Human Umbilical Cord Mesenchymal Stem Cells in Combination with Hyaluronic Acid Ameliorate the Progression of Knee Osteoarthritis

The aim of this study is to evaluate the feasibility and usefulness of the human umbilical cord mesenchymal stem cells (hUC-MSCs) and hyaluronan acid (HA) combination to attenuate osteoarthritis progression in the knee while simultaneously providing some insights on the mitigation mechanism. In vitro, the effect of hUC-MSCs with HA treatment on chondrocyte cell viability and the cytokine profile were analyzed. Additionally, the antioxidation capability of hUC-MSCs-CM (conditioned medium) with HA towards H2O2-induced chondrocyte cell damage was evaluated. The HA addition increased the hUC-MSC antioxidation capability and cytokine secretion, such as Dickkopf-related protein 1 (DKK-1) and hepatocyte growth factor (HGF), while no adverse effect on the cell viability was observed. In vivo, the intra-articular injection of hUC-MSCs with HA to a mono-iodoacetate (MIA)-induced knee osteoarthritis (KOA) rat model was performed and investigated. Attenuation of the KOA progression in the MIA-damaged rat model was seen best in hUC-MSCs with a HA combination compared to the vehicle control or each individual element. Combining hUC-MSCs and HA resulted in a synergistic effect, such as increasing the cell therapeutic capability while incurring no observable adverse effects. Therefore, this combinatorial therapy is feasible and has promising potential to ameliorate KOA progression.


Introduction
Osteoarthritis (OA) is a debilitating progressive disease affecting the joints that could be divided into two types: primary and secondary. Both types involved the degeneration of cartilage in joints; primary OA is mostly caused by wear and tear on joints that becomes increasingly apparent as people age, while specific triggers resulting in cartilage breakdown cause secondary OA to manifest, such as genetics [1], obesity [2], inflammation [3], etc.
Knee osteoarthritis (KOA) has a higher incidence rate compared to other OA types, which marks its importance [4]. A recent study in Belgium, conducted over a period of 20 years (1996-2015), showed that there was an increasing prevalence of KOA, which called for the need to prioritize improvement towards KOA care [5]. As for today, there is no disease-ameliorating agent that could cure KOA, and the current treatment would only slow down the disease progression and other inflammatory conditions [6].
Hyaluronic acid (HA) is a compound that is naturally produced by the body and one of the main components of synovial fluid, responsible for reducing friction during synovial joints movement [7]. Intra-articular (IA) HA injections yield positive outcomes. Therefore, the administration of HA to KOA patients is recommended when other pharmacologic treatments, such as glucocorticoid injections, fail to show results [8].
Mesenchymal stem cells (MSCs) therapy for KOA has been increasingly considered due to its unique properties, such as inflammation site homing, immune suppression, self-renewal and the restoration of cartilage tissue, making it a promising candidate for a gradual KOA treatment [9]. MSCs are multipotent stromal cells that could perform multiple functions due to their capacity to differentiate into specific cell types, such as the abundant secretion of soluble growth factors and cytokines. Moreover, MSCs can migrate to injury and inflammation sites plus exert beneficial effects through immunomodulatory, antifibrosis, antioxidation and anti-apoptosis effects [10][11][12].
Combining MSCs with HA has been shown to result in synergistic effects that are beneficial for treating KOA. Outcomes like an increase in the osteogenic differentiation capability [13], improved inflammation site homing [14] and enhanced cartilage repair [15] have been reported.
The aim of this paper was to investigate the effect of MSCs-particularly, human umbilical cord-derived MSCs (hUC-MSCs) and HA combinatorial treatment for KOA. Furthermore, we tried to examine whether HA affected the expression of cytokines in hUC-MSCs and if that, in turn, had beneficial impacts on the treatment function.

