Cryo-EM-based discovery of a pathogenic parvovirus causing epidemic mortality by black wasting disease in farmed beetles

SUMMARY We use cryoelectron microscopy (cryo-EM) as a sequence-and culture-independent diagnostic tool to identify the etiological agent of an agricultural pandemic. For the past 4 years, American insect-rearing facilities have experienced a distinctive larval pathology and colony collapse of farmed Zophobas morio (superworm). By means of cryo-EM, we discovered the causative agent: a densovirus that we named Zopho-bas morio black wasting virus (ZmBWV). We conﬁrmed the etiology of disease by fulﬁlling Koch’s postulates and characterizing strains from across the United States. ZmBWV is a member of the family Parvoviridae with a 5,542 nt genome, and we describe intersubunit interactions explaining its expanded internal volume relative to human parvoviruses. Cryo-EM structures at resolutions up to 2.1 A˚ revealed single-strand DNA (ssDNA) ordering at the capsid inner surface pinned by base-binding pockets in the capsid inner surface. Also, we demonstrated the prophylactic potential of non-pathogenic strains to provide cross-protection in vivo .


In brief
Cryo-EM, used as a sequence-and culture-independent diagnostic tool, identifies the pathogenic parvovirus responsible for an epidemic in farmed beetles and paves the way for countermeasures.The virus structure reveals extensive capsid/ssDNA interactions.

INTRODUCTION
Technologies for detection of uncultivatable viral pathogens by classical electron microscopy, 1,2 serology, 3 or molecular biology techniques 4 have been largely supplanted by metagenomic or other sequence-based methods, 5 leading to a renaissance of diagnostic virology. 6However, metagenomic approaches harbor pitfalls, including dependence on databases of known, annotated sequences, and methodological biases.For example, rolling circle amplification is heavily biased toward circular single-stranded DNA (ssDNA) viruses, while adapter-ligation sequencing is biased toward double-stranded DNA viruses.Because cryoelectron microscopy (cryo-EM) can now provide near-atomic-resolution 3D maps of chemical structures of viruses from a few virions, 7 because recent advances in cryo-EM throughput have reduced the cost and time of acquiring such data, and because new methods allow amino acid sequence inference from density maps, [8][9][10] we reasoned that cryo-EM is mature for use in diagnostic virology (at genus level or finer) and pathogen discovery without need for cultures or sequencing.
The superworm (Zophobas morio [Z.morio]) and its close relative, the mealworm (Tenebrio molitor [T.molitor]), are species of darkling beetles whose protein-rich larvae are dietary staples for captive reptiles, birds, and amphibians worldwide.Superworms are also a potential alternative protein source. 11We received reports from commercial and hobbyist growers of Z. morio that farmed larval populations were experiencing up to 100% mortality.Using larval corpse homogenate as a diagnostic specimen, we determined the structure of three variants of the agent at 2.1, 2.7, and 2.7 A ˚resolution, which we named Z.morio black wasting virus (ZmBWV); the initial structure was sufficient to identify it as a virus of subfamily Densovirinae, family Parvoviridae.][14][15][16] Parvoviruses (PVs) have ssDNA 3.8-6.4kb genomes.Two major expression cassettes-rep, which expresses two to six non-structural (NS) proteins, and cap, which expresses between one and four isoforms of the structural protein (VP)-are flanked by hairpin-like, partially double-stranded ''inverted terminal repeats'' (ITRs).The best-characterized PV, adeno-associated virus (AAV), is used in human gene therapy.Translational challenges in AAV biology include increasing the length of DNA that can be accommodated in a capsid. 17Herein, we identify the causative agent of Z. morio black wasting disease (ZmBWD) by cryo-EM, confirm its etiological role by experimental satisfaction of Koch's postulates, and describe its pathogenesis, nationwide distribution, and a potential countermeasure.Among the virion's unique structural features is a striking ordering of the genomic DNA.S3. (B) Sucrose step gradient from blackened Z. morio larvae displaying two fractions (arrows) at the 20% and 30% sucrose interfaces, respectively, which appear blue in fluorescent light due to the presence of viral particles.(C) Electron micrographs displaying the particles of $28 nm from those fractions.Insets emphasize the absence of genome inside particles of the 20% band and presence inside particles of the 30% band.(D) Isosurface rendering (left) and central slice through (right) 3D reconstructions of the particles purified from diseased larvae.The gold line separates reconstructions of the 20% band (upper right) and 30% band (lower left).The protein structure is almost identical between the two bands, but only the 30% band contains genome.The initial polyalanine chain is displayed as a ribbon diagram, with each T = 1-related subunit in a different color.Results of searches based on this map are presented in Tables S1 and S2.(E) Comparison of the capsid surface of the densovirus described herein associated with ZmBWD and consequently designated as Z. morio black wasting virus (ZmBWV) with those of other members of the Parvoviridae family: subfamilies Densovirinae (Galleria mellonella densovirus [GmDV], Acheta domesticus (legend continued on next page)

ZmBWD
A small, commercial insect-rearing facility in Utah experienced repeated Z. morio colony collapse in 2022 and contacted us in March of that year for investigation.At approximately 8 weeks of age and 25 mm in length, Z. morio larvae showed signs of distressed locomotion, uncoordinated wiggling (Video S1), and rigor followed by death (we later reproduced this clinical progression by experimental inoculation; see below).Moribund larvae blackened, and their inner organs lost structure, essentially becoming liquefied (Figure 1A).Mortality of larvae in the colony was approximately 90%, but the surviving larvae could pupate successfully and emerge as mature beetles.We tentatively named the observed pathology ''Zophobas morio black wasting disease" (ZmBWD).

Identification of a PV by cryo-EM
We homogenized Z. morio carcasses received from the Utah facility and obtained two bands by sucrose gradient fractionation: at the interfaces between the 20% and 25% sucrose steps and between the 30% and 35% sucrose steps (Figure 1B).Both contained isometric viral particles of approximately 28 nm in diameter.The more-buoyant particles were hollow, but the less-buoyant particles contained dense intracapsid material corresponding to the viral nucleic acid (Figure 1C).We named the virus ZmBWV.We obtained 3D maps of the empty capsids and full virions at 2.9 and 2.7 A ˚resolution, respectively (Figure 1D).Both were T = 1 icosahedral capsids with a single capsid protein per asymmetric unit (ASU) with a jelly roll fold.We manually traced the C a backbone by building polyalanine chains of 423 and 425 residues in length per ASU, respectively (Figure 1D).Qualitatively, we assessed that the backbone model exhibited a prototypical PV fold.Quantitatively, we queried all extant protein structures from the Protein Data Bank (PDB) using DALI 18 to detect homologs and obtained three hits with a Z score > 20, all of which are members of the Densovirinae subfamily of the Parvoviridae (Figure 1E).Although we manually traced and interpreted the density map in this case, cryo-EM would be most useful as a diagnostic technique if it did not rely on labor-intensive tracing or expert interpretation of folds.Therefore, we also used automated, background-knowledge-agnostic tools to reproduce this determination.We automatically traced and assigned sequence to the cryo-EM density using ModelAngelo. 10ALI search of the longest detected chain revealed the same result pattern, followed by PVs of other subfamilies (Table S1).Meanwhile, HHblits search 19 of the sequence profile from ModelAngelo against UniProt detected similarity to Blatella germanica DV (BgDV) with a p value of 10 À77 (Table S2).Sequencing would later confirm that ZmBWV is a DV whose closest sequenced relative is BgDV and whose closest structurally characterized relative is Galleria mellonella DV (GmDV) capsid, in complete accordance with the structure-based identification.
We conclude that cryo-EM can be used for sequencing-free discovery of novel viral species from clinical or environmental samples.

ZmBWV genome and structural proteins
Because cryo-EM had revealed that ZmBWV is a PV, we knew the genome must consist of linear ssDNA.Linear ssDNA is refractory to ligase-based next-generation sequencing (NGS) preparation but amenable to transposase-based NGS preparation, 20,21 and PVs need few reads for good coverage.Because the cryo-EM result informed the sequencing strategy, we obtained a complete genome of the index case of ZmBWV by commercial transposase-based NGS at a cost of $15.The genome is 5,452 nucleotide (nt) long, with I-shaped ITRs of 180 nt at both termini (Figure 2A).Its coding region harbored five major open reading frames (ORFs) over the (+) and (À)-sense frames, indicating an ambisense replication strategy (Figure 2B).Three of these, located on the right strand, were homologs of the NS1, NS2, and NS3 proteins of DVs classified to genus Blattambidensovirus of the Densovirinae subfamily (protein sequence identity of 45%-98%).Blattambidensoviruses such as BgDV (the type species) express three capsid proteins encoded by two ORFs: cap1 and cap2.ZmBWV contains homologs of cap1 and cap2 (amino acid identity of 41% and 49.5%, respectively, with BgDV).Analyzing either empty capsids or full virions by SDS-PAGE, four protein bands were observed at sizes of 85, 74, 50, and 48 kDa, with the 50 kDa fraction being the most abundant (Figure 2C).Protein sequencing by tandem mass spectrometry (MS/MS) revealed that the three lighter bands (designated as VP4, VP3, and VP2) were products of cap2 exclusively.Based on the peptide coverage map, VP2 is translated from the first ATG start codon of the frame, whereas likely start sites for VP3 and VP4 are Met110 and Met156, respectively.The heaviest band encompassed the almost-complete cap1, which is spliced to cap2 (Figure S1A).BgDV minor proteins are ubiquitinated, and BgDV VP2 is produced by splicing from cap1, 22 while by contrast ZmBWV minor proteins are not ubiquitinated, and ZmBWV VP2 is produced by use of the second ATG start codon of cap2 (Figure 2B).

