosomal membranes, promoting destabilization and eventual rupture (Figure S10).

The proposed model suggests that direct contact with Mycobacteria affects the maturation of the MCV. This interaction reduces the bending rigidity of the MCV membrane, which might alter the kinetics of maturation machinery, besides enhancing ESAT-6-mediated membrane disruption in a bacterial load-dependent manner.

Discussion

Many intracellular pathogens, including Mtb, manipulate host membranes via secreted effector proteins. However, the role of direct physical contact between bacteria and host membranes and the mechanisms that dictate the pathogen residency time remain unknown.^1,41 Here, we show that bacterial internalization is highly dynamic and heterogeneous, with both compact and spacious LAMP1-positive vacuoles persisting over time, independent of bacterial viability (Figure 1). Previous studies indicate that internalization is governed not only by host or pathogen specificity but also by the degree of interaction.⁴² For Mtb, the extent of interaction influences intracellular uptake and contributes to heterogeneous compartmentalization.^43,44,45 Our findings reveal that Mtb exploits direct physical contact with host phagosome membranes to mediate a time-dependent shift in the population of spacious and compact phagosomal states (Figures 1B, 1D, 1I, and 1J). We demonstrate that this transition is governed by membrane remodeling by bacterial surface-mediated interaction, where bacterial load and contact area serve as determinants of phagosome fate. At low-to-moderate bacterial loads mimicking spacious phagosome-like compartments, a significant reduction in both bending and area stretch moduli is observed, accompanied by membrane fluidization and vesiculation. This process initiates when mycobacterial contact induces membrane bending, lipid wrapping, and reorganization at the membrane interface, creating diffusion barriers that promote phase⁴⁶ (Figure 2). Notably, these effects are amplified when host membranes are near their lipid demixing point. The relocalization of Rh-PE and Top-Fluor Cholesterol signals indicates that bacterial contact induces localized changes in host membrane properties, consistent with a wetting-like phenomenon at the interaction site. These regions may serve as key sites for cholesterol extraction into MCVs, supporting the known ability of mycobacteria to utilize host cholesterol for their metabolism.⁴⁶ Indeed, intracellular vacuolar compartments represent soft confinements that can be subjected to remodeling and deformation by localized surface contacts and forces such as the active propulsion by motile bacteria.⁴⁷

The extent of phagosome-mimicking membrane phase separation depends on bacterial load (Figures 2, 3, and 4). Under spacious phagosome mimics, freely moving bacteria contacting the membrane at multiple planes induced fluidization and disordered protrusions (Figures 3D and 3E). In contrast, compact phagosome mimics with higher bacterial load cause global reshaping and complete phase separation driven by bacterial clumping (Figure 4B). The phagosome membrane’s proximity to a lipid demixing point further enhances phase separation upon bacterial contact (Figure 3B). Indeed, previous studies^{48,49,50,51,52} have suggested that cell membranes lie close to the lipid demixing point.⁵³ Bacterial surface may induce condensation of phagosome-mimicking membrane interfaces that are close to the lipid demixing point, driving phase separation. Such a phenomenon might also explain the lipid zippering proposed for bacterial surface lectin LecA with the host cell glycosphingolipid Gb3, sufficient for driving complete internalization of the Pseudomonas aeruginosa.⁵ Self-propelling motile bacteria are known to induce stronger non-equilibrium fluctuations in the membrane, resulting in the formation of long protruding tubes.³⁰ Unlike motile bacteria, host membrane deformation by mycobacteria is driven by inter-surface interactions and dictated by the ratio of available surface area of the host membrane and the area of direct contact. Importantly, this mechanical priming synergizes with ESAT-6 activity, as the softened membrane becomes more susceptible to the effects of the virulence factor. Load-dependent shape deformations of encapsulated giant vesicles have also been observed for self-propelled Janus colloids that induce highly branched tether-like protrusions at low/moderate loads and globular reshaping at higher loads, capturing vesicle shapes that do not exist in equilibrium systems.⁵⁴

