Thermostability of FAdV-C4 virions

We examined the thermostability of FAdV-C4 virions in comparison to HAdV-C5 using extrinsic fluorescence of the DNA-intercalating agent YOYO-1, to characterize capsid disruption as a function of temperature [18,44]. The fluorescence emission of YOYO-1 increased with temperature for both specimens, as genomes became exposed to the solvent (Fig 1a). The half-transition temperature indicating the maximum rate of DNA exposure was T_0.5 = 57.3 ± 0.01 ºC for FAdV-C4, 10 °C higher than for HAdV-C5 (T_0.5 = 47.5 ± 0.06 ºC). In comparison, the highest T_0.5 reported in human adenoviruses is that of enteric human adenovirus HAdV-F41, with approximately 51ºC [38]. Hence, similar to the results previously reported for FAdV-A1 (CELO) [31], we found that FAdV-C4 virions are highly thermostable compared to human adenoviruses with either respiratory or enteric tropisms.

Download:

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

Fig 1. Thermostability and general organization of the FAdV-C4 capsid.

(a) Effect of heating on HAdV-C5 and FAdV-C4 virions analysed by extrinsic fluorescence. Circles represent the normalized fluorescence ± STD for HAdV-C5 and FAdV-C4. Continuous lines correspond to the Boltzmann sigmoid fitting to estimate the T_0.5 for each condition tested. (b) Radially coloured surface rendering of the 3.3 Å resolution FAdV-C4 map viewed down the 2-fold icosahedral axis. One facet is identified by a white triangle. Icosahedral symmetry axes are indicated with white symbols: 5-fold (pentagon), 3-fold (triangle) and 2-fold (oval). (c) Molecular models of the major and minor coat proteins traced in the icosahedral FAdV-C4 facet. Top: view from outside the capsid. Bottom: view from inside the capsid. Protein colours as indicated by the legend at the left-hand side. The four hexon trimers in the AU are numbered H1-H4.

https://doi.org/10.1371/journal.ppat.1013553.g001

Absence of genus specific capsid or core proteins in FAdV-C4 virions

We considered the possibility that genus-specific aviadenovirus proteins play a role in capsid stability, as previously reported for mastadenovirus proteins IX and V [13,32–34] and atadenovirus protein LH3 [4,6,7]. To investigate this possibility, we analysed the protein composition of purified FAdV-C4 virions by liquid chromatography- tandem mass spectrometry (LC-MS/MS) (S1 Table). All the expected capsid (hexon, penton base, fibre-1, fibre-2, IIIa, VI, VIII) and core (VII, X) proteins were detected, along with packaging factors IVa2 and L1 52/55 kDa, the terminal protein (TP, with only two copies per virion) and the adenovirus maturation protease (AVP), also packaged together with the genome [2,45]. Analysis of tryptic and non-tryptic peptides supports the universality of the AVP cleavage specificity across the Adenoviridae family (S1 Fig). Although FAdVs have extensive left and right ends in their long genomes accommodating unique open reading frames that are genus- and even type-specific (including ORFs 1B, 20A and 28 in FAdV-C4) [1,35,46,47], we did not detect any peptides corresponding to products of these genes in our assay.

FAdV-C4 overall structure

We obtained high resolution 3D maps for two different FAdV-C4 strains: KR5, and AG234 (S2 Table and S2a and S2b Fig). Model building (S3 Table) was carried out for the KR5 map (Fig 1b and 1c), given that the AG234 map presented a disrupted particle lacking pentons (S2c Fig and S1 Text). The structure shows the general characteristics common to all previously solved adenovirus structures: hexons (major capsid protein) and penton base (vertex protein) arranged in a pseudo T = 25 icosahedral particle, with inner cementing proteins IIIa and VIII on the inner capsid surface (Fig 1b and 1c and S4 Table). S5 Table summarizes the main differences between the FAdV-C4 virion proteins and their counterparts in HAdV-C5. Given the known lack of the gene coding for protein IX in aviadenovirus genomes, it was not surprising that we could not trace this protein in the cryo-EM map. In agreement with the LC-MS/MS data, no densities that could be interpreted as genus specific minor coat proteins were present in the map. These results corroborate the lack of genus specific proteins compensating for absence of proteins IX or V in the FAdV-C4 capsid.

Major coat protein: Hexon

As expected, the general architecture of the FAdV-C4 hexon monomer (937 amino acids vs. 952 in HAdV-C5) is highly conserved compared to other adenoviruses, with two jelly-rolls (P1 and P2, Fig 2a) forming the pseudo-hexagonal base, and flexible loops that contain the hyper variable regions (HVRs) [48] on the surface-exposed towers (Fig 2a).

Download:

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

Fig 2. Hexon and penton structures in FAdV-C4.

(a) Superposition of the HAdV-C5 (PDB ID 6b1t, in white) and FAdV-C4 (coloured by RMSD) hexon monomers. P1 and P2 indicate the two jelly rolls. The horizontal dashed line indicates the equatorial region. (b) Structural superposition of available hexon structures (PDB IDs: 6cgv: HAdV-C5, 5tx1: HAdV-D26, 6yba: HAdV-F41, 2obe: SAdV-25, 3zif: BAdV-3, and 6qi5: LAdV-2). Conserved residues (with < 5Å RMSD compared to the rest of structures) are shown in dark blue. Residues with RMSD higher than 5 Å are represented in white for all hexons, except for the > 5 Å RMSD FAdV-C4 residues, which are represented in red, to be distinguished from the rest. Conserved residues are mainly forming the jelly rolls in the base. Variable regions are mostly located at the top of the hexons, conforming the specific HVRs. Note the unique big insertion of FAdV-C4 (zoom at the right), not present in any of the other hexon structures. (c) Superposition of the HAdV-C5 (PDB ID 6b1t, white) and FAdV-C4 (coloured by RMSD) penton base monomers. The hypervariable and variable loops (HVL, VL and VL’) along with other diff regions are indicated. (d) Top view of the penton base pentamers of HAdV-C5 and FAdV-C4, shown with one monomer rainbow-colored and with the HVL, VL and VL’ indicated, and the rest of monomers in grey tones.