hUC-MSC Isolation and Identification
Umbilical cord tissue was obtained from individual mothers that had signed informed written consent, which was approved by the Institutional Review Board of National Cheng Kung University Hospital (IRB No. A-BR-104-045) and conducted in accordance with the Helsinki Declaration. Human umbilical cord mesenchymal stem cells were isolated and identified from umbilical cord tissue as previously described [16,17]. Briefly, umbilical cord tissue was cut to small pieces ranging 1 to 2 mm 3 , then, 2-mg/mL Collagenase NB6 (SERVA) was used for digestion at 37 • C for 120 min while shaking. Collected tissue samples were centrifuged, and the cell pellet was resuspended in minimal essential medium α (MEMα, Thermo Fisher Scientific, Waltham, MA, USA) containing 18% fetal bovine serum (FBS; Thermo Fisher Scientific), 4-ng/mL fibroblast growth factor basic (bFGF; Thermo Fisher Scientific) and 50-mg/mL gentamicin (Thermo Fisher Scientific) at 37 • C, 5% CO 2 with saturated humidity. To remove nonadherent cells from plastic-adherent colonies, after seven days, cultures were washed with PBS three to five times and then further cultured up to eleven days, with medium changes every 3 days. According to the cell growth state, e.g., large colony formation, cells were passaged. Finally, hUC-MSCs were maintained in MEMα supplemented with 10% FBS (Thermo Fisher Scientific) and 4-ng/mL bFGF (Thermo Fisher Scientific), 5% CO 2 with saturated humidity, passaged every time it reached approximately 70-90% confluence. hUC-MSCs were characterized by flow cytometry to analyze the CD surface marker expressions (CD44, CD73, CD90, CD105, CD11b, CD19, CD34, CD45 and HLA-DR; BD Stemflow hMSC Analysis Kit; BD Biosciences, San Jose, CA, USA). Additionally, hUC-MSCs were also examined for their trilineage differentiation capability (osteocytes, chondrocytes and adipocytes), yielding positive results.

Hyaluronic Acid (HA) Preparation
HA was prepared by mixing high molecular weight HA (1700-2500K Da, Bloomage Freda Biopharma, Jinan, China) and low molecular weight HA (850-1300K Da, Bloomage Freda Biopharma) with a ratio of 1:2. The mixed HA solution was further diluted using MEMα basal medium (Thermo Fisher Scientific) to reach the desired concentration for each experiment, respectively.

Cell Viability Analysis of Human Umbilical Cord Mesenchymal Stem Cells (hUC-MSCs) with HA Treatment
Briefly, hUC-MSCs were seeded at a density of 16,000 cells/cm 2 in complete culture medium consisting of MEMα (Thermo Fisher Scientific), 10% fetal bovine serum (FBS; Thermo Fisher Scientific) and 4-ng/mL fibroblast growth factors-basic (bFGF; Thermo Fisher Scientific) for 24 h. The cells were subsequently cultured in a humidified incubator with 5% CO 2 at 37 • C. After 24 h, the culture medium was replaced with MEMα in the presence or absence of HA (range from 0% to 0.5%) and cultured for another 48 h. Then, cells were detached with TrypLE (Thermo Fisher Scientific), labeled with trypan blue and counted as the number of viable cells. Harvested cells and cell culture supernatants were stored at −80 • for further analysis.

Cytokine Array Analysis
A semi-quantitative evaluation of the cytokines, chemokines and other soluble proteins in the cell culture supernatant was performed by using a Proteome Profiler Human XL Cytokine Array Kit (R&D Systems, Minneapolis, MN, USA). hUC-MSCs were seeded at a density of 16,000 cells/cm 2 in a T75 flask in the presence or absence of 0.1% HA. After 48 h of culturing, cell culture supernatants were collected for a cytokine array analysis. Briefly, array membranes containing 105 captured antibodies were incubated overnight at 4 • C with cell culture supernatants. Following incubation with detection antibodies, Streptavidin-HRP, chemiluminescent detection procedures were performed according to the manufacturer's instructions. Signals produced by the chemiluminescent reaction were detected and analyzed using the ChemiDoc-It™ Imaging System (UVP LLC, Upland, CA, USA).

ELISA Analysis
hUC-MSCs were cultured at a density of 16,000 cells/cm 2 in a T75 flask with complete culture medium for 24 h. Cells were then treated with various concentrations of HA (range from 0.025% to 0.5%) and incubated for another 48 h. Supernatants were then collected for Dickkopf-related protein 1 (DKK-1) and hepatocyte growth factor (HGF) quantification by specific enzyme-linked immunosorbent assays (ELISAs; R&D Systems). The cytokine concentrations were normalized to the relative viable cell number.

