**1. Introduction**

The genus *Orthohantavirus*, within the *Hantaviridae* family of the *Bunyavirales* order, comprises over 36 species of enveloped segmented negative stranded RNA viruses [1] that are found throughout the globe, with many capable of causing devastating human disease [2]. Orthohantaviruses segregate into either New World (NW) and Old World (OW) clades based on their country of isolation, with viruses of the OW clade being widespread throughout Eurasia, whereas NW viruses are found within the Americas. Orthohantaviruses are mostly associated with a specific rodent host, in which they cause persistent infections [3]. In contrast, many OW and NW orthohantaviruses are associated with pathologically distinct disease outcomes in humans. Typically, viruses of the OW clade, such as the Hantaan virus (HTNV) and the Seoul virus (SEOV), are associated with a haemorrhagic fever with renal syndrome (HFRS), whereas NW viruses including Andes virus (ANDV) and Sin Nombre virus (SNV) are the causative agents of hantavirus cardiopulmonary syndrome (HCPS). Both syndromes are characterised by extensive vascular leakage, with human mortality rates ranging from 0.1–10% for HFRS to around 40% for HCPS [4]. In rodent hosts, few cytopathic effects are apparent and the

animals develop asymptomatic or subclinical persistent infections [5]. This outcome is mirrored in cell culture systems that become persistently infected [6], reflecting the ability of hantaviruses to evade pathogen surveillance and innate immune defences.

The orthohantavirus genome comprises three virion-associated negative sense RNA segments (vRNAs) named small (S), medium (M) and large (L) based on their relative sizes [7]. The S segmen<sup>t</sup> encodes the nucleocapsid protein (NP), which coats the vRNAs forming helical ribonucleoprotein (RNP) complexes that act as templates for RNA synthesis. The M segmen<sup>t</sup> encodes a glycoprotein precursor (GPC) that is processed by cellular proteases into Gn and Gc proteins that form envelope spikes with primary roles in virus attachment and entry. The L segmen<sup>t</sup> encodes the large (L) protein that is the catalytic component of the RNA-dependent RNA polymerase (RdRp), responsible for two distinct RNA synthesis activities; mRNA transcription to yield a single 5 capped mRNA from each vRNA template, and RNA replication to amplify the three vRNA segments via complementary (cRNA) intermediates. Hantavirus transcription involves the acquisition of short 5 capped oligoribonucleotides from host cell mRNAs by an RdRp-associated endonuclease, which are then used by the RdRp to prime mRNA synthesis using a 'prime-and-realign' mechanism [8]. The source of these primers for hantaviruses is thought to be processing bodies (PBs), that are non-enveloped cytoplasmic RNP granules where non-sense mRNAs are stored awaiting degradation, and with which the SNV NP has been shown to closely associate [9]. NP is thought to bind the mRNA cap and protect it from degradation by the PB resident de-capping enzymes including DCP1a and DCP2, host factors that are required for other members of the order *Bunyavirales* [10].

The replication of hantaviruses, as with all bunyaviruses, is thought to occur within cytoplasmic viral factories [11]. Specifically, the replication factory of the prototypical bunyavirus, Bunyamwera virus (BUNV) has been visualized and described in detail. In mammalian cells, BUNV establishes large virus factories around the Golgi stack, recruiting organelles, cytoskeletal proteins and other cellular factors that together form elongated membranous structures known as viral tubes [12]. These tubes are lined with multiple membrane-bound polymerase molecules that are the sites of viral RNA synthesis, subsequently leading to the generation of RNPs for virion assembly. RNPs are thought to travel to adjacent assembly sites in the Golgi with the aid of actomyosin motors [12], where they interact with the viral glycoproteins and bud into the Golgi lumen. Virions subsequently mature as they tra ffic within Golgi-derived vesicles to the plasma membrane for release. BUNV replication and assembly in insect cells, in which the virus establishes a persistent infection, similarly occurs in the Golgi [13]. However, it was found that in mosquito cells, viruses do not accumulate intracellularly and exit the cell immediately after assembly, with NP and L proteins accumulated in non-membrane-bound cytoplasmic aggregates [13].

In contrast to BUNV, the precise intracellular site of hantavirus replication has not been defined. As NP is the major protein component of RNPs, its location is expected to coincide with sites of RNA synthesis and previous work has shown that NP from the NW hantaviruses, ANDV and Black Creek Canal virus (BCCV), accumulates within membranous structures surrounding the Golgi [14,15]. Consistent with this, NP from the OW hantavirus Tula virus (TULV) localizes in perinuclear regions [16] and HTNV NP resides within both perinuclear regions and the endoplasmic reticulum (ER)–Golgi intermediate compartment [17,18]. Further clues as to the location of sites of hantaviral RNA synthesis were revealed by the finding that the GFP-tagged L protein of TULV co-localized with the Golgi [19] and that a recombinant L protein from ANDV accumulated in perinuclear regions, co-localizing with both its cognate NP and also PB marker DCP1a [20]. Overall, with some notable NW hantavirus exceptions [21,22], hantavirus assembly is thought to occur by budding into the Golgi [7], suggesting the above description of BUNV replication and assembly sites may also broadly apply to those of hantaviruses.

In this study, we used confocal immunofluorescence microscopy alongside fluorescent single molecule in-situ hybridization (FISH) and electron microscopy (EM) to identify and characterise the cellular sites where NP as well as S segmen<sup>t</sup> RNAs were localised throughout an extended infection time course of the model OW hantavirus, TULV. We revealed that at peak and persistent stages of infection, NP is present as large filamentous and tubular structures, co-localizing with TIA-1, a marker for stress granules (SGs), and the Golgi, and that the FISH analysis strongly suggested that these TIA-1-containing and Golgi-derived structures are the sites of TULV RNA synthesis.
