Single molecule super resolution microscopy reveals formation of Dengue NS2B3 cluster on mitochondrial network and its effect on fragmentation

Abstract

It is challenging to understand viral proteins at a single-molecule level and visualize their dynamics in a cellular environment. This calls for sophisticated microscopy techniques (such as,SMLM) and single-molecule analysis that can reveal details with the precision of a single molecule. In Dengue type-2(Denv − 2) infection, NS2B3 is recognized as a critical protein complex responsible for proteolytic activity and processing of viral polyprotein. We study the distribution of NS2B3 protein and its clusters. Accordingly, two distinct photoactivatable probes(mEos − NS2B3 and PAGFP − NS2B3) and a fluorescent probe (eGFP-NS2B3) were constructed. Results confirmed the formation of NS2B3 clusters in close vicinity to the mitochondrial network. Statistical analysis of single-molecule data (images) facilitated the estimation of critical cluster parameters(such as cluster density, number of molecules/cluster,and its spread). In addition, regional analysis suggests a correlation between NS2B3 cluster formation and fragmentation of mitochondrial network which is further supported by time-lapse confocal imaging. In other words, the presence of cluster depletes mitochondrial segment. To further demonstrate, time-lapse super-resolution imaging is carried out that enabled visualization of the formation of NS2B3 clusters. Understanding the underlying dynamics of NS2B3 clustering at the single-molecule level may facilitate the identification of potential drug targets,and methods to disrupt them.

Introduction

Super-resolution imaging is perceived as the next-generation imaging technique for cell and disease biology. Advanced imaging techniques have shed critical insights (at single-molecule level) in the broad field ranging from cell biology to virology^1–6. In recent years, dengue has emerged as a pandemic in many parts of the world, with approximately 13 million reported dengue cases worldwide since the beginning of the year 2024. Of these, nearly 42,000 were classified as severe, and more than 8,700 dengue-related deaths have been reported globally⁷. Dengue virus is an enveloped virus that consists of capsid protein (C) and a positive single-stranded RNA molecule. The dengue genome encodes a polyprotein which, upon cellular processing, releases ten proteins, of which three are structural proteins (C, prM/M, E), and seven are non-structural proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B, and NS5)⁸. Out of these proteins, it is the non-structural proteins that are directly involved in viral replication and assembly^8–12. NS2B, a non-structural protease sub-unit, promotes conformational changes in the NS3 structure^13–15. In addition, accumulation of NS2B is also observed in ER¹⁶ However, NS3 protein has the distinct ability to cleave parts of the polyprotein and is actively involved in viral replication due to its ATPase/helicase and RNA triphosphatase domain in the C-terminal region^17–19. Together, NS2B/NS3 (in short, NS2B3) is a highly conserved protein and forms a weak non-covalent complex. The NS2B3 complex is responsible for proteolytic activity and processing of viral polyprotein at the junctions with other non-structural proteins (NS2B-NS3, NS2A-NS2B, NS3-NS4A, NS4B-NS5)^20,21. Specifically, NS2B-NS3 protease is crucial in correctly processing viral polyproteins during dengue pathogenesis.

Like most flaviviruses, dengue virus uses the host cellular mechanism to multiply and infect other healthy cells^8,22. However, little is known about what transpires at the single-molecule level, be their distribution during infection. Such information is vital to understanding the underlying mechanism and selective drug targets to disrupt the onset of these processes, hoping to inhibit the chain of viral infection.

Studies over the years suggest that the NS2B3 protein complex interacts with several cellular proteins, which is essential for effective infection of dengue virus to host cell^23–25. Dengue infection begins with virion E protein binding to the host cell surface ligands²⁶. The dengue virus, up on receptor-mediated endocytosis, releases viral RNA genome into host cytosol and starts producing single viral polyprotein²⁷. In conjunction with host cellular proteins, the viral proteins facilitate RNA replication, budding, and complete virion maturation^28,29. Dengue virion utilizes all the host organelles involved in protein expression to maturation. The virion matures in the Golgi bodies, and the Golgi network helps to release into the extracellular region^30,31.

Recent studies reported that NS2B3 protease targets host mitochondria, which induces matrix-localized GrpE protein homolog 1 (GrpEL1) cleavage. The GrpEL1 protein, a cochaperone of the Hsp-70 protein, cleavage led to the dysfunction of the Hsp-70 protein, hampering overall protein folding. Thus, GrpE1 cleavage leads to mitochondria and mitochondrial-associated membranes (MAMs) disruption, causing mitochondrial fragmentation³². NS2B3 proteases induce mitochondrial disruption responsible for inducing apoptosis in human medulloblastoma cells. It also blocks RIG-1 and MDA-5 translocation to the mitochondria from the cytosol, leading to suppression of the host immune response³³. However, the exact consequences of NS2B3 protease dysfunction at the molecular level are yet to be studied. Most of the existing studies are performed by diffraction-limited microscopy techniques that are known to give ensemble information. Hence, the dynamics and related kinetics at a single-molecule or cluster level were never revealed. Recent confocal microscopy study show mitochondrial damage due to dengue infection^34–37. Dengue protease, NS2B3 interferes with mitochondrial fusion and mitochondrial dynamics by cleaving mitochondrial outer membrane protein mitofusin 1 and mitofusin 2 (MFN1/2)³⁴. Specifically, mitofusins help in mitochondrial fusion by tethering adjacent mitochondria³⁸, and decreased mitofusin activity leads to mitochondrial fragmentation³⁹. It is found that MFN1 positively regulates host antiviral signaling, and MFN2 is responsible for mitochondrial membrane potential during dengue infection⁴⁰. The cleavage of both MFNs by dengue protease would attenuate interferon production and promotes dengue-induced cell death³⁴. But still, it is not well known about the mode of attachment of NS2B3 protease complex on mitochondrial membrane and its fate after attachment. There is no information available at the single-molecule level and thus, the underlying mechanism is unknown. It requires sophisticated super-resolution microscopy and single-molecule-based biological protocols. A better understanding of the infection process may lead to precision drug targeting and selective therapy for the disease.

