Acapsular Staphylococcus aureus with a non-functional agr regains capsule expression after passage through the bloodstream in a bacteremia mouse model

Selection pressures exerted on Staphylococcus aureus by host factors during infection may lead to the emergence of regulatory phenotypes better adapted to the infection site. Traits convenient for persistence may be fixed by mutation thus turning these mutants into microevolution endpoints. The feasibility that stable, non-encapsulated S. aureus mutants can regain expression of key virulence factors for survival in the bloodstream was investigated. S. aureus agr mutant HU-14 (IS256 insertion in agrC) from a patient with chronic osteomyelitis was passed through the bloodstream using a bacteriemia mouse model and derivative P3.1 was obtained. Although IS256 remained inserted in agrC, P3.1 regained production of capsular polysaccharide type 5 (CP5) and staphyloxanthin. Furthermore, P3.1 expressed higher levels of asp23/SigB when compared with parental strain HU-14. Strain P3.1 displayed decreased osteoclastogenesis capacity, thus indicating decreased adaptability to bone compared with strain HU-14 and exhibited a trend to be more virulent than parental strain HU-14. Strain P3.1 exhibited the loss of one IS256 copy, which was originally located in the HU-14 noncoding region between dnaG (DNA primase) and rpoD (sigA). This loss may be associated with the observed phenotype change but the mechanism remains unknown. In conclusion, S. aureus organisms that escape the infected bone may recover the expression of key virulence factors through a rapid microevolution pathway involving SigB regulation of key virulence factors.

Staphylococcus aureus is a transient, sometimes permanent member of the human microbiota in the nares and skin of a significant number of healthy individuals. As predisposing conditions emerge in the host, this opportunistic species may cause infections with different severity, ranging from mild skin and soft tissue infection to severe disseminated disease 1,2 . Treatment of S. aureus infections is hampered by widespread dissemination of methicillin-resistant S. aureus (MRSA) 3 and by the frequent emergence of S. aureus with low level resistance to vancomycin 4 . Whereas MRSA was primarily considered a nosocomial pathogen 5 , it is now unanimously accepted that MRSA also affects individuals of the general community with no previous exposure to health care settings 6,7 . Up to 1.5-2% of patients receiving an orthopedic prosthetic device becomes infected and a significant number of these infections are caused by S. aureus. Most osteomyelitis caused by S. aureus become refractory to antibiotic treatment soon after bacteria settles on the prosthetic device surface and in bone tissue 1 . S. aureus has the ability to swiftly adapt to the conditions encountered at the infection niche by adjusting its metabolism and/or Isolate P3.1 characterization. Phenotypic characterization revealed that strains HU-14 and P3.1 have almost identical mean generation times (MGT) (40 min and 39 min, respectively). Biofilm production was significantly higher (p < 0.0001, Student t test) in the P3.1 derivative [OD B /OD G = 0.60 ± 0.02 (n = 42)] when compared with the parental HU-14 strain [OD B /OD G = 0.35 ± 0.01 (n = 42)], as assessed by the crystal violet test. Total hemolytic activity was negative for both strains as well as for alpha-and beta-hemolysins. Proteolytic activity was negative for both strains too. Both strains were vancomycin susceptible. Transcriptional evaluation of the Agr system in HU-14 and P3.1 revealed that the expression of RNAIII and agrA was still abolished in P3.1 as ascertained through qRT-PCR (Fig. 2a). Amplification of the P3.1 agrC gene produced an amplicon of identical size (2,793 bp) to that observed in HU-14, which suggested that IS256 remained positioned in the agrC gene of isolate P3.1 (Fig. 2b). Sequence analysis revealed that the insertion site of IS256 in agrC was identical in both strains and positioned after base 218. In synthesis, S. aureus regained CP expression in the absence of a functional Agr system (lack of RNAIII and agrA expression). Another noticeable phenotypic feature was that P3.1 produced the classical yellowish-golden colonies that give the name to this bacterial species whereas its parental HU-14 strain produced whitish colonies on TSA (Fig. 3a). Methanolic extracts from HU-14 exhibited a significantly lower absorbance at 450 nm when compared with those of P3.1, which suggested that P3.1 produced increased levels of staphyloxanthin (Fig. 3b).
