Full assembly of HIV-1 particles requires assistance of the membrane curvature factor IRSp53

Article activity feed

1. 

   Version published to 10.7554/elife.67321 on eLife

   Jun 11, 2021
2. 

   eLife

   May 19, 2021

   Author Response:

   Reviewer #1:

   After infection, new HIV-particles assemble at the host cell plasma membrane in a process that requires the viral protein Gag. Here, Inamdar et al. showed that a component of the host cell, the membrane curvature-inducing protein IRSp53, contributes to efficiently promote the formation of viral particles in synergy with the viral Gag protein.

   In cells depleted of IRSp53, the formation of HIV-1 Gag viral-like particles (VLPs) was compromised. The authors showed in compelling electron micrographs that the formation of VLPs was arrested at about half stage of particle budding. Biochemical data (co-IPs and analysis of VLPs and HIV particle content), super-resolution nanoscopy (single molecule localization microscopy) data, and in vitro biophysics measurements (in GUVs), all seem to indicate a functional …

   Author Response:

   Reviewer #1:

   After infection, new HIV-particles assemble at the host cell plasma membrane in a process that requires the viral protein Gag. Here, Inamdar et al. showed that a component of the host cell, the membrane curvature-inducing protein IRSp53, contributes to efficiently promote the formation of viral particles in synergy with the viral Gag protein.

   In cells depleted of IRSp53, the formation of HIV-1 Gag viral-like particles (VLPs) was compromised. The authors showed in compelling electron micrographs that the formation of VLPs was arrested at about half stage of particle budding. Biochemical data (co-IPs and analysis of VLPs and HIV particle content), super-resolution nanoscopy (single molecule localization microscopy) data, and in vitro biophysics measurements (in GUVs), all seem to indicate a functional connection between Gag and the iBAR-domain containing protein IRSp53. The combination of the different techniques and approaches is a clear strength of this manuscript. However, to my opinion, the interpretation of some of the experimental data is somehow limited by the lack of some appropriate controls (that are lacking for different reasons, as the authors state in some parts of the text). These are:

   1. Specificity of the IRSp53 siRNA. Although the authors showed that the siRNA used can deplete the expression of the protein (both endogenous and ectopic), they did not presented any rescue experiments of the phenotypes (or corroboration with different siRNA oligoes).

   We have tried several different commercial and home-designed siRNA targeting IRSp53 from different companies (providing single siRNA and multiple siRNA mix): we have summarizing all in the Figure R1 (see below). One can see that indeed only 2 siRNA were effective in extinguishing IRSp53 gene: one from Invitrogen on endogenous IRSp53 and ectopic IRSp53-GFP and one from Dharmacon that was only effective on ectopic IRSp53-GFP, as revealed by Western Blot (Fig R1A). Furthermore, the specificity of the siRNA was challenge by testing siRNA IRSp53 on human IRSp53-GFP and on mouse I-BAR-GFP in HEK293T transfected cells and visualized by fluorescence microscopy. Results show in figure R1B that only siIRSp53 is able to extinguished human IRSp53-GFP and not mouse I- BAR-GFP. SiIRTKS and siCtrl are not extinguishing any of these genes. Overall these results confirm the specificity of IRSp53 siRNA-mediated knockdowns.

   

   Figure R1: Specificity of siRNA-mediated knockdowns: (A) Western blots of HEK293T cells lysates probed with anti-IRSp53 antibody (and house-keeping gene GAPDH) showing a series of different siRNA IRSp53 (and siRNA Control, CTRL from Invitrogen, Dharmacon or Sigma) on endogenous and ectopic IRp53 genes in human HEK293T cells and their efficacy in specifically down regulating IRSp53. (B) siRNA IRSp53 from Invitrogen was tested for its specificity in extinguishing human IRSp53-GFP protein expressed in transfected HEK293T cells, but not mouse I-BAR-GFP, and as compare to siRNA control and IRTKS, revealed by fluorescence imaging (GFP).

