Abstract

   Mammarenaviruses are a diverse genus of emerging viruses that include
   several causative agents of severe viral hemorrhagic fevers with high
   mortality in humans. Although these viruses share many similarities,
   important differences with regard to pathogenicity, type of immune
   response, and molecular mechanisms during virus infection are different
   between and within New World and Old World viral infections. Viruses
   rely exclusively on the host cellular machinery to translate their
   genome, and therefore to replicate and propagate. miRNAs are the
   crucial factor in diverse biological processes such as antiviral
   defense, oncogenesis, and cell development. The viral infection can
   exert a profound impact on the cellular miRNA expression profile, and
   numerous RNA viruses have been reported to interact directly with
   cellular miRNAs and/or to use these miRNAs to augment their replication
   potential. Our present study indicates that mammarenavirus infection
   induces metabolic reprogramming of host cells, probably manipulating
   cellular microRNAs. A number of metabolic pathways, including valine,
   leucine, and isoleucine biosynthesis, d-Glutamine and d-glutamate
   metabolism, thiamine metabolism, and pools of several amino acids were
   impacted by the predicted miRNAs that would no longer regulate these
   pathways. A deeper understanding of mechanisms by which
   mammarenaviruses handle these signaling pathways is critical for
   understanding the virus/host interactions and potential diagnostic and
   therapeutic targets, through the inhibition of specific pathologic
   metabolic pathways.

   Keywords: microRNAs, mammarenaviruses, cellular metabolism, amino acid
   metabolism, metabolism of cofactors and vitamins

1. Introduction

   Mammarenaviruses (Bunyavirales: Arenaviridae: Mammarenavirus) are
   enveloped bi-segmented ambisense RNA viruses and each segment encodes
   for two oppositely oriented non-overlapping reading frames separated by
   an intergenic region, a small (S) segment that encodes the envelope
   glycoprotein precursor (GPC) and the nucleoprotein (NP) and a large (L)
   segment coding for the matrix protein (Z) and the RNA-dependent RNA
   polymerase (RdRp) [[30]1,[31]2]. Based on their genomic features and
   antigenic properties, mammarenaviruses are historically classified into
   two large monophyletic clades—the Old World (OW) and New World (NW)
   groups, both containing important zoonotic pathogens affecting humans.
   NW mammarenavirus, or Tacaribe serocomplex, includes viruses indigenous
   to the Americas and can be further divided into Clades A, B, C, and D
   while the OW or Lassa-lymphocytic choriomeningitis virus (LCMV)
   serocomplex includes viruses from Africa and, recently, some viruses
   isolated in Asia [[32]1,[33]3,[34]4].

   Some of these emerging, zoonotic viruses are pathogens of major
   clinical importance to humans, as (i) Lassa virus (OW) is endemic in
   West Africa and is estimated to cause up to 300,000 infections each
   year, (ii) Junín (Clade B NW) that periodically cause hemorrhagic fever
   (HF) outbreaks in Argentina, and (iii) LCMV (OW) which is distributed
   worldwide and is a neglected human pathogen of clinical significance,
   to immunocompromised individuals and pregnant women, due to
   teratogenicity [[35]5,[36]6,[37]7]. Although these two groups share
   many similarities, important differences with regard to pathogenicity,
   type of immune response, and molecular mechanisms during virus
   infection recovery and clearance of the virus is different between and
   within NW and OW infections. For example, whilst both the OW Lassa
   virus (LASV) and Junín virus (JUNV) can cause disruption of the
   vascular endothelium, which is an important pathological feature, the
   immune responses for LASV infection results in an overall generalized
   immune suppression, and patients infected with JUNV seem to develop a
   cytokine storm [[38]8,[39]9]. Understanding the potential differences
   on a cellular level could aid the development of new vaccines and
   treatment strategies against these deadly viral infections.

   MicroRNAs (miRNAs) are the smallest endogenous regulatory non-coding
   RNAs that play a central role in cell differentiation, proliferation,
   and survival by binding to complementary target mRNAs leading to
   translational inhibition or degradation. miRNAs are the crucial factor
   in diverse biological processes such as antiviral defense, oncogenesis,
   and cell development [[40]10]. There are complex interaction networks
   between long non-coding RNAs (lncRNAs), viral genome, and miRNAs. While
   certain miRNAs can regulated the stability and half-life of lncRNAs and
   viral genome, lncRNAs and viral genome can also compete for miRNAs
   acting as miRNA sponges and epigenetic modifications [[41]11,[42]12].
   The miRNA-binding sites within viral genomes are mostly located in the
   5′ and 3′ non-translated regions (NTRs) but have recently been found in
   the coding regions of viral proteins [[43]12,[44]13,[45]14]. Viral
   genomes and gene transcripts affect with host gene expression
   exploiting passive mechanisms to deregulate host miRNA activity.
   Various RNA viruses mimic or block the binding between a host miRNA and
   its target transcript, a phenomenon mediated by the miRNA seed site at
   the 5′ end of miRNA [[46]12,[47]15]. Based on bioinformatics, we
   predicted the potential miRNAs, their target genes, and related
   signaling pathways. To the best of our knowledge, this is the first
   study to systemically analyze and predict the potential miRNAs and
   target genes for Mammarenavirus. Our research could help to further
   assess the roles of miRNAs in virus–host interactions, highlighting
   potential targets during infection, leading to a better understanding
   of the metabolic alterations required for the replication of each virus
   that may lead to novel therapeutic approaches through targeted
   inhibition of specific cellular metabolic pathways.

2. Materials and Methods

   The genomic sequences used in the study were all retrieved from the
   GenBank^® database of NCBI ([48]http://www.ncbi.nlm.nih.gov/nuccore/),
   including all species of the genus Mammarenavirus officially recognized
   by the International Committee on Taxonomy of Viruses (ICTV)
   ([49]https://talk.ictvonline.org/taxonomy/). Herein, for a more clear
   presentation of our results, we considered Clade D New World
   mammarenaviruses (formally known as Clade A recombinant) as members of
   Clade A, as well as Xapuri virus that is associated with Clades B, and
   C was didactically classified here as belonging to Clade B NW
   mammarenaviruses.