Mono-Iodoacetate (MIA)-Induced KOA Model
Wistar rats (male, 8 weeks old) obtained from BioLASCO Taiwan Co. Ltd. (Taipei, Taiwan) were used in this study. Rats were housed in standard environmental conditions (1 to 2 animals/cage) with a 12-h light/dark cycle at 21 ± 2 • C and 50 ± 15% humidity and supplied with a normal diet and sterilized water ad libitum. MIA (Sigma) was used to induce unilateral osteoarthritis in rats. Under 2.5-3% isoflurane (Panion & BF biotech, Taipei, Taiwan) anesthesia, 80 mg/kg and 40 mg/kg of MIA were delivered into the articular cavity (through infrapatellar ligament) of the right knee on week 0 and week 1, respectively. MIA was dissolved in 50 µL of 0.9% normal saline and administered using 26-G needles (Terumo, Tokyo, Japan).
Twenty-seven rats were randomly divided into four groups: Vehicle control (n = 7), 1.5% HA (n = 6), hUC-MSCs (n = 7) and hUC-MSCs combined with HA (n = 7). On week 3, a single dose of 0.75% HA (Bloomage Freda Biopharma) or hUC-MSCs (2.5 × 10 6 cells/knee joint) in 100 µL of 0.9% normal saline was injected into the knee joint by intra-articular injection. For hUC-MSCs combined with HA groups, 50 µL of 2.5 × 10 6 hUC-MSCs and 50 µL of 1.5% HA were mixed before administration. Vehicle control rats were injected with 100 µL of 0.9% normal saline under the same treatment conditions. All animal experiments were approved by the Institutional Animal Care and Use Committee of Taipei Medical University (IACUC Approval No. LAC-2018-0426).

Histopathological Analysis
After experimental study completion on week 19, all rats were euthanized by carbon dioxide inhalation. The right knee tissues were collected and fixed in 10% (m/v) neutral buffered formalin (NBF) for histological investigation. The tissues were decalcified, trimmed, embedded in paraffin, sectioned and stained with hematoxylin and eosin (H&E). The prepared slides were examined microscopically. Damage of the articular cartilage was evaluated using an Osteoarthritis Research Society International (OARSI) score on a scale of 0-24 points. The histopathological analysis was performed by a veterinarian pathologist of BioLASCO Taiwan Co. Ltd. who was familiar with cartilage histopathology and the OARSI scoring system in a blinded manner. Osteoarthritis synovial fluid (OA SF) was drawn aseptically from KOA patients undergoing total knee replacement (TKR) surgery. Synovial fluid was collected and centrifuged to remove cellular debris, and supernatants were aliquoted and stored at −80 • C until use. hUC-MSCs were cultured at a density of 16,000 cells/cm 2 in a T75 flask with a complete culture medium for 24 h, replaced with MEMα basal medium containing 20% OA SF or MEMα basal medium containing 20% OA SF and 0.1% HA and then incubated for another 48 h. Media was collected and centrifuged at 1000 g for 10 min at 4 • C. Supernatants were aliquoted and stored at −80 • C until the cytokine assay analysis.

Statistical Analysis
Data were presented as mean ± SD (standard deviation). Using GraphPad Prism 6.0. Software (Version 6.07, GraphPad, San Diego, CA, USA, 2015), data were analyzed with a Student's t-test and one-way ANOVA; the differences were considered to be statistically significant at p < 0.05.

Different Concentrations of HA Had no Adverse Effects on the Viability of hUC-MSCs
In order to determine the effects of the HA treatment on hUC-MSCs survival, hUC-MSCs were cultured for 48 h with increasing concentrations of HA, ranging from 0.025% to 0.5%. As shown in Figure 1, all HA concentration treatments did not induce any significant changes in the cell viability, even at the highest concentration. These results indicated that the direct interaction of hUC-MSCs with HA would not affect the viability of the hUC-MSCs. It is noteworthy that an increasing concentration of HA (0.025-0.5%) did not seem to affect the cell growth and proliferation in our study.

