MicroRNA-based control of tick-borne flavivirus neuropathogenesis: challenges and perspectives

Abstract

In recent years, microRNA-targeting has become an effective strategy for selective control of tissue-tropism and pathogenicity of both DNA and RNA viruses. Previously, we reported the successful application of this strategy to control the neurovirulent phenotype of a model chimeric tick-borne encephalitis/dengue type 4 virus (TBEV/DEN4), containing the structural protein genes of a highly virulent TBEV in the genetic backbone of non-neuroinvasive DEN4 virus. In the present study, we investigated the suitability of this approach for the attenuation of the more neurovirulent chimeric virus (TBEV/LGTV), which is based on the genetic backbone of the naturally attenuated member of the TBEV serocomplex, a Langat virus (LGTV). Unlike the TBEV/DEN4, the TBEV/LGTV virus retained the ability of its parental viruses to spread from the peripheral site of inoculation to the CNS. We evaluated ten potential sites in the 3′NCR of the TBEV/LGTV genome for placement of microRNA (miRNA) targets and found that the TBEV/LGTV genome is more restrictive for such genetic manipulations compared to TBEV/DEN4. In addition, unlike TBEV/DEN4 virus, the introduction of multiple miRNA targets into either the 3′NCR or ORF of the TBEV/LGTV genome had only a modest effect on virus attenuation in the developing CNS of highly permissive newborn mice. However, simultaneous miRNA-targeting in the ORF and 3′NCR had synergistic effect on control and silencing of virus replication in the brain and completely abolished the virus neurotropism. Furthermore, neuroinvasiveness of miRNA-targeted TBEV/LGTV viruses in very sensitive immunodeficient SCID mice was significantly limited. Immunocompetent animals immunized with such viruses were completely protected against challenge with the neurovirulent LGTV parent. These findings support the rationale of the miRNA-targeting approach to develop live attenuated virus vaccines against various neurotropic viruses.

1. Introduction

Viruses within the Flavivirus genus, such as Japanese encephalitis (JE), St. Louis encephalitis (SLE), West Nile (WN), and tick-borne encephalitis (TBEV) viruses, are important zoonotic pathogens, typically causing in humans severe neurological diseases, of which up to 40% are fatal (Sips et al., 2012). They represent a serious public health problem in many regions of the world. Due to changes in climate and increased human population and travel, both mosquito- and tick-borne flaviviruses have expanded their geographic range and emerged in areas where they previously did not exist (Artsob et al., 2009; Mackenzie et al., 2004; Roth et al., 2010). The spread of viruses into new areas resulted in a severe impact on wildlife and caused outbreaks of disease in humans. This increase in flavivirusassociated illness emphasizes the need for efficacious and immunogenic vaccines. Although inactivated vaccines are available for JE and TBEV, they require multiple primary immunizations followed by repeated booster vaccinations due to the relatively short duration of immunity. Live-attenuated virus vaccines are known to induce a more multifaceted and long-lasting immune response (Plotkin, 2015), suggesting that the development of a new safe live TBEV vaccines will be important improvement for public health in prevention of TBEV-caused illness in populations living in endemic areas.

The neuropathogenesis of neurotropic flavivirus infection includes two distinct properties: neurovirulence and neuroinvasiveness (Mandl, 2005). In the CNS, the primary targets of encephalitic flaviviruses are neurons. Nevertheless, key aspects of neuronal damage are poorly understood: neuronal injury and death can be caused by direct viral infection and/or initiated by a virus-induced inflammatory immune response (Cho and Diamond, 2012). Thus, successful attenuation of the pathogenesis associated with a number of neurotropic flaviviruses requires the prevention of virus entry into the CNS, as well as restriction of its replication in the neurons.

Flaviviruses are positive-stranded RNA viruses with genome of ∼11 kb that contains a single open reading frame (ORF) flanked by 5′ and 3′ noncoding regions (NCRs). The polyprotein encoded by the ORF is processed by viral and cellular proteases to produce three structural proteins (capsid [C], precursor-membrane [prM] and envelope [E]) and seven nonstructural proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B and NS5) (Lindenbach and Rice, 2003). Thus, the flavivirus positive-sense RNA genome serves as a long messenger RNA and can be exploited as a template for miRNA-mediated silencing of viral protein synthesis and genome replication in order to control virus pathogenesis.

Recently, we and others demonstrated that RNA viruses engineered to contain the specific target sequences for endogenous miRNAs were restricted for replication in tissues expressing these miRNAs (reviewed in: (Barnes et al., 2008; Kelly and Russell, 2009; Lauring et al., 2010; tenOever, 2013). Using chimeric tick-borne encephalitis/dengue virus (TBEV/DEN4) as a model neurotropic flavivirus, we showed that cellular miRNAs abundantly expressed in the brain could be successfully exploited to selectively restrict replication of this virus in the CNS by inserting miRNA targets into the viral genome (Heiss et al., 2011; Heiss et al., 2012; Teterina et al., 2014). The present study focused on investigating whether this approach could effectively attenuate the more pathogenic flavivirus, a tick-borne encephalitis/Langat virus (TBEV/LGTV). Unlike non-neuroinvasive TBEV/DEN4, the TBEV/LGTV virus demonstrates both neurovirulent and neuroinvasive properties characteristic of neurotropic flaviviruses and, as such, is a more fitting model for studies of these viruses' attenuation. We demonstrated that combined targeting of the TBEV/LGTV genome within the ORF and 3′NCR led to significant attenuation of both neurovirulent and neuroinvasive virus properties, as assessed in the most susceptible mouse models.