   Currently, the human genome contains 2654 mature sequences of microRNAs
   identified in the miRBase database ([50]http://www.mirbase.org). We
   used BLAST with the Geneious R11.1 software
   ([51]https://www.geneious.com) to search for miRNAs that interact with
   the mammarenavirus genome with a perfect alignment of 11 nucleotides
   encompassing the 8 mer seed region, important specific gene silencing
   motifs, and regions from the viral RNA [[52]16]. We also sought
   interactions of miRNAs with transcribed mRNA for the viral proteins and
   antigenomic viral RNA that is produced during the replication in the
   Mammarenavirus genus.

   For microRNA bioinformatic target prediction, we used miRabel, a tool
   that shows significantly better predictions than other important
   algorithms such as MBSTAR, miRWalk, ExprTarget, and miRMap [[53]17].
   This tool adds all human results of four important prediction
   algorithms, miRanda [[54]18], PITA [[55]19], SVMicrO [[56]20], and
   TargetScan [[57]21]. Each of them uses different and complementary
   features with interactions such as seed match, free energy, site
   accessibility, and target-site abundance. The miRabel contains data for
   2587 human miRNAs which have target mRNAs, 19,799 genes and 275
   pathways, representing more than 14.7 million predicted interactions
   from which 351,298 are experimentally established. These experimentally
   validated interactions were annotated using miRTarBase [[58]22] and
   miRecords [[59]23], whereas 5′UTR and CDS predictions were identified
   with the miRWalk database [[60]24]. miRabel uses genes and pathways
   information as well as their relationships recovered from the KEGG
   database [[61]25] while miRNA data were from miRbase (release 22.1)
   [[62]26] and linked with miRNA target predictions.

   miRabel produce pathways linked to the resulting interactions that can
   be retrieved and ranked according to the proportion of its interactions
   regulated by a given microRNA. Moreover, for each pathway, the number
   of validated interactions for this miRNA is also indicated [[63]17]. To
   result in a robust prediction, we consider that the pathways that had
   25% of proportion of its interactions are regulated by a given
   microRNA.

   The data of mammarenaviruses–miRNA interaction and miRabel results were
   integrated using the python library pandas [[64]27]. The resulting
   dataset was used to generate heatmaps to visualize the data with the
   python library seaborn. We used KEGG pathway classification and its
   hierarchy relation plotted against all mammarenaviruses. The scale
   demonstrates how many miRNA–pathway relationships were found for each
   virus species. If a miRNA targeted multiple pathways that were under
   the same higher order classification and if multiple miRNA targeted the
   same pathway, each time would count as a miRNA–pathway relationship in
   order to represent the possible relevance of miRNAs in that pathway. We
   started showing higher order pathway classification and subdivided it
   for better visualization. All procedures were performed using python
   programming language version 3.6.

3. Results

3.1. MicroRNAs

   We retrieved a total of 39 genome sequences from GenBank^®: 20 NW
   species, divided into Clade A (8), Clade B (10), and Clade C (2)
   viruses and 19 OW species. Among the 2654 mature miRNAs identified in
   the miRBase database, we found that 566 miRNAs can bind to certain
   regions of the mammarenavirus genome in a total of 755 binding-sites
   ([65]Supplementary Table S1). We found 316 miRNAs (386 binding-sites)
   for NW mammarenavirus: 131 of Clade A species (147), 169 Clade B (194),
   and 45 for Clade C (45). Some of these miRNAs bind to more than one
   clade, totaling 345. Of these, 231 miRNAs (277) strongly bind in the L
   segment and 97 to the S segment (109). In total, 313 miRNAs (369) were
   able to bind in the genome of OW mammarenaviruses: 213 for the L
   segment (244) and 109 to the S segment (125) ([66]Table 1). The average
   amount of miRNAs binding-sites by specie was nineteen, with Loie River
   virus (OW) presenting the lowest amount of predicted miRNAs (8) and
   Cupixi virus (Clade B NW) the highest (30).

Table 1.

   Total miRNAs found with the potential to bind to mammarenavirus genome.
   New World Mammarenaviruses
   Clade A Viruses Total miRNAs (Binding-Sites) Clade B Viruses Total
   miRNAs (Binding-Sites)
   L Segment S Segment L Segment S Segment
   Allpahuayo virus 12 04 Amapari virus 14 03
   Bear Canyon virus 10 05 Aporé virus 15(16) 06
   Flexal virus 11 03 Chapare virus 15 04
   Paraná virus 09 07 Cupixi virus 23(25) 05
   Pichindé virus 14 04 Guanarito virus 12 07
   Pirital virus 13 09 Junín virus 13 05
   Tamiami virus 12(13) 03(04) Machupo virus 17 08
   Whitewater Arroyo virus 22(23) 06 Sabiá virus 09 06(07)
   Total binding-sites 105 42 Tacaribe virus 14 02
   Clade C viruses Total microRNAs Xapuri virus * 07 05
   L segment S segment Total binding-sites 140 54
   Latino virus 16 03 * Didactically classified here as belonging to Clade
   B NW mammarenaviruses.
   Oliveros virus 16 10
   Total 32 13
   Old World Mammarenaviruses
   Viruses Total miRNAs (Binding-Sites) Viruses Total miRNAs
   (Binding-Sites)
   L Segment S Segment L Segment S Segment
   Alxa virus 08 12 Mariental virus 12 03
   Gairo virus 11 12 Merino Walk virus 14 04
   Ippy virus 17 05 Mobala virus 12 08
   Lassa virus 16 08 Mopeia virus 14 04
   Lijiang virus 12(13) 07 Okahandja virus 09 09(10)
   Loie River virus 07 01 Ryukyu virus 19(21) 03
   Lujo virus 20 06 Solwezi virus 010 04
   Luna virus 13 08 Souris virus 05 09
   Lunk virus 18(22) 10 Wenzhou virus 09 06
   Lymphocytic choriomeningitis virus 11 05 Total binding-sites 244 125
   [67]Open in a new tab

