Molecular determinants of Arc oligomerization and …

 

 

bioRxiv preprint

this version posted June 12, 2019.
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
CC-BY-NC-ND 4.0 International license
.

The copyright holder for this preprint (which was not

https://doi.org/10.1101/667956

doi:

a

;

Molecular determinants of Arc oligomerization

and formation of virus-like capsids

Maria Steene Eriksen1,2, Oleksii Nikolaienko1,2, Erik Ingmar Hallin1, Sverre Grødem1,2, Helene J.
Bustad1,2, Marte Innselset Flydal1,2, Rory O’Connell3, Tomohisa Hosokawa4, Daniela Lascu1,2,
Shreeram Akerkar1,2, Jorge Cuéllar5, James J. Chambers6, Gopinath Muruganandam7,8, Remy Loris7,8,
Tambudzai Kanhema1,2, Yasunori Hayashi4, Margaret M. Stratton3, José M. Valpuesta5,
Petri Kursula1,9, Aurora Martinez1,2, Clive R. Bramham1,2,*

From:
1Department of Biomedicine, University of Bergen, Bergen, Norway
2KG Jebsen Centre for Neuropsychiatric Disorders, University of Bergen, Bergen, Norway
3Department of Biochemistry and Molecular Biology, University of Massachusetts Amherst, Amherst MA, USA
4Department of Pharmacology, Kyoto University Graduate School of Medicine, Kyoto, Japan
5Centro Nacional de Biotecnología (CNB-CSIC), Darwin 3, 28049 Madrid, Spain
6Institute for Applied Life Sciences, University of Massachusetts, Amherst, MA, USA
7VIB-VUB Center for Structural Biology, Vlaams Instituut voor Biotechnologie, 1050 Brussels, Belgium
8Structural Biology Brussels, Department of Bioengineering Sciences, Vrije Universiteit Brussel, 1050 Brussels,
Belgium
9Faculty of Biochemistry and Molecular Biology, University of Oulu, Oulu, Finland

*Correspondence to: Clive Bramham, MD, PhD. Department of Biomedicine, University of Bergen, 5009 Bergen,
Norway. Phone: +47 55 58 60 32. E-mail: clive.bramham@uib.no

Keywords: synaptic plasticity, memory, protein oligomerization, protein biochemistry, single-molecule TIRF
microscopy, electron microscopy, small-angle X-ray scattering

1

bioRxiv preprint

this version posted June 12, 2019.
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
CC-BY-NC-ND 4.0 International license
.

The copyright holder for this preprint (which was not

https://doi.org/10.1101/667956

doi:

a

;

ABSTRACT

Expression of activity-regulated cytoskeleton-associated protein (Arc) is critical for long-term synaptic

plasticity, memory formation, and cognitive flexibility. The ability of Arc to self-associate and form

virus-like capsid structures implies functionally distinct oligomeric states. However, the molecular

mechanism of Arc oligomerization is unknown. Here, we identified a 28-amino-acid region necessary

and sufficient for Arc oligomerization. This oligomerization region is located within the second coil of a

predicted anti-parallel coiled-coil in the N-terminal domain (NTD). Using alanine scanning mutagenesis,

we found a 7-amino-acid motif critical for oligomerization and Arc-mediated transferrin endocytosis in

HEK cells. Intermolecular fluorescence lifetime imaging in hippocampal neurons confirmed self-

association mediated by the motif. To quantify oligomeric size, we performed a single-molecule

photobleaching analysis of purified Arc wild-type and mutant. This analysis revealed a critical role for

the NTD motif in the formation of higher-order Arc oligomers (30-170 molecules). Moreover, assembly

of higher-order wild-type Arc oligomers was significantly enhanced by addition of GFP RNA. Purified

wild-type Arc formed virus-like capsids, as visualized by negative-stain EM, and was estimated by light

scattering analysis to contain 40-55 Arc units. In contrast, mutant Arc formed a homogenous dimer

population as demonstrated by single-molecule TIRF imaging, size-exclusion chromatography with

multi-angle light scattering analysis, small-angle X-ray scattering analysis, and single-particle 3D EM

reconstruction. Thus, the dimer appears to be the basic building block for assembly. Herein, we show

that the NTD motif is essential for higher-order Arc oligomerization, assembly of virus-like capsid

particles, and facilitation of oligomerization by exogenous RNA.