2. Materials and Methods

2.1. Viruses

All viruses were generated from corresponding infectious cDNA clones. The full-length cDNA clone (pE5-651) of attenuated strain E5 of LGTV (GenBank access #AF253420) and cDNA clone of chimeric TBEV/DEN4 virus (GenBank access # FJ28986) have been described previously (Campbell and Pletnev, 2000; Pletnev, 2001; Pletnev et al., 1992). The attenuated LGTV strain E5 was derived from parental LGTV strain TP-21 by serial passages in chicken embryos (Thind and Price, 1966a, b). The E5 virus retained significant neurovirulence in suckling mice inoculated intracerebrally and had reduced peripheral virulence in monkeys and adult mice (Thind and Price 1966a). To create the chimeric TBEV/LGTV clone, the cDNA gene segment encoding the structural prM and E proteins of the Far Eastern TBEV (FE-TBEV strain Sofjin) from pTBEV/DEN4 (nt 400 -2360) was used to replace the corresponding LGTV sequences in pE5-651 using conventional cloning methods (Sambrook and Russell, 2001). All recombinant viruses carrying miRNA targets were derivatives of a TBEV/LGTV cDNA clone generated using reverse genetics technology. The TBEV/LGTV cDNA contains a unique Xho I site at nucleotide (nt) position 2381, a junction site between the TBEV and LGTV sequences in the E protein gene (Fig. 1B). Insertion of the synthetic sequence TL6 (Blue Heron Biotechnology) (Supplemental Fig. S1) in this site of the TBEV/LGTV genome resulted in TBEV/LGTV-TL6 virus (designated vTL6) carrying three complementary miRNA targets between sequences encoding the TBEV E protein and LGTV non-structural protein NS1 (Fig. 1).

Figure 1. Schematic representation of viral genomes used in this study.

(A) Predicted secondary structure of the LGTV 3′NCR part of TBEV/LGTV genome and the schematic diagram of ten sites that were evaluated for miRNA target insertions. Blue arrows indicate sites where insertions of target sequences were stable; red arrows indicate sites from which target sequences were eliminated during virus growth in Vero cells and black arrow indicates the site that was restricted for virus growth; numbers indicate nucleotide position in viral genome. (B) Schematic illustration of the parental TBEV/LGTV chimeric virus and insertion of target sequences for miRNAs in the area coding for the C-terminal end of protein E and in the 3′NCR. Orange area represents TBEV sequences; green area indicates LGTV virus sequences. Dashed lines mark sequences inserted at the Xho I site of the parental genome; red arrows indicate duplicated signalase cleavage site. Positions of miRNA targets for brain-expressed mir-124 and mir-9 are indicated by red and blue boxes, respectively. Recovered viruses and location of miRNA target sequences in their genomes are listed in the table.

To generate TBEV/LGTV viruses with miRNA targets located in the 3′NCR, ten sites for miRNA target insertions (Fig. 1A) were selected in the 3′NCR between nucleotides: 10375-76 (site 1), 10389-90 (site 2), 10412-13 (site 7), 10493-94 (site 8) 10570-71 (site 9) 10594-95 (site 3), 10619-20 (site 10), 10668-69 (site 11), 10700-1 (site 4), and 10836-37 (site 5). The SspI-KpnI DNA fragment TL1 that encoded the 3′NCR (nt 10366 - 10950) with mir-9 and mir-124a target sequences inserted at sites 1 and 2, correspondingly, was synthesized by Blue Heron Biotechnology (Supplemental Fig. S2) and cloned into the TBEV/LGTV cDNA genome. Recombinant cDNA clones with additional mir-124a target inserted at site 3, 4, 5, 7, 8, 9, 10 or 11 (Fig. 1) were constructed using synthetic cDNA segments (Blue Heron Biotechnology) and cloned into the TL1 cDNA genome. RNA transcripts derived from these modified TBEV/LGTV cDNA clones were generated by transcription with SP6 polymerase and transfected into Vero cells (Engel et al., 2010; Rumyantsev et al., 2006b). Only five viruses (designated vTL1, vTL7, vTL6, vTL8, and vTL10) carrying 2 or 3 copies of targets for mir-124a and mir-9 in the 3′NCR or ORF (Table in Fig. 1B) were recovered from Vero cells. The DNA fragment from vTL6 cDNA that contained miRNA targets in the ORF was used for insertion into the cDNA genomes of vTL1, vTL7, and vTL10. Three newly engineered viruses (designated vTL6+1, vTL6+7, and vTL6+10) carrying miRNA targets in the ORF and 3′NCR were also recovered from Vero cells. All viruses were biologically cloned by two or three terminal dilutions, and amplified by two passages in Vero cells. Full-genome consensus sequencing of biologically cloned viruses was performed as previously described (Engel et al., 2010) to ensure genetic integrity. Viral RNA was extracted from cell culture supernatants and subjected to reverse transcription (RT) and polymerase chain reaction (PCR) using the SuperScript III one-step RT-PCR kit (Life technologies) with LGTV- and TBEV-specific primers to produce 5-8 overlapping fragments (primer sequences available upon request). RT-PCR products were purified using PCR product purification kit (Roche) and sequence analysis of the PCR fragments was carried out as described previously (Engel et al., 2010, Heiss et al., 2012).

2.2 Evaluation of viruses in mice

Studies in mice were performed at the Animal Biosafety Level 3 facilities of the NIAID, in compliance with the guidelines of the NIAID/NIH Institutional Animal Care and Use Committee.

2.2.1 Neurovirulence and virus replication in brain

Suckling (3-day-old) Swiss Webster (SW) mice (Taconic Farms) in litters of 9 - 11 were inoculated intracerebrally (IC) with 1, 10, 10², or 10³ PFU of virus and monitored 21 days for signs of morbidity, including tremor, seizure, and paralysis. The LD₅₀ was determined by the Reed & Muench method (Reed and Muench, 1938). Moribund (paralyzed) mice were euthanized, and their brain homogenates (10% w/v) were prepared and used for viral RNA extraction as described previously (Engel et al., 2010; Pletnev et al., 2006). Consensus sequences of the genomic regions containing the engineered miRNA targets were determined as described above. Virus presence in brains harvested from surviving mice on day 22 p.i. was evaluated by direct titration and after one blind passage in Vero cells.

To evaluate virus replication in the CNS, litters of 9 suckling SW mice were IC inoculated with 10³ PFU of vTL6+7, vTL6+10 or parental TBEV/LGTV. Three mice from each group were euthanized on days 2, 4, and 6 p.i. for TBEV/LGTV virus or on days 6, 13, and 20 p.i. for vTL6+7 or vTL6+10 viruses, and virus loads in brain homogenates were quantitated by titration in Vero cells.