   Among the 521 binding-sites found within the L segment, 208 were found
   in the RNA-dependent RNA polymerase (RdRp) region for OW
   mammarenaviruses and 259 for NW (129 for Clade B, 100 Clade A, and 30
   Clade C). We found 27 binding-sites in the Z protein (Z), being 17 for
   OW and 10 for NW (4 Clade A, 4 Clade B, and 2 Clade C). We identified
   19 binding-sites for the noncoding intergenic region of the L segment
   (L-IGR), with 13 for OW mammarenaviruses, 5 for Clade B, and 1 for
   Clade A. We have not identified any interaction to L-IGR of the Clade
   C. Regarding 5′UTR and 3′UTR untranslated region, we found four
   binding-sites to each region. In total, 233 binding-sites were found
   within the S segment, being 114 for glycoprotein precursor (GPC), 109
   for nucleoprotein (NP), 9 for noncoding intergenic region of the S
   segment (S-IGR), and 1 bind in the 3′UTR untranslated region. Further
   information can be found in [68]Supplementary Table S1, where the
   region with which each miRNA interacts is informed.

   Fifty-two miRNAs were found binding to more than one virus in the same
   genomic segment. For example, the hsa-miR-2052 can bind to the S
   segment of Allpahuayo virus, Bear Canoyn virus, Pichinde virus, and
   Pirital virus, all belonging to Clade A NW ([69]Table 2). Another
   interesting example is the hsa-miR-3120-3p binding to L segment of Lujo
   virus, Lunk virus, Merino Walk virus, and Mopeia virus, all from OW.
   The complete table with all miRNAs is available in [70]Supplementary
   Table S2.

Table 2.

   List of top 10 miRNAs binding to different regions of same the RNA
   segment of different mammarenavirus species.
   miRNAs Mammarenaviruses (Genomic Position)
   Clade A New World Viruses (L Segment)
   hsa-miR-122b-3p Flexal (2745–2755: RdRp)/Paraná (2814–2824: RdRp)
   Clade A New World Viruses (S Segment)
   hsa-miR-2052 Pirital (954–944: GPC)/Bear Canyon (2422–2412:
   NP)/Pichindé (948–938: GPC)/Allpahuayo (945–955: GPC)
   Clade B New World Viruses (L Segment)
   hsa-miR-122b-3p Aporé (2977–2967: RdRp/7195–7185: 3′UTR)/Machupo
   (2920–2910: RdRp)
   hsa-miR-147b-5p Aporé (6826–6816: RdRp)/Machupo (1424–1434:
   RdRp)/Tacaribe (6771–6761: RdRp)
   hsa-miR-3149 Machupo (3195–3185: RdRp)/Tacaribe (3173–3163:
   RdRp)/Xapuri (4629–4639: RdRp)
   hsa-miR-12122 Amapari (3524–3534: RdRp)/Chapare (3552–3562:
   RdRp)/Tacaribe (3553–3563: RdRp)
   Old World Viruses (L Segment)
   hsa-miR-3120-3p Lujo (3497–3487: RdRp)/Lunk (3599–3589: RdRp)/Merino
   Walk (5697–5687: RdRp)/Mopeia (3591–3581: RdRp)
   hsa-miR-3134 Alxa (5200–5190: RdRp)/Wenzhou (2378–2388:
   RdRp)/Lymphocytic choriomeningitis (1975–1985: RdRp)
   Old World Viruses (S Segment)
   hsa-miR-4327 Lunk (2716–2706: NP)/Merino Walk (2721–2711: NP)/Mariental
   (2721–2711: NP)/Souris (2777–2767: NP)
   hsa-miR-6516-3p Lujo (1548–1538: NP)/Lijiang (1655–1645: NP)/Mariental
   (1673–1663: NP)
   hsa-miR-6740-5p Alxa (2834–2824: NP)/Gairo (2788–2778: NP)/Lassa
   (2784–2774: NP)/Lymphocytic choriomeningitis (2787–2777: S-IGR)
   [71]Open in a new tab

   RdRp: RNA-dependent RNA polymerase; GPC: glycoprotein precursor; NP:
   nucleoprotein; S-IGR: noncoding intergenic region of S segment; 3′UTR:
   untranslated region.

   Thirty-four miRNAs can bind to different regions of L and S segments.
   For example, hsa-miR-122b-3p can bind to two different positions of
   Aporé virus L segment, Flexal virus, Machupo virus, Paraná virus, and
   also to Alxa virus S segment. Another interesting miRNA was miRNA
   hsa-miR-9-5p that can bind in Pichindé virus L segment, Ryukyu virus,
   Mobala virus, and to Tacaribe virus S segment ([72]Table 3). The
   complete table with all miRNAs is available in [73]Supplementary Table
   S3.

Table 3.