Here, we studied the NS2B3 cluster formation and its effect on mitochondrial network at the single-molecule level using single-molecule localization microscopy (SMLM). SMLM has advantages as compared to other super-resolution microscopy techniques for quantifying individual protein molecules and their collective dynamics. These viral proteins are known to undergo processes such as migration, cluster formation, and localization to the specific organelle^32,34. We study the dynamics of NS2B3 complex post-transfection in the cellular system, and investigate the collective dynamics of single-molecules by suitably fusing the NS2B3 protein with a photoactivable protein. This has allowed us to understand processes both at the single-molecule and ensemble (cluster) levels. The biophysical parameters (cluster size, density, and copy number) related to the viral protein complex may be critical in assessing the progress of the Denv-2 infection process.

Results

In the present study, we investigate the role of NS2B3 complex to the transfected host cell. To accomplish this, we constructed appropriate fluorescent probes to localize the site or organelle of activity. Specifically, two photoactivable probes (PAGFP-NS2B3 and mEos-NS2B3) and a fluorescent probe (eGFP-NS2B3) are designed. The generation of recombinant probes eGFP-NS2B3, PAGFP-NS2B3 and mEos-NS2B3 was confirmed by sequencing. Results of PCR, restriction analysis and sequencing along with recombinant vector map were given in Supplementary Notes 1-4. The recombinant plasmids (eGFP-NS2B3, PAGFP-NS2B3, and mEos-NS2B3) were used to transfect NIH3T3 cells (mouse embryonic fibroblast cell line) to study the effect of dengue NS2B3 with the cellular organelles.

Cell Transfection and Localization Study

Figure 1 shows NIH3T3 cells transfected with eGFP-NS2B3 / mEos-NS2B3 / PAGFP-NS2B3 (individual transfections) recombinant plasmids followed by the standard protocol of washing, culture, and incubation for 48 hr. Here, eGFP − NS2B3 is used for carrying out normal fluorescence study using confocal microscopy whereas, mEos-NS2B3 and PAGFP-NS2B3 are used for single molecule studies using super-resolution microscopy. This is largely due to the fact that photoactivable probes suits better for SMLM which can be attributed to their stochastic photoactivation property. The transfection was confirmed by exposing it simultaneously to 405nm light and blue light (470-490 nm) for which the fluorescence is observed in green. It is clear that the protein complexes (eGFP / PAGFP-NS2B3 / mEos-NS2B3) are expressed in the cell. A few enlarged images of individual cells are also shown in Fig. 1A. The corresponding transfection efficiency is discussed in Supplementary Note 5. To ascertain the presence of NS2B3 protein complex and its effect on mitochondrial network, we have carried out additional studies. The experiment involves transfecting NIH3T3 cells with eGFP − NS2B3 recombinant plasmid. Upon expression of the eGFP − NS2B3 recombinant protein, the cells were washed and labeled with MitoTracker orange (Invitrogen, USA). The specimen is then fixed following standard protocol and imaged using confocal microscopy. The individual fluorescence images and their overlay images along with the transmission image are shown in Fig. 1B. The eGFP − NS2B3 recombinant protein are in the vicinity of mitochondria (labeled with MitoTracker Orange (see, enlarged images R1 and R2 in Fig. 1B). In fact, eGFP-NS2B3 clusters are in green blobs in Fig. 1B, while the mitochondrial network is in magenta color. This makes mitochondria as a possible target organelle as reported in other studies^32,34. To ensure that the observed clusters are not artifacts caused by fluorescence probe-induced aggregation or self-quenching, we carried out additional confocal studies. Accordingly, cell transfection experiments were performed with fluorescent protein alone (i.e., only eGFP). Figure 1B (right) shows the confocal image of cells transfected with the eGFP only. The uniform distribution of fluorescence throughout the cell is quite evident, indicating that the clusters are not artifacts. In addition, experiments were carried out to check the localization of unconjugated NS2B3 as discussed in Supplementary Note 6. This suggests that the clusters observed in eGFP-NS2B3 transfected cells are indeed due to the underlying molecular processes. In addition, we have carried out anticorrelation analysis that use pixel-based correlation map as shown in Fig. 1C and Supplementary Note 13. The details of pixel-based correlation analysis is discussed in the methods section. A few enlarged regions (R1 and R2) are chosen that indicate values close to 0 in the map, suggesting positive correlation between NS2B3 clusters and the mitochondrial segments. This implies the presence of NS2B3 cluster in close vicinity to the local mitochondrial network lead to its fragmentation. This is also supported by intensity analysis on more cells as detailed in Supplementary Note 13.

Fig. 1. Transfection and Localization studies of NS2B3 using confocal microscope.