ZmBWV is endemic in the United States and is associated with disease
We gathered specimens of Z. morio (16 pools) and of T. molitor (9 pools) from breeders who had experienced mass mortality events clinically consistent with ZmBWD (9 breeders; 11 farms), or from mail-order services (3 pools), or from local stores (2 pools) (summarized in Table S3).Using the whole-genome sequence, we developed a diagnostic PCR targeting the NS1 gene (Figure 3D).100% of Z. morio pools obtained from breeders with symptomatic larvae tested positive by PCR.Insects from local stores tested positive, and staff or customers reported observing this pathology.Mail-order Z. morio from two vendors did not exhibit symptoms and were PCR-negative.(legend continued on next page) Furthermore, one breeder who culled his colony, thoroughly bleached, and recommenced operations with fresh stock sent asymptomatic larvae that were PCR-negative (we received comparable anecdotes from other states that culling and bleaching were effective).As ZmBWV was found only in symptomatic Z. morio populations, Koch's First Postulate is satisfied.
ZmBWV was detected in 11 geographically dispersed states, so nationwide presence should be presumed.
In two pools, we observed non-ZmBWV particles (Figure 1F).For example, from pool NJ2-molitor, we detected 351 particles with 32 nm diameter (vs.37,850 ZmBWV particles of diameter 28 nm in the same dataset).The 3.55 A ˚resolution reconstruction was automatically modeled identified as an iflavirus.The triangulation number (pseudo-T = 3) and surface morphology are consistent with this identification.Identification of this iflavirus establishes that the limit of detection (LOD) of this assay is at least two orders of magnitude below ZmBWV abundance.
Breeders reported that mealworms would occasionally exhibit black wasting but never economically significant mortality.We detected a ZmBWV-like virus from all 3 mail-order T. molitor batches tested, although no signs or symptoms were observed (Table S3).Likewise, 3 of 4 T. molitor pools from breeders of Z. morio were PCR-positive; two of these colonies had a few symptomatic individuals and one had none.We conclude that the ZmBWV-like virus of T. molitor is of mild pathogenicity in its native host.Genomic sequences were obtained for 13 Z.morio and 7 T. molitor samples.We did not attempt to obtain samples from other countries.4][25] This implies a worldwide distribution of ZmBWV.The presence of ZmBWV in these vertebrate metagenomes does not indicate vertebrate infection; it can be explained by insectivory.Full genome phylogenetic calculations (Figure 2D) revealed no geographic clustering, nor did sequences cluster by breeder.Paired T. molitor and Z. morio strains from the same breeding facility did not cluster with each other but segregated by host species.Of the Z. morio contributed directly from breeders (as opposed to being purchased on the open market), only one pool was believed by the breeder to be ZmBWV-negative at time of shipment.These larvae did sicken and die a few days later, and ZmBWV was detected.Interestingly, the genome of that strain designated FL-morio (but not of any other strain isolated from Z. morio) clustered with T. molitor strains (Figure 2D).It is possible that this is a low-pathogenicity strain whose virulence was activated by stress (such as cold) during shipping.All pathogenic ZmBWV strains clustered together.
Two more epidemiologically relevant species were co-housed at one of the sampled farms.Alphitobius diaprenius, commonly known as lesser mealworm or the buffalo beetle, is another species of the Tenebrionidae and is frequently co-housed with superworms as a cleaning aid.These insects were co-housed with the species Blaptica dubia, commonly known as the Dubia roach.Both larvae (8.17 3 10 9 gc/mL) and beetles (1.13 3 10 9 gc/mL) of A. diaprenius tested positive for ZmBWV, as well as B. dubia nymphs (9.99 3 10 11 gc/mL) and adults (1.05 3 10 10 gc/mL) (Figure 4D).Both colonies were reproductively thriving and asymptomatic.
Although most Z. morio samples were received fresh, one was a frozen diagnostic specimen from February 2020, whose breeders observed ZmBWD in their colonies as early as 2019.We retrospectively located social media posts by exotic pet keepers between May 2019 and September 2020 claiming that Z. morio were no longer available from their usual vendors, as well as posts from breeders detailing mass mortality events with blackened carcasses.14 of these were between January 2019 and September 2020, and three were earlier (2013, 2015,  2017).Descriptions were insufficient to conclusively implicate ZmBWV, but it seems likely that ZmBWV had achieved nationwide spread by late 2019 and that the economic impact was felt beyond Z. morio breeders in the form of shortages and loss of product quality.

Pathology of ZmBWV infection
We administered ZmBWV (reference strain UT-morio; see Table S3) to healthy, 4-week-old, PCR-negative Z. morio larvae.Three forms of administration were contrasted: injecting virus into the fat body (Figure 3A), dripping virus suspension onto larval cuticles, or (exploiting the naturally cannibalistic tendency of Z. morio) feeding blackened carcasses of infected insects (the ZmBWV titer in carcasses was $10 16 gc/mL) (Figure 3B).By all three methods, larvae exhibited the same pathology as described above in natural infection, viz.: distressed locomotion, loss of coordination in wiggling (Video S1), blackening, and eventually liquefaction and death.This confirms Koch's Third Postulate.Time to symptoms and death varied with titer (Figure 3A) and route of (Figure 3B).Direct injection was the most lethal; the LD 50 by injection is below 10 5 gc but above 10 9 gc by dripping.Administration by cannibalism led to a slower course of infection, but full mortality was observed.No symptomatic individual recovered in any experiment.At 14 days post infection (d.p.i.), among carcass-fed larvae, viral load was 6.6 3 10 12 gc/mL in presymptomatic larvae and about 33 higher in symptomatic and deceased larvae (Figure 3C).Meanwhile, (B) Line diagrams representing the manifestation of symptoms and the survival rate of Z. morio larvae after introduction of ZmBWV strain UT-morio by adding blackened carcasses of ZmBWV-killed larvae, which were ingested by the subjects (purple); or by dripping a suspension of virus in PBS onto the larval cuticle (gray and green).Controls as in (A).(C) Box-and-whisker plots showing the qPCR viral titers of the larvae from (A) and (B), sampled at 12 d.p.i.(halfway through the experiment).Each plot summarizes the results of three biological replicates.(D) Viral yield of infected Z. morio specimens at various life stages.The box and whiskers plots summarize the qPCR results of three infected colonies, located in New York, Arkansas, and Georgia.(E) MicroCT reconstructions of the midgut of a healthy (left) and freshly deceased (right) 4-week-old Z. morio larvae.The specimen in the right panel, fixed immediately upon becoming non-responsive, had been inoculated with ZmBWV strain UT-morio by ingestion of an infected carcass and showed the typical pathology (including blackening) associated with ZmBWV.Both images show the cross section of the midgut, tilted obliquely, revealing both the lumen and outer surface of the organ.(legend continued on next page) dead larvae infected by injection at 10 15 or 10 13 gc/mL had viral loads of 7.9 3 10 12 and 3.9 3 10 12 gc/mL, respectively.No sequence differences were identified in ZmBWV recovered from larvae experimentally inoculated with the UT-morio strain, confirming Koch's Fourth Postulate.ZmBWV could be cultured in BGE-2 cells (Blatella germanica) but not C6/36 (Aedes albopictus), Aag2 (Aedes aegypti), Sf9 (Spodoptera frugiperda), or S2 (Drosophila melanogaster).Strain UT-morio but not NY3-molitor caused cytopathic effect (CPE) in BGE-2 cells (Figure S2A), although replication was confirmed by qPCR for both strains.The propagation of ZmBWV in pure culture confirms Koch's Second Postulate.When injected into healthy superworms, supernatant from infected BGE-2 cells reproduced ZmBWD symptoms and caused death.
To assess the time course of natural infection, we obtained Z. morio individuals in various stages of life from three affected farms in Arkansas, Georgia, and New York, viz., 8-week-old symptomatic or asymptomatic larvae, surviving pupae and beetles, as well as 1-week-old larvae that are the offspring of surviving beetles.Virus yield (by NS1 qPCR) varied extensively by life stage, with titers from $1 3 10 8 gc/mL in newly hatched larvae to $2 3 10 16 gc/mL for blackened carcasses (Figure 3D).Surviving beetles exhibited a titer of $7 3 10 12 gc/mL, and the breeder reported that beetles who survived through pupation continued to reproduce at the farm.
Symptomatic larvae exhibited a dark area of discoloration along the midgut prior to gross blackening.This, and the high viral load in larvae infected by feeding, suggested that the midgut plays a crucial role in the ZmBWV infection.Microscale X-ray computed tomography (microCT) showed that while healthy larvae had an intact midgut wall, the midgut wall of dead larvae was extremely thin and was disrupted by frequent fenestrations (Figure 3E).The preserved ring-shaped structures suggest that the longitudinal muscles of the outer midgut wall remained intact, while the inner layer (which is composed mainly of columnar cells) was destroyed by ZmBWV infection.