   List of top 10 miRNAs binding to small (S) and large (L) segments of
   different mammarenaviruses.
   miRNAs Species (Genomic Position)
   L Segment S Segment
   hsa-miR-122b-3p Aporé (2977–2967: RdRp/7195–7185: 3′UTR) Alxa
   (1328–1338: GPC)
   Flexal (2745–2755: RdRp)
   Machupo (2920–2910: RdRp)
   Paraná (2814–2824: RdRp)
   hsa-miR-9-5p Pichindé (4923–4933: RdRp) Tacaribe (2639–2649: NP)
   Ryukyu (4996–5006: RdRp)
   Mobala (5050–5060: RdRp)
   hsa-miR-3611 Aporé (3887–3897: RdRp) Xapuri (1849–1859: NP)
   Flexal (3774–3784: RdRp)
   hsa-miR-3617-5p Pichindé (7003–6993: RdRp) Mopeia (2392–2402: NP)
   Paraná (1198–1208: GPC))
   hsa-miR-1229-3p Cupixi (4426–4436: RdRp) Lassa (2783–2793: NP)
   Oliveros (6625–6635: RdRp)
   hsa-miR-3085-5p Amapari (3334–3324: RdRp) Lujo (1316–1306: GPC)
   Lunk (1709–1699: RdRp)
   hsa-miR-4256 Tamiami (1844–1834: RdRp) Chapare (856–866: GPC)
   Xapuri (2518–2528: RdRp)
   hsa-miR-3913-5p Chapare (2959–2969: RdRp) Luna (2453–2463: NP)
   Lujo (2961–2971: RdRp)
   hsa-miR-4735-3p Lassa (2553–2563: RdRp) Whitewater Arroyo
   (1734–1744: NP)
   Merino Walk (2369–2359: RdRp)
   hsa-miR-4762-5p Chapare (1305–1315: RdRp) Paraná (2680–2870: NP)
   Latino (3939–3949: RdRp)
   [74]Open in a new tab

   RdRp: RNA-dependent RNA polymerase; GPC: glycoprotein precursor; NP:
   nucleoprotein; 3′UTR: untranslated region.

   Ten miRNAs were predicted to bind different regions of the same virus,
   especially the miRNA has-miR-8485 binding in three different positions
   of Ryukyu virus L segment and five positions of Lunk virus L segment,
   both belonging to OW mammarenaviruses group ([75]Table 4).

Table 4.

   MicroRNAs biding to different regions of the same mammarenavirus
   genomic segment.
   microRNAs Mammarenaviruses Genomic Position
   Clade A (L Segment)
   hsa-miR-6083 Whitewater Arroyo 1597–1607: RdRp/5470–5480: RdRp
   hsa-miR-7856-5p Tamiami 3946–3956: RdRp/4552–4562: RdRp
   Clade B (L Segment)
   hsa-miR-122b-3p Aporé 2977–2967: RdRp/7195–7185: 3′UTR
   hsa-miR-376a-3p Cupixi 1543–1553: RdRp/3120–3130: RdRp
   hsa-miR-376b-3p Cupixi 1543–1553: RdRp/3120–3130: RdRp
   Clade B (S Segment)
   hsa-miR-5700 Sabiá 1039–1029: GPC/1030–1040: GPC
   Old World (L Segment)
   hsa-miR-4460 Lijiang 209–199: Z/4747–4737: RdRp
   hsa-miR-8485 Ryukyu 354–364/356–366/388–398 (L-IGR)
   hsa-miR-8485 Lunk 418–428/420–430/422–432/424–434/426–436 (L-IGR)
   [76]Open in a new tab

   RdRp: RNA-dependent RNA polymerase; GPC: glycoprotein precursor; Z: Z
   protein; L-IGR: noncoding intergenic region of L segment; 3′UTR:
   untranslated region.

3.2. miRNAs and Target Metabolic/Cellular Pathways

   To further understand the biological implications of miRNAs binding to
   the mammarenavirus genome, miRabel was used to predict genes targeted
   by these miRNAs (data not shown). After genes prediction analysis, we
   performed pathway enrichment analysis of miRNAs target genes based on
   the KEGG database using miRabel and generated a report of the pathway
   mapping. Our analysis found a total of 75 pathways ([77]Supplementary
   Figure S1).

   Overall, d-Glutamine and d-glutamate metabolism, thiamine metabolism,
   and valine, leucine, and isoleucine biosynthesis were the main
   signaling pathways predicted to be regulated by miRNAs ([78]Figure 1).
   A total of 23 pathways were predicted to Clade A where valine, leucine,
   and isoleucine biosynthesis, d-Glutamine and d-glutamate metabolism,
   and biotin metabolism were the main signaling pathways; 55 pathways for
   Clade B, the top three being thiamine metabolism, biotin metabolism,
   and valine, leucine, and isoleucine biosynthesis, both in second place
   and d-Glutamine and d-glutamate metabolism in third place; and 29 for
   Clade C viruses, where degradation of aromatic compounds and thiamine
   metabolism were the most common followed by d-Glutamine and d-glutamate
   metabolism, valine, leucine, and isoleucine biosynthesis, and biotin
   metabolism, all in second. For miRNAs binding to Old World
   mammarenavirus, we found a total of 62 predicted pathways, d-Glutamine
   and d-glutamate metabolism, degradation of aromatic compounds, and
   thiamine metabolism were main ones ([79]Supplementary Figure S2).

Figure 1.

   [80]Figure 1
   [81]Open in a new tab

   Top 20 main signaling pathways predicted to be regulated by microRNAs
   targeting mammarenavirus genome. The representative proportion of each
   pathway individually for the four clades (Clade A, B, C, and Old World)
   is shown. The graphic scale is represented by the number of microRNAs
   that were found interacting with each pathway. The larger the scale,
   the greater the number of microRNAs found that are regulating a
   specific pathway.

   The miRabel tool through the KEGG pathway analysis demonstrated that
   the targets of miRNAs were more associated with amino acid metabolism
   and metabolism of cofactors and vitamins. d-Glutamine and d-glutamate
   metabolism pathway was regulated by miRNAs for a total of 49 times
   (miRNAs encountered in two or more clades were considered multiple
   times), 8 miRNAs that were found for Clade A, 10 for Clade B, 4 for
   clade C, and 27 for OW mammarenaviruses, including miRNAs that were
   found in more than one group of viruses. The thiamine metabolism was
   targeted 43 times, 3 miRNAs for Clade A, 13 for clade B, 6 for Clade C,
   and 21 for OW viruses. The results showed a total of 40 miRNAs
   interactions with the pathway with the valine, leucine, and isoleucine
   biosynthesis, 12 for Clade A, 12 for Clade B, 4 for Clade C, and 12 for
   OW viruses.

