The small non-coding RNA Vaultrc5 is dispensable to mouse development

Vault RNAs (vRNAs) are evolutionarily conserved small non-coding RNAs transcribed by RNA polymerase lll. Initially described as components of the vault particle, they have since also been described as noncanonical miRNA precursors and as riboregulators of autophagy. As central molecules in these processes, vRNAs have been attributed numerous biological roles including regulation of cell proliferation and survival, response to viral infections, drug resistance, and animal development. Yet, their impact to mammalian physiology remains largely unexplored. To study vault RNAs in vivo, we generated a mouse line with a conditional Vaultrc5 loss of function allele. Because Vaultrc5 is the sole murine vRNA, this allele enables the characterization of the physiological requirements of this conserved class of small regulatory RNAs in mammals. Using this strain, we show that mice constitutively null for Vaultrc5 are viable and histologically normal but have a slight reduction in platelet counts pointing to a potential role for vRNAs in hematopoiesis. This work paves the way for further in vivo characterizations of this abundant but mysterious RNA molecule. Specifically, it enables the study of the biological consequences of constitutive or lineage-specific Vaultrc5 deletion and of the physiological requirements for an intact Vaultrc5 during normal hematopoiesis or in response to cellular stresses such as oncogene expression, viral infection, or drug treatment.


INTRODUCTION
Vault RNAs (vRNAs) are small non-coding RNAs initially identified in rat livers 1 through their association with the vault complex, the largest naturally occurring cellular particle described to date [2][3][4][5] .Both vaults and vRNAs are evolutionarily conserved, and the particle itself is present at high copy numbers in many cellular contexts suggesting it has important functions in eukaryotes, though these remain to date poorly defined 6,7 .Although initially identified via its association with the vault, only a minority of the vRNA is typically bound to that complex 8 and many alternative functions for this RNA have been proposed.In humans, vRNAs are expressed from four related loci (vtRNA1-1, vtRNA1-2, vtRNA1-3, and vtRNA2-1) and are thought to serve as noncanonical precursors for miRNAs whose processing is DICER-dependent but DROSHA-independent [9][10][11][12] .Processing of vault RNAs into these miRNAs (also known as small-vaultRNAs; svRNAs) seems to be a regulated process 9,10 suggesting they may act in a contextdependent manner.Studies in cell culture have suggested these svRNAs are important for proper cell differentiation 10 , required for the regulation of cell cycle progression and apoptosis 13,14 , implicated in the regulation of drug-metabolism 11 , and dysregulated in both cancer [15][16][17] and neural diseases 18 .More recently, studies in human and mouse cells have shown that unprocessed vRNAs from both species can act as riboregulators of p62 to control autophagy 19,20 a process with critical roles in cell differentiation, animal development, and human disease 21,22 .Surprisingly, despite the mounting evidence implicating both full length and processed vRNAs in the regulation of fundamental biological processes, their physiological requirements are not known.In contrast to humans, mice have a single vRNA (Vaultrc5 or Mvg1) 23 .The existence of a sole vRNA gene in murine genomes represents an invaluable opportunity to define the physiological requirements of an abundant but mysterious class of non-coding RNA molecules in mammals.With this purpose, we have generated a new mouse model carrying a conditional loss-of-functional allele for Vaultrc5 and performed the initial characterization of Vaultrc5-null animals.We show that constitutive loss of Vaultrc5 in vivo is compatible with animal development and survival.We also find that Vaultrc5-null animals are histologically indistinguishable from their wild-type littermates with no detectable gene expression changes in either livers or brains, two organs where vRNAs have been proposed to play important functions.Nevertheless, we have observed a minor reduction in platelet counts in the absence of Vaultrc5, suggesting vRNAs may have important roles in hematopoiesis.Our work represents the first step towards defining the functions of mammalian vRNAs in vivo.This mouse model will enable more detailed studies using targeted and acute deletions of this conditional allele to further define the physiological roles of vRNAs during tissue development and homeostasis or in response to viral infections, oncogenic insults, or other cellular stresses.