https://doi.org/10.1371/journal.ppat.1013553.g002

Structural superposition of HAdV-C5 and FAdV-C4 hexon monomers revealed seventeen regions of difference (diff) with RMSD > 5 Å (S6 Table and Fig 2a). The N-terminal region of the FAdV-C4 hexon is five amino acids longer (S6 Table, diff 1), a unique feature of aviadenovirus hexons [41] which could have an impact in the interaction with other proteins (see section “Protein networks in FAdV-C4”). We could fully trace the seven HVRs of FAdV-C4 (S4 Table), which as expected differ from the corresponding regions in HAdV-C5 (S3a Fig and S6 Table, diff 2–4 and 6–9). Diff 11, 12 and 16 are located in the equatorial region of the hexon monomer, between the base and the tower regions. In the three-dimensional structure, these equatorial differences come together forming a sort of pincer above the second jelly roll (Fig 2a and S6 Table, “pincer”). The pincer is located close to the position occupied by the outer cementing protein IX in HAdV-C5 [3], suggesting that the differences in this region could be related to the absence of outer cementing proteins in FAdV-C4. Indeed, several residues in FAdV-C4 hexon would clash with protein IX if this protein was present in FAdV-C4 (S3b Fig). Multiple sequence [48] and structural alignments indicate that the pincer is one of the variable regions when comparing all available hexon structures (Fig 2b). HAdV-F41 hexon shows a subtle difference in this area when compared to HAdV-C5, proposed to be involved in the unique conformation of protein IX in the enteric virus [38]. Likewise, LAdV-2 hexon exhibits differences in this region [6], and its genus specific cementing protein LH3 also follows a different conformation when compared to that of protein IX in HAdV-C5. These observations highlight the pincer region as an indicator for evolutionary variability in adenoviruses.

Another difference likely related to the absence of protein IX concerns the structure of HVR2. In HAdV-C5, this loop reaches out from the towers towards the neighbouring hexons (S3c Fig), and interacts with protein IX C-terminal helix bundle in hexons 2 and 4 across the icosahedron edge [3]. HVR2 in FAdV-C4 does not spread out from the tower, but folds back towards the main body of the hexon trimer (S3c Fig). It has previously been proposed that hexon HVR2 and the pincer region jointly modulate the organization of protein IX on the capsid surface [49].

Diff 14 corresponds to a big insertion (14 residues) in FAdV-C4 hexon (S6 Table and Fig 2a). Multiple comparison of available hexon structures shows that this feature is unique for aviadenoviruses (Fig 2b). The big insertion forms part of the loops arising from the P2 jelly roll, but it is not located in the hexon towers as the HVRs. Instead, it contours the equatorial region of the monomer surface and reaches towards the loops arising from the P1 jelly roll, and may be the cause of rearrangements in the neighbourhood (diff 10 and diff 15, Fig 2a). Additionally, the big insertion forms a negatively charged surface patch (S3c Fig). Since all contacts established by the big insertion are intra-monomer (S7 Table), its presence is not expected to impact on the stability of the hexon trimer or the interactions between capsomers in the virion. Finally, diff 13 (S6 Table) is located on the surface of the inner hexon cavity— Tyr 756 in FAdV-C4, Tyr 787 in HAdV-C5, and might be involved in specific interactions with proteins VI and VII [16].

We also analysed the differences between the twelve hexon monomers in the FAdV-C4 asymmetric unit (AU). Regions showing a RMSD >2 Å are located at the trimer base and comprise N-terminal and C-terminal residues (S3d Fig). In other reported AdV structures the hexon N- and C-termini also exhibit variability, related to the hexon monomer position in the AU and its neighbouring proteins [3,6]. Variability among the twelve monomers in the AU at regions such as the HVRs or the jelly roll bases has been observed in other AdV structures [6, 36], but not in FAdV-C4 hexons.

Vertex capsomer: Penton base

Penton base topology is conserved when compared with other AdVs, with an extended N-terminal arm preceding a single jelly roll domain at the pentamer base, and a protruding domain containing variable regions exposed on the capsid surface (Fig 2c) [2]. Nevertheless, structural superposition of HAdV-C5 and FAdV-C4 penton base monomers (571 residues in FAdV-C4 vs. 525 in HAdV-C5) revealed eight diff regions (S8 Table and Fig 2c). The N-terminal arm of FAdV-C4 is longer (~65 amino acids) compared to that of mastadenoviruses (approx. 50 aa) and atadenoviruses (approx. 15 aa) [3,6]. We were able to start tracing the FAdV-C4 penton base at Ala59, but not the upstream residues traced in HAdV-C5 (Asp37-Gly50) and LAdV-2 (Glu2-Gly16). A conspicuous polyproline stretch (residues 11–16, S4a Fig) may impact on the flexibility of the FAdV-C4 N-terminal arm, and therefore its interactions with neighbouring proteins or cell factors during the viral cycle. The variable loop (VL, diff 3) is quite long in FAdV-C4 (17 vs 6 aa in HAdV-C5) and folds back towards the main body of the monomer, augmenting intramolecular interactions (S9 Table), in sharp contrast to the protruding orientation of the VL in HAdV-C5 (Fig 2c and 2d). FAdV-C4 lacks the integrin-binding RGD motif, present in the hypervariable loop (HVL) in HAdV-C5 (diff 4) [50]. The HVL is notably shorter in FAdV-C4 (5 vs 74 aa in HAdV-C5) allowing us to completely trace it. Diff 5 forms another variable loop (VL’) located between the HVL and the VL and three amino acids longer in FAdV-C4 than in HAdV-C5. We hypothesize that the architectural differences in penton base loops VL, HVL and VL’ could be related with the need to attach two fibres to the same penton base [30]. The arrangement of two trimeric fibres bound to the pentameric penton base is one of the most intriguing features of the aviadenovirus capsid, which cannot be elucidated using icosahedrally averaged maps (S1 Text and S4 Fig). Finally, diff 2 and diff 7, located at the base domain, contact the peripentonal hexons and may contribute to the specificity of hexon-penton interactions during assembly [17,38].