   When considering pathways that have at least one experimentally
   validated interaction for miRNAs, we found a total of 69 pathways out
   of the 75 initials ([82]Supplementary Figure S3). Dorso-ventral axis
   formation, degradation of aromatic compounds, and circadian rhythm were
   the main validated signaling pathways predicted to be regulated by
   miRNAs ([83]Figure 2). According to each mammarenavirus group, 20
   pathways were found in Clade A viruses and valine, leucine, and
   isoleucine biosynthesis, dorso-ventral axis formation, and d-Glutamine
   and D-glutamate metabolism were the three main predicted pathways; for
   Clade B a total of 49 pathways were found with dorso-ventral axis
   formation first of all, and d-Glutamine and d-glutamate metabolism,
   Lipoic acid metabolism, valine, leucine, and isoleucine biosynthesis,
   and Thyroid cancer, equally represented, stand out in proportion.
   Twenty-two were found for Clade C with degradation of aromatic
   compounds and thiamine metabolism, biotin metabolism, adherens
   junction, long-term depression, and thyroid cancer, highlighted in
   miRNA proportion by pathway. For those miRNAs that were predicted in OW
   mammarenaviruses, we found a total of 59 pathways, especially
   dorso-ventral axis formation, degradation of aromatic compounds, and
   circadian rhythm.

Figure 2.

   [84]Figure 2
   [85]Open in a new tab

   Top 20 main signaling pathways predicted to be regulated by microRNAs
   biding to mammarenavirus genome that had at least one target validated
   experimentally in proportion for each of the four clades (Clade A, B,
   C, and Old World viruses). The graphic scale is represented by the
   number of microRNAs that were found interacting with each pathway. The
   larger the scale, the greater the number of microRNAs found that are
   regulating the pathway.

   Among all predicted pathways, only seven showed no experimentally
   validated interactions for the set of miRNAs: caffeine metabolism;
   phenylalanine, tyrosine, and tryptophan biosynthesis; primary bile acid
   biosynthesis; proximal tubule bicarbonate reclamation; shigellosis; and
   synthesis and degradation of ketone bodies.

3.3. Mammarenavirus Species, miRNAs, and Target Metabolic/Cellular Pathways

   We used KEGG pathway classification in order to predict biological
   relevance of each miRNA found for the different mammarenavirus included
   in this study. Target genes of miRNAs found were associated with six
   branches of KEGG pathways: cellular processes, environmental
   information processing, genetic information processing, human disease,
   metabolism and, organismal systems. Notably, target genes of miRNAs
   found are more frequently related to pathways that belong to the
   metabolism branch, followed by human diseases and organismal systems
   ([86]Figure 3). Strikingly, the set of miRNAs found for Junín virus,
   Machupo virus, Lassa virus, and Lymphocytic choriomeningitis virus
   (LCMV) act strongly in these three branches of KEGG pathways.

Figure 3.

   [87]Figure 3
   [88]Open in a new tab

   Heatmap of predicted miRNA–pathway group interactions for each
   mammarenavirus. Scale demonstrates how many times a pathway that
   belonged to a KEGG pathway group was targeted by a miRNA (if a miRNA
   targeted more than one pathway in the same group it was counted as many
   times).

   Within the KEGG metabolism classification, the predicted areas to be
   more affected by mammarenavirus infection are metabolism of cofactor
   and vitamins, metabolism of other amino acids, and amino acid
   metabolism. Pathways related to cancer studies are highlighted within
   human diseases group. For organismal systems, interactions occurs more
   frequently in the development and regeneration pathway. We also
   observed that the miRNAs found for the Junín, Machupo, Oliveros, Lassa,
   and Lymphocytic choriomeningitis viral genome were the ones with the
   greatest impact on different pathways ([89]Figure 4). For example,
   miRNAs that target certain regions of the Lassa genome act in six of
   the eight metabolism pathways, also acting in four of the five
   organismal systems pathways, and in three of the five of human diseases
   pathways.

Figure 4.

   [90]Figure 4
   [91]Open in a new tab

   Heatmap of predicted miRNA–pathway subgroup interactions for each
   mammarenavirus. Scale demonstrates how many times a pathway that
   belonged to a KEGG pathway subgroup was targeted by a miRNA (if a miRNA
   targeted more than one pathway in the same subgroup it was counted as
   many times).

   Individually evaluating each virus, we found different sets of mapped
   pathways. For example, for Pirital virus, miRNAs are mainly associated
   with the valine, leucine, and isoleucine biosynthesis. Olivero virus
   degradation of aromatic compounds followed by biotin metabolism were
   the main predicted pathways. MicroRNAs found for Lujo virus were mainly
   related to thiamine metabolism ([92]Supplementary Figure S4).

4. Discussion

   MicroRNAs are noncoding RNAs which downregulate a large number of
   target mRNAs and modulate cell activity [[93]10]. MicroRNAs functions
   under normal physiological conditions might be integrated into
   multilayered control circuits ensuring proper development and cellular
   homeostasis. However, the dysregulation of miRNA expression or function
   in response to intrinsic factors (genetic or epigenetic) or extrinsic
   factors (environmental cues or stress, such as viral infection) may
   contribute to aberrant gene expression patterns underlying abnormal
   developmental patterning or metabolic dysfunction [[94]28,[95]29].
   Viral genomes could have evolved to directly interact with host miRNAs
   to facilitate certain steps of their replication and progression. The
   cellular miRNA composition in infected cells is likely to indirectly
   affect viruses, because many pathways that promote or limit viral
   replication or the survival of infected cells are likely to be
   regulated by cellular miRNAs. In either case, viruses may therefore
   gain an advantage by reshaping the cellular miRNAs availability
   [[96]12,[97]15,[98]30,[99]31].

