A 198 bp gene fragment, in which the original ORF3 initiation codon ATG at genome nucleotide positions 5347-5349 (nt numbering according to GenBank acc.-no. KC618403) was exchanged with GCA, was synthesized by IDT (Newark, NJ, USA) based on a previous publication (Yamada et al., 2009). First, this fragment was inserted into the subgenomic plasmid p47832fc2 (Scholz et al., 2020) by exchange of the corresponding ClaI/KpnI fragment. Thereafter, a PciI/SwaI fragment of the resulting plasmid was replaced with the corresponding fragment from plasmid p47832mc, which contains the complete genome of HEV genotype 3c strain 47832c (Scholz et al., 2020), resulting in the Δ-ORF3-plasmid. The sequence of the plasmid was verified by Sanger sequencing of the corresponding region using a commercial provider (Eurofins Genomics GmbH, Ebersberg, Germany).

After linearization of the pT7HEVwt- or pT7HEVΔORF3-plasmid using SwaI, the in-vitro transcription was carried out with the MEGAscript™ T7 Transcription Kit (Invitrogen, Carlsbad, USA) using 1µg DNA according to the manufacturer’s protocol for 4 h at 37°C, followed by the removal of the remaining DNA with TurboTM DNase at 37°C for 1.5 h. For the subsequent RNA purification, the MEGAclear™ Transcription Clean-Up Kit (Invitrogen, Carlsbad, USA) was used according to the manufacturer’s protocol. The provided elution solution was heated to 95°C and 50 µL were applied to the column before centrifugation at 15,000 × g for 1 min. This step was repeated to enhance the RNA recovery. Capping of the obtained RNA was performed with the Vaccinia Capping System (New England Biolabs, Frankfurt am Main) as described in the manufacturer’s protocol with an extended incubation time of 40 min. Due to the fact that the RNA was intended for transfection, the capped RNA was purified according to the RNA Cleanup protocol of the RNeasy Mini Kit (Qiagen, Hilden).

Three days before electroporation, the human hepatoma cell line PLC/PRF/5 (CLS cell lines service), cultivated in Minimum Essential Medium (MEM) Eagle supplemented with 10% heat-inactivated fetal calf serum (FCS), 2 mM L-glutamine, 1% non-essential amino acids (NEAA), 100 U/mL penicillin G and 100 μg/mL streptomycin (all reagents from PAN-Biotech, Aidenbach, Germany) at 37°C and 5% CO₂, were passaged in a ratio of 1:3 to reach 90-100% confluency. After trypsinization, a total 2 × 10⁶ cells in suspension were centrifuged for 4 min at 4°C and 500 × g. The cell pellet was resuspended in 400 µL DPBS (PAN-Biotech, Aidenbach, Germany), transferred to a 1.5 mL tube and the transcribed RNA was added. The suspension was mixed thoroughly before it was transferred to a 4 mm electroporation cuvette, which was stored on ice. The Gene Pulser Xcell™ (Bio-Rad, München, Germany) was used for electroporation set to the following parameters: square wave, 270 V, one pulse for 20 ms. Thereupon, the cuvette was incubated for 5 min at room temperature before the cells were resuspended in 10 mL DMEM supplemented with 5% FCS and 2mM L-glutamine and transferred to a T-75 cell culture flask. A change of the complete cell culture medium was performed after 5 hs and at three days post transfection. At day 7, the cells as well as the supernatant were harvested by three freeze-thaw-cycles and afterwards the suspension was centrifuged for 7 min at 3,220 × g. The virus-containing supernatant was harvested and used to infect native PLC/PRF/5 cells, which were already seeded two weeks prior to infection in a concentration of 1 × 10⁴ cells/cm². The cells were infected for 1 h at room temperature, which was followed by removal of the inoculum and washing three times with DPBS. Cells were then supplemented with DMEM incubated at 34.5°C and 5% CO₂ for five weeks with media exchange being performed twice a week. Finally, cell culture supernatant was cleared of debris via centrifugation for 7 min at 3,220 × g and stored at -80°C in 1 mL aliquots for the following experiments.