A NIH3T3 cells expressing, eGFP − NS2B3, mEos − NS2B3, and PAGFP − NS2B3 proteins. The transmision image is overalyed with fluorescence image. The transfected cells in each sample is marked with yellow box and their enlarged images are shown in C1, C2 and C3 respectively. B Separate and overlayed fluorescence image of the the cell sample (eGFP − NS2B3 transfected NIH3T3 cells (cell1 and cell2) stained with MitoTracker Orange dye). Few enlarged sections (R1 and R2) are also shown in the overlay image which clearly indicates the vicinity of eGFP-NS2B3 on mitochondrial network. More cells are given in Supplementary Note 7. In addition, cells transfected with EGFP plasmid alone is shown in the right panel. C Pixel-based correlation map showing the presence of NS2B3 clusters depletes mitochondria. Additional data is shown in Supplementary Note 13.

NS2B3 Protease and Mitochondrial Fragmentation

As a next step, we studied the effect of these NS2B3 blobs on the mitochondrial network. To do so, we conducted image-based analysis to access changes in mitochondrial network. Similar studies are carried out for analyzing mitochondrial fragment in cellular system^41–45. Here, we have carried out transfection of cells followed by staining to understand the effect of NS2B3 protease on mitochondrial network. Specifically, we have transfected the cells by eGFP-NS2B3 plasmid and upon 48hr post-transfection, the cells were stained and fixed. Here, we have considered two independent controls (cells stained with mitotracker orange and cells transfected with eGFP (without NS2B3) followed by staining with mitotracker orange) and compared them. The transfection, staining and fixing was performed by following the protocols mentioned in the methods section. The fixed cells are imaged using confocal microscope were shown in Fig. 2A. Two different spectral windows are selected to record eGFP transfection (green) and mitochondrial network (labelled with mitotracker orange, magenta). The combined image is shown along with the skeletonized image. The eGFP proteins and mitochondria networks images are captured sequentially with Z-step size of 250nm. Subsequently, mitochondria network is analyzed after skeletonization and the results are compared with the cells that are only stained with mitotracker orange dye (without transfection). Two key parameters (mean fragment length and fragments/area) are compared for the cells with only Mitotracker staining and cells transfected with eGFP followed by Mitotracker staining as shown in Fig. 2B. We have used non-parametric t-test (Mann Whitney) for statistical comparison that shows statistical significance as ns (not-significant). This clearly suggests that the transfection does not alter or affect mitochondrial network. The cells shown in Fig. 2C is control and Fig. 2D is first transfected by eGFP-NS2B3, followed by Mitotracker staining before carrying out skeletonization and fragment analysis. The transfected cells are identified from the fluorescence (at 507 nm) observed from the eGFP-NS2B3 molecules as discussed in Fig. 1. Subsequently, the acquired images are subjected to skeletonization using standard inbuilt Matlab scripts and self-developed algorithms (see, Fig. 2C, D). Visually, the presence of small segments is evident in NS2B3 transfected cell, suggesting fragmentation of mitochondrial network. This can be further confirmed by comparing it with a control experiment that shows a relatively long and connected mitochondrial network. To quantify, we carried out statistical analysis related to number of fragments per unit area, fragment length, and entropy (measure of randomness). Analysis shows that, the number of fragments/area has increased by ~ 3.6-fold for NS2B3 transfected cells (see, Fig. 2E). Specifically, the number of mitochondrial fragments per area is found to be significantly large (27.05 ± 0.593 for eGFP − NS2B3) as compared to the control (7.52 ± 0.207). Additional cell data is shown in Supplementary Note 16. This confirms the breakdown of the mitochondrial network by the viral complex NS2B3. Moreover, we have used the Shannon entropy measure to assess randomness (distribution of fragments) in each case. For control, mitotracker stained cells were expected to preserve the connectedness in the mitochondrial network, which is quite evident for control suggesting low entropy. On the other hand, eGFP-NS2B3 transfected cells show a significantly higher entropy as compared to the control cells. In addition, we have carried out region-wise site-specific analysis to demonstrate the effect of NS2B3 clusters on mitochondrial network. This is discussed in Supplementary Note 12.

Fig. 2. Fragmentation study for control cells and NS2B3 transfected cells showing shorter mitochondrial fragment effected by NS2B3 protease.

A The mitochondria network (magenta) of the cells transfected with eGFP (green) and skeletonized images of the mitochondria network. Alongside, the combined image is also shown. B The controls are compared based on mitochondrial fragment analysis (mean fragment length and fragments/area). C Fluorescence image for NIH3T3 cells labeled with Mitotracker Orange (Control-without NS2B3) along with skeletonized images suggesting a connected mitochondrial network. A few enlarged images regions are also shown for visualizing better fragmentation. D Fluorescence image for NS2B3 transfected NIH3T3 cells labeled with Mitotracker Orange. Alongside skeletonized images are also shown that indicates fragmented mitochondrial network. E Statistical analysis for the number of fragments per unit area, fragment length and entropy are also shown. The analysis suggests a decrease in mean fragment length (0.694 ± 0.019 μm) for NS2B3 transfected cells as compared to non-transfected cells (1.273 ± 0.065μm). We have used non-parametric t-test (Mann Whitney) for statistical comparison. The corresponding statistical significance (P-values) is indicated as, ***p < 0.0001, and ns indicates non-significant. The statistical analysis presented here are based on 15 cells each from 3 different experiments.