Analysis of viral antigen distribution and assessment of the extent of virus-induced neuronal damage in the brain were performed as previously described (Heiss et al., 2012; Teterina et al., 2014). Three brains of suckling SW mice inoculated IC with 10³ PFU of the virus were collected on days 13 and 20 p.i. for vTL6+7 or vTL6+10-infected group and on day 6 for TBEV/LGTV group when the mice became moribund. For detailed analysis of viral antigen distribution in the brain, we performed immunohistochemistry analysis using standard immunoperoxidase method with primary rabbit anti-TBEV polyclonal antibodies (1:30,000) (Heiss et al., 2012).

2.2.2. Neuroinvasiveness

To investigate the neuroinvasive phenotype of TBEV/LGTV virus and its derivatives, two studies were performed. First, 3-week-old SW mice in groups of 10 were inoculated intraperitoneally (IP) with 10³, 10⁴, or 10⁵ of vTL6+1, vTL6+7, or vTL6+10, respectively and mice were observed for 28 days for signs of morbidity typical of CNS involvement, including paralysis. In a second study, groups of 3-week-old SCID mice (ICRSC-M; Taconic Farms) were inoculated IP with 10³, 10⁴, or 10⁵ PFU of vTL6+1, vTL6+7, or vTL6+10 virus, or with 1, 10, or 100 PFU of parental LGTV or TBEV/LGTV virus and observed for 50 days. To test for viremia, mice were bled on day 6, 10, 14, 18, and 28, and the amount of virus in serum was determined by titration on Vero cells. For sequence analysis, viruses present in the serum harvested on day 18 and day 28 were amplified in Vero cells.

2.2.3. Protective immunity

Immunogenicity and protective efficacy of miRNA-target viruses were assessed in 3-week-old mice. In experiment #1, groups of 5 SW mice were immunized IP with 10³, 10⁴, or 10⁵ PFU of vTL6+7 or vTL6+10; in experiments #2 and #3, groups of 10 SW mice (exp. #2) or C3H mice (exp#3) were immunized IP with 10⁵ PFU of each virus. Sera were collected on day 28 p.i. for measurement of TBEV-specific neutralizing antibody titer using the 50% plaque reduction neutralization assay (PRNT50) in Vero cells as previously described (Pletnev et al., 2001; Rumyantsev et al., 2006b). To demonstrate protection, mice were challenged IP on day 29 with 10⁵ PFU (experiments #1) or 10⁶ PFU (experiments #3) of parental TBEV/LGTV virus or with 10⁶ PFU (experiments #2) of LGTV virus. Challenged mice were observed for 28 days for signs of encephalitis.

2.3. Statistics

Statistical analysis was performed using GraphPad Prism 6 software (GraphPad Software Inc.). Differences between groups were examined using analysis of variance (ANOVA) and Student's unpaired t-test. Differences between two groups were considered statistically significant if the two-tailed p value was <0.05. Mouse survival analyses were performed using the Log-rank (Mantel-Cox) test, with p<0.05 required for significance.

3. Results and Disscussion

3.1 Effect of antigenic chimerization of TBEV with LGTV on viral virulence

A full-length cDNA clone of chimeric TBEV/LGTV was generated by replacing the prM and E structural protein genes of the attenuated E5 strain of tick-borne LGTV (Pletnev, 2001) with the corresponding region from FE-TBEV (Fig. 1B).

The TBEV/LGTV pathogenicity in mice was compared with that of parental LGTV and chimeric TBEV/DEN4 virus (Pletnev et al., 1992) that contains the same TBEV prM and E protein genes in the genetic background of mosquito-borne DEN4 virus. Neurovirulence of these viruses was assessed by estimating the LD₅₀ values after IC inoculation of either adult or suckling SW mice. TBEV/LGTV chimeric virus retained the neurovirulence of its LGTV parent with LD₅₀ values of 3.1 and 0.9 PFU for adult and newborn mice, respectively. These levels of TBEV/LGTV neurovirulence were similar to those observed for the TBEV/DEN4 virus or parental LGTV, but were lower than mouse neurovirulence of all previously studied TBEV strains (Table 1).

Table 1.

Neurovirulence and neuroinvasiveness of TBEV/LGTV virus in mice.

Virus	Neurovirulencea		Neuroinvasivenessb
	IC LD50 (PFU)		IP LD50 (PFU)
	Suckling SW mice	Adult mice	Immunocompetent adult SW mice	Immunodeficient adult SCID mice
TBEV/LGTV	0.9	3.1	60	5
LGTV	0.4	3.2	25	0.1
TBEV/DEN4c	0.8	6	>10,000,000	24,000
TBEVd (Absettarov)	0.06	0.1	1	-
TBEVe (Neudoerfl)	0.1	-	1	-
TBEVf (Hypr)	-	0.15	1

^aThree-day-old suckling or 21-day-old adult Swiss Webster mice were inoculated intracerebrally (IC) with serial 10-fold dilutions of indicated viruses and then monitored for morbidity for 21 days.

^bThree-week-old Swiss Webster or SCID mice were inoculated intraperitoneally (IP) with 10-fold dilutions of indicated viruses and then monitored for signs of neurological disease and morbidity for 4 or 7 weeks, respectively.

