Enveloped viruses distinct from HBV induce dissemination of hepatitis D virus in vivo

Hepatitis D virus (HDV) doesn’t encode envelope proteins for packaging of its ribonucleoprotein (RNP) and typically relies on the surface glycoproteins (GPs) from hepatitis B virus (HBV) for virion assembly, envelopment and cellular transmission. HDV RNA genome can efficiently replicate in different tissues and species, raising the possibility that it evolved, and/or is still able to transmit, independently of HBV. Here we show that alternative, HBV-unrelated viruses can act as helper viruses for HDV. In vitro, envelope GPs from several virus genera, including vesiculovirus, flavivirus and hepacivirus, can package HDV RNPs, allowing efficient egress of HDV particles in the extracellular milieu of co-infected cells and subsequent entry into cells expressing the relevant receptors. Furthermore, HCV can propagate HDV infection in the liver of co-infected humanized mice for several months. Further work is necessary to evaluate whether HDV is currently transmitted by HBV-unrelated viruses in humans.

Hepatitis delta virus (HDV) has long been known to co-infect people in a manner dependent on the hepatitis B virus and, indeed, was discovered based on its association with severe hepatitis in HBV patients. This dependence is due to the need for the hepadnavirus envelope to release infectious particles containing the HDV RNA and protein enveloped by the hepadnavirus surface antigen. Perez-Vargas et al describe results indicating that HDV might be able to use the envelope proteins of other viruses to produce infectious particles. The observations involve combinations of cell culture based experiments in which infectious HDV was found to be released from cells co-expressing the envelope of other viruses besides HBV, as well as experiments using mice with "humanized" livers in which HDV spread occurred via HCV rather than HBV, although at much lower levels. The authors findings not only have the potential to dramatically expand our understanding of the egress of HDV RNA-protein complexes from cells but also imply that HDV might affect the pathogenesis of numerous other viruses, as it is known to do for Hepatitis B virus. Thus, the potential impact of this study is quite large. However, given this potential impact, it is imperative that several weaknesses in the cell-culture based experiments several be addressed to convincingly demonstrate that the authors' conclusions are correct. These weaknesses (addressed specifically below), in addition to inconsistencies in some of the data and the use of somewhat unorthodox cell culture conditions undermine confidence in the robustness of the experimental results.
Major points 1. The conclusion that infectious HDV particles are secreted from cells expressing VSV or HCV glycoproteins is indeed suggested by the results in Figs. 1 and 2; however, there are concerns. The biggest is that the purported secreted viral particles are not characterized sufficiently and the characterization that is provided -immunoprecipitation -raises questions about their identity. The immunoprecipitation of the particles is very inefficient -less than 10%. This inefficiency is masked by the log representation of the graph in Fig. 1B. There is no need for a log representation in this figure; it should be changed. It seems possible that the observed release of what are presumed to be enveloped particles might be just HDV RNPs. Further analysis of the particles, in particular determining whether the RNA is genome, antigenome or some combination of both, would help to conclude one way or the other. In addition, in Fig. 1B, the authors suggest a very unusual HDAg composition for the HDV particles released from cells co-expressing different virus glycoproteins (even for regular "HDV"). Numerous studies of HDV particles obtained from cells, infected animals and patients consistently show similar amounts of S-HDAg and L-HDAg. The authors, from labeling of the immunoblot, propose that the particles consist mostly of L-HDAg, with very little S-HDAg present (except following immunoprecipitation with anti-VSV G antibody). There is no discussion of this highly unusual HDAg composition. On the other hand, without showing specific L-HDAg and S-HDAg controls on the immunoblot, it is possible the electrophoresis is not resolving the two species (that is, the labeling is incorrect) and that it is not possible to determine the relative amounts of S-HDAg and L-HDAg from the blot shown. 2. A second major concern is that the authors have not conclusively shown that the particles they have obtained are actually infectious. Several additional experiments are necessary to prove that the authors are observing actual infections rather than, for example, adherence of material to cells. A time course showing accumulation of HDV RNA and protein over several days following incubation would be more convincing than a single time point on day 7. Also more convincing would be demonstration of antigenomic RNA accumulation (assuming that the released RNA is genome) and an increase with time in the amount of L-HDAg, which only occurs during replication.
3. The authors included 2% DMSO in the cell cultures used to produce HDV particles and for infection experiments. The stated purpose was to retard cell growth. However, this treatment is unnecessary (2% DMSO is not typically included in experiments used to produce HDV nor to analyze infection in Huh 106 cells) and its use here raises questions about the generality of the results. Perhaps 2% DMSO is somewhat toxic (it does retard cell growth) and leads to the release of HDV from cells stressed by expression of certain viral glycoproteins. At least some of the production and infection experiments should be repeated in the absence of DMSO.
Additional points 1. There was considerable variability between some similar experiments, or unexplained inconsistencies. In Fig. 3, for example, the levels of HDV RNA detected in mock infected cells varied by almost 10-fold between panel A and panel B. While the RNA analyses suggest that HDV obtained from cells expressing VSV G replicates to 10-fold higher levels than HDV with an HBV envelope (Fig 2A), the immunofluorescence analysis shows approximately 5 times more cells positive for HDV with the HBV envelope ( Fig. 2C). 2. Page 3, second paragraph, second sentence. The statement that HDV does not meet the criteria for the definition of a virus is overstated. HDV is a satellite virus, with HBV as the only known helper. 3. Figure 5. The legend describes black and gray bars, there are only solid black and hashed bars. 4. Supplemental Fig. 1; Group 9. According to the table in Fig. 6, only 4 animals were in this group, yet there are about 9 lines on the graph. There is no indication of what the colors represent in this graph nor on any of the others in this figure.