Cytokine Profiling of hUC-MSCs Treated with Various Concentrations of HA
A multitude of cytokines and chemokines secreted by hUC-MSCs are known to have critical roles in biological processes such as immunomodulation, vascular stabilization, angiogenesis and tissue repairs. We profiled the cytokine expression of the supernatant from hUC-MSCs treated with or without 0.1% HA for 48 h. Based on Figure 2A,B, the cytokine secretion patterns of hUC-MSCs and hUC-MSCs treated with 0.1% HA were similar. Twenty-two out of 105 cytokines or chemokines were detected in both groups. Among these factors, eight are known for their roles in cell growth and proliferation, such as hepatocyte growth factor (HGF) and growth differentiation factor-15 (GDF-15); twelve for immunomodulation, such as interleukin-6 (IL-6), monocyte chemoattractant protein-1 (MCP-1) and Serpin Family E Member 1 (Serpin E1) and two for anti-apoptosis, such as urokinase-type plasminogen activator receptor (uPAR) ( Figure 2C).    However, it was noted that, in the 0.1% HA-treated group, the expression of Chitinase-3-like protein 1 (CHI3L1), DKK-1, HGF, interleukin-8 (IL-8), interleukin-17A (IL-17A), Osteopontin (OPN) and uPAR were elevated higher than 5% ( Figure 2C). Among the upregulated protein, DKK-1 and HGF were known to ameliorate osteoarthritis (OA) pathogenesis. DKK-1 exerts a protective effect on its capacity to inhibit the Wnt-mediated expression of catabolic factors, such as matrix metalloproteinase 13 (MMP-13), and inhibit OA cartilage destruction [18,19]. It has been noted that HGF stimulates the proliferation, motility and proteoglycan synthesis of chondrocytes and is involved in the repair mechanism of matrix damage [12,20,21]. Taken together, DKK-1 and HGF might serve as therapeutic targets in osteoarthritis.
The results from the single cytokine array experiment showed that the expression of DKK-1 and HGF were elevated in hUC-MSC treatments with 0.1% HA for 48 h ( Figure 2C). To investigate whether the cytokine secretion of the hUC-MSCs was dependent on the concentration of HA, we tested six different supernatants from the culture of hUC-MSCs with 0, 0.025, 0.05, 0.1, 0.2 and 0.5% of HA.
Next, the amount of DKK-1 and HGF were quantified by a specific ELISA to validate our single cytokine array result. Our data demonstrated that HA treatment significantly induced the expression of DKK-1 secreted by hUC-MSCs. DKK-1 expression was increased in the 0.1%, 0.2% and 0.5% HA groups but not in the 0.05% and 0.025% HA groups. Interestingly, the secretion of DKK-1 was elevated in a dose-dependent manner in the 0.1%, 0.2% and 0.5% HA groups ( Figure 2D). Subsequently, we sought to examine the expression of HGF under HA treatment. It could be noted that the expression of HGF was only increased in the 0.1% HA group ( Figure 2E). These results were similar to the cytokine array results, in which hUC-MSCs cultured under the 0.1% HA condition increased the expression of DKK-1 and HGF. Notably, both cytokines have been shown to regulate the pathogenesis of osteoarthritis [12,[18][19][20][21]. The treatment of 0.1% HA and above seems to be able to induce the hUC-MSC secretion of beneficial cytokines.

Cytokine Profiling of hUC-MSCs Cocultured with Osteoarthritis Synovial Fluid and Hyaluronic Acid
In order to gain insights on how hUC-MSCs and hUC-MSCs with a HA combination might behave in vivo, we profiled the cytokine expressions of OA patient's synovial fluid (OA SF) only or OA SF with hUC-MSCs or OA SF with hUC-MSCs and 0.1% HA ( Figure 3A). The densitometric analysis results showed six cytokines that had their expression elevated above 5% in the OA SF + hUC-MSCs + HA compared to OA SF + hUC-MSCs or OA SF-only group; among these were DKK1, Apolipoprotein A1 (ApoA1), CHI3L1, CD105, growth differentiation factor 15 (GDF-15) and complement component 5a (C5/C5a) ( Figure 3B). We found that, compared to OA SF, OA SF with hUC-MSCs had higher expressions of cytokines (>5%) that were beneficial towards OA amelioration, such as DKK-1 and GDF-15 expression elevated 41% and 13%, respectively. Meanwhile, in OA SF with hUC-MSCs and HA media, the DKK-1 and GDF-15 expressions were increased further than the OA SF group by 59% and 19%, respectively, along with a 7% upregulated expression of ApoA1 ( Figure 3C). These results indicated that hUC-MSCs could lessen KOA deterioration through paracrine mechanisms, which were further boosted by the presence of HA.

