New tools for chloroplast genetic engineering allow the synthesis of human growth hormone in the green alga Chlamydomonas reinhardtii
- First Online:
- Cite this article as:
- Wannathong, T., Waterhouse, J.C., Young, R.E.B. et al. Appl Microbiol Biotechnol (2016) 100: 5467. doi:10.1007/s00253-016-7354-6
- 2k Downloads
In recent years, there has been an increasing interest in the exploitation of microalgae in industrial biotechnology. Potentially, these phototrophic eukaryotes could be used for the low-cost synthesis of valuable recombinant products such as bioactive metabolites and therapeutic proteins. The algal chloroplast in particular represents an attractive target for such genetic engineering, both because it houses major metabolic pathways and because foreign genes can be targeted to specific loci within the chloroplast genome, resulting in high-level, stable expression. However, routine methods for chloroplast genetic engineering are currently available only for one species—Chlamydomonas reinhardtii—and even here, there are limitations to the existing technology, including the need for an expensive biolistic device for DNA delivery, the lack of robust expression vectors, and the undesirable use of antibiotic resistance markers. Here, we describe a new strain and vectors for targeted insertion of transgenes into a neutral chloroplast locus that (i) allow scar-less fusion of a transgenic coding sequence to the promoter/5′UTR element of the highly expressed endogenous genes psaA or atpA, (ii) employ the endogenous gene psbH as an effective but benign selectable marker, and (iii) ensure the successful integration of the transgene construct in all transformant lines. Transformation is achieved by a simple and cheap method of agitation of a DNA/cell suspension with glass beads, with selection based on the phototrophic rescue of a cell wall-deficient ΔpsbH strain. We demonstrate the utility of these tools in the creation of a transgenic line that produces high levels of functional human growth hormone.
KeywordsChlamydomonasChloroplastGenetic engineeringHuman growth hormone
Microalgae represent attractive biotechnology platforms for the synthesis of recombinant therapeutic proteins and other high-value products (Gong et al. 2011; Rasala and Mayfield 2015). Key advantages when compared to one or more of the traditional heterotrophic platforms such as Escherichia coli, yeast, or CHO cells include (i) the low-cost phototrophic cultivation of the alga in sterile, controlled photobioreactors using simple and inexpensive medium; (ii) the generally recognized as safe (GRAS) status of a number of algal species including the chlorophyte Chlamydomonas reinhardtii, making possible the oral delivery of bioactive products in whole algae and therefore avoiding expensive downstream purification; and (iii) the chloroplast organelle as a unique biosynthetic and storage compartment that contains its own minimal genetic system. Importantly, the use of the chloroplast for expression of foreign genes rather than the algal nucleus confers several benefits including precise insertion into the chloroplast genome (or “plastome”) via homologous recombination, high-level expression that is not subject to any gene-silencing mechanisms, the possibility of expressing multiple transgenes as operons, and accumulation of the recombinant proteins in a benign compartment where the formation of disulfide bonds occurs readily (Tran et al. 2009) but undesirable glycosylations are avoided (Gregory et al. 2013).
Chloroplast transgenics were first reported in 1991 when a bacterial antibiotic-resistance gene was expressed in C. reinhardtii (Goldschmidt-Clermont 1991). Since that time, there have been many reports describing the synthesis of functional therapeutic proteins in the C. reinhardtii chloroplast including monoclonal antibodies (Mayfield et al. 2003; Tran et al. 2009), growth factors (Rasala et al. 2010), antigens (Dreesen et al. 2010; Michelet et al. 2011; Jones et al. 2013), gut-active proteins (Manuell et al. 2007; Yoon et al. 2011), anti-bacterial proteins (Braun-Galleani et al. 2015), immunotoxins (Tran et al. 2013a, 2013b), and anti-toxins (Barrera et al. 2015). In addition, efforts are being made to manipulate chloroplast biosynthetic pathways in order to synthesize novel bioactive compounds such as diterpenoids (Gangl et al. 2015; Zedler et al. 2015). This encouraging progress in the development of the algal chloroplast as a viable platform has recently led to the establishment of start-up companies seeking to exploit the technology and the demonstration of pilot-scale production of a bioactive protein (Gimpel et al. 2015).