Whole genome sequence analysis. Analysis using Illumina short read sequences confirmed that HU-14 and P3.1 were isogenic, with only one single nucleotide polymorphism (SNP) in P3.1. It consisted of replacement of C by T in position 589 of the icaA gene, leading to a change from alanine to threonine in position 197. Phenotypical analysis of the icaADBC product, namely the polysaccharide intercellular adhesin (PIA) revealed negligible PIA production with no significant differences between HU-14 and P3.1 (Fig. 4). In addition to agrC, which was mutated by the IS256 insertion in both strains, further short read sequence analysis determined that no genomic modification of any other known major regulator occurred which may explain the P3.1 capsulated phenotype. Furthermore, no SNP was found in the intergenic regions which may affect the expression of any regulator.
Pacific Biosciences sequencing was performed to assess whether there was any structural difference between the HU-14 and P3.1 genomes. Hybrid Unicycler assembly of PacBio and Illumina reads resulted in two genome fragments. The smaller fragment in the HU-14 and P3.1 hybrid Unicycler assemblies was a circular sequence that shared 99.98% and 99.80% identity respectively with plasmid pCM05 (PLSDB database). Annotation with Prokka Scientific RepoRtS | (2020) 10:14108 | https://doi.org/10.1038/s41598-020-70671-1 www.nature.com/scientificreports/ using the pCM05 plasmid from the NCBI GenBank Database (NC_013332.1) as a reference and SnapGene sideby-side analysis with the pCM05 GenBank file confirmed that the circular sequence was indeed pCM05 in both assemblies, and contained all the same genes and restriction sites. The plasmid sequences were also aligned using ProgressiveMaude, but these did not show any rearrangements. The size of the larger fragment (2,972,382 bp and 2,971,047 bp in HU-14 and P3.1, respectively) was consistent with that of the S. aureus chromosome. Whole genome alignment of the chromosome sequences was performed using progressiveMauve, showing no chromosomal rearrangements between the two. However, annotation with Prokka revealed that P3.1 has lost one IS256 copy originally located in the non-coding region downstream of DNA primase (dnaG, LJDFIFNA_00432) and upstream of RNA polymerase sigma factor RpoD After 24 h mice were sacrificed and blood obtained by cardiac puncture. Blood was plated quantitatively and after an overnight culture all colonies from a plate were harvested and suspended in PBS, and the suspension adjusted to a density of 1 × 10 9 per ml for further injection to another group of mice. Suspension aliquots were plated for CFU count and colony immunoblots were performed on the grown TSB plates to detect the emergence of any colony producing CP5. (b) Colony immunoblot of strain HU-14 after three passages through mice. The arrows indicate strains Reynolds CP5, Reynolds CP8 and Reynolds NT. These three control strains were added by impregnation of the blot membrane. (c) Magnification of the photograph shown in (b www.nature.com/scientificreports/ (rpoD, LJDFIFNA_00434) (sigA) (Suppl Fig. 1). IS256 excision in P3.1 was clean and left a conserved region between dnaG and rpoD identical to those of N315, USA300 and other ST5 isolates from our collection. Interestingly, when both hybrid assembly and only short read analysis was performed the IS256 was definitively missing in P3.1 but such excision was not pinpointed by performing long read analysis only. Finally, SNPcalling between the two closed hybrid genomes was carried out using Snippy, with the Prokka-annotated HU-14 closed reference genome and the P3.1 Illumina reads, and vice-versa. This revealed the same single SNP as the same analysis using just the Illumina reads: an inconsequential change from alanine in HU-14 to threonine in P3.1, as described above.
Regulation of CP5 expression. It is known that regulation of CP5(8) expression in S. aureus is complex and involves a number of redundant mechanisms that act in concert. Evaluation of main S. aureus regulators sae and sarA at the transcriptional level by qRT-PCR revealed that no functional modification occurred (data not shown). However, transcription of asp23, a gene that is solely regulated by SigB, was significantly increased in P3.1 when compared with HU-14 (Fig. 5). Since staphyloxanthin production by S. aureus is positively regulated by SigB, the increased pigmentation displayed by P3.1 supports the hypothesis that S. aureus P3.1 reacquired the capacity to produce CP5 through a mechanism related to recovery of SigB functionality, in the absence of an operative Agr system.
