NudCL2 is an Hsp90 cochaperone to regulate sister chromatid cohesion by stabilizing cohesin subunits
Sister chromatid cohesion plays a key role in ensuring precise chromosome segregation during mitosis, which is mediated by the multisubunit cohesin complex. However, the molecular regulation of cohesin subunits stability remains unclear. Here, we show that NudCL2 (NudC-like protein 2) is essential for the stability of cohesin subunits by regulating Hsp90 ATPase activity in mammalian cells. Depletion of NudCL2 induces mitotic defects and premature sister chromatid separation and destabilizes cohesin subunits that interact with NudCL2. Similar defects are also observed upon inhibition of Hsp90 ATPase activity. Interestingly, ectopic expression of Hsp90 efficiently rescues the protein instability and functional deficiency of cohesin induced by NudCL2 depletion, but not vice versa. Moreover, NudCL2 not only binds to Hsp90, but also significantly modulates Hsp90 ATPase activity and promotes the chaperone function of Hsp90. Taken together, these data suggest that NudCL2 is a previously undescribed Hsp90 cochaperone to modulate sister chromatid cohesion by stabilizing cohesin subunits, providing a hitherto unrecognized mechanism that is crucial for faithful chromosome segregation during mitosis.
KeywordsCochaperone cohesin Hsp90 NudCL2 Sister chromatid cohesion
Bovine serum albumin
Breast cancer 2
Checkpoint kinase 1
Establishment of cohesion protein 1
GFP-fused histone H2B
Histone H3-associated protein kinase
Heat shock protein 90
Lissencephaly 1 gene product
Nuclear distribution gene C
NudC-like protein 2
Shugoshin 1 and protein phosphatase 2A
Small interfering RNAs
During mitosis, the proper segregation of duplicated chromosomes into daughter cells is required for maintaining genome integrity. Sister chromatid cohesion, which is mediated by the highly conserved protein complex cohesin, plays an essential role in chromosome segregation . In mammals, the cohesin complex involved in mitosis is composed of four highly conserved subunits. Two structural maintenance of chromosomes proteins (Smc1α and Smc3) form a heterodimer by interacting their hinge regions. One α-kleisin subunit protein (Rad21) interacts with the head domains of Smc1α and Smc3 and effectively closes the ring. One stromal antigen (SA1 or SA2) directly interacts with Rad21 via its C-terminal region to stabilize the cohesin ring structure [1, 2].
The cohesin complex is dynamically regulated by various factors during cell cycle progression in mammalian cells . In telophase and G1, cohesin is loaded onto chromatin by the Nipbl-Mau2 (Nipped-B like-Mau2) complex, which is counteracted by Wapl (wings-apart like) [3, 4]. During S phase, Smc3 is acetylated by Esco1 (establishment of cohesion protein 1) to enhance the cohesin-DNA interaction and promote the recruitment of Sororin to cohesin. This recruitment displaces Wapl from its binding partner Pds5 (precocious dissociation of sisters protein 5) to further stabilize the cohesin ring that entraps the sister chromatids during S/G2 phase [5, 6]. In early mitosis, phosphorylation of SA2 by Plk1 (polo-like kinase 1) and opening of the cohesin ring by Wapl result in the removal of the majority of cohesin from the chromosome arms [7, 8]. At the same time, the centromeric cohesin is protected by the Sgo1-PP2A (shugoshin 1-protein phosphatase 2A) complex and Haspin (histone H3-associated protein kinase) until metaphase [9, 10]. The cleavage of Rad21 by separase and the deacetylation of Smc3 by histone deacetylase Hdac8 at the metaphase–anaphase transition promote the complete removal of cohesin from chromatin [11, 12, 13]. This precise regulation of the cohesin complex ensures the accurate chromosome segregation in cell cycle progression. However, little is known about the regulation of cohesin subunits stability.
Heat shock protein 90 (Hsp90) is a highly conserved and abundant chaperone in eukaryotic cells, functioning as a key regulator of proteostasis by mediating the folding and maturation process of its diverse client proteins . Hsp90 has been well documented to be essential for many cellular activities, including cellular homeostasis, transcriptional regulation, chromatin remodeling, and genomic stability . Multiple components of the DNA repair machinery that regulates genomic stability are reported as Hsp90 clients . Inhibition of Hsp90 leads to the degradation of BRCA2 (breast cancer 2), Rad51, and CHK1 (checkpoint kinase 1) and results in genomic instability. However, it is unclear whether Hsp90 is required for the stability of cohesin subunits.