^cData from ref. (Pletnev et al., 1992; Rumyantsev et al., 2006a)

^dData from ref. (Pripuzova et al., 2009)

^eData from ref. (Mandl et al., 1998)

^fdata from ref. (Rumyantsev et al., 2013)

The neuroinvasive properties of TBEV/LGTV virus were investigated following IP inoculation of either adult immunocompetent SW mice or more susceptible immunodeficient SCID mice. Unlike TBEV/DEN4, which is greatly restricted in its ability to invade the CNS and cause fatal encephalitis (Rumyantsev et al., 2006a), the TBEV/LGTV chimeric virus was found to be neuroinvasive in both animal models and demonstrated a modest (2 or 50 fold, correspondingly) increase in IP LD₅₀ levels when compared with LGTV, that has an IP LD₅₀ of 25 and 0.1 PFU, respectively, and significantly longer average survival times (AST) (Table 1 and Supplemental Fig. 3). It is interesting to note that while chimerization between TBEV and LGTV had little effect on virus replication in cell culture, it yielded attenuating effect in vivo in mice. Each of the gene products of LGTV or TBEV has been selected during the evolutionary history to act efficiently with other viral and cellular proteins in the complex program of virus replication. The TBEV/LGTV chimera contains the sequence for pre-membrane and envelope proteins of FE-TBEV that have the 15% and 12% divergence in amino acid sequence of prM and E proteins of LGTV; moreover, the amino acid identity among the M proteins of LGTV and FE-TBEV is only 64% (Mandl et al., 1991). It is likely that replacement in the chimera of two LGTV genes for the corresponding genes of TBEV created LGTV-TBEV protein incompatibilities that compromised chimeric virus replication in vivo, altered virus-host interactions, and resulted in the reduction of neuroinvasiveness. Our data are in a good agreement with those observed by Sakai et al., 2014 for chimeric TBEV viruses constructed between TBEV Sofjin (highly pathogenic) and Oshima (low pathogenic) strains of TBEV (Sakai et al., 2014). Replacement of structural protein genes of Oshima genome with corresponding genes derived from Sofjin genome reduced chimeric virus replication in mouse neuroblastoma cells and exhibited neuroinvasiveness similar to that of the parental low-pathogenic Oshima strain”.

3.2 Generation of TBEV/LGTV viruses carrying multiple miRNA target sites

The secondary RNA structure for LGTV 3′NCR was predicted using MFOLD (Fig. 1A) and correlates well with a previously proposed structure (Gritsun et al., 2014). The TBEV 3′NCR consists of two distinct regions: the 5′-end, named the “variable” region that varies in length and sequence among various tick-borne flaviviruses and within their strains, and the 3′-terminal “core” region (∼340 nts at the extreme 3′ end), which contains several elements essential for viral replication (Gritsun et al., 1997; Wallner et al., 1995). Previously, we showed that the LGTV 3′NCR can tolerate a large deletion (nts 10379-10700, starting from the UAA stop codon and extending into the variable region) without a negative effect on virus viability (Pletnev, 2001), confirming the plasticity of this variable region. Two sites (sites 1 and 2 in Fig. 1A and B) located immediately after the stop codon were utilized for insertion of the perfectly complementary target sequences for brain-expressed mir-9 and mir-124a miRNAs to generate a vTL1 genome. Eight additional sites (numbered 3, 4, 5, 7, 8, 9, 10 and 11; Fig. 1A) selected in the predicted accessible, A/U nucleotide-rich loop-structures of the 3′NCR were evaluated for insertions of a second mir-124a target sequence in the vTL1. Seven viruses (designated vTL3, vTL4, vTL5, vTL7, vTL8, vTL9, and vTL10) were recovered after transfection of RNA transcripts of viral cDNA genomes in Vero cells. Each genome carried mir-9T and mir-124T at sites 1 and 2 and a second target for mir-124a at sites 3 through 10 (Fig. 1 A), correspondingly. The insertion of a foreign sequence into site 11 (between nts 10668-9) appeared to be lethal for virus replication, probably due to an effect on the local stem-loop structures of the core region that are conserved among different flaviviruses (Gritsun et al., 2014). Based on sequence analysis data, recombinant miRNA-targeted vTL1, vTL7, vTL8 and vTL10 viruses (Fig.1 B) were genetically stable during recovery and 5 passages of replication in Vero cells and contained the desired miRNA target insertions in the 3′NCR. However, the second target for mir-124a inserted at site 3, 4, 5, or 9 was mutated or deleted from virus progeny recovered after transfection of Vero cells with TL3, TL4, TL5 or TL9 RNA genomes.

TBEV/LGTV virus carrying the miRNA targets in the ORF, in the region between protein E and NS1 genes, was generated as previously described (Bonaldo et al., 2007; Teterina et al., 2014). Three tandem targets (mir-124T-9T-124T) for miRNAs were placed between duplicated sequences encoding the C-terminal stem-anchor (SA) region of protein E (Fig. 1B).

To target virus genomes in two distantly located regions, we engineered genetically stable TBEV/LGTV viruses (designated vTL6+1, vTL6+7, and vTL6+10) in which vTL6 sequence was combined with the 3′NCR derived from the vTL1, vTL7, or vTL10 genome (Fig. 1B). These viruses demonstrated smaller plaque morphology on Vero cells and delay in the attainment of peak titers of 8.1, 7.6 and 7.1 log₁₀ PFU/ml for vTL6+1, vTL6+7 and vTL6+10, respectively (Supplemental Fig. S4).

3.3 Neurovirulence of the miRNA-targeted TBEV/LGTV viruses in suckling mice

The unmodified TBEV/LGTV virus is highly neurovirulent with an estimated IC LD₅₀ level of 0.9 PFU in newborn mice (Table 1). Introduction of the miRNA targets into either the 3′NCR or ORF of the TBEV/LGTV genome had a modest attenuating effect. We observed an 11- to 55-fold increase in the LD₅₀ of viruses with target sequences inserted in the 3′NCR (vTL1, vTL7, vTL8 and vTL10) compared to that of parental virus (Table 2); similarly, neurovirulence of the vTL6 virus, carrying miRNA targets only in the ORF, was reduced by 20-fold. In addition, mice became moribund at later time points after inoculations with miRNA-targeted viruses, and the AST for the mice inoculated with either 1 or 10 PFU of virus was significantly longer as compared to mice infected with the same dose of TBEV/LGTV (Table 2 and Supplemental Fig. 5) (p<0.02 or 0.001 for 1 and 10 PFU, correspondingly). Interestingly, the level of attenuation of TBEV/LGTV virus with miRNA target sequences in the ORF (vTL6) was comparable with attenuation level of TBEV/DEN4 virus, having a similar set of miRNA-target sequences (vE1) (Teterina et al., 2014), whereas the attenuating effect of miRNA-targeting in the 3′NCR of TBEV/LGTV (vTL1, vTL7, vTL8 and vTL10) on viral neurovirulence in suckling mice was significantly lower than that observed for miRNA-targeted TBEV/DEN4-derived viruses (for example, TBEV/DEN4 mir-9T-124aT-124aT(1,2,3) virus has the IC LD₅₀ of 490 PFU that indicates a 613- fold reduction of its neurovirulence compared to that of parental TBEV/DEN4 virus) (Heiss et al., 2012). The reason for such difference is not clear and might be related to the fact that TBEV/LGTV 3′NCR was found to be more restrictive for the placement of target sequences. Sequence analysis of TBEV/LGTV progeny viruses derived from the brains of moribund mice revealed that viral escape from miRNA-mediated suppression occurred due to partial or complete deletions of all introduced miRNA-binding sites in either the ORF or the 3′NCR.

