ARTICLE
`
`DOI: 10.1038/s41467-018-03432-4
`
`OPEN
`Structural basis for the recognition of LDL-receptor
`family members by VSV glycoprotein
`1, Laura Belot1, Hélène Raux1, Pierre Legrand
`2, Yves Gaudin
`1 & Aurélie A. Albertini1
`
`Jovan Nikolic
`
`Vesicular stomatitis virus (VSV) is an oncolytic rhabdovirus and its glycoprotein G is widely
`used to pseudotype other viruses for gene therapy. Low-density lipoprotein receptor (LDL-R)
`serves as a major entry receptor for VSV. Here we report two crystal structures of VSV G in
`complex with two distinct cysteine-rich domains (CR2 and CR3) of LDL-R, showing that their
`binding sites on G are identical. We identify two basic residues on G, which are essential for
`its interaction with CR2 and CR3. Mutating these residues abolishes VSV infectivity even
`though VSV can use alternative receptors, indicating that all VSV receptors are members of
`the LDL-R family. Collectively, our data suggest that VSV G has specifically evolved to
`interact with receptor CR domains. These structural insights into the interaction between
`VSV G and host cell receptors provide a basis for the design of recombinant viruses with an
`altered tropism.
`
`1234567890():,;
`
`1 Institute for Integrative Biology of the Cell (I2BC), CEA, CNRS, Univ. Paris-Sud, Université Paris-Saclay, 91198 Gif-sur-Yvette cedex, France. 2 Synchrotron
`SOLEIL, 91192 Gif-sur-Yvette cedex, France. These authors contributed equally: Jovan Nikolic, Laura Belot. Correspondence and requests for materials should
`be addressed to Y.G. (email: yves.gaudin@i2bc.paris-saclay.fr) or to A.Albertini. (email: aurelie.albertini@i2bc.paris-saclay.fr)
`
`NATURE COMMUNICATIONS | (2018) 9:1029
`
`| DOI: 10.1038/s41467-018-03432-4 | www.nature.com/naturecommunications
`
`1
`
`Page 1 of 12
`
`KELONIA EXHIBIT 1018
`
`

`

`ARTICLE
`
`Vesicular stomatitis virus (VSV) is an enveloped, negative-
`
`strand RNA virus that belongs to the Vesiculovirus genus
`of the Rhabdovirus family. It is an arbovirus which can
`infect insects, cattle, horses, and pigs. In mammals, its ability to
`infect and kill tumor cells, although sparing normal cells makes it
`a promising oncolytic virus for the treatment of cancer1–3. VSV
`genome encodes five structural proteins among which a single-
`transmembrane glycoprotein (G). G plays a critical role during
`the initial steps of virus infection4. First, it is responsible for virus
`attachment to specific receptors. After binding, virions enter the
`cell by a clathrin-mediated endocytic pathway5,6. In the acidic
`
`NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03432-4
`
`the endocytic vesicle, G triggers the fusion
`environment of
`between the viral and endosomal membranes, which releases the
`genome in the cytosol for the subsequent steps of infection.
`Fusion is catalyzed by a low-pH-induced large structural transi-
`tion from a pre toward a post-fusion conformation, which are
`both trimeric7,8.
`The polypeptide chain of G ectodomain folds into three dis-
`tinct domains which are the fusion domain (FD), the pleckstrin
`homology domain (PHD), and the trimerization domain (TrD).
`During the structural transition, the FD, the PHD, and the TrD
`retain their tertiary structure. Nevertheless, they undergo large
`
`a
`
`sp
`
`1
`
`2
`
`3
`
`4
`
`5
`
`6
`
`7
`
`a
`
`cßb
`
`Ectodomain
`
`TM
`
`CR domains
`
`b
`GST-CR 1
`
`+ Gth pH 8
`2 3 4 5 6
`
`7
`
`GST
`
`O-link
`
`EGF
`c
`GST-CR
`
`+ Gth pH 8 + Gth pH 6
`1
`2 3
`1
`2 3
`
`d
`
`GST-CR
`
`+ VSV pH 8
`1
`2 3
`
`+ VSV pH 6
`1
`2 3
`
`L
`
`G N
`
`CRx-GST
`
`Gth
`
`CRx-GST
`GST
`
`75
`
`25
`
`75
`
`Gth
`
`CRx-GST
`
`25
`
`BSR
`
`Virus
`
`4 h
`
`or
`
`Anti VSV N
`GST-CRATTO550
`
`+VSVΔG-CHAV G
`
`f
`
`Anti N
`
`Microscopy
`
`+VSV
`
`GST-CR1ATTO550
`
`Merge
`
`Anti N
`
`GST-CR2ATTO550
`
`Merge
`
`Anti N
`
`GST-CR2ATTO550
`
`Merge
`
`75
`
`25
`
`e
`
`g
`
`Anti N
`
`GST-CR3ATTO550
`
`Merge
`
`Anti N
`
`GST-CR3ATTO550
`
`Merge
`
`Fig. 1 VSV G interacts specifically with CR2 and CR3 in its pre-fusion conformation. a Scheme of the modular organization of the LDL-R indicating the 7 CR
`modules (1–7), the 3 EGF repeats (a,b and c) , the seven-bladed β-propeller domain (β) of the epidermal growth factor precursor like domain (EGF), and the
`C-terminal domain containing O-linked oligosaccharides (O-link). SP signal peptide, TM transmembrane domain. b SDS–PAGE analysis of interaction
`experiments between the 7 GST-CR proteins, bound to GSH magnetic beads, and Gth at pH 8. c, d Coomassie-stained SDS–PAGE of interaction
`experiments between GST-CR1, GST-CR2 and GST-CR3, bound to GSH magnetic beads, and Gth (c) or VSV (d) at pH 8 and 6, respectively. Purified GST-
`CR bound to GSH magnetic beads were incubated with either Gth or VSV in the appropriate pH condition in presence of Ca2+ for 20 min at 4 °C. Then,
`after wash, the beads were directly loaded on a gel. As a control in b, GST alone bound to the GSH coated beads was incubated in presence of Gth.
`e Cartoon that illustrates the experiments presented in f and g. After 4 h of infection, BSR cells were labeled with an antibody directed against VSV
`nucleoprotein (anti-VSV N) to visualize the infection (green fluorescence) and a GST-CRATTO550 to probe CR domain recognition by the surface displayed
`glycoprotein (red fluorescence). f Labeling of G at the surface of BSR cells infected with VSV using fluorescent GST-CR1ATTO550, GST-CR2ATTO550, and
`GST-CR3ATTO550. At 4 h post-infection (p.i.), cells were incubated with the appropriate GST-CRATTO550 at 4 °C during 30 min prior fixation and
`permeabilization and then immuno-labeled using an anti-VSV N antibody to visualize the infection. g Labeling of CHAV G at the surface of BSR cells
`infected with a recombinant VSV expressing CHAV G (VSVΔG-CHAVG) using fluorescent GST-CR2ATTO550 and GST-CR3ATTO 550. Infected cells are
`labeled using anti-VSV N antibodies. In f and g, DAPI was used to stain the nuclei. Scale bars=20 µm
`
`2
`
`NATURE COMMUNICATIONS | (2018) 9:1029
`
`| DOI: 10.1038/s41467-018-03432-4 | www.nature.com/naturecommunications
`
`Page 2 of 12
`
`