NudC (nuclear distribution gene C) is a highly conserved protein from yeasts to humans . In the filamentous fungus Aspergillus nidulans, NudC has been demonstrated to be an upstream regulator of NudF (a homolog of human LIS1, a key regulator of cytoplasmic dynein) . We cloned and characterized two mammalian homologs of NudC, NudCL (NudC-like protein) and NudCL2 [18, 19]. NudC and NudCL participate in many biological processes, including cell cycle progression and cell migration [18, 20, 21, 22]. Recently, our group has found that NudCL2 is able to stabilize LIS1 by enhancing the interaction between LIS1 and Hsp90 . All the NudC homologs share a conserved core domain of p23 (p23 domain), which modulates Hsp90 ATPase activity to promote the folding of client proteins, and are involved in the regulation of protein stability [18, 19, 21, 23, 24]. However, whether the members of the NudC family are the Hsp90 cochaperones is largely unknown.
In this study, we provide evidence that NudCL2 is required for stabilizing cohesin subunits and ensuring accurate sister chromatid separation during mitosis. Our data reveal that NudCL2 significantly inhibits Hsp90 ATPase activity and enhances its chaperone function. Moreover, ectopic expression of Hsp90 efficiently rescues the cohesion defects in NudCL2-depleted cells. Thus, we propose that NudCL2 acts as an Hsp90 cochaperone to maintain cohesin subunit stability, which is crucial for faithful chromosome segregation in mammalian cells.
Materials and methods
Plasmids and small interfering RNAs (siRNAs)
Human Flag-NudCL2, Flag-NudCL2* (with a silent mutation of three nucleic acids in the RNAi targeting region: ACCTTGAAAAGTGACTGCT), GST-NudCL2, His-p23 and yeast His-Hsp90 vectors were constructed previously [19, 21]. Human Hsp90α was cloned by RT-PCR and inserted into pcDNA 3.1/Myc-His C (Myc/His-tag vector, Invitrogen). Full-length NudCL2 was cloned by PCR using Flag-NudCL2 as a template and subcloned into pET-28a (His-tag vector, Novagen) and pcDNA 3.1/Myc-His C. Full-length human SMC1A, SMC3, RAD21 and STAG2 cloned by RT-PCR were inserted into pFastBac-HT A (His-tag vector, Invitrogen). All of these constructs were confirmed by DNA sequencing.
LIS1: 5′-GAACAAGCGAUGCAUGAAGTT-3′ 
Plk1: 5′-GGGCGGCUUUGCCAAGUGCTT-3′ 
STAG2: 5′-GUACGGCAAUGUCAAUAUATT-3′ 
RAD21: 5′-AUACCUUCUUGCAGACUGUTT-3′ 
SMC1A: 5′- GGAAGAAAGUAGAGACAGATT-3′ 
SMC3: 5′-UGUCAUGUUUGUACUGAUATT-3′ .
Cell culture and transfections
HeLa and HEK-293 cells were maintained in Dulbecco’s Modified Eagle’s Medium (DMEM, Corning) supplemented with 10% fetal bovine serum (Invitrogen) at 37 °C in 5% CO2. Plasmids were transfected with PolyJet (SignaGen) and the siRNA duplexes were transfected with Lipofectamine RNAiMAX (Invitrogen). The transfection processes were performed according to the manufacturer’s instructions.
Geldanamycin (GA, Tocris) was stored at − 20 °C as a stock solution of 1.78 mM in DMSO. Cells were treated with geldanamycin for the indicated times as described in the text. The final concentration of GA was 1.78 μM for HeLa cells and 0.5 μM for HEK-293 cells. For the chromosome spread experiments, 100 ng/ml of colcemid (Sigma) was used to treat cells for 2.5 h.