HA Enhanced Antioxidation Capacity of MSC-CM on Human Chondrocyte Cells
To assess the rescue effect of MSC-CM on damaged chondrocytes caused by oxidative stress, the H 2 O 2 -induced chondrocyte cell damage was evaluated. In cultured chondrocyte cells, H 2 O 2 exposure induced cell death. As shown in Figure 4, H 2 O 2 -induced cell death was quantified as a fold of relative cell recovery; the MSC-CM treatment increased the cell viability by 1.13-fold, while MSC-CM with 0.1% or 0.2% HA further improved the rescue capability to 1.22-or 1.32-fold, respectively. On the other hand, HA-only treatments did not show any significant improvements in the cell viability. This result showed that hUC-MSCs combined with HA could rescue the oxidation stress induced by H 2 O 2 through the paracrine mechanism.

hUC-MSCs-HA Combinatorial Treatment Reduced Cartilage Damage in KOA Rats
To investigate the therapeutic effects of the hUC-MSCs and HA treatments in KOA, we established MIA-induced KOA rat models as shown in Figure 5A. Briefly, eight-week-old male Wistar rats were injected with MIA to induce cartilage damage. A knee edema was observed on the right-hind knee of each rat after MIA injection, which indicated that an inflammatory response had occurred within the affected joint. At week 3, randomly divided rats were treated with a single dose of HA, hUC-MSCs, hUC-MSCs combined with HA or 0.9% normal saline (vehicle control). For the hUC-MSC treatments, a dose level of 2.5 × 10 6 cells/knee joint was used in this study. All rats were sacrificed at 16 weeks after treatment, and the knee joints were processed for a histological analysis and OARSI scoring. As shown in Figure 5B, H&E staining revealed cartilage damage and structural degradation in the vehicle group. The cartilage tissue was denuded and broken down to shatter and almost eroded into subchondral bone (SB). These results together suggested that an injection of MIA successfully induced OA. In contrast, the hUC-MSC and HA treatments respectively reduced the cartilage damage significantly caused by MIA. In the hUC-MSC-HA combinatorial treatment, the degree of cartilage damage was significantly reduced, where the superficial and mid and deep zones were not damaged compared to the vehicle control, hUC-MSC-alone and the HA-alone groups. The OARSI scoring system, with a scale ranging from 0 (normal architecture) to 24 (over 50% joint involvement with deformation), was used to evaluate the extent of MIA-induced articular cartilage degeneration in this study.
The highest degree of cartilage damage was evident in the vehicle group, with a mean OARSI score of 12.14 ± 2.04, which was significantly decreased when compared to the hUC-MSC-only group (7.14 ± 2.97), HA-only group (9.50 ± 2.26) and hUC-MSC + HA group (4.43 ± 1.81). It was noted that the hUC-MSC + HA group reduced the articular cartilage degeneration, which, induced by MIA, was superior to the hUC-MSCor HA-only groups ( Figure 5C). Taken together, our results suggested that all the treatment groups successfully attenuated the articular cartilage damage induced by MIA, whereas the combination of hUC-MSCs and HA exhibited the highest potential therapeutic benefits in the MIA-induced knee OA animal model.

