Recombinant Escherichia coli produces tailor-made biopolyester granules for applications in fluorescence activated cell sorting: functional display of the mouse interleukin-2 and myelin oligodendrocyte glycoprotein
© Bäckström et al; licensee BioMed Central Ltd. 2007
Received: 20 July 2006
Accepted: 04 January 2007
Published: 04 January 2007
Fluorescence activated cell sorting (FACS) is a powerful technique for the qualitative and quantitative detection of biomolecules used widely in both basic research and clinical diagnostic applications. Beads displaying a specific antigen are used to bind antibodies which are then fluorescently labelled using secondary antibodies. As the individual suspension bead passes through the sensing region of the FACS machine, fluorescent signals are acquired and analysed. Currently, antigens are tediously purified and chemically cross-linked to preformed beads. Purification and coupling of proteins often renders them inactive and they will not be displayed in its native configuration. As an alternative, we genetically engineered Escherichia coli to produce biopolyester (polyhdroxyalkanoate=PHA) granules displaying diagnostically relevant antigens in their native conformation and suitable for FACS analysis.
Hybrid genes were constructed, which encode either the mouse interleukin-2 (IL2) or the myelin oligodendrocyte glycoprotein (MOG) fused via an enterokinase site providing linker region to the C terminus of the PHA granule associated protein PhaP, respectively. The hybrid genes were expressed in PHA-accumulating recombinant E. coli. MOG and IL2 fusion proteins were abundantly attached to PHA granules and were identified by MALDI-TOF/MS analysis and N terminal sequencing. A more abundant second fusion protein of either MOG or IL2 resulted from an additional N terminal fusion, which did surprisingly not interfere with attachment to PHA granule. PHA granules displaying either IL2 or MOG were used for FACS using monoclonal anti-IL2 or anti-MOG antibodies conjugated to a fluorescent dye. FACS analysis showed significant and specific binding of respective antibodies. Enterokinase treatment of IL2 displaying PHA granules enabled removal of IL2 as monitored by FACS analysis. Mice were immunized with either MOG or OVA (ovalbumin) and the respective sera were analysed using MOG-displaying PHA granules and FACS analysis showing a specific and sensitive detection of antigen-specific antibodies within a wide dynamic range.
E. coli can be genetically engineered to produce PHA granules displaying correctly folded eukaryotic proteins and which can be applied as beads in FACS based diagnostics. Since PHA granule formation and protein attachment occurs in one step already inside the bacterial cell, microbial production could be a cheap and efficient alternative to commercial beads.
Polyhydroxyalkanoate (PHA) granules (biopolyester particles) are formed inside bacterial cells based on the activity and biochemical properties of the PHA synthases and specific biosynthesis enzymes which are involved in PHA precursor supply [1, 2]. Biologically, PHA serves as a reserve material. The PHA granule core is composed of PHA and the surface of a phospholipid membrane with embedded or attached proteins. Amphipathic phasin proteins are one group of proteins specifically and hydrophobically interacting with the PHA core (for review see [1, 2]). The functional role of the phasins impacting on PHA granule structure has been studied in detail [3–5].
Phasins and their fusion proteins have been increasingly considered for protein production at the PHA granule surface [2, 6–8]. Recently, PHA synthase engineering enabled production of the beta-galactosidase and GFP fusion proteins, respectively, at the PHA granule surface [9, 10]. The GFP-PHA synthase fusion even enabled monitoring of in vivo PHA granule formation indicating that PHA granule formation starts at the cell poles. Only recently, PHA granules have been considered as spherical biopolyester particles which can be stably maintained outside the bacterial cell exerting a size range from about 100 nm to several μm [2, 8].
Here we co-expressed the PHA biosynthesis operon from Cupriavidus necator with a hybrid gene encoding a phasin fusion protein in Escherichia coli in order to mediate the formation of PHA granules efficiently displaying the respective fusion partner. In this study, eukaryotic antigen displaying PHA granules were designed and their application performance with respect to diagnostic applications using fluorescence activated cell sorting (FACS) was evaluated. The displayed myelin oligodendrocyte glycoprotein (MOG) was used as an example depicting diagnostic analysis of the autoimmune disease multiple sclerosis (MS). It was for the first time observed that surface-engineered and antigen displaying PHA granules can be efficiently used for FACS based diagnostics. Thus designed PHA granules, which combine cheap one step production with facilitated folding of proteins, might in future replace commercial beads.