Strain differences and cross-protection
We inoculated Z. morio with a ZmBWV strain from Z. morio (UTmorio) and T. molitor (NJ2-molitor) to establish whether the avirulence of T. molitor strains is host-specific or strain-specific (Figure 4A).Fat body injections by the pathogenic UT-morio strain recapitulated the previous inoculation experiment: all larvae died 10-21 d.p.i. in a dose-dependent manner (Figure S2B).By contrast, larvae inoculated with 10 5 or 10 9 gc of NJ-molitor exhibited 90% or 70% survival to 30 d.p.i., respectively.Larvae inoculated at 10 13 gc tended to die within 3 days or survive through 30 d.p.i., suggesting acute toxicity at extreme dose rather than virulence.We repeated the feeding experiment by offering the healthy larvae either blackened Z. morio larvae infected by strain UT-morio or blackened T. molitor larvae containing $10 16 gc/mL of NJ-molitor (Figure 4B).Although the UT-morio-infected group all died by 25 d.p.i., the NJ2-molitorfed group was 100% asymptomatic through 30 d.p.i.These larvae were kept alive for another 4 months, and the virus titer was assessed at 30, 60, and 120 d.p.i.There was no significant difference in the virus yield of the two earlier time points (2.68 3 10 9 and 1.15 3 10 10 gc/mL, respectively) (Figure 4C).The virus remained PCR-detectable at 4 months post-inoculation, albeit at 1.78 3 10 7 gc/mL.
To investigate the possibility of cross-protection by different strains of ZmBWV, we inoculated the healthy larvae with the NJ2-molitor strain at a dose of 10 9 gc, or saline control, by injection (Figure S2C).No morbidity was observed through 21 d.p.i.The larvae were challenged at 21 d.p.i. by injection with 10 7 gc of UT-morio or with saline.Larvae that received saline followed by UT-morio challenge all died within 21 d.p.i.Of larvae that received NJ2-molitor followed by UT-morio challenge, 30% survived to experiment termination at 32 d.p.i.This group also showed a 7-day delay in the onset of the first symptoms.Mechanisms of cross-protection in invertebrates include adaptive immunity, induction of a generalized antiviral state, or superinfection exclusion.We repeated the aforementioned experiment but sacrificed and pooled five individuals from each group semiweekly.A polymorphic region of the NS3 gene was amplified by PCR and Sanger-sequenced.The ''protecting'' strain NJ2-molitor persisted through 2 days after UT-morio challenge (Figure S3).From 6 d.p.i., UT-morio outcompeted NJ2-molitor to the extent that UT-morio alone was detectable in the NJ2-molitor/UT-morio group in either healthy or symptomatic individuals.NJ2-molitor persisted in the NJ2-molitor/saline group through 30 d.p.i.
To investigate whether all non-virulent T. molitor-derived strains provided cross-protection, we inoculated healthy larvae with five ZmBWV strains derived from asymptomatic T. molitor populations nationwide.Three of these provided some crossprotection: MN-molitor, NY3-molitor, and PA-molitor.OR-molitor, however, displayed mild virulence during the 3-week-long pre-challenge at 30% mortality.It also accelerated the pathogenesis of UT-morio challenge.
Persistence of the protective strains at 4 months post-inoculation bodes well for the durability of protection, as farmed Z. morio are typically sold to consumers at under 3 months of age.

The ZmBWV capsid structure and intersubunit interactions
We resolved the capsid structure of virions of NJ2-molitor (2.1 A ˚), UT-morio (2.7 A ˚), and OR-molitor (3.3A ˚) as well as the empty capsids of UT-morio (2.9A ˚) and OR-molitor (2.7 A ˚) (Table S4), corresponding to a non-virulent, virulent, and mildly virulent ZmBWV strain, respectively.ZmBWV is remarkably flat, even for a DV (which are already smoother than human-infecting PVs such as AAV).Spikes around the 5-fold pore in DVs are shorter and blunter in ZmBWV (Figures 1E and S5A).ZmBWV is antigenic upon injection into rabbits (Figure S5C).Cross-reactivity between ZmBWV and the human-infecting PV AAV9 was not detected (Figures S5C and S5D).
Map density begins at Ala165, which is shared between VP4 and the longer isoforms: regions unique to VP1, VP2, and VP3 (legend continued on next page) are not seen (Figure 5A).ZmBWV and other DVs can package more DNA than AAV due to their larger capsids: the ZmBWV capsid interior is 25% larger (interior volume 3.6 3 10 6 A ˚3) than AAVhu.37 (2.8 3 10 6 A ˚3), so the ZmBWV genome is not more densely packed (1.94 vs. 1.75-1.89A ˚3/Da for AAV) despite ZmBWV's longer genome length.A major feature enlarging the capsid is the elongation of the first b strand (bB) of the Rossmann fold [26][27][28] (Figure S5A).The elongated strand forms an intersubunit b sheet across the 2-fold axis, resulting in a stable dimer interaction with a large interface (Figure 5B).ZmBWV elaborates on this motif: in addition to the interaction across the 2-fold axis, the first five ordered residues of the N terminus form a second intersubunit b sheet with the bF strand of a different, 3-fold-related neighbor (Figure 5B).The bA/bF intersubunit sheet is unique to ZmBWV and stabilizes the luminal region near the 5-fold pore.These residues (Ala165-Arg170) are disordered in genome-free capsids, and the N terminus is free to move underneath the 5-fold pore; this could explain the presence of density not accounted for by the atomic model under the 5-fold pore in empty particles (Figures 5B and 5C).
0][41] The 20 A ˚-long channel was occupied by protein density even in the absence of viral genome (Figure 5C).In full virions, the density was rod-shaped and was mostly limited to the lower regions of the channel.Empty particles displayed a more diffuse density present throughout the channel (albeit weaker in the middle).Large areas of disordered density not attributable to VP4 were observed underneath and over the 5-fold pore in empty particles particularly.Also, the DE loop (which forms the 5-fold pore) was flexible in empty particles and unmodelable despite the resolution of the map (Figure 5C).In diverse PVs, studies have shown the 5-fold channel is filled after genome packaging. 27,35,42,43However, in ZmBWV, the 5-fold axis has increased mobility and externalization in empty capsids, in a reversal of the 5-fold dynamics relative to AAV and other PVs (Figure 5C).We assayed protease accessibility of the capsid proteins in full virions and empty capsids under the same conditions as used for vitrification (Figures S4B and  S4C).In accordance with the structural data, VP1 and VP2 were more accessible in empty than full particles.
A pore at the center of the 3-fold axis of ZmBWV is surrounded by a b-annulus structure (Figure 5D).Unlike the 5-fold pore, the 3-fold pore is quite rigid.An ion coordinated by three symmetry-related arginine residues-probably by both the primary and secondary amines-glues the pore together.ZmBWV virions do not contain iron, cobalt, or nickel, as assayed by inductively coupled plasma mass spectrometry (ICP-MS).Copper and zinc are present in pellets of both empty and full particles, but not in control PVs (Figure S1C).The charge and elemental identity of the ion have not been established.[29][30]43,44 However, the pore in the center of this annulus has been thought to be a flexible portal due to the disorder observed in GmDV 26 and other viruses (Figure S5B).Key residues are conserved between GmDV and ZmBWV.We dialyzed ZmBWV virions into the phosphate-free buffer used for GmDV purification and observed that many capsids became disrupted and most lost some or all of their DNA (Figure 5E).It is plausible that native GmDV contains this coordinated ion but that the ion was lost in the crystal structure due to biochemical manipulations such as crystallization.
ZmBWV capsid polymorphism 12 VP3 sites had an amino acid variant that was present in two or more strains, and all 12 were surface-exposed.Sites were concentrated around the 3-fold axis and the blunt spikes around the 5-fold pore (Figure S6).The surface near the 5-fold pore plays a role in DV virulence and tropism in lepidopteran DVs.For example, in Junonia coenia DV, these residues govern midgut transcytosis. 45,46In this vicinity, additional density not attributable to the capsid protein was observed in strains UT-morio and NJ2-molitor (Figures 6A and 6B)-but not OR-molitor (Figure 6C), which lacks a hydrophobic residue at the 5-fold spike-terminating at capsid residue Asn235.
Capsid-genome interactions While 0, 1, or 2 nt per ASU are observed in a typical PV structure, 29 the genome-containing virions of ZmBWV contain an unexpected number of ordered nucleotides (Figure 6D).The density attributed to the nucleic acid was present in all three strains, exhibiting only minor differences in the ordering of certain nucleotide densities (summarized in Table S5).We modeled 17 nt per subunit in the 2.1-A ˚-resolution structure of ZmBWV strain NJ2-molitor.Several nucleobases are visible at the same isosurface threshold as capsid amino acids, indicating that these bases have high occupancy; therefore, almost onefifth of the ZmBWV genome is icosahedrally ordered!
The ordered DNA consists of three discontiguous stretches per ASU, with 3, 9, and 5 nt, respectively.These stretches interact with inner capsid surface as well as with each other.
Each has at least 1 nt inserted into a pocket formed by the capsid inner surface (Figures 6F-6H) as well as nucleotides engaged in base-pairing: the 3-mer and 9-mer form an interchain Watson-Crick pair and the 5-mer an intra-chain Hoogsteen pair (Figure 6D).Two coaxial stacks of 4 nt project out from the surface; one, which sits atop Tyr400, contains a nucleotide from the 9-mer as well as three from the 5-mer (Figure 6E).
There must be some degeneracy in the nucleotide sequence of the bound DNA because the most likely sequence of none of the three stretches is present at $60 copies per ZmBWV genome.Nevertheless, certain nucleotides have more clearly resolved density and form base-specific interactions with the capsid, such as a guanine residue (third nucleotide of the 9-mer) that hydrogen bonds with the main chain of a b strand G of the Rossmann fold or a cytosine base occupying a negatively charged luminal pocket.The base composition of ordered nucleotides is strongly enriched in purines, despite the fact that the ZmBWV genome as a whole is only 49.6% purines.