   To further answer the reviewers’ comments, we also perform one rescue experiment of the phenotype as shown in Figure R2 below. We observed that, upon co-transfection of pGag+pIRSp53- GFP+siRNA IRSp53 (lane 2), about 50% of the ectopic IRSp53-GFP was extinguished (since this construct is not siRNA resistant), leaving 50% of this ectopic protein expressed in the cells. In this context, one can observe that Gag-VLP release is ~50% (lane 2), similar to the condition pGag+siCTRL (lane 3). When we compare this to pGag+siIRSp53 (lane 4) which is reduced by 2-3 fold (data from Figure 1b of the manuscript), we can say that the remaining IRSp53-GFP in the Lane 2 seems to rescue the defect caused by extinction of the endogenous IRSp53. In the condition pGag+pIRSp53- GFP +siCTRL, VLP-Gag release was slightly reduced. This is an atypical rescue experiment since we do not have an IRSp53-GFP that is resistant to the siRNA IRSp53 used in this study (Figure R1B), but it suggests that if IRSp53-GFP is overexpressed in the presence of Gag and the siRNA IRSp53, VLP-Gag release is at a normal 50% level in contrast to the absence of IRSp53-GFP (compare lane 2 with lane 4). Unfortunately, due to limited time and by the siRNA IRSp53 out of stock, and the delay in supply, we could only provide one experiment. We thus decided to show it for answering the reviewers but not as part of a figure in the final manuscript.

   

   Figure R2: Rescue of siRNA IRSp53 knock-down with overexpression of IRSp53-GFP: 293T cell were transfected with pGag, pIRSp53 and siRNA control (siCTRL, lane 1) or siRNA IRSp53 (lane 2); cell lysat and VLP wre loaded on SDS-PAGE gels and immunoblots were revealed with anti-GFP (for IRSp53-GFP) and anti-CAp24 (for HIV-1 Gag). One graph on the left shows the percentage of IRSp53-GFP expression upon siRNA IRSp53 cell treatment (lane 2) as compare to the siRNA CTRL (lane 1). The graph on the right shows the resulting gel quantification for the % of Gag-VLP release upon siRNA IRSp53 cell treatment (lane 2) as compare to the siRNA CTRL (lane 1) in the presence of IRSp53-GFP over-expression, or without (lane 3 and 4, as in Figure 1b). N=1 rescue experiment.

   1. In the co-IPs (IRSp53 IP + Gag co-IP) there is no assessment of the IRSp53 IP efficiency in the different conditions. The authors argued that IgG signal masking precluded them from doing that.

   See the new figure 2. In the new figure 2b, we have assess the IP/co-IP of IRSp53-GFP/Gag efficiency by adding a complete experiment showing that an anti-GFP is able to pull down IRSp53- GFP very efficiently (lanes 2 and 3) and co-IP Gag efficiently (lane 3) accordingly to the input and remaining flowthrough. Using IRSp53-GFP and an anti-GFP antibody, we could bypass the IgG signal masking the endogenous IRSp53 with the IRSp53 antibody’s IP.

   1. The authors observed an increase in the membrane-bound pool of IRSp53 when Gag is present (Fig. 2c). It is not clear whether this is specific for IRSp53 or other IBAR proteins can also be more membrane-bound as a result of Gag expression.

   See the new figure 2. In the new figure 2d, we have re-loaded all the gel fractions on new SDS- PAGE gels and probed the corresponding immunoblots for Gag, IRSp53, IRTKS, Tsg101 and the cellular markers, Lamp2 (for membrane fractions) and ribosomal S6 protein (for cytosolic fractions). One can see that after quantification of the IRSp53 versus IRTKS bands in the HEK293T cell control and in the Gag expressing cells, only IRSp53 is increasing at the cell membranes upon Gag expression and not IRTKS.