MATERIALS AND METHODS
Mouse husbandry and transgenic lines.Mouse beta-actin-Cre line has been previously described 24 and was obtained from the Jackson Laboratory (strain 019099).Vaultrc5 flx mice (carrying loxP sites around Vaultrc5 gene) will be made available through the Jackson Laboratory as JAX Stock No. 037602.These animals were generated by zygotic injection of a single-stranded donor DNA template ordered from IDT and in vitro assembled Cas9-gRNA ribonucleoprotein complexes.Reagents were generated and tested by the Genome Modification Core at Frederick National Lab for Cancer Research and used in targeting experiments by the Mouse Modeling & Cryopreservation Core.Super-ovulated C57Bl6NCr female mice were used as embryo donors.Animals were genotyped using PCR followed by Restriction Fragment Length Polymorphism (RFLP).The 3' loxP site or the 5' loxP were amplified using (3'LoxP, 5'-GAATCCGCGGAACTTTGG-3', 5'-AATGCATACACAGGAGAGTTTCA-3'; 5'LoxP, 5'-AGGCAACCCATCTCTTATT-3', 5'-GAGATGACAGACCAATCGG-3'), which amplify a 1140-bp band from (3' LoxP) or a 1113-bp band (5' LoxP).PCRs were cleaned up and digested with Xmnl for one hour.The resulting fragments were visualized by agarose gel electrophoresis.Correct integration in the founder male was confirmed by amplifying genomic DNA from tail clippings using primers flanking the targeting construct (5'-AGGCAACCCATCTCTTATT-3', 5'-TGCATGTTAAAAACCCTCAGAAC-3'), cloning the resulting amplicon into the TOPO vector, followed by sanger sequencing of multiple clones.A colony for this mouse strain was established by crossing the founder to C57Bl6 females obtained from The Jackson Laboratory (Sock No: 000664).All animal procedures were conducted according to the NIH Guide for the Care and Use of Laboratory Animals, under Animal Study Proposal no.390923 approved by relevant National Institutes of Health Animal Care and Use Committees.

Conservation tracks
Vertebrate and Placental Mammal basewise conservation tracks were generated by PhyloP, downloaded from the UCSC genome browser, and visualized on IGV.

TargetScan predictions
Datasets and Perl scripts were used from 7.2 release of TargetScanMouse and TargetScanHuman 26 .First, the conserved miRNA targets and non-conserved sites were identified using a custom set of data available on TargetScan for both mouse and human 27,28 .Second, conserved branch length (sum of phylogenetic branch lengths between species that contain a site) and P CT (probability of preferentially conserved targeting) for each predicted target in a custom set of data was calculated 29 .Third, the context++ scores for a set of predicted miRNA sites in a custom set of data was calculated.The context++ score for a specific site is the sum of the contribution of 14 features described in 30 .A list of targets with predicted miRNA binding sites in their 3' UTRs was generated based on conserved branching length, P CT and context++ score.The entire analysis was done on Homo sapiens, Mus musculus, Macaca mulatta, Pan troglodytes, Bos taurus, Rattus norvegicus and Monodelphis domestica which were available on TargetScan.
Histology and blood collection Eosin-Hematoxylin staining was performed on 5 μm sections of tissues fixed in 10% neutral buffered formalin and embedded in paraffin.Blood was collected by cardiac puncture in anesthetized animals.Animals were euthanized immediately upon completion of blood collection in K2-EDTA collection tubes (Thomas scientific).Complete blood counts as well as histopathology analysis were performed at MD biosciences.
Small RNA sequencing analysis.
Total RNA sequencing.

Data availability.
RNA sequencing datasets have been deposited to GEO and are available under GSE269048.