`

`NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03432-4
`
`ARTICLE
`
`CR1
`0
`
`10
`
`Time (min)
`20
`30
`40
`
`50
`
`60
`
`CR2
`0
`
`10
`
`Time (min)
`20
`30
`40
`
`50
`
`60
`
`CR3
`Time (min)
`0
`10 20 30 40 50
`
`60 70 80
`
`Kd=3.75 μM
`n = 0.8
`
`0.05
`
`0.00
`
`–0.05
`
`–0.10
`
`–0.15
`
`–0.20
`
`–0.25
`
`0.0
`
`–2.0
`
`–4.0
`
`–6.0
`
`μcal/s
`
`kcal mol–1 of injectant
`
`Kd=7.5 μM
`n = 1
`
`0.00
`
`–0.04
`
`–0.08
`
`–0.12
`0.0
`
`–1.0
`
`–2.0
`
`–3.0
`
`μcal/s
`
`kcal mol–1 of injectant
`
`0.0
`
`0.5
`
`1.0
`
`1.5
`
`2.0
`
`0.0
`
`0.5
`
`1.0
`
`1.5
`
`2.0
`
`2.5
`
`0.0
`
`0.5
`
`1.0
`
`1.5
`
`2.0
`
`2.5
`
`Molar ratio
`
`Molar ratio
`
`Molar ratio
`
`+VSVeGFP
`
`GST-CR1
`100 nM
`
`GST-CR2
`100 nM
`
`GST-CR3
`100 nM
`
`a
`
`0.20
`
`0.10
`
`0.00
`
`–0.10
`
`–0.20
`
`1.00
`
`0.50
`
`0.00
`
`μcal/s
`
`kcal mol–1 of injectant
`
`b
`
`CR1
`100 μM
`
`CR2
`100 μM
`
`CR3
`100 μM
`
`GST-CR2
`GST-CR3
`
`CR2
`CR3
`
`1
`
`4
`2
`3
`Receptor concentration (log nM)
`
`5
`
`6
`
`100
`
`50
`
`c
`
`% Neutralization
`
`0
`
`0
`
`Fig. 2 Characterization of VSV G-CR2 and VSV G-CR3 interaction a Isothermal titration calorimetry (ITC) analysis between Gth and CR1, Gth and CR2, Gth
`and CR3 at 20 °C. Representative plots of each ITC experiments are illustrated with raw data in the upper panel. Binding parameters were determined by
`curve fitting analysis with the single-site binding model. The values indicated in the panel are those corresponding to the curves that are presented. Kd
`values given in the text are means of three independent experiments±standard errors. b, c Inhibition of VSV infection by soluble forms of CR domains.
`b BSR cells were infected with VSV-eGFP preincubated with GST-CR1, GST-CR2, GST-CR3 (upper part), CR1, CR2, or CR3 monovalent domains (lower
`part) at the indicated concentrations. Cells were fixed 4 h p.i. Only infected cells are expressing eGFP. Note that neither CR1 nor GST-CR1 construction
`protect cells from infection. DAPI was used to stain the nuclei. Scale bars=100 µm. c VSV-eGFP was preincubated with increasing concentrations of GST-
`CR2, GST-CR3, CR2, or CR3 monovalent domains. At 4 h p.i., the percentage of infected cells was determined by counting the number of cells expressing
`eGFP using a flow cytometer. The percentage of neutralization was equal to 100 × [1−(% of infected cells in presence of CR)/(% of infected cells in the
`absence of CR domains)]. Data depict the mean with standard error for experiments performed in triplicate
`
`rearrangements in their relative orientation due to secondary
`changes in hinge segments (S1 to S5), which refold during the
`low-pH induced conformational change7–10.
`VSV G has been widely used for pseudotyping other viruses11–
`13 and VSV-G-pseudotyped lentiviruses (VSV-G-LVs) exhibit the
`same broad tropism as VSV. Recently it has been shown that low-
`density lipoprotein receptor (LDL-R) and other members of this
`receptor family serve as VSV receptors14. This result explains why
`VSV-G-LVs do not allow efficient gene transfer into unstimulated
`
`T cells, B cells, and hematopoietic stem cells, as they have a very
`low expression level of LDL-R15.
`The LDL-R is a type I transmembrane protein which regulates
`cholesterol homeostasis in mammalian cells16. LDL-R removes
`cholesterol carrying lipoproteins from plasma circulation. Ligands
`bound extracellularly by LDL-R at neutral pH are internalized
`and then released in the acidic environment of the endosomes
`leading to their subsequent lysosomal degradation. The receptor
`then recycles back to the cell surface. LDL-R ectodomain is
`
`NATURE COMMUNICATIONS | (2018) 9:1029
`
`| DOI: 10.1038/s41467-018-03432-4 | www.nature.com/naturecommunications
`
`3
`
`Page 3 of 12
`
`