The striking change in the area stretch modulus and the bending modulus of the compartmental membrane with respect to the bacterial load not only renders it more deformable but also might influence the dynamics of phagosome maturation. Under spacious phagosome mimic, ESAT-6 activity induces membrane vesiculation, reducing the available area of the phagosome-mimicking membrane (Figure 4H), which may eventually progress into a compact phagosome mimic (Figure 5I). This ultimately leads to phagosome rupture, similar to the effects observed under high bacterial load conditions (Figure 5I). This observation is consistent with the observation in the compact phagosome aspiration experiment: the bacterial surface wetting phenomenon results in increasing fluidity for the phagosome-mimicking membrane, where we see significant deposition of the vesicle lipids on the pipette surface. (Figure 5C). Vesiculation may represent a transient intermediate in the conversion from spacious to compact phagosomes. Such compact phagosome compartments preferentially facilitate the activity of secreted virulence factors, thereby promoting Mtb access to the host cytosol.⁴⁵ It is important to note that such a surface-driven biophysical mechanism is expected to be broadly applicable to other intracellular bacteria with similar cell-wall rigidity, while Mtb-specific factors (ESAT-6, PDIMs) may modulate or amplify the outcome. ESAT-6 has been reported to induce independent events such as vesiculation and leakage.⁵⁵ The ESX-1 secretion system facilitates direct interaction with host membranes,²³ and targeting sterol-rich microdomains to disrupt the MCV.⁵⁶ The bending modulus and tension of the membrane are known to regulate the binding kinetics of proteins, the shape generation of the membrane during endocytosis, as well as the rate of budding for inter-organelle trafficking.^57,58,59 The direct-contact mediated manipulation of the phagosomal compartment could facilitate ESAT-6-mediated disruption while potentially delaying phagosome maturation. Indeed, the binding of Ras family proteins is known to be sensitive to membrane tension.^57,60 The mechanical properties of the membranes are critical for processes involving the fission and fusion of membranes.^58,61,62 While our reconstitution and infected-cell imaging support a minimal biophysical mechanism underlying MCV remodeling, it is noteworthy that the magnitude and threshold of these effects in cells are likely influenced by the more complex lipid and protein composition of the MCV. Together, we propose that the ability to transition from spacious to compact states may represent an evolved strategy allowing Mtb to either persist within protective compartments or escape into the cytosol when conditions favor rupture, regulating its intracellular residency time.

Limitations of the study

Although GUVs do not fully reproduce the biochemical complexity of MCVs, they exhibit a highly complex and dynamic composition; however, they are an indispensable reductionist model that enables the investigation of bacterial interactions with host membrane regulatory systems in a controlled condition. Further, fixed cells limit our ability to track individual MCV dynamics, and thus we cannot directly establish or confirm real-time transition from spacious to compact phagosomes at the single-vacuole level that would make an important question for future research. However, population-level analysis of the dynamics of spacious and compact pools of phagosomes indirectly captures overall morphological changes as a result of bacterial contact, including phagosome rupture. Lastly, while our data support a diffusion barrier model, we cannot fully exclude contributions from metabolism-dependent lipid uptake or remodeling by live bacteria.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Mohammed Saleem(saleem@niser.ac.in).

Materials availability

This study did not generate any unique reagents. The plasmids used in this study are available upon request.

Data and code availability

• •All data reported in this paper will be shared by the lead contact upon request.
• •The code for the Monte Carlo simulation used in this study is available on GitHub (https://github.com/nani-git2/Membrane_phase_separation).
• •Any additional information required to reanalyze the data reported in this paper is available upon request.

Acknowledgments

We thank Rohan Dhiman for the kind gift of the ESAT-6 plasmid and GFP-M. smegmatis strain. We thank Shubhay Arvind Dikkar and Sajag Kumar for their inputs during the technical discussion. We also thank Abik Hameem P.M. and Dipam Naskar for their help with micropipette experiments. We acknowledge all the members of the M.S. lab for their critical reading of the manuscript. We thank the Biosafety and Central Imaging and Flow Cytometry (CIFF) facilities at NCBS. SC is supported by the Ramalingaswami Re-entry Fellowship (BT/HRD/35/02/2006), Ministry of Science and Technology, Government of India. MS gratefully acknowledges the DBT-Wellcome Trust India Alliance (grant- IA/I/20/2/505212) and the Department of Atomic Energy for financial support. VS acknowledges core funding from NCBS-TIFR and the Department of Atomic Energy.

