Abstract

   MicroRNAs (miRNAs) are single-stranded RNA molecules that control gene
   expression in various processes, such as cancers, Alzheimer’s disease,
   and bone metabolic diseases. However, the regulatory roles of miRNAs in
   osteoporosis have not been systematically analyzed. Here, we performed
   a comprehensive analysis to identify the differentially expressed
   miRNAs involved in osteoporosis. MiRNAs associated with osteoporosis
   were collected through literature retrieval and further screened based
   on specific inclusion and exclusion criteria. The osteoporosis
   therapeutic targets of miRNAs were obtained by the integration of
   miRWalk 3.0 database and five human disease therapeutic target
   databases. Then, the network analysis and functional enrichment
   analysis of miRNAs and their targets were performed. As a result, 11
   eligible miRNAs were identified highly associated with osteoporosis.
   MiRNA-mRNA network demonstrated there were the complex mutual
   interactions between miRNAs and their targets. Besides, ADRB2, AR,
   ESR1, FGFR1, TRAF6, etc., were identified as the top hub genes in
   protein-protein interaction (PPI) network. Functional enrichment
   analysis revealed that miRNAs and their targets were mainly mapped on
   processes associated with bone and immune system, such as bone
   remolding, bone mineralization, PI3K/AKt, TNF signaling pathways and
   Th17 cell differentiation. RT-PCR results showed that the expression of
   miR-335-3p was significantly down-regulated in hind limb unloading
   (HLU) mice tibia samples compared with controls, the remaining 10
   miRNAs were significantly up-regulated after HLU (P < 0.01). In
   summary, we identified 11 differentially expressed miRNAs and their hub
   target genes in osteoporosis, which may be novel diagnostic biomarkers
   for osteoporosis.

   Keywords: osteoporosis, microRNAs, bioinformatics analysis, biomarker,
   target genes

Background

   Osteoporosis is a subclinical chronic bone metabolic disease
   characterized by low bone mineral density (BMD), progressive bone loss,
   and destroyed bone microstructure. The typical signs and symptoms of
   osteoporosis mainly include pain, stiffness, skeletal hypofunction, and
   limitation of movement, even companied by an increased risk of
   fractures ([49]1). Recently, osteoporosis has been aroused considerable
   attention all over the world given the growing morbidity and disability
   as well as the substantial cost to health care and social services
   worldwide ([50]2). According to World Health Organization (WHO), more
   than 69.4 million people > 50 years are hampered by osteoporosis and
   approximately 687,000 populations develop to osteoporotic fractures
   each year each year in China ([51]3). In US, it has been estimated that
   the financial costs associated with bone fractures will reach $25.3
   billion by the end of 2025 ([52]4). Therefore, accurate early diagnosis
   of onset is the key to the prevention and effective therapy for
   osteoporosis. If increased risk of bone loss is diagnosed prior to the
   first occurrence of osteoporosis, the morbidity may be significantly
   reduced by preventive pharmacologic treatments and/or lifestyle
   interventions. Presently, despite that BMD measurement by dual-energy
   X-ray absorptiometry (DXA) is the most recommended tool for
   osteoporosis risk monitoring, it involves complex equipment, which is
   bulky, expensive, and emits radiation ([53]5). Some bone metabolism
   biochemical indicators, such as alkaline phosphatase (ALP), osteocalcin
   (OCN), N-terminal propeptide of type I collagen (PINP, a recommended
   biomarker of bone formation), β-isomerized C-terminal telopeptide of
   type I collagen (β-CTX), and parathyroid hormone (PTH), are also
   applied in the early diagnosis of osteoporosis, exhibiting
   high-sensitive, and strong-specific in reflecting bone homeostasis
   ([54]6). However, owing to that these conventional biochemical
   indicators above are unstable inherently; the detection results of
   these biomarkers are not entirely reliable and stable. The mentioned
   problems of these existing osteoporosis diagnosis methods underscored
   the utmost importance for the identification of more specific and
   reliable biomarkers in the early diagnosis of osteoporosis.

   Fortunately, over the past few decades, a novelty kind of RNA,
   microRNAs (miRNAs) have attracted tons of attention from researchers
   and clinicians. miRNAs, endogenous, single-stranded non-coding RNA (~20
   nucleotides long), regulate gene expressions at post-transcriptional
   level through recognizing the complementary miRNA recognition element
   (MRE) or seed-matched sequences located in the 3’-untranslated region
   (UTR) of mRNAs ([55]7). Despite that miRNAs share several sources of
   variability with enzyme/peptide markers, such as biological factors,
   technical issues, standardization of methods, and the use of internal
   controls, multiple researches have been conducted to measure the
   sources of miRNA variability and explore effective strategies to
   improve the detection power of the changes in miRNA expression ([56]8,
   [57]9). Moreover, a great deal of researches showed that abnormal
   expression of miRNAs was closely associated with the occurrence of
   various diseases, such as cancers, Alzheimer’s disease, and bone
   metabolic diseases, especially in osteoporosis ([58]10–[59]12). During
   the past decades, there were already more than 200 articles directly
   working on miRNAs in osteoporosis. For example, Wang et al. found that
   miR-133a was a potential biomarker in circulating monocytes for
   postmenopausal osteoporosis ([60]13). Another study revealed that
   miR-218 exerted a negative regulatory role in osteoclastogenesis and
   bone resorption by suppressing the p38MAPK-c-Fos-NFATc1 pathway
   ([61]14). In addition, Bedene and his colleagues identified the
   positive regulatory role of miR-148a-3p in osteoclast differentiation
   and bone homeostasis ([62]15). All these studies hinted at the
   potential indicative roles of miRNAs in the diagnosis of osteoporosis.
   However, the regulatory effects of miRNAs involved in the bone
   homeostasis of osteoporosis have not been systematically explored until
   now.

   Herein, we attempted to carry out an integrated bioinformatics analysis
   to identify the expression profiles of miRNAs and their potential
   targets in osteoporosis, which may provide novel miRNA profiles as
   diagnostic biomarkers for osteoporosis.

Materials and Methods

Literature Retrieval Strategy

   Wide-scale literature retrieval was conducted on PubMed, Springer Link,
   Web of Science and CNKI (China National Knowledge Infrastructure) from
   December 2008 to December 2019. Literature searching full-process
   consists of four steps: First, to be as inclusive as possible, single
   terms such as microRNAs, miRNAs, mirna*, or microrna* were applied to
   obtain all miRNAs-related studies from the databases mentioned above.
   The same retrieval approach was used for osteoporosis: terms including
   “osteoporosis” and other more specific terms such as “osteoblasts”,
   “osteoclasts”, “low bone mineral density (BMD)”, “bone loss”, or “bone
   quality” were used as the keywords. Second, a combination of controlled
   vocabulary and text words were applied to retrieve literatures
   concerned miRNAs in osteoporosis by entering “terms relevant to
   osteoporosis” AND “microRNAs-related text” in the search box of the
   databases. Despite the adopted systematic search strategy, some
   relevant articles still had not been included. Third, to ensure the
   comprehensive of this analysis, some missing additional articles were
   added into the list of included literatures and labeled as ‘not
   identified from search strategy’ for transparency. Finally, to
   facilitate the collection and consultation of information, full-texts
   of all retrieved results were downloaded into Endnote (EndNote X8, Bld
   7072, Thomson Research Soft, Stamford), and duplicates were removed.

Selection Criteria and Data Extraction

   To further screen the eligible literatures, titles, abstracts, and, if
   necessary, full texts of the initially retrieved references were