Table 2.

Neurovirulence of chimeric TBEV/LGTV miRNA mutants in 3-day-old Swiss mice inoculated IC.

Virus (miRNA-targets insertion sites)a	Doseb (PFU)	No.of moribund mice on the indicated day (dpi)c	ASTd	Morbidity (%)	LD50 (PFU)
vTL1 (3′NCR: site 1, 2)	1			0
	10			0	49
	102	6(d8), 2(d9)	8.3	72.7
vTL7 (3′NCR: site 1, 2, 7)	1			0
	10	1(d15), 1(d17)	16e	18.2	25
	102	6(d8), 4(d9)	8.4	100
vTL8 (3′NCR: site 1, 2, 8)	1	1(d11)		9.1
	10	3(d11)	11e	30	16g
	102	6(d8), 2(d10), 2(d12)	9.2	100
vTL10 (3′NCR: site 1, 2, 10)	1			0
	10	2(d11), 2(d12), 1(d16)	12.4	50	10g
	102	3(d8), 5(d10), 2(d11)	9.6	100
vTL6 (ORF)	1	1(d14), 1(d15)		18.2
	10	1(d12), 1(d13), 1(d15)	13.3e	72.7	18g
	102	5(d11), 2(d12), 1(d13), 1(d15), 1(d17)	12.4	91
vTL6+1 (ORF and 3′NCR: site 1, 2)	10			0
	102	1(d12), 2(d14)	13.3e	30	776g
	103	1(d11), 2(d12), 1(d16)	12.8	40
vTL6+7 (ORF and 3′NCR: site 1, 2, 7)	10			0
	102			0	>1000
	103			0
vTL6+10 (ORF and 3′NCR: site 1, 2, 10)	10			0
	102			0	>1000
	103			0
TBEV/LGTV	10	6(d6),	6	66.7	0.9f

^aThe place where target sequences for miRNAs are inserted as shown in Fig. 1B.

^bGroups of 9-11 3-day-old SW mice were inoculated IC with the indicated dose of virus.

^cThe brains of paralyzed mice were collected on the indicated day, and viral RNA was isolated from brain homogenate and directly used for sequence analysis. Sequence analysis of 2-3 brain-derived viruses was performed for each group with moribund mice.

^dAverage survival times (AST) of moribund mice

^eSurvival time is statistically different from TBEV/LGTV by Log-rank (Mantel-Cox) test (determined for the IC dose 10 PFU)

^fThe IC LD₅₀ of TBEV/LGT from previous studies (Table 1).

^gThe LD₅₀ is statistically different from TBEV/LGTV by nonlinear regression curve fit analysis (p< 0.05).

The miRNA co-targeting at two distant sites of viral genome (ORF and 3′NCR) led to a more than additive attenuating effect on virus neurovirulence in suckling mice as demonstrated by a delay in onset of encephalitis and a substantially increased survival rate (Table 2). The IC LD₅₀ was 860-fold higher for vTL6+1 and greater than 1110-fold higher for vTL6+7 or vTL6+10 compared with that of the TBEV/LGTV parent.No death or neurological signs were observed in newborn mice inoculated IC with 10, 100 or 1000 PFU of vTL6+7 or vTL6+10. These two viruses (vTL6+7 and vTL6+10) demonstrated the most attenuated neurovirulence and were selected for further assessment in mouse models.

3.4 Multiple site miRNA co-targeting within the ORF and 3′NCR greatly restricts virus replication and prevents neuroinflammation and neurodegeneration in the CNS

To quantitate the levels of virus replication in mouse brain, groups of suckling mice were inoculated IC with 10³ PFU of vTL6+7, vTL6+10, or parental TBEV/LGTV virus, and brains of three mice from each group were harvested on day 2, 4, and 6 for parental TBEV/LGTV or on day 6, 13, and 20 for mutant viruses to determine virus loads. Following parental TBEV/LGTV inoculation, mice showed neurological signs (hunched back and limb weakness) by day 6 p.i. leading to paralysis or death within the next day. TBEV/LGTV replicated in the brain efficiently and rapidly reached an extremely high titer (8.0 log₁₀ PFU/g of brain on day 2 and >10 log₁₀ PFU/g of brain on days 4 and 6 p.i.) (Fig. 2A). In contrast, replication of viruses with multiple miRNA targets was severely restricted. Virus vTL6+7 replicated to 3.4 log₁₀ PFU/g of brain by day 6 p.i. decreasing to below limit of detection on day 13 and 20. Further attenuation was observed for TL6+10 virus, whose replication was detected only at very low levels (2 log₁₀ PFU/g of brain) in two of three mice on day 6 p.i. with no virus detected in mouse brains on days 13 and 20 p.i. (Fig. 2B). In addition, all mice infected with miRNA-targeted viruses remained asymptomatic during the course of the study.

Figure 2. Replication kinetics of recombinant TBEV/LGTV viruses in mouse brain.