   MiRNAs are characterized by variable expression in cells and tissues,
   which is influenced by the molecular cell environment. Some different
   miRNAs and miRNA families are predominantly expressed in certain
   tissues. However, the majority of miRNAs (>80%) is not specific for
   single tissues [[100]32]. A great example is miR-122 that is highly
   abundant in liver, with over 60,000 copies in hepatocyte cells
   [[101]33]. The impact of miR-122 binding on the hepatitis C virus (HCV)
   genome is critical for viral replication. Upon binding of miR-122, HCV
   genome translation and initiation of replication are increased by
   stabilizing and protecting the uncapped HCV RNA genome from degradation
   [[102]34,[103]35]. MiR-122 in vivo knockout studies have revealed that
   reduced miR-122 expression in hepatocellular carcinoma correlates with
   metastasis and poor prognosis [[104]36,[105]37]. These discoveries
   provide insight into the importance of the roles of miRNAs in
   maintaining normal cellular function and how disruptions in miRNA
   expression profiles may heavily impact the development,
   differentiation, and control of growth leading to diseases.

   In this article, we predicted several cellular miRNAs potentially
   regulated by mammarenavirus via direct binding to viral RNAs. Aside
   from virus specific binding sites, we also predicted cellular pathways
   that could be dysregulated during viral infection. A large intersection
   of predicted miRNA sets was found on miRNAs targeting cell metabolism
   for all mammarenavirus groups, including those considered as human
   pathogens. As known, viruses are entirely dependent on host metabolism
   to support its replication, and virally infected cells seem to require
   complex metabolic alterations in order to deal with the high anabolic
   demands essential during viral replication [[106]38,[107]39,[108]40].

   However, there are highly exclusive patterns of virus-induced
   remodeling of host cell metabolic machineries, and the mode of cell
   manipulation appears to be different between RNA and DNA viruses
   [[109]40,[110]41]. Metabolic studies were successfully applied to a
   diversity of virus infections in mammals, insects, and plants, besides
   chronic and acutely infected cell cultures. In all cases, important
   changes in host cellular metabolism were observed, allowing the
   identification of specific metabolites and pathways involved in viral
   infection, and potential cross-talk with the immune system and virus
   pathogenesis [[111]42,[112]43,[113]44,[114]45,[115]46,[116]47,[117]48].
   We identified that d-Glutamine and d-glutamate metabolism, thiamine
   metabolism, and valine, leucine, and isoleucine biosynthesis seems to
   be the most affected pathway during mammarenavirus infection. As
   previous studies have shown for other viral families, mammarenaviruses
   also could control host-cell metabolism via pos-transcriptional
   regulations to cope with the pace of the corresponding replication
   cycles [[118]38,[119]49].

   Glutamate metabolism plays a vital role in biosynthesis of nucleic
   acids and proteins [[120]50]. Glutamine is the most abundant and
   versatile amino acid in the body and is of fundamental importance to
   intermediary metabolism, interorgan nitrogen exchange via ammonia (NH3)
   transport between tissues, and pH homeostasis [[121]51]. It is believed
   that the viral infection induces glutamine uptake and that
   glutaminolysis is required to generate cellular energy during
   infection, allowing the survival of infected cells [[122]52,[123]53].
   In fact, LCMV infection of mice was found to change significantly the
   concentration of α-ketoglutarate, in a concerted fashion, following a
   response and recovery pattern [[124]54]. The α-ketoglutarate is a
   product of a set of metabolic reactions that degrade glutamine
   (glutaminolysis), allowing it to be used anaplerotically to form the
   intermediates of the tricarboxylic acid cycle (Kreb’s cycle) supporting
   oxidative phosphorylation [[125]55]. The plasma concentrations followed
   mice response and recovery dynamics, decreasing at day 3, reaching a
   minimum concentration at day 7, when viral infection peaks, and
   returning to the initial concentration at day 14 [[126]54]. These
   results were also correlated to immune response, including the rise and
   fall of natural killer cell populations, serum soluble TNF receptor
   concentration, and viral clearance, providing potential miRNA-mediated
   pathway targets for molecular diagnostics or therapeutics of
   mammarenavirus infection and immune response.

   Similar results have been reposted, where glutamine is important to
   support HIV-1 replication, when comparing HIV-1 infected with
   uninfected activated primary human CD4^+ T cells. Glutamine
   concentrations are elevated in HIV-1-infected cells, implying that
   HIV-1 infection leads to considerable changes in the cellular glutamine
   metabolism [[127]56]. In another study using metabolomics, researchers
   showed that the latently infected cells with Kaposi’s
   Sarcoma-associated herpesvirus (KSHV) have higher levels of glutamine
   as compared to their mock counterparts [[128]48]. Lysine is another
   amino acid that has been shown to play an important role on viral
   infections. It is essential for the replication of viruses and
   progression of infections [[129]57,[130]58,[131]59]. The extra- and
   intracellular concentrations of l-lysine amino acid play a limiting
   role in the synthesis of the virus proteins and in transcription
   initiation of the retrovirus life cycle, and the deficiency of this
   essential nutritional element can reduce viral load [[132]60]. Studies
   have demonstrated serious changes in plasmatic concentrations of
   l-lysine in HIV-infected patients, where it was found that plasma
   concentrations of l-lysine were negatively correlated with HIV-1 RNA
   levels and inversely with CD4 lymphocytes count. Therefore, an excess
   of l-lysine concentration leads to active HIV replication and reduces
   the concentration of plasma amino acid [[133]59,[134]61].