`

`ARTICLE
`
`NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03432-4
`
`a
`
`PHD
`
`b
`
`PHD
`
`N
`
`CR2
`
`S3
`
`S2
`
`C
`
`S1&S4
`
`TrD
`
`PHD
`
`S3
`
`N
`S2
`
`S5
`
`CR3
`
`N
`S3
`S2
`
`S4
`
`S1
`
`S5
`
`FD
`
`CTer
`
`CR2
`
`C
`
`TrD
`
`Top view
`
`FD
`
`CTer
`
`C
`
`S1&S4
`TrD
`
`S5
`
`PHD
`
`S3
`
`S2
`N
`
`CR3
`
`C
`
`TrD
`
`Top view
`
`S4
`
`S1
`S5
`
`c
`
`Side view
`
`PHD
`
`Y209
`
`TrD
`H8
`R354
`
`S2&S3
`
`K47
`
`Pre-fusion
`
`Side view
`e
`
`Viral membrane
`
`Viral membrane
`
`PHD
`
`S2&S3
`
`CR1
`CR2
`CR3
`
`CR4
`
`CR5
`
`CR6
`CR7
`
`CR1
`CR2
`CR3
`
`CR4
`
`CR5
`
`CR6
`CR7
`
`d
`
`TrD
`
`PHD
`
`Y209
`
`K47
`
`S2&S3
`
`TrD
`
`H8
`
`R354
`
`180°
`
`Post-fusion
`
`Cell membrane
`
`Cell membrane
`
`Fig. 3 X-ray structures of Gth-CR2 and Gth-CR3 complexes a, b Overview of GthCR2 (a) and Gth CR3 (b) crystalline structures in ribbon representation. G is
`depicted by domains and CR domains are in two shades of gray. The conserved disulfides bonds of each CR that maintain their secondary structure are in
`yellow. In both complexes the CR domain is nested in the same cavity of G. N and C-terminal extremities of each CR are indicated. Color code for Gth: the
`trimerization domain (TrD) is in red, the pleckstrin homology domain (PHD) is in orange, the fusion domain (FD) is in yellow. Those domains are connected
`by segments (S1 to S4) which refold during conformational change: segments S1 and S4 are in cyan, segments S2 and S3 are in green, S5 and the C-
`terminal segment (CTer) are in purple. The calcium ion of the CR domains is depicted as a green sphere. c Footprint of CR2 domain on G pre-fusion
`conformation. G is in full atoms view and depicted by domains. Residues of G that establish contacts with CR are shown in black on the surface of the
`protein. d Location of residues interacting with CR domains on G post-fusion conformation. Two views at 180° are shown. Note that the interaction patch is
`scattered when G is in this conformation. e Scheme showing the two complexes that can be formed between VSV G and LDL-R. At the cell surface, at
`neutral pH, the LDL-R adopts an open extended conformation19 and VSV G can bind either CR2 or CR3. Note that the LDL-R in this extended conformation
`has the appropriate orientation to interact with G anchored in the viral membrane
`
`composed of a ligand-binding domain, an epidermal growth
`factor (EGF) precursor homology domain and a C-terminal
`domain enriched in O-linked oligosaccharides. The ligand
`binding domain is made of 7 cysteine-rich repeats (CR1 to CR7,
`Fig. 1a and Supplementary Fig. 1). Each repeat is made of
`approximately 40 amino acids and contains 6 cysteine residues,
`engaged in 3 disulfide bridges, and an acidic residues cluster
`that coordinates a Ca2+
`ion17. The intracellular release of the
`cargo is driven by a low-pH-induced conformational change of
`LDL-R from an open to a closed conformation (Supplementary
`Fig. 1)17–19.
`The LDL-R gene family consists of trans-membrane receptors
`that reside on the cell-surface, are involved in endocytic uptake of
`lipoproteins, and require Ca2+
`for ligand binding. All these
`receptors have in common several CR repeats (up to several tens),
`EGF precursor-like repeats, a membrane-spanning region and an
`intracellular domain containing at least one internalization signal
`sequence20. They are found ubiquitously in all animals including
`insects21.
`
`Here we show that VSV G is able to independently bind two
`distinct CR domains (CR2 and CR3) of LDL-R and we report
`crystal structures of VSV G in complex with those domains. The
`structures reveal that the binding sites of CR2 and CR3 on G are
`identical. We show that HAP-1 cells in which the LDL-R gene has
`been knocked out are still susceptible to VSV infection con-
`firming that VSV G can use receptors other than LDL-R for entry.
`However, mutations of basic residues, which are key for inter-
`action with LDL-R CR domains, abolish VSV infectivity in
`mammalian, as well as insect cells. This indicates that the only
`receptors of VSV in mammalian and in insect cells are members
`of the LDL-R family and that VSV G has specifically evolved to
`interact with their CR domains.
`
`Results
`LDL-R CR2 and CR3 bind G and neutralize viral infectivity.
`We have expressed individually each LDL-R CR domain in fusion
`with the glutathione S-transferase (GST) in Escherichia coli.
`Proteins were solubilized in the presence of sarkosyl (acting as a
`
`4
`
`NATURE COMMUNICATIONS | (2018) 9:1029
`
`| DOI: 10.1038/s41467-018-03432-4 | www.nature.com/naturecommunications
`
`Page 4 of 12
`
`