   Reviewer #3:

   Inamdar et al. used biochemical and microscopy assays to investigate the role of I-BAR domain host proteins on HIV-1 assembly and release from HEK 293T and Jurkat cells. They show that siRNA knockdown of IRSp53, but not a similar I-BAR domain protein IRTKS, inhibits HIV-1 particle release from 293T cells after transfection of the HIV-1 provirus or HIV-1 Gag in cells. The authors then show that HIV-1 Gag associates with IRSp53 in the host cell membrane and cytoplasm, using biochemical assays and super resolution microscopy. In addition, IRSp53 is incorporated into HIV-1 particles along with other previously identified host proteins. Then using in vitro-derived membrane vesicles ("giant unilamellar vesicles" or GUVs), the authors indicate that HIV-1 Gag can associate with IRSp53, particularly on highly curved structures.

   The conclusions are largely supported data, with the virology and biochemical results being particularly strong, but the mechanistic studies in GUVs appear somewhat preliminary and are not entirely clear. The GUV experiments would benefit from better quantification of measurements and manipulation to simulate actual cellular scenarios. In addition, while it is appreciated that the HEK 293T cell line is convenient for biochemical and imaging studies, they are not biologically relevant HIV-1 target cells. While the authors present examples of reproducibility of their results in a CD4+ T cell line, these data are buried in the supplemental figures, whilst it would have been better to highlight them and perhaps include primary CD4+ T cells.

   1. Immortalized cell lines do not always recapitulate primary cells. It is unclear what the role of IRSp53 is in the membrane curvature of CD4+ T cells and whether expression levels and localization are consistent with Jurkat T cells.

   Please consider the general responses to the Editors, which is:

   We have published that IRSp53 (using siRNA) is involved in HIV-1 particle release on primary T cells (PBMC derived T cells) in Thomas et al, JVI 2015, so high probability is that it would be the same in different cell type, transfected HEK293T cells, transfected or infected Jurkat T cells and infected primary T cells. But we have not done the extensive super-resolution microscopy on infected primary T cells because this would require time overconsuming study. We are currently proceeding in setting up condition with an infectious HIV-1 virus carrying mEOS2 photoactivable protein for being able to infect primary T cells and go on for further research using infectious relevant system and super- resolution microscopy, but it is not ready for this current manuscript as it would require months of extra work and experiments.

   Although, we agree with the reviewer #3 that the localization of Gag in Jurkat T cells and in primary CD4 Tc cells is different at the cell level (in primary T cells HIV-1 Gag is more polarized at uropods, as referred in the literature – see for an example Bedi et al/Ono’s Lab), but at the nanoscopic level of the budding sites, chances are that it would be similar but it need to be checked in future studies.

   1. Description of some of the microscopy measurements could be improved. In lines 204-206 of the text and Figure S5, it is unclear how the localization of precision was determined to be approximately 16 nm for PALM-STORM.

   These lines have been changed in the main text as they were not mandatory to understand how we determine the size of the VLP clusters. However, we have now detailed in figure S5 how we measure localisation precision.

   The following text has been added to the legend of the figS5:

   “Distribution of localisation precisions for PALM (in green) or STORM (in red) as given by Thunderstorm analysis in Fiji : Localisation precision distribution exhibit maxima at 16 nm and a mean±sd value of 20±5 nm for PALM, and a maxima of 26 nm, corresponding to a mean±sd value of 27±10 nm for STORM. The localization precision is obtained by eq 17 of (Thompson et al., 2002).”

   As well as the reference of the original paper (Thompson et al. 2002, Biophysical Journal).

   In Figure 4b, it is understood from the text (lines 252-256) that the red bars denote the Mander's coefficient for colocalization of the GFP-tagged proteins with Gag-mCherry (presumably the average of multiple experiments with standard deviations or errors of the mean, although this is not stated in the figure legend), it is unclear what the green bars are showing.

   Yes, the red bars denote the Mander's coefficient for colocalization of the Gag-mCherry with the GFP-proteins, and the green bar denote for colocalization of the GFP-tagged proteins with Gag- mCherry, showing for more than 300 green and red vesicles, thant indeed all the Gag-VLP are green in the case of IRSp53-GFP (red bar) but that not all the GFP-IRSp53-GFP “green” vesicles are (+) for Gag: this indicates that vesicles produced by transfected HEK cells produced GAG/IRSp53 VLP but also IRSp53-GFP vesicles. Thanks to the reviewer to point this out. We added the explanation in the main text (page 12, lanes 272-282) and in the figure legend of Figure 4b.