   In a recent review, Keshavarz and colleagues, based on available
   metabolomic studies, argued that influenza virus infection can affect
   with cellular metabolic pathways either directly or indirectly via
   stimulation of immune system mediators [[135]62]. Curiously, through
   enhancing the activity of the mTORC1 complex, the influenza virus
   strengthens numerous metabolic pathways, including glycolysis,
   glutaminolysis, pentose phosphate, and fatty acid synthesis, to provide
   more ATP and structural materials for viral replication
   [[136]62,[137]63,[138]64]. Mammalian target of rapamycin (mTOR) is a
   conserved serine/threonine kinase that plays a critical role in the
   control of cellular growth and metabolism [[139]65]. Mapping of the
   mTOR signaling pathways has revealed that mTOR controls biomass
   accumulation and metabolism by modulating key cellular processes,
   including protein synthesis and autophagy [[140]66]. Some important
   studies have revealed interplays between miRNAs and mTOR pathway during
   cancer development. These interactions appear to provide a fine-tuning
   of various cellular functions and contribute qualitatively to the
   behavior of cancer [[141]67,[142]68]. Furthermore, excessive
   upregulation of mTOR complex 1 (mTORC1) protein complex pathway could
   be detrimental because it leads to enhanced production of particular
   miRNAs [[143]69].

   Vital metabolic pathways of host cells are one of the most widely used
   mechanisms targeted by viruses, resulting in extensive modifications.
   Some studies have revealed that many different human viruses, such as
   dengue [[144]70], cytomegalovirus [[145]44,[146]71], and rubella
   [[147]72], can strongly affect host cell glycolysis, lipid metabolism,
   and glutaminolysis. Cheng and collaborators demonstrated that
   enterovirus 71-infected Vero cells had significant changes in
   glutathione and its related metabolites, and several amino acids, such
   as glutamate and aspartate. Furthermore, they associated the presence
   of glutamine in culture medium with an increase in viral replication,
   and a dimethyl α-ketoglutarate treatment partially mimicked the effect
   of glutamine supplementation [[148]73].

   The small molecule profile of serum from Lassa fever-infected patients
   indicated a physiological dysregulation affecting pathways mediating
   blood coagulation, and lipid, amino acid, nucleic acid metabolism
   changing the levels of numerous metabolites in the circulation
   [[149]74]. Serum lipids were the most frequently identified molecular
   class and also the most frequently identified as decreased in fatal
   Lassa fever and LCMV infected mice as well as proteolytic breakdown
   products as the dipeptides γ-glutamyl-Valine, γ-glutamyl-Leucine, and
   prolyl-hydroxyproline, which were present in lower amounts over the
   course of LCMV infection. Suggesting possible changes in the
   concentration or activity of enzymes, Gamma glutamyl transferase
   transfers the gammaglutamyl moiety to various amino acids during viral
   immune response [[150]54,[151]74]. In this regard, our study predicted
   several miRNA related to amino acid and fatty acid biosynthesis and
   metabolic pathways accounting as a potential mammarenavirus strategy to
   subvert the cell metabolism to its benefits, increasing catabolism of
   unnecessary compounds and favoring synthesis or suppressing metabolism
   of those necessary for replication and evasion of immune system.

   Biotin metabolism was among the most frequent pathways regulated by
   miRNAs that target mammarenavirus genome. Biotin may affect
   transcription of genes, biotinylation of proteins in cells, cellular
   growth, proliferation, and differentiation. The vitamin acts as a
   co-factor for five carboxylases that are critical for fatty acid,
   glucose, and amino acid metabolism [[152]75,[153]76]. Biotin deficiency
   is associated with various diseases, and vitamin deficient mice display
   enhanced inflammation. Moreover, abnormal cellular growth and
   differentiation is the underlying cause of fetal malformations
   [[154]77,[155]78] and impaired immune function observed in
   biotin-deficient animals [[156]79]. The predicted miRNAs can play an
   important role in congenital LCMV infection that are marked by
   permanent, lifelong neurologic deficits, the most common are
   macrocephaly (secondary to hydrocephalus), and microcephaly
   [[157]80,[158]81]. Biotin was also shown to be important for the
   activity of human natural killer lymphocytes, for the generation of
   cytotoxic T lymphocytes, and for the maturation and responsiveness of
   immune cells [[159]75,[160]82,[161]83].

   One important pathway predicted to be associated with miRNAs targeting
   mammarenavirus genome was the circadian rhythm which is an internal
   biological clock that enables to sustain an approximately 24-h rhythm
   in the absence of environmental cues. The circadian clock system is a
   main regulatory factor for nearly all physiological activities and its
   disorder has severe consequences on human health [[162]84,[163]85].
   Interestingly, Miller and colleagues performed a study where they
   infected mice with LCMV and measured plasma corticosterone and
   cytosolic glucocorticoid receptor (GR) binding at multiple time points
   throughout the day and throughout infection. Despite a vigorous immune
   response to LCMV, the infection was associated with minimal and
   transient increases in corticosterone secretion. However, significant
   decreases in cytosolic GR were detected in immune tissues. Receptor
   decreases were characterized by a significant reduction of GR binding
   during the diurnal rise in corticosterone in the spleen and thymus of
   infected but not uninfected animals on days 5–10 post infection. In
   addition, in the morning on these days, GR binding in the spleen of
   infected mice was decreased compared to uninfected control mice
   [[164]86]. This indicates effects of LCMV infection in the circadian
   rhythm with interactions between the neuroendocrine and immune systems
   modifying at the level of the GR the context of an ongoing immune
   response during a viral infection.