However, there remains a need to develop improved molecular tools that address some of the current technical limitations in the generation of C. reinhardtii transgenic lines (Purton et al. 2013). Specifically, there is a need for a simple and reliable method of rapidly generating homoplasmic transformant lines that also avoids the use of bacterial antibiotic-resistance genes as selectable markers. Currently, transformation typically involves bombardment of an algal lawn with DNA-coated microparticles (= biolistics) and the use of the aadA or aphA6 bacterial genes as selectable markers conferring resistance to spectinomycin and kanamycin, respectively (Goldschmidt-Clermont 1991; Bateman and Purton 2000). Resistant colonies are then checked for the presence of the gene of interest (GOI) and taken through multiple rounds of single-colony selection in order to ensure that the transformant lines attain a stable, homoplasmic state in which all copies of the polyploid genome contain the marker and the GOI. A simpler alternative to microparticle bombardment involves agitating a suspension of cells and transforming DNA in the presence of glass beads, although this method requires the prior removal of the cell wall either by digestion or mutation (Kindle et al. 1991; Economou et al. 2014).
Similarly, one alternative method for selection employs non-photosynthetic mutants as recipient strains where the genetic lesion is in a key photosynthetic gene on the chloroplast genome. Selection is based on the use of a wild-type copy of the gene as the marker with successfully transformed cells able to grow phototrophically on minimal medium through replacement of the mutated gene with the wild-type version (Purton 2007; Michelet et al. 2011; Chen and Melis 2013). This selection strategy therefore allows the introduction of a GOI as the only transgene, avoiding the use of any antibiotic-resistance gene. Such “marker-free” transgenic lines are appropriate for industrial cultivation since they circumvent the regulatory and environmental issues associated with the possible horizontal transfer of such resistance genes into other microorganisms. Other issues associated with antibiotic-based selection include (i) the occurrence of “false-positive” colonies due to natural resistance mutations arising in C. reinhardtii genes, (ii) the additional metabolic burden on the chloroplast of replicating and expressing the marker, and (iii) the challenge of using antibiotic selection to drive transformant lines efficiently to a homoplasmic state.
Our recent work on C. reinhardtii chloroplast engineering (Braun-Galleani et al. 2015; Young and Purton 2014, 2016) has involved the use of a new recipient strain and expression vectors that allow the generation of marker-free transformants using the glass bead transformation method. In this paper, we describe these tools in detail. The recipient strain carries the nuclear cw15 mutation resulting in a cell wall-deficient phenotype, together with a deletion of the essential photosystem II gene psbH to allow selection based on phototrophy. The expression vectors carry the psbH marker and are designed to allow simple cloning of a GOI as a perfect translational fusion with the promoter and 5′ untranslated region (UTR) from various chloroplast genes. We find that the vector that uses the psaA promoter/5′UTR element gives high expression levels in the chloroplast and is able to serve as a dual expression vector, since expression is also readily detectable in E. coli. Finally, we describe a simple three-primer PCR strategy to confirm both the correct integration of the GOI into the plastome and the establishment of homoplasmy. To validate these new tools, we have created a transgenic line expressing a synthetic gene encoding human growth hormone (hGH) and shown that the hGH is biologically active.
Materials and methods
C. reinhardtii strains and culture conditions
The TN72 strain is a psbH::aadA knockout mutant of the cell wall-deficient strain cw15.mt+ (Davies and Plaskitt 1971) and was created as described below. TN72 has been deposited in the publicly available collection of the Chlamydomonas Resource Center (chlamycollection.org) as strain CC-5168. TN72 transformant lines were initially selected on 2 % agar plates containing high salt minimal (HSM) medium but then maintained on Tris-acetate phosphate (TAP) agar plates (Harris et al. 2009) under dim light (5–10 μE/m2/s) at 20 °C. For liquid cultures, cells were grown in TAP medium in conical flasks at 25 °C, 120 rpm at 5 μE/m2/s (for TN72) or 50 μE/m2/s (for transformants). Spectinomycin (100 μg/ml) and erythromycin (10–50 μg/ml) were added to TAP agar medium as required.