Reviewer #2: Remarks to the Author: Perez-Vargas et al. report that hepatitis D virus (HDV), which was thought to be a cognate satellite virus of hepatitis B Virus (HBV), may be enveloped by the surface glycoproteins (GPs) from a variety of viruses for its assembly and the release. The authors also found that the production and infection of HDV is not restricted to liver cells, but is restricted by the GPs enveloping HDV and the corresponding receptors expressed on cells. The authors performed various cell culture and animal experiments to support their observations. Co-expression of HDV RNPs and individual viral GP (HBV, VSV, HCV, etc) in hepatoma cell lines led to release of infectious HDV particles. The infection of different GP-enveloped HDV could be blocked by antibodies to the corresponding viral envelop proteins. Notably, in both HBV or HCV infected mice HDV viremia was detected, confirming the observations in a physiologically relevant experimental system. Based on these lines of evidence, the authors propose that HDV may have an origin independent of HBV, and could potentially be propagated by viruses other than HBV. The authors observation is surprising, given the scarcity of literature hinting HDV infection in the absence of an HBV infection. The authors have carved a unique angle to investigate the HDV propagation and provided both cell culture and animal evidences to support their hypothesis, which if proven true will enrich our understanding of HDV origin and the interplay of different viruses. However, several issues need to be addressed to further validate their hypothesis.
RNA was mostly DNA-directed rather than RNA-directed, and since this does not recapitulate the authentic HDV replication, an improved construct was made to have slightly larger than one unit length to produce RNA-derived transcripts (Lazinski & Taylor, 1994, J. Virol). The data produced in Figure 1 and 4 may not hold if the production of HDV RNPs is formed in way that deviates from the authentic mechanism. Two methods could be used: 1) NTCP expressing cells/cell lines expressing various viral GPs is infected by recombinant HBV enveloped HDV virus to test if infectious HDV particles is secreted, and 2) use the improved constructs to validate the findings.
2. In figure 1a, the release of virus through cell death should be ruled out.
Minor issues: 1. Figure 1b, the VSV-∆p 41A1 has a different pattern, e.g, much more s-HDAg and a 55 kDa band. A discussion of the difference, cause, and implication would be helpful.
2. Figure 2c should include light-field or nuclei staining.
3. Figure 5, it is not clear what is the HDV-RNA expressing cell, what construct is used to drive HDV-RNA expression? 4. Is there any explanation why the HDV infection is not reported to be dependent on HCV, if HCV propagates HDV indeed rather efficiently? 5. At numerous occasions the authors refer to "data not shown". This reviewer regards it as important to show all data relevant to the study in the manuscript 6. Page 4: "While HDV was expressed…" rephrase as RNA is not expressed but transcribed or in this case replicated. Proteins are expressed. 7. Page 8, line: please correct/spell out the mutant alleles in the FRG mice: fumarylacetoacetate hydrolase (fah-/-), recombinase activating gene 2 (rag2-/-), interleukin 2 receptor gamma chain (IL2RgNULL) 8. Page 12: subheading, delta particle NOT particles Reviewer #3: Remarks to the Author: The origin of HDV attracts many interesting speculations. Current study proposed it coming from cellular circular RNAs that captured by HBV envelopes, so it is possible other viruses also enabling the packaging of naked HDV RNPs. They tested several viruses, such as HCV, dengue, HIV, et al, and showed HCV, denge and WNV, among others, coudl package the HDV RNPs into their envelope proteins and pass the HDV into next round of infections in cell cultures or in human hepatocyte chimera mice.
Though the results appeared to be interesting, their validation of pseudo-typed HDV and its infections is incomplete and many basic HDV RNA and proteins analysis are missing.
The comments are for authors' reference. 1. The claimed of packaging of HDV RNP by HBV Surface proteins, or by VSV or HCV GPs are interesting by in vitro co-transfection. This finding is supported by their immune-precipitation of these pseudo-typed HDVs, using anti-HBs or anti-VSV or anti-HCV GP. However, the validation of these packaged HDV falls far behind. To confirm the packaged HDV RNP, the authors only showed an ambiguous detection of so-called HDV large HDAg, but not HDV RNA. It is essentially to include a HDV virus packaged by HBV surface protein as a positive control. Such virion has to contain the HDV small delta antigen, and genomic HDV RNA, other than the so-called large HDAg. A Northern blot to confirm an intact, full-sized HDV RNA is required. The RT-PCR quantitation cannot distinguish the viral genemic vs. antigenomic HDV RNAs, either about the size. (In fact, the large HDAg detected in western blot appeared to be suspicious. Other HDAg-specific antibody is required, as it is difficult to understand why no small HDAg is co-packaged). 2. The packaging of HDV RNP by HBsAg required specific isoprenylation of large HDAg. Do the rescue of HDV RNPs require the same modification or not ? This can be easily studied by using isoprenylation inhibitor currently available. 3. The authors tried to band the VSV or HCV GP-packaged HDV virions by CsCL gradient analysis. They succeeded in identifying the putative pseudo-typed HDV in unique density fractions. Again, their only data based upon RT-PCR assay for HDV RNA. Northern and western blots to show HDV genomic RNA and both large and small delta antigens are essential. Finally, as the HDV virions are so abundant, it is necessary to do a simple EM study for these fractions to visualize the size, distribution of these pseudo-type HDV particles. 4. In their co-infection experiments, though HCV or other viruses appeared able to rescue the intracellular HDV RNPs and resulted in efficient next round infections, the data are not comprehensive. The authors relied only HDV RNA quantification by RT-PCR, however, they failed to provide either northern blot or western blot to show the simultaneous presence of HDV RNA or delta antigens. These are easily to show, as the HDV RNA titiers are so high by their data. Besides, it is important to document the co-presence of HDAg and HCV antigen or dengue virus antigen in the same human hepatocytes from the chimera mice. Without these collaborating data, the HDV RNA RT-PCR seems shaky. It should be pointed that currently there is no approved HDV RNA assays, and many in-house assays suffer from varying or inconsistent performance. 5. Finally, the HCV or dengue virus infections in humanized chimera mice took a lot of effort and showed intriguing results. Other than insufficient virological data as mentioned in point 3, the authors may need to study the natural HCV/HDV coinfection in human intravenous drug abusers who frequently co-infected by HBV/HDV/HCV. Do these patients carry HDV RNA within HCV envelope ? We greatly appreciated the Editor and Reviewers' helpful and constructive comments, which we have all taken into account to improve our manuscript by performing additional experiments. Overall, we believe that we have succeeded to provide a more detailed description of these novel HDV particles and to clarify most issues raised by all Reviewers in this revised version of our manuscript. We have revised the manuscript accordingly (see manuscript copy with changes underlined) and we provide a point-by-point response to these comments below (in blue).