Production of PhaP fused to IL2 or MOG at the PHA granule surface
Bacterial strains, plasmids and oligonucleotides used in this study
Source or reference
recA1, endA1, gyrA96, thi-1, hsdR17 (rk-, mk+),
supE44, relA1, -, lac [F', proAB, lacI q , lacZΔM15, Tn10(Tcr)]
Cupriavidus necator H16
Amp, ColE1 origin
PHB biosynthesis operon from C. necator in pBluescriptSK-
pET-14b containing NdeI/BamHI inserted phaC gene from C. necator
DNA fragment encoding N terminal 1–173 amino acids of MOG from mouse inserted in SmaI site of pUC57
DNA fragment encoding N terminal 60–169 amino acids of IL2 from mouse inserted in SmaI site of pUC57
pCRII containing phaP gene from C. necator
pHAS containing the phaP gene inserted into XbaI/NdeI sites
pHAS-phaP containing the MOG encoding DNA fragment inserted into NdeI/BamHI sites
pHAS-phaP containing the IL2 encoding DNA fragment inserted into NdeI/BamHI sites
DNA fragment encoding PhaP inserted into XbaI and NdeI site of plasmid pUC57-MOG
DNA fragment encoding PhaP inserted into XbaI and NdeI site of plasmid pUC57-IL2
DNA fragment encoding PhaP-MOG fusion protein subcloned from pUC57-phaP-MOG via XbaI and BamHI into pBHR68
DNA fragment encoding PhaP-IL2 fusion protein subcloned from pUC57-phaP-IL2 via XbaI and BamHI into pBHR68
Identified peptide fragments of proteins analyzed by MALDI-TOF/MS
A91-Y120, E96-A114, E121-E135, E190-T218,
A57-Y86, E62-A80, E87-E101, E156-T184, L300-G320
N15-M34, W18-M34, L89-H118, A94-H118, L119-N155, L113-V133, A156-S169, T185-T216, A188-T216, T222-G232, G247-L255, T273-P283, V289-R308
L55-H84, A60-H84, L85-N121, L79-V99, A122-S135, A154-T182, A173-T188, G247-L255, T273-P283, T275-V288, V289-R308.
Batch cultivations led to the production of about 1 × 1011 PHA granules/L cultivation broth corresponding to about 3.0 g protein displayed at the PHA granules surface.
Native IL2 and MOG proteins were displayed on PHA granules
PhaP-IL2 and PhaP-MOG fusion proteins containing the Asp-Asp-Asp-Asp-Lys recognition sequence are cleavable with enterokinase
PHA granules displaying PhaP-MOG fusion proteins were used in FACS-based assays to detect antigen-specific serum antibodies
Engineered PHA granules were subjected to FACS analysis using antibodies specifically recognizing the respective native protein folds, which showed the display of the natively folded eukaryotic proteins MOG and IL2 at the PHA granule surface as well as the applicability of designed PHA granules for FACS-based diagnostics (Fig. 3). Although MOG and IL2 are secreted proteins, the derived protein domains were properly folded attached to the PHA granule in the reducing cytosol of E. coli while avoiding the formation of protein inclusion bodies. Although E. coli strains (e.g. Origami) are available, which provide an oxidative cytosol, protein overproduction often leads to inclusion body formation. Enterokinase treatment of MOG or IL2 displaying PHA granules and the complete release of the antigen suggested that the respective antigen was exposed at the PHA granule surface (Fig. 4). This surface exposure enables efficient antigen specific antibody binding. To evaluate the suitability of MOG or IL2 displaying PHA granules for qualitative and quantitative FACS-based antibody detection, mice were immunized with MOG or OVA (control) and induction of specific antibody production was assessed and confirmed by ELISA (Fig. 5). These MOG or OVA antisera were analysed using the MOG displaying PHA granules and FACS technology, which clearly indicated that anti-MOG antibodies can be specifically detected at least up to an antisera dilution of 1:100,000 (Fig. 5). In this study, it was demonstrated that eukaryotic proteins can be functionally displayed at the PHA granule surface using protein engineering of PhaP. Evidence was provided that the fusion proteins (antigens) were exposed at the PHA granule and thus PHA granules could be used to capture antibodies. This feature in combination with the particle properties of the PHA granules led to an outstanding performance in FACS based diagnostics, particularly considering the signal to noise ratio and dynamic range of antibody detection. It was recently shown that ZZ domain displaying PHA granules, which were produced via protein engineering of the PHA synthase, were suitable for IgG purification . When compared to commercial beads, engineered PHA granules showed increased sensitivity with similar distribution of signal intensities in FACS. Overall, protein engineering of PHA granule surface proteins provides a novel molecular tool for the display of antigens for FACS-based diagnostics.