RESULTS
Analysis of non-canonical miRNAs derived from the murine Vaultrc5 locus.Mouse genomes encode a sole vault RNA gene, Vaultrc5 (also known as Mvg1) 32 .Vaultrc5 is transcribed as a single exon-transcript from chromosome 18 (18qB2) from a locus located immediately downstream of Zmat2 (Supplementary Figure 1A) and that is in a syntenic region conserved across mammals 23 .All vRNAs are characterized by their ability to associate with the Vault particle, are often transcribed from analogous genomic locations, and have been described in eukaryotes ranging from amoebas to humans.However, the Vaultrc5 locus shows relatively low sequence conservation to other vertebrates or placental mammals when compared to that of protein coding genes (Supplementary Figure 1A).Nevertheless, a few features of the locus show some conservation.First, like vault RNAs of other species, the regulatory elements in its polymerase III (pol III) promoter are well defined (Supplementary Figure 1B).These include a proximal element and a TATA-like box 32 .Both of these are also found in the promoters of human vtRNA1 genes (vtRNA1-1, vtRNA1-2, vtRNA1-3) 32 , which like Vaultrc5 are located downstream of Zmat2.Moreover, at least the proximal element shows broad conservation amongst placental mammals and vertebrates as does the polymerase III termination sequence (Supplementary Figure 1B) (see also 23 ).Within the gene body itself, Vaultrc5 contains three regulatory sequences in the form of the A and B boxes 5 , which characterize pol III type 2 promoters such as those found in tRNA genes 33 .Only two of these (A box and B2 box) are broadly conserved, as are the sequences that surround them (Supplementary Figure 1B).In contrast, the central region of vRNAs varies substantially between species 23 .In the case of Vaultrc5, it contains a third pol III internal sequence known as B1 box, with low overall conservation with other species (Supplementary Figure 1B) with the notable exception of rats 32 .Despite these differences, the 3' and 5' ends of Vaultrc5, are predicted to form a double stranded region that is identical to that predicted for human vtRNA1 genes, as exemplified here by vtRNA1-1 (Figure 1A).Importantly, although the sequences predicted to fold into this double stranded structure are highly conserved between vRNAs of the two species (Supplementary Figure 1C), they share only 83-89% of nucleotide identity.Yet, every position at which the sequence of the mouse Vaultrc5 differs from that of one of more vtRNA1 gene, the change either has little impact on base pairing (e.g.G•C pairing between bases 3 and 136 of Vaultrc5 versus G•U wobble pairing between bases 3 and 92 in vtRNA1-1) or is compensated by changes in the complementary RNA sequence (Supplementary Figure 1C, Figure 1A).This suggests that the secondary structure of the stem may play important roles for the functions of vRNAs, and that those are conserved between mice and humans.Given that for human vRNA1-1 this stem is processed by DICER into non-canonical miRNAs 11 (Figure 1B), we thought it was likely that identical molecules are produced from Vaultrc5 as well (Supplementary Figure 2A).To investigate this possibility, we analyzed small RNA sequencing datasets from a variety of adult mouse tissues 31 .We found that reads matching the predicted location of the mature miRNA sequences were relatively low abundant (Figure 1C, 1D, Supplementary Figure 2 B-D).Of all four predicted noncanonical miRNAs derived from this locus, svRNA* had the highest number of reads.Nonetheless, they were much less abundant than those matching canonical miRNAs.as an example, miR-19a-3p a miRNA with functions in hematopoietic tissues reached about 9,000 RPM in bone marrow and spleen samples, with a mean of around 2,600 RPM across all tissues (Figure 1D).In contrast, reads for the predicted vaultRNA-derived miRNA-like fragments had an abundance similar to that of miR-19a-5p, the passenger strand of miR-19a which is degraded following loading of the mature miRNA into an Argonaute protein (miR-19a-5p max=39.3RPM, mean=6.3RPM; svRNAa max=5.2RPM, mean=0.7 RPM; svRNAa* max=130.8RPM, mean=20.7;svRNAb max=3.8RPM, mean=0.5 RPM; svRNAb* max=3.8RPM, mean=0.5 RPM).Low counts were observed even in lung, where the levels of vaultRNA-derived miRNAs were reported to be high in humans (mean counts svRNAa=0.4RPM, svRNAa*=25.5 RPM, svRNAb=0.2RPM, svRNAb*=0.2RPM) 11 .Given that the strength of repression imparted by a miRNA depends strongly on its abundance 34 and that even well-expressed miRNAs like miR-19a have typically only modest effects on target gene expression these data suggest that under homeostatic conditions Vaultrc5 is not a source of non-canonical miRNAs with physiologically relevant functions in the tissues we have analyzed.Previous work has suggested that in humans, svRNAb plays a role in drug resistance through the regulation of target genes such as CYP3A4 11 .Based on this, it is possible that vRNA-derived miRNAs perform conserved regulatory functions only under stress conditions.We think that is unlikely to be the case.First, miRNAs recognize their targets primarily through base-pair complementarity via their extended seed regions (miRNA nucleotides 2-8) 35 .Although the sequence variations between the vtRNA-1 and Vaultrc5 genes preserve the secondary structure of the vaultRNA stem, they do not preserve the seed of all predicted svRNAs, specifically svRNAa and svRNAb* (Supplementary Figure 2A).Second, although the predicted seed sequences of svRNAb and svRNAa* is shared between the two species, there is poor overlap between their predicted targets, based on our own implementation of the targetscan algorithm 26 (Supplementary Table 1).
Generation of a Vautrc5 conditional loss-of-function allele.Despite the lack of evidence that Vaultrc5 functions as a non-canonical precursor for murine miRNAs, the conserved secondary structure of its stem across species suggests this non-coding RNA plays important functions in vivo, potentially as a component of the Vault particle or in riboregulatory role of p62.In fact, Vault RNAs have been functionally implicated in processes such as cell survival 36 , proliferation 37 , and differentiation 10,38 , all of which are essential for successful embryonic development.The expression of full length vRNAs also seems to be regulated during cellular differentiation 38 .Together these reports suggest that full-length vault RNAs may be essential during mammalian embryogenesis.To help define what those functions might be we generated a conditional loss-offunction allele that would allow both constitutive as well as spatiotemporally controlled deletion of the locus enabling the dissection of the physiological requirements for Vaultrc5 expression in mice.We designed a construct in which loxP sites were inserted directly upstream and downstream of the annotated promoter and termination elements respectively without disrupting them (Figure 2A, Supplementary Figure 1B).Both loxP sequences were inserted at genomic sites with no sequence conservation amongst vertebrates or placental mammals suggesting they do not unintentionally disrupt regulatory elements of neighboring genes.Moreover, no conserved sequences aside from those of Vaultrc5 are predicted to be affected by Cre-mediated recombination of the loxP sites.Of note, putative enhancer elements as well as CTCF binding sites are found in the vicinity of the Vaultrc5 but, as above, none are affected by our targeting or by the subsequent CRE-mediated genomic deletion.To facilitate the genotyping of the Vaultrc5 locus following targeting we included an XmnI restriction site next to each loxP sequence (Supplementary Figure 1D).