NS2B3 Clusters and Biophysical Parameters

Accordingly, we set on to understand the distribution of these viral proteins (NS2B3 complex) at the single-molecule level. We used super-resolution localization microscopy to understand the distribution of these viral proteins on mitochondria that lead to their fragmentation. The effect of NS2B3 on mitochondrial network is evident (see, Fig. 1), which is also independently reported in other studies^32,34. Taking this as the basis, super-resolution studies were carried out to understand the distribution of viral proteins. Figure 3 shows super-resolved images of PAGFP-NS2B3 and mEos-NS2B3 transfected cells along with the respective transfected cell images (yellow box). The studies were performed on fixed cells 48 hrs after transfection. We have used point-based clustering technique for analyzing the clusters⁴⁶. In addition, a density-based method (termed as, DBSCAN) is also used for comparison (see, Supplementary Note 10). One can readily see the appearance of single-molecule clusters indicating the formation of NS2B3 clusters. Corresponding localization precision and FRC analysis is discussed in Supplementary Note 17. The analysis shows a mean localization precision of 20nm, and the resolution is calculated using FRC analysis see, methods section. It may be noted that the localization precisions for PAGFP-NS2B3 and mEos-Ns2B3 are different. This is mainly due to the photophysical properties of the respective fluorescent proteins, including the blinking period and photobleaching^47–49. It is due to the high-resolvability of super-resolution microscopy that we are able to see and analyze distinct disconnected viral clusters. Statistical analysis shows a cluster size (Area) of ≈ 0.05 μm², with a cluster density of ≈ 3500mol/μm² and an average of ≈ 120mol/cluster.

Fig. 3. Formation of NS2B3 clusters and its characteristics.

A, B Super-resolution study on the viral plasmid DNA (PAGFP − NS2B3 and mEos − NS2B3) transfected cells (yellow box) suggests clustering of NS2B3 proteins. This is further confirmed by enlarged section of selected regions (R1 and R2). Alongside, transfected cell image (brightfield) is also shown. The raw data for PAGFP-NS2B3 and mEos-NS2B3 are shown in Supplementary movie 1 and 2. C Biophysical parameters associated with the estimated clusters. D Statistical analysis showing an average cluster area of 0.056 ± 0.006μm² for PAGFP-NS2B3 and 0.049 ± 0.006μm² for mEos - NS2B3 respectively. The corresponding cluster density is 2906 ± 772mol/μm² and 3926 ± 891mol/μm² with an average number of 96 ± 12 and 136 ± 21 molecules per cluster for PAGFP-NS2B3 and mEos - NS2B3 respectively. The estimated cluster parameters in cells are transfected with NS2B3 gene indicate the formation of Dengue NS2B3 protein with single-molecule precision. Other cells that showing NS2B3 cluster are discussed in Supplementary Note 8. Here, the analysis is performed on an entire cell, and in total we have used 15 cells per condition.

To understand the repercussions of NS2B3 clusters on the mitochondrial network, we overlayed or superimposed super-resolved image on the confocal image (Fig. 4A). The regional analysis of biophysical parameters are carried out to determine the effect of cluster formation in close vicinity to the mitochondrial network locally (regions, R1-R4). So, the critical parameters (derived from confocal and super-resolution studies) such as entropy, #fragments/ area, number of clusters, and cluster density were estimated as shown in Fig. 4B and C. The visual inspection and estimated parameters show an impact on the mitochondrial dynamics related to both fission and fusion, when compared to control (non-transfected cells stained with mitotracker orange dye) (see, Fig. 4B,C). Specifically, the number of fragments/area and local entropy (regions, R1-R4) are estimated to be 7.52/μm² and 1.73, respectively for control, whereas, in NS2B3 transfected cells, it ranges from 31 − 33.5μm² and 4.35 − 5.18, respectively. This overall suggests that NS2B3 clustering promotes mitochondrial fragmentation leading to a relatively larger number of fragments, which in turn results in an overall increase in local entropy (see Fig. 4D). Moreover, we found both the constructs to have similar performance, although for long time imaging mEos-NS2B3 showed better performance. We used PAGFP-NS2B3 for Fig. 4 since its spectra doesn’t overlap with mitotracker orange. Overall, these critical biophysical parameters are related to clustering in a cellular system.

Fig. 4. Localization of NS2B3 clusters on the site of mitochondrial fragmentation.

A Overlaying super-resolution data (image) on the confocal data (image) of PAGFP-NS2B3 transfected cells stained with mitotracker orange. The corresponding raw movie is shown in Supplementary movie 3, B, C Regional parameter estimation (R1-R4) of NS2B3 clusters for confocal and SMLM indicating increase in the number of fragments per area and regional entropy with respect to control (represented by red-dotted line showing an average 7.52 and 1.73, respectively). D Graphical representation showing the mean parameters for different regions. Additional data for more cells can be found in Supplementary Note 14. E Mean statistical analysis of single-molecule clusters.