Minor coat protein VIII

Based on the mastadenovirus AVP consensus sequence motif [45], protein VIII in FAdV-C4 is predicted to be cleaved at least at four sites during maturation (S1 Fig). We have traced the two largest fragments (residues 1–115 and 174–241) almost in their entirety (S4 Table and Fig 3a). As in all previously solved AdV structures, there are two independent copies of protein VIII per AU: one gluing the GOS (Group of Six: penton base + five surrounding hexons) and the other beneath the GON (Group of Nine: nine hexons in the facet) (Figs 1c, bottom and S5) [3]. Both copies are virtually identical.

Download:

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

Fig 3. Protein VIII divergent structure in FAdV-C4.

(a) Sequence alignment and secondary structure elements of protein VIII in HAdV-C5 (top) and FAdV-C4 (bottom). Squiggles indicate α-helices and arrows indicate β-strands. Sequence numbering corresponds to HAdV-C5. Residues are coloured according to similarity: black (low similarity), red with blue frame (high similarity) and white on red background (strict identity). N- and C-terminal fragments are delimited by the AVP-processed peptides shadowed in grey. An arrowhead below the FAdV-C4 sequence indicates where the C-terminal fragment tracing starts, seven amino acids downstream of the AVP site. Note the divergence in α2/α2’ and that FAdV-C4 α2” and α2”’ are displaced with respect to β3 and β4 in HAdV-C5. (b) Superposition of protein VIII in HAdV-C5 (white, PDB ID: 6b1t) and FAdV-C4 (coloured by RMSD). (c) Rainbow coloured structures of protein VIII in HAdV-C5 (left) and FAdV-C4 (right). The body, neck (dashed box) and head domains are indicated. Dashed gaps corresponding to the AVP cleaved segment (not traced) are indicated with scissors. Note that α2 in FAdV-C4 is shorter than α2 in HAdV-C5, and along with α2’ follows a different path compared to α2 and subsequent coil structure in HAdV-C5 (curved red arrows). β3 and β4 in HAdV-C5 differ from α2” and α2”’ in FAdV-C4 (dashed frame). Helix α2’ in FAdV-C4 is bent by 115 º with respect to α2. (d) Detail of the HAdV-C5 (white) and FAdV-C4 (orange) C-terminal fragment of protein VIII.

https://doi.org/10.1371/journal.ppat.1013553.g003

Superposition of the HAdV-C5 and FAdV-C4 protein VIII structures (247 amino acids in FAdV-C4 vs. 227 in HAdV-C5) reveals several major differences (S10 Table and Fig 3b). The α2 helix in the body domain of HAdV-C5 protein VIII is 22 amino acid long, while its counterpart in FAdV-C4 is shorter (13 residues) and followed by a bend of 115º and another 12-residue α-helix (α2’) (Fig 3c). This bend determines a completely different path for the protein backbone as it leaves the body to enter the neck domain (Fig 3c, red arrows). The neck domain, which contains two antiparallel β-strands forming a narrow sheet in HAdV-C5 (β3 and β4), is broader and formed by two helices (α2” and α2”’) in FAdV-C4 (Fig 3b and 3c, dashed box). The head domain is intercalated between these two helices and contains the gap left by the protease cleavage, separating the N- and C-terminal fragments (Fig 3c, scissors). The structure of the C-terminal fragment is more conserved and only differs in the secondary structure elements of the neck domain (β4 in HAdV-C5 vs α2”’ in FAdV-C4 (Fig 3d). The structure of protein VIII in FAdV-C4 considerably differs from all the previously reported, including those of HAdV-C5, HAdV-D26 and HAdV-F41, and the atadenovirus LAdV-2, where only a small distortion of the neck domain fold was observed [3,6,16,17,36,38]. This conformational divergence impacts on its interactions with neighbouring proteins (see section Results. Protein interaction networks in FAdV-C4). Interestingly, in immature particles of HAdV-C5, the neck domain of protein VIII also presents a small α-helix similar to those found in LAdV-2 and FAdV-C4, which turns into the β4 strand upon AVP cleavage at the head domain (S6a Fig) [37].

Minor coat protein IIIa

There is one copy of IIIa per AU, located beneath the penton (Fig 1c, bottom). Similarly to all previously reported adenovirus structures, we were able to trace only 244 amino acids of protein IIIa (out of a total of 590 in FAdV-C4 vs. 585 in HAdV-C5, S4 Table). From N- to C-termini, the traced structure folds into the same domains previously defined for HAdV-C5: GOS-glue (cementing penton and peripentonal hexons), connecting helix, VIII-binding (interacting with the peripentonal copy of protein VIII), and core proximal [3]. The structural superposition of HAdV-C5 and FAdV-C4 protein IIIa structures reveals several major differences (S11 Table and Fig 4a). The N-terminus of protein IIIa is ten amino acids shorter in FAdV-C4 than in HAdV-C5 (diff 1). Tracing started at Thr16 for FAdV-C4 (Fig 4a). In the atadenovirus LAdV-2 this region is also shorter than in HAdV-C5, but its structure is ordered from Gln2 (Fig 4b). As a result, in FAdV-C4 the GOS-glue domain shows only one tier of ordered amino acids, compared with the two tiers present in the other solved IIIa structures (Fig 4a and 4b, arrowheads). Since in other adenoviruses this region interacts with peripentonal hexons [3,6] and the N-terminal arm of penton base [6], its lack of order suggests a different network of interactions stabilizing the vertex region in FAdV-C4.