Virulence study. The virulence of strain HU-14 and derivative P3.1 was tested in a mouse model of intraperitoneal injection. Mortality 5 days after challenge with P3.1 was higher than that in the group challenged with HU-14 but was not significantly different (Suppl Fig. 2a). Body weight was recorded over 5 days after challenge and the results showed that mice challenged with the P3.1 derivative lost more weight than those injected with the parental strain HU-14, but the differences were again not significant (Suppl Fig. 2b). Interestingly, mice challenged with P3.1 exhibited more prominent signs of distress, such as hunched posture and spiky fur, when compared with those challenged with HU-14.   www.nature.com/scientificreports/ osteoclastogenesis and cytokine production studies. The differential capacity of strains HU-14 and P3.1 to trigger osteoclastogenesis was investigated in vitro. Osteoclast precursors were stimulated with S. aureus in the presence of M-CSF. Forty-eight hours after stimulation cells were fixed and multinucleated TRAP ( +) cells were counted. Rapid differentiation into osteoclasts was induced by HU-14 with formation of multinucleated cells (3 or more nuclei) (Fig. 6a,b). In contrast, significantly decreased numbers of osteoclasts were observed when the osteoclast precursors were stimulated with P3.1 (Fig. 6a,b). S. aureus HU-14 and its P3.1 derivative were tested for their capacity to induce the production of TNF-α, a proinflammatory cytokine involved in osteoclastogenesis, by osteoclast precursors in culture. The P3.1 derivative exhibited a decreased capacity to trigger TNFα secretion (Fig. 6c).

Discussion
Staphylococcus aureus has the potential to adapt to diverse niches with defined microenvironments 21 . In a recent study we showed that a stable non-encapsulated derivative emerged during chronic osteomyelitis and displaced a more virulent CP8 parental strain which had been isolated several months before from the same infection site 9 . The S. aureus derivative, which bore a frameshift mutation in agrC that caused loss of both RNAIII transcription and CP8 expression, was considered a microevolution endpoint adapted to bone. Lack of CP expression benefits the interaction between surface S. aureus adhesins and eukaryotic cell receptors thus promoting internalization of bacteria into the eukaryotic cell milieu 8 . Intracellular location, reduced capacity to cause inflammation 13,22 , increased biofilm production and SCV formation 23 may explain why not only S. aureus pulmonary infection of cystic fibrosis patients 24 but also S. aureus osteomyelitis become refractory to antibiotic treatment in spite of appropriate antibiotic treatment. From our bacterial collection we selected non-encapsulated S. aureus isolate HU-14 with a non-functional agr due to an IS256 insertion in agrC. Although HU-14 may be considered a microevolution endpoint, in the present study we show that the passage of this isolate through blood in an experimental mouse model induced the reversion from a NT to a CP5 phenotype. Interestingly, it was not a true genetic reversion since IS256 remained inserted at the same location in the agrC of the CP5 isogenic derivative. Therefore, loss of the agr functionality due to a stable mutation (in this case caused by an IS256 insertion) may only be considered a microevolution endpoint at the precise niche where selection initially took place. A variable number of relevant S. aureus isolates are non-reactive to antibodies anti-CP5 and anti-CP8 as phenotypically ascertained in vitro in the clinical bacteriology laboratory. Diverse mechanisms responsible for the loss CP5(8) expression have been reported, one of which is related to dysfunction of the main global regulator Agr 17 . There is ample evidence that agr-deficient strains may not only be selected during infection 25 but also colonize healthy hosts 26 . In a study on 195 S. aureus isolates collected from different sources it was found that 49% of the isolates carried an inactive agr locus, as shown by the lack of a RNAIII transcript 27 . In the same study, interestingly, 55% of the strains with a non-functional agr produced capsule (CP5 or CP8). Such finding permits to speculate that the type of S. aureus derivative obtained after passage through the bloodstream described in the present report, which produce CP5 in the absence of a functional agr, may not be a rare event in clinical practice after all. CP5(8) is a strong candidate for construction of a multicomponent vaccine to prevent S. aureus severe infections 28 . Indeed, the most promising preparations contain CP conjugated to a toxoid plus other relevant antigens 16,[29][30][31] . The finding that prevalent clone USA300 in the USA is non-encapsulated due to mutations in the capsular genes rather than in a capsule regulator 32 appeared to cast doubt on the inclusion of CP as vaccine component. But on the other hand, many S. aureus clinical isolates carry a conserved cap cluster and, furthermore, it was recently demonstrated that clinical strains that do not produce CP in vitro are able www.nature.com/scientificreports/ to produce CP in vivo 20 . Our present study demonstrates the feasibility that S. aureus may recover in a stable manner CP5 production, which facilitates phagocytosis evasion once bacteria reach the bloodstream, in spite of a non-functional Agr. The results of our study suggest that the inclusion of CP in a multicomponent vaccine preparation should not be dismissed. Staphylococcus aureus possesses an intricate regulatory system that governs the production of soluble virulence factors and exoenzymes. In addition to Agr, main regulators include a number of two-component systems, such as SaeRS, transcriptional factors 33 , such as SarA and its homologues 34 and factor σB (sigB) 35 . Bioinformatic analysis of strains HU-14 and P3.1 sequences revealed that the sequences of the aforementioned regulators were