Dr. François-Loïc Cosset
The additional results, as per Reviewers' requests, are: -Northern blot analysis of HDV particles produced with unconventional GPs ( Figure 1C) and of cells infected with these particles ( Figure 3D). -Determination by strand-specific RT-PCR of HDV RNA unit size in HDV particles produced with unconventional GPs ( Figure 1B, Figure 1H), with live viruses (Figure 6), and in sera of coinfected animals (Supplemental Figure 7). -Western blot analysis of HDV particles produced with HBV, VSV and HCV glycoproteins ( Figure  1D) and of cells infected with these particles ( Figure 3E). -Electron microscopy analysis of HDV particles produced with HBV, VSV and HCV glycoproteins ( Figure  immuno-fluorescence assays ( Figure 5D and Supplemental Figure 5). -Demonstration that HDV particles can be formed with non-HBV glycoproteins both via transfection of pSVLD3 plasmid and via infection with "helper-free" HDV particles in GPexpressing cells (Supplemental Figure 6).
The other important changes in our revised manuscript are: -Statistical analysis of the data.
-Assessment of production of HDV particles by HDV/DENV-co-infected mosquito cells (Supplemental Figure 6). -Demonstration that identical production of HDV particles can be achieved from cells cultured in media containing, or not, 2% DMSO (Supplemental Figure 2). -Results of co-infection by HDV and HCV in a second cohort of human liver mice (N=24; Supplemental Figure 8).
-Reorganization of the manuscript text, figures and supplementary figures to address, on the whole, all comments of the Reviewers.
We thank you very much for your interest and time in considering our revised manuscript for publication in Nature Communications.

Reviewer #1 (Remarks to the Author):
Hepatitis delta virus (HDV) has long been known to co-infect people in a manner dependent on the hepatitis B virus and, indeed, was discovered based on its association with severe hepatitis in HBV patients. This dependence is due to the need for the hepadnavirus envelope to release infectious particles containing the HDV RNA and protein enveloped by the hepadnavirus surface antigen. Perez-Vargas et al describe results indicating that HDV might be able to use the envelope proteins of other viruses to produce infectious particles. The observations involve combinations of cell culture based experiments in which infectious HDV was found to be released from cells co-expressing the envelope of other viruses besides HBV, as well as experiments using mice with "humanized" livers in which HDV spread occurred via HCV rather than HBV, although at much lower levels. The authors findings not only have the potential to dramatically expand our understanding of the egress of HDV RNA-protein complexes from cells but also imply that HDV might affect the pathogenesis of numerous other viruses, as it is known to do for Hepatitis B virus. Thus, the potential impact of this study is quite large. However, given this potential impact, it is imperative that several weaknesses in the cell-culture based experiments several be addressed to convincingly demonstrate that the authors' conclusions are correct. These weaknesses (addressed specifically below), in addition to inconsistencies in some of the data and the use of somewhat unorthodox cell culture conditions undermine confidence in the robustness of the experimental results.