Cryo-EM in agent discovery
We have definitively established ZmBWV as the cause of superworm liquefaction.Methods for identifying a viral pathogen have undergone a remarkable transformation in recent decades: culture techniques dominated for over a century until being largely supplanted by metagenomics. 47Despite its advantages, NGS still harbors pitfalls that hamper the discovery of divergent pathogens.As a diagnostic tool, electron microscopy still plays a role alongside other methods because of its speed, lack of reliance on reference databases, and lack of genome chemistry bias.For example, SARS-CoV was first identified by thin-section transmission electron microscopy. 2 We demonstrated that cryo-EM is sufficient to reveal the identity of a viral pathogen at the genus level.This information was at our disposition within a week of receiving the first diagnostic specimen.We publicly disseminated the identity of the agent and appropriate decontamination protocol 48 days (Table S3) after receipt of the initial specimen (before any sequencing could be performed). 48Many isolates sequenced herein were obtained from breeders who contacted us after reading that communication 48 because they observed the characteristic pathology of ZmBWD in their insects; all pools of diseased larvae obtained thereby tested positive.
Furthermore, the cryo-EM structure informed our NGS strategy, facilitating the development of a PCR-based assay that we used to investigate the extent and pathogenesis of the ZmBWV epidemic.The high abundance of ZmBWV in the organism (up to 2 3 10 16 gc/mL in some cases) makes cryo-EM-based identification facile, whereas human pathogens achieve lower titers (e.g., 2 3 10 9 gc/g for SARS-CoV-2 in the lung). 49Furthermore, the LOD for non-icosahedral virions (e.g., SARS-CoV-2) will be worse than for icosahedral (e.g., ZmBWV) or helical pathogens because fewer averageable subunits are present per virion and subunits are harder to align.The LOD can be improved trivially by increasing the number of micrographs, using a detector with a larger field of view, or concentrating the input material (e.g., 1003 with tangential flow filtration), but whether this would be enough to reliably discover human pathogens is not yet clear.Technologies in development such as direct cryo-EM of heterogeneous cell lysates 8 and improvements in cryo-EM throughput would improve the LOD and ensure that human viruses are discoverable by cryo-EM.
ZmBWD: Epidemiology, pathology, and control Our results suggest that ZmBWV had become a nationwide epidemic in the United States by 2019.The DV Acheta domesticus densovirus (AdDV) regularly disseminates in cricket-rearing facilities, causing mass mortality across Europe since 1977 50 and in North America since 2009. 51AdDV has impacted cricket-rearing practices worldwide, requiring the rearing of less susceptible orthopteran species. 16Although we did not test any specimens from outside the United States, the presence of sequences from metagenomic datasets that fall within the ZmBWV clade suggests that this virus is not limited to North America.ZmBWV's broad host spectrum, subclinical presence, and multiple genotypes suggest that the epidemic may remain active for a long a time.
There are several different ZmBWV strains circulating currently in the United States, affecting both T. molitor and Z. morio.Despite this, we did not observe a comparable epidemic of disease in T. molitor.The lack of virulence exhibited by the NJ2-molitor strain in the Z. morio host suggest that the current epidemic has a distinct origin.This is also supported by the phylogenetic calculations, suggesting a single introduction event.Our experiments indicate that there is cross-protection between strains.This phenomenon could be further developed to design a successful prophylaxis.Although the prophylactic mechanism is unclear when challenged 3 weeks after prophylactic inoculation with an avirulent strain, protection from symptoms endures after the virulent strain outcompetes the prophylactic strain.
ZmBWV can spread via the oral-fecal route.The midgut has been shown to play an important role in DV infection.Lepidopteran protoambidensoviruses cross the midgut wall by transcytosis in order to reach the true site of replication, which may be the fat bodies or the wall of the visceral trachea and hemocytes, [52][53][54] while lepidopteran iteradensoviruses and bidnaviruses replicate exclusively in the columnar midgut cells. 55,56In both cases, however, larvae die quickly: within 7-10 d.p.i. 52,53,55By contrast, larvae die of ZmBWV 18-21 d.p.i. with dietary or superficial inoculation.Moreover, newly hatched larvae were already infected, yet symptoms manifested only at 8 weeks of age.Once the larvae display the initial signs of ZmBWD, death is expected in less than 5 days.We suggest a model of pathogenesis wherein the virus replicates in the midgut columnar cells following uptake through contaminated feed, which eventually leads to fat body invasion once the midgut wall is too damaged to fulfill its barrier function.The invasion may happen immediately prior to the onset of symptoms.Although DV-infected caterpillars fail to pupate, 53,55 Z. morio larvae that pupate before the onset of symptoms complete their entire lifecycle.The ability of ZmBWV to cause a chronic infection in reproducing beetles and asymptomatic phases may be major reasons why the epidemic has been difficult to control.
Those rearing Z. morio should be aware that PVs are notably resistant to alcohol-based sanitizers.Therefore, ethanol is not sufficient for cleaning enclosures and other items that have come into contact with infected beetles.We recommend that bleach be used in these cases.When bringing new beetles into a breeding colony, avoiding overtly symptomatic individuals is insufficient.
ZmBWV structure A unique aspect of ZmBWV structure is the reversal of the N terminus exposure pattern.ZmBWV, as well as members of all DV genera with a split VP expression strategy, contains at least two capsid proteins with a unique N-terminal region.The channel density and the protease accessibility assay results indicate that while the N termini of the major capsid protein (VP3) and its shorter minor counterpart (VP4) are constantly surfaceexposed, the unique VP1 and VP2 N termini are more exposed in the absence of a packaged genome.ZmBWV VP1u must be at least 400-aa long, which is more than twice the length of the corresponding region ($180 aa) in protoparvoviruses of similar genome sizes.This accords with our findings that some of the unique VP1 and VP2 N termini appear to be surface exposed even in the mature virions.The only PV thus far that has been shown to harbor a constitutively external VP1u is B19, 39 the human pathogen.It is possible that, in ZmBWV, the packaged genome supports intraparticle retention of the longer N termini, bringing them adjacent to bF (which would reinforce the retention by hydrophobic interactions).
ZmBWV has a second potential channel: the 3-fold annulus.Its diameter exceeds 10 A ˚only in GmDV, 26 where it is open and flexible, ZmBWV, where a triad of arginine residues coordinates an unidentified ion, and divergent Penaeus monodon metallodensovirus (PmMDV), where a triad of histidines coordinates an ion. 30The GmDV annulus was hypothesized to be the portal for DNA packaging or uncoating due to its flexibility and positively charged sidechains. 26The structure of the ZmBWV portal, however, contraindicates this.In the T = 3 ssRNA family Tombusviridae, the 3-fold axis is also covered by a b-annulus, but, as in Parvoviridae, the annulus varies in size and is stabilized either by a bound ion, or by hydrogen bonds, or by hydrophobic interactions depending on the species. 57Cucumber necrosis virus's annulus coordinates Zn 2+ , and the b strands comprising the walls of the opening are twice as long as in other tombusviruses.This suggests that PVs independently evolved the same strategy to stabilize the 3-fold axis: while small diameters can be stabilized by hydrophobic interactions (as in AdDV) 27 or hydrogen bonds (as in Bombyx mori densovirus [BmDV]), 28 chelated ions can bridge a larger opening (Figure S5B).Given the similarity between ZmBWV and GmDV and effect on ZmBWV of dialysis into the buffer used with GmDV, it is plausible that the coordinated ion is also present in the GmDV annulus under physiological circumstances but absent in the GmDV structure due to biochemical manipulations, such as crystallization.
Of the 13 polymorphic residues linked to the ZmBWV capsid surface, 10 are near the 3-fold annulus, while three are linked to the 5-fold spikes.9][60][61][62][63] Because of the strain-specific variation in the presence/absence of the unmodeled moiety near Asn235 and 5-fold spikes, it would be interesting to examine the effect of polymorphism in this region on how each strain interacts with its host.
Little is known about the structure of the parvoviral genome following packaging, as in general few ordered nucleotides are present. 29In genus Protoparvovirus, however, large stacked columns of 11-21 nt could be modeled for each subunit, which establish sparse, non-sequence-specific hydrogen bonds with the interior capsid surface near the 3-fold symmetry axis.The ordering is attributed to the chelation of ions. 64,65In Acheta domesticus segmented DV, five of the six ordered nucleotides per ASU form sequence-specific interactions with four subunits simultaneously stabilizing the 2-fold interface. 43The ZmBWV virion harbors a surprisingly high number of ordered nucleotides, with each ordered DNA stretch harboring at least one base that occupies base-specific pockets of the capsid interior.AdDV shares 1 nt-binding site with ZmBWV.Genome packaging into preformed PV capsids stands in stark contrast with other ssDNA virus families, such as the Circoviridae, where interactions between the nucleic acid-binding domain of each subunit and the ssDNA genome are required to maintain virion integrity. 66Due to its I-shaped terminal hairpins, the ZmBWV genome is partially double-stranded.In an asymmetric reconstruction of the ssDNA Chaetoceros tenuissimus DNA virus type II genome (family Bacilladnaviridae), the partially dsDNA regions display a spooled structure and form an inner capsule around the less-ordered ssDNA. 67In our C 1 reconstruction of the ZmBWV capsid, no such regions could be observed, suggesting that the ITRs may not establish a direct connection with the inner capsid wall.