Three-day-old SW mice were inoculated IC with 10³ PFU of indicated viruses. Brains of three mice were harvested at the indicated times, mean virus titers (± SD) of brain homogenates are shown for TBEV/LGTV on days 2, 4 and 6 (A), or TBEV/LGTV (black bar), vTL6+7 (red bars) and vTL6+10 (blue bars) on days 6, 13 and 20 post-inoculation (B). The dashed line indicates the limit of virus detection (1.7 log₁₀ PFU/g of brain). Differences in replication kinetics were compared between TBEV/LGTV virus and miRNA targeted viruses using 2-way ANOVA implemented in Prism 6 software (**P<0.001).

We next analyzed the distribution of viral antigens and virus-induced histopathology in the brains of infected mice. Simultaneous immunostaining of multiple mouse brains for the TBEV antigens corroborated the data on virus replication (Fig. 3). Various neuronal populations throughout the brains of TBEV/LGTV-infected mice displayed massive amounts of viral antigens (Fig. 3, A and B). No viral antigens were detected in neurons of mice inoculated with the vTL6+7 or vTL6+10 (Fig. 3, vTL6+7: C and D; vTL6+10 E and F). In addition, no signs of neuroinflammation, as assessed by analysis of microglial activation (Fig. 4) or neuronal damage were detected in mice inoculated with the TL6+7 (data not shown) or TL6+10 (Fig. 4).

Figure 3. Immunohistochemical detection of TBEV antigens in neurons of infected mice.

Representative brain sections from a mouse inoculated IC with the TBEV/LGTV virus (A and B; 6 dpi), TL6+7 virus (C and D; 20 dpi), TL6+10 virus (E and F; 20 dpi), or mock (G and H; 6 dpi. Note: mock at 20 dpi was identical in appearance). Boxed areas of the cortex in (A), (C), (E) and (G) are shown at higher magnification in (B), (D), (F) and (H). Bar in (A) panel also applies to (C), (E) and (G); bar in (B) panel also applies to (D), (F) and (H).

Figure 4. Morphological appearance of microglia in the CNS of infected mice.

Immunostaining for the ionized calcium binding adapter molecule 1 (Iba1) shows differences in microglial morphology in response to the infection with the parental TBEV/LGTV or miRNA-targeted TL6+10 virus. Representative sections of the cortex are shown from a mock-inoculated mouse (A; 20 dpi), and mice infected with the TBEV/LGTV (B; 6 dpi) or TL6+10 virus (C; 20 dpi). Note: a ramified morphology of surveying microglia in the cortex of mock and TL6+10-inoculated mice (bottom: higher magnified insets in A and C) in contrast to the activated microglia (increased Iba1-IR, amoeboid morphology and phagocytic end-bulbs) in the cortex of a TBEV/LGTV-infected mouse (inset in B). Bar (50 μm) in (A) also apply to (B) and (C). Bars in A and insert: 50 μm.

3.5 miRNA-targeted TBEV/LGTV viruses demonstrate highly attenuated neuroinvasive properties in vivo

Since parental TBEV/LGTV virus retained the ability to spread from a peripheral site of inoculation into the CNS of immunocompetent mice, demonstrating an LD₅₀ level of 60 PFU, we initially evaluated miRNA-targeted viruses for neuroinvasion in this animal model. For that, adult SW mice were inoculated IP with 10³, 10⁴ or 10⁵ PFU of TL6+7 or TL6+10 virus followed by daily monitoring for morbidity. Even at a high dose of 10⁵ PFU (1670-fold higher than the IP LD₅₀ of the TBEV/LGTV parent) all viruses with miRNA targets were restricted in the ability to invade the CNS and cause fatal encephalitis, as all mice survived 28 days observation with no signs of neurological symptoms.

We next assayed neuroinvasiveness of these viruses in SCID mice, a more sensitive model for assessment of this property (Charlier et al., 2004; Halevy et al., 1994; Pletnev and Men, 1998). Groups of 3-week-old SCID mice were inoculated IP with 1, 10, or 10² PFU of parental LGTV or unmodified TBEV/LGTV viruses or with 10³, 10⁴, or 10⁵ PFU of miRNA-targeted vTL6+1, vTL6+7 or TL6+10 virus and monitored 7 weeks for morbidity. In addition, to assess virus replication and genetic stability in the periphery, serum samples were collected on days 6, 10, 14, 18 and 28 p.i. from surviving mice that received the highest dose of virus (10² PFU for LGTV and TBEV/LGTV or 10⁵ PFU for vTL6+1, vTL6+7 and vTL6+10). LGTV virus rapidly replicated in the periphery with a mean peak titer of 7.5 log₁₀ (PFU/ml of serum) by day 10 p.i., at which time the majority of SCID mice were paralyzed (Table 3). Its estimated IP LD₅₀ was 0.2 PFU, and the AST of mice that succumbed to LGTV infection was 10.5 days for animals inoculated with 10² PFU. Similarly, TBEV/LGTV replicated to greater than 6 log₁₀ (PFU/ml of serum) by day 10 p.i., and its calculated IP LD₅₀ was 5 PFU; all mice infected with 10² PFU of this virus succumbed to infection within 14 days. Introduction of miRNA targets into the TBEV/LGTV genome resulted in a substantial decrease of viral neuroinvasiveness as evidenced by a significant increase in the AST (longer than 32 days; Table 3) and a reduction in morbidity/mortality rate of mice. vTL6+7 and vTL6+10 had 330- and 13,450-fold higher IP LD₅₀ when compared with TBEV/LGTV and 8,300- and 500,000-fold increases of the IP LD₅₀ when compared with LGTV, respectively (Table 3). Although development of viremia in mice inoculated with either vTL6+7 or vTL6+10 virus was delayed, virus titers increased progressively during the course of infection (Table 3 and Fig. 5) and reached a high level (7.2 and 5.8 log₁₀ PFU/ml of serum by day 18 p.i. for vTL6+7 and vTL6+10, respectively). Despite the high viremia, the majority of mice infected with high doses of vTL6+7 or vTL6+10 exhibited no signs of neurological disease (ruffled hair, slow movement, lethargy, or hunched back) that could progress to a moribund state. The onset of neurological signs such as paralysis of limbs was significantly delayed and observed between days 29 and 35 p.i. in 20% or 80% of SCID mice inoculated with the highest dose (10⁵ PFU) of vTL6+10 or vTL6+10, respectively. The remaining mice did not show any signs of neurological disease during 50 days of observation.