Download:

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

Fig 4. FAdV-C4 minor coat protein IIIa.

(a) Superposition of protein IIIa in HAdV-C5 (PDB ID: 6b1t, white) and FAdV-C4 (coloured by RMSD), presented in their native position in the capsid. The view is from inside the capsid, with the 5-fold symmetry axis at the top of the page. The GOS-glue, connecting helix, VIII-binding and core proximal domains are indicated. (b) As in (a), but comparing with LAdV-2 protein IIIa in grey (PDB ID: 6qi5). The GOS-glue domains and two thirds of the connecting helices mostly overlap in the three structures (FAdV-C4, HAdV-C5 and LAdV-2). The N-terminal region in FAdV-C4 presents fewer ordered amino acids (filled arrowheads), lacking a ~ 12 residue stretch ordered in HAdV-C5 and LAdV-2 (hollow arrowheads). The VIII-binding domain is significantly rotated with respect to its position in HAdV-C5 and slightly rotated with respect to LAdV-2. (c) Centre: superposition of the three protein IIIa structures: FAdV-C4 (orange), HAdV-C5 (white) and LAdV-2 (grey), represented in their native position in the capsid. Left: comparison of the three connecting helices. Horizontal black lines indicate the region where the three helices overlap; dashed lines indicate their bending away from each other (diff 2). The dashed black rectangles show the VIII-binding domains in the three viruses, with the small handle formed by diff3 labelled as a landmark. A cylinder indicates the first helix of the VIII-binding domain in FAdV-C4. (d) Focus on the connecting helix and the first helix in the VIII-binding domain showing how their relative orientations vary between the three viruses. For convenience a cartesian coordinate system is represented with the connecting helix of FAdV-C4 along the Y axis and the first helix of the VIII-binding domain of HAdV-C5 along the X axis. The origin of coordinates is located at the last amino acid of the FAdV-C4 connecting helix.

https://doi.org/10.1371/journal.ppat.1013553.g004

The next point of structural divergence between protein IIIa in the different AdV genera is the C-terminal part of the connecting helix (diff 2). The helix in FAdV-C4 is straight, compared to the curved helices in HAdV-C5 and LAdV-2. While the first two thirds of the helix overlap perfectly in the three viruses, at the end the helix in the human and reptilian viruses deviate by ca. 20º, each one in a different direction, from the avian one (Fig 4c, left). The helix is followed by a flexible loop that, along with the bend of the connecting helix, determines the relative orientation of the VIII-binding domain with respect to the rest of the protein. As a result, the VIII-binding domain is differently oriented in the human, reptilian and avian adenoviruses. In both LAdV-2 and FAdV-C4, the VIII-binding domain swings away from its position in HAdV-C5 by more than 180º (Fig 4c, center). Additionally, while in the human and lizard viruses the first helix of the VIII-binding domain is roughly perpendicular to the connecting helix, in FAdV-C4 this helix adopts an oblique orientation (Fig 4d), which results in a completely different conformation of the protein in each one of the genera (S1 Video). This difference in conformation directly impinges on the network of contacts established by protein IIIa in the virion (see section Results. Protein interaction networks in FAdV-C4).

In spite of this drastic change in the relative position of domains, the backbone of protein IIIa VIII-binding domain is conserved when compared to the other adenoviruses, except for two small regions (diff3 and diff4) (S11 Table and S6b Fig) that show variability among the three structures. Finally, the core proximal domain, formed by the last traced residues in HAdV-C5, does not seem to be ordered in FAdV-C4 (diff 5) (S6b Fig), as similarly reported for the atadenovirus LAdV-2 [6].

Additional internal densities

Analysis of densities that did not allow unequivocal polypeptide tracing (remnant densities, RD) revealed several features of interest (Fig 5). By comparison to other AdVs, RD1, with three copies located at the mouth of each hexon, presumably corresponds to peptides cleaved during maturation from the precursor forms of proteins VI and VII [16]. L-shaped densities at the local 3-fold symmetry axis between hexons 2, 3 and 4 (RD2a), and an equivalent 3-fold averaged density at the icosahedral 3-fold (RD2b) have also been observed in human adenovirus maps [16,17,36,38]. In HAdV-F41, RD2a was interpreted as part of core protein V [17]. However, FAdV-C4 lacks protein V and does not have genus-specific proteins substituting for V. This observation suggests that a virion component other than protein V, and present in both mastadenoviruses and aviadenoviruses, must be filling this position. The fact that HAdV-C5 particles devoid of protein V also present these RD2a densities further support this hypothesis [13].

Download:

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

Fig 5. Additional internal components of the FAdV-C4 capsid.

(a) FAdV-C4 map, as viewed from inside of the capsid. Left: only the density corresponding to the traced proteins is shown. The two-fold (oval), three-fold (triangle) and five-fold (pentagon) symmetry axes, the four hexons in one AU (H1-H4), and the facet (dashed triangle) are indicated for reference. Right: remnant densities (RD) are coloured in red. (b) Remnant map components colour-coded as RD1-4. Right: a zoom showing only the colour-coded RDs. RDs considered too small to be analysed are left in grey. Densities correspond to the unsharpened map contoured at 1.5 σ. Hexagon contours indicate hexon positions.

https://doi.org/10.1371/journal.ppat.1013553.g005

Four α-helices can be modelled in RD3, located at the vertex near proteins IIIa and VIII. (S6c Fig). This is reminiscent of the appendage domain (APD) of protein IIIa, a region downstream of the core proximal domain observed in HAdV-D26 and also in immature particles of HAdV-C5 [36,37,51]. We tentatively traced amino acids 305–384 of FAdV-C4 protein IIIa in RD3 (S4 Table). The topology of the α-helices and the position of RD3 in the FAdV-C4 capsid do not overlap with those of HAdV-D26, possibly due to the presence of flexible linkers between the helices and to the differently oriented VIII-binding domains. In any case, this RD3 element would be reinforcing the link between the GOS and the GON. Finally, RD4 densities are located at the two-fold axes between H2 trimers. Due to the intrinsic curvature of the capsid shell, hexon interactions at the two-fold axes are reduced, compared to the rest of inter-hexon interactions in the capsid (S13 Table). RD4 might be contributing to reinforce these weaker hexon-hexon interactions.