For infection of PLC/PRF/5 and the human lung carcinoma-derived subclonal cell line A549/D3 (Schemmerer et al., 2016), native A549/D3 cells were cultivated in Dulbecco’s Modified Eagle Medium (DMEM) and native PLC/PRF/5 cell were cultivated in MEM Eagle, both supplemented with 10% fetal calf serum (FCS), 2 mM L-glutamine, 100 U/mL penicillin G and 100 μg/mL streptomycin (all reagents from PAN-Biotech, Aidenbach, Germany) at 37°C and 5% CO₂ (Schemmerer et al., 2016; Gremmel et al., 2022). Cells were seeded in a concentration of 1 × 10⁴ cells/cm² two weeks prior to infection with medium exchanges being performed twice weekly. The infection proctocol was the same as for the generation of infectious virus, but only supernatant from infected cells was used as inoculum. A medium exchange was performed one week post infection. For the infection kinetics, the cells were incubated for a total of three weeks at 34.5°C and the cell culture medium was renewed twice a week. To determine the genome copies in cell culture supernatants, nucleic acid was extracted by using the KingFisher Duo Prime System (Thermo Fisher Scientific) and the IndiMag Pathogen Kit (Indical Bioscience, Leipzig, Germany) and a previously described ORF3-detecting qPCR protocol was applied in combination with an in-house HEV RNA copy standard (Jothikumar et al., 2006). To determine the viral titer, the number of focus-forming units (FFU) was assessed as described elsewhere (Gremmel et al., 2022). Briefly, PLC/PRF/5 cells were seeded in 96-well plates two weeks prior infection. The sample to be investigated was diluted in a 10-fold series and tested in quadruplicates and incubated on the cells for 1 h at room temperature. After removal of the inoculum and another incubation period of one week at 34.5°C the plates were evaluated by immunofluorescence staining (Gremmel et al., 2022) and specific foci were counted in the highest dilutions showing positive signals.

To test for the presence or absence of pORF2 and pORF3 in infected PLC/PRF/5 cells, cell lysates of HEVΔORF3- and HEVwt-infected PLC/PRF/5 cells were analyzed via Western Blot (WB). To generate cell lysates, culture supernatants were discarded and cells were washed once with DPBS. These were then lysed in NP-40 lysis buffer for 10 min on ice, which was followed by freezing at -80°C. Lysates were then sonicated (Bandelin Sonoplus HD2070.2) for 10 s and cleared of debris via centrifugation at 4°C, 10,000 × g for 10 min. Samples were then supplemented with SDS sample buffer and denatured at 95°C for 10 min and subjected to WB analyses. To determine HEV pORF2 and host proteins in cell culture supernatant and density gradients, WB analyses were performed as well. Aliquots of cell culture supernatant and density gradient fractions were therefore directly supplemented with reducing SDS sample buffer, denatured at 95°C for 10 min and subjected to WB analyses.

Cell lysates or density-gradient fractions were separated on a 12.5% (lysates) or 10% (fractions) SDS-PAGE (sodium dodecyl sulfate-polyacrylamide gel electrophoresis). Proteins were then blotted onto PVDF (polyvinylidene fluoride) membranes via a semidry electroblotting procedure. Blocking of membranes was performed by overnight incubation at 4°C (lysates) using the Amersham ECL Prime Blocking Reagent (GE Healthcare, RPN418) diluted 1:50 in TBS-T (TBS supplemented with 0.02% Tween 20) or by incubation for 30 min at room temperature (RT) (fractions) in ROTI^®Block (Carl Roth, A151.1) diluted 1:10 in ddH₂O. Membranes were then probed with the primary antibodies rbαpORF2 (dilution 1:500, kindly provided by R. Ulrich, FLI for lysates; dilution 1:2,000, M. Glitscher, PEI for fractions), rbαpORF3 (dilution 1:100, M. Glitscher, PEI) or mαTGN46 (Thermo, MA3-063) for 1 h at room temperature (lysates) or overnight at 4°C (fractions) in blocking buffer. After washing three times in blocking buffer (lysates) or TBS-T (fractions), HRP-conjugated (lysates) or fluorophore-conjugated (IRDye680RD or IRDye800CW; fractions) secondary antibodies (Invitrogen, 31466, dilution 1:5,000; LI-COR Biosciences, 926–32213 or 926-68072, dilution 1:5,000) were incubated for 1 h at room temperature in blocking solution. After washing, chemiluminescence was imaged using the Amersham ECL Select Western Blotting Detection Reagent (GE Healthcare, RPN2235) was used to visualize the bound antibody via chemiluminescence, while fluorescence was measured using a LI-COR Odyssey CLx instrument.