Conclusions
We demonstrated sequence-and culture-independent, genuslevel identification and molecular characterization of a viral etiological agent-ZmBWV-by means of cryo-EM, using material from deceased animals as the direct input.Structural features of the capsid/DNA interaction indicate a partially degenerate, base-composition-biased element in the DNA that engages base-binding pockets on the capsid interior.Extension of the bB strand to form an intersubunit b sheet with the 3-foldadjacent subunit allows for increased capsid radius and interior volume-a feature that would be pharmaceutically valuable if it could be conferred to AAV, increasing its packaging capacity.

Limitations of the study
We introduce atomic structure determination by cryo-EM as a diagnostic method in virology.This method was facile in this study because of abundant, icosahedral virions.We demonstrated agent identification at $2 3 10 12 gc/mL and estimated that with existing methods, extension to 1 3 10 9 particles/mL seems realistic for icosahedral agents (>6 3 10 10 for asymmetric agents), so the LOD is not yet competitive with deep sequencing.Biosafety considerations complicate diagnostic cryo-EM, particularly for select agents or biosafety level 3 specimens. 68Cryo-EM equipment is expensive, and we spent over $3,000 in equipment fees in imaging the initial strain in 2022; however, the maturation of low-voltage cryo-EM 69 and other cost-saving techniques means a comparable experiment in 2024 would cost around $1,000.
We showed that certain avirulent ZmBWV strains provide a degree cross-protection and ameliorate morbidity and mortality from virulent-strain challenge.Only injection was tested as a method of administration in cross-protection studies, and injection is not viable in the field or at scale.Furthermore, the optimal dose and formulation to optimize cross-protection have not been defined, and it is unclear whether an attenuated live virus will achieve acceptable cross-protection.
Structural understanding of PV virions is limited by asymmetric or pleomorphic features that are obscured during icosahedral averaging, blurred density for the moiety near Asn235 due to its flexibility, and uncertainty around the identity of bound ions.

Lead contact
Further information and requests for resources and reagents may be directed to and will be fulfilled by either corresponding author: Judit J. Pe ´nzes (judit.penzes@rutgers.edu)or Jason T. Kaelber (jason.kaelber@rutgers.edu).JTK is the lead contact.

Materials availability
Zophobas morio black wasting virus has been deposited at, and may be obtained from, the World Reference Center for Emerging Viruses and Arboviruses.
Data and code availability d Sequences have been deposited in Genbank, cryoEM maps have been deposited in the Electron Microscopy Data Bank, and atomic models have been deposited in the Protein Data Bank.All accessions are publicly available.Accession numbers are listed in Table S3 and in the key resources table.d This paper does not report original code.d Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS Animals
Zophobas morio larvae were housed in plastic insect breeder boxes at room temperature and were given a piece of fresh carrot every second day.Oatmeal bran was used as bedding.Experimentally inoculated larvae were housed in single use plastic dessert cups, also on oat bran bedding.Animals were euthanized prior to processing in either dry ice or by freezing at -80 C. Sexing Z. morio by microscopic investigation of the spermathecae/parameres was not practical in this study, but the parent population were mixed gender.Age is noted separately in each experiment.Institutional permission is not required for studies on this species.

Cell lines
Five insect cell lines were cultured for this study.The Aag2, C6/36, Sf9 and S2 cell lines were previously obtained from ATCC and were authenticated at time of purchase with a certificate of analysis.The BGE-2 cell line was a gift from Prof. Timothy Kurtti from the Department of Entomology, University of Minnesota and was not authenticated.All cell lines were cultured at 28 • C. Aag2 and S2 cells were grown in Schneider's Drosophila Medium (Gibco), supplemented by 5% fetal bovine serum (FBS), with additional 1% L-glutamine and 1% MEM non-essential amino acid solution (NEAAS) (Gibco) in case of the former.Sf9 cells were grown in serum free SF-900 II medium (Gibco).C6/36 cells were grown in L15 medium (Gibco) supplemented with L-glutamine, 1% MEM-NEAAS and 5% FBS in the presence of 5% CO 2 .BGE-2 cells were cultured in L15-B medium supplemented with D-glucose (Sigma-Aldrich), 5% FBS, 5% tryptose phosphate broth (Sigma-Aldrich) and 0.1% bovine lipoprotein concentrate (Sigma-Aldrich).Experimental infection of cell lines with the ZmBWV UT-morio strain was carried out in a 10-fold dilution series, at a multiplicity of infection (MOI) of 10 5 to an MOI of 0.1, in the company of a saline control.The culturing medium included an additional 10 mg/ml gentamicin.Cells were seeded the day prior to infection and the virus suspension in 100 ml phosphate-buffered saline (PBS) was added to 1 ml culturing medium.5 hours after virus addition, the medium was aspirated and replaced by fresh gentamycin-supplemented culture medium.

METHOD DETAILS
Origin of the samples Specimens of Z. morio were shipped directly to our laboratory from 11 farms, located in the states of Arkansas, Florida, Georgia, Maryland, Mississippi, Ohio, New Jersey, New York, and Utah, with two farms located in the state of New York.From three of these farms we also obtained T. molitor samples, namely from Ohio, New Jersey and one of the farms in New York.We purchased and online ordered Z. morio larvae from two facilities, located in the states of Oregon and Pennsylvania.T. molitor samples were obtained the same way from both facilities, as well as from Louisiana.Samples of both T. molitor and Z. morio bred in the states of Indiana and Minnesota were picked up at local pet stores.Dubia roaches and buffalo beetles originated from one of the farms located in New York.

