Detection of a novel TAMV variant in host-seeking Amblyomma americanum ticks

The advent of novel powerful NGS approaches sharply increased the discovery rates of new emerging viruses [44]. We used an unbiased NGS approach to screen samples of A. americanum ticks trapped in Florida [41], which were previously found positive for TCRV, for the presence of additional mammarenavirus sequences. To this end, 100 ticks were pooled to prepare cleared homogenates [45] and subjected to three blind passages on VeroE6 cells [41]. We used tissue culture supernatants (TCS) of the third passage to prepare cDNA libraries for Illumina-based NGS analysis. To facilitate detection of under-represented viral sequences, we employed a step-wise strategy (Fig 1A). In our NGS run, we found 99158 and 621 out of 105780 total filtered reads corresponding to RNA of TCRV and TAMV, respectively (S1 Table). For further analysis, candidate arenavirus sequence reads were assembled in contigs longer than 500 bp and subjected to BLAST analysis against available reference viral genomes (Fig 1A and S2 Table). Obtained reads were assembled into 4 contigs corresponding to the complete L and S segments of TCRV and TAMV (S2 Table). Comparison between the obtained TCRV genomic nucleotide sequences (TCRV-FL. Genbank accession numbers MW150032.1 (L) and MW150033.1 (S), respectively) with the TCRV reference strain 11573 (MT478050.1 and MT478051.1) [46], reveals overall identities of 99.87% and 99.91% in segments L and S, respectively (Fig 1B and 1E, and S2 and S3 Tables), confirming the remarkably high similarity between tick-derived TCRV-FL and the TCRV reference strain. Nevertheless, despite the similar origin of the TCRV-FL and the TCRV Florida isolate [41], the comparison with other available TCRV sequences revealed TCRV-BEI strain [46] as the closest relative of TCRV-FL (S3 Table). Alignment of the obtained reads against the rodent-borne TAMV reference strain W-10777 (TAMV-Ref. NC_010701.1 and NC_010702.1) [47] reveals a high sequence variation despite uneven and limited coverage (Fig 1C). Therefore, we next assembled a full-length tick-derived TAMV genome (TAMV-FL. GenBank access numbers MK500936 and MK500937) from the reads obtained in our NGS run (Fig 1D). In contrast to the tick-derived TCRV, the nucleotide and amino acid sequences of the assembled tick-derived TAMV-FL differ significantly from the TAMV-Ref isolated from the cotton rat Sigmodon hispidus in 1965, in Florida, and other available TAMV sequences [48] (Table 1 and Fig 1E and S1 Fig). The complete sequences of the S and L segments of TAMV-FL contained 3526 and 7143 nucleotides, respectively (S2 Table). The S segment shows 85.28% identity at the nucleotide level and 89.93% and 90.33% at the amino acid level of NP and GPC, respectively, compared to TAMV-Ref (Fig 1E and Table 1). The identity in the L segment at the nucleotide level is 87.18% and at the amino acid level 95.29% and 91.77% for the Z protein and the viral polymerase L, respectively (Fig 1E). The extent of sequence variation is incompatible with a laboratory contamination and raised the possibility of TAMV-FL being a new mammarenavirus species. The current species definition of mammarenaviruses is based on several criteria, including association with a specific host species, geographic distribution, status as human disease agent, lack of serological cross-neutralization, and the results of pairwise sequence comparisons (PASC) of coding-complete genomes using >12% divergence in NP amino acid sequence as one of the criteria for species demarcation [12]. Despite significant sequence deviation, TAMV-FL fulfills the criteria to be considered a strain of the same species as TAMV.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1.

(A) Work-flow followed for the NGS analysis. (B) TCRV read coverage (Grey area) and relative abundance of SNV (red dots) aligned against TCRV strain 11573 in our tick-derived isolate. (C) TAMV read coverage (Grey area) and relative abundance of SNVs (red dots) aligned against TAMV-Ref in tick-derived isolate stock. (D) TAMV read coverage (Grey area) and relative abundance of SNVs (red dots) aligned against TAMV-FL in our tick-derived isolate. (E) TCRV (red bars) and TAMV-Ref (blue bars) sequence homology at nucleotide and amino acid level in our tick-derived isolate. Alignments were performed with Blastn.

https://doi.org/10.1371/journal.pntd.0009004.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Sequence comparison of TAMV-FL, TAMV-Ref, TAMV- W10777 and AV97140103 strains.

Single Nucleotide Polymorphisms (SNPs) are shown for each TAMV strain. *: Only S segment sequence was available for TAMV AV97140103 strain.

https://doi.org/10.1371/journal.pntd.0009004.t001

TAMV-FL utilizes human transferrin receptor-1 more efficiently than TAMV-Ref

The ability of NW mammarenaviruses to use hTfR1 is a critical factor for their potential for zoonotic spillover and ability to cause human disease [21,22,25,26,28,29,50,51]. Although several additional factors are involved during zoonotic events, the importance of hTfR1 in this process prompted us to examine whether TAMV-FL was able to use hTfR1 to infect human cells. Since mammarenavirus cell attachment and entry are both mediated by the viral envelope, we used a bio-contained pseudotype platform to study cell entry mediated by the GP of selected mammarenaviruses. Recombinant vesicular stomatitis virus (rVSV)-derived pseudoviruses (PVs) displaying heterologous viral GPs have become major tools to study cell entry of highly pathogenic emerging viruses [52–59]. We successfully generated PVs decorated with the GPs of JUNV, TAMV-Ref, TAMV-FL, and several other NW mammarenaviruses mentioned in this study. Moreover, we quantified GP incorporation in our PV preparations by calculating the ratio of TAMV GP (HA-tagged) and the VSV matrix protein (VSV-M) by immunoblotting. We found that TAMV-FL PV contained more GP than TAMV-Ref PV (S2A Fig). To further evaluate the impact of the amount of GP incorporated into PV preparations, we produced TAMV-FL PVs using ranged amounts of GP (S2B Fig). We found that, although the amount of GP in PV preparations affects PV titers (S2C Fig), infections performed with inoculums normalized with the detected amount of GP results in comparable PV infectivity (S2D Fig). These results indicate that the amount of GP in our PV does not affect infectivity and further validate the use of VSV-based PVs to compare GP-mediated entry. Furthermore, aiming to perform faithful comparisons between TAMV-Ref and TAMV-FL, minimizing any possible bias of previous virus passage history, for our TAMV-Ref PV production, we in vitro synthetized TAMV-Ref GP from publicly available sequence in GenBank NC_010701.1.