In this study, it was demonstrated that correctly folded eukaryotic proteins can be abundantly produced at the PHA granule surface as phasin fusion proteins. Isolated PHA granules displaying the respective eukaryotic proteins could be used as beads for specific and sensitive antibody detection using FACS technology. These native antigen displaying PHA granules were manufactured by recombinant E. coli without the need of antigen purification and chemical cross-linking to independently produced beads. Often purification of these proteins requires tedious refolding at low efficiency. The production of functional eukaryotic proteins at the PHA granule surface represents a novel in vivo matrix-assisted protein folding system avoiding aggregation of protein folding intermediates. Moreover, PHA granules could be stored for at least one year at 4°C without loss of performance in FACS supporting their potential use in diagnostic applications.
This work opens up an alternative route for the production of protein displaying beads harnessing nature's capacity to produce spherical polymer beads which surface can be functionalized by engineering of specifically bead associating proteins (Fig. 6). Multiple functionality might be easy achievable by co-expression of various hybrid genes suggesting that this biotechnological bead production strategy might represent a versatile platform technology.
Bacterial strains and growth conditions
Bacterial strains, plasmids and oligonucleotides used in this study are listed in Table 1. Cupriavidus necator and E. coli XL1 Blue were grown at 37°C. When required, ampicillin 75 μg/ml was added. All chemicals were purchased from Sigma-Aldrich (St. Louis, Mo., USA).
Isolation, analysis and manipulation of DNA
General cloning procedures and isolation of genomic DNA were performed as described elsewhere . DNA primers, deoxynucleoside triphosphate, Taq and Platinum Pfx polymerases were purchased from Invitrogen™ (CA, USA). DNA sequences of new plasmid constructs were confirmed by DNA sequencing according to the chain termination method using the model ABI310 automatic sequencer. Plasmids used in this study are listed in Table 1.
Construction of plasmids mediating production of either IL2 or MOG displaying PHA granules
The DNA fragments encoding either the full length mature IL2 protein (amino acids 60–169, accession no. AAN38301) or the extracellular part of the MOG protein (amino acids 1–117, accession no. Q61885) from mouse were synthesized by GenScript Corp. (USA). The codon usage was optimized for expression in E. coli [see Additional file 1]. Each DNA fragment contained an NdeI site at the 5' end and a BamHI site at the 3' end. These DNA fragment were inserted into the SmaI site of pUC57 directly after synthesis by GenScript Corp. (USA) resulting in plasmids pUC57-MOG or pUC57-IL2, respectively. The coding region of phaP1 gene was amplified by PCR from chromosomal DNA of C. necator using oligonucleotides phaP-XbaI-NdeI and phaP-NdeI listed in Table 1 and introducing an XbaI site at the 5' end and an NdeI site including an enterokinase site at the 3' end. The PCR product was subcloned into TA cloning plasmid pCRII. The phaP coding region was again amplified from plasmid pCR-phaP using oligonucleotides phaP-XbaI including an E. coli ribosomal binding site and phaP-NdeI. The resulting PCR product was subcloned into the XbaI and NdeI sites of pHAS.
Either pUC57-MOG or pUC57-IL2 was hydrolyzed with NdeI and BamHI and the corresponding DNA fragments were subcloned into pHAS-phaP resulting in plasmids pHAS-phaP-MOG and pHAS-phaP-IL2, respectively. The respective fusion protein encoding region was then subcloned into pBHR68 using XbaI and BamHI sites downstream of the lac promoter and upstream of the PHB biosynthesis operon (Fig. 1).