   Also, the histograms for IRSp53 and IRTKS colocalized with Gag look similar in Figure S10, suggesting that they are not different in Jurkat cells, but this is not addressed.

   Yes. We have now addressed this particular point in the global response to the reviewers. Indeed, the figure 3 and 4 were remodelled into new figure 3 showing, in the same figure, HEK and Jurkat cells results and in figure 4 the simulations results. Overall, the PALM/STORM microscopy analysis results on Gag/IRSp53 colocalization are very similar in both cell types.

   1. GUVs are first referenced on page 7 after description of Figure 2, the significance of which is confusing to the reader. However, the actual experimental data are described on pages 12-13 and Figures 5 and S11. A better description of these structures would be warranted for an audience that is unfamiliar with them. In addition, the biologic concentrations of I-BAR proteins at cell membranes are not provided and it is unclear what conditions used in Figures 5 and S11 represent a "normal CD4+ T cell" situation. It appears that the advantage of this in vitro system is that different factors can be provided or removed to simulate different cellular scenarios. For example, relatively low IRSp53 concentrations may simulate siRNA knockdown experiments in Figure 1, which could recapitulate those results that less viral particles are released from the membrane. In addition, the authors state that HIV-1 Gag preferentially colocalizes with IRSp53 as the tips of the GUV tubular structures (Figure 5b,c), but this is not actually shown or quantified. Similar quantification as shown in Figure 1e could be performed to strengthen this argument.

   We thank the review for pointing this out. We now described all the GUV result in section 5.

   Considering the biological concentrations of I-BAR proteins in cells, to the best of our knowledge, there is no measurement of it. We thus could not relate concentrations used in the GUV experiments with those in cells.

   We could not perform quantification as in Figure 1e because the majority of the tubes in GUVs were moving too rapidly, preventing us from acquiring images with higher spatial resolution (see Fig. S11, and Movie 2 and 3). However, we would like to point out that the Gag signals appeared dotty inside GUVs (see Fig. S11, and Movie 2 and 3), which is very different from the signals of I-BAR that are clearly along the tubes (see Fig. S10c). Moreover, for tubes that were not moving too fast, we found that for all the tubes (17 tubes), Gag signals are exclusively located at the tips of the tubes (see new Fig. 6d). Also, the sorting maps shown in Fig. 6c and Fig. S10 d indicate the relative accumulations of Gag at the tips of the tubes. To make it clearer that the Gag signals were located at the tips of the tubes, in the current manuscript, we have added the new Fig. S11, Movie 1, 2 and 3, and included zoom-in images in Fig. 6b, 6c and a new Fig. 6d. Also, we have included the quantitation results (17 tubes) in the manuscript.

   Read the original source
3. 

   eLife

   May 19, 2021

   Reviewer #3 (Public Review):

   Inamdar et al. used biochemical and microscopy assays to investigate the role of I-BAR domain host proteins on HIV-1 assembly and release from HEK 293T and Jurkat cells. They show that siRNA knockdown of IRSp53, but not a similar I-BAR domain protein IRTKS, inhibits HIV-1 particle release from 293T cells after transfection of the HIV-1 provirus or HIV-1 Gag in cells. The authors then show that HIV-1 Gag associates with IRSp53 in the host cell membrane and cytoplasm, using biochemical assays and super resolution microscopy. In addition, IRSp53 is incorporated into HIV-1 particles along with other previously identified host proteins. Then using in vitro-derived membrane vesicles ("giant unilamellar vesicles" or GUVs), the authors indicate that HIV-1 Gag can associate with IRSp53, particularly on highly curved …

   Reviewer #3 (Public Review):

   Inamdar et al. used biochemical and microscopy assays to investigate the role of I-BAR domain host proteins on HIV-1 assembly and release from HEK 293T and Jurkat cells. They show that siRNA knockdown of IRSp53, but not a similar I-BAR domain protein IRTKS, inhibits HIV-1 particle release from 293T cells after transfection of the HIV-1 provirus or HIV-1 Gag in cells. The authors then show that HIV-1 Gag associates with IRSp53 in the host cell membrane and cytoplasm, using biochemical assays and super resolution microscopy. In addition, IRSp53 is incorporated into HIV-1 particles along with other previously identified host proteins. Then using in vitro-derived membrane vesicles ("giant unilamellar vesicles" or GUVs), the authors indicate that HIV-1 Gag can associate with IRSp53, particularly on highly curved structures.