Major points
1. The conclusion that infectious HDV particles are secreted from cells expressing VSV or HCV glycoproteins is indeed suggested by the results in Figs. 1 and 2; however, there are concerns. The biggest is that the purported secreted viral particles are not characterized sufficiently and the characterization that is provided -immunoprecipitation -raises questions about their identity. The immunoprecipitation of the particles is very inefficient -less than 10%. This inefficiency is masked by the log representation of the graph in Fig. 1B. There is no need for a log representation in this figure; it should be changed. It seems possible that the observed release of what are presumed to be enveloped particles might be just HDV RNPs. Further analysis of the particles, in particular determining whether the RNA is genome, antigenome or some combination of both, would help to conclude one way or the other. In addition, in Fig. 1B, the authors suggest a very unusual HDAg composition for the HDV particles released from cells co-expressing different virus glycoproteins (even for regular "HDV"). Numerous studies of HDV particles obtained from cells, infected animals and patients consistently show similar amounts of S-HDAg and L-HDAg. The authors, from labeling of the immunoblot, propose that the particles consist mostly of L-HDAg, with very little S-HDAg present (except following immunoprecipitation with anti-VSV G antibody). There is no discussion of this highly unusual HDAg composition. On the other hand, without showing specific L-HDAg and S-HDAg controls on the immunoblot, it is possible the electrophoresis is not resolving the two species (that is, the labeling is incorrect) and that it is not possible to determine the relative amounts of S-HDAg and L-HDAg from the blot shown. Reply: We have addressed these points and we are happy to provide a more detailed characterization of VSV-∆p and HCV-∆p particles, as described below : 1 -Demonstration that HDV RNAs in particles are genomes rather than antigenomes. We used a strand-specific RTqPCR assay (Li et al., 2006) to quantify HDV genomic RNA (gRNA) and antigenomic RNA (agRNA) in lysates and supernatants of transfected and/or infected cells (see new Supplemental Figure 1A). The enrichment of HDV gRNAs (panel A) in secreted particles is reflected by the gRNA/agRNA ratios (panel C), which were up to 800-fold higher in HDV, VSV-∆p or HCV-∆p particles than in the lysates of their corresponding producer cells. As shown in panels B and E, we noted no significant increase over time post-transfection of the low amounts of HDV agRNAs detected in the supernatants, in sharp contrast to the extracellular HDV gRNAs that increased by up to 1,000-fold (panels A and D). Owing to the high sensitivity of the RTqPCR assay, these low levels HDV agRNAs could be due to some background of cell death induced by the combination of GP transfection and extended culture conditions of these cells (up to 9 days). Note that identical extracellular HDV agRNAs levels were detected for VSV-∆p and HCV-∆p particles as compared to "normal" HDV particles (i.e., with HBV GPs). We believe that these new results show that HDV particles generated with unconventional GP incorporate full-length genomic RNA. 2 -Improvement regarding the detection of HDAg species present in viral particles. We replaced the HDAg co-IP analysis, which was not possible to improve at this stage, by a Western blot analysis of the different types of HDV particles that were purified by ultracentrifugation on a sucrose cushion. The results clearly show that both L-HDAg and S-HDAg are incorporated at similar levels and ratios in the purified viral particles generated with HCV and VSV GPs as compared to "normal" HDV particles produced with HBV GPs. 3 -Concerning the RNA coIP that precipitated 5 to 11% of HDV RNAs, we do not expect that the efficiency could be higher because of the competition exerted by the SVPs for either type of particles that outnumber the infectious particles. We respectfully request to keep the log representation of the graph in Figure 1E, in order to better show the results of the Flow through RTqPCR values.

2.
A second major concern is that the authors have not conclusively shown that the particles they have obtained are actually infectious. Several additional experiments are necessary to prove that the authors are observing actual infections rather than, for example, adherence of material to cells. A time course showing accumulation of HDV RNA and protein over several days following incubation would be more convincing than a single time point on day 7. Also more convincing would be demonstration of antigenomic RNA accumulation (assuming that the released RNA is genome) and an increase with time in the amount of L-HDAg, which only occurs during replication.

Reply:
We have performed all these experiments. The panels C-E of the new Figure 3 provides a time course analysis in infected cells over several days following inoculation. Using a strand-specific RTqPCR assay for HDV RNA, we show that not only genomic HDV RNAs but also antigenomic RNAs are amplified from day 3 to day 9 post-infection and accumulate in infected cells (panel C). Likewise, we confirm by Northern blot analysis of these infected cells that HDV RNAs accumulate in these cells (Panel D). Finally, we show that HDAg protein levels also increase with a progressive appearance of L-HDAg, which marks productive infection (panel E).
In addition to the other pieces of evidence such as i) inoculation of cell expressing vs. not expressing the receptors (now Figure 4A), ii) co-infection and transmission assays with live helper viruses (now Figure 6 and Supplemental Figure 6) and iii) propagation in experimentally infected animals (now Figure  7 and Supplemental Figures 7 & 8), we believe that altogether, these results convincingly show that the particles are infectious.
3. The authors included 2% DMSO in the cell cultures used to produce HDV particles and for infection experiments. The stated purpose was to retard cell growth. However, this treatment is unnecessary (2% DMSO is not typically included in experiments used to produce HDV nor to analyze infection in Huh 106 cells) and its use here raises questions about the generality of the results. Perhaps 2% DMSO is somewhat toxic (it does retard cell growth) and leads to the release of HDV from cells stressed by expression of certain viral glycoproteins. At least some of the production and infection experiments should be repeated in the absence of DMSO. Reply: We used Williams E-based medium for production or infection with HDV particles in Huh-7derived cells (including Huh-106 and Huh-7.5 cells). While this Reviewer is correct to say that DMSOcontaining medium is not typically included in such experiments, we supplemented our medium with 2% DMSO in this study since, when we started the project, the current procedures to grow hepatocytederived cell lines for several days recommended such media both to maintain cell differentiation (Bauhofer et al., 2012, Sainz & Chisari, 2006 and to induce or maintain NTCP expression , Yan et al., 2012. For example, it was shown that DMSO-containing media strongly increase HDV and HBV infection efficacy in HepG2 NTCP and Huh-7 NTCP cells, as compared to DMSO-free media (Iwamoto et al., 2014, Ni et al., 2014. Importantly, as requested by this Reviewer, we now show in the new Supplemental Figure 2, first, that 2% DMSO does not induce more cell toxicity as compared to cell cultures grown in DMSO-free medium (panel A) and, second, that omitting DMSO does not change the production levels and infectivity of HDV, VSV-∆p and HCV-∆p particles (panels B and C).