Uninfected and HEV-infected PLC/PRF/5 and A549/D3 cells were seeded on glass coverslips in 12-well plates in a concentration of 6 x 10⁴ cells/cm² and grown for 72 h. Cell culture supernatant then was aspirated and cells were washed with DPBS, which was followed by fixation for 30 min at -20°C in ice-cold 100% (v/v) ethanol. Samples were then washed with DPBS and blocked for 15 min at RT in 5% (w/v) bovine serum albumin (BSA; Carl Roth, T844.4) diluted in TBS-T. Subsequently, diluted primary antibodies mαpORF2 (dilution 1:200, clone 5G5, kindly provided by J. Meng, Southeast University, Nanjing, China), rbαpORF2 (dilution 1:500, M. Glitscher, PEI), rbαpORF3 (dilution 1:250, M. Glitscher, PEI), mαCD63 (dilution 1:200, abcam, ab59479), rbαGM130 (dilution 1:200, Cell signaling, 12480S), mαTGN46 (dilution 1:100, Thermo, MA3063) or rbαRab11 (dilution 1:100, Cell signaling, 5589T) were incubated in blocking buffer for 1 h at RT. Samples were then washed three times with TBS-T and incubated with diluted fluorophore-conjugated secondary antibodies (dilution 1:500, Alexa Fluor™ 488/546/633, Invitrogen, A-21202, A-21206, A-10036, A-10040, A-21052 or A-21071) in blocking buffer supplemented with 250 ng/mL DAPI (4′,6-diamidino-2-phenylindol; Carl Roth, 6335.1) for 1 h at RT. After washing three times with TBS-T, samples were mounted onto glass slides using Mowiol mounting medium (Sigma Aldrich, 324590-100G). Image acquisition was performed with a Leica TCS SP8 System using a 100× oil immersion objective (numerical aperture = 1.4) and the LAS X Control Software (Leica). The scan speed was set to 100 with an image size of 2048 × 2048 pixels, while the pinhole set to 1.0 airy units resulting in a z-slice thickness of 0.4 μm. Deconvolution was achieved by applying the lightning algorithm (Leica) at default settings. Equal channel intensities were set and images were transformed to grayscale afterwards, if needed. Quantification of CTCF (corrected total cell fluorescence) and tMOC (thresholded Mander’s overlap coefficient) were performed using FIJI (Schindelin et al., 2012) as described previously (Glitscher et al., 2024). Similarly, outlines of structures were extracted using FIJI by thresholding grayscale images and using the find edges processing tool, which were then merged to an RGB composite.

Immune electron microscopy was performed as described earlier (Horvatits et al., 2021). The HEV particles from cell culture supernatant were identified by using the above-mentioned primary antibodies rbαpORF2 and rbαpORF3 and detected by a secondary gold-labelled antibody. Negative staining of the labelled samples was done using 1% uranyl acetate. The samples were thereafter examined through transmission electron microscopy using a JEM 1400 microscope (Jeol, Tokyo, Japan) operated at 120-kV acceleration voltage. Images were captured using a Veleta G2 camera.