Constitutive loss of Vautrc5 is compatible with mouse development.
To test the broad requirement for vRNA expression during mouse development, we crossed Vaultrc5 flx/+ mice to the general beta-actin-Cre deleter line 24 .This led to widespread recombination of the floxed alleles and to offspring that were heterozygous for a Vaultrc5 null allele (Vaultrc5 -/+ ; Figure 2A).Subsequent intercrossing of heterozygous Vaultrc5 -/+ mice generated full Vaultrc5 loss-of-function animals, carrying the null allele in homozygosity (Vaultrc5 -/-) (Figure 2B).Quantification of Vaultrc5 in cells from these animals by RT-qPCR confirmed a genotype-dependent reduction of Vaultrc5 RNA levels with no detectable transcripts in Vaultrc5 -/-cells (Figure 2C).In contrast, the expression of Zmat2 and Pcdha1 was not altered in these samples (Figure 2D), confirming that our targeting strategy did not disrupt their regulation.The recovery of Vaultrc5 -/-animals at weaning suggests that complete loss of vRNA expression in mice is compatible with animal development.Nevertheless, many genes that have well established essential functions during embryogenesis -including those in the canonical miRNA pathway-lead to lethal phenotypes with incomplete penetrance when disrupted [39][40][41][42] .We found no evidence of that being the case for Vaultrc5.Specifically, data collected over multiple litters showed that Vaultrc5 -/-animals were recovered at the expected ratio at weaning (Figure 2E).Similarly, we found no deviation from the expected ratios between sexes (Figure 2F).Together, these data indicate that mice are able to successfully complete development and survive in the absence of vRNAs.
Vaultrc5-null mice are histologically normal.Aside from being viable, Vaultrc5 -/-animals were morphologically indistinguishable from littermate controls suggesting that constitutive absence of vRNAs does not lead to gross phenotypic abnormalities in mice.Furthermore, we found no weight differences-a common indicator of suboptimal animal health-between genotypes either in male (median weights of 18.25g, 18.05g, 18.55g for wild-type, heterozygous, and homozygous respectively) or female animals (median weights of 15.75g, 15.9g, 15.8g for wild-type, heterozygous, and homozygous respectively) (Figure 3A).To test if loss of Vaultrc5 caused more subtle phenotypes, we collected tissues from Vaultrc5 -/-and Vaultrc5 +/+ littermates of both sexes at 8 weeks of age and subjected them to a histopathological evaluation (Figure 3B-4E, Supplementary Note 1).We were particularly interested in brain tissues as previous studies had implicated vRNAs in neuronal differentiation 43,44 .In line with a potential role for vRNAs in the brain, their dysregulation has also been implicated in neurodegenerative diseases 45 .Despite these reports, we found that in the absence of Vaultrc5, the neural tissues were seemingly uncompromised.Specifically, cortex, medulla, hippocampus, brain stem, corpus callosum and cerebellum all showed no histological differences between Vaultrc5 -/-and Vaultrc5 +/+ animals (Figure 3B).Meninges were present on the cortex, and the parenchyma of the grey and white matter consisted of fine capillaries, glial cells, neurons, and abundant neuropil.The cerebellum of mutant animals was equally unremarkable and characterized by a very cellular granular layer and a less cellular molecular layer, with the prominent Purkinje cells at the interface of the two layers.Finally, the delicate pia mater lined the cerebellum along the molecular layer in animals of both genotypes.In sum, we found no evidence that constitutive loss of vRNA impacts neural tissues in mice, with the histologic findings in both Vaultrc5-null and wild-type controls being within normal limits and consistent with normal brain development and tissue differentiation.In addition to neural tissues, Vaultrc5 may also impact liver biology since vRNAs were initially identified as components of the Vault particle from rat liver extracts 1 , an organ in which the murine Vaultrc5 is well expressed 32 .Moreover, this organ is characterized by high levels of autophagy 46 , a process that seems to be regulated by vRNAs through their interaction with p62 via the central region (Figure 1B) 19 .Although the predicted structure of this region has limited similarity between vtRNA1-1 and Vaultrc5 (Figure 1A), binding of p62 to vault RNAs seems to be conserved between the two species 19 .Yet, despite the importance of autophagy to hepatic functions 46 , histopathological analysis showed no evidence that the development or structure of the liver was compromised in the absence of Vaultrc5.Specifically, the liver in animals of both genotypes was encapsulated and composed of hepatocytes with abundant granular pink cytoplasm.In both cases the organ was organized as cords of cells with a zonal arrangement around both portal triads and central veins.Macrophages and scattered mononuclear cells were also identified throughout the parenchyma (Figure 3C).Similarly, in our analysis of other major organs including the lung, heart, kidney, spleen, intestine, pancreas (Figure 3C) as well as ovaries, and testis (Figure 3D) we found no evidence of histological differences between Vaultrc5 -/-and Vaultrc5 +/+ animals.A detailed description of the histological findings in these organs can be found in Supplementary Note 1. Finally, we analyzed the histology of bone and spleen sections in our mice as these are major hematopoietic and lymphoid organs and vRNAs have been previously implicated in immune cell functions [47][48][49] .We found no histological differences between the spleens of Vaultrc5 wild-type and histologically identical (Figure 3E).In both cases bone marrows showed high cellularity and tri-lineage hematopoiesis.Within this cell population, megakaryocytes were the most abundant cells with erythroid and granulocytic precursors easily observed at higher magnifications.In all cases these cells showed no morphological abnormalities in the absence of Vaultrc5.As part of our pathology analysis, we also performed a complete blood count.We found that measurements for Vaultrc5 -/-animals as well as for Vaultrc5 +/+ littermate controls were within the normal physiological ranges expected for the C57Bl/6 strain (Mouse Phenome Database, www.jax.org/phenome).They were also for the most part identical between wild-type and mutant mice with two notable exceptions (Supplementary Figure 3).First, although we found no statistical difference in white and red blood cell counts (Supplementary Figures 3A and 3B), red blood cell size (Supplementary Figure 3E, 3F), or hemoglobin levels (Supplementary Figure 3G-I) between Vaultrc5 -/- and Vaultrc5 +/+ animals, hematocrit values-the percentage of red blood cells in the blood-were slightly reduced in mutant mice (Supplementary Figure 3D).This suggests that Vaultrc5 may have essential albeit subtle roles in erythropoiesis.Second, the platelet counts were reduced in the absence of Vaultrc5 to approximately 78% of the values in wild-type animals (Supplementary Figure 3C).This may point to a potential role for Vaultrc5 in platelet development from megakaryocyte precursors in the bone marrow or in the regulation of platelet survival.Nevertheless, platelet values remained well within the normal physiological ranges expected for the C57Bl/6 strain (Vaultrc5 +/+ =977-997 x 10 3 platelets/μl; vautrc5 -/-=745-792 x 10 3 platelets/μl; C57Bl/6=562-2159 x 10 3 platelets/μl).
RNA sequencing profiles of wild-type and Vaultrc5 -/-animals.We reasoned that even in the absence of major histological changes, loss of Vaultrc5 could result in molecular phenotypes that would be reflected in transcriptome dysregulation.To determine if that was the case we performed RNA sequencing to the brain and liver of vautrc5 -/-animals, alongside those of age-and sex-matched wild-type littermate controls (Figure 4).As before, we chose to focus on these organs because previous studies pointed to a role for vRNA in their development or homeostasis.For example, in vitro studies both human vtRNA1-1 43 and mouse Vaultrc5 44 promote synapse formation by modulating MAPK signaling and deregulation of the human vtRNA2-1 was proposed to induce neural dysfunction in humans 45 .Despite this, we found no evidence of gene dysregulation caused by absence of vRNA in mouse brains (Figure 4A).Similarly, human vtRNA1-1 was shown to promote cell proliferation and tumorigenesis in human liver cell lines 37 , which together with the abundant expression of vRNAs in the liver 32 suggests a potential role for vault RNAs in this organ.Yet, we again found no gene significantly dysregulated in the Vaultrc5 -/-samples (Figure 4B).Together with the histological analysis to these organs, these data suggest that in the mouse, vRNA has no significant impact to the development or homeostasis of the brain or liver.