For immunofluorescence, antibodies against α-tubulin (Sigma, T6199), cyclin B1 (Santa Cruz, sc-70898), Bub1 (Abcam, ab900), Mad2 (Santa Cruz, sc-28261), Hec1 (Abcam, ab3613) and SMC2 (Abcam, ab10412) and human CREST autoimmune serum (Antibodies, Inc., 15-235-0001) were used. For Western blot analysis, antibodies against Rad21 (Bethyl, A300-080A), SA2 (Bethyl, A300-581A), Smc1α (Bethyl, A300-055A), Smc3 (Bethyl, A300-060A), Mau2 (Abcam, ab46906), Sgo1 (Santa Cruz, sc-54329), Plk1 (Santa Cruz, sc-53418) and Hsp90 (Abcam, ab59459) were utilized. The anti-NudCL2 antibody was generated as described previously . An antibody against actin (Sigma, T1978) was used as the loading control. The secondary antibodies for immunofluorescence analyses were Alexa Fluor 488-conjugated anti-rabbit or anti-mouse IgG and Alexa Fluor 568-conjugated anti-human IgG (Invitrogen). Goat anti-mouse or anti-rabbit secondary antibodies (LI-COR) conjugated with either Alexa Fluor 680 or IRDye 800 were used for Western blot analysis.
HeLa or HEK-293 cells grown on glass coverslips were fixed for 15 min with cold methanol (− 20 °C) and then incubated with primary antibodies for 2 h and secondary antibodies for 1 h at room temperature. DNA was stained with DAPI (Sigma).The mounted coverslips were analyzed by confocal fluorescence microscopy with an oil immersion 60 × objective (LSM510, Zeiss).
HeLa cells stably expressing GFP-H2B were treated with either siRNA for 72 h or GA for 48 h. Then, the cells were analyzed with time-lapse video microscopy to track the cell cycle progression. Fluorescence images were captured at 3-min intervals for at least 3 h with LSM510 software (Zeiss). The videos were further edited using ImageJ software (NIH).
GST pull-down assays
GST pull-down assays were performed as described previously [20, 29]. GST and GST-NudCL2 were purified from bacteria. His-tagged Smc1α, Smc3, Rad21 and SA2 were purified from insect Sf9 cells using the Bac-to-Bac baculovirus expression system (Invitrogen). To detect the association among NudCL2, cohesin subunits and Hsp90, the blots were probed with the respective antibodies as indicated in the text.
Immunoprecipitation and Western blot assays
Immunoprecipitation for endogenous proteins was performed as described previously . Briefly, whole-cell extracts were generated in TBSN buffer (20 mM Tris [pH 8.0], 150 mM NaCl, 0.5% Nonidet P-40, 5 mM EGTA, 1.5 mM EDTA, 0.5 mM Na3VO4, 20 mM p-nitrophenyl phosphate) supplemented with protease inhibitors and subjected to coimmunoprecipitation analysis with the indicated antibodies. Western blot analyses were performed with the indicated antibodies and analyzed using the LI-COR Odyssey (LI-COR) system.
Chromosome spread assays
Chromosome spread assays for Giemsa staining were prepared as previously described . In brief, cells were treated with 100 ng/ml of colcemid (Sigma) for 2.5 h, and mitotic cells were collected by shaking-off. Cells were incubated in a hypotonic solution (DMEM: H2O at a ratio of 2:3) for 5.5 min at room temperature and fixed with freshly prepared Carnoy’s solution (75% methanol, 25% acetic acid). Cells in Carnoy’s solution were dropped onto glass slides, stained with 5% Giemsa (Solarbio) and analyzed by bright field microscopy (BX51, Olympus). To perform chromosome spreads for immunofluorescence staining, HeLa cells grown on coverslips were treated with 100 ng/ml of colcemid (Sigma) for 2.5 h to enrich mitotic cells. After washing three times with PBS, the cells were incubated in 75 mM KCl hypotonic solution for 15 min at 37 °C and centrifuged for 5 min at 2000g to spread the chromosomes onto coverslips. The samples on coverslips were washed with PHEM buffer (60 mM PIPES, 25 mM HEPES, 10 mM EGTA, 2 mM MgCl2, pH 6.9), fixed for 5 min with cold methanol (− 20 °C) and processed for immunofluorescence microscopy (LSM510, Zeiss).