We first performed a siRNA knock down of 293T endogenous hTfR1 expression (S3A Fig) and investigated the consequences for GP-mediated cell entry of JUNV, TAMV-Ref and TAMV-FL. Entry of JUNV, which in addition to its rodent host TfR1, Calomys musculinus, can utilize hTfR1 as an entry receptor [28], is significantly inhibited by hTfR1 silencing (Fig 2A). Intriguingly, TAMV-FL entry, but not TAMV-Ref, GP-mediated entry is significantly reduced in cells with decreased hTfR1 expression (Fig 2A). To confirm these results, we performed infection assays in airway epithelial A549 cells, which are suitable cellular systems to address possible human infection by inhalation of virus-containing aerosols [16–19], and in hTfR1-partially-depleted cells. To this aim, we used CRISPR/Cas9 to generate A549 cells with a knockdown for hTfR1. Due to observed significant reduction of cell growth rate and survival that could bias our results, we generated A549 cells in which hTfR1 expression was reduced to residual levels (A549/hTfR1 KD) but not completely abrogated, which behaved as parental A549 cells (S3B Fig). Using A549/hTfR1 KD cells, we confirmed that TAMV-FL, compared to TAMV-Ref, is better able to utilize hTfR1 for cell entry (Fig 2B). To validate that the observed hTfR1 dependence was a consequence of specific interactions between the viral GP and hTfR1, we next examined the effect of a monoclonal blocking antibody against hTfR1 on JUNV and TAMV cell entry into 293T and A549 cells, which express high levels of hTfR1 (S3C and S3D Fig). In the presence of the anti-hTfR1 blocking antibody [23,28], the relative entry of JUNV in 293T and A549 cells is inhibited 95% and 82.4%, validating our experimental setup. Confirming previous results, TAMV-FL is inhibited by the anti-hTfR1 antibody more efficiently than TAMV-Ref (Fig 2C and 2D). We next examined the ability of TAMV-Ref and TAMV-FL to use the TfR1 orthologue of the TAMV natural reservoir, Sigmodon hispidus (ShTfR1), using 293T cells expressing ShTfR1 (S3E Fig). Whilst JUNV GP-mediated infectivity increase in ShTfR1 expressing cells is negligible, consistent with its inability to use this receptor orthologue [15,21], TAMV-Ref and TAMV-FL entry is increased 4.6- and 9.6-fold, respectively in ShTfR1-transfected 293T cells compared to parental cells (Fig 2E). These data are consistent with the ability of both TAMV-Ref and TAMV-FL to use the TfR1 of the native host species.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Transferrin receptor-1 (human and Sigmodon hispidus orthologues) dependence of TAMV-FL.

(A) Relative entry of NW arenavirus in siRNA-reduced hTfR1 expression in 293T cells. hTfR1-targeted or non-targeted siRNA siRNAs were transfected to transiently reduce hTfR1 expression or used as reference control of infection (S3A Fig). Error bars represent standard deviations (n = 5). (B) Relative entry of NW arenavirus in CRISPR/Cas9-reduced hTfR1 expressing A549 cells. (A549/hTfR1 KD). hTfR1 expression in un-transduced (A549) and A549/hTfR1 KD cells was monitored by FACS (S3B Fig). Error bars represent standard deviations (n = 4). (C) Relative infection of NW mammarenavirus in 293T cells in presence of anti-hTfR1 blocking antibody. Reference values were obtained from infections performed in presence of identical concentrations of isotype control. Error bars represent standard deviations (n = 12). (D) Relative infection of NW mammarenavirus in A549 cells in presence of anti-hTfR1 blocking antibody. Reference values were obtained from infections performed in presence of identical concentrations of isotype control. Error bars represent standard deviations (n = 3). (E) Relative entry of NW mammarenaviruses in Sigmodon hispidus (Sh)TfR1 transiently expressing 293T cells. ShTfR1 surface expression was confirmed by FACS analysis at the moment of the infection. (S3E Fig). Error bars represent standard deviations (n = 4). Asterisks in all panels denote statistical significance in ANOVA test. (n.s.: p>0.05; *: p<0.05; **: p<0.01).

https://doi.org/10.1371/journal.pntd.0009004.g002

Isolation of replicating TAMV-FL from tick-derived samples

Despite the proven ability of the TAMV-FL GP to support viral entry, it was unclear if the TAMV-FL sequences identified in A. americanum ticks represented a replicating virus. A major challenge for the isolation of TAMV-FL was the vast excess of replicating TCRV present in our samples. However, TAMV is capable of exerting an unusually broad non-reciprocal superinfection exclusion of other NW arenaviruses in macaque Vero cells, preventing productive secondary infections of phylogenetically distant species like TCRV [43]. Therefore, we hypothesized that, if non-reciprocal superinfection exclusion by TAMV is not a cell-type dependent trait, TAMV-FL could outcompete excess of TCRV during serial passages. To this end, we used A549 cells, which are known to be highly permissive for NW mammarenaviruses and have been extensively used in experimental studies with mammarenaviruses [60–63]. While TCRV is known to be sensitive to type I interferon (IFN-I) [63,64], sensitivity of TAMV-FL to IFN-I is unknown. Therefore, we reasoned that an IFN-induced antiviral state could favor TAMV to outcompete TCRV during serial passages. Briefly, A549 monolayers were left untreated or IFN-pretreated and were initially infected with the TCS from the tick-derived sample passaged three times in VeroE6 cells. This inoculum contained predominantly TCRV and 0.58% of TAMV-FL (S1 Table). To allow for multiple rounds of virus infection during each passage, cells were infected at low TCRV multiplicity of infection (MOI; 0.05 PFU/cell) and TCS were collected either after 2 (short passages) or 5 days (long passages). Infectious TCRV titers from collected TCS were determined by immunofocus assay (IFA) using monoclonal antibody MA03, which recognizes TCRV NP but not TAMV NP [65]. Due to lack of highly specific TAMV antibodies, it was impossible to monitor TAMV infectious viral titers along the passages. During serial passaging, we observe a progressive drop in infectious TCRV titers, resulting in undetectable levels between passages 2 and 6 (Short passages format) or 2 and 4 (Long passages format), corresponding to a total of 96–288 hpi and to 240–480 hpi, respectively (Fig 3A). To examine if the loss of infectivity of TCRV in TCS correlates with the presence of TAMV-FL, we infected fresh VeroE6 cells with undiluted TCS from passage 6 (IFN-pretreated cells, short format passage) and 48 hpi, fixed the cells and examined them by IFA. In order to distinguish between TCRV and TAMV infected cells, we used the aforementioned MA03 antibody and the broadly cross-reactive IC06 antibody, which recognizes both TCRV and TAMV NP [65]. Consistent with undetectable TCRV titers in TCS, cells infected with TCS of passage 6 show no detectable staining for TCRV NP above background (Fig 3B). In contrast, probing with the cross-reactive IC06 antibody reveals a cytosolic staining pattern characteristic of mammarenavirus NP expression, demonstrating the presence of a replicating NW arenavirus other than TCRV (Fig 3B). To verify the genetic identity of the selected virus, we examined TCS from passage 5 in IFN-pretreated cells (short and long passage format; TAMV-FLp5s and TAMV-FLp5l respectively) by NGS as depicted in Fig 1A. The obtained sequence data confirms the extinction of TCRV and selection of TAMV-FL that represents 100% of all arenavirus reads in both samples, TAMV-FLp5s and TAMV-FLp5l (Fig 3C), demonstrating the presence of replicating TAMV-FL in our tick-derived samples. The NGS analysis allowed us to assemble the entire TAMV-FL genome in both samples. Consistent with mammarenaviruses quasispecies dynamics [14,30,31], the alignment of TAMV-FLp5s and TAMV-FLp5l (Fig 3D) against TAMV-FL genome reveals 43 single nucleotide variants (SNVs) in both segments of TAMV-FL genome (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. TAMV isolation from tick-derived sample.