Production of phasin fusion proteins at the PHA granule surface
Cells of E. coli XL1 Blue were transformed with plasmids pBHR68-PhaP-IL2 and pBHR68-PhaP-MOG, respectively. Transformants were grown at 37°C and induced by adding isopropyl-β-D-thiogalactopyranoside (IPTG) to a final concentration of 1 mM. After growth for 48 h, cells were harvested by centrifugation and subjected to PHA granule isolation.
Isolation of PHA granules
Cells were harvested by centrifugation for 15 min at 5,000 × g and 4°C. The sediment was washed and suspended in 3 volumes of 50 mM phosphate buffer (pH7.5). Cells were passed through French Press four times at 8000 psi. The cell lysate (0.75 ml) was loaded onto a glycerol gradient (88 % and 44 % (v/v) glycerol in phosphate buffer). After ultracentrifugation for 2.5 h at 100,000 × g and 10°C, granules could be isolated from a white layer above the 88 % glycerol layer. The PHA granules were washed with 10 volumes phosphate buffer (50 mM, pH7.5) and centrifuged at 100,000 × g for 30 min at 4°C. The sediment containing the PHA granules was suspended in phosphate buffer and stored at 4°C.
Sodium dodecyl sulfate (SDS) polyacrylamide gel electrophoresis (PAGE)
Protein samples were routinely analyzed by SDS-PAGE as described elsewhere . The gels were stained with Coomassie brilliant blue G250. Protein bands of interest were cut off the gel and analyzed by matrix assisted laser desorption/ionization time-of flight mass spectrometry (MALDI TOF-MS).
MALDI-TOF mass spectrometry
Mass spectrometric analyses of tryptic peptides were carried out on a MALDI VOYAGER DE-PRO time of flight mass spectrometer from PerSeptive BioSystems (Framingham, MA) utilizing a nitrogen laser, emitting at 337 nm, and an accelerating voltage of 25 kV. Measurements were performed in the delayed extraction mode using a low mass gate of 2000. The mass spectrometer was used in the positive ion detection and linear mode. Samples of the digestion mixture were placed directly on a 100-position sample plate, and allowed to air-dry after the addition of an equal volume of saturated solution of 3,5-dimethoxy-4-hydroxycinnaminic acid (sinapinic acid) in 50% acetonitrile and 0.3% TFA.
C57BL/6 mice were originally purchased from the Jackson Laboratory (Bar Harbor, ME). All mice were maintained by the Biomedical Research Unit, Malaghan Institute of Medical Research, Wellington, New Zealand. Experimental protocols were approved by the Victoria University of Wellington Animal Ethics Committees and performed according to the guidelines of their guidelines. In all experiments, sex and age matched mice were used between 8–12 weeks of age.
Mouse serum IgG responses
C57BL/6 mice were immunized with 150 μg recombinant mouse MOG protein consisting of amino acid 1–117 (MOG1–117) of the matured protein  or 150 μg ovalbumin protein (OVA, Sigma-Aldrich) emulsified in Complete Freund's adjuvant (CFA), containing 500 μg Mycobacterium tuberculosis (both DIFCO Laboratories). The emulsion (100 μl) was injected subcutaneously over the flanks. Blood was collected from tail bleeds 28 days post-immunization, centrifuged at 13,000 g for 1 min, and then the top layer of serum was removed and stored at -20°C. Serum was tested for antigen-specific total IgG, using the PhaP-MOG granules and flow cytometry or ELISA (see below).
Cleavage of IL2 from IL2 displaying PHA granules
The enterokinase recognition sequence Asp-Asp-Asp-Asp-Lys was introduced between PhaP and IL-2 to enable cleavage of the IL-2 protein from the granules. Twenty-five μg of bovine enterokinase (Sigma-Aldrich, E-5144) in 10 mM Tris/HCl, 10 mM CaCl2 (pH8) were incubated with 2 × 1010 PhaP-IL2 or PhaP-MOG granules at 37°C, and at indicated time points samples were removed, washed with PBS, and stored at 4°C until use. The relative amount of native IL2 and MOG at the surface of the granules was determined using flow cytometry.