Hsp90 ATPase assays
Hsp90 ATPase assays were performed as described previously . Briefly, His-Hsp90 (1 µM) was incubated with His-NudCL2 (1 µM), His-p23 (1 µM) or geldanamycin (1.78 µM) at 37 °C for 20 min in reaction buffer (50 mM Tris, pH 7.4, 20 mM KCl, 6 mM MgCl2, 1 mM dithiothreitol, 0.5 mM ATP). The released inorganic phosphate was determined by measuring the absorbance at 650 nm using a Cyto Phosphate Assay BIOCHEM kit (Cytoskeleton).
The aggregation reactions of CS (Sigma) or luciferase (Promega) were carried out as described previously . Briefly, CS (0.15 μM) or luciferase (0.15 μM) was incubated alone or with BSA (0.15 μM), NudCL2 (0.15 μM) or Hsp90 (0.15 μM) at 43 °C (for CS) or 42 °C (for luciferase) for 30 min in 40 mM HEPES-KOH (pH 7.5). To monitor the kinetics of thermal aggregation, light scattering was measured at 370 nm by a DU 800 spectrophotometer (Beckman).
CS thermal inactivation assays
The enzyme inactivation assay of CS was performed as described previously [31, 32]. Briefly, CS (0.15 μM) was incubated at 43 °C in the absence or presence of IgG (1.2 μM), NudCL2 (0.6 μM) or Hsp90 (0.6 μM) in the inactivation buffer (40 mM HEPES-KOH, 0.1 mM EDTA, pH 7.5). Aliquots (100 μl) were taken at the indicated times and mixed with 650 μl of 100 mM Tris (pH 8.1), 50 μl of 3 mM acetyl-CoA (Sigma), 100 μl of 1 mM DTNB (Sigma), and 100 μl of 5 mM oxaloacetate (Sigma), and then incubated at 30 °C for 1 min to eliminate the false readings. To monitor CS activity, the readings were measured at 30 °C for 1 min with 20-s intervals at 412 nm by SpectraMax (Molecular Devices).
Luciferase refolding assays
Luciferase refolding assays were carried out as described previously . Luciferase (0.2 μM) was incubated either alone or in the presence of IgG (2 μM), NudCL2 (0.2 μM) or Hsp90 (0.2 μM) at 22 °C or 42 °C for 15 min in refolding buffer (5 mM MgCl2, 10 mM KCl, 2 mM DTT, 50 mM HEPES-KOH, pH 7.5). After incubation, 10 μl of each mixture was immediately added to a solution containing 18 μl of rabbit reticulocyte lysate (RRL) (Promega) and 2 μl of 0.1 M ATP. During the incubation at 30 °C, the luciferase enzyme activity was measured by SpectraMax (Molecular Devices) using the Luciferase Assay System (Promega) at the indicated times. The tubes were treated with 1 mg/ml BSA for 15 min to prevent luciferase adsorption to the walls.
Data are representative of at least three independent experiments. The mean and standard deviations (SD) were calculated for all quantitative experiments. The Student’s t test was used to determine statistically significant differences between groups.
Depletion of NudCL2 induces chromosome misalignment
To determine whether the misalignment of chromosomes in NudCL2-depleted cells is due to defects in chromosome congression during prometaphase or chromosome segregation at anaphase, we examined the expression and localization of cyclin B1 and the spindle checkpoint proteins Mad2 and Bub1 in these cells. Approximately, 90% of NudCL2-depleted mitotic cells with misaligned chromosomes were positive for cyclin B1 throughout the cell and were also positive for Mad2 and Bub1 at kinetochores (Fig. 1e–m). Furthermore, live-cell imaging analysis of HeLa cells stably expressing GFP-H2B (GFP-fused histone H2B) revealed that chromosomes failed to align at the metaphase plate in cells depleted of NudCL2 (Fig. 1n, Supplementary Movies. 1, 2). These results strongly indicate that NudCL2 depletion causes defects in chromosome congression during prometaphase, resulting in the misalignment of chromosomes.
NudCL2 depletion causes premature sister chromatid separation
Given that LIS1 is crucial for mitosis and stabilized by NudCL2 [19, 25], we examined whether the role of NudCL2 in sister chromatid cohesion is mediated by LIS1. Chromosome spreads with Giemsa staining clearly revealed that knockdown of LIS1 had no significant effect on sister chromatid cohesion (Supplementary Fig. 6), suggesting that the defects in sister chromatid cohesion induced by NudCL2 depletion are not due to LIS1 downregulation.