Additional points
1. There was considerable variability between some similar experiments, or unexplained inconsistencies. In Fig. 3, for example, the levels of HDV RNA detected in mock infected cells varied by almost 10-fold between panel A and panel B. While the RNA analyses suggest that HDV obtained from cells expressing VSV G replicates to 10-fold higher levels than HDV with an HBV envelope (Fig 2A), the immunofluorescence analysis shows approximately 5 times more cells positive for HDV with the HBV envelope (Fig. 2C).

Reply:
The previous immunofluorescence analysis of Figure 2C was displayed to provide a qualitative assessment of HDAg nuclear localization upon infection. As requested by this Reviewer, we have repeated the experiment and we provide in the revised Figure 2B images that are more consistent with the quantification of infectivity (now shown in Figure 2C) and that include nuclei staining with Hoechst. As for the variability between panel A and panel B of previous Figure 3 (now Figure 4), i.e., there is 3to 5-fold differences in the HDV RNA RTqPCR results obtained for the mock-treated supernatants for panel A vs. panel B. This is explained by the fact that either experiment type (i.e., neutralization in panel A and receptor blocking in panel B) was performed at different sessions of the study, which induces experimental variations. Yet, we had made sure that the sizes of the HDV, VSV-∆p and HCV-∆p particles inputs in either panel A or panel B were identical in order to provide accurate comparisons between conditions in each panel.
2. Page 3, second paragraph, second sentence. The statement that HDV does not meet the criteria for the definition of a virus is overstated. HDV is a satellite virus, with HBV as the only known helper.

Reply:
We have removed this sentence as requested.
3. Figure 5. The legend describes black and gray bars, there are only solid black and hashed bars. Reply: We have corrected the legend in this Figure (now revised Figure 6). Fig. 1; Group 9. According to the table in Fig. 6, only 4 animals were in this group, yet there are about 9 lines on the graph. There is no indication of what the colors represent in this graph nor on any of the others in this figure.

Supplemental
Reply: We thank this Reviewer for pointing this. Like for the "Mocks" group (Group #10), but also in other groups (not shown, for sake of clarity), we tested the samples for all three viruses (HBV, HCV and HDV) in Group #9, whose mice were infected with HDV only, and this is now indicated in the revised legend with the indicated color codes (blue: HBV; black: HCV; red: HDV). Note that one animal died after week 8 (P4), which explains that there are 9 lines after P4 (12 lines until P4) in this graph.

Reviewer #2 (Remarks to the Author):
Perez-Vargas et al. report that hepatitis D virus (HDV), which was thought to be a cognate satellite virus of hepatitis B Virus (HBV), may be enveloped by the surface glycoproteins (GPs) from a variety of viruses for its assembly and the release. The authors also found that the production and infection of HDV is not restricted to liver cells, but is restricted by the GPs enveloping HDV and the corresponding receptors expressed on cells. The authors performed various cell culture and animal experiments to support their observations. Co-expression of HDV RNPs and individual viral GP (HBV, VSV, HCV, etc) in hepatoma cell lines led to release of infectious HDV particles. The infection of different GP-enveloped HDV could be blocked by antibodies to the corresponding viral envelop proteins. Notably, in both HBV or HCV infected mice HDV viremia was detected, confirming the observations in a physiologically relevant experimental system. Based on these lines of evidence, the authors propose that HDV may have an origin independent of HBV, and could potentially be propagated by viruses other than HBV. The authors observation is surprising, given the scarcity of literature hinting HDV infection in the absence of an HBV infection. The authors have carved a unique angle to investigate the HDV propagation and provided both cell culture and animal evidences to support their hypothesis, which if proven true will enrich our understanding of HDV origin and the interplay of different viruses. However, several issues need to be addressed to further validate their hypothesis.
Major issues: 1. In figure 1 and figure 4, the key plasmid pSVLD3 was first developed in the Taylor lab (Kuo, et al, 1989, J. Virol) as a trimer of HDV genome and used for HDV replication. However, as it was indicated later (Taylor, 2006, Curr Top Microbiol Immunol), that the unit-length genomic or antigenomic HDV RNA was mostly DNA-directed rather than RNA-directed, and since this does not recapitulate the authentic HDV replication, an improved construct was made to have slightly larger than one unit length to produce RNA-derived transcripts (Lazinski & Taylor, 1994, J. Virol). The data produced in Figure 1 and 4 may not hold if the production of HDV RNPs is formed in way that deviates from the authentic mechanism. Two methods could be used: 1) NTCP expressing cells/cell lines expressing various viral GPs is infected by recombinant HBV enveloped HDV virus to test if infectious HDV particles is secreted, and 2) use the improved constructs to validate the findings. Reply: We thank this Reviewer for pointing out the original reference to this key construct, which is duly cited now (Kuo et al., 1989). Regarding his/her specific point, we had used in Figure 6 a variation of the method #1 he/she suggested. Accordingly, following infection of naïve cells with VSV-∆p particles (i.e., "helper-free" VSV-G enveloped HDV particles), the cells were superinfected with live HCV, HBV or DENV viruses. The results show that we could rescue infectious HDV particles, indicating that both pSVLD3 transfection and "helper-free" HDV infection (such as VSV-∆p) processes leads to expression of HDV RNA that can be transmitted as infectious particles. Note that a similar conclusion can be deduced from the mouse infection data since these animals were inoculated with "helper-free" HBsAgenveloped HDV virus before, after or concomitantly with live HCV or HBV (Figure 7). Finally, we show in the new Supplemental Figure 6 the results of transmission experiments using both Huh-7.5 human hepatoma and C6/36 mosquito cells that were infected with supernatants from HDV/DENV co-infected cells. We found that these secondary HDV/DENV-infected Huh-7.5 and C6/36 cells could replicate, assemble and transmit infectious HDV particles to tertiary cells. In our opinion, these results also demonstrate that HDV RNA can be transmitted via processes that involve authentic HDV replication.
2. In figure 1a, the release of virus through cell death should be ruled out. Reply: We now provide indirect evidence to address this point, which is very difficult to formally rule out since most of the GPs studied here can intrinsically cause cell death (albeit through different pathways). We evaluated the cytotoxicity in transfected cells at different time points of collection of HDV, VSV-∆p, and HCV-∆p particles as well as in No GP (i.e., pSVLD3-transfected cells) and non-transfected control cells. Using the Pierce Cytotoxicity Assay Kit, we obtained similar levels of LDH release for both VSV-∆p or HCV-∆p particles and "normal" HDV particles but also for the No GP control and the nontransfected condition as shown at day 6 post-transfection in the new Supplemental Figure 2. Thus, we conclude that while the combination of long-term culture and transfection procedure is somehow harmful to cells, the release of particles may not occur through cell death first, since otherwise, the No GP control (transfected with pSVLD3 only) would induce secretion of HDV RNAs from these cells, and second, since one may conclude that "classical" HDV particles would also be released through cell death. Note that slightly increased cytotoxicity levels were obtained when producing VSV-∆p particles; this was expected owing to the previously known fusogenic activity of VSV-G (see e.g., (Arai et al., 1998) for VSV-G-pseudotyped lentiviral vectors), which, ultimately, does not preclude production of such vectors.
Minor issues: 1. Figure 1b, the VSV-∆p 41A1 has a different pattern, e.g, much more s-HDAg and a 55 kDa band. A discussion of the difference, cause, and implication would be helpful.