DISCUSSION
We describe here the generation of Vaultrc5 conditional knockout mice and the initial characterization of animals with constitutive deletion of this locus.Given that mice-like most mammals-have a single vtRNA1 gene and have in addition lost vtRNA2 23 , Vaultrc5 -/-animals are null for vRNAs.This is to our knowledge the first characterization of the physiological requirements for this evolutionarily conserved but poorly understood class of small RNAs in vertebrates.Despite the myriad of roles assigned to vRNA genes in humans, ranging from regulation of cell proliferation, survival, and differentiation, we found that animals null for Vaultrc5 are viable and histologically normal.This suggests that vRNAs may not be essential in mice under unchallenged conditions.This is similar to what has been described for the other components of the Vault particle.Indeed, mice null for the Major Vault Protein (Mvp -/-), the telomerase-associated protein 1 (Tep1 -/-), or the vault poly-(adenosine diphosphate-ribose) polymerase (Vparp -/-) are all seemingly phenotypically normal [50][51][52] even when multiple genes are deleted 52 .Yet, when challenged with specific pathogens Mvp - /-animals show increased resistance to infection 53 .Similarly, Varp -/-animals show increased susceptibility to carcinogen induced tumorigenesis 54 .Thus, as is the case for many other genes, it may be that vRNA expression only becomes indispensable under particular conditions such as aging, oncogenic stress, or viral infections all of which have been previously linked to vRNA functions.Hematopoiesis may be particularly sensitive to loss of vRNAs since we have observed a small but significant reduction in the platelet counts in our animals compared to littermate controls.Follow-up studies breeding the conditional Vaultrc5 fx/flx animals with lineage restricted deleters such as Vav-Cre (leading to recombination of the locus in all cells of the hematopoietic system) 55 or Gp1ba-Cre (leading to recombination in megakaryocytes) 56 combined with both immune challenges and a more detailed characterization of the immune system will be essential to fully characterize these phenotypes, define the extent to which Vautrc5 is required for hematopoiesis, and understand the predominant mechanism through which this evolutionarily conserved small non-coding RNA regulates mammalian physiology.To facilitate those studies, we are making our mice freely available to the community through The Jackson Laboratory mouse strain repository.

Figure 1 .
Figure 1.Structure and conservation of the murine Vaultrc5 locus.(A) Minimum free

Figure 2 .
Figure 2. A conditional loss of function allele for Vaultrc5.(A) Schematic representation of conditional Vaultrc5 locus (Vaultrc5 flx ) and its recombination after Cre expression to generate a

F
Figure 2