SIGNIFICANCE

Arc protein is rapidly expressed in neurons in response to synaptic activity and plays critical roles in

synaptic plasticity, postnatal cortical developmental, and memory. Arc has diverse molecular functions,

which may be related to distinct oligomeric states of the protein. Arc has homology to retroviral Gag

protein and self-assembles into retrovirus-like capsid structures that are capable of intercellular transfer

of RNA. Here, we identified a motif in the N-terminal coiled-coil domain of mammalian Arc that

mediates higher-order oligomerization and formation of virus-like capsids. The basic building block is

the Arc dimer and exogenous RNA facilitates further assembly. The identified molecular determinants

of Arc oligomerization will help to elucidate the functional modalities of Arc in the mammalian brain.

2

bioRxiv preprint

this version posted June 12, 2019.
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
CC-BY-NC-ND 4.0 International license
.

The copyright holder for this preprint (which was not

https://doi.org/10.1101/667956

doi:

a

;

INTRODUCTION

Activity-regulated cytoskeletal associated protein (Arc) has emerged as a key regulator of neuronal

plasticity, memory, and postnatal cortical development (Bramham et al., 2010; Nikolaienko et al., 2018;

Shepherd and Bear, 2011). Arc is induced as an immediate early gene, and the neuronal activity-induced

Arc RNA and protein are subject to rapid turnover, indicating a highly dynamic mode of action. Loss of

function studies support a role for Arc in long-term potentiation (LTP) and long-term depression (LTD)

of synaptic transmission and homeostatic synaptic scaling. These diverse responses are mediated by

distinct Arc protein-protein interaction complexes in the postsynaptic dendritic compartment and the

neuronal nucleus (Chowdhury et al., 2006; DaSilva et al., 2016; Korb et al., 2013; Nair et al., 2017;

Okuno et al., 2012; Zhang et al., 2015). Thus, convergent lines of evidence support a role for Arc as a

signaling hub protein and cell-autonomous organizer of synaptic plasticity (Nikolaienko et al., 2018).

The mechanisms that dictate Arc function at the molecular level are poorly understood. Post-

translational modification of Arc by SUMOylation (Craig et al., 2012; Nair et al., 2017) and ERK-

catalyzed phosphorylation (Nikolaienko et al., 2017) are implicated in the regulation of protein-protein

interactions and subcellular localization, while Arc turnover is regulated by ubiquitination, acetylation,

as well as GSK-catalyzed phosphorylation (Gozdz et al., 2017; Greer et al., 2010; Lalonde et al., 2017;

Mabb et al., 2014; Wall et al., 2018). In addition, biochemical studies show that recombinant Arc protein

is capable of reversible self-association (Byers et al., 2015; Myrum et al., 2015). It has also been shown

that purified human Arc forms higher-order oligomeric species, dependent on ionic strength, but

reverses to monomers and dimers at low ionic strength (Myrum et al., 2015). This reversible

oligomerization, as opposed to nonspecific aggregation, raised the possibility that Arc function is related

to its oligomeric state.

Recent advances highlight a structural and functional relationship between Arc and retroviral

Gag polyprotein. Arc was identified in a computational search for domesticated retrotransposons

harboring Gag-like protein domains (Campillos et al., 2006). Biochemical studies showed that

mammalian Arc has a positively charged N-terminal domain (NTD) and a negatively charged C-

terminal domain (CTD), separated by a flexible linker (Myrum et al., 2015). Crystal structure analysis

of the isolated CTD revealed two lobes, both with striking 3D homology to the capsid (CA) domain of

HIV Gag (Zhang et al., 2015). In retroviruses, self-association of CA allows assembly of Gag

polyproteins into the immature capsid shell (Lingappa et al., 2014; Perilla and Gronenborn, 2016).

3

bioRxiv preprint

this version posted June 12, 2019.
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
CC-BY-NC-ND 4.0 International license
.

The copyright holder for this preprint (which was not

https://doi.org/10.1101/667956

doi:

a

;

Remarkably, recombinant Arc from fruit fly and rat was subsequently shown to self-assemble into

spheroid particles with resemblance to HIV Gag capsids (Ashley et al., 2018; Pastuzyn et al., 2018). The

Arc capsids are released in extracellular vesicles and capable of transmitting RNA cargo to recipient

cells (Ashley et al., 2018; Pastuzyn et al., 2018). These discoveries implicate Arc as an endogenous

neuronal retrovirus, and oligomeric assembly of Arc into virus-like capsids mediates the capture and

intercellular transfer of RNA (Parrish and Tomonaga, 2018; Shepherd, 2018).