   The conclusions are largely supported data, with the virology and biochemical results being particularly strong, but the mechanistic studies in GUVs appear somewhat preliminary and are not entirely clear. The GUV experiments would benefit from better quantification of measurements and manipulation to simulate actual cellular scenarios. In addition, while it is appreciated that the HEK 293T cell line is convenient for biochemical and imaging studies, they are not biologically relevant HIV-1 target cells. While the authors present examples of reproducibility of their results in a CD4+ T cell line, these data are buried in the supplemental figures, whilst it would have been better to highlight them and perhaps include primary CD4+ T cells.

   1. Immortalized cell lines do not always recapitulate primary cells. It is unclear what the role of IRSp53 is in the membrane curvature of CD4+ T cells and whether expression levels and localization are consistent with Jurkat T cells.

   2. Description of some of the microscopy measurements could be improved. In lines 204-206 of the text and Figure S5, it is unclear how the localization of precision was determined to be approximately 16 nm for PALM-STORM. In Figure 4b, it is understood from the text (lines 252-256) that the red bars denote the Mander's coefficient for colocalization of the GFP-tagged proteins with Gag-mCherry (presumably the average of multiple experiments with standard deviations or errors of the mean, although this is not stated in the figure legend), it is unclear what the green bars are showing. Also, the histograms for IRSp53 and IRTKS colocalized with Gag look similar in Figure S10, suggesting that they are not different in Jurkat cells, but this is not addressed.

   3. GUVs are first referenced on page 7 after description of Figure 2, the significance of which is confusing to the reader. However, the actual experimental data are described on pages 12-13 and Figures 5 and S11. A better description of these structures would be warranted for an audience that is unfamiliar with them. In addition, the biologic concentrations of I-BAR proteins at cell membranes are not provided and it is unclear what conditions used in Figures 5 and S11 represent a "normal CD4+ T cell" situation. It appears that the advantage of this in vitro system is that different factors can be provided or removed to simulate different cellular scenarios. For example, relatively low IRSp53 concentrations may simulate siRNA knockdown experiments in Figure 1, which could recapitulate those results that less viral particles are released from the membrane. In addition, the authors state that HIV-1 Gag preferentially colocalizes with IRSp53 as the tips of the GUV tubular structures (Figure 5b,c), but this is not actually shown or quantified. Similar quantification as shown in Figure 1e could be performed to strengthen this argument.

   Read the original source
4. 

   eLife

   May 19, 2021

   Reviewer #2 (Public Review):

   Inamdar and colleagues present a convincing manuscript identifying the unique role of IRSp53 in the successful assembly of HIV-1 particles. The study provides insight into the molecular mechanisms underlying the membrane curvature generation associated with virion budding from infected cells. Notably, the authors postulate a model in which the HIV-1 machinery "hijacks" host functions to generate fully-assembled viral particles by recruiting a central virion-assembly factor, HIV-1 Gag, to the luminal extremities of nascent extracellular vesicles generated by endogenous IRSp53. This is achieved through interactions between the two proteins, resulting in supramolecular complexes containing host and pathogen factors.

   A significant positive aspect of this study is the implementation of an experimental approach …

   Reviewer #2 (Public Review):

   Inamdar and colleagues present a convincing manuscript identifying the unique role of IRSp53 in the successful assembly of HIV-1 particles. The study provides insight into the molecular mechanisms underlying the membrane curvature generation associated with virion budding from infected cells. Notably, the authors postulate a model in which the HIV-1 machinery "hijacks" host functions to generate fully-assembled viral particles by recruiting a central virion-assembly factor, HIV-1 Gag, to the luminal extremities of nascent extracellular vesicles generated by endogenous IRSp53. This is achieved through interactions between the two proteins, resulting in supramolecular complexes containing host and pathogen factors.