Purification of viral particles
Deceased or euthanized Z. morio or T.molitor larvae were subjected to tissue homogenization in PBS (137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , 1.8 mM KH 2 PO 4 ), followed by three cycles of freeze-thaws.Following this, the homogenate was combined with an equal volume of 13 TNTM pH8 (50 mM Tris at pH 8, 100 mM NaCl, 0.2% Triton X-100, 2 mM MgCl 2 ) and the debris were removed by centrifugation at 3700 3 g at 4 • C for 15 min intervals until the supernatant was sufficiently cleared.The supernatant was mixed with 13 TNET pH8 (50 mM Tris pH8, 100 mM NaCl, 0.2% Triton X-100, 1 mM EDTA) in a 1:3 ratio and concentrated on a cushion of 20% w/v sucrose in TNET using a type 45 Ti rotor for 3 h at 4 C at 207,500 3 g on a Beckman Coulter S-class ultracentrifuge.The pellet was resuspended in 1 mL of 13 TNTM pH8 and, following overnight incubation, purified on a 5 to 60 % w/v sucrose step gradient for 3 h at 4 C at 210,000 3 g, using the same instrument with a SW 41 Ti swinging bucket preparative ultracentrifuge rotor.Both visible bands were aspirated by a single needle puncture and a 10-mL volume syringe.The purified fractions were dialyzed against PBS in order to remove the sucrose.

Experimental inoculation of Z. morio larvae
For direct fat body injections, a titered ZmBWV virus stock diluted to the desired concentration was used.Approximately 10 ml of diluted virus was injected into each larvae using a 1 ml insulin syringe with a delicate needle.The fat bodies of the first five abdominal segments were targeted by the needle.Mock-infected individuals were injected the same way with saline control (viz., 10 ml of PBS).To inoculate the larvae with contaminated food, we used deceased individuals that showed clear blackening.In case of Z. morio larval carcasses, one carcass was provided for every 10 healthy larvae, while one carcass was provided per every three healthy larvae in case the deceased T. molitor larvae.To infect the Z. morio larvae by dripping virus suspension on their cuticles, the ZmBWV virus stocks were diluted to the desired concentration with PBS to a volume of 100 ml, and this 100 ml aliquot was applied dropwise onto 10 healthy larvae (i.e., mean volume per larva is 10 ml).

Complete genome sequencing and phylogenetic calculations
Viral DNA was extracted either from purified virus particles or directly from insect specimens.In the first case 0.25 mg of full ZmBWV particles were incubated for three hours in 13 TE puffer, pH 8.7 (10 mM tris-HCl, 1mM EDTA) supplemented with 10 ml 10% sarcosyl and 4 ml proteinase K in 10 mg/ml concentration.The DNA was extracted by the DNeasy kit from Qiagen.To isolate viral DNA directly from insects, we pooled five individuals in a 5-ml conical tube and homogenized them in 1 ml TE buffer, using a handheld homogenizer with a sterile pestle.DNA was isolated from 100 ml homogenized suspension, which was digested overnight with the same components as the virions.The isolation step was carried out by the Monarch PCR & DNA Cleanup kit from New England Biolabs, utilizing the single-stranded DNA specific protocol.The acquired DNA preps were shipped to a commercial Oxford nanopore sequencing service, provided by Plasmidsaurus (Eugene, OR, USA), where they were processed as linear amplicons.
The obtained complete genomes were aligned by T-Coffee 80 and converted to Nexus format using Unipro Ugene. 81Model selection was carried out by the nucleic acid model selection subroutine of IQ-Tree. 82Phylogenetic inference was calculated by BEAST. 83ercent identity at the protein level was computed based on BLASTP homology searches.

PCR and DNA quantification
We designed the primer pair 5'-GACAGCGGATACTATGTGTCA-3' and 5'-AATTTCAAGAGGAAGTCTTTG-3', targeting an approximately 300-nt-long highly conserved region of the NS2 gene.The primers were designed to be capable of amplifying the respective genome region of all members within the Blattambidensovirus incertum1 species.Amplification was executed in a 25 mL final reaction volume, including 2 mL of purified DNA target, 0.5 mL of both primers in 50 pmol concentration, 0.5 mL dNTP mix with 8 mmol of each nucleotide, and 0.1 mL of DreamTaq DNA polymerase enzyme (Thermo Fisher).PCR reactions were executed under a program of 5 min denaturation at 95 C followed by 35 cycles of 30 s denaturation at 95 C, 30 s annealing at 50 C, and 1 min of elongation at 72 C. The final elongation step was 5 min long at 72 C. The same program was executed to analyze a variable region (positions 291-681 nt) of the NS3 gene, using the primer pair of 5' -ATATCACCCTCAGAGCTTGT -3' and 5'-TTCATTAGCAGATCCTGGAC-3'.These products were analyzed by Sanger sequencing via a commercial platform, provided by Genwiz (Azenta).
Quantification of the viral DNA was carried out by real-time PCR amplification (qPCR), using an Applied Biosystems QuantStudio 5 instrument.A 300-bp-long target sequence was amplified by the primers mentioned above.For dsDNA quantification the SYBR TM Green PCR Master Mix (Applied Biosystems) was used, with an amplification program of 5-min denaturation at 95 C followed by 45 cycles of 30 s denaturation at 95 C, 15 s annealing at 55 C, and 30 s of elongation at 72 C. Results were analyzed by the QuantStudio Pro software (Applied Biosystems).All reactions were carried out in triplicate, using three separate pools of five specimen to obtain biological triplicates.
Microscale X-ray computed tomography Z. morio larvae were fixed in 4% paraformaldehyde.To preserve the healthy larvae in a straightened-out position, they were first placed to 4 C for 15 min, and only in this immobile state were they placed in the paraformaldehyde solution, incubated at room temperature for the first 24h, and then stored at 4 C. Specimens were washed in PBS three times and then stained in aqueous 1% iodine with 2.5% potassium iodide to enhance internal features.Staining was carried out for 24 hours.Following staining, the sample was washed in a 0.9% sodium chloride solution.The specimens were scanned in a Skyscan 1272 instrument, at a voltage of 60kV with a 166 mm source current.Images were collected at the pixel size of 4.5 mm, as the frame average of three per a 4 rotation step.The completed 2D scan images were reconstructed by the Skyscan Nrecon software and rendered in the Amira software (Thermo Fisher).
For ion mass spectrometry, virus suspensions were mixed with concentrated double-distilled nitric acid in metal-free polypropylene centrifuge tubes, sonicated for 60 min, then digested using a CEM microwave system (CEM, Matthews, NC).Samples were diluted to 5% nitric acid with Milli-Q water, followed by serial dilution with 5% HNO 3 for analysis via inductively coupled plasma mass spectrometry (ICP-MS).Three replicates at masses 58 Fe, 59 Co, 60 Ni, 65 Cu, and 67 Zn were measured in deflector jump mode using a Nu AttoM High-Resolution ICP-MS.Calibration standards were prepared daily with metals concentrations ranging from 0.001-10 ppb, in 5% HNO3, with an instrument limits of detection of 0.01 (Co, Cu), 0.1 (Fe, Ni), and 0.25 (Zn).The limit of quantification (LOQ) was calculated as the limit of detection multiplied by the dilution factor, yielding 0.8 ppb (Co, Cu), 8 ppb (Fe, Ni), and 20 ppb (Zn).Quality control standards (NIST calibrants SM-1811-001 and SM-1811-005) were measured after every sixth sample to account for instrument drift and reproduced with RSD < 3.2% (n=3).A spiked blank was included to account for analyte recovery (83-102% for all metals), and a procedural blank was included to account for reagent background contribution and contamination, measuring below the LOQ for all metals except Cu (1.78ppb).Control parvoviruses used were Penaeus monodon metallodensovirus (PmMDV) and Acheta domesticus segmented densovirus (AdSDV).

ELISA
Anti-ZmBWV purified polyclonal rabbit antibodies were generated by GenScript by prime and boost injections of ZmBWV empty capsids of the UT-morio strain.Experimental samples were tested in duplicate.Post-boost antibodies were tested at a constant concentration of 1:5000 (0.1 mg/mL) against several dilutions of empty ZmBWV particles or AAV9.Untreated, purified total rabbit IgG was included as a negative control.Plates were coated for 1 hour at 37 C, blocked three times with superblock TBS (Thermo 37535) according to the manufacturer's protocol, and incubated with primary antibody at 4 C overnight.After three washes and incuation with 1:20000 (0.5mg/mL) anti-rabbit HRP secondary antibody for 1 hour at room temperature, the assay was developed with 1-Step Ultra TMB ELISA Substrate Solution (Thermo #34028) and absorbance read on a SpectraMax M5 (Molecular Devices).Preboost antigenicity was confirmed similarly.