Reply:
We agree with this issue of this Reviewer, which was also pointed out by the two other Reviewers. Accordingly, we have replaced this HDAg co-IP analysis, which was difficult to improve, by a Western blot analysis of the different types of HDV particles that were purified by ultracentrifugation on a sucrose cushion. The results unambiguously show that both L-HDAg and S-HDAg are incorporated at similar levels and ratios in the purified viral particles generated with HCV and VSV GPs as compared to "normal" HDV particles produced with HBV GPs.
2. Figure 2c should include light-field or nuclei staining. Reply: As requested by this Reviewer, we have repeated the experiment and we provide in the revised Figure 2B images that show nuclei staining (Hoechst) in inoculated cells.
3. Figure 5, it is not clear what is the HDV-RNA expressing cell, what construct is used to drive HDV-RNA expression? Reply: We infected these cells with VSV-G-coated HDV particles (VSV-∆p). Cells were then superinfected with the indicated live viruses (see our reply to Major issue #1 of this Reviewer).

Is there any explanation why the HDV infection is not reported to be dependent on HCV, if HCV
propagates HDV indeed rather efficiently? Reply: We are deeply interested by finding an explanation to this question, which is also raised by Reviewer #3. Indeed, one of our future plans is to attempt detection of HCV-dependent HDV propagation in selected patient cohorts, which is difficult clinically and logistically. As for tentative explanation, it is possible that direct or indirect (e.g., immune-mediated) interference mechanisms may impede in the long term HDV/HCV co-infections in vivo, though they may occur in different contexts. For example, HDV-induced activation of the innate immune response, which is known to have little/no effect on HDV itself (Alfaiate et al., 2016, Zhang et al., 2018, may impact several markers of co-infection, such as for HCV that is interferon-sensitive in contrast to HBV (Lutgehetmann et al., 2011. Thus, overall, we propose that what might determine eventual successful vs. sporadic transmission and propagation of HDV with non-HBV helper viruses would reside in the balance between biochemical/virological HDV RNP compatibility with the GP of these helper viruses vs. potential (immunological) mechanisms of interference, though the immune status of individuals, such as immunesuppression, may favor transmission of such HDV co-infections. This putative explanation is included in the revised Discussion.
5. At numerous occasions the authors refer to "data not shown". This reviewer regards it as important to show all data relevant to the study in the manuscript Reply: We have complied with the request of this Reviewer and we now display these data not shown in Figure 3 (kinetics of HDV replication in infected cells), in Figure 5 (results of infectivity for all HDV/GP combinations), in Supplemental Figure 5 (assessment of co-infection by IF), in Supplemental Figure 6 (infection of mosquito cells), and in Supplemental Figure 8 (results of infection from a second cohort of HuHep mice). The only results that are not shown are the FAH staining of humanized livers, as they have been published previously by us (Calattini et al., 2015) and others (Bissig et al., 2010).
6. Page 4: "While HDV was expressed…" rephrase as RNA is not expressed but transcribed or in this case replicated. Proteins are expressed.
8. Page 12: subheading, delta particle NOT particles Reply: We have introduced the requested change in this subheading.