   Similar results were found by Edgar and colleagues in infected
   wild-type mice with a strain of herpes virus and influenza A at
   different phases of the animals’ circadian clocks. These researchers
   demonstrated that the time of day of virus infection has an important
   impact on disease progression, in cellular models as well as in
   animals. In addition, they observed that clock disruption leads to
   increased virus replication and dissemination, indicating that severity
   of acute infections is influenced by circadian timekeeping [[165]87].
   Since there is a possibility of circadian rhythms affecting the immune
   response in the animals, the timing of vaccination should also be
   important. Indeed, timing of influenza vaccine administration has been
   shown to be a determinant in the systemic immunization response in
   people over 65 years of age. In a cluster-randomized trial design, Long
   and collaborators examined whether manipulating the time of day an
   older adult received their influenza vaccination would have an effect
   on the magnitude of the antibody response at one month. Results showed
   that antibody responses to two of three influenza strains were higher
   when the vaccination was given in the morning [[166]88].

   The identification of the molecular determinants underlying the
   distinct virulence of pathogenic and nonpathogenic arenaviruses is
   still a question of great importance for epidemiology and public health
   [[167]89]. In our study, there was no clear difference in total miRNAs
   found with the potential to bind to pathogenic and nonpathogenic
   mammarenavirus genomes, which could be explained by the close genetic
   relationship of nonpathogenic arenaviruses with their highly pathogenic
   cousins [[168]1,[169]89,[170]90]. Although the role of miRNA in
   mammarenavirus pathogenesis is not clear, its role in regulating
   important molecular pathways during infection cannot be ruled out as an
   important mechanism to be further investigated. In addition, many
   studies have suggested that a multifactor pathway could be involved in
   the different outcomes from arenaviral infection including a relatively
   low number of adaptive mutations, as well as the incapacity to bind to
   human receptors and different immune responses elicited by these
   viruses [[171]91,[172]92,[173]93].

   Interaction between host and pathogen has a profound effect on the
   outcome of an infection. Knowledge about host–pathogen interactions is
   critical for understanding the pathogenesis of infection. However, it
   generally overshadows knowledge of metabolic cross-talk between host
   and pathogen [[174]43,[175]94]. At a cellular level, host and pathogen
   share similar nutritional substrates generating common metabolic
   products. The host depends upon nutritional substrates to support its
   immune responses against the pathogen, while the pathogen is also
   highly dependent on nutritional substrates being unable to synthesize
   some or all substrates for its replication [[176]49,[177]94,[178]95].
   Post-transcriptional regulation of gene expression plays a pivotal role
   in various gene regulatory networks including but not limited to
   metabolism, embryogenesis, and immune responses. Recently, the role of
   post-transcriptional gene regulation during pathogenic infections and
   host immune responses has become increasingly important. The role of
   cellular miRNAs is crucial when they are proviral, or when a longer,
   persistent infection is established. miRNAs can be used as an entry
   gate into regulatory networks that could be explored to find new
   unconventional therapeutic targets.

5. Conclusions

   Our present study indicates that mammarenavirus infection induces
   metabolic reprogramming of host cells, probably manipulating cellular
   microRNAs. A number of metabolic pathways, including valine, leucine,
   and isoleucine biosynthesis, d-Glutamine and d-glutamate metabolism,
   thiamine metabolism, and pools of several amino acids were impacted by
   the predicted miRNAs that would no longer regulate these pathways. The
   change caused by the effect of the mammarenavirus genome acting as a
   sponge to miRNAs probably allows activation of anabolic pathways
   necessary for the production of the viral nucleic acids, capsids, and
   eventually membrane envelopes. While some studies have demonstrated
   that viruses reprogram cell metabolism and rely on metabolic changes
   for optimal virus replication, significant work remains to determine as
   viral genome interact with host cell machinery to induce such
   alterations and characterize whether the same metabolic perturbations
   occur during infection. As well as other viruses, the mammarenavirus
   has likely evolved to modify the host metabolism for multiple purposes,
   facilitating viral replication and potentially also reflect antiviral
   defense mechanisms. The study of metabolism, including metabolomics
   approach, should not be restricted to energy supply and biosynthetic
   materials, but should also extend to a better understanding of its role
   during viral infection and pathogenesis. A deeper understanding of
   mechanisms by which mammarenaviruses handle these signaling pathways is
   critical for better understanding the virus/host interactions and
   potential targets for diagnostic and therapeutic, through inhibition of
   the specific pathologic metabolic pathways.

Supplementary Materials

   The following are available online at
   [179]https://www.mdpi.com/2073-4409/9/11/2525/s1, Figure S1: List of
   all 75 pathways found, Figure S2: All pathways found for each clade,
   Figure S3: All 69 pathways with at least one experimentally validated
   interactions for miRNAs, Figure S4: All 75 pathways distributed
   according to amount of microRNA interactions for each virus, Table S1:
   All microRNAs found, their binding sites and the location on the viral
   genome, Table S2: List of all miRNAs binding to different regions of
   same the RNA segment of different mammarenavirus species, Table S3:
   List of all miRNAs binding to S and L segments of different
   mammarenaviruses.
   [180]Click here for additional data file.^ (5.3MB, zip)

Author Contributions

   Conceptualization, J.F., R.L.M., and A.G.; methodology, R.L.M. and
   A.G.; formal analysis, J.F., R.L.M., and A.G.; data curation, A.G.;
   writing—original draft preparation, J.F., R.L.M., and A.G.;
   writing—review and editing, E.R.S.d.L. and A.G.; funding acquisition,
   E.R.S.d.L. All authors have read and agreed to the published version of
   the manuscript.

Funding

   This research was funded by Coordenação de Aperfeiçoamento de Pessoal
   de Nível Superior—Brazil (CAPES)—Finance Code 001, and Conselho
   Nacional de Desenvolvimento Científico e Tecnológico (CNPq)—Brazil. The
   APC was funded by Instituto Oswaldo Cruz—Fundação Oswaldo Cruz
   (Fiocruz)—Brazil.

Conflicts of Interest

   The authors declare no conflict of interest.

Footnotes

   Publisher’s Note: MDPI stays neutral with regard to jurisdictional
   claims in published maps and institutional affiliations.

References