Protease accessibility assay
In each treatment aliquot, 10 mg of ZmBWV virions or empty capsids were digested with 1 mg mass spectrometry grade trypsin (Thermo Scientific) or 16.6 U thermolabile proteinase K (New England Biolabs) in PBS at room temperature.Following incubation for 15, 30, 120, or 240 minutes, the aliquots were denatured at 95 • C for 10 min in the presence of 2-mercaptoethanol and sodium dodecyl sulfate and were subjected to SDS-PAGE and also to capillary gel electrophoresis (CGE) on a 2100 Bioanalyzer system (Agilent) for quantitation.For CGE, Agilent Protein 230 reagents were equilibrated to room temperature; thawed samples were combined with denaturing solution and heated at 95 C; samples were then quenched deionized water; and prepared samples were run on the bioanalyzer chip and analyzed.The proteinase-free control was denatured after 4 hours of incubation in PBS at room temperature.
Two replicates with each protease were executed.Band identity was confirmed by LC-MS/MS conducted as described above.In trypsinzed capsids, full-length bands completely disappeared for empty capsids but were only mildly reduced in intensity (relative to no-trypsin control) in full virions (Figure S4 A).The additional band $65 kDa band was confirmed by MS/MS to be a C-terminal fragment of cap2, comprising residues 141-593 (vs.4-593 in the uncleaved parent polypeptide).Proteinase K treatment produced a similar result; empty capsid VPs were cleaved already at the 15 min timepoint (Figure S4B).Control empty particles incubated in the absence of proteinase K still contained reduced VP1 compared to full particles (Figure S4B), even though freshly-purified empty particles have a significant amount of VP1 (Figure 2C).The exposed VP1 in empty particles may be being cleaved, either by the autoproteolytic activity of the VP1 unique region 84 or by residual cellular proteases.
Grid preparation, vitrification, and cryo-transmission electron microscopy Quantifoil R1.2/1.3 300 mesh grids were glow discharged and coated with a 2.62-nm-thick carbon film.The film was fabricated by electron-beam deposition on cleaved mica using a Leica EM ACE600 instrument and floated onto a surface of ultrapure water through which the discharged grids were lifted.In case of the NJ2-molitor and OR-molitor strains UltrAufoil R1.2/1.3 300 grids were used.Samples were plunged-frozen into liquid ethane using a Vitrobot Mark IV (FEI) at 100% humidity and ambient temperature, after blotting away the excess liquid for 5 seconds at an arm force of 0. The grids were clipped into autoloader grids and imaged using a Talos Arctica transmission electron microscope (TEM) (Thermo Fisher), equipped with a Gatan BioQuantum energy filter and K2 direct electron detector, operated in low dose mode with a slit width of 20eV.Data collection parameters and refinement statistics are shown in Table S5.In all cases a 100 mm objective aperture was employed.CryoEM grids were imaged using the aforementioned electron microscope operated at 200 kV, with a 10-s-long exposure and a total dose of 43.16 e -/A ˚2, using a frame length 0.2 s (UTmorio) and 32.38 e -/A ˚2 with a frame length of 0.1 s with an exposure of 3 s (NJ2-molitor).The OR-molitor dataset was collected at a total dose of 33.93 e -/A ˚2 with a 3-s-long exposure time and 0.1-s-long frame time.Movie frames were recorded in counting mode using the Serial EM suite 70 at a sampling of 1.038 A ˚/pixel (UT-morio) and 0.658 A ˚/pixel (NJ2-molitor, OR-molitor).To assess the impact of GmDV buffer on ZmBWV structure, ZmBWV strain UT-morio full virions were dialyzed against 6L of TBS buffer (10 mM Tris, 100 mM NaCl, 1.0 mM CaCl 2 , 1.0 mM MgCl 2 , pH 7.5) for 24 hours at 4 C with three changes of buffer.Dialyzed virions were vitrified on UltrAuFoil grids as above and imaged on a ThermoFisher Tundra TEM.

CryoEM data processing
The collected movies were aligned by the MotionCorr2 application with dose-weighting. 71The cisTEM software was used for singleparticle image reconstruction to obtain an initial model. 72High resolution single particle reconstruction was carried out by Relion 4.0 73 and cryoSPARC. 3,74Micrograph quality was assessed by CTF estimation using a box size of 512.The subset of micrographs with the best CTF fit values were included in further processing.Particles were automatically boxed by the particle selection subroutine of CisTEM, at a threshold value of 2.0 or by the blob picking subroutine of cryoSPARC Live, during on-the-fly processing.Boxed particles were subjected to 2D classification, imposing icosahedral symmetry at 35 classes.Particles of classes that failed to display a clear 2D-class average of the icosahedral particle were eliminated from the reconstruction.Ab initio model generation was carried out for 40 iterations, imposing icosahedral symmetry.The obtained startup volume was subjected to automatic refinement under icosahedral constraints and underwent iterations until pseudoconvergence.To improve the resolution, corrections for higher-order aberrations, beam tilt, trefoil and anisotropic magnification were applied, as well as per-micrograph astigmatism and per-particle CTF parameters.In case of Relion processing, particles were subjected to Bayesian polishing, with training on 10000 particles.The automatic high-resolution refinement was repeated in the presence of a soft-edged mask.The maps were subjected to sharpening or post-processing to obtain the final reconstructions.The resolution of each reconstructed map was calculated based on a Fourier shell correlation (FSC) of 0.143.
For the asymmetric genome reconstruction an initial model was created, using CisTEM, applying a C 1 symmetry operator.Particle subtraction was carried out in Relion 4 in order to remove the capsid shell.These particles were used with the C 1 initial model to perform automatic refinement preserving the C 1 symmetry.The obtained cryoEM maps were visualized in coot 76 to model the backbone of one subunit.Visualization was carried out by UCSF Chimera. 77The density was modeled and refined using coot and ISOLDE. 78The final refinement step was carried out by PHENIX, 79 refining all 60 copies of the icosahedral particle into the electron density, using icosahedral NCS restraints.The symmetry matrix was added in UCSF Chimera.
To construct difference maps, the maps were aligned and resampled onto a common grid, the radially-averaged structure factor of the lower-resolution map was applied to the higher-resolution map, intensities of one map were multiplied by a scale constant to set capsid intensities equal between the two maps, and then the maps were subtracted.All operations were performed in EMAN2. 75o segment densities of the unidentified moiety near Asn235 (Figures 6B and 6C), first, a low-pass filter with Gaussian falloff at 2 nm -1 frequency was applied to the map (to reduce the contribution of high-resolution noise and to highlight potentially flexible moieties).Second, the atomic model was used to split the map into three: density >2.5A ˚from any modeled atom, density <2.5A ˚from an atom of Asn235 and closer to that than any other modeled atom, and all other density <2.5A ˚from a modeled atom.
To calculate capsid interior volume, a theoretical map was computed from the protein structure using molmap (implemented in UCSF Chimera based on EMAN code), then the map was expanded/gaps filled to form a mask by adding two single-pixel shells in EMAN2, and a radial cutoff was applied to truncate connecting density through the five-fold pores.The mask was inverted and applied to a sphere of constant intensity, and the resultant blob whose shape and size correspond to the capsid interior was measured in Chimera.As an alternative approach, the theoretical map was inverted and subjected to watershed segmentation 85 and the segmented interior density was measured.This latter approach yielded larger interior volume estimates of 3.3310 6 A ˚3 for AAVhu.37 and 3.95310 6 A ˚3 for ZmBWV.
To reconstruct the 3D structure of the unknown iflavirus capsid, the particles of the NJ2-molitor dataset were subjected to 2D classification in cryoSPARC 3,74 with a larger mask diameter of 350 A ˚.The only class projection of the 35 classes that represented the slightly larger particles was re-extracted and used to construct an ab initio model volume.This volume was used in homogenous refinement to obtain the high-resolution structure with icosahedral symmetry imposed.A subvolume of the final map was extracted using UCSF Chimera, 77 approximately corresponding to the presumptive asymmetric unit, and was subjected to de novo atomic model building by ModelAngelo. 10The longest amino acid chain was selected for homology and structural similarity search by HHblits 19 and DALI, 18 respectively.This study was conducted as an identical setup to the experiment in Figure S2C, with samples being pulled and subjected to PCR amplification twice every week, following a 3-week-long pre-inoculation interval with the non-pathogenic NJ2-molitor strain only.