Table 3. Neuroinvasiveness of parental and miRNA-targeted viruses in immunodeficient SCID mice.

Virusa	Dose (PFU)b	ASTb	Viremia on indicated day, log10 (PFU/ml of serum)b					No. moribund/No.tested (%)b	LD50 (PFU)b
			d6	d10	d14	d18	d28
LGTV	102	10.5	7.1	7.5				5/5 (100)	0.2
TBEV/LGTV	102	13.4	4.7	6.2				5/5 (100)	5
vTL6 +1	105	37	3.0	4.4	5.1	6.0	7.2	4/5 (80)	102.9
vTL6+7	105	32.6	4.1	5.2	6.3	7.2	8.1	4/5 (80)c	103.2
vTL6+10	105	32	3.0	4.1	5.0	5.8	5.6	1/5 (20)c	105

^aGroups of 5 3-week-old SCID mice were inoculated IP with 10³, 10⁴, or 10⁵ PFU of vTL6+1, vTL6+7, or vTL6+10 virus or with 1, 10, or 100 PFU of parental LGTV or TBEV/LGTV virus and observed for 50 days to determine their IP LD₅₀.

^bResults for virus titers in the serum (viremia), percent of survivals, and average survival times (AST) of moribund mice are shown for groups of mice inoculated with a highest dose of indicated virus.

^cThe brains of paralyzed SCID mice were collected (one harvested from vTL-6+10-infected group on day 32 p.i. and three from vTL-6+7-infected group on days 32, 33, and 35), and sequence analysis of brain-derived viruses was performed.

Figure 5. Viremia profile of viruses in SCID mice following IP inoculation.

Mice in groups of 5 were inoculated with 10² PFU of LGTV or TBEV/LGTV virus or with 10⁵ PFU of TL6+7 or TL6+10 virus and titer of virus present in the serum (viremia) collected on days 6, 10, 14, 18 and 28 post-inoculation was determined by plaque assay on Vero cells. Each time point represents the mean titer (± SD) from 5 mice in two replicates.

3.6 Genetic stability of miRNA-targeted viruses in SCID mice

Because of the high mutation rate of RNA viruses caused by the low fidelity of the RNA polymerase, the long-term persistence of vTL6+7 and vTL6+10 viruses observed in the blood of SCID mice could lead to accumulation of mutations in the viral genome. We examined whether the delayed onset in the development of viral encephalitis in SCID mice was a result of the appearance of such mutations within the inserted miRNA target sequences. We sequenced the 3′NCR and the region encoding the C-prM-E-NS1 proteins of viruses that were present in individual serum samples collected on days 18 and 28 from vTL6+7- and vTL6+10-inoculated mice. In addition, we sequenced these regions in viruses isolated from brains of mice that succumbed to vTL6+7 or vTL6+10 infection. The results of the sequence analysis are shown in Fig. 6. All viruses derived from mouse sera harvested on day 18 p.i. had the 3′NCR sequence identical to the sequence of vTL6+7 or vTL6+10 input virus, although short deletions (21, 69, or 99 nts in the length) in the region encoding the duplicated SA structural elements (H2 and TM1; Fig. 1B and Fig. 6) of LGTV E protein were detected in viruses isolated from some mice. However, all miRNA target sequences placed in this region remained unchanged.

Figure 6. Location and size of deletion mutations identified in the vTL6+10 and vTL6+7 genomes isolated from serum of asymptomatic SCID mice and from the brain of moribund mice.

Schematic diagram of the insertion of miRNA targets in the area encoding the C-terminal end of protein E and in the 3′NCR. TBEV sequences shown in orange and LGTV sequences shown in green; red and blue boxes represent inserted mir-124 and mir-9 targets, correspondingly. The size and location of the deletions detected in mutant viruses present in serum are shown by double-headed arrows. Blue arrows depict deletions found in sera on day 18, black arrows illustrate deletions found in sera on day 28 and red arrows represent deletions found in brains of mice that succumbed to infection.

On day 28 p.i., we observed emergence of short deletions in the 3′NCR that partially removed miRNA target sequences. vTL6+7-derived viruses isolated from the sera of two animals had a deletion of 26 nts that began at a UAA stop codon of the ORF and extended into the central part of the mir-9 target. Due to this deletion, the UUU codon at the C-terminal end of the NS5 protein was changed to UUA (introducing mutation Phe > Leu), followed by the new UAA stop codon. Similarly, in one vTL6+10 serum-derived isolate, we identified a single 26-nt deletion that eliminated the mir-124 target located in site 10 (Fig. 6). Nevertheless, in all serum-derived viruses, miRNA targets located in the ORF and at least two of three miRNA targets in the 3′NCR remained identical to the sequence of the input virus, indicating target genetic stability during the 28-day persistence of viruses in the blood and peripheral tissues.

Conversely, all viruses derived from brains of mice (one from vTL6+10- and three from vTL6+7-infected group) that succumbed to infection contained significantly larger deletions in both ORF and 3′NCR (Fig. 6). vTL6+10 isolated on day 32 p.i. from the brain of a paralyzed mouse contained three separate deletions that accumulated progressively in the genome: in addition to an extension of the 26 nt deletion (observed in the serum of this mouse on day 28) to 28 nts, the viral genome acquired a second 107-nt deletion including both mir-9T and mir-124T sequences and extending into the 3′NCR, and a 300-nt deletion in the ORF, which included SA structural elements of the LGTV E protein and all miRNA targets inserted in this region of the viral genome. Similarly, three escape mutant viruses isolated from the brains of mice that succumbed to vTL6+7 infection contained large deletions both in the ORF and 3′NCR that eliminated SA structural elements of the TBEV or LGTV E protein and all introduced miRNA target sequences. The discovery of the two or three simultaneous large deletions present in escaped viruses isolated from brain of succumbed mice is intriguing and raises questions regarding the possible mechanism involved. Our data indicate that in the absence of potent B and T cell responses in SCID mice, viruses were able to persist in the periphery for over 4 weeks and might have endeavored multiple attempts to invade and establish replication in the CNS, during which several independent deletions had occurred under miRNA-mediated pressure. Interestingly, recently Mlera et al. (Mlera et al., 2015) reported accumulation of LGTV defective interfering (DI) particles during long-term virus propagation in cell culture that resulted in defective genomes with deleted regions of E and NS1 protein genes. Although the observed deletions in the DI virus genome differed significantly from deletions in the duplicated stem-anchor structural elements of the E protein described here, it is intriguing that deletions were observed in E-NS1 area of viral genome in both studies.