Protein networks in FAdV-C4

Interactions between major and minor capsid proteins have been studied for different adenoviruses, looking for the basis of capsid assembly and stability. Given that FAdV-C4 proteins IIIa and VIII present the largest deviations from the rest of traced capsid components, we focus here on the interactions between these proteins and the rest of the capsid, and how they change with respect to the rest of available AdV structures.

Contacts between neighbouring IIIa monomers are similar in number and location for HAdV-C5 and FAdV-C4, and mostly involve the GOS-glue domain of one monomer and the connecting helix of the other (Fig 6a and S16 Table). However, the contacts between IIIa and the copy of protein VIII located beneath the GOS diverge noticeably, mostly as a consequence of the large conformational difference found in IIIa. Although the interaction is still established via the VIII-binding domain of protein IIIa, the number of contacts decreases by ca. 75% (Fig 6b and S16 Table). Another consequence of the different conformation of the VIII-binding domain is the loss of interactions between IIIa and hexon 4 (Fig 6c and S16 Table). Therefore, the change in the orientation of the VIII-binding domain in protein IIIa could in principle weaken the attachment between the GOS and the rest of the capsid.

Download:

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

Fig 6. Differences in interactions established by protein IIIa in HAdV-C5 and FAdV-C4.

(a) Protein-protein interfaces between IIIa (yellow surface) and its neighbouring copy IIIa’ (symmetry copy coloured in white) for HAdV-C5 (left) and FAdV-C4 (right). (b) Interfaces between IIIa (yellow surface) and VIII (orange surface). (c) Protein-protein interfaces between IIIa (yellow surface) and neighbouring hexons (blue surface). In each panel, the interacting molecules are shown at the top as seen from inside the vertex region (indicated with a black pentagon), and one of them is separated and rotated as indicated at the bottom, to reveal the interfacing residues (coloured in red). Dashed arrows indicate that the molecule orientation is maintained between the top and bottom panels. The number of contacting residues is indicated in red inside circles.

https://doi.org/10.1371/journal.ppat.1013553.g006

On the other hand, contacts between both copies of protein VIII and the surrounding hexons are slightly more numerous for FAdV-C4 than for HAdV-C5 (Fig 7a and S17 Table). The “footprint” of VIII on the hexons is broader, reflecting the protein shape. Interestingly, the N-terminal region of one monomer in hexon 1 (chain B) interlaces with the divergent neck region of VIII, while the N-terminus of one monomer in hexon 2 (chain F) crosses over the head domain of protein VIII, further securing it in a clasp near the AVP cleavage site (Fig 7a, 7b, ovals, and 7c). Thus, the long N-terminal tail unique to the aviadenovirus hexon in combination with the different fold of protein VIII could play a stabilizing role in the capsid. If the structure of protein VIII in FAdV-C4 was the same as in mast- or atadenoviruses, it would clash with the long hexon N-terminal tails.

Download:

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

Fig 7. Interfaces between protein VIII and hexons in HAdV-C5 and FAdV-C4.

(a) Protein- protein interfaces between VIII (orange surface) and hexons (blue surface), with contacting residues highlighted in red, and their number indicated as in Fig 6. The N-termini of HAdV-C5 hexons are shorter than their FAdV-C4 counterparts (ovals). (b) Zoom in on protein VIII. The traced N-termini of the HAdV-C5 hexons (blue) are not interlaced with protein VIII (orange) (left, ovals) whereas the N-termini of the FAdV-C4 hexons (most notably H2) are clasping protein VIII (right, ovals). (c) Ribbon representation of the molecules represented as surfaces in (a) and (b), with the non-traced first residues of the hexons indicated by dashed lines. Blurry curved arrows represent the flexibility of the hexon N-termini. Scissors indicate the AVP sites. Dashed frames are drawn to facilitate comparisons. Hexon chain IDs (according to S4 Table) are indicated in parenthesis.

https://doi.org/10.1371/journal.ppat.1013553.g007

Insights into the FAdV-C4 core

Aviadenovirus genomes (including that of FAdV-C4) are among the longest in the Adenoviridae family, approximately 25% longer than those of human AdV (approx. 45 kbp vs 36 kbp dsDNA) [1]. So far, a longer genome has only been reported in the single known ichtadenovirus (48 kbp) [52]. All AdV genomes code for core proteins VII and µ, but only mastadenovirus genomes contain the gene for protein V. Although they appear in all AdV genera, proteins VII and µ differ substantially in sequence and length (Fig 8a). For example, in mastadenoviruses protein VII is considerably longer than µ (198 vs 80 amino acids for the immature polypeptides) while the length relation is reversed in aviadenoviruses (72 residues in VII vs 176 aa in µ). In spite of this variability, the N-terminus of pVII and the C-terminal region of protein pre-µ (pµ) are well conserved across all the genera. Remarkably, pVII and pµ in ichtadenoviruses [52] are reduced to just these two conserved domains. The conserved domains are delimited by consensus maturation cleavage sites (Fig 8a, dashed lines). Since AVP is a genus-common gene, proteolytic maturation seems to be an ancient feature of the adenovirus replicative cycle. These observations point towards a hot spot of evolution in the region coding for the core proteins.