Time-lapse Imaging of NS2B3 Clusters

Visualizing NS2B3’s dynamics (formation/migration) provides direct evidence of clustering in a live cell. To facilitate, time-lapse imaging (both confocal and super-resolution) is carried out at various time points post-transfection. Time-lapse confocal imaging of eGFP-NS2B3 transfected live cells, along with Mito-PamKate labelled mitochondrial network. Figure 5 shows time-lapse images at regular interval, revealing steady increase in NS2B3 clusters and increased fragmentation. We have carried out fragmentation analysis (#fragments/area bar-plots) for an extended 6.5 h time-lapse confocal imaging (see Fig. 5). Specifically note that the number of fragments per unit area increased from 4.619 ± 0.079 to 5.442 ± 0.086. This gives a positive correlation between the increase in number of clusters and number of mitochondrial fragments. Additional images and related movies are incorporated as additional Supplementary Note 15, and Supplementary movie 5, respectively. The control experiments are discussed in Supplementary Note 15 and the time-lapse movie is shown in (Supplementary movie 6). Overall, this gives the imaging-based evidence of the cluster formation. However, note that fragmentation may also be a consequence of other factors too. Extending this finding to time-lapse super-resolution imaging helps us to visualize the overall evolution of clusters, including their numbers, size, and the number of molecules per cluster. The time-lapse super-resolved image, along with the estimated biophysical parameters, are shown in Fig. 6. At each time point, the data (5000 frames) is acquired by the system and processed to reconstruct the super-resolved image (see, Fig. 6A). The corresponding raw data are shown in Supplementary movie 4. One can immediately see the dynamic changes that occur during cluster formation. In fact, we observe re-appropriation of few newly-formed clusters and formation of new ones. Specifically, we show a few enlarged regions (R1 and R2) exhibiting the time development of a couple of clusters (from 1.5h − 3h). Subsequently, the clusters are analyzed using point-based clustering algorithm, and the parameters are estimated (see, Fig. 6B). Super-resolved image and analysis shows that the clusters are formed dynamically over a time period of 3h, showing their evolution. Moreover, a steady increase is noted in the number of clusters (including the small ones) and the number of molecules per cluster, along with a significant increase in the cluster area (from 0.034μm² to 0.068μm²). So, the time-lapse imaging and the associated analysis indicate an overall increase in the size of a few clusters over time. This is incredible, considering that diffraction-limited microscopy techniques cannot generate single-molecule maps and determine parameters that are critical to visualize the dynamics of Dengue NS2B3 protein.

Fig. 5. Time-lapse confocal imaging of NS2B3 clusters in transfected cells: NS2B3 cluster evolution revealed by time lapse confocal imaging.

Time-lapse confocal imaging of NIH3T3 cells transfected with eGFP-NS2B3(green clusters) and Mito-PanKate (magenta mitochondria) plasmids. Here, the time-point 0hrs refers to 36hrs post-transfection. The images show steady increase in NS2B3 clusters over a period of 6.5 hrs. Fragment analysis depicts an increase in the number of fragments per unit area over time. Additional data are presented in Supplementary Note 15 and the cluster evolution along with mitochondria network is shown in Supplementary movie 5.

Fig. 6. Time lapse super-resolution imaging of dynamic NS2B3 clusters in live cell.

A Raw images are collected at regular time interval (from 0 to 3 hr) post 36 hrs of transfection and super-resolved images are reconstructed to visualize dynamically changing NS2B3 clusters. Few clusters are chosen (marked by blue, red and green boxes) and followed for 3 hr to visualize its dynamics. Specifically, region R3 shows depletion of number of localization overtime and essentially vanish to negligible. Alongside transfected cell image is also shown. B Enlarged image of a few NS2B3 clusters during their developmental stages. C The estimated biophysical parameters (# clusters, # molecules/cluster, and cluster area) are evaluated over time indicating steady increase. In addition, conventional (non-SMLM) fluorescence images of NS2B3 aggregates are shown in Supplementary Note 11.

Proposed Mechanism

Based on the study, a schematic diagram is shown that summarizes the role of NS2B3 protein in a cellular system. Specifically this shows clustering in close vicinity to mitochondrial network and it correlates with fragmentation (see, Fig. 7). The study may provide insight to the dynamics of NS2B3 (both at ensemble and single molecule level), post entry in a cellular system that may be harnessed for targeted drug-study.

Fig. 7. Proposed Mechanism.

The schematic diagram summarizing the findings of the study in a cellular system. The experimental evidence suggests formation of NS2B3 clusters in close vicinity to mitochondrial network, and its effect on fragmentation during Dengue (type-2) infection.

Discussion

Understanding Dengue biophysics and pinning down the location of key protein complexes (NS2B3) has yet to be successful. With the availability of single-molecule-based super-resolution imaging, it is now possible to decipher the process at a single-molecule level. Specifically, the key protein complex NS2B3, a protease, is known to act on the mitochondrial membrane. Previous studies have shown NS2B3 using innate mitochondrial transport signal (MTS) to get into mitochondrial network³². The in-vitro study using purified proteins, with reduced levels of matrix-localized GrpEL1, a cochaperone of mitochondrial Hsp70 (mtHsp70), is noticed in NS2B3 expressing dengue infected cells³². However, in-vitro detailed study of the biophysical properties of NS2B3 clustering on mitochondrial membrane leading to mitochondrial fragmentation have not been studied due to the limited resolution of existing microscopes. In this study, we used super-resolution microscopy that can decipher details at a single-molecule level to reveal the presence of NS2B3 protein clusters (accumulation and clustering) leading to mitochondrial fragmentation.

Our study using single-molecule imaging on NS2B3 transfected cells indicates clustering of NS2B3 single-molecules. Specifically, the NS2B3 clustering lead to mitochondrial fragmentation. It may be noted that clustering is a known phenomenon that occurs in a number of viral diseases and shows the region of its biological activity in a cellular system. Specifically, the present study on Dengue virus (type 2) suggest the clustering of NS2B3 protease molecules. As far as other viral infection processes are concerned, clustering as an indicator of viral infection was also reported in Respiratory syncytial virus (RSV) and Influenza Virus^50,51. Moreover, abnormal mitochondrial fragmentation due to Zika viral infection led to neuronal cell death was reported by Yang et al.⁵². Further, transient-transfection and protein expression studies using NS2B3 plasmids demonstrate a relatively large number of fragments, more randomness (large entropy), and shorter fragment length as compared to control (see Fig. 1 and Fig. 2). Super-resolution study and associated single molecule analysis estimated the biophysical parameters (cluster size (Area) of ≈ 0.05μm², cluster density of ≈ 3500mol/μm² and an average of ≈ 120mol. /cluster) that are critical for mitochondrial fragmentation (see, Fig. 3). The localization of NS2B3 clusters on the site of mitochondrial fragmentation marked the direct evidence of NS2B3 protease involvement in mitochondrial fragmentation. These parameters are important for determining the role of the NS2B3 complex during Denv-2 viral infection.