3.7 miRNA targeted TBEV/LGTV viruses are immunogenic and protect mice against challenge

Immunogenicity of viruses carrying miRNA targets was assessed in adult mice (Table 4). In experiment #1, 3-week-old SW mice were immunized with vTL6+7 or vTL6+10, and TBEV-specific neutralizing antibody titers were determined on day 28. Immunization with 10³, 10⁴, or 10⁵ PFU of these viruses resulted in seroconversion ranging from 60% to 100% that was dose dependent. All mice immunized with 10⁵ PFU of vTL6+7 or vTL6+10 virus developed detectable TBEV-specific neutralizing antibodies with a GMT of 426 or 64, respectively (Table 4). Mice were challenged with 10⁵ PFU of parental TBEV/LGTV virus on day 29. All mice immunized with vTL6+7 or vTL6+10 survived challenge; however, it is not possible to evaluate the protective efficacy of these viruses in this experiment, as the mock-inoculated group of mice manifested only 20% mortality, likely due to development of age-dependent resistance to virus infection.

Table 4. Immunogenicity and protective efficacy in mice.

Virus	Dose (PFU)a	Exp. #1 (SW n=5)			Exp.#2 (SW n=10)	Exp.#3 (C3H n=10)
		GMT (range)b	Sero conversion (%)	No. moribund/No.testedc	No. moribund/No.testedc	GMT (range)b	Sero conversion (%)	No. moribund/No.testedc
vTL6+7	103	73 (9-124)	80	0/5
	104	130 (18-353)	80	0/5
	105	426 (143-731)	100	0/5	0/10	174 (25-1035)	100	0/10
vTL6+10	103	50 (9-128)	60	0/5
	104	57 (25-134)	100	0/5
	105	64 (20-103)	100	0/5	0/10	161 (127-653)	100	0/10
Mock		<5	0	1/5	3/10	<5	0	10/10

^aGroups of 3-week-old SW or C3H mice were inoculated IP with the indicated dose of virus and observed for 28 days for morbidity.

^bSerum samples were collected on day 28 for neutralization assay, and plaque reduction (50%) neutralizing antibody titers were determined using TBEV/LGTV as target virus. The reciprocal dilution is reported. Geometric mean titers (GMT) are calculated for each group of mice. Seroconversion defined as a 4-fold or greater increase in serum neutralizing antibody level to TBEV/LGTV on day 28.

^cOn day 29, mice were challenged IP with 10⁵ PFU of TBEV/LGTV (exp. #1) or 10⁶ PFU of LGTV (exp.#2) or TBEV/LGTV (exp.#3), respectively, and observed for additional 28 days for morbidity.

In experiment #2, immunized SW mice were challenged with a higher dose (10⁶ PFU) of neuroinvasive wild-type LGTV virus, strain TP-21. Mock-inoculated mice manifested 30% mortality. All mice immunized with vTL6+7 or vTL6+10 survived challenge; however, the level of protection was not statistically different from the mock group.

In experiment # 3, C3H mice, which are more sensitive to flavivirus infections (Perelygin et al., 2002; Wang and Deubel, 2011), were immunized with 10⁵ PFU of vTL6+7 or vTL6+10, resulting in 100 % seroconversion and GMTs of 174 and 161, respectively. On day 29 post-immunization, mice were challenged with a 10⁶ PFU dose of unmodified TBEV/LGTV virus and observed for an additional 28 days for morbidity. All immunized mice survived this challenge while 100% mice in the mock-inoculated group became paralyzed or died during the 8 days post-inoculation. These data strongly indicate that miRNA-targeted TBEV/LGTV viruses are immunogenic in mice and able to induce protection against neuroinvasive TBEV/LGTV virus.

4. Conclusion

In this study we show that incorporation of multiple target sequences for two brain specific miRNAs into the genomes of highly neuroinvasive TBEV/LGTV virus significantly attenuated its neuropathogenic properties. We demonstrated that miRNA targeting of the viral genome in two distantly located areas successfully prevented virus replication in the CNS. At the same time, viruses retained the ability to replicate in peripheral tissues, induced a strong immune response and provided immune protection in immunocompetent mice. However, studies in immunocompromised mice revealed that in the absence of potent B and T cell responses, prolonged replication of viruses allowed emergence of deletions of the inserted target sequences, leading to neurological disease. Likely, effective suppression of neuroinvasive virus may depend on increased cellular miRNA pressure on the viral genome in both CNS and peripheral tissues. Hence, improved design of genetically stable and safe vaccine candidates against neurotropic viruses will require additional down-regulation of viral replication in the periphery. Once an immunogenic and suitably attenuated vaccine candidate will be identified, the protective efficacy against wild-type TBEV strains will need to be studied further.

• We developed an approach to selectively restrict tick-borne flavivirus pathogenesis even in highly permissive animal models.

• We showed that multiple co-targeting of viral genome for brain-specific miRNAs silenced virus replication in neurons.

• We showed that simultaneous miRNA co-targeting had strong attenuating effects on virus neuroinvasiveness.

• This approach represents an alternative to traditional strategies for the control of viral tissue tropism.

• We suggest that miRNA-targeting is a rationale approach to reinforce the safety of novel or existing vaccines.