Author contributions

Conceptualization, J.B.A. and M.S.; methodology, J.B.A., A.N., J.B., D.K., and J.S.; investigation, J.B.A., A.N., J.B., D.K., and J.S.; formal analysis, J.B.A., A.N., J.B., D.K.; resources, A.N., J.B., S.C., and V.S.; writing – original draft, J.B.A., D.K and M.S.; writing – review and editing, S.C., V.S., and M.S.; data curation, J.B.A. and M.S.; funding acquisition, M.S.; overall supervision, M.S.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Mouse Monoclonal anti-Lamp1 (Human)	DHSB	RRID: AB_626853
Goat anti-Mouse IgG (H + L) Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 647	Invitrogen™	RRID: AB_2535804
Bacterial and virus strains
E. coli BL21 (DE3, plysS) strain	This paper	NA
GFP-M. smegmatis () strain	This paper	NA
Chemicals, peptides, and recombinant proteins
1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC)	Avanti Polar Lipids	850375C
1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE)	Avanti Polar Lipids	850725C
Brain SM	Avanti Polar Lipids	860062
di-stearoyl phosphatidyl ethanolamine-PEG (2000)-biotin	Avanti Polar Lipids	880129
TopFluor™ Cholesterol	Avanti Polar Lipids	810255P
Cholesterol	Avanti Polar Lipids	700000P
lissamine rhodamine B sulfonyl	Avanti Polar Lipids	810150C
Avidin	Sigma–Aldrich	A9275
FITC labeling KIT	Sigma–Aldrich	343210
Phosphate Buffer Saline	Sigma–Aldrich	P2272
Nickel nitrilotriacetic acid (Ni-NTA) Resin	QIEGEN	30410
Triton X-	Himedia	MB031
Bovine serum albumin	MP BIOMEDICALS	162313
DAPI	Invitrogen™	D1306
Sucrose	Sigma-Aldrich	RDD023
Formaldehyde solution	Sigma-Aldrich	F8775
Saponin (from quillaja bark)	Sigma-Aldrich	S4521
Gelatin	HiMedia	TC041
Sodiuim Azide	Sigma-Aldrich	S2002
Software and algorithms
SIGMAPLOT	Sigma plot software	NA
GraphPad Prism	GraphPad Prism Software	NA
Imaris Viewer	Imaris Viewer 10.2.0	NA
ImageJ	ImageJ Software	NA
ORIGIN	National institute of science education and research, India	NA
Python Simulation Code	GitHub Repository	https://github.com/nani-git2/Membrane_phase_separation

Experimental model and study participant details

Mammalian cell culture and infection

THP-1 monocytes were cultured in RPMI 1640 (Gibco™ 31800022) supplemented with 10% fetal bovine serum (Gibco™ 16000-044). THP1 monocytes were differentiated to macrophages by treatment with 20 ng/ml of phorbol myristate acetate (PMA) (Sigma-Aldrich, P8139) for 24 h, followed by incubation in PMA-free RPMI 1640 medium for 24 hours, and used for infections. THP-1 monocyte-derived macrophages were incubated with mycobacteria at an MOI of 10 for 2 h, followed by the removal of extracellular bacteria by multiple washes. Infected cells were fixed with 4% paraformaldehyde (Sigma 158127) at 2, 6, 12, 24, and 48 hpi and were used for subsequent experiments.