In SMLM, background consideration is important as this lead to low signal-to-noise ratio. The background is largely due to light from the off-focal plane, electronic noise, thermal electrons, intensity fluctuations and autofluorescence from the sample itself. The background contribute significantly and reduce signal-to-noise ratio and must be minimized for quality single molecule imaging^53–56. However, to decrease the background, the experiments were performed in dark lab conditions. Moreover, for detecting single molecule spots, we subtract background noise using active intensity threshold. This avoids false detection.

One of the findings of the present study is the formation of cluster with specific characteristics with respect to number of clusters per unit area, cluster density, fragmentation and entropy (see, Fig. 4). Previous studies have indicated that NS3 complex-induced cleavage of GrpEL1, a matrix protein is a major cause of mitochondrial matrix fragmentation^32,57. Moreover, NS2B3 protease causes depletion of GrpEL1 that leads to dysfunctional mitochondria^58–62. Time lapse confocal imaging of NS2B3 cluster evolution revealed a positive corelation between clustering and mitochondrial fragments. A strong association of NS2B3 protease with mitochondrial membrane is revealed by Yu et al. (2015)³⁴. Further, time-lapse super-resolution imaging is carried out to visualize the dynamics of NS2B3 molecules and the associated clustering process (see Fig. 6). The associated cluster parameters indicated an average increase in the size of a few clusters and the number of molecules per cluster over a period of 3 hr (post-transfection). Although more direct evidence may be necessary to establish the entire mechanism, the basic understanding of NS2B3 clustering can be of the therapeutic potential that may be exploited. Specifically, the potential drugs capable of reversing or disrupt the clustering process are of great value.

The study hypothesizes that the formation of NS2B3 clusters in a cellular system leads to the fragmentation of the mitochondrial network. In this respect, the super-resolution studies have played key role, suggesting the formation of NS2B3 clusters. The technique allows determination of cluster parameters such as cluster area, density, and number of molecules per cluster (see Fig. 3). In addition, the clusters are not restricted to a specific plane but span the whole cell. It is the time-lapse super-resolution imaging of live transfected cells that showed the kinetics of single viral particle (NS2B3), and the formation of these NS2B3 aggregates over time. This eventually leads to stable clusters with a steady increase in the number of clusters over time and the number of NS2B3 molecules stabilizes to about 120, resulting in increased cluster area. This is evident from the time-lapse super-resolution imaging of live cells as shown in Fig. 6. Overall, the super-resolved study suggests the formation of NS2B3 clusters over time after 36 hrs of transfection.

The single molecule based super-resolution imaging has enabled direct visualization of single viral protein molecules (NS2B3), their clustering, and their evolution with time. It is hoped that the proposed study may inspire the use of single-molecule techniques in virology and disease biology.

Methods

Recombinant Plasmids

Recombinant plasmid PAGFP-NS2B3 was generated by subcloning of NS2B3 protease from pcDNA-DENV2-NS2B3, a Gift from Alan Rothman⁶³ at Pst1- BamH1 site of PAGFP C1 plasmid, a gift from Jennifer Lippincott-Schwartz⁶⁴. NS2B3 protease with PstI- BamHI overhangs were generated by PCR using NS2B3-specific oligonucleotide primers, NSPstFP with Pst 1 site as forward primer and NSBamRP with Bam H1 site as reverse primer. Recombinant plasmid mEos_NS2B3 was generated by subcloning of mEos fragment from mEos3.2-TOMM20-N-10, a gift from Michael Davidson⁶⁵ at Nhe I- Pst I site of PAGFP-NS2B3 by removing PAGFP fragment. mEos with NheI-PstI overhangs were generated by PCR using oligonucleotide primers, mEosNheFP with Nhe 1 site as forward primer mEosPstRP with Pst 1 site as reverse primer. Similary the recombinant plasmid eGFP-NS2B3 was generated by subcloning of eGFP from eGFP-N1 with Nhe I-Sac I overhangs by removing dendra 2 from Dendra 2 NS2B3 plasmid (In house plasmid where the backbone is Dendra2-Tubulin plasmid, a gift from Samuel Hess (University of Maine, USA).) The list of the sequence of primers used are given in Table 1. PCR was performed using Phusion™ High-Fidelity DNA Polymerase (Thermo Fisher Scientific, India). All restriction enzymes used were purchased from Thermo Fisher Scientific, India. Subcloning and transformation were performed using the standard protocol (Sambrook and Russel, 2001).

Table 1.