(A) TCRV viral titers along short (2 days per passage) and long (5 days per passage) passages in A549 cells pretreated or not with 100 IU/ml of IFN. Grey dotted line represent the detection limit of the experiment. Error bars represent standard deviations (n = 4). Biological replicas (A, B and C) are shown independently. (B) Identification of TAMV by IFA in VeroE6 infected cells (48h after infection) with TCS from passage 6 (short format) in presence of IFN (replica A). MA06 antibody recognizes TCRV NP and IC06 antibody recognizes TCRV and TAMV NPs. Scale bars represent 125 μm. (C) Relative abundance of TCRV and TAMV-FL reads obtained by NGS analysis of tick-derived original sample (passage 0, TCRV/TAMV-FL) and of TCS from passage 5 in presence of IFN of short (TAMV-FLp5s) or long (TAMV-FLp5l) passages. (D) TAMV-FL, TAMV-FLp5s and TAMV-FLp5l read coverage (Grey area) and relative abundance of SNV (red dots) of aligned against TAMV-FL sequence.

https://doi.org/10.1371/journal.pntd.0009004.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Quasispecies composition of TAMV-FLp0, TAMV-FLp5FLp5s and TAMV-FLp5FLp5l.

α indicates SNV identifier, consisting of the viral gene, the observed substitution, and position of the substitution. β indicates relative abundance for each particular SNV. δ indicates the amino acid change resulting from the SNV (empty fields denote synonymous or insertion/deletion mutations). * stands for p<10⁻⁶ and ** for p<10⁻¹². SNVs GP540A and GP553A are highlighted in bold.

https://doi.org/10.1371/journal.pntd.0009004.t002

Acquisition of non-synonymous mutations upon short-term TAMV-FL passage

At the amino acid level, the 43 SNVs found in TAMV-FL5s and TAMV-FLp5l upon serial passage result in 18 amino acid substitutions (Table 2).

In TAMV-FLp5s quasispecies, the frequency of the GP1 variants N151K and D156N increase from 0.11% and 0% to 18.75% and 20.03%, respectively (p<10⁻¹² in both cases), with respect to TAMV-FL. In TAMV-FLp5l, the substitution N151K enriches to 32.39% (p<10⁻¹²), while D156N only accumulates to 0.33% (p>10⁻⁶) (Table 2). While 569 and 623 reads, out of a total of 2997, harbor T540A (N151K) and G553A (D156N) substitutions, respectively, only 2 reads contain both mutations (p<0.01, χ² value = 6.3x10^-12), suggestive of a strong evolutionary disadvantage against the TAMV-FL double mutant T540A/G553A (N151K/D156N). The remaining non-synonymous SNVs show only limited changes in relative abundance during our serial passage experiment and were not further analyzed.

Mutations acquired in TAMV-FLp5s/l GP lie outside the predicted transferrin receptor 1 -binding interface

Recent structures of the attachment-mediating GP1 glycoprotein from the clade D WWAV have revealed structural diversity within the GP1 of TfR1-tropic viruses [25,66], principally within GP1 loop regions that are likely involved in TfR1 recognition [26]. Such structural variation across NW mammarenaviruses with shared receptor-tropism may indicate diverse modes of TfR1 recognition [67]. WWAV-AV96, which was associated with fatal human illness [20], utilizes hTfR1 for viral entry [25,51]. Interestingly, sequence alignment of WWAV and TAMV strains reveals common D to N substitutions at positions 156 and 154 of the GPs in TAMV-FL and WWAV-AV96, respectively (Fig 4A). We therefore thought to assess the potential functional effects of the TAMV-FL GP1 N151K and D156N substitutions, enriched during serial passaging of TAMV-FL in human cells. To that end, we performed structure-based mapping, utilizing the crystal structure of the closely related WWAV GP1 (Protein Data Bank [PDB] code 6HJ5) as a proxy. Our model localizes the N151K and D156N substitutions to the helical face of the molecule, distal to the putative TfR1-binding interface [26] (Fig 4B). To further assess the spatial relationships of these substitutions in the context of mature GP, a model of the trimeric clade D NW mammarenaviral GP was constructed based upon superposition of WWAV GP1 onto the structure of the LASV GP ectodomain (PDB: 5VK2), as has been proposed previously for MACV GP [7]. Our model places both amino acid positions, N151K and D156N, near the trimeric axis of the spike, in a sterically constrained environment that would likely be inaccessible to large protein ligands, such as TfR1 (Fig 4C and 4D). Furthermore, superposition of the C-terminal region of LASV GP1 (E228–S255^LASV), which is truncated in all available isolated NW arenavirus GP1 structures, suggests that both N151K and D156N substitutions may be wholly or partially buried in the context of full-length GP1 within mature GP. Given the spatial distinction from the putative TfR1-binding site and their likely position near the sterically constrained trimeric axis of GP, the N151K and D156N substitutions are unlikely to modulate TfR1 utilization by directly affecting the GP1-TfR1 interface. We thus decided to conduct functional analyses in order to clarify the functional effects of these mutations.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Structure-based mapping of acquired TAMV mutations.