To determine the buoyant density of HEV particles, isopycnic density-gradient centrifugation was performed as described previously (Glitscher et al., 2021b). In brief, samples originating from PLC/PRF/5 and A549/D3 cell culture supernatant, thus representing extracellular samples, were cleared of debris via centrifugation at 10,000 x g for 10 min at 4°C. Only for A549/D3 cells, exosomes were pelleted at 150,000 xg for 2 h at 4°C in an SW-41 swing-out rotor (Beckman Coulter, USA) and resuspended in PBS prior to detergent-treatment. Subsequently, samples were treated by addition of 10% (w/v) sodium deoxycholate (NaDoC) dissolved in H₂O to reach a final concentration of 1% (v/v) and incubated at 37°C for 2 h. Samples were then loaded on an iodixanol (OptiPrep; 1114542; Progen Biotechnik, Heidelberg, Germany) gradient consisting of six layers 10-60% (w/v) iodixanol diluted in water. Gradients were centrifuged at 255,000 x g for 4 h at 4°C in a TLS-55 swing-out rotor (Beckman Coulter, USA). Gradients were then fractionated in ten fractions, density and subjected to determination of refractive indices, viral and host proteins and genomic HEV copies. The latter were quantified by diluting fractions 1:5 in ddH₂O supplemented with 0.1% (w/v) DEPC and subjecting them to RT-qPCR via the LightMix Modular Hepatitis E Virus Kit (53-0638-96; TIB MolBiol, Berlin, Germany) in combination with LightCycler Multiplex RNA Virus Master (6754155001; Roche Diagnostics, Mannheim, Germany) according to the manufacturer’s instructions in a LightCycler 480 Instrument II (Roche Diagnostics).

The neutralization assay was performed as described previously (Gremmel et al., 2023). Briefly, PLC/PRF/5 cells were seeded two weeks prior to inoculation. Log₄ dilution series of serum samples were prepared and incubated with defined infectious dose of NaDoC-treated and untreated virus to compare the neutralization capability. The virus treatment was carried out with 1% NaDoC for 2 hs at 37°C followed by a centrifugation of the virus through Vivaspin 20, 50,000 MWCO PES Ultrafiltration Units (Sartorius, Göttingen, Germany) for 10 min at 3,320 × g (Gremmel et al., 2023). Afterwards the inoculum was applied onto the PLC/PRF/5 cells and incubated for 1 h before washing the cell layer and incubation for one week at 34.5°C. At 7 dpi plates were evaluated by immunofluorescence staining and the neutralization titer (neutralization doses 50%; ND₅₀) was calculated.

If not stated otherwise, three independent experiments were used to collect data for this study. Values are displayed as mean ± SEM. If a control group does not display an error bar, relative comparisons led to the control being arbitrarily set to 1, thus representing a standardization performed for each independent dataset. Data distribution was controlled for normality using the Shapiro-Wilk test (alpha = 0.05). In case of a Gaussian distribution, an ungrouped, parametric t test was performed to compare groups. Else, an ungrouped, non-parametric t test (Mann-Whitney test) was applied. Graphs and statistics were generated using GraphPad Prism v9.0.

As an approach to study nHEV, we aimed to generate a pORF3-deficient virus strain based on the human genotype 3c strain 47832c (Johne et al., 2014). The rationale was to diminish quasi-envelopment by deleting the start codon of ORF3 in a reverse-genetic system. Unexpectedly, we noticed that the resulting virions from this HEVΔORF3 variant still behaved surprisingly similar to the wild-type HEV (HEVwt) in terms of its infection kinetics on the hepatocellular carcinoma cell line PLC/PRF/5, which is frequently used in studies of HEV.

To make sure the start codon of the ORF3 protein was successfully mutated, Sanger sequencing was performed from the original plasmid, the ΔORF3 plasmid and from virus harvested from cell culture supernatant after a first virus passage on PLC/PRF/5 cells. Further, cells replicating HEVwt and HEVΔORF3 were analysed for the presence or absence of pORF2 and pORF3 in WB analyses and confocal microscopy. Additionally, the infection kinetics of both virus forms was monitored on PLC/PRF/5 cells ( Figure 1 ).