List of primers used for subcloning protein of interest (NS2B3) and the marker proteins (eGFP, PAGFP & mEos3.2)

S. No.	Primer Name	Sequence ($5^{\prime }$...... $3^{\prime }$)
1	NSPstFP	AACTGCAGTCATGGCAAGCTGGCC
2	NSBamHRP	ATGGATCCACCGGTACGCGTAGAATCG
3	mEosNheIFP	CTAGCTAGCCATGAGTGCGATTAAGCC
4	mEosPstIRP	GTTCTGCAGTTCGTCTGGCATTGTCAG
5	GFPNheFP	CTGCTAGCGCTACCGGTCGCCACCATG
6	GFPSacRP	GCTCCTTGTACAGCTCATCCATG

Cell Line and Transfection

The recombinant plasmids,eGFP-NS2B3, PAGFP-NS2B3, and mEos-NS2B3 were used to transfect NIH3T3 cells (mouse embryonic fibroblast cell line) to study the distribution of dengue NS2B3 clusters. The cells used in this study were obtained from our collaborator Dr. Upendra Nongthomba (Biological Sciences, Indian Institute of Science, Bangalore, India). Transfection was performed using 1μg of plasmid DNA with Lipofectamine 3000 (Invitrogen, USA) according to manufactures instructions. The cells were cultured on a coverslip (No.0) (Blue star, India) in a 35mm dish with a density of 10⁵ cells/ml. After 12h incubation cells were transfected with respective plasmids. After 48 h of post-transfection cells were washed twice with 1X PBS, and cells were fixed with 3.7% (W/V) paraformaldehyde for 15 min. It was followed by washing with 1X PBS two times. Cover slip was removed from the 35mm dish and placed on a clean slide with mounting media fluorosave (Thermofisher, USA). The fixed cells were made airproof by applying nail paint on the edges for long time preservation and further investigation. The NIH 3T3 cells without transfection are considered as control. Inorder to visualize mitochondria network, the cells are stained with 175nM Mitotracker orange dye for 1 hr 15min at at 37^oC with 5% CO2 and fixed in the same way mentioned above.

Image acquisition using confocal microscopy and data processing

Fixed samples were imaged in Andor Dragonfly microscope (Bioimaging Facility, Indian Institute of Science, Bangalore, India) using 100X oil immersion objective. The cells were illuminated with a 488nm Laser, and the emission of eGFP-NS2B3, PAGFP-NS2B3 and mEos-NS2B3 was captured in the interval 502–540 nm by using a band pass filter and a 40 μm pin hole. The images obtained from confocal microscopy were skeletonized using image J^66,67. The skeletonized image was then processed using a script (inbuilt entropy function) written in MATLAB. The length of each fragment was measured and then the statistical analysis was performed. The entropy measures randomness of the signal in the image and it is calculated for the sample and control (mitotracker stained cells) using equation (1)

$$Entropy=-\sum p*lo{g}_{2}p$$ 1

where p contains the normalized histogram counts from the image.

For fragmentation analysis, the confocal images were initially Gaussian-blurred with a sigma radius of 1 pixel. After that, the images are filtered using an unsharp mask with a sigma radius of 2 pixels and a mask weight of 0.8. Further, the images are binarized using the maximum entropy method (ImageJ). The resulting binarized structures were subsequently skeletonized, and mitochondrial network parameters were calculated from the skeletonized image.

Image Resolution Analysis using FRC and SQUIRREL

In optical microscopy, FRC is widely used to quantify the image resolution. Specifically, FRC is better suited for single molecule localization microscopy since it strongly depends on the localization uncertainty and density of single fluorescent labels^68,69. The analysis involves dividing a super-resolution image (set of single emitter localizations) into two independent sub-images (statistically independent subsets of emitters), followed by determining the statistical correlation of their Fourier transforms (in the frequency domain). Accordingly, FRC is calculated using the equation (2),

$$FRC=\frac{{\sum }_{r\in r_i}{F}_1\left(r\right).F_2^{*}\left(r\right)}{\sqrt{{\sum }_{r\in r_i}{F}_1^2\left(r\right){\sum }_{r\in r_i}{F}_2^2\left(r\right)}}$$ 2

where, F₁ and F₂ are Fourier transform of two sub-images and r_i is the i^th frequency bin. This method gives the resolution of entire super resolution image.

Along with FRC analysis, the resolution of final super resolution image is also analyzed using an ImageJ pluggin (NanoJ-SQUIRREL)⁷⁰ and this method calculates the region wise FRC map. For the SQUIRREL analysis, the final super resolution image is transformed into two sub-images by separating the molecules equally. The sub-images are stacked and block wise FRC(Fourier Ring Correlation) is performed using NanoJ-SQUIRREL pluggin with 25 block per axis.

Labeling and Transfection Study

Mitotracker orange is a dye that selectively labels mitochondria by reacting with the thiol group of cysteine residues. Then it gets accumulated in the mitochondrial matrix. Mito tracker orange (Invitrogen, USA) was dissolved in DMSO to obtain 1 mM stock solution. The stock solution was diluted in DMEM to get a working concentration of 175 nM, was used for staining. Cells transfected with eGFP-NS2B3 for 48 h and then stained with 175 nM mitotracker orange followed by incubation for 1 h 15 min at 37 °C with 5% CO2. It was followed by fixing and preservation as described above. The fixed samples were imaged in confocal microscopy using two colors. We imaged eGFP-NS2B3 using laser of 488nm wavelength and the fluorescence signal is recorded with the band pass having pass band wavelength of 502–540 nm. The Mito-tracker was excited using a 561 nm wavelength laser and emission signal was collected with the band pass filter(572.5–615.5 nm).