Method details

Preparation of GUV

Giant unilamellar vesicles (GUVs) that imitate the phagosomal membrane were made up of 40% DOPC, 5.5% DOPE, 22% BSM, 32.5% cholesterol, 0.03% PEG-Biotin, and 0.1% Rhod PE. GUVs with a phagosome-like composition were created by following and slightly altering the protocol given in.⁵⁸ In summary, conductive ITO-coated glass from Nanion Technologies, GmbH was equally covered with 15μL of the lipid mixture at a concentration of 5 mg/ml, and it was vacuum-dried for a minimum of three hours. After rehydrating the dried lipid film in a 300 ± 5 mOsm sucrose solution of pH (6-6.5, the GUVs were developed for three hours at 40-45°C with a sine voltage of 2V and 10 Hz. Using 7H9 media containing precultured Mycobacteria, the lipid film was rehydrated in order to encapsulate live Mycobacteria within the GUVs. GUVs are stable in the experimental conditions. For encapsulation experiments, Mycobacterium cultures were grown in 7H9 medium to an optical density OD of 1.0–1.2. The bacterial suspension was subsequently used for encapsulation within giant unilamellar vesicles (GUVs). To assess bacterial viability under the experimental conditions, colony-forming unit (CFU) assays were performed, which indicated that approximately 60% of the encapsulated bacteria remained viable.

Expression and purification of ESAT-6

The ESAT-6 protein was expressed using the Escherichia coli BL21 (DE3, pLysS) strain. The strain was transformed with the plasmid pET22b containing the Mycobacterium tuberculosis gene Rv3875, which encodes the ESAT-6 protein. Transformed cells were grown in Luria-Bertani (LB) medium at 37°C until the optical density at 600 nm (OD600) reached approximately 0.6. Protein expression was induced by adding 1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG), followed by overnight incubation at 18°C. Post-induction, the bacterial cells were harvested by centrifugation and lysed via sonication. The cell lysate was then subjected to affinity purification using nickel-nitrilotriacetic acid (Ni-NTA) resin, per the manufacturer’s instructions (GE Healthcare, UK). The eluted fractions containing the His-6-tagged ESAT-6 protein were analysed using SDS-PAGE to confirm protein purity. Purified fractions were subsequently dialyzed against an appropriate buffer, concentrated, and stored at -80°C in aliquots for further use.

Light microscopy

An 8-well Ibidi chamber slide was coated with 10μl of 1 mg/ml streptavidin, followed by the addition of 200μL of biotinylated GUVs. The GUVs were then incubated for 15 minutes to cause immobilisation. Leica TCS SP8 LCSM imaging was carried out with the proper Rhodamine-PE (561 nm) and Alexa-488 (488 nm) lasers. Every experiment was conducted using the same laser power and gain settings.

Initially, images were taken with a lower laser intensity for the Fluorescence Recovery after Photobleaching (FRAP) measurements. A 5μm radius of bleaching was achieved within the chosen ROI after 30 seconds of photobleaching at maximum laser power. The recovery curve was then measured, and the laser intensity was adjusted to the attenuated level. The GUV under visualisation was photobleached at its equatorial plane. Following three repetitions, the FRAP curves were standardised for each condition.

Micropipette aspiration

Using a 100X objective lens, all measurements were made using an Olympus IX83 epifluorescence microscope fitted with Hoffman modulation optics. Three-axis hydraulic micromanipulators are utilised to position the pipette. Capillaries with a diameter of 10μm are used to extract micro-pipettes (Narashige Micro Forge MF-900, Sutter Instruments CO P-87). On top of the unfinished DIY coverslip, 100 μl of GUVs was put. The suction pressure P applied to a bilayer vesicle produces a uniform membrane tension (τ), which is described by a geometric relation based on the pipette (Radius)R_pipette and vesicle (Radius)R_GUV of the vesicle spherical segment exterior to the pipette.

$$\tau ={\left(1/2\right)}^{\ast }\left({\mathrm{R}}_{\text{pipette}}\Delta \mathrm{P}/1-\left({\mathrm{R}}_{\text{pipette}}/{\mathrm{R}}_{\text{GUV}}\right)\right)$$