Figure 1 .
Figure 1.Discovery of a pathogenic parvovirus via cryo-EM (A) Top: deceased superworms (Zophobas morio) typifying the blackening in Z. morio black wasting disease (ZmBWD).Bottom: progression of ZmBWD in 4-week-old Z. morio larvae.Larva source is presented in Table S3.(B) Sucrose step gradient from blackened Z. morio larvae displaying two fractions (arrows) at the 20% and 30% sucrose interfaces, respectively, which appear blue in fluorescent light due to the presence of viral particles.(C) Electron micrographs displaying the particles of $28 nm from those fractions.Insets emphasize the absence of genome inside particles of the 20% band and presence inside particles of the 30% band.(D) Isosurface rendering (left) and central slice through (right) 3D reconstructions of the particles purified from diseased larvae.The gold line separates reconstructions of the 20% band (upper right) and 30% band (lower left).The protein structure is almost identical between the two bands, but only the 30% band contains genome.The initial polyalanine chain is displayed as a ribbon diagram, with each T = 1-related subunit in a different color.Results of searches based on this map are presented in Tables S1 and S2.(E) Comparison of the capsid surface of the densovirus described herein associated with ZmBWD and consequently designated as Z. morio black wasting virus (ZmBWV) with those of other members of the Parvoviridae family: subfamilies Densovirinae (Galleria mellonella densovirus [GmDV], Acheta domesticus densovirus [AdDV], and Bombyx mori densovirus [BmDV]), Hamaparvovirinae (Penaeus stylirostris densovirus [PstDV]), and Parvovirinae (adeno-associated virus 2 [AAV2]), colored by radial distance.(F) Discovery of an iflavirus at low concentration alongside NJ2-molitor.Left: extract of a micrograph showing an iflavirus virion (arrow) alongside a ZmBWV virion.Right: automated model of the reconstructed density, used for structure-based identification.

Figure 2 .Figure 3 .
Figure 2. Characterization of the genome, proteins, and phylogenetic relationships of the densovirus, ZmBWV (A) The I-shaped ssDNA secondary structure of the ITR, flanking the ZmBWV genome.The presented DNA region represents the entire left ITR, and the bases forming the terminal hairpin are highlighted in green.(B) Top, diagram of the ZmBWV genome showing ITRs (rectangles) and ORF locations (arrows) The pink and teal boxes indicate the respective size and position of the superfamily 3 helicase (SF3) and the phospholipase A2 (PLA2) domains, respectively.Bottom, diagram showing the putative expression pattern of the structural proteins (VPs) from cap1 and cap2 based on the tandem mass spectrometry data (Figure S1A).(C) SDS-PAGE gel showing the four structural proteins of ZmBWV, present in both the empty and virion fractions.Mass spectrometry is presented in Figure S1.(D) Inferred phylogeny of the complete coding sequences of all ZmBWV strains identified in this study (shown in bold), plus members of the species Blattambidensovirus incertum1, derived mostly from metagenomic studies.The tree is outgroup-rooted to the closest related species of the same genus.Scale bar: substitutions per kilobase.Node labels: posterior probability.See also Figure S1.

Figure 4 .
Figure 4. Cross-species pathogenesis of ZmBWV (A) Line diagrams of the survival rate and the prevalence of symptoms in case of 4-week-old Z. morio larvae inoculated by direct fat body injections of T. molitorderived NJ2-molitor strain.Figure S2B presents the control experiment to this study, inoculating the negative larvae of the same stock using the pathogenic representative UT-morio ZmBWV strain.Negative control received no injection.Mock-infection was with sterile PBS.(B) Line diagrams comparing the survival rate and prevalence of symptoms in z4-week-old Z. morio larvae inoculated by UT-morio or NJ2-molitor ZmBWV strains via feeding.Controls as (A).(C) Box-and-whisker plot showing viral titers of Z. morio larvae following inoculation by the strain NJ1-molitor.Titer values were determined by qPCR.t.p.i., time post-inoculation.(D) Box-and-whisker plot showing the ZmBWV titer in the tenebrionid beetle Alphitobius diaprenius and the blattodean roach Blaptica duibia.Both were obtained from a facility in New York that experienced Z. morio colony collapse due to ZmBWV.* indicates uninfected larva.

Figure 5 .
Figure 5. Intersubunit interactions of the ZmBWV capsid (A) Ribbon diagrams of the atomic model of one subunit of the capsid.Colored arrows mark the first ordered residue of each model.The b strands comprising the jelly roll core are labeled.The 5-fold symmetry axis is labeled by a pentagon, the 3-fold axis by a triangle, and the 2-fold axis is marked by the ellipsoid.(B) Ribbon diagram of adjacent subunits.Note the domain-swapping conformation of the N termini, composing an additional, intersubunit b sheet (cyan).In this conformation, the bA strand can interact with the bF strand of the 3-fold neighboring subunit (red box).This interaction only forms in the presence of a packaged genome (see inset).Comparison to homologs is shown in Figure S5A.

Figure 6 .
Figure 6.Non-protein densities in the cryo-EM map of ZmBWV (A) Excess exterior density (magenta) not accounted for by protein (cyan) in the ZmBWV icosahedral map is concentrated at one location per ASU.(B) NJ2-molitor map filtered to 5 A ˚resolution showing capsid residue Asn235 (blue), other capsid density (gray), and non-protein density (orange).(C) No ordered density for the unidentified moiety is seen adjacent to Asn235 in the map of OR-molitor filtered to 5 A ˚(colored as above).(D) Virions of the NJ2-molitor strain include 17 ordered nucleotides per subunit, surrounding the 3-fold symmetry axis.Interchain hydrogen bonds in yellow.Nucleotide properties are described in TableS5.(E) A column of stacked bases, including a single base from the 9-mer under three bases from the 5-mer.(F-H) Three base-binding pockets on the capsid interior surface: a pyrimidine (modeled as cytosine) of the 3-mer apposes two lysine residues (F); an adenosine of the 9-mer is base-flipped into a pocket (G); and a cytosine of the 5-mer base-flips to p-stack against Tyr404 while oriented to form a hydrogen bond with Glu493 (H).In (F)-(H), density attributable to DNA is translucent pink (3-mer), purple (9-mer), or dark purple (5-mer), protein density is gray, and water density is red; in (B), (C), and (E)-(H), protein carbons are displayed in cyan, DNA carbons in pink/purple/dark purple (as for the DNA density), and heteroatoms as per standard conventions.See also FigureS6.

Figure S1 .
Figure S1.Mass spectrometry analyses of the ZmBWV proteins, related to Figures 2 and 5

(
A) Mass spectra derived from the protein sequencing using tandem mass spectrometry (MS/MS).The peaks are annotated by the position of each peptide sequence within the cap2 ORF (marked by green if present in all bands, marked by red in case located in the N-terminal region of the cap2-dericed putative protein) or within the cap1 ORF (peptides marked by blue).(B) Mass spectrum of the 65 kDa band, which is produced upon trypsin treatment of the ZmBWV virions, as a cleaved product of the surface-exposed VP1 and VP2 N termini.(C) Results of the inductively coupled plasma mass spectrometry (ICP-MS), targeting the identification of metal ions present in the intact Z. morio black wasting virus (ZmBWV) capsid.As control, the Acheta domesticus segmented densovirus (AdSDV) virions were used.

Figure S2 .
Figure S2.Experimental characterization of the ZmBWV pathogenesis, related to Figure 4

(
A) BGE-2 cells at 203 magnification, 7 days post-inoculation with either saline (control) or with ZmBWV UT-morio at an MOI of 10 4 .(B) Results of the cross-strain experiment control to Figure4A, obtained by administering direct fat body injections using the UT-morio strain.(C) Results of the direct fat body injections leading up to the cross-protection study (and the pilot experiment leading up to Figure4C), involving only NJ2-molitor as the protective strain.

Figure S3 .
Figure S3.Electrophoretograms, obtained by Sanger sequencing, of a short PCR product targeting a polymorphic region of the NS3 gene, related to Figure4

Figure S4 .
Figure S4.Characterization of the ZmBWV structural protein N-terminal externalization dynamics, related to STAR Methods (A) Coomassie-stained SDS-PAGE image of the results of the trypsin protease accessibility assay to investigate the externalization pattern of the four capsid protein N termini in case of empty capsids and virions, respectively.The green arrows mark cleavage products, while the small black (VP1), blue (VP2), purple (VP3), and magenta (VP4) arrows mark the ZmBWV structural proteins.The time intervals refer to how long the aliquot was subjected to trypsin treatment before denaturation.(B) Capillary gel electrophoresis analyses of thermolabile Proteinase K treated ZmBWV empty capsids and virions.Capsid proteins and cleavage products are labeled with arrows, as in (B).The electrophoretograms display the results of the incubated yet untreated control as well as the results of the longest time point.

Figure S6 .
Figure S6.Polymorphic sites on the ZmBWV capsid surface, related to Figure 6 (A) Summary of the positions and variations of each polymorphic residue.The numbers parenthesis indicate the number genotypes possessing the given residue variant.Positions marked with an asterisk are only polymorphic in one genotypic variant.(B) Polymorphic residues are shown at the capsid surface of ZmBWV.Position 335 is highlighted in blue, as it plays a critical role in the ordering of the branching glycan associated with Asn235.Polymorphic residues are highlighted in magenta, and residues that appear polymorphic in only one genotypic variant are shown in green.

TABLE
(Continued on next page)