conserved, with no differences between the parental strain and the derivative. Since it was recently demonstrated that CP production is up regulated by SigB we focused our attention on this regulator 36 . The SigB regulon is known to respond to different stress signals and accordingly regulates stress responses 33 . One of the marker genes upregulated by SigB is asp23 37 . An increase of SigB activity in P3.1 was suspected by noteworthy pigmentation of P3.1 colonies. Indeed, another known SigB-regulated trait is production of staphyloxanthin 38 . Further assessment revealed that P3.1 produced pigments whereas the parental strain HU-14 did not, as determined spectrophotometrically. In the present report, the significant increase in asp23 expression revealed that the increase in SigB functionality was associated with the recovery of CP5 production. It is hypothesized that adaptation of S. aureus through a long-term process in the host may have led to reduced SigB expression and perhaps loss of the Agr function seen in HU-14. Experimental passage of this isolate through the bloodstream induced reacquisition of CP expression with parallel increase of SigB expression, as ascertained by increased asp23 expression. In favor of this hypothesis, Marbach and coworkers 39 has found that S. aureus infecting the bovine udder may not require sigB expression once it is established in the mammary gland of the cow with chronic mastitis. Interestingly, as we hypothesized before, loss of SigB expression in the infected cow was eventually fixed by mutation 8 . In this case the mutated gene was rsbU 39 .
Whole genome sequence analysis from reads obtained by Illumina sequencing did not reveal any genetic lesion directly or indirectly responsible for the observed SigB increased expression. Therefore, the presence of structural changes in P3.1 was then investigated by long-read sequencing. Hybrid (PacBio and Illumina) sequence analysis of the HU-14 and P3.1 chromosomes as well as of plasmid pCM05 suggested a structural change in www.nature.com/scientificreports/ P3.1 compared with HU-14, which consisted of an IS256 excision located between dnaG (DNA primase) and rpoD (sigA). This preliminary observation was not surprising since mobile genetic elements are key drivers of evolution in S. aureus 40 . Giulieri and coworkers 41 have indeed demonstrated that insertion of IS256 elements enhances genetic diversity during infection thus representing an effective driver of within-host microevolution. Furthermore, insertion of IS256 in agrC was found to promote increased fitness of S. aureus as a compensatory mechanism for the biological cost resulting from acquisition of a high number of multiresistant traits by hospital isolates 42 . In this regard, it is speculated that reacquisition of SigB activity and CP5 as well as staphyloxanthin production was perhaps achieved by restoration of SigA activity since IS256 was excised without affecting the original conserved intergenic non-coding region between dnaG (DNA primase) and rpoD (sigA). In support of this hypothesis, it has been shown that the sigB operon is transcribed from at least two differentially controlled promoters, one of which is a putative σ A -dependent promoter termed sigB P1 43 . Therefore, sigB expression is driven by SigA, which leads to the transcription of the whole operon, as well as through autoregulation by SigB itself, leading to transcription of rsbV and rsbW 33 . Further research is required to better understand the complex mechanisms that permit concomitant reacquisition of CP5 and pigment production by S. aureus in the presence of an inactive agr and a functional sigB operon, which is known to be driven by SigA.
Altman and coworkers have shown that within-host loss of agr function was associated with increased genetic divergence between distinct S. aureus subclones 44 . In synthesis, whereas agr dysfunction is adaptive for survival in the infected bone, it may be counter adaptive outside such niche. As observed before, agr mutations do not tend persist in natural S. aureus populations 45 . In the present report we show that S. aureus can regain CP production when subjected to selection pressure in vivo through a mechanism that does not require RNAIII expression. Indeed, in a previous study it has been shown that genetic changes occurring in an agr-defective infecting S. aureus strain resulted in increased virulence in a murine model of bloodstream infection thus bypassing the agr mutation 44 . The S. aureus derivative obtained after passage through the bloodstream displayed a significant reduction in the capacity to trigger TNFα production and became less adapted to bone as ascertained by its decreased capacity to induce osteoclastogenesis. The S. aureus strain adapted to blood not only conserved but increased its capacity to produce biofilm. Recovery of at least two important traits such as CP5 and pigment production permit bacteria to evade innate immune mechanisms of defense in the blood [46][47][48][49] and readapt to a new environment elsewhere in the host with the potential to cause methastatic infection. The fact that S. aureus can adapt to blood by regaining the capacity to produce CP5 and pigments does not necessarily mean that virulence should increase. P3.1 exhibited a trend to be more virulent, as ascertained in the mouse model of intraperitoneal challenge, but the differences were not significant. This was not a surprising fact since the Agr remained non-functional and many other Agr-regulated virulence factors are not being expressed, such as the hemolysins. Therefore, as much as S. aureus is a multifactorial pathogen, it is speculated that its adaptation to a given niche in the host is most probably multifactorial too. In the present study, modification of virulence factor expression due to increased SigB activity was enough to reduce adaptation to bone but not to increase virulence in a significant fashion.