Super-resolution imaging for Single-molecule analysis

Transfected cells (PAGFP-NS2B3 and mEos-NS2B3) were fixed, and super-resolution studies were carried out. We used a 405 nm laser for the activation of the samples and 488 nm and 561 nm lasers were used for PAGFP-NS2B3 and mEos-NS2B3, respectively. The transfected cells were seen by 470–490 nm blue light. The system is a homemade system⁷¹. The details of optical system and data processing methods are discussed in Supplementary Note 9. The fact that NS2B3 form clusters needs further investigation to determine the cluster area, number of molecules (number of localizations) and density of clusters. During single molecule analysis and rendering, the repeated localizations are merged into single localization as it appears for more than one frame. Then, cluster analysis is performed to find the clustered molecules. This information is then used to calculated area of the cluster and number of molecules per cluster. Subsequently, the cluster density is computed by dividing number of molecules in a cluster by the cluster area. The cluster analysis was done using the point-based clustering algorithm⁴⁶ and compared with DBSCAN (see, Supplementary Note 10). For point-clustering method, the centroid information of molecules is used. For a set of molecules to be considered as a cluster, we hace chosen a maximum Euclidean distance between two centroids as 60 nm and and the minimum number of molecules as 50.

aSMLM Super-resolution Microscopy

aSMLM is an adaptive technique that allows the user to change the field of view, thereby facilitating imaging over a large area. The method uses an electrically tunable lens (ETL) in the illumination subsystem. The focal length of the ETL can be changed by applying an electric current, and the change in the focal length of the ETL ensures overfilling/underfilling of the back aperture of the objective lens. The technique is beneficial for the present study, as this allows imaging of a large portion of the cell without the need for image stitching. Moreover, aSMLM allows uniform distribution of light during sample illumination. These two factors are essential for high-quality super-resolution imaging over a large FOV without any intensity bias.

Overlaying confocal and super resolution images

Cells transfected with PAGFP-NS2B3 for 48 h and then stained with 175 nM mitotracker orange followed by incubation for 1 h 15 min at 37 °C with 5% CO2. It was followed by fixing as mentioned above. In order to visualize mitochondrial network, the fixed samples were imaged for Mito-tracker orange using a 561nm wavelength laser in confocal microscopy. To find out the intra-cellular localization of NS2B3 clusters super resolution studies of the same cell was carried out. We used a 405 nm laser for the activation, followed by excitation at 488 nm (for PAGFP fluorescent protein), for which the fluorescence signal is collected in the wavelength range 500–540 nm. The effect of NS2B3 cluster formation on mitochondrial network was analysed by comparing regional entropy, number of fragments, number of clusters and cluster density in a given region of transfected cell.

Time-Lapse Imaging

To understand NS2B3 cluster dynamics, we have carried out time-lapse super-resolution imaging. The transfection was performed using 1 μg of plasmid DNA with Lipofectamine 3000 (Invitrogen, USA). The cells were cultured on a live imaging dish (35mm dish, Mattek, USA) with a density of 30, 000 cells/ml. After 12h incubation cells were transfected with mEos-NS2B3 plasmid. Post 36h of transfection, the cells were imaged every 30 min for a total duration of 3h. We observed that clusters evolve after 36 h of transfection, but exact reason is not well understood. For live cell imaging during time-lapse, the incubation temperature is maintained at 37 °C along with 5% of CO2. The recorded images are then subjected to the reconstruction process. This is followed by cluster analysis and the determination of cluster parameters (number of clusters, number of molecules/cluster and cluster area).

Statistical Analysis

All the statistical analyses were performed using GraphPad Prism 10. Data for statistical analysis were copied directly from Matlab files. Statistical significance between the test group and control were calculated by a nonparametric t-test(Mann-Whitney test). The mitochondrial branch lengths of control (stained with Mitotracker only) and sample(cells transfected with eGFP-NS2B3 and stained with Mitotracker) were considered as same (null hypothesis). From the t-test distribution table, p-value is determined for the calculated t-value. If the obtained p value is less than 0.0001, it leads to the rejection of the null hypothesis and indicates the significant difference between the control and the sample. The statistical the analysis is done on much larger number of cells (i.e., 15 cells per condition).

Transfection Efficiency

The transfection efficiency is calculated by imaging the samples in confocal microscopy with low magnification objective lens (20x). From the recorded images the number of transfected and non-transfected cells are counted and the transfection efficiency is calculated using the following expression,

$${\eta }_T=\frac{N_{trans}}{N_{tot}}\times 100\%$$ 3

where, N_trans and N_tot are the number of transfected and total cells, respectively.

Pixel-based Correlation Analysis

A block size of 10 × 10 pixels is considered for the correlation study, and the Pearson correlation coefficient is determined. The obtained correlation coefficients are mapped as a ‘correlation map’. It is known that the output of the Pearson correlation coefficient varies from -1 to +1, but for visualization purposes, +1 is added to all the pixels in the correlation map, which shifts the correlation values to [0, 2]. In the correlation map, 0 represents anti-correlation (dark regions), and 2 represents perfect correlation (bright regions). It is evident from the dark spots in the correlation map that the presence of clusters depletes mitochondrial network (see, Fig. 1C and Supplementary Note 13).

Author contributions

P.P.M. and J.M.V. conceived the idea. J.M.V., P.J., A.S., P.P.M. carried out the experiments. J.M.V., P.J., A.S., P.P.M. prepared the samples. P.P.M. wrote the paper by taking inputs from all the authors.

Data Availability

The data and code that support the findings of this study are available from the corresponding author upon request.

Code availability

The code that support the findings of this study are available from the corresponding author upon request.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s44298-026-00202-8.