Depletion of NudCL2 destabilizes cohesin subunits
Inhibition of Hsp90 leads to precocious sister chromatid separation
NudCL2 regulates cohesin subunit stability via Hsp90
NudCL2 interacts with Hsp90 and cohesin subunits
NudCL2 functions as an Hsp90 cochaperone
Both CS and luciferase enzymes are active in their native state and lose activity upon unfolding or denaturation and this property has been utilized to study the chaperoning function of Hsp90 influenced by its cochaperones [36, 37]. To confirm whether NudCL2 acts as an Hsp90 cochaperone to regulate the enzymatic activity of Hsp90 client proteins, we carried out CS thermal inactivation and luciferase refolding assays in vitro. The results showed that upon Hsp90, NudCL2 not only significantly reduced the heat-induced CS inactivation (Fig. 7e), but also promoted the luciferase activity recovery after heat-induced inactivation (Fig. 7f). Interestingly, in the absence of Hsp90, NudCL2 did not suppress heat-induced CS inactivation or enhance the recovery of luciferase activity after heat-induced inactivation (Fig. 7e, f). Taken all together, these data indicate that NudCL2 may function as an Hsp90 cochaperone to enhance the stability and function of client proteins.
The Hsp90 cochaperone p23 plays a critical role in the stabilization of client proteins, including steroid hormone receptors and telomerase, by inhibiting Hsp90 ATPase activity and stabilizing the ATP-induced conformational change of Hsp90 [35, 38, 39]. Knockdown of p23 destabilizes a number of Hsp90 client proteins [40, 41]. Here, we provide evidence that NudCL2 interacts with Hsp90 and suppresses the ATPase activity of Hsp90 (Figs. 6a–c, 7a, b). Further results reveal that NudCL2 inhibits the heat-induced aggregation and enzymatic inactivation of Hsp90 substrates (Fig. 7c–f). Moreover, we also show that either NudCL2 depletion or Hsp90 inhibition by GA decreases the stability of cohesin subunits and induces premature sister chromatid separation (Figs. 2, 3, 4). Ectopic expression of Hsp90 efficiently rescues the defects induced by NudCL2 depletion, but not vice versa (Fig. 5). These results indicate that the p23 domain-containing protein NudCL2 is an Hsp90 cochaperone to promote its chaperone function and stabilize cohesin subunits (Fig. 7g).
In this study, we find that NudCL2 knockdown and GA’s inhibition of Hsp90’s ATPase cause similar defects (Figs. 2, 3, 4). GA is a natural compound which competes with ATP to bind with high affinity into the ATP binding pocket of Hsp90 and thereby inhibits the ATPase activity of Hsp90 . When cells are treated with GA, Hsp90 loses the ability to fold and activate its client proteins, which results in the degradation of client proteins , including cohesin subunits (Fig. 4). Therefore, although NudCL2 functions as a cochaperone to inhibit Hsp90’s ATPase activity, its knockdown and GA treatment induce similar effects.
NudC was first identified in Aspergillus nidulans as a key upstream regulator of the LIS1/dynein complex [17, 44, 45]. Our group and others have found that vertebrate NudC has three principal homologs, NudC, NudCL and NudCL2, all of which contain the conserved p23 domain [18, 19, 21, 23, 24]. NudC is not only able to interact with Hsp90 and regulate LIS1 stability via the Hsp90 pathway, but also stabilizes cofilin 1 to regulate ciliogenesis in an Hsp90-independent manner [21, 24]. Additionally, NudCL has been documented to play an important role in the stability regulation of the dynein intermediate chain . Recently, our data have shown that NudCL2 promotes LIS1 stability by enhancing the interaction between Hsp90 and LIS1 . In this report, we further find that NudCL2 acts as an Hsp90 cochaperone to stabilize cohesin subunits by modulating the ATPase activity of Hsp90. Intriguingly, NudC and NudCL, but not NudCL2, possess an intrinsic chaperoning activity independent of Hsp90 to regulate the stability of client proteins [18, 23, 24]. Taken together, these data imply that the members of NudC family may act as a new type of regulator of protein stability possibly via the Hsp90-dependent or -independent pathway.