`

`NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03432-4
`
`ARTICLE
`
`a
`
`CR2 res 44–84
`CR3 res 85–123
`
`44
`
`85
`
`50
`
`90
`
`I
`
`70
`
`II
`
`IVIII
`80
`
`110
`
`120
`
`60
`
`100
`
`c
`
`b
`
`d
`
`K47
`
`Y209
`
`Q71
`
`II(D73)
`
`Y209
`
`D110
`
`K47
`
`II(D112)
`
`Ca2+
`
`R67
`
`C68
`
`H8
`
`I(D69)
`
`F65
`
`R354
`
`Ca2+
`
`I(D108)
`
`Q104
`
`H8
`
`V106
`C107
`
`R354
`
`e
`
`R354
`
`R354
`
`Ca2+
`
`Ca2+
`
`Y209
`
`K47
`
`H8
`
`W66
`
`A51
`
`Y209
`
`K47
`
`H8
`
`F105
`
`A51
`
`Fig. 4 Molecular basis of G/CR interaction. a Sequence alignment of LDL-R CR2 and CR3. Conserved residues are in a red box and similar residues are
`shown by red letters boxed in blue. Acidic residues involved in the binding of the Ca2+ ion are indicated by I, II, III, and IV. CR residues involved in polar
`contacts with G are labeled with gray symbols (light gray for CR2 and dark gray for CR3; dots when the contact is established via the lateral chain and
`triangles when the contact is established via the main chain) on each CR sequence. The aromatic residue which protrudes from the CR modules and
`establishes hydrophobic interactions with G is indicated by a blue arrow. b, c Close-up view on the Gth-CR interface showing the docking of G basic
`residues on the acidic patch of both CR2 (b) and CR3 (c). In both cases, the same G residues (H8, K47, Y209, and R354) are involved in the interaction.
`Residues labels on each CR domain are in italic letters when the contact is established via the main chain; putative bonds are shown as light blue dashed
`lines. d, e Close-up view on the Gth-CR interface showing the hydrophobic interactions between the aromatic residue W66 of CR2 (d) and F105 of CR3
`(e) and residues K47 and A51 of G.The color code is the same as in Fig. 3a and b
`
`mild denaturing agent), DTT and Ca2+ and renatured by dilution
`in a Ca2+
`containing buffer. The presence of Ca2+
`was manda-
`tory for correct folding of the proteins. Individual CR domains
`were then obtained by cleavage of the GST tag by prescission
`protease. All purified CR domains behave similarly in gel filtra-
`tion experiments (Supplementary Fig. 2).
`Each fusion protein was incubated at pH 8 with magnetic beads
`coated with glutathione before addition of a soluble form of the
`ectodomain of G (VSV Gth, amino acid (AA) residues 1–422,
`generated by thermolysin limited proteolysis of viral particles22)
`
`(Fig. 1b). After 20 min of incubation at 4 °C, the beads were
`washed and the associated proteins were analyzed by SDS/PAGE
`followed by Coomassie blue staining. This revealed that only CR2
`and CR3 domains are able to directly bind VSV G (Fig. 1b) at pH
`8. These results are consistent with previous data indicating that a
`monoclonal antibody (Mab) directed against LDL-R CR3 almost
`completely inhibited the VSV-triggered cytopathic effect which
`was not the case with a MAb directed against LDL-R CR614. The
`binding of Gth or VSV to fusion proteins GST-CR2 and GST-CR3
`was pH dependent and no interaction was detected at pH 6
`
`NATURE COMMUNICATIONS | (2018) 9:1029
`
`| DOI: 10.1038/s41467-018-03432-4 | www.nature.com/naturecommunications
`
`5
`
`Page 5 of 12
`
`