DNA manipulations were carried out using standard protocols (Green and Sambrook 2012), including transformation of chemically competent E. coli DH5α using ampicillin selection. Oligonucleotide primers were obtained from Eurofins Genomics (Ebersberg, Germany); restriction enzymes and T4 DNA ligase for cloning procedures were purchased from NEB (Ipswich, MA, USA) and Thermo Scientific (Waltham, MA, USA). Details of oligonucleotide primers used for plasmid construction are given in Table S1. Plasmid p72B-aadA-BM, which was used to create TN72, was made by replacing the 0.36-kb MluI-BstXI region of p72B (Bateman and Purton 2000) with the “aadA cassette” (comprising the aadA coding sequence fused to the atpA promoter/5′UTR and rbcL 3′UTR) amplified from pUC-atpX-AAD (Goldschmidt-Clermont 1991) using primers that introduced an MluI and BstXI site at the ends of the cassette. Transformation vector pASapI was created from p72B-SH (Bateman and Purton 2000), by first cutting p72B-SH with SapI (within the pUC8 vector) and EcoRI, end-filling, and religating, thereby removing a 240-bp region of the vector sequence including the unwanted SapI site. Then, the MluI-NcoI region containing the rbcL promoter/5′UTR element was replaced with an equivalent region amplified from atpA to give pASapI in which a SapI site was created immediately downstream of the start codon. Vectors pSRSapI, pPSapI, and pCSapI were created from pASapI by replacing the MluI to SapI region of pASapI with the promoter/5′UTR region of psaA exon 1, psbA, and chlL, respectively. The sequences of pASapI and pSRSapI are given in Fig. S1. The ereB coding sequence was amplified from plasmid pAT72 (Arthur and Courvalin 1986) using primers that introduce an NcoI and SphI site at the start and end of the coding region, respectively, and cloned into these sites in pSRSapI to create pRY129a. The synthetic gene for hGH was codon-optimized for the C. reinhardtii chloroplast using the Codon Usage Database (www.kazusa.or.jp/codon) with SapI and SphI sites added for cloning (Fig. S2). The gene was synthesized by Life Technologies (Carlsbad, CA, USA) and cloned into the four expression vectors.
Chloroplast transformation was based on a previously described method (Kindle et al. 1991) and involved the agitation of an algal/DNA suspension with glass beads of 400–625 μm diameter. A 400-ml culture grown to early log phase (approx. 2 × 106 cells/ml) was concentrated by centrifugation and resuspended in TAP medium to 4 ml. Three hundred microliters of cells were added to a sterilized 5-ml test tube containing 300 mg sterile glass beads, followed by 5–10 μg circular plasmid DNA. The mixture was agitated vigorously at the maximum speed of a Vortex Genie II (Fisher Scientific, Loughborough, UK) for 15 s. The cells were spread on selective agar plates (TAP + spectinomycin at 100 μg/ml for aadA selection; HSM for psbH selection) after mixing with 0.5 % molten (42 °C) agar of the same selective medium. The plates were incubated at 25 °C in dim light (~2 μE/m2/s) overnight then transferred to a moderate light (~50 μE/m2/s) the next day. Transformant colonies were picked after 2–3 weeks and restreaked to single colonies several times on selective media to ensure homoplasmicity, although this was often achieved after the first streaking. Homoplasmy was determined by PCR using a combination of three primers (Table S1). Total genomic DNA was extracted from a loopful of cells using the Chelex 100 method as described by Werner and Mergenhagen (1998), and PCR amplification was carried out with Phusion DNA polymerase (Thermo Scientific) according to the manufacturer’s instructions. The photosystem II-deficient phenotype of TN72 was confirmed by measuring the photosynthetic capacity through chlorophyll fluorescence (Maxwell and Johnson 2000). Putative transformant lines were spotted onto TAP agar plates, incubated for 5 days and scanned using a pulse-modulated imaging fluorometer (FluorCam 700MF, Photon Systems Instruments, Czech Republic) as described by Wingler et al. (2004).
Western analysis of C. reinhardtii and E. coli transformants
Protein samples from C. reinhardtii were prepared, fractionated on a 15 % acrylamide SDS-PAGE gel and blotted onto nitrocellulose membrane as described by Young and Purton (2014). For immunodetection, membranes were blocked in TBS-T (20 mM Tris, 137 mM NaCl, 1 M HCl pH 7.4, 0.1 % Tween-20) with 0.5 % milk powder overnight at 4 °C. Membranes were incubated with primary antibody in TBS-T with 0.5 % milk (1 h at RT), washed 3 × 5 min in TBS-T, incubated with secondary antibody in TBS-T with 0.5 % milk (1 h at RT), and washed as before. The probed blots were scanned and quantified using an Odyssey Infrared Imaging System (LI-COR). Alternatively, for Fig. 5a, the antibody binding was detected by enhanced chemiluminescence (ECL) using SuperSignal West Pico Chemiluminescent substrate (Thermo Scientific) and Hyperfilm ECL (