(A) Amino acid sequence alignment of NW clade D GP1s, numbered according to WWAV-Ref. Polymorphisms common to TAMV and WWAV are highlighted in yellow and numbered in accordance with TAMV-FL numbering. Secondary structural elements of WWAV GP1 are illustrated above the sequences. Alpha (α) and 3₁₀ (η) helices are shown as coils, and β-sheets (β) as arrows. Conserved residues are shown in red boxes and semi-conserved residues are shown in red font within a blue box. Alignments were displayed using ESpript (http://espript.ibcp.frhttp://espript.ibcp.fr) [109]. (B) Amino acid substitutions (yellow) acquired by TAMV_FL map to the putatively trimeric axis-facing helical region of GP1, distal to the expected TfR1 receptor binding site (red). Features are mapped onto the structure of WWAV GP1. (C) Localization of TAMV mutants in the context of a composite model of trimeric GP, excluding SSP and transmembrane domains. The GP2 subunit from the trimeric LASV GP structure is shown in light grey, with GP1 from WWAV (colored as in panel B) in place of LASV GP1. The C-terminal region of LASV GP1, which is not present in the structure of WWAV GP1, is shown overlaid in blue. (D) View of the trimeric axis-facing side of a composite GP1-GP2 protomer, colored as in panel B.

https://doi.org/10.1371/journal.pntd.0009004.g004

Single N151K and D156N substitutions increase hTfR1 entry dependence of TAMV-FL

To assess the possible impact of N151K and D156N changes on hTfR1 usage, we introduced the aforementioned substitutions into TAMV-FL GP1 and produced the corresponding PVs (TAMV-FL N151K, TAMV-FL D156N and TAMV-FL DM, respectively) as well as WWAV-Ref and WWAV-AV96 PVs. Quantification of GP incorporated into the PVs indicates that TAMV-FL N151K, TAMV-FL D156N and TAMV-FL DM PVs contained less GP than the TAMV-FL (S4 Fig). Next, we infected A549 and A549/hTfR1 KD cells and found that the introduction of either substitutions, N151K or D156N, increases hTfR1 usage compared to the parental TAMV-FL, whilst the double mutant exhibits decreased hTfR1 usage, comparable to TAMV-FL. This suggests that these mutations cancel out their individual functional advantage when combined (Fig 5A). To confirm these results, we infected 293T and A549 cells in presence of an anti-hTfR1 blocking antibody (Fig 5B and 5C, respectively), and observed that either mutation, N151K or D156N, increases hTfR1 usage of TAMV-FL GP, whereas this effect was abrogated if both mutations are simultaneously present in the same genome. These results also confirmed that WWAV-AV96 utilizes hTfR1 more efficiently than WWAV-Ref (Fig 5B and 5C) [25,51].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Transferrin receptor-1 (human and Sigmondon hispidus orthologues) use of TAMV-FL mutants.

(A) Relative entry of NW mammarenavirus in CRISPR/Cas9-reduced hTfR1 expressing A549 cells (A549/hTfR1 KD). hTfR1 expression in non-transduced (A549) and hTfR1hA549/TfR1 KD cells was monitored by FACS (S3B Fig). Error bars represent standard deviations (n = 4). Antibody perturbation against endogenous hTfR1 with specific blocking antibody in (B) 293T cells and (C) A549 in infections of TAMV-FL and TAMV-FL mutants. Reference values were obtained from infections performed in presence of identical concentrations of isotype control. Error bars represent standard deviations (panel B, n = 12; panel C, n = 3) (D) Antibody perturbation experiments, with anti-hTfR1 specific antibody or isotype control, in infections with TAMV-Ref and TAMV-Ref mutants in A549 cells. Error bars represent standard deviations (n = 3). (E) Relative entry of NW mammarenavirus on Sigmodon hispidus (Sh)TfR1 transiently expressing 293T cells. ShTfR1 surface expression was monitored by FACS analysis at the moment of the infection (S3E Fig). Error bars represent standard deviations (n = 4). Asterisks in all panels denote statistical significance in ANOVA test (ns: p>0.05; *: p<0.05; **: p<0.01).

https://doi.org/10.1371/journal.pntd.0009004.g005

We then investigated whether the GPC genetic backbone is important in determining the effect of the N151K and D156N substitutions upon hTfR1 binding. To this end, we introduced either N151K, D156N, or both mutations, in TAMV-Ref GPC backbone, produced the corresponding PVs (TAMV-Ref N151K, TAMV-Ref D156N and TAMV-Ref DM, respectively) and infected A549 cells in presence of the anti-hTfR1 blocking antibody. We found that the mutations, either alone or in combination, had a negligible impact on hTfR1 dependence in the TAMV-Ref genetic backbone (Fig 5D).

Furthermore, we investigated the ability of TAMV-FL variants to bind ShTfR1. We infected 293T cells transiently expressing ShTfR1 (S3E Fig) with parental TAMV-FL, and TAMV-FL mutant PVs. Either individual substitutions, N151K or D156N, reduces the ability of GP1 to bind ShTfR1, but the changes antagonize each other when present together (TAMV-FL DM), leading to a ShTfR1 usage comparable to that of the parental TAMV-FL (Fig 5E).

Substitutions N151K and D156N increase affinity for heparan sulfate proteoglycans

Efficient heparan sulfate binding has been reported in several viruses [35,36,40,68,69]. Amino acid changes N151K and D156N lead to charge alterations in the viral GP, and therefore may modulate the HSPG binding and promote viral entry. Therefore, we examined the infectivity of TAMV-Ref, TAMV-FL, TAMV-FL N151K, TAMV-FL D156N and TAMV-FL DM PVs, along with WWAV-Ref and WWAV-AV96 PVs, in presence of increasing concentrations of competing heparin (Fig 6A and S5A Fig). As reference, we used Ebola virus (EBOV) PV, which is known to bind HSPG with high affinity [40]. Entry of both TAMV-FL mutant PVs, N151K and D156N, is more strongly inhibited by heparin than entry of the parental TAMV-FL, indicating increased ability to bind HSPG. Although virtually absent in any natural TAMV-FLp5 population analyzed, the double mutant N151K/D156N TAMV-FL PV shows the highest binding capacity to HSPG among the viruses tested. Interestingly, WWAV-AV96 is also more inhibited in the presence of heparin than WWAV-Ref (Fig 6A). To validate these results, we also performed PV infections in A549 cells treated with heparinase III, which cleaves and removes the polysaccharide chains from the HSPG protein component (Fig 6B and S5B Fig). The results confirm an increased HSPG binding capacity of TAMV-FL N151K, D156N and DM variants as well as WWAV-AV96 compared to TAMV-FL and WWAV-Ref, respectively.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. HSPG binding modulation by N151K and D156N substitutions.