`

`NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03432-4
`
`wt
`
`H8A
`
`0.5%
`
`80.1%
`
`0%
`
`83.5%
`
`0.1%
`
`Y209A
`87.0%
`
`K47A
`
`K47Q
`
`R354A
`
`R354Q
`
`0.1%
`
`42.6%
`
`0.2%
`
`7.7%
`
`0.3%
`
`4.4%
`
`0%
`
`2.6%
`
`13.8%
`
`0%
`
`5.5%
`
`0.2%
`
`16,2.7%
`
`0.8%
`
`12.0%
`
`93.4%
`
`0%
`
`96.0%
`
`0.4%
`
`96.1%
`
`2%
`
`0%
`
`45.7%
`
`2.8%
`
`89.3%
`
`11.5%
`
`3.2%
`
`0%
`
`0.7%
`
`0%
`
`83.7%
`
`0.7%
`
`0.3%
`
`0%
`
`97.1%
`
`0.3%
`
`94.1%
`
`0.4%
`
`99.3%
`
`ARTICLE
`
`CR2-GST
`
`a
`
`0.6%
`
`6%
`
`0.1%
`
`3.8%
`
`0.7%
`
`2.9%
`
`2.0%
`
`94.7%
`
`3.2%
`
`96.0%
`
`5.2%
`
`CR3-GST
`
`VSV G
`
`CR2
`CR3
`
`c
`
`VSV G
`
`P-GFP
`
`BSR
`
`or
`
`Transfection
`24 h
`
`Incubation
`at different pHs
`
`Empty
`
`wt
`
`H8A Y209A K47A K47Q R354A R354Q
`
`pH 5.5
`
`pH 6.0
`
`pH 6.5
`
`pH 7.0
`
`pH 5.5
`
`pH 6.0
`
`pH 6.5
`
`pH 7.0
`
`R354A
`
`K47Q
`
`R354Q
`
`Y209A
`
`b
`
`100
`
`80
`
`60
`
`40
`
`20
`
`0
`
`% labeled cells
`
`d
`
`Empty
`
`wt
`
`H8A
`
`K47A
`
`Fig. 5 K47 and R354 on G are crucial for interaction with CR domains but not required for fusion. a Flow cytometry analysis of the expression of WT and
`mutant glycoproteins at the surface of HEK293T cells and of the binding of fluorescent GST-CR2 (top) and GST-CR3 (bottom). After 24 h of transfection,
`cell surface expression of WT and mutant G was assessed using monoclonal anti-G antibody 8G5F11 directly on living cells at 4 °C during 1 h. Cells were
`then incubated simultaneously with anti-mouse Alexa fluor 488 and the indicated GST-CRATTO550. Cells transfected with a G construct that was still able
`to bind GST-CR proteins exhibited red fluorescence due the ATTO550 dye. In each plot, the percentage of ATTO550 positive cells is indicated. b WT and
`mutant G ability to bind CR domains. The histogram indicates the mean percentage of ATTO550 positive cells for each point mutation on G (n = 3). The
`error bars represent the standard deviation. c Cartoon describing the cell–cell fusion assay. BSR cells are co-transfected with plasmids expressing VSV G
`(either WT or mutant G) and P-GFP. After 24 h of post-transfection cells are exposed for 10 min to media adjusted to the indicated pH which is then
`replaced by DMEM at pH 7.4. The cells are then kept at 37 °C for 1 h before fixation. Upon fusion, the P-GFP diffuses in the syncytia. d Cell–cell fusion
`assay of WT and mutant glycoproteins. Scale bar=200 µm
`
`(Fig. 1c, d) suggesting that only G pre-fusion conformation is able
`to bind LDL-R. Finally, GST-CR2 and GST-CR3 (but not GST-
`CR1) fluorescently labeled with ATTO550 (Fig. 1e) specifically
`recognized VSV G expressed at the surface of infected cells but
`not
`the glycoprotein of Chandipura virus (CHAV, another
`vesiculovirus), which shares 40% AA identity with VSV G
`(Fig. 1f, g). We also used isothermal titration calorimetry (ITC) to
`investigate the binding parameters of CR1, CR2, and CR3 to Gth
`in solution (Fig. 2a). Here again, no interaction between Gth and
`CR1 was detected. On the other hand, for both CR2 and CR3, the
`binding
`reactions
`appear
`to
`be
`exothermic,
`show a
`
`1:1 stoichiometry and exhibit similar Kds in the micromolar
`range (4.3±1 µM for CR3 and 7.3±1.5 µM for CR2, mean±SEM of
`three independent experiments).
`Recombinant soluble CR2 and CR3 domains, either alone or in
`fusion with GST, are also able to neutralize viral infectivity when
`incubated with the viral inoculum prior infection (Fig. 2b). In
`order to determine the IC50 of the different constructions, we
`incubated 5 × 104 VSV-eGFP infectious particles with serial
`dilutions of GST-CR2, GST-CR3, CR2 or CR3. After 15 min,
`the mixtures were transferred onto 2 × 104 BSR cells for 30 min of
`adsorption. After 4 h,
`the percentage of
`infected cells (i.e.,
`
`6
`
`NATURE COMMUNICATIONS | (2018) 9:1029
`
`| DOI: 10.1038/s41467-018-03432-4 | www.nature.com/naturecommunications
`
`Page 6 of 12
`
`