As depicted above, sequence analyses revealed an over 280 bp spanning fragment of the ORF3 region (nt 5320–5357 not shown). Here, the third ATG is responsible for initiation of translation and retains its mutation, ATG to GCA, in both the ΔORF3 plasmid and the infectious virus from the first virus passage ( Figure 1A ). This went along with the absence of pORF3 in both WB analyses and confocal microscopy ( Figures 1B, E ). The modestly reduced pORF2 signal in the Western blot reflects a lower proportion of HEV-positive cells in that lysate, caused by normal sample-to-sample variation between independently maintained persistently infected cultures (HEVwt vs. HEVΔORF3). In the latter, an efficient infection can be seen even in the case of HEVΔORF3, with micrographs notably being performed on persistently infected PLC/PRF/5 cells (after cryo-conservation and passaging several times). The distribution of pORF2 seems to change in most of the infected cells upon deletion of ORF3, where instead of smaller or bigger dot-like structure, additional fibre-like structures can be seen ( Figure 1E ), which will become apparent in a later section of the study. In the ORF2 staining of the non-infected cells, minor artifacts (red) are visible, which do not represent a specific ORF2 signal. The capability of HEVΔORF3 to still maintain such persistent infection correlates with its similarly pronounced infection kinetic as compared to HEVwt ( Figure 1C ). Over the course of up to 25 days, viral genome-load in cell culture supernatants increase to a similar extent in both wild-type and mutant strain, reaching up to 10^8–10⁹ copies/mL. Lastly, the relative infectivity was assessed to monitor whether virions of both variants were similarly efficient. With the result that there is a significant decrease in relative infectivity – the number of genome copies required to form one FFU (focus-forming unit) ( Figure 1D ). Effectively, infection efficiency therefore is reduced by the deletion of ORF3 with an infectious titer of 2.0 × 10⁵ FFU/mL (relative infectivity of 4.3 × 10³ RNA copies/FFU) compared to an infectious titer of 5.3 × 10⁵ FFU/mL for HEVwt (relative infectivity of 8.8 × 10² RNA copies/FFU).

Taken together, these findings suggest that, although viral particles originating from HEVΔORF3 are less infectious as compared to HEVwt, viral release is not significantly hampered. This majorly is visible by the unaltered infection kinetics and the capability of the variant to build up a persistent infection in PLC/PRF/5 cells.

Quasi-envelopment by the ESCRT machinery is, even up until today, the sole described route of viral egress in case of HEV. As for the similarities in infection kinetics and the differences in relative infectivity, the virion morphology was studied in a virus neutralization assay. Essentially, the aim was to answer the question whether HEV virions originating from HEVΔORF3 still retain an envelope or form in fact nHEV particles, as was the general state of research.

As known from literature, eHEV is susceptible to detergent-treatment. Here, the viral envelope is stripped off and makes virions susceptible to a neutralization with anti-pORF2 antibodies and sera (Gremmel et al., 2023). To compare the neutralizability of HEVΔORF3 viral particles in a similar fashion, virions were treated with sodium deoxycholate (NaDoC), a bile acid, or were left untreated. Subsequently, four wild boar sera as well as a rabbit anti-HEV hyper immune serum were used to determine the ND₅₀ ( Figure 2 ).

As shown in Figure 2B , no clear neutralization could be seen when using the non-treated virus as test virus applying the scheme depicted in Figure 2A . Instead, there were randomly positive wells showing ≥ 3 foci in the immunofluorescence. In contrast, the NaDoC-treated HEVΔORF3 can be neutralized efficiently in serum dilutions up to 1:1,024 (s/473). For the serum sample s/180 a clear neutralization could be proven in dilutions up to 1:4,096, but one well in serum dilution 1:16 contained ≥ 3 foci. This was in line with rabαHEV sera. The quantitative data of the neutralization assay is shown in Figures 2C–H as FFU in each well in relation to the serum dilution. Additionally, the neutralizing capacity of control serum rabαHEV regarding HEVwt in untreated and treated form is in line with the neutralization of HEVΔORF3 – no clear neutralization of untreated HEVwt, but complete neutralization of treated HEVwt in serum dilutions of up to 16,384.