(A) Relative entry of TAMV-FL mutants in infections performed in presence of increasing concentrations of heparin. Error bars represent standard deviations (n = 4). (B) Relative entry of PV infections performed in heparinase III-A549-treated cells. Error bars represent standard deviations (n = 4). (C) Relative entry of TAMV-Ref mutants in infections performed in presence of increasing concentrations of heparin. Error bars represent standard deviations (n = 4). Asterisks in all panels denote statistical significance in ANOVA test (ns: p>0.05; *: p<0.05; **: p<0.01).

https://doi.org/10.1371/journal.pntd.0009004.g006

In order to evaluate the effect of the genetic backbone on the HSPG binding, we used TAMV-Ref and TAMV-Ref mutants to infect A549 cells in the presence of different concentrations of heparin, and monitored relative entry (Fig 6C). The results show that TAMV-Ref N151K, TAMV-Ref D156N and TAMV-Ref DM exhibit increased HSPG dependence, comparable to the analogous TAMV-FL variants.

N151K and D156N substitutions delay GP-mediated endosomal escape

After receptor-mediated endocytosis, NW mammarenaviruses are delivered to late endosomes [70]. The transition from early to late endosomal compartments is accompanied by progressive acidification of the endosomal lumen, resulting in a pH gradient that serves as guidance cue for membrane fusion triggering (Fig 7A). The transmembrane GP2 undergoes low pH-induced conformational changes that include transition from a metastable pre-fusion state to an energetically favorable post-fusion state with a six-helix bundle architecture reminiscent to other class I viral fusion protein [71,72]. Given that N151K and D156N map to the sterically constrained axis of GP (Fig 4C), we hypothesized that the substitutions may affect the overall conformational stability of GP and hence modulate pH-dependent GP2-mediated fusion (Fig 4C and 4D). To interrogate this hypothesis, we investigated the effects of N151K and D156N on the pH-triggered GP-mediated fusion activity of TAMV-FL in the context of productive PV entry into A549 cells. Ammonium chloride rapidly depletes the endosomal pH gradient in a concentration-dependent manner, without causing overall cytotoxicity and preventing pH-mediated viral exit from endosomal compartments [73,74]. Thus, sensitivity to ammonium chloride is inversely proportional to the viral fusion pH, with late fusing viruses showing higher sensitivity. To evaluate pH-dependent fusogenic activity of TAMV-FL mutants, we raised the endosomal pH by adding increasing concentrations of ammonium chloride to cells and monitored productive PV entry. To validate our system, we established dose-response curves of ammonium chloride for VSV, which is known to escape from early endosomes at pH >6 [75] and JUNV, which escapes from late endosomes at pH <5.5 [76]. As expected, based on the lower fusion pH, JUNV PV shows higher sensitivity to ammonium chloride than VSV PV (IC₅₀ values 1.07 mM and 2.4 mM, respectively) (Fig 7B and Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Endosomal escape of clade D NW mammarenavirus.

(A) Schematic virus internalization, compartment acidification, and pH-dependent endosomal escape. (B) Ammonium chloride sensitivity of TAMV-FL mutants and WWAV strains. PV infections were performed in presence titrated ammonium chloride concentrations (0–5 mM). Error bars represent standard deviations (n = 6). (C) Endosomal escape of TAMV-FL mutants and WWAV strains. At indicated time points, endosomal exit was prevented with 20 mM of ammonium chloride. Error bars represent standard deviations (n = 3). (D) Endosomal escape of TAMV-Ref mutants. Error bars represent standard deviations (n = 7).

https://doi.org/10.1371/journal.pntd.0009004.g007

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Ammonium chloride sensitivity (IC₅₀) of PV of VSV, JUNV and clade D NW arenaviruses.

Standard deviations are indicated between parentheses (n = 6). Asterisks (**) indicate statistical significance (p<0.01) in ANOVA test when compared to TAMV-Ref (For TAMV-FL), TAMV-FL (For TAMV-FL mutants) or WWAV-Ref (For WWAV-AV96).

https://doi.org/10.1371/journal.pntd.0009004.t003

Ammonium chloride dose-response curves for TAMV-Ref and TAMV-FL PV reveal similar IC₅₀ values of 1.34 mM and 1.54 mM, respectively (Fig 7B and Table 3). Interestingly, introduction of either the substitution N151K or D156N into TAMV-FL GP1 significantly increases ammonium chloride sensitivity, resulting in IC₅₀ values of 1.14 mM and 0.99 mM, respectively (Fig 7B and Table 3). The presence of both changes (TAMV-FL DM) yields the lowest IC₅₀ (0.62 mM) of all tested PVs. WWAV-AV96 PV also shows reduced ammonium chloride sensitivity and lower IC₅₀ than WWAV-Ref PV (1.04 mM and 1.65 mM, respectively). Although our ammonium chloride titration assay is semi-quantitative in nature, it nonetheless evidences the ability of the N151K and D156N substitutions to modulate fusion pH in vitro.

The timed gradual decrease in pH delimits the endosomal conditions available for viral escape and release into the cytoplasm (Fig 7A). Hence, under the assumption of identical participating host factors, the timing of sensitivity to ammonium chloride is a proxy for the timing of endosomal escape. In order to determine the half-escape time of selected clade D NW mammarenaviruses, we added ammonium chloride to TCS at different time points after PV infection, preventing any pH-dependent exit from endosomal compartments. VSV and JUNV PV exit from endosomal compartments approximately 15.5 and 20.5 minutes after infection, respectively (Fig 7C and Table 4). TAMV-FL N151K and TAMV-FL D156N exhibit delayed endosomal escape kinetics and higher half-escape times (23.6 and 25.4 minutes) compared to the parental TAMV-FL, which has a half-escape time of 20.3 minutes. (Fig 7C and Table 4) Endosomal escape of TAMV-FL DM is delayed to a half-escape time of approximately 27 minutes after infection. Intriguingly, despite the absence of significant differences in ammonium chloride sensitivity, TAMV-Ref escapes earlier than TAMV-FL (Fig 7B and 7C, and Tables 3 and 4). WWAV-AV96 shows no delay in half-escape time when compared to WWAV-Ref, whereas the differences in ammonium chloride sensitivity are substantial (Fig 7B and 7C, and Tables 3 and 4). These apparent discrepancies suggest the possible participation of additional host factors during viral entry such as intracellular receptors that have been described for other emerging viruses, including the OW mammarenaviruses LASV and Lujo virus [52,53,55,77]. In summary, these results indicate that N151K and D156N substitutions delay viral release, lead to release from later, and more acidified endosomal compartments.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Half-escape time of PV of VSV, JUNV and clade D NW arenaviruses.