`

`NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03432-4
`
`ARTICLE
`
`n.s.
`
`n.s.
`n.s.
`
`*
`
`Ctrl
`+50 nM RAP
`
`*
`
`b
`
`1.0
`
`0.5
`
`Infectivity
`
`0.0
`
`H A P-1
`
`H A P-1
`LDL-RK O 1
`
`H A P-1
`LDL-RK O 2
`
`150
`
`100
`
`75
`
`50
`
`HAP-1
`
`HAP-1
`LDL-RKO1
`
`HAP-1
`LDL-RKO2
`
`HAP-1 LDL-RKO1
`LR P3
`LR P4
`LR P5
`LR P6
`
`LR P1
`
`LR P2
`
`LR P8
`
`SorLA
`
`vLDLR
`
`HAP-1 LDL-RKO2
`LR P4
`LR P5
`LR P6
`LR P8
`
`SorLA
`
`vLDLR
`
`LR P1
`
`LR P2
`
`LR P3
`
`a
`
`LDL-R
`
`Tub
`
`c
`
`600
`500
`400
`300
`200
`
`100
`
`Fig. 6 LDL-R presence on the cell surface is not mandatory for VSV infection. a Analysis of LDL-R expression in wild-type HAP-1, HAP-1 LDL-RKO1 and
`HAP-1 LDL-RKO2 cells. The immunoblot was performed on crude cell extracts and revealed with chicken anti LDL-R (GW22458A from Sigma). α-Tubulin
`(tub) was also immunoblotted as a loading control. The LDL-R band migrates with an apparent MW of 150 kDa due to the presence of glycosylation sites
`(predicted MW without oligosaccharides is 95 kDa). b Effect of the RAP protein on the susceptibility of LDL-R deficient HAP-1 cells to VSV-eGFP infection.
`VSV-eGFP was used to infect HAP-1 and HAP-1 LDL-RKO cells in the presence or not of RAP (50 nM). Infectivity was determined by counting the number
`of cells expressing eGFP using a flow cytometer. Data depict the mean with standard deviation for experiments performed in triplicate. p values were
`determined using an unpaired Student's t test (*p < 0.05; n.s. non-significant). c Expression of LDL-R family members in HAP-1 LDL-RKO cell lines evaluated
`by RT-PCR
`
`expressing the eGFP) was determined by flow cytometry. For
`both CR2 and CR3, the IC50 is about 15 µM and decreases to
`about 20 nM when fusion GST-CR constructions are used, thanks
`to their dimeric nature which induces avidity-enhanced binding
`to the viral glycoproteins (Fig. 2c).
`
`Crystal structures of Gth in complex with LDL-R CR domains.
`We crystallized Gth in complex with either CR2 or CR3. The
`crystal of the complex Gth-CR3 (diffracting up to 3.6 Å) belongs
`to space group P622. In this crystal form, the lattice organization
`of Gth molecules is identical to that of the crystal of Gth alone in
`pre-fusion state8 (Supplementary Fig. 3B). On the other hand, the
`Gth-CR2 crystal form (diffracting up to near 2.2 Å) belongs to the
`H32 space group (Supplementary Fig. 3A).
`The binding site of CR domains on G is the same in both
`crystal forms (Fig. 3a, b). CR2 (resp. CR3) interaction with G
`buries 1590 Å2 (resp. 1450 Å2) of the two molecules’ surfaces. It is
`essentially constituted by segments going from residues 8 to 10
`and 350 to 354 in the trimerization domain (TrD), 180 to 184 in
`the pleckstrin homology domain (PHD) and 47 to 50 in segment
`S2. Those three segments are rearranged in G post-fusion
`conformation7, which explains the inability of CR domains to
`bind G at low pH (Fig. 3c, d). It is worth noting that the
`orientation of CR domains in both complexes is optimal for the
`interaction between G and the open conformation of LDL-R
`when both proteins are anchored in their respective membranes
`(Fig. 3e).
`Two basic residues of G (H8 from the TrD and K47 from
`PHD) are pointing toward two acidic residues which belong to
`
`the octahedral calcium cage of the CR domains (D69 and D73 on
`CR2; D108 and D112 on CR3 labeled I and II—Fig. 4a). The side
`chain of K47 also establishes an H-bond with the amide group
`of Q71 (in CR2) and a salt bridge with the acidic group of D110
`(in CR3). Two other residues, Y209 (which makes hydrogen
`bonds with the C=O group of D69 on CR2, and with the C=O
`group of D108 and the carboxyl group of D110 on CR3) and
`R354 (which establishes contacts with the main chain of both
`CR domains), seem also to be key for the interaction (Fig. 4b,
`c). In both CR domains, the side chain of an aromatic residue
`(W66 in CR2 and F105 in CR3), which protrudes from the
`receptor module, also contributes to the stability of
`the
`complex by establishing hydrophobic interactions with the
`aliphatic part of K47 side chain and with the side chain of A51
`of G (Fig. 4d, e).
`
`K47 and R354 of G are crucial for LDL-R CR domains binding.
`To investigate their contribution to LDL-R CR domains binding,
`we mutated residues H8, K47, Y209, and R354 of G into an
`alanine or a glutamine. HEK cells were transfected with pCAGGS
`plasmids encoding wild-type or mutant VSV G glycoproteins
`(WT, H8A, K47A, K47Q, Y209A, R354A, and R354Q). After 24 h
`of post-transfection, the cells were incubated with MAb 8G5F11
`directed against G ectodomain. Then, green fluorescent anti IgG
`secondary antibodies and GST-CR proteins fluorescently labeled
`with ATTO550 were simultaneously added. Immunofluorescence
`labeling indicated that WT and all G mutants are efficiently
`transported to the cell surface (Fig. 5a). Mutants H8A and Y209A
`bind GST-CR proteins as WT G whereas the other mutants are
`
`NATURE COMMUNICATIONS | (2018) 9:1029
`
`| DOI: 10.1038/s41467-018-03432-4 | www.nature.com/naturecommunications
`
`7
`
`Page 7 of 12
`
`