   A significant positive aspect of this study is the implementation of an experimental approach that encompasses complementary techniques, namely biochemistry, super-resolution microscopy, and advanced computational analysis and modelling. This is particularly relevant because several critical studies in the field of HIV-1 often rely on either one or the other set of methods and consequently lack the depth and cross-validation power achieved here. Also, the authors take advantage of experimental models that fall into opposite sides of the natural-artificial spectrum and use them adequately to test hypotheses and make conclusions.

   Read the original source
5. 

   eLife

   May 19, 2021

   Reviewer #1 (Public Review):

   After infection, new HIV-particles assemble at the host cell plasma membrane in a process that requires the viral protein Gag. Here, Inamdar et al. showed that a component of the host cell, the membrane curvature-inducing protein IRSp53, contributes to efficiently promote the formation of viral particles in synergy with the viral Gag protein.

   In cells depleted of IRSp53, the formation of HIV-1 Gag viral-like particles (VLPs) was compromised. The authors showed in compelling electron micrographs that the formation of VLPs was arrested at about half stage of particle budding. Biochemical data (co-IPs and analysis of VLPs and HIV particle content), super-resolution nanoscopy (single molecule localization microscopy) data, and in vitro biophysics measurements (in GUVs), all seem to indicate a functional …

   Reviewer #1 (Public Review):

   After infection, new HIV-particles assemble at the host cell plasma membrane in a process that requires the viral protein Gag. Here, Inamdar et al. showed that a component of the host cell, the membrane curvature-inducing protein IRSp53, contributes to efficiently promote the formation of viral particles in synergy with the viral Gag protein.

   In cells depleted of IRSp53, the formation of HIV-1 Gag viral-like particles (VLPs) was compromised. The authors showed in compelling electron micrographs that the formation of VLPs was arrested at about half stage of particle budding. Biochemical data (co-IPs and analysis of VLPs and HIV particle content), super-resolution nanoscopy (single molecule localization microscopy) data, and in vitro biophysics measurements (in GUVs), all seem to indicate a functional connection between Gag and the iBAR-domain containing protein IRSp53. The combination of the different techniques and approaches is a clear strength of this manuscript. However, to my opinion, the interpretation of some of the experimental data is somehow limited by the lack of some appropriate controls (that are lacking for different reasons, as the authors state in some parts of the text). These are:

   1. Specificity of the IRSp53 siRNA. Although the authors showed that the siRNA used can deplete the expression of the protein (both endogenous and ectopic), they did not presented any rescue experiments of the phenotypes (or corroboration with different siRNA oligoes).

   2. In the co-IPs (IRSp53 IP + Gag co-IP) there is no assessment of the IRSp53 IP efficiency in the different conditions. The authors argued that IgG signal masking precluded them from doing that.

   3. The authors observed an increase in the membrane-bound pool of IRSp53 when Gag is present (Fig. 2c). It is not clear whether this is specific for IRSp53 or other IBAR proteins can also be more membrane-bound as a result of Gag expression.

   Read the original source
6. 

   eLife

   May 19, 2021

   Evaluation Summary:

   This manuscript combines cell biology, biochemistry, and quantitative biophysics to understand a new host cell factor, the human I-BAR domain protein IRSp53, promotes HIV type 1 (HIV-1) assembly and release. Since this new factor is a protein involved in the generation and sensing of negative membrane curvature, this manuscript will be of interest not only for retrovirologists and virologists in general but also for membrane biologists and biophysicists.

   (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #1, Reviewer #2 and Reviewer #3 agreed to share their names with the authors.)

   Read the original source
7. 

   Version published to 10.1101/2021.02.10.430663v1 on bioRxiv

   Feb 10, 2021