Standard deviations are indicated between parentheses (n = 3). Asterisks (**) indicate statistical significance (p<0.01) in ANOVA test when compared to TAMV-FL.

https://doi.org/10.1371/journal.pntd.0009004.t004

Next, we examined the impact of N151K and D156N on endosomal escape in the TAMV-Ref GPC genetic backbone. When introduced in TAMV-Ref GPC, N151K and D156N substitutions delay the endosomal escape by 1.7 and 0.8 minutes, respectively, while when introduced in TAMV-FL GP1, N151K and D156N substitutions delay the endosomal escape by 3.3 and 5.1 minutes, respectively (Fig 7C and 7D, and Tables 4 and 5). Incorporation of both mutations results in comparable endosomal escape delay in both TAMV-FL and TAMV-Ref backbones (6.6 and 6.3 minutes, respectively). Taken together, the TAMV-FL GP1 substitutions, N151K and D156N, not only impact attachment factor usage, but also fusion kinetics and subcellular fusion compartments of the virus, suggesting that these mutations may be adaptive.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Half-escape time of PV of TAMV-Ref and TAMV-Ref mutants.

Standard deviations are indicated between parentheses (n = 7). Asterisks (**) indicate statistical significance (p<0.01) in ANOVA test when compared to TAMV-Ref (**: p<0.01).

https://doi.org/10.1371/journal.pntd.0009004.t005

Antibodies, plasmids and reagents

Mouse monoclonal antibodies (mAb) MA03-BE06 and IC06-BE10 [65] were obtained from BEI Resources (Manassas, VA). Alexa Fluor 488 F(ab′)2 fragment of goat anti-mouse IgG and Alexa Fluor 594 goat anti-mouse IgG were purchased from Life Technologies (Carlsbad, CA). Anti hTfR1 (CD71, Ref#555534) was obtained from BD Biosciences and isotype control (Ref#11711) were purchased from R&D systems. Anti HA (High Affinity) rat monoclonal IgG1 antibody (Cat. No. 11 867 423 001) and recombinant human IFN (Interferon-αA/D human, Cat# I4401) were purchased from Sigma-Aldrich (St. Louis, Missouri, USA). Plasmid lentiCRISPR v2 was a gift from Feng Zhang (Addgene plasmid # 52961; http://n2t.net/addgene:52961). Lentiviral packaging plasmid pCMV-VSV-G was a gift from Bob Weinberg (Addgene plasmid # 8454; http://n2t.net/addgene:8454) and pLJM1-EGFP was a gift from David Sabatini (Addgene plasmid # 19319; http://n2t.net/addgene:19319). HA-tagged GPCs of TAMV-Ref (NC_010701.1), TAMV-FL (MK500937.1), and WWAV-AV96 (EU 123330.1) were synthetized in vitro (Genscript, Piscataway, NJ, USA). All TAMV GPs were cloned into pCAGGS plasmid, between BglII and XhoI restriction sites and HA-tagged. HA-tagged ShTfR1 was in vitro synthetized by Genscript (Piscataway, NJ, USA) and subcloned into pcDNA3.1(+) vector.

Viruses and cells

Human lung adenocarcinoma epithelial cells (A549), human embryonic kidney cells (293T) and African green monkey kidney epithelial cells (VeroE6) were maintained in Dulbecco’s modified Eagle medium containing high glucose (4.5 mg/l) and GlutaMAX (DMEM, Gibco BRL) supplemented with 10% (vol/vol) fetal calf serum (FCS) at 37°C and 5% (vol/vol) CO₂. Baby hamster kidney cells (BHK)-21 were maintained in DMEM supplemented with 5% (vol/vol) FCS and non-essential amino acids (Gibco BRL) at 37°C and 5% (vol/vol) CO₂. All cells were regularly tested for mycoplasma with MycoAlert Mycoplasma detection kit (Lonza). The tick-derived isolate was kindly provided by Dr. Katherine Sayler, and was the very same sample previously reported to contain TCRV [41]. Tick homogenates were used to infect VeroE6 cells, and passaged 3 times. Supernatants were cleared by brief centrifugation (1500 rpm, 4°C, 5 min) [41]. Viral stocks were produced by infecting BHK-21 cells with passage 3 (on VeroE6 cells) of tick-derived isolate, collecting the conditioned TCS after 4 days after infection and cleared by centrifugation at 1500 rpm for 5 minutes at 4°C. Virus stocks were mixed 1:1 with a sterile solution of 140 g/l polyethylene glycol (PEG)-8000 in PBS and incubated overnight in rocking station at 4°C. Samples were centrifuged at 8000 x g for 1h at 4°C, TCS were discarded and precipitated material was resuspended in supplemented DMEM. PEG-precipitated samples were then layered on 30% (wt/vol) sucrose cushion and centrifuged at 37900 x g for 2 h at 4°C in an Optima XPN-80 ultracentrifuge (Beckman Coulter) equipped with a SW-55 Ti rotor. After ultracentrifugation, pellets were resuspended in complete DMEM and stored at -80°C.

RNA extraction and library preparation

For RNA extraction, samples were layered on 30% (wt/vol) sucrose cushion and centrifuged at 100000 x g for 2 h at 4°C in an Optima XPN-80 ultracentrifuge (Beckman Coulter) with a SW-55 Ti rotor. After ultracentrifugation, pellets were resuspended in nucleases buffer (Tris-HCl, pH 7.5 40mM, MgCl₂ 6mM, NaCl 10mM, CaCl₂ 1mM) and treated for 30 minutes at room temperature with nucleases cocktail to remove non-encapsidated nucleic acids. Nucleases cocktail included DNase I (Catalog number EN0521) and RNAse A (Catalog number EN0531A) from Thermo Fisher Scientific (Waltham, Massachusetts, USA) and Nuclease S7 from Sigma-Aldrich (St. Louis, Missouri, USA) (Catalog number N5386). Reaction was stopped by addition of EDTA (28 mM) and EGTA (2 mM). RNA was then extracted with TRIzol reagent (Ambion) followed by ethanol precipitation and resuspended in 20 μl of nuclease-free water. cDNA libraries were prepared with TruSeq Stranded mRNA LT Sample Prep Kit or Kit and following manufacturer’s instructions. Briefly, RNA was fragmented and first and second cDNA strands were synthetized. Then, 3’ ends of cDNA were adenylated and adaptors were incorporated. Samples were then run in a MiSeq instrument (2x250 reads) (Illumina).