^63,64 Movement of the projection length L_p inside the pipette directly measures the area increase. Using measurements of the vesicle spherical segment and cylindrical projection dimensions, we numerically computed changes in apparent area Δa for each displacement ΔL_p of the projection length with simple geometric formulae and the constraint of a fixed vesicle volume.⁶⁵ The proportionality between apparent area and length is easily seen in the following first-order approximation:

$$\Delta \mathrm{a}=2{\mathrm{\pi }}^{\ast }{{\mathrm{R}}_{\text{pipette}}}^{\ast }{\Delta \mathrm{L}}_{\mathrm{p}}+2\pi {{\mathrm{R}}_{\text{pipette}}}^2$$

While the reference area a₀= 4π∗R_GUV² corresponds to a spherical vesicle radius R_GUV. These formulas combine to give the following expression for the areal strain:

$$\Delta \mathrm{a}/{\mathrm{a}}_0=\left({{\mathrm{R}}_{\text{pipette}}}^{2\ast }\left(1+{\mathrm{L}}_{\mathrm{p}}/{\mathrm{R}}_{\text{pipette}}\right)\right)/{2{\mathrm{R}}_{\text{GUV}}}^2$$

Displacement of the projection length of an aspirated vesicle under a change in suction pressure is shown in Figure 4D. Even though the apparent area can be measured over a tension range from 76 μN/m to 550 μN/m as demonstrated in Figure 4E to determine the area stretch modulus (K_a), we plot the change in tension (τ) against direct area of expansion, and the resulting slope is K_a.to determine the bending modulus (K_b), we plot the semi-log of tension(τ) against the direct area of expansion, and resulting slope is

$$\text{slope}=\frac{8\pi K_b}{k_{B}T}$$

Using this, we calculate the bending modulus (K_b), All the experiments were performed at room temperature.⁶⁵

Mycobacterial strains and growth conditions

M. tuberculosis H37Rv expressing mCherry strains were grown in Middlebrook 7H9 (BD Difco 271310) supplemented with 10% OADC (BD Difco 211886) at 37 °C. Before infection, bacterial clumps were removed by centrifugation at 80 × g, and supernatant was pelleted, re-suspended in RPMI media, and used for infection, as discussed previously.⁶⁶ All the experiments were performed inside a BSL-3 facility.

Immunostaining, imaging, and image analysis

For immunostaining of LAMP1 marker, differentiated THP-1 macrophages after infection were fixed with 4% paraformaldehyde, washed with PBS, and permeabilized with SAP buffer (0.2% saponin (Sigma-Aldrich S4521), 0.2% gelatin (Himedia Laboratories TC041) in PBS) for 10 min at room temperature. Primary antibody Lamp1 (DSHB, clone H4A3) was prepared in SG-PBS (0.02% saponin, 0.2% gelatin in PBS) and incubated at room temperature for 2 hours. After washing with SG-PBS, cells were incubated in AlexaFluor 647 (Life Technologies, Invitrogen) prepared in SG-PBS for 1 h at room temperature, washed, stained with 1 μg/ml of 4[prime],6-diamidino-2-phenylindole (DAPI) for 30 minutes at room temperature. The coverslips were then mounted using a Mowiol-based mountant and imaged using Olympus FV3000 confocal microscope using a 60x objective.

Images were analyzed by CellProfiler 4.2.8. and pipelines similar to those previously established⁶⁷ were used. In all cases, images were segmented to identify nuclei, cells, bacteria, and Lamp1-positive compartments. Objects such as bacteria and Lamp1-positive vesicles were related to individual cells to obtain single cell statistics, and multiple parameters relating to their numbers, sizes, intensities, as well as Mtb-Lamp1 associations were extracted. MS Excel and GraphPad 8.0.1 were used for data analysis and plotting graphs. Statistical significance between different sets was determined using the Mann-Whitney unpaired test, the t-test with unequal variance, and the paired t-test for data sets with normal distributions. Pearson’s correlation coefficients and the corresponding p-values were calculated with a confidence interval of 95%.