Download:

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

Fig 8. Organization of the core varies among adenovirus genera.

(a) Multiple protein sequence alignment of core proteins pVII (top) and pµ (bottom), showing their respective length (in amino acids) encircled. NCBI GenBank accession numbers for the genomes chosen as representative members of each genus are as follows: MastAdV, HAdV-C5 (AC_000008); AviAdV, FAdV-A1 (AC_000014); SiAdV, FrogAdV-1 (NC_002501); AtAdV, OAdV-7 (NC_004037); IchtAdV, WSAdV (MK_101347). Predicted AVP cleavage sites are indicated by dashed vertical lines following the consensus patterns for proteins pVII and pµ. Amino acids are coloured by polarity, and the top histograms indicate the mean pairwise identity over all pairs in the column. Notice that the number of cleavage sites also varies between genera, but the sites delimiting the conserved domains of protein VII and µ are also conserved. Blue rectangles highlight high similarity regions corresponding to the pVII N-terminal peptide (pVII_n1) and C-terminal region of pµ. (b) Central section of a HAdV (as an example we use HAdV-F41, EMDB code:10768, left) and FAdV-C4 (right) cryo-EM maps. Arrows point to more pronounced concentric shells in FAdV-C4. (c) Radial average profiles of FAdV-C4 and HAdV-F41 maps. The arrows indicate the 25 Å spacing between shells.

https://doi.org/10.1371/journal.ppat.1013553.g008

In previously published mastadenovirus [13,18,38,53] and atadenovirus [6,7] cryo-EM maps where the core has not been masked away, there is mostly featureless density inside the capsid. This indicates that the core does not have an icosahedral organization. In FAdV-C4, the unsharpened map presents a few, weak concentric shells near the inner wall of the capsid (Fig 8b). The distance between shells is 25 Å (Fig 8c, arrows), similar to the distances observed between the concentric layers of naked dsDNA in tailed bacteriophages [54]. The same shells are observed in the FAdV-C4 map lacking pentons (S2c and S2d Fig). The presence of concentric shells suggests that the longer aviadenovirus genome may have a tendency to pack in a more ordered fashion than the mastadenovirus genome. Similar weak shells have been observed in HAdV-C5 particles lacking protein VII, suggesting that they may originate from changes in the DNA/protein balance in the core [18]. Another hint of a different core organization in FAdV-C4 is the occasional observation of material released from disrupted particles in cryo-EM grids, which appears as condensed yarn ball-like objects (S7 Fig). These are reminiscent of the cores released from immature HAdV-C2 ts1 particles [53,55] in contrast to cores released by HAdV-C5 virions which show a diffuse aspect. This observation suggests that the mature aviadenovirus core proteins have a stronger DNA condensing action than those in mature mastadenoviruses. Since in HAdV-C2 ts1 particles the higher core condensation correlates with the higher stability of the immature particle [15,55–57], this observation prompts the hypothesis that core organization also has a role in the increased thermostability of the aviadenovirus virion. In fact, extrinsic fluorescence analyses show that genome exposure to intercalating dyes requires higher temperatures in FAdV-C4 than in HAdV-C5 not only at neutral pH (Fig 1a), but also in a range of acidic conditions (S8 Fig). Lack of genome accessibility to dyes is also a sign of stronger core condensation [44].

Virus production and purification

Two very similar strains (98.1% global pairwise nucleotide identity) of FAdV-C4 were used. Strain KR5 [23,64,65] [NCBI GenBank: HE608152.1] was used for the extrinsic fluorescence experiments, mass spectrometry analysis and cryo-EM structure determination of the complete FAdV-C4 virion. KR5 was propagated in primary CEL (chicken embryo liver) cells infected at a multiplicity of infection (MOI) = 1 and harvested at 67 hours post infection (hpi). Strain AG234 [22,26] [NCBI GenBank: MK572849.1] yielded a penton-less cryo-EM map and was propagated in LMH (ATCC: CRL-2117, Leghorn Male hepatocellular carcinoma from chicken epithelial liver) cells infected at a MOI = 4 and harvested at 27 hpi. The HAdV-C5/attP variant described in [66] was propagated as previously reported [18,67] and used as a control for extrinsic fluorescence experiments.

For both FAdV-C4 and HAdV-C5, infected cells were centrifuged in a Heraeus 1.0R megafuge for 40 min at 4000 rpm and 4°C, resuspended in 42 ml of medium (from the supernatant) and lysed by four freeze-thaw cycles. Cell lysates were clarified to remove cellular debris by centrifugation in a Heraeus 1.0R megafuge at 4000 rpm for 30 min at 4°C. For double CsCl gradient ultracentrifugation, the supernatant was distributed in 6 tubes containing a discontinuous gradient of 1.25 g/ml and 1.40 g/ml CsCl in TD1X buffer (137 mM NaCl, 5.1 mM KCl, 0.7 mM Na₂HPO₄.7H₂O, 25 mM Tris-HCl pH 7.4) (2.5 ml of each CsCl buffer and 7 ml of supernatant) and centrifuged at 35700 rpm for 90 min at 18°C in a Beckman Optima L-100 XP ultracentrifuge using a Beckman SW41Ti swinging bucket rotor. The low and high density virus bands from each tube were collected (~1 ml of band by tube) and independently pooled. In the second gradient, the low and high density pooled materials (6 ml) were laid onto 6 ml of a CsCl 1.31 g/ml solution in TD1X buffer and centrifuged for 18 hours at 35700 rpm and 18°C in a Beckman SW41Ti swinging bucket rotor. Bands containing viral particles from each tube were collected and buffer exchanged into HBS (20 mM HEPES, 0.15M NaCl pH7.8) using dialysis filters (10 kDa cut off, Sigma #D9277) for FAdV-C4, or Econo-Pac 10DG disposable chromatography columns (Biorad Cat#732–2010) with molecular weight cutoff of 6000 Dalton for HAdV-C5. Samples were stored at -80°C after adding glycerol to a final concentration of 10%.