NGS analysis

Raw reads were first quality filtered using PRINSEQ v0.20.4 [94]. A minimum average quality of 25 and a minimum length of 120 nt were required. Only paired reads were considered for further analysis. The taxonomy binning of the reads was carried out using Kraken v1.0 [95] using the pre-built DustMasked MiniKraken DB 8GB (built on 18/10/17). Spades v3.12.0 [96] was used for contig assembly (using—careful option). For contig classification, contigs were aligned using Blastn against the reference viral genomes NCBI database (downloaded on 07/01/19), and only those hits of at least 500 bp with and e-value < 1x10^-5 were considered. Quality filtered reads were aligned against reference viral genomes (TCRV and TAMV) o against new sequence genomes/contigs using Bowtie2 v2.3.4.3 [97] using “local” mode. Alignment pileups were built using Samtools v 0.1.19 [98] (minimum required quality 20). Then, pileups were used for SNVs detected using VarScan v2.4.0 [99] (minimum average quality 20, minimum coverage 20 and strand-filtering 80). Sequence conservation between TAMV strains was calculated using mVista program (http://genome.lbl.gov/vista/mvista). Raw reads where submitted to European Nucleotide Archive (ENA) under project accession PRJEB31100.

Pseudotyped virus production

The pseudotype viral system was based on the recombinant VSVΔG-EGFP/Luc (Indiana strain) vector in which the glycoprotein gene (G) had been deleted and replaced with genes encoding green fluorescent protein (EGFP) and luciferase (Luc) [100]. To produce VSV-based pseudoviruses (PV) decorated with viral GP of interest, we performed as described in [101]. Briefly, 293T cells were plated in 10cm dishes and transfected with plasmids encoding for selected viral glycoproteins using JetPrime transfection kit (PolyPlus) and followed manufacturer’s instructions. After 36 hours post transfection, cells were infected with VSVΔG-EGFP/Luc at MOI of 3 PFU/cell and incubated at 37°C and 5% (vol/vol) CO₂ for 90 minutes. After incubation, fresh supplemented DMEM was added with neutralizing anti-VSV monoclonal antibody I-1 (conditioned hybridoma supernatant diluted 1:100). After 16h post infection, conditioned TCS were collected, cleared by centrifugation at 1500 rpm for 5 minutes at 4°C, mixed 1:1 with a sterile solution of 140 g/l PEG-8000 in PBS, and incubated overnight in rocking station. Samples were centrifuged at 8000 x g for 1h at 4°C. TCS were discarded and pellets were resuspended in supplemented DMEM.

Immunoblotting

For GP and VSV-M quantitation, PV preparations were cleared by brief centrifugation (1500 rpm, 5 min, 4°C) and purified by ultracentrifugation in 30% sucrose cushion (in PBS). Pellets were re-suspended in DMEM, and a fraction was lysed with Laemli buffer. Then, proteins were separated by SDS-PAGE and transferred to nitrocellulose membrane. After blocking in 5% (wt/vol) skim milk in PBS, 0.1% (vol/vol) Tween-20 (PBST), membranes were incubated with primary antibody (Anti VSV-M monoclonal antibody from Merck, reference MABF2347, and anti HA high affinity antibody from Merck, reference 11867423001) in 5% (wt/vol) skim milk, PBST, for 2h at room temperature. Then, membranes were washed 3 times with PBST, and HRP-coupled secondary antibodies were applied 1:5000 in PBST (1 h, room temperature). Membranes were then washed with PBST, blots were developed by enhanced chemiluminescence (ECL) using LiteABlot kit (EuroClone). Signals were acquired by ImageQuant LAS 4000Mini (GE Healthcare Lifesciences). Signals were acquired with an ImageQuant LAS 4000Mini (GE Healthcare Lifesciences, Glattbrugg, Switzerland) instrument and quantified with ImageJ Software.

Virus and pseudovirus infections and titrations

Virus infections were performed as described in [64]. Briefly, viral infections were performed by removing cell TCS, adding viral inoculum and incubation during 90 minutes at 37°C and 5% (vol/vol) CO₂. Inoculums were then removed and fresh supplemented DMEM was added. For viral titrations, samples were 10-fold serially diluted in supplemented DMEM and used to infect VeroE6 cells (96 well-format plates). After 16-20h, cells were washed once with PBS, fixed with 2% formaldehyde/PBS and stained for viral NP with MA03 or IC06 antibodies. Positive infectious foci were scored using an EVOS Floid Cell Imaging Station 20X Plan fluorite lens (Thermo Fisher Scientific, Waltham, Massachusetts, USA). PV infections were performed adding 150–300 PFU/well (96 well-format plate) diluted in supplemented DMEM and used as inoculum. Cells and inoculum were incubated for 90 minutes in 37°C and 5% (vol/vol) CO₂. Then, TCS were removed and fresh supplemented DMEM was added to each well. After 16–20 hours, cells were washed once with PBS and fixed with 2% formaldehyde/PBS for EGFP-positive scoring or assayed for luciferase activity. Positive infectious foci were scored using an EVOS Floid Cell Imaging Station 20X Plan fluorite lens (Thermo Fisher Scientific, Waltham, Massachusetts, USA) or luciferase activity measured by ONE-Glo Luciferase Assay System, from Promega (Madison, Wisconsin, USA), as described by the manufacturer.

Antibody perturbation assays

A549 or 293T cells were plated in 96 well-plate format 24h before infection. Cells were preincubated with 200 nM of anti-hTfR1 or isotype control antibodies for 1h at RT. Cells were then infected with 150–300 PV PFU in presence of indicated antibodies. After adsorption, TCS were removed and 20mM NH₄Cl in supplemented DMEM was added to each well. After 16–20 hours, cells were washed once with PBS and fixed with 2% formaldehyde/PBS for EGFP-positive scoring or assayed for luciferase activity as described above.

Fluorescence-activated cell sorting (FACS)

Cells were detached, resuspended, and washed once with supplemented DMEM, washed once with PBS, and fixed with 2% (wt/vol) formaldehyde in PBS for 30 minutes at room temperature. After washing with PBS, cells were permeabilized (for MA03 and IC06 antibodies) or left un-permeabilized (for anti-hTfR1 or anti HA antibodies) for 30 minutes at room temperature with 1% (vol/vol) FCS in PBS with 0.1% (wt/vol) saponin (permeabilization solution) or without (surface staining solution). Primary and secondary antibodies were diluted in permeabilization or surface staining solution and incubated for 1h and 45 minutes, respectively, at room temperature. Cells were washed three times with PBS and analyzed with a FACS Calibur flow cytometer (Becton Dickinson, San Jose, CA).