Emerging studies have demonstrated the molecular regulation underlying the stability of the cohesin-DNA interaction [4, 5, 6, 7, 11]. However, how the stability of cohesin subunits is regulated remains unclear. Recent studies have shown that Smc3 knockdown leads to Smc1α instability, but Smc1α depletion has no significant effect on Smc3 stability . Downregulation of either Smc1α or Smc3 causes Rad21 degradation, while depletion of Rad21 only decreases SA protein levels [46, 47]. Here, our data reveal that knockdown of either Smc3 or Smc1α reduces the stability of the other three cohesin subunits (Supplementary Fig. 9). Rad21 depletion decreases the protein levels of SA2, but not Smc1α or Smc3. SA2 depletion does not affect the stability of the other three subunits. These results indicate that either Smc1α or Smc3 is essential for the stability of Rad21 and SA2, but not vice versa. Moreover, we show that depletion of NudCL2 decreases the protein levels of all cohesin subunits (Fig. 3). Purified GST-NudCL2 directly interacts with His-tagged Smc1α and Smc3, but not Rad21 or SA2 (Fig. 6d–g). Together, these data suggest that NudCL2 may promote the stability of cohesin subunits by directly interacting with Smc1α and Smc3.
In summary, this study provides evidence that NudCL2 is essential for sister chromatid cohesion by acting as an Hsp90 cochaperone to stabilize cohesin subunits. Depletion of NudCL2 causes cohesin subunits degradation and results in premature sister chromatid segregation during mitosis. This work reveals a previously uncharacterized mechanism for the regulation of chromosome segregation in mammalian cells.
We are grateful to Dangsheng Li and Xueliang Zhu for helpful suggestions, to Lynn W. Enquist for his comments on the manuscript, and to Dante Neculai for his careful language correction. We thank the members of Zhou’s lab for the helpful comments and suggestions during the work. This work was supported by the National Natural Scientific Foundation of China (31671394, 31471259, 31701214, 91740205, 31620103911, 31571446 and 31771540) and the National Key Research and Development Program of China (2016YFA0100301).
YY, WW, and ML designed and performed the majority of the experiments. WZ provided technical assistance for the aggregation assays. YH provided assistance for GST pull-down assays. YG, WZ, XY, WL, and FW discussed the results and commented on the manuscript. YY and WW wrote the original draft. DC and TZ edited the manuscripts. TZ supervised the project.
Compliance with ethical standards
Conflict of interest
The authors declare that they have no conflict of interest.
- 1.Nasmyth K, Haering CH (2009) Cohesin: its roles and mechanisms. Annu Rev Genet 43:525–558. https://doi.org/10.1146/annurev-genet-102108-134233 CrossRefGoogle Scholar
- 5.Zhang J, Shi X, Li Y, Kim BJ, Jia J, Huang Z, Yang T, Fu X, Jung SY, Wang Y, Zhang P, Kim ST, Pan X, Qin J (2008) Acetylation of Smc3 by Eco1 is required for S phase sister chromatid cohesion in both human and yeast. Mol Cell 31(1):143–151. https://doi.org/10.1016/j.molcel.2008.06.006 CrossRefGoogle Scholar
- 29.Lu Y, Xie S, Zhang W, Zhang C, Gao C, Sun Q, Cai Y, Xu Z, Xiao M, Xu Y, Huang X, Wu X, Liu W, Wang F, Kang Y, Zhou T (2017) Twa1/Gid8 is a beta-catenin nuclear retention factor in Wnt signaling and colorectal tumorigenesis. Cell Res 27(12):1422–1440. https://doi.org/10.1038/cr.2017.107 CrossRefGoogle Scholar
- 30.Gimenez-Abian JF, Diaz-Martinez LA, Wirth KG, Andrews CA, Gimenez-Martin G, Clarke DJ (2005) Regulated separation of sister centromeres depends on the spindle assembly checkpoint but not on the anaphase promoting complex/cyclosome. Cell Cycle 4(11):1561–1575. https://doi.org/10.4161/cc.4.11.2146 CrossRefGoogle Scholar
- 38.Dittmar KD, Demady DR, Stancato LF, Krishna P, Pratt WB (1997) Folding of the glucocorticoid receptor by the heat shock protein (hsp) 90-based chaperone machinery. The role of p23 is to stabilize receptor.hsp90 heterocomplexes formed by hsp90.p60.hsp70. J Biol Chem 272(34):21213–21220CrossRefGoogle Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.