Reviewer #3 (Remarks to the Author):
The origin of HDV attracts many interesting speculations. Current study proposed it coming from cellular circular RNAs that captured by HBV envelopes, so it is possible other viruses also enabling the packaging of naked HDV RNPs. They tested several viruses, such as HCV, dengue, HIV, et al, and showed HCV, denge and WNV, among others, coudl package the HDV RNPs into their envelope proteins and pass the HDV into next round of infections in cell cultures or in human hepatocyte chimera mice.
Though the results appeared to be interesting, their validation of pseudo-typed HDV and its infections is incomplete and many basic HDV RNA and proteins analysis are missing.
The comments are for authors' reference.
1. The claimed of packaging of HDV RNP by HBV Surface proteins, or by VSV or HCV GPs are interesting by in vitro co-transfection. This finding is supported by their immune-precipitation of these pseudo-typed HDVs, using anti-HBs or anti-VSV or anti-HCV GP. However, the validation of these packaged HDV falls far behind. To confirm the packaged HDV RNP, the authors only showed an ambiguous detection of so-called HDV large HDAg, but not HDV RNA. It is essentially to include a HDV virus packaged by HBV surface protein as a positive control. Such virion has to contain the HDV small delta antigen, and genomic HDV RNA, other than the so-called large HDAg. A Northern blot to confirm an intact, full-sized HDV RNA is required. The RT-PCR quantitation cannot distinguish the viral genemic vs. antigenomic HDV RNAs, either about the size. (In fact, the large HDAg detected in western blot appeared to be suspicious. Other HDAg-specific antibody is required, as it is difficult to understand why no small HDAg is co-packaged). Reply: We have performed these experiments and we provide in this revised manuscript a more detailed characterization of VSV-∆p and HCV-∆p particles : 1 -Incorporation of genomic HDV RNA in viral particles. First, we provide in revised Figure 1C a result of Northern blot that confirm the presence of an intact, full-sized HDV RNA is pellets of particles purified by ultracentrifugation on a 30% sucrose cushion. Second, we used a strand-specific RTqPCR assay (see revised Material and Methods) to quantify HDV genomic RNA (gRNA) and antigenomic RNA (agRNA) in lysates and supernatants of transfected and/or infected cells (see new Supplemental Figure 1A). The enrichment of HDV gRNAs (panel A) in secreted particles is reflected by the gRNA/agRNA ratios (panel C), which were up to 800-fold higher in HDV, VSV-∆p or HCV-∆p particles than in the lysates of their corresponding producer cells. Third, we designed a RT-PCR strand-specific "banding" assays with primers that allow amplification of the HDV genomic RNA. We found that HDV particles contains HDV RNA at the expected size of 1.7 kb, whether they were produced by transfection with pSVLD3 and GP-expression plasmids ( Figure 1B, 1H) or by co-infection with live HCV, HBV or DENV in vitro ( Figure 6) and in experimentally-infected animals (Supplemental Figure 7). Fourth, showing that VSV-∆p and HCV-∆p form particles, we performed Electron Microscopy analysis of particles ( Figure 1C and Supplemental Figure 3), which were obtained from cell supernatants purified with heparin beads, as discussed below in Point #3.

-Detection of HDAg species present in viral particles.
We replaced the HDAg co-IP analysis, which was not possible to improve, by a Western blot analysis of the different types of HDV particles that were pelleted by ultracentrifugation on a sucrose cushion. The results ( Figure 1D) clearly show that both L-HDAg and S-HDAg are incorporated at similar levels and ratios in the purified viral particles generated with HCV and VSV GPs as compared to "normal" HDV particles produced with HBV GPs.
2. The packaging of HDV RNP by HBsAg required specific isoprenylation of large HDAg. Do the rescue of HDV RNPs require the same modification or not ? This can be easily studied by using isoprenylation inhibitor currently available.