PhaP-MOG or PhaP-IL2 granules (~5 × 108/well) were added to 96-well round-bottom plates (Becton, Dickinson and Company) and washed twice with FACS-buffer (PBS, 1% Foetal calf serum, 0.1% sodium azide, 5 mM EDTA, pH8). IL2-phaP or MOG-phaP granules were then incubated with phycoerythrin conjugated anti-IL2 monoclonal antibody (clone PC61, PharMingen) or a mouse anti-MOG monoclonal antibody (clone 8.18-C5, kindly provided by C. Bernard). Following 15 minutes incubation at ambient temperature, granules were washed twice in fluorescence activated cell sorting (FACS) buffer. Allophycocyanin-conjugated anti-mouse IgG (Jackson ImmunoResearch Laboratories Inc.) polyclonal antibodies were added to the MOG displaying PHA granules and incubated for 15 minutes, and then washed twice. At least 10,000 events for each sample were collected and analysed, using a BD FACsCAlibur and FlowJo analysis software (Tree Star, San Carlos, CA, USA). For detection of MOG-specific antibody responses, sera from MOG or OVA immunized mice were diluted in FACS-buffer and then incubated for 30 minutes at ambient temperature with MOG-phaP granules. The granules were then washed and incubated with biotinylated anti-mouse total IgG, followed by PE-conjugated streptavidin. At least 10,000 events for each sample were collected and analysed on a BD FACsCAlibur and FlowJo software (Tree Star).
Round-bottom 96-well plates (Becton, Dickinson and Company) were coated overnight at 4°C (50 μl/well) with 3 μg/ml of recombinant mouse MOG1–117or OVA protein (Sigma-Aldrich) in PBS. Supernatant was discarded and wells blocked by adding 1% BSA in PBS (100 μl/well) for 1 h at ambient temperature. Plates were then wash 3 times with 10 mM Tris/HCl, pH 7.5, 0.05% Tween 20 (ELISA buffer). Diluted sera (50 μl/well) in 0.1% BSA/PBS from MOG1–117 or OVA immunized mice were added and after 2 h, plates were washed 3 times with ELISA buffer. Biotin-conjugated anti-mouse IgG antibodies (Southern Biotechnology, Ltd) diluted 1:4,000 in 0.1% BSA/PBS were added for 1 h, and wells then wash 3 times with ELISA buffer. Amdex Streptavidin-HRP (Amersham Biosciences) diluted 1:3,000 in 0.1% BSA/PBS, was added to each well (50 μl/well) and incubated for 30 minutes. Plates were then wash 3 times and 3,3',5,5'-tetramethylbenzidine (TMB) added for 5–30 minutes. Colour development was stopped using 2 M H2SO4 and the ELISA plates were then read at 450 nm on a Benchmark microplate reader (Bio-Rad Laboratories Inc. Hercules, CA, USA).
The authors would like to thank Verena Peters for scientific discussion and Roleen Lata for technical assistance. Recombinant mouse MOG1–117 was a kind gift from Dr Claude Bernard, Monash University, Melbourne, Australia. This study was supported by research grants from Massey University and PolyBatics Ltd. B.T.B. is the recipient of the Wellington Medical Research Foundation Malaghan Fellow. Proteomic analysis was performed by Simone König (Integrated Functional Genomics, Interdisciplinary Center for Clinical Research, University of Münster, Germany).