A recent structural analysis of full-length monomeric human Arc shows a compact shape, in

which the oppositely charged NTD and CTD interact (Hallin et al., 2018). Interestingly, Drosophila Arc

has a CA-like CTD domain but lacks the NTD found in mammals (Zhang et al., 2015). The Arc NTD is

expected to have evolved from the matrix (MA) domain of the Gag polyprotein (Campillos et al., 2006).

There is also a functional similarity, as both the Arc NTD and the Gag MA domain mediate binding to

phospholipid membranes (Hallin et al., 2018; Mailler et al., 2016). However, important differences may

exist, as structural analysis of the Arc NTD indicates an antiparallel coiled-coil, which is not present in

retroviral MA (Hallin et al., 2018).

A major goal is to elucidate the relationship between Arc as a dynamic mediator of intracellular

signaling versus its newly described function as a virus-like capsid. We therefore sought to identify

molecular determinants of Arc self-association and assembly into capsids. Our approach combines

biochemical and physicochemical methods (affinity purification, in situ protein crosslinking, size-

exclusion chromatography (SEC), dynamic light scattering (DLS)), as well as fluorescence lifetime

imaging, single-molecule photobleaching analysis, and electron microscopy (EM). We identified a 28-

amino-acid stretch in the Arc NTD that is both necessary and sufficient for oligomerization. The critical

oligomerization region is located within the second coil of the predicted anti-parallel coiled-coil.

Alanine scanning mutagenesis identified a 7-amino-acid motif underlying oligomerization above the

dimer stage and leading to capsid formation, as validated by EM. Surprisingly, similarly to the

nucleocapsid domain of HIV Gag, Arc oligomerization is greatly enhanced by exogenous RNA in a

manner dependent on the second coil motif. Our findings shed light on the molecular mechanisms of

mammalian Arc oligomerization and capsid assembly.

4

bioRxiv preprint

this version posted June 12, 2019.
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
CC-BY-NC-ND 4.0 International license
.

The copyright holder for this preprint (which was not

https://doi.org/10.1101/667956

doi:

a

;

RESULTS

Arc oligomerization is mediated by the NTD second coil

First, we sought to map the regions of the Arc protein mediating oligomerization using a stringent

affinity purification assay. We co-expressed two variants of Arc in a human embryonic kidney cell line

(HEK293FT):1) Arc with an N-terminal fusion to mTurquoise2 (mTq2) and 2) Arc with a a C-terminal

fusion to StrepII tag (Fig.1A). Next, mTq2-fused full-length Arc (1-396) was coprecipitated with

StrepII-tagged Arc, indicating complex formation (Fig. 1C). Specificity was confirmed by the absence

of mTq2-positive bands in purifications from cells transfected with StrepII-Arc and the empty mTq2

vector. Arc contains five cysteine residues (C34, C94, C96, C98, C159), which we suspected might form

disulfide-linked oligomeric species. However, substituting all five cysteines with serines did not affect

the Arc-Arc interaction, showing that oligomerization is not a result of disulfide bond formation (Fig.

1C). We next co-expressed truncated versions of Arc fused to mTq2 to determine the minimal regions

needed for association (Fig.1B). When StrepII-tagged full-length Arc was coexpressed with mTq2-fused

N-terminal region (1-140), linker (135-216) or C-terminal region (208-396), only the N-terminal region

showed binding (Fig. 1D). The involvement of the N-terminal region in self-association agrees with our

previous SAXS data showing that the isolated Arc CTD is fully monomeric (Hallin et al., 2018).

According to secondary structure predictions, the Arc NTD has a predicted coiled-coil region

which can be divided into two alpha coils, with residues 20-77 and 78-140 corresponding to the first and

second coil, respectively. Our previous structural work indicates an elongated, anti-parallel coiled-coil

structure of the NTD (Hallin et al., 2018). We therefore examined co-purification of full-length Arc with

each of the NTD coils and found that only the second coil (residues 78-140) interacted with Arc (Fig.

1E). Similar binding was obtained with a Cys to Ser mutant (3C/3S, residues 94, 96, and 98) of the

second coil, again ruling out formation of disulfide-linked oligomers (Fig 1E). The second coil co-

purified with the isolated Arc NTD but not with the isolated linker region or CTD (Fig. 1F).