In summary, the inefficient neutralization of non-treated HEVΔORF3 virions by anti-pORF2 antibodies and the neutralizability of virions treated with a detergent suggest that quasi-enveloped HEV species may still exist even in the absence of the accessory protein pORF3.

As neutralization of HEVΔORF3 virions only was achieved after applying a detergent, the possibility of quasi-envelopment still being in place for this variant was addressed further. Conventionally, quasi-envelopment of HEV virions takes place at the surface of MVBs. Here, pORF3 mediated contact to ESCRT-components, thus leading to incorporation of HEV capsid into ILVs and to subsequent exosomal release. Therefore, HEV virions are classified by an exosome-membrane tightly wrapping the capsid structure, thus giving rise to an exosomal buoyant density of ~1.15 g/mL (Nagashima et al., 2017). Accordingly, pORF2 in eHEV virions is not accessible to anti-pORF2 antibodies.

To further refine the findings made for virus neutralization, the buoyant density of eHEV virions originating from both HEVwt and HEVΔORF3 therefore was assessed through isopycnic density-gradient centrifugation. Additionally, virions were aimed to be labelled with antibodies targeting pORF2 and pORF3 in an immunogold stain, which was followed by negative contrasting and TEM analysis. Both analyses were performed in the absence or presence of NaDoC to pinpoint whether effects seemingly being due to a quasi-envelope can be reverted by a detergent-treatment ( Figure 3 ).

As indicated by isopycnic density-gradient centrifugation, untreated HEVwt virions display the most prominent viral RNA peak at a density of ~1.17 g/mL ( Figure 3A ). Upon detergent-treatment, this pORF2-population shifts to a density of ~1.24 g/mL. These densities and changes upon detergent-treatment can also be observed for HEVΔORF3 virions by the same shifting from ~1.17 g/mL to ~1.24 g/mL ( Figure 3B ). This strictly correlates, as indicated by neutralization assays in advance, with the accessibility of pORF2 and pORF3 to antibodies. For HEVwt virions, pORF3 can be detected in particles displaying a size of ~30–40 nm, yet no pORF2-specific stain is visible. Only upon treatment with NaDoC, pORF2 is stainable, yet pORF3 is no longer ( Figure 3C ). Again, just as for HEVwt virions, this is the case for HEVΔORF3 virions with the exception of pORF3 not being stainable by an anti-pORF3 antibody for obvious reasons ( Figure 3C ).

These findings clearly provide evidence for an envelope still being in place in case of HEVΔORF3 and shielding the capsid structure from the surrounding.

As delineated above, quasi-envelopment of HEV is dependent on MVBs. As quasi-envelopment evidently is present once pORF3 is deleted, the organelle harboring pORF2, thus the capsid structure, was aimed to be assessed. As shown above, we observed differences in infectivity of virions. Further, recruitment of pORF2 to MVB-resident ESCRT-components should theoretically not be possible if pORF3 is deleted. This may also hold true for recycling endosomes playing a role in viral morphogenesis, as postulated by another study (Bentaleb et al., 2022). We therefore expanded the readout on a possible egress via the Golgi network, which might present an alternative to the exosomal route. Firstly, eHEV virions have been described to have the potential containing Golgi-components (Nagashima et al., 2014b) and secondly, the fiber-like distribution of pORF2 upon deletion of ORF3 ( Figure 1E ) was somewhat reminiscent of the Golgi apparatus.

In order to analyze the subcellular localization of pORF2 with respect to these different organelles, confocal microscopy was used. Infected cells were stained for CD63 (as MVB-marker), GM130 (as cis-Golgi marker), TGN46 (TGOLN2, a trans-Golgi marker) or Rab11 (a marker for recycling endosomes) alongside pORF2. Notably, this strategy was used to successfully validate functional experiments and gain fundamental knowledge about HEV based on the pORF2 localization, regardless of the isoform, before (Glitscher et al., 2021a, 2021b; Schwickert et al., 2024; Woytinek et al., 2024). These were then analyzed for colocalization with each other by calculating the thresholded Mander’s overlap coefficient ( Figure 4 ).