Discussion
Through the MIA-induced OA animal model conducted in this study, we observed that a combination of hUC-MSCs and HA could synergistically mitigate the cartilage degradation process better when compared to each individual administration. Where hUC-MSCs exposed to 0.1% HA increased the secretion of DKK-1, HGF and a higher antioxidation capability by might provide an insight on the mechanism of this event. Moreover, the HA addition to hUC-MSCs did not result in any adverse effect, such as a reduction in cell viability.
In addition to serving its role as a vehicle during MSC administration, several studies reported that HA provides beneficial effects by enhancing the MSC chondrogenic differentiation ability [22] and delaying cellular senescence [23], which might lead to better attenuation capability.
Here, we showed that 0.1% HA could not only increase DKK-1 and HGF secretion from hUC-MSCs ( Figure 2D,E) but also reduce the effect of H 2 O 2 oxidative stress towards human chondrocyte viability ( Figure 4). Interestingly, in vitro studies using 0.1% HA also showed beneficial effects toward human articular chondrocyte proliferation and viability [24] due to the binding of HA with CD44, a cell surface receptor expressed in chondrocytes, resulting in the induction of c-myc and TGF-β mRNA expression, as indicated by Ishida et al. [25]. hUC-MSCs were also identified to express high levels of CD44 on the cell surface [26], which indicates a possibility of a similar HA-binding mechanism as chondrocytes but needs to be proven with further studies.
The binding of HA with the hUC-MSC CD44 surface marker triggers various signaling cascades. One event resulted in an increase of GSK3β expression, which, in turn, induced β-catenin expression and its downstream target. β-catenin was reported to be able to translocate into the cell nucleus and, in concert with the coactivators BCL9, Pygo and CBP/p300, regulated Wnt target genes, such as DKK-1, c-Myc, cyclin D1 and Axin2 but not HGF [27][28][29][30][31]. In our study, the HGF expression did not follow a dose-dependent manner to increase the HA concentration, which might suggest that only a specific concentration of HA could increase the HGF expression in hUC-MSCs.
A KOA patient might have aberrant synovial fluid (SF) biochemical expression due to a loss of normal joint homeostasis [32]. A reduced expression of DKK-1 in the plasma and SF of the KOA patient was inversely correlated with the disease severity [19,33], while the GDF-15 [34] and ApoA1 [35] expressions were also found to be impaired in the os-teoarthritic cartilage and chondrocytes, respectively. DKK-1 is recognized as an inhibitor of canonical Wnt signaling [36], and the addition of DKK-1 to OA SF reduced the degradation of the cartilage oligomeric matrix protein (COMP) by decreasing the ADAMTS-7 expression, a metalloproteinase, through blocking Wnt signaling in the cartilage explant [37]. DKK-1 also elicits protective effects to inhibit H 2 O 2 -induced oxidative stress and DNA damage via Wnt/ β-catenin pathway blockage [38]; this observation might provide some explanation for the results observed shown in Figure 4, where MSC-CM could reduce the damage caused by H 2 O 2 . ApoA1 is a protein that has an important role in the metabolism of lipoproteins [39]. Osteoarthritis is associated with irregular lipid metabolism, such as the intracellular accumulation of lipid in chondrocytes, and a study done by Tsezou et al. demonstrated that restoring ApoA1 promotes a chondrocytes cholesterol efflux [35]. Therefore, the therapy targeting restoration of DKK-1 and ApoA1 might be beneficial for the OA patient. On the other hand, a lower-than-normal GDF-15 expression in OA SF suggests it could be used as a biomarker, although more sample studies should be conducted. A critical regulatory system for cartilage maintenance is the β-catenin-mediated canonical Wnt signaling pathway. This pathway is important during cartilage development, and its activation in adult cartilage tissue leads to hypertrophy, the initiation of calcification and tissue degradation via an increased expression of ECM-degrading components.
Several studies investigating the MSC and HA ability to repair cartilage defects were done in the following animal models: intra-articular administration of MSC with HA has been proven to be beneficial in a spontaneous OA Hartley strain guinea pig model through articular cartilage regeneration [40]. Meanwhile, another study conducted using an ACL rat model suggested that multiple injections might be needed due to the temporary effect of MSC with HA therapy, as, after 6 weeks, a significantly better International Cartilage Repair Score was evident when compared to the control group, but none was observed after 12 weeks [41]. MSC and HA combination therapy on an induced chondral defect porcine animal model showed improved cartilage healing both histologically and morphologically at 6 and 12 weeks compared to the control groups [42]. The cartilage repair effect of the MSC and HA combination therapy observed from these studies was similar to the results shown in our study.
Healthy knee synovial fluid contains HA with various molecular weights (MW) [43], which suggests that administration using a combination of high and low MW HA in the same manner might provide better results by mimicking the normal physiological properties. Some studies on KOA treatment showed that combined high and low MW HAs resulted in greater improvements in patient conditions compared to either high or low MW HA-only treatments [44,45]. Therefore, a high and low MW HA combination was used in this study. More studies need to be conducted to provide evidence on whether there is a combined high and low MW HA superior efficacy compared to high or low MW HA-only for OA attenuation, which can also be applied to hUC-MSC therapy.  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data that support the findings of this study are available from the corresponding author, [W.L.], upon reasonable request.

Conflicts of Interest:
The authors declare no conflict of interest.