`

`ARTICLE
`
`NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03432-4
`
`affected in their binding ability (Fig. 5a, b). Mutants K47Q,
`R354A and R354Q bind neither GST-CR2 nor GST-CR3. Finally,
`although no interaction is detected between mutant K47A and
`CR3, a residual binding activity is observed between this mutant
`and CR2 (Fig. 5a, b).
`
`We also checked the fusion properties of all the mutants
`(Fig. 5c). For this, BSR cells were transfected with pCAGGS
`plasmids encoding wild-type or mutant VSV G glycoproteins.
`The cells expressing mutant G protein have a fusion phenotype
`similar to that of WT G (Fig. 5d). This confirms that the
`mutant glycoproteins are correctly folded and demonstrates
`
`a
`
`HEK
`
`VSV Gmut
`
`VSVΔG-GFP/VSV Gwt*
`
`24 h
`
`Infection
`
`16 h
`
`VSVΔG-GFP / VSV Gmut
`
`wt
`
`H8A
`
`K47A
`
`K47Q
`
`Y209A
`
`R354A
`
`R354Q
`
`BSR
`
`1
`
`1.3
`
`1.9
`
`2
`
`1.5
`
`1
`
`0.5
`
`0
`
`c
`
`Normalized G/M ratio
`
`17
`
`10
`
`9.6
`
`15
`
`471
`
`wt
`
`H8A
`
`K47A
`
`E m pty
`
`w T
`
`H8A
`
`K47A
`
`K47Q
`
`Y209A
`
`R354A
`
`R354Q
`
`G M
`
`1.0
`
`0.8
`
`0.6
`
`0.4
`
`0.2
`
`0.0
`
`Infectivity
`
`1
`
`HEK
`
`1.6
`
`3
`
`355
`
`84
`
`71
`
`116
`
`77
`
`wt
`
`H8A
`
`b
`kD
`
`250
`150
`100
`75
`
`50
`37
`25
`20
`15
`10
`
`d
`
`Infectivity
`
`1.0
`
`0.8
`
`0.6
`
`0.4
`
`0.2
`
`0.0
`
`E m pty
`
`K47A
`
`Y209A
`
`R354A
`
`K47Q
`
`R354Q
`
`E m pty
`
`Y209A
`
`R354A
`
`K47Q
`
`R354Q
`
`1
`
`S2
`
`5.6
`
`210
`
`E m pty
`
`15
`
`122
`
`wt
`
`H8A
`
`K47A
`
`*
`
`I
`
`II
`
`Y209A
`*
`
`87
`
`56
`
`76
`
`R354A
`
`K47Q
`
`R354Q
`
`IVIII
`
`*
`
`1.0
`
`0.8
`
`0.6
`
`0.4
`
`0.2
`
`0.0
`
`*
`
`Infectivity
`
`1
`
`1.5
`
`CHO
`
`1.9
`
`40
`
`28
`
`25
`
`25
`
`1188
`
`E m pty
`
`1.0
`
`0.8
`
`0.6
`
`0.4
`
`0.2
`
`0.0
`
`Infectivity
`
`e
`
`wt
`
`H8A
`
`K47A
`
`Y209A
`*
`
`R354A
`
`K47Q
`
`R354Q
`
`*
`
`LDLR CR2
`LDLR CR3
`
`47
`88
`
`VLDLR
`VLDLR
`LRP1
`LRP1
`LRP1B
`LRP2
`LRP2
`LRP2
`LRP3
`LRP4
`PGBM
`Vit. Rec
`
`33
`239
`3575
`3613
`967
`183
`1109
`1149
`455
`148
`285
`1283
`
`H H H H H H H H H H H H H
`
`D
`
`8
`
`NATURE COMMUNICATIONS | (2018) 9:1029
`
`| DOI: 10.1038/s41467-018-03432-4 | www.nature.com/naturecommunications
`
`Page 8 of 12
`
`