As seen in Figure 4A , pORF2 in HEVwt-infected cells localizes in large, dot-like structures. These are often found overlapping or in close proximity to similarly shaped CD63-positive vesicular structures, indicating the presence of virions in or at MVBs. While there are still some of these large, dot-like pORF2 structures visible in case of HEVΔORF3, these do appear more elongated. Especially once extracting the outlines of these structures, it becomes apparent that pORF2 of HEVΔORF3-infected cells does not colocalize with CD63 to the same extent, which can also be confirmed by a significantly lower tMOC for pORF2::CD63 in HEVΔORF3-infected cells than in HEVwt-infected ones. Effectively, deletion of pORF3 leads to a significant loss in MVB-resident pORF2 ( Figure 4B ). When addressing the cis-Golgi compartment through GM130, large tubular structures representing this organelle can be seen ( Figure 4C ). There is some overlap of perinuclear pORF2 in case of HEVwt with this compartment, which likely reflects the non-capsid associated isoform as for its recruitment to the ER, the subsequent glycosylation and release via the secretory route. This is far less pronounced in case of HEVΔORF3. As for MVBs, there is a significant reduction in pORF2::GM130 colocalization upon deletion of pORF3, although it becomes apparent that pORF2 in this setting resides next to cis-Golgi structures. These structures next to the cis-Golgi network share a high degree in colocalization with the trans-Golgi marker TGN46. In comparison to GM130, the TGN46-positive structure form a finer network of fibre-like stretches alongside some dot-like patterns. For HEVwt-infected cells, occasional overlaps between pORF2 and the trans-Golgi network can be observed. In case of HEVΔORF3 however, the fibre-like distribution of pORF2 often completely matches the trans-Golgi structure ( Figure 4E ). Notably, this heavy change in pORF2 localization upon deletion of ORF3 likely means that all pORF2 isoforms are changing their localization. This observation correlates with a significant increase in pORF2::TGN46 colocalization after deletion of pORF3, shown by a significantly higher tMOC pORF2::TGN46 in HEVΔORF3-infected cells compared to HEVwt-infected cells ( Figure 4F ). This change in distribution does not significantly affect the population of pORF2 being associated with recycling endosomes ( Figures 4G, H ), which highlights the role of the TGN compartment in this context.

To test these observations in a different cell culture model, essential experiments for infection kinetic, release of eHEV virions and subcellular distribution of pORF2 were repeated on A549/D3 cells ( Figure 5 ).

Analyses of the viral variants on A549/D3 cells revealed substantial differences to PLC/PRF/5 cells. While HEVwt replicated well and released increasing amounts of viral RNA nearly reaching 10⁷ copies/mL in cell culture supernatants, HEVΔORF3 appeared to be dramatically attenuated ( Figure 5A ). While both variants started at RNA titers of around 10⁵ copies/mL, the latter declined in titre over time. Similarly, exosomal HEV RNA was significantly reduced by nearly four orders of magnitude in case of HEVΔORF3 ( Figure 5B ). In line with this, HEVwt RNA peaked at a density of around 1.12 g/mL in density gradient centrifugation, displaying a slightly lower density as virions originating from PLC/PRF/5 cells ( Figure 3A ), yet similarly were susceptible to detergent-treatment ( Figure 5C ). Opposingly, HEV RNA peaked at very high densities of around 1.28-1.33 g/mL, thus not matching exosomal density and likely representing other RNA-containing complexes such as naked capsids ( Figure 5D ). In terms of subcellular localization, pORF2 of the HEVΔORF3 variant appeared differently as compared to PLC/PRF/5 cells. In A549/D3 cells, pORF2 was often found inside the nucleus and in dot-like structures, yet not in fibre-like compartments. Interestingly, there was only a minor reduction in MVB-resident pORF2 upon deletion of pORF3 ( Figures 5E, F ). While, similar to PLC/PRF/5 cells, a decrease in cis-Golgi association ( Figures 5G, H ) and no major change in association to recycling endosomes ( Figures 5K, L ) upon deletion of pORF3 could be observed, dramatic differences between the cell lines can be seen in the TGN-association. Here, a significant reduction in pORF2-TGN46 colocalization was observed ( Figures 5I, J ), thus representing the exact opposite of the situation in PLC/PRF/5 cells.