Quantification of physical viral particles

Capsid protein concentration was quantified using hexon fluorescence emission spectra obtained in a Hitachi Model F-2500 FL spectrophotometer. Sample volumes of 0.150 ml were examined in sealed quartz cuvettes. The sample was excited at 285 nm, and the emission was monitored from 310 to 375 nm using excitation and emission slit widths of 10 nm. The spectra were corrected by subtraction of the buffer spectrum. The maximum emission intensity for each spectrum was found at 333 nm and recorded. The concentration (in viral particles per ml, vp/ml) was determined from a calibration curve calculated from a sample with known concentration.

Extrinsic fluorescence thermostability assays

Capsid thermostability was analysed by measuring extrinsic fluorescence spectra in the presence of an intercalating dye (YOYO-1), thereby detecting accessibility of the virus genome to the medium. The data for HAdV-C5 have been reported previously [18,38]. The data for FAdV-C4 were collected in the same time period, as part of a large study comparing the thermostability of different adenovirus specimens. Stock virus samples (~10¹² vp/ml) of HAdV-C5 as control or FAdV-C4 were diluted to a final concentration of 5 × 10⁹ particles/ml in a final volume of 800 µl of Sodium-Potassium Phosphate buffer (8 mM Na₂HPO₄, 2 mM KH₂PO₄, 150 mM NaCl, and 0.1 mM EDTA, pH 7.4) and incubated overnight at 4 °C at 300 rpm shaking. To analyse thermostability in acidic conditions, samples were incubated in citric acid-sodium citrate buffers: 9.5 mM C₆H₈O_7, 41.5 mM C₆H₅O₇Na₃.2H₂O at pH 6; 20.5 mM C₆H₈O₇, 29.5 mM C₆H₅O₇Na₃.2H₂O at pH 5, and 33 mM C₆H₈O₇, 17 mM C₆H₅O₇Na₃.2H₂O at pH 4; all with 150 mM NaCl and 0.1 mM EDTA.

A volume of 8.8 μl of YOYO-1 diluted 1:100 in milliQ water from the initial stock (Thermo Fisher Scientific, stock composition: 1 mM YOYO-1 in DMSO) was added to the virus sample and equilibrated for 5 min at 20ºC (or 30ºC for the experiments in acidic conditions) before data acquisition. Fluorescence emission spectra were obtained employing a Horiba FluoroLog spectrophotometer equipped with a Peltier temperature control device and continuous shaking. The dye was excited at 490 nm wavelength, and fluorescence emission was monitored from 500 to 580 nm using an excitation slit of 2 nm and an emission slit of 5 nm, integration time 2 s and 950 V. Maximal emission intensity occurred at 509 nm. The temperature in the sample cell was increased from 20ºC to 70ºC for neutral pH assays; for acidic pH assays, temperatures ranged from 30 to 70ºC for HAdV-C5, or to 90ºC for FAdV-C4. Emission was recorded every 2ºC after holding for 0.5 min for equilibration. Each run took approximately one hour and a half to be completed. The fractional change in fluorescence (normalized fluorescence, F_n (T)), was calculated according to equation (1).

(1)

where T indicates the independent variable temperature (ºC) and F_exp refers to the raw fluorescence signal in counts per second (c.p.s.) at each temperature. F_i and F_f represent the linear extrapolates, at this temperature, of pre-transition and post-transition base lines, respectively. Buffer signal was almost negligible and was not subtracted.

For assays at neutral pH, a total of six (HAdV-C5) and five (FAdV-C4) independent experiments were considered for each averaged curve, normalized separately according to equation (1) and averaged to obtain the average curve for each condition. Two independent experiments for each virus were carried out for thermostability assays in acidic conditions. Normalized fluorescence F_n (T) curves were plotted as average values ± standard deviation (STD). The half transition temperatures (T_0.5) were calculated from the fitting of the averaged fluorescence change fraction (F_n) as a function of temperature (T) to a Boltzmann sigmoid (Origin software package) according to equation (2):

(2)

Where F₀ 0, F_max 1 and dT = steepness of the curve.

T_0.5 are expressed as fit values of each averaged curve ± standard error (SE).