Immunofocus assay (IFA)

Infected samples were washed once with PBS and fixed with 2% (wt/vol) formaldehyde in PBS for 30 minutes at room temperature. Cells were washed with PBS and permeabilized for 30 minutes at room temperature with 0.1% (wt/vol) saponin and 1% (vol/vol) FCS in PBS (permeabilization solution). Primary and secondary antibodies were diluted in permeabilization solution and incubated for 60 and 45 minutes, respectively, at room temperature. Cells were washed three times with PBS. Positive infectious foci were scored using an EVOS Floid Cell Imaging Station 20X Plan fluorite lens (Thermo Fisher Scientific, Waltham, Massachusetts, USA).

Endosomal escape and ammonium chloride titration

To monitor endosomal escape kinetics, we performed as described in [102]. Briefly, to ensure synchronized infections, A549 cells were cooled down to 4°C and infected with pre-cooled PVs as described above. To allow attachment but not internalization, cells were incubated on ice for 90 minutes and then quickly shifted to 37°C. Upon infection, equal volume of ammonium chloride 40 mM in supplemented DMEM was added to each well (20mM final concentration) at indicated time points. After 16–20 hours, cells were washed once with PBS and fixed with 2% formaldehyde/PBS for EGFP-positive scoring. Relative entry was referenced to values obtained 30 minutes after infection. To determine ammonium chloride sensitivity, PV infections were made in A549 cells. Cells were pre-treated with indicated concentrations of ammonium chloride for 30 minutes before infection. Cells were PV infected as described above, in presence of indicated concentrations of ammonium chloride. After inoculum removal, fresh supplemented DMEM with indicated concentrations of ammonium chloride were added. After 16–20 hours, cells were washed once with PBS and fixed with 2% formaldehyde/PBS for EGFP-positive scoring. Relative entry was referenced to infectivity in absence of ammonium chloride.

Structure-based mapping analysis

Amino acid substitutions identified within the TAMV quasispecies were mapped onto the crystal structure of the GP1 glycoprotein from WWAV (PDB: 6HJ5) using PyMOL (The PyMOL Molecular Graphics System, Version 2.0 Schrödinger, LLC), based upon positions inferred from amino acid sequence alignment calculated using MultAlin [103]. A model for the putative trimeric arrangement of NW GP1s, representative of a mature mammarenavirus GP ectodomain, was generated by superposition of WWAV GP1 onto the structure of the LASV GP ectodomain (PDB: 5VK2), using secondary-structure matching superposition [104] implemented in COOT [105]. The putative TfR1 binding interface on the clade D GP1 surface was predicted using PDBePISA [106], from a model constructed by superimposing WWAV GP1 onto the structure of MACV GP1 in complex with hTfR1 (PDB: 3KAS).

CRISPR/Cas9

A549 cells express hTfR1, which mediates entry of hTfR1-trofic mammarenaviruses [28]. To address the changes in viral entry, we aimed to deplete hTfR1 from A549 cells using CRSIPR/Cas9 editing. To reduce hTfR1 expression, A549 cells were subjected to CRISPR/Cas9 engineering as described in [107,108]. Briefly, single guide RNA (sgRNA) sequences targeting the exon regions of hTfR1 (5′- ATCACTATAGATCCATTCAC-3′). Annealed oligonucleotides were cloned into pLenti CRISPRv2 ccdB by digesting oligonucleotides and vector with BsmBI. In order to transduce A549 cells with the respective gRNA sequence, lentiviral VSV-G pseudotyped particles were produced in the cells. 293T cells were then transfected with plasmid DNA encoding the human immune deficiency virus (HIV) Gag and Pol proteins, VSV-G protein, and the respective pLenti CRISPR construct. Lentiviruses were harvested 48 h after transfection, and A549 cells were transduced with the lentiviruses for 8 h. 37°C and 5% (vol/vol) CO₂. After 48 h after transduction, cells were subjected to puromycin selection (2 μg/ml) for 14 days. Upon puromycin selection, cells were sorted with a Beckman Coulter MoFlo Astrios EQ instrument, discarding those cells that showed higher fluorescence than the isotype control-stained population. Samples were routinely characterized for hTfR1 expression by FACS for every experiment performed.

siRNA transient gene knockdown

To address the impact of transient reduced expression of hTfR1, we used 293T cells, which express hTfR1 and are susceptible to high-efficiency transfection. siRNA TFRC (7037) RNA probe was obtained from Dharmacon/Horizon Discovery (Cambridge, UK). 293T cells were transfected with siRNA probe with RNAiMAX transfection reagent (Thermo scientific, Waltham, Massachusetts, USA), following manufacturer’s instructions. 24 hours upon transfection, cells were plated in poly-lysine-treated 96 plates. 24 hours after plating, cells were infected with PV and tested for hTfR1 expression. After 16–20 hours, cells were washed once with PBS and fixed with 2% formaldehyde/PBS for EGFP-positive scoring or assayed for luciferase activity.

Heparin inhibition and heparinase III treatment

In order to evaluate the binding capacity of selected viral GP, PV infections were performed in presence of heparin or in heparinase III-treated A549 cells. For infections in presence of heparin, PV were diluted in supplemented DMEM with indicated heparin concentrations (Catalog number H9267. Sigma-Aldrich, St. Louis, Missouri, USA) for 1 hour at 4°C. Upon incubation, TCS was removed and PV infections were performed as described above. To remove heparan sulfate proteoglycans from A549 cell surface we performed as described in [69]. Briefly, we plated A549 cells in 96 well-plate format 24 hours before infection. TCS were removed and heparinase III buffer (NaCl 100 mM, Tris-HCl 20mM and CaCl₂ 1.5mM) with 10 IU heparinase III (Catalog number P0737S. New England Biolabs, Ipswich, Massachusetts, USA) was added and incubated for 2 hours at 30°C and 5% (vol/vol) CO₂. After heparinase III treatment, cells were washed there times with PBS and PV infections were performed as described above.

Statistical analysis

Data analyses were performed with Excel software using ANOVA test (Significance 0.05 if not otherwise stated). The half-escape time and IC₅₀ were calculated using Prism software version 8.2.1 (GraphPad Software, San Diego, CA, USA, Version 8.2.1).