conclusion
Here we demonstrated that S. aureus adapted to infected bone can regain the ability to produce CP, even in the presence of a mutated, non-functional Agr when transferred to the bloodstream. S. aureus organisms that escape the infected bone can potentially become a threat because they can recover in a stable manner the expression of traits that permit survival in blood, such as CP and staphyloxanthin. Therefore, a supposedly harmless endpoint-of-microevolution may turn back into a dangerous infective derivative. Given that CP is one of the factors identified in this microevolution process, our findings support the idea that S. aureus CP should not be dismissed as a multifactorial vaccine candidate component.

Materials and methods
Bacterial strains and cultures. Staphylococcus aureus clinical isolate HU-14 from our strain collection was originally recovered in 2005 from a 34-year old male patient with chronic osteomyelitis of the right femur with a prosthetic implant ("Hospital Ramón Carrillo", Buenos Aires, Argentina). Species was confirmed by a species-specific PCR 50 . Strain HU-14 is ST5, CC5, SCCmec type 1, Agr type II and Spa t149 51 . All strains utilized in the study were kept frozen in trypticase soy broth (TSB) with 20% glycerol at − 80 °C and S. aureus was routinely cultured at 37 °C for 24 h on TSB unless otherwise indicated. In addition, S. aureus strain 6850 was used as a qRT-PCR positive control for Agr and SigB functionality testing. S. aureus strains RN6911 and SH1000 were utilized as negative and positive controls, respectively, for staphyloxanthin production. S. aureus strains Sa113 and MBD034 were utilized as positive and negative controls, respectively, for PIA production. Strain HU-8 (CC5, SCCmec type 1, Agr type II, Spa t149 and produces CP5), which bears a functional agrC, was also utilized as control. www.nature.com/scientificreports/ by exposure to CO 2 and blood was drawn by cardiac puncture and quantitatively plated on trypticase soy agar (TSA). Plates were incubated for 24 h at 37 °C and colonies were suspended PSS to ca. 1 × 10 9 CFU to prepare a 10 8 CFU inoculum to be re-injected in the next group of mice. This cycle was repeated seven times. To assess the differential virulence of the strains under investigation mortality curves were compared. Groups of 10 CF1 outbred mice were injected with 1 × 10 8 CFU of the strain to be tested suspended in 0.5 ml of 2% (w/v) Brewer´s yeast (Sigma Chemical Co.) in TSB broth. Mortality and body weight were assessed daily for 5 days. Surviving mice were euthanized by cervical dislocation. The Kaplan-Meier mortality curve was obtained using the Graph-Pad Prism software (GraphPad Software, Inc., La Jolla, USA; version 6.00).
illumina whole genome sequencing and analysis. DNA 57 , followed by long-read polishing with Racon v1.4.11 58 and short-read polishing with Pilon v1.22 59 . The completeness of the closed genome sequences was assessed using BUSCO v4.0.2 60 . The "bacteria" database was used to search the genomes for 120 universal bacterial genes, and the more specific "Bacillales" database was used to search for 450 genes common to all Bacillales species. Prokka v1.14.5 54 was used to annotate the closed genome sequences, using the protein sequences from the S. aureus strain N315 (NC_00275.2) as a reference. The annotated GenBank format files were then used to call SNPs between the two samples with Snippy v4.5.1 (https ://githu b.com/tseem ann/snipp y); the HU-14 GenBank file was used as the reference with the P3.1 Illumina reads, and vice-versa. The plasmid sequence for each sample was compared to the PLSDB plasmid database v.2019_10_07 61 (11), and annotated with Prokka (1.14.5) 54 to confirm its identity. The closed chromosome and plasmid sequences from each sample were rotated to start with the same genes (dnaA for the chromosome, repA for the plasmid) and aligned using progressiveMauve (build version: Sep 16 2015). The output files were inspected visually for any rearrangements using the progressiveMauve GUI.