Lipid monolayer assay

The monolayer experiments were performed in a KSV NIMA Langmuir balance. The instrument consisted of two barriers used during the compression isotherms assays, and as a surface pressure sensor a Wilhelmy microbalance with a filter paper plate was used. Monolayers of M.tb H37Ra and M.smegmatis lipids were formed by spreading the lipid/chloroform solution (1mg/ml) in a dropwise manner and kept for 15 mins to equilibrate the monolayers and for the evaporation of the chloroform. Autoclaved water was used as the subphase with the maintained temperature of 25°C for all the Langmuir experiments. Isotherms were recorded with uniform compression of the monolayer at a constant speed of 1 mm/min. Isotherms were recorded until the collapse pressure (πc) was reached. Each experiment was performed thrice.

Image analysis

Image analysis and processing were done using the ImageJ program. The average fluorescence intensity was quantified using the line intensity profiling at the direct contact sites along the equatorial plane of GUVs. The same detection configuration was followed for every experiment. Every experiment set used the same detection configuration. ImageJ was used to manually quantify the diameters of the vesicles as well as the THP1 cells experiment. The relative area between the bacterial surface and the phagosome inner membrane is quantified with manual segmentation using ImageJ. The populations of compact and spacious phagosomes were classified based on the relative area ratio between the phagosome and the bacteria at the imaging plane. Phagosomes with an area ratio ≤ 1 were defined as compact, whereas those with an area ratio > 1 were classified as spacious.

Quantification and statistical analysis

Image acquisition and processing were performed using ImageJ software. Fluorescence intensity profiles along the equatorial plane of giant unilamellar vesicles (GUVs) were quantified by measuring the mean membrane fluorescence intensity (Fvesicle) after background subtraction (Fbackground). Identical imaging and detection settings were maintained across all experimental conditions to ensure consistency. Vesicle diameters were measured manually using ImageJ.

Statistical analyses were carried out using one-way analysis of variance (ANOVA) in SigmaPlot software and Mann-Whitney unpaired test t test with unequal variance in GraphPad Prism. Data are presented as mean ± standard deviation (SD) unless otherwise stated. A p-value of < 0.05 was considered significant. ∗ represents p-value < 0.05. ∗∗ represents p-value < 0.01. ∗∗∗ represents p-value < 0.001. ∗∗∗∗ represents p-value < 0.0001.

Supplemental information

Video S1. 3D-reconstruction of Mtb-containing LAMP1-positive compact vesicles in THP1 cells, related to Figure 1

3D-reconstruction of confocal images of Mtb-mCherry (red) inside Lamp1-positive vesicles (green) inside THP1 cells. The images were acquired in the Olympus FV3000 system and reconstructed using Imaris Viewer 10.2.0 software, where the rotation is initiated after zooming into the entrapped bacteria at a scale of around 4-6μm. Compact vesicle formation can be seen around the bacteria where there is close association of LAMP1 with the Mtb, both in the video as well as the representative snapshots.

Video S2. 3D-reconstruction of Mtb-containing LAMP1-positive spacious vesicles in THP1 cells, related to Figure 1

3D-reconstruction of confocal images of Mtb-mCherry (red) inside Lamp1-positive vesicles (green) inside THP1 cells. The images were acquired in the Olympus FV3000 system and reconstructed using Imaris Viewer 10.2.0 software, where the rotation is initiated after zooming into the entrapped bacteria at a scale of around 4–6 μm. Spacious vesicle formation can be seen around the bacteria, where there is a comparatively loose association of LAMP1 with the Mtb, as there is a pocket for the bacteria, but the volume of the vesicle seems to exceed that of the Mtb, which is emphasized by the excess space inside the vesicle.

Video S3. Real-time dynamics of direct contact of bacteria with phagosome-like membranes, related to Figure 2

Time-lapse confocal imaging of phagosomal membrane mimic GUVs labeled with 0.1% Rhodamine-PE (Rh-PE), encapsulating live bacteria, were immobilized using 0.01% Biotinylated-PEG. Upon monitoring over 60 min (0.0508 fps). Scale bars, 10 μm. with accelerated Video.

Video S4. Phagosome-like membranes without bacteria, related to Figure 2

Time-lapse confocal imaging of GUVs of phagosomal membrane composition (without bacteria) labeled with 0.05% Rh-PE (red) and 0.05% Top Fluor cholesterol (green) and immobilized using 0.03% PEG-Biotin. Upon monitoring over 60 min (0.0083fps). Scale bars, 10 μm. with accelerated Video.