It is reasonable to assume that the difference in viral performance lies within a difference in cell types – specifically, in their different behaviour when it comes to the crosstalk with the TGN compartment. These findings were further set into context by assessing the morphology of both MVBs and the TGN compartment in both PLC/PRF/5 and A549/D3 cells ( Figure 6 ).

A direct comparison of the possible release-competent compartments revealed staggering differences in the TGN morphology between the cell lines used in this study. MVBs, as marked by CD63-positivity, expectedly appeared as vesicular structures scattered throughout the cell in both cell lines. While this held true for the majority of TGN46-positive structures in A549/D3 cells, the TGN compartment was much more fibre-like in PLC/PRF/5 cells. While these fibres were sometimes observed in perinuclear regions of A549/D3 cells, the network in PLC/PRF/5 cells was found to be far more elaborate and stretched throughout the cells. These fibres again are very comparable to the pORF2-fibres found for HEVΔORF3 in PLC/PRF/5 cells, which also colocalize with this compartment ( Figures 4E, F ). This major difference in TGN-morphology may explain the difference in pORF2 localization and virion release, thus viral fitness, between A549/D3 and PLC/PRF/5 cells when it comes to the HEVΔORF3 variant.

If pORF2 of HEVΔORF3 is less associated to MVBs, but more associated to the trans-Golgi network in PLC/PRF/5 cells, a switch in route of egress and a subcellular re-arrangement once ORF3 is deleted can be hypothesized. To test for the possibility of HEVΔORF3 eHEV being released or accompanied by markers of the trans-Golgi network, samples of density-gradients were subjected to WB analyses. This scheme was used to track whether trans-Golgi-markers co-migrate with HEV capsids once the quasi-envelope is removed ( Figure 7 ).

WB analyses performed on eHEV virions from the different viral variants provide evidence that shifts in densities upon detergent-treatment seen for viral RNA ( Figure 3 ) do match with shifts seen for capsid-associated pORF2 ( Figure 7A ). Here, pORF2 of a molecular weight of ~70 kDa peaks at densities of ~1.17 g/mL in untreated samples. Conversely, detergent-treatment results in a shift in density to ~1.24 g/mL ( Figure 7B ). The capsid independent, glycosylated and secreted form of pORF2 at molecular weights greater than 70 kDa is not affected by this and bands at lower densities of ~1.12 g/mL. Most interestingly however, TGN46 is present in supernatants of both virus variants ( Figure 7A ). The striking difference lies within fractions containing most capsid-associated pORF2 in case of HEVΔORF3. Here, a clear enrichment of TGN46 can be seen, possibly due to being part of a protein complex still being attached to virions, which matches the overall distribution of capsid-associated pORF2 at ~1.17 g/mL. This, however, is far less pronounced in case of HEVwt. Upon detergent-treatment, this peak is shifted to 1.24 g/mL ( Figure 7C ) in case of HEVΔORF3, just as is the case for HEV pORF2. This aligns with the immunofluorescence findings, showing that TGN46::ORF2 colocalization occurs significantly more often in HEVΔORF3-infected PLC/PRF/5 cells than in those infected with HEVwt ( Figures 4E, F ).

In summary, these data provide intriguing evidence for deletion of pORF3 leading to: (i) a re-routing of pORF2 from MVBs to the trans-Golgi network and (ii) an induction of trans-Golgi dependent release of eHEV. This may very well explain why there is still an envelope present even in the absence of a pORF3-mediated recruitment to ESCRT-components, which is an entirely new finding.