phenotypic studies. CP5 production was evaluated by colony immunoblot on TSA plates as described previously 62 . Phenotypic expression of α-and β-hemolysin was performed by evaluating the production of the hemolysis in rabbit and goat blood agar (α-and β-hemolysin, respectively). Proteolytic activity was assessed in TSA supplemented with 10% whole milk. After 18 h incubation at 37 °C the hemolytic and proteolytic halos were evaluated, respectively. Ethanolic extracts of the carotenoid pigment from bacterial suspensions with equal optical density (OD) at 600 nm were quantified spectrophotometrically at 450 nm 63 . The quantitative assessment of biofilm formation was performed as previously described with modifications 64 . Briefly, S. aureus strains were grown for 18 h and diluted 1:100 in TSB supplemented with 0.25% of glucose (TSBg). After 24 h of static incubation at 37 °C in 96-well microtiter plates, bacterial growth in each well was measured by optical density reading at 595 nm (OD G ) using a microplate reader (Multiskan EX). Then, the wells were washed two times with PBS, the biofilms were fixed with 100% methanol for 15 min, stained with 0.5% crystal violet for 20 min, and washed twice again gently under running tap water. The amount of biofilm biomass was measured after addition of 30% glacial acetic acid by OD reading at 595 nm (OD B ). The biofilm biomass was expressed relative to the final cell density measured prior to crystal violet staining (biofilm OD B /OD G ). Three independent experiments were performed in sixtuplicate measurements. PIA production was assessed by ELISA according to a protocol described elsewhere 64 . Briefly, the procedure was similar to that described above for biofilm except that, after blocking solution removal,100 μl per well was added of a 75 ng/ml wheat germ agglutinin (WGA)-HRP conjugate, a lectin that binds to PIA sugar residues.
Real-time quantitative reverse transcription PCR (qRT-PCR). Bacterial RNA was extracted from S. aureus grown in TSB until post-exponential phase, using TRIzol Reagent (Invitrogen Life Technologies), according to the manufacturer's protocol. RNA was subjected to DNAse treatment using a RQ1 RNAse free DNAse (Promega). cDNA synthesis was performed with an ImProm-II Reverse Transcriptase kit (Promega). qRT-PCR Scientific RepoRtS | (2020) 10:14108 | https://doi.org/10.1038/s41598-020-70671-1 www.nature.com/scientificreports/ for RNAIII, agrA and asp23 expression were performed using the primers and conditions described in a previous publication 9 and the SYBR Green PCR Master Mix (Applied Biosystems) equipment and kits. The 16S gene was used to normalize data. The (− ΔCt) value represents the difference in threshold cycle (Ct) between the target and control(16S) genes 65 .
Osteoclastogenesis assay and TNFα detection. To determine the differential capacity of strains HU-14 and P3.1 to trigger osteoclastogenesis RAW 264.7 cells were plated on coverslips (2,5 × 104 cells/well, 24 well plate) in the presence of M-CSF 30 ng/ml (MACS Miltenyi Biotec) and 107 CFU/ml or each S. aureus strain or media alone. At 48 h after stimulation cells were fixed in 4% paraformaldehyde and stained for Tartrate-Resistant Acid Phosphatase (TRAP) (Sigma-Aldrich, St. Louis, MO, USA). Multinucleated (more than three nuclei), TRAP-positive cells were defined as osteoclasts and counted 66 . TNFα concentrations were determined quantitatively in culture supernatants collected at 24 h after stimulation by ELISA using specific antibody pairs (Beckton-Dickinson) as described previously 67 .
Statistical considerations. Multiple data statistical comparison was performed by ordinary one-way ANOVA using the Tukey´s multi comparisons test and the Bartlett-Forsythe post-test. Statistical comparison of two data groups with normal distribution was performed with the Student t test for unpaired samples. Statistical comparison of two data groups with non-normal distribution was performed with the Mann-Whitney test. P < 0.05 was considered statistically significant. Graph-Pad Prism software (GraphPad Software, Inc., La Jolla, USA; version 6.00) was used for all statistical analyses.

Data availability
The sequences analyzed in this report have been deposited in the European Nucleotide Archive (ENA) under Bioproject PRJEB35732. Other datasets generated and/or analyzed during the current study are available from the corresponding author upon request.