- Rehm BHA: Polyester synthases: natural catalysts for plastics. Biochem J. 2003, 376 (Pt 1): 15-33. 10.1042/BJ20031254.View ArticleGoogle Scholar
- Rehm BHA: Genetics and Biochemistry of Polyhydroxyalkanoate Granule Self-assembly: The Key Role of Polyester Synthases. Biotechnol Lett. 2006, 28 (4): 207-213. 10.1007/s10529-005-5521-4.View ArticleGoogle Scholar
- York GM, Stubbe J, Sinskey AJ: New insight into the role of the PhaP phasin of Ralstonia eutropha in promoting synthesis of polyhydroxybutyrate. J Bacteriol. 2001, 183 (7): 2394-2397. 10.1128/JB.183.7.2394-2397.2001.View ArticleGoogle Scholar
- York GM, Junker BH, Stubbe JA, Sinskey AJ: Accumulation of the PhaP phasin of Ralstonia eutropha is dependent on production of polyhydroxybutyrate in cells. J Bacteriol. 2001, 183 (14): 4217-4226. 10.1128/JB.183.14.4217-4226.2001.View ArticleGoogle Scholar
- Hanley SZ, Pappin DJ, Rahman D, White AJ, Elborough KM, Slabas AR: Re-evaluation of the primary structure of Ralstonia eutropha phasin and implications for polyhydroxyalkanoic acid granule binding. FEBS Lett. 1999, 447 (1): 99-105. 10.1016/S0014-5793(99)00235-5.View ArticleGoogle Scholar
- Banki MR, Gerngross TU, Wood DW: Novel and economical purification of recombinant proteins: intein-mediated protein purification using in vivo polyhydroxybutyrate (PHB) matrix association. Protein Sci. 2005, 14 (6): 1387-1395. 10.1110/ps.041296305.View ArticleGoogle Scholar
- Barnard GC, McCool JD, Wood DW, Gerngross TU: Integrated recombinant protein expression and purification platform based on Ralstonia eutropha. Appl Environ Microbiol. 2005, 71 (10): 5735-5742. 10.1128/AEM.71.10.5735-5742.2005.View ArticleGoogle Scholar
- Rehm BHA: Biopolyester particles produced by microbes or using polyester synthases: Self-assembly and potential applications. In Microbial Bionanotechnology: Biological self-assembly systems and biopolymer-based nanostructures; ed. Bernd H.A. Rehm. 2006, horizon bioscience, pp 1-34, 1-34.Google Scholar
- Peters V, Rehm BHA: In vivo monitoring of PHA granule formation using GFP-labeled PHA synthases. FEMS Microbiol Lett. 2005, 248 (1): 93-100. 10.1016/j.femsle.2005.05.027.View ArticleGoogle Scholar
- Peters V, Rehm BHA: In vivo enzyme immobilization by use of engineered polyhydroxyalkanoate synthase. Appl Environ Microbiol. 2006, 72 (3): 1777-1783. 10.1128/AEM.72.3.1777-1783.2006.View ArticleGoogle Scholar
- Linnington C, Webb M, Woodhams PL: A novel myelin-associated glycoprotein defined by a mouse monoclonal antibody. J Neuroimmunol. 1984, 6 (6): 387-396. 10.1016/0165-5728(84)90064-X.View ArticleGoogle Scholar
- Lowenthal JW, Corthesy P, Tougne C, Lees R, MacDonald HR, Nabholz M: High and low affinity IL 2 receptors: analysis by IL 2 dissociation rate and reactivity with monoclonal anti-receptor antibody PC61. J Immunol. 1985, 135 (6): 3988-3994.Google Scholar
- Sambrook J, Fritsch EF, Maniatis T: Molecular cloning: a laboratory manual. 1989, Cold Spring Harbor Lab. Press, Plainview, NYGoogle Scholar
- Laemmli UK: Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature. 1970, 227 (5259): 680-685. 10.1038/227680a0.View ArticleGoogle Scholar
- Clements CS, Reid HH, Beddoe T, Tynan FE, Perugini MA, Johns TG, Bernard CC, Rossjohn J: The crystal structure of myelin oligodendrocyte glycoprotein, a key autoantigen in multiple sclerosis. Proc Natl Acad Sci U S A. 2003, 100 (19): 11059-11064. 10.1073/pnas.1833158100.View ArticleGoogle Scholar
- Bullock WO, Fernandez JM, Short JM: XL1-Blue - a high efficiency plasmid transorming recA Escherichia-coli strain with beta-galactosidase SELECTION. Biotechniques. 1987, 5 (4): 376-Google Scholar
- Spiekermann P, Rehm BHA, Kalscheuer R, Baumeister D, Steinbüchel A: A sensitive, viable-colony staining method using Nile red for direct screening of bacteria that accumulate polyhydroxyalkanoic acids and other lipid storage compounds. Arch Microbiol. 1999, 171 (2): 73-80. 10.1007/s002030050681.View ArticleGoogle Scholar
- Yuan W, Jia Y, Tian J, Snell KD, Muh U, Sinskey AJ, Lambalot RH, Walsh CT, Stubbe J: Class I and III polyhydroxyalkanoate synthases from Ralstonia eutropha and Allochromatium vinosum: characterization and substrate specificity studies. Arch Biochem Biophys. 2001, 394 (1): 87-98. 10.1006/abbi.2001.2522.View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.