`

`NATURE COMMUNICATIONS | DOI: 10.1038/s41467-018-03432-4
`
`ARTICLE
`
`that it is possible to uncouple G fusion activity and receptor
`recognition.
`
`Other LDL-R family members are alternative receptors of VSV.
`We infected two HAP-1 cell-lines (HAP-1 LDL-RKO1 and HAP-1
`LDL-RKO2) in which the LDL-R gene has been knocked out
`(Fig. 6a and Supplementary Fig. 4). Both were as susceptible to
`VSV infection as WT HAP-1 cells (Fig. 6b). This demonstrates
`that VSV receptors other than the LDL-R are present at the
`surface of HAP-1 cells.
`To evaluate the role of other LDL-R family members23 as VSV
`receptors, we took advantage of the properties of the receptor-
`associated protein (RAP), a common ligand of all LDL-R family
`members24–26 which blocks ligand binding to all LDL-R family
`members with the exception of LDL-R itself14,26. RAP signifi-
`cantly inhibits VSV infection in HAP-1 LDL-RKO cell lines but
`not in WT HAP-1 cells (Fig. 6b). Those results are consistent with
`previous data suggesting that VSV can use other LDL-R family
`members as alternative receptors14. Indeed, RT-PCR on purified
`RNA revealed the expression of several LDL-R family members
`including LRP1, LRP3, LRP4, LRP6, and sortilin (SorLA) in both
`HAP-1 LDL-RKO cell
`lines. Messengers corresponding to the
`vLDL-R were only detected in HAP-1 LDL-RKO2 (Fig. 6c).
`
`G mutants affected in CR binding cannot rescue VSVΔG-GFP.
`We then examined whether the mutant glycoproteins described
`above are able to sustain viral infection. We used a recombinant
`VSV (VSVΔG-GFP) in which the G envelope gene was replaced
`by the green fluorescent protein (GFP) gene and which was
`pseudotyped with the VSV G glycoprotein27,28. This pseudotyped
`recombinant was used to infect HEK cells either transfected or
`not transfected by a plasmid encoding WT or mutant glycopro-
`teins29. After 16 h, the infected cells supernatant was collected
`(Fig. 7a). Mutant glycoproteins incorporation into the envelope of
`the particles present in the supernatant was verified by western
`blot (Fig. 7b, c) and the infectivity of the pseudotyped particles
`was analyzed in different cell lines (mammalian HEK, BSR, CHO
`and Drosophila S2 cells) by counting the cells expressing GFP by
`flow cytometry 4 h p.i. (Fig. 7d). Mutants K47A, K47Q, R354A,
`and R354Q did not rescue the infectivity of VSVΔG-GFP.
`Compared to WT G, the infectivity decreased by a factor of 10 up
`to 120 (Fig. 7d). The decrease was more important in HEK and S2
`cell lines than in the two hamster cell lines. In mammalian cell
`lines, mutants H8A and Y209A can rescue the infectivity of
`VSVΔG-GFP at a level similar to that of WT. This is not the case
`in S2 cell line, where their infectivity significantly decreased (by a
`factor of 15 for mutant H8A and ~6 for Y209A) (Fig. 7d).
`
`As the fusion activity of the mutants is unaffected (Fig. 5d), the
`loss of infectivity of pseudotypes bearing a mutant glycoprotein
`can be safely attributed to their disability to recognize a cellular
`receptor. These results indicate that mutants K47A, K47Q,
`R354A, and R354Q which are unable to bind LDL-R CR domains
`are also severely impaired in their ability to bind other VSV
`receptors.
`
`Discussion
`LDL-R has been demonstrated to be the major entry port of VSV
`and lentivirus pseudotyped by VSV G14. Here, we demonstrate
`that VSV G is able to bind two CR domains of the LDL-R with
`similar affinities. The biological relevance of this interaction was
`demonstrated by the ability of both CR2 and CR3 to inhibit VSV
`infection. The crystal structures of VSV G in complex with CR2
`and CR3 reveal that they both occupy the same site at the surface
`of the glycoprotein in its pre-fusion conformation and that the
`same G residues ensure the correct anchoring of the CR domains.
`This binding site is split apart when G is in its post-fusion con-
`formation, which explains why G is unable to bind CR domains at
`low pH. This may disrupt the interaction between G and LDL-R
`in the acidic endosomal lumen and favor the transport of the
`virion to an appropriate fusion site.
`CR domain recognition by VSV G involves basic residues H8
`and K47, pointing toward the calcium-coordinating acidic resi-
`dues, and R354 which interacts with C=O groups of the main
`chain. This mode of binding is very similar to what is observed
`for endogenous ligand recognition by CR domains of the LDL-R
`family members17,30,31 and,
`indeed, mutant glycoproteins in
`which either K47 or R354 is replaced by an alanine or a gluta-
`mine, are unable to bind CR domains. It is worth noting that
`those key residues are not conserved among vesiculoviruses.
`Therefore, the use of LDL-R as a viral receptor cannot be gen-
`eralized to the other members of the genus. Indeed, we have
`shown that CHAV G, which does not possess basic residues in
`positions corresponding to VSV residues 47 and 354, does not
`bind CR domains.
`VSV G binds only CR2 and CR3. All CR domains have the
`same fold and all form a calcium cage involving conserved acidic
`residues. The acidic residues in position I and II, which play a key
`role in the interaction with G, are conserved except in CR7 which
`has an asparagine in position I (Supplementary Fig. 1). Therefore,
`they are poorly discriminant for the interaction between a given
`CR domain and G. Similarly, residue R354 of G interacts with the
`main chain of both CR2 and CR3 and the influence of CR
`domains amino acid sequence on this interaction is difficult to
`predict. However, the aromatic residues (W66 in CR2 and F105
`in CR3) establishing hydrophobic interactions with the aliphatic
`
`Fig. 7 Infectivity of VSVΔG-GFP virus pseudotyped with mutated G in vario