Reply:
We have performed the requested experiment and we show, in Figure 3A & B, that Lonafarnib, an isoprenylation inhibitor that prevents HDV assembly and secretion (Bordier et al., 2003), could readily inhibit production and hence, transmission and replication of HDV RNA from HDV, VSV-∆p and HCV-∆p particles, suggesting a shared pathway of the early assembly process leading to production of all HDV particle types.
3. The authors tried to band the VSV or HCV GP-packaged HDV virions by CsCL gradient analysis. They succeeded in identifying the putative pseudo-typed HDV in unique density fractions. Again, their only data based upon RT-PCR assay for HDV RNA. Northern and western blots to show HDV genomic RNA and both large and small delta antigens are essential. Finally, as the HDV virions are so abundant, it is necessary to do a simple EM study for these fractions to visualize the size, distribution of these pseudo-type HDV particles. Reply: We performed the requested experiment but failed to detect HDV RNA by Northern blot in that specific case, owing to their insufficient concentrations in fractions from density gradients. Indeed, the HDV RNA copy number required to perform Northern blots (>10 7 copies/lane -i.e., 20µl loaded -are needed) allowed detection of full-sized HDV RNA in the 100-fold pelleted viral particles shown in Figure  1C but not in the fractions from iodixanol density gradients, owing to dilution of the sample in the gradient. Note that the aim of this latter experiment was to performed density gradient analysis from unprocessed, crude supernatants (in order to maintain native state of the sample) and that these supernatants contain ca. 4x10 7 copies /mL for HCV-∆p particles (thus, the most HDV RNA-enriched fraction contain less than 2x10 5 copies/20µl, which is below the threshold level). As for detection of L-HDAg and S-HDAg, while we concentrated the fractions with methanol/acetone, the signals were not of sufficient quality for being displayed in Figure 1 owing to BSA levels that interfered with migration in SDS-Page.
To overcome these technical issues and address the request of this Reviewer regarding the packaging of full-sized HDV RNA, we used the above-mentioned RT-PCR banding assays and we show that the RTqPCR-positive fractions contain HDV RNA at the expected genomic size of 1.7 kb ( Figure 1G). As for EM studies, we provide in Figure 1F and Supplemental Figure 3 images of particles which were obtained from cell supernatants purified with heparin beads. We observed two types of spheres with diameters of 35-40 and 25-30 nm (Gudima et al., 2007). The small spheres likely corresponded to subviral particles since they were also detected when VSV-G and HCV-E1E2 were expressed alone, similar to HBV GPs (Supplemental Figure 3C & D) whereas the large spheres, that were only detected when HDV RNA were co-expressed with either GP (Supplemental Figure 3A & B), could correspond to VSV-∆p and HCV-∆p particles. While the concentration of these particles appeared insufficient to allow simple EM studies from the fractions of density-gradients, we believe that this consolidate the characterization of these novel HDV particles.
northern blot or western blot to show the simultaneous presence of HDV RNA or delta antigens. These are easily to show, as the HDV RNA titers are so high by their data. Besides, it is important to document the co-presence of HDAg and HCV antigen or dengue virus antigen in the same human hepatocytes from the chimera mice. Without these collaborating data, the HDV RNA RT-PCR seems shaky. It should be pointed that currently there is no approved HDV RNA assays, and many in-house assays suffer from varying or inconsistent performance. Reply: We show in the revised set of figures a more detailed characterization of these co-infection experiments.
First, regarding the presence of full-sized HDV RNA, using the above-mentioned RT-PCR banding assays, we show that cells co-infected with HDV and HCV or HBV or DENV express and secrete RNAs at the expected genomic size of 1.7 kb, which matches the detection of these RNAs by RTqPCR. Second, we performed an immunofluorescence (IF) analysis of these co-infected cells to document the co-presence of HDAg with HCV, HBV or DENV antigens in the same human hepatocytes. As shown in the revised Figure 6 and new Supplemental Figure 5, in addition to cells that were mono-infected by either virus type, we could readily detect cells co-expressing the antigens of either virus combinations (HDAg with HBV core, HDAg with HCV NS5A or HDAg with DENV E), which indicates that cells were co-infected by HDV and HCV, HBV or DENV. Third, regarding the chimera mice, we performed an IF analysis of liver sections from the co-infected animals. While we could readily detect mono-and co-infected hepatocytes in HBV/HDV co-infected mice, as shown previously by others (Lutgehetmann et al., 2012), the IF analysis of HDV/HCV infected animals was more difficult (see Figure below) and displayed rare co-infected cells with dull HDAg immunofluorescence, which may be explained at this stage by the following reasons. First, as the levels of HCV RNAs in this human liver mouse model are typically ca. 10-30 fold less elevated than for HBV, the propagation of HDV is less favorable with HCV helper virus than with HBV and results in a smaller proportion of infected hepatocytes. Second, at the time these animals were sacrificed (i.e., at week 14 post-infection ( Figure 7A)), the levels of HDV RNAs had decreased by over two-logs as compared to previous time-points (see e.g., week 8 for Group#8 in Figure 7), which made the analysis difficult to do. Third, we think that HDV and HCV interfere with each other, perhaps in a stronger manner that the previously known HDV/HBV interference (Alfaiate et al., 2016, Lutgehetmann et al., 2012, and this will be the subject of a further study of our team. Indeed, it is possible that direct or indirect (e.g., immunemediated) interference mechanisms may impede in the long term HDV/HCV co-infections in vivo, though they may occur in different contexts. For example, HDV-induced activation of the innate immune response, which is known to have little/no effect on HDV itself (Alfaiate et al., 2016, Zhang et al., 2018, may impact several markers of co-infection, such as HCV that is interferon-sensitive in contrast to HBV (Lutgehetmann et al., 2011. Thus, if this is acceptable, we respectfully request to this Reviewer that we do not display the IF results from the infected mice as they warrant further studies beyond the scope of this first study. We also believe that the results IF analyses of co-infection in vitro requested by this Reviewer are meaningful to establish the co-presence of HDV and live helper viruses. Finally, using the above-mentioned PCR banding assays, we show that sera from mice co-infected with HDV and HCV or HBV contain RNAs at the expected genomic size of 1.7 kb, in agreement with the presence these RNAs by RTqPCR, which supports our conclusion that the unconventional HDV particles can be secreted and propagated in vivo. 5. Finally, the HCV or dengue virus infections in humanized chimera mice took a lot of effort and showed intriguing results. Other than insufficient virological data as mentioned in point 3, the authors may need to study the natural HCV/HDV coinfection in human intravenous drug abusers who frequently coinfected by HBV/HDV/HCV. Do these patients carry HDV RNA within HCV envelope ? Reply: This is clearly a highly important and interesting question that we wish to pursue in the follow up work of this pioneer study. Indeed, how to best address this and reach statistical significance, given that natural HCV/HDV coinfections in human must be very rare in our opinion, is the subject of on-going discussions with clinicians who may collaborate with us on this issue. Yet, identifying and forming the different patient cohorts (e.g., from human intravenous drug abusers, as proposed by this Reviewer) or just accessing to collections of samples will require time, not only because we do not yet know exactly which are the best types of individuals to screen but also because getting the necessary ethical permits (and funding) is a difficult enterprise, particularly when it deals with countries where HDV is currently prevalent, like Mongolia and South America in the Amazonian basin. Finally, while Western countries, particularly Italia, have been severely hit by HDV infection in the 80's, recovering the collections of specimens from infected patients and their complete clinical description is difficult and will also take several months.