Video S5. Direct contact of bacteria-induced local phase separation in phagosome-like membranes, related to Figure 2

Time-lapse confocal imaging of GUVs of phagosomal membrane composition labeled with 0.05% Rh-PE (red) and 0.05% Top Fluor cholesterol (green) and immobilized using 0.03% PEG-Biotin encapsulated with live M. Smegmatis (green). Upon monitoring over 60 min (0.0083fps). Scale bars, 10 μm. with accelerated Video.

Video S6. Real-time dynamics of spacious phagosome-mimicking GUVs encapsulating live Mycobacterium smegmatis, related to Figure 3

Time-lapse confocal imaging of GUVs of phagosomal membrane composition labeled with 0.1% Rh-PE (red) and immobilized using 0.03% PEG-Biotin mimicking spacious phagosome encapsulated with live M. smegmatis (green). Upon monitoring over 60 min (0.0083fps). Scale bars, 10 μm. with accelerated Video.

Video S7. Real-time dynamics of compact phagosome-mimicking GUVs encapsulating live Mycobacterium smegmatis, related to Figure 4

Time-lapse confocal imaging of GUVs of phagosomal membrane composition labeled with 0.1% Rh-PE (red) and immobilized using 0.03% PEG-Biotin mimicking Compact phagosome encapsulated with live M. smegmatis (green). Upon monitoring over 60 min (0.0083fps). Scale bars, 10 μm. with accelerated Video.

Video S8. Micropipette aspiration of phagosome-like membrane compartment, related to Figure 5

Time-lapse Epi-fluorescence imaging of GUVs of phagosomal membrane composition labeled with 0.1% Rh-PE (red). Scale bars, 10 μm. Aspirated with Pressure (Δp = 50Pa) and monitored over time.

Video S9. Micropipette aspiration of spacious phagosome-mimicking compartment, related to Figure 5

Time-lapse Epi-fluorescence imaging of GUVs of phagosomal membrane composition labeled with 0.1% Rh-PE (red) mimicking spacious phagosome encapsulated with (green). Scale bars, 10 μm. Aspirated with Pressure (Δp = 50Pa) and monitored over time.

Video S10. Micropipette aspiration of compact phagosome-mimicking compartment, related to Figure 5

Time-lapse Epi-fluorescence imaging of GUVs of phagosomal membrane composition labeled with 0.1% Rh-PE (red) mimicking a compact phagosome encapsulated with live M. smegmatis (green). Scale bars, 10 μm. Aspirated with Pressure (Δp = 50Pa) and monitored over time.

Video S11. Interaction of EAST-6 with phagosome-like membrane compartment, related to Figure 5

Time-lapse confocal imaging of GUVs of phagosomal membrane composition labeled with 0.1% Rh-PE (red) and immobilized using 0.03% PEG-Biotin incubated with 2.5 μM doped FITC-ESAT-6 (green). Upon monitoring over 120 min (0.0667 fps). Scale bars, 10 μm. with accelerated Video.

Video S12. Interaction of EAST-6 with a spacious phagosome-mimicking compartment, related to Figure 5

Time-lapse confocal imaging of GUVs of phagosomal membrane composition labeled with 0.1% Rh-PE (red) and immobilized using 0.03% PEG-Biotin mimicking spacious phagosome encapsulated with live M. smegmatis (green) and incubated with 2.5 μM doped FITC-ESAT-6 (green). Upon monitoring over 120 min (0.0667 fps). Scale bars, 10 μm. with accelerated Video.

Video S13. Interaction of EAST-6 with compact phagosome-mimicking compartment, related to Figure 5

Time-lapse confocal imaging of GUVs of phagosomal membrane composition labeled with 0.1% Rh-PE (red) and immobilized using 0.03% PEG-Biotin mimicking Compact phagosome encapsulated with live M. smegmatis (green) and incubated with 2.5 μM doped FITC-ESAT-6 (green). Upon monitoring over 120 min (0.033 fps). Scale bars, 10 μm. with accelerated Video.