Furthermore, high-affinity binding between StrepII-tagged 78-140 and mTq2-tagged 78-140 was

observed, demonstrating self-association of the NTD second coil (SI Appendix, Fig. S1).

5

bioRxiv preprint

this version posted June 12, 2019.
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
CC-BY-NC-ND 4.0 International license
.

The copyright holder for this preprint (which was not

https://doi.org/10.1101/667956

doi:

a

;

Identification of an oligomerization motif in the Arc NTD second coil

Next, we sought to delineate which residues within the NTD second coil (residues 78-140) mediate Arc-

Arc interaction. We first generated a series of deletion mutants of the second coil in which 18 amino

acid stretches were removed, starting with every 9th consecutive residue, giving an overlap of 9 residues

between the deleted regions (SI Appendix, Table S1). Deletion mutants Δ78-95 and Δ114-131 did not

bind full-length Arc (Fig. 2A), however, the Δ78-95 mutant was associated with decreased expression of

the peptide. We then generated substitution variants, in which non-overlapping stretches of seven amino

acids were mutated to alanine (Fig. 2B, SI Appendix, Table S1). With the exception of s78-84A, which

exhibited impaired expression, the substitution variants expressed at levels similar to wild-type (WT)

control. Affinity purification analysis showed that all substitutions within the central region of the

second coil, from residue Gln99 to Trp126, inhibited binding of the second coil to full-length Arc (Fig.

2B). Importantly, oligomerization was similarly inhibited or abolished when mutations were introduced

into full-length Arc (Fig.2C). Most notably, alanine substitution of the sequence 113MHVWREV119

(Arcs113-119A) abolished binding to Arc-StrepII.

As a complementary approach, we mapped the NTD second coil interaction site using a peptide-

tiling array of Arc region 71-147. Each spot on the array displays a 21-residue peptide, with 7-residue

overlap between sequences. The array was incubated with purified GST-fused Arc 78-140 (complete

second coil) and immunoblotted with anti-GST antibodies. Binding was observed with tiling peptides

spanning Arc region 99-126 (Fig. 2D). Pure GST did not bind to immobilized peptides. These in vitro

binding results identify an oligomeric interface in the second coil and corroborate findings from affinity

purification of tagged Arc from HEK cells.

Taken together, Arc region 99-126 is necessary and sufficient for Arc oligomerization (Fig. 2E

and 2F). We refer to this 28-amino-acid stretch as the oligomerization region, while the critical Arc 113-

119 sequence is referred to as the oligomerization motif. The position of the oligomerization region in

the context of the SAXS hybrid 3D model of the Arc monomer (Hallin et al. 2018) is shown in Fig 2E.

The oligomerization region is amphipathic with hydrophobic and hydrophilic residues on opposite sides

of the predicted helix (Fig. 2F). Notably, the oligomerization region has an isoelectric point of 8.6 and is

flanked by patches of dense positive charge in the highly basic NTD (pI of 9.6; SI Appendix, Fig. S2).

6

bioRxiv preprint

this version posted June 12, 2019.
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
CC-BY-NC-ND 4.0 International license
.

The copyright holder for this preprint (which was not

https://doi.org/10.1101/667956

doi:

a

;

The Arc oligomerization region forms hexamers

The oligomerization region is predicted to adopt an α-helical conformation and self-associate. To assess

secondary structure and detect possible oligomeric forms of the oligomerization region, we performed

circular dichroism (CD) spectrum analysis, SAXS, and SEC with multiangle light-scattering (SEC-

MALS) analysis on a synthetic peptide (residues 99-132) containing the defined oligomerization region

(99-126) and six additional C-terminal residues, which based on our analysis of substitution mutants

may contribute to oligomerization. The CD spectrum indicated high α-helical content as expected (SI

Appendix, Fig. S3A). SEC showed the formation of low-order oligomers by the peak position at an

elution volume of ~16 ml (SI Appendix, Fig. S3B) and a lack of larger complexes. MALS indicated a

size of 27 kDa, corresponding to a hexamer. SAXS analysis of the same peptide also showed a

hexameric assembly and an elongated shape with P32 symmetry (SI Appendix, Fig. S3C-F). These

experiments demonstrate a strong intrinsic property of the peptide to oligomerize in the absence of

cellular factors, such as protein binding partners, membranes, or nucleic acids.