Mass spectrometry analyses

The proteins present in purified FAdV-C4 virions were identified by shotgun proteomic analysis. Briefly, the sample was reduced with 10 mM DTT and alquilated with 55 mM iodoacetamide. To digest the sample, trypsin was added in a 1:25 enzyme:protein ratio (w/w) and incubated overnight. Interfering salts were washed using C18 Zip Tips (Millipore) and the eluting peptides were dried prior to analysis by mass spectrometry. LC-MSMS analysis was performed using an Eksigent 1D-nanoHPLC coupled to a 5600 TripleTOF QTOF mass spectrometer (Sciex, Framinghan, MA, USA). The analytical column used was a C18 Waters nanoACQUITY UPLC 75 µm × 15 cm, 1.7 µm particle size. The trap column was an Acclaim PepMap 100, 5 µm particle diameter, 100 Å pore size, switched on-line with the analytical column. The loading pump delivered a solution of 0.1% formic acid in 98% water/ 2% acetonitrile (Scharlab, Barcelona, Spain) at 3 µL/min. The nanopump provided a flow-rate of 250 nL/min and was operated under gradient elution conditions, using 0.1% formic acid (Fluka, Buchs, Switzerland) in water as mobile phase A, and 0.1% formic acid in 100% acetonitrile as mobile phase B. Gradient elution was performed according the following scheme: isocratic conditions of 96% A: 4% B for five minutes, a linear increase to 40% B in 25 min, a linear increase to 95% B in two minutes, isocratic conditions of 95% B for five minutes and return to initial conditions in 10 min. Injection volume was 5 µL. The LC system was coupled via a nanospray source to the mass spectrometer. Automatic data-dependent acquisition using dynamic exclusion allowed obtaining both full scan (m/z 350–1250) MS spectra followed by tandem MS CID spectra of the 15 most abundant ions. Acquisition time was 250 ms and 100 ms for MS and MSMS spectra, respectively. MS and MS/MS data were used to search against a customized database containing all the proteins encoded by the FAdV-C4 strain KR5 genome [NCBI GenBank: HE608152.1] and a collection of typical laboratory protein contaminants. Database searches were done using a licensed version of Mascot v.2.6 [68], and search parameters were set as follows: carbamidomethyl cysteine as fixed modification and acetyl (protein N-term), NQ deamidation, Glu to pyro-glutamic and oxidized methionine as variable ones. Peptide mass tolerance was set at 25 ppm and 0.1 Da for MS and MS/MS spectra, respectively, and 2 missed cleavages were allowed. Restriction enzyme was set to none. Peptides with individual Mascot ions scores indicating identity or extensive homology (p < 0.05) were used for protein identification. Only those proteins with at least one unique peptide were considered. The software Skyline v22.2 [69] was used to extract the signal corresponding to the identified protein-specific peptides. The area (arbitrary units) corresponding to each peptide was integrated and summed at the protein level.

Cryo-EM sample preparation and data collection

Cryo-EM image processing

All image processing and 3D reconstruction tasks (S2 Table) were performed within the Scipion framework [71]. Frames were aligned using Motioncor2 and weighted according to the electron dose received before averaging [72]. The contrast transfer function (CTF) was estimated using gCTF [73]. Particles were semi-automatically picked from micrographs corrected for the phase oscillations of the CTF (phase-flipped), extracted into 758 × 758 (FAdV-C4 KR5) or 752 x 752 (FAdV-C4 AG234) pixel boxes, normalized and downsampled by a factor of 2 using Xmipp [74]. All 2D and 3D classifications and refinements were performed using RELION [75]. 2D classification was used to discard low quality particles, and run for 25 iterations, with 100 classes, angular sampling 5 and regularization parameter T = 2. Classification in 3D was run in rounds of 10 iterations, with three classes, starting with an angular sampling of 3.7° and sequentially decreasing to 0.5, and regularization parameter T = 4. Icosahedral symmetry was imposed throughout the refinement process. The initial reference for 3D classification was a lizard adenovirus cryo-EM map [6], low-pass filtered to 60 Å resolution. The class yielding the best resolution was individually refined using the original particles and the map obtained during the 3D classification as a reference, yielding final maps at 3.3 Å and 3.2 Å resolution for FAdV-C4 KR5 and FAdV-C4 AG234 respectively. Resolution was estimated according to the gold-standard FSC = 0.143 criterion as implemented in RELION auto-refine and postprocess routines [76]. The actual sampling for the map was estimated by comparison with a HAdV-C5 model (PDB ID 6b1t) [16] in UCSF Chimera [77].

Grey values of the maps were standardized with UCSF ChimeraX [78] (volume scale shift -mean factor 1/Standard Deviation) to have an average = 0 and standard deviation σ = 1 within a region comprising the hexon 1 trimer. Map density thresholds are given as multiples of σ. An Xmipp protocol implemented in Scipion was used to calculate radial average profiles, normalized to minimum and maximum values in the [0, 1] interval.

Model building and structure analysis

Interpretation of the FAdV-C4 KR5 3D map was performed using the molecular modelling workflow based on sequence homology in Scipion [79]. The initial model for each polypeptide chain was predicted with Modeller [80], using as input template the structure of the respective HAdV-C5 homolog chain (PDB ID 6b1t). UCSF Chimera was used to fit each chain initial model as a rigid body into the sharpened map. Next, the fitted model of each chain was iteratively refined using Coot [81] and Phenix real space refine [82]. Validation metrics to assess the model quality were computed with the Phenix comprehensive validation (cryo-EM) algorithm.

After generating the whole AU structure, Chimera findclash (integrated in the Scipion protocol chimera-contacts) was executed to identify possible interactions between pairs of chains contained in the AU or between a chain in the AU and a chain from a neighboring AU. The default parameters (cutoff = −0.4 and allowance = 0.0) were used to report as possible bonds all pair of atoms separated by a distance no higher than the sum of the corresponding van der Waals radii plus 0.4 Å. To identify additional unmodeled densities, we generated remnant maps by masking off from the initial map a region within a 3, 4 or 8 Å radius from the modelled atoms using ChimeraX (vol zone invert true). Cα backbones of the unmodeled densities were built by automatic placing of poly-Ala helices in Coot. Assignment of sequence of probable candidates was semi-automatically performed with the docking GUI in Coot and manually assessed.

Protein charge at pH 7.0 was calculated as described in http://isoelectric.org/index.html, with amino acid pKa values taken from [83], and general pKa values for terminal amino and carboxy groups taken from [84]. Protein sequence alignments were performed with Clustal Ω (http://www.clustal.org/omega/) integrated in Geneious version 2019.0 (https://www.geneious.com). Secondary structure elements and similarity depictions were performed with ESPript 3 (https://espript.ibcp.fr/ESPript/cgi-bin/ESPript.cgi).

Molecular graphics and analyses including molecule superposition (matchmaker command), all-atom RMSD calculation, map rendering, map normalization and colouring by surface electrostatic potential were performed with UCSF Chimera or UCSF ChimeraX. UCSF Chimera Hide dust was used for clarity when composing figures representing density maps.