Fluorescence lifetime FRET imaging in hippocampal slices confirms self-association of the second

coil mediated by the oligomerization motif

To assess the second coil-mediated interaction in live neurons, we performed fluorescence lifetime

FRET imaging (FLIM-FRET) in CA1 pyramidal cells of organotypic rat hippocampal slice cultures.

Slices were transfected by gene gun with plasmids expressing the second coil (residues 78-140) fused to

GFP (donor) and mCherry (acceptor) (Fig. 3A). In FLIM-FRET, a decrease in fluorescence lifetime of

the GFP donor, as measured by time-correlated single-photon counting, indicates increased protein-

protein interaction. The fluorescence lifetime of GFP-Arc second coil WT was equal to the positive

control, a GFP-mCherry fusion that gives constitutive FRET (Fig. 3B), but significantly lower than the

negative control (GFP-P2A-mCherry) in which the two fluorophores are separated by a self-cleaving

peptide. Next, we examined the role of the 113-119 oligomerization motif in mediating the interaction

between coils. When the mutant coil was used as the FRET acceptor, the interaction was significantly

reduced compared to WT control (Fig. 3B). These live-imaging data show strong self-association of the

second coil in hippocampal neurons and confirm the critical role of the 113-119 motif in

oligomerization. However, we also note that the FRET signal in the second coil mutant did not reach the

level of the negative control, indicating residual oligomerization.

7

bioRxiv preprint

this version posted June 12, 2019.
certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under
CC-BY-NC-ND 4.0 International license
.

The copyright holder for this preprint (which was not

https://doi.org/10.1101/667956

doi:

a

;

In situ protein crosslinking reveals persistence of Arc dimer in oligomerization motif mutant

The findings from hippocampal slices suggested that alanine substitution of the 113-119 motif does not

completely abolish Arc interactions, which contrasts with our findings from affinity purification of Arc

expressed in HEK cells. We therefore considered that cell lysis and homogenization might disrupt weak

oligomeric interactions, as previously shown for α-synuclein (Dettmer et al., 2013). To detect oligomeric

species in situ, we examined Arc interactions in live HEK cells exposed to protein crosslinkers. Cells

were transfected with constructs expressing mTq2-fused ArcWT, Arcs92-98A, Arcs113-119A, or empty vector.

In the main protocol used, cells were treated for 10 min with 100 μM disuccinimidyl glutarate (DSG)

before harvesting (Fig. 4). A strong Arc (and GFP) immunoreactive band was detected at 75 kDa,

corresponding to monomeric mTq2-tagged Arc (Fig. 4A). In crosslinked preparations only, Arc

immunoreactive bands were detected at approximately 180 kDa and 280 kDa, corresponding to dimers

and trimers, respectively (Fig 4A). The dimer was prominent for all constructs, and there was no

difference between constructs in the dimer/monomer ratio, indicating no impact of the 113-119 motif at

the level of dimer formation (Fig. 4B). However, the trimer/dimer ratio was significantly reduced in

Arcs113-119A relative to both Arcs92-98A and ArcWT (Fig. 4B). These results reveal prominent cellular

expression of an Arc dimer and further imply a specific role for the Arc 113-119 motif in the assembly

of oligomers above the dimer stage.

Mutation of Arc oligomerization domain impairs transferrin endocytosis.

Arc is an adaptor protein for clathrin-mediated endocytosis and facilitates cellular uptake of transferrin

(Chowdhury et al., 2006). The Arc NTD second coil harbors an endophilin binding site determinant

(residues 89-100) involved in endocytosis (Chowdhury et al., 2006), and the same region contains a

cysteine cluster 94CLCRC98 involved in palmitoylation and membrane targeting of Arc (Barylko et al.,

2017). We therefore explored whether oligomerization mediated by the NTD second coil contributes to

endocytosis, as assessed by uptake of Alexa Fluor 647-conjugated transferrin in HEK cells. As expected,

expression of ArcWT enhanced transferrin uptake relative to empty vector control (SI Appendix, Fig. S4).

Expression of the endophilin binding site mutant, Arcs92-98A, failed to increase transferrin uptake (SI

Appendix, Fig. S4), although the protein successfully forms oligomers, as shown by affinity purification

(Fig 2B and 2C). Notably, the oligomerization-deficient Arc mutants (Δ114-131 and s113-119A) failed

to facilitate transferrin uptake (SI Appendix, Fig. S4). As levels of monomer and dimer expression did

not differ between Arcs113-119A, Arcs92-98A and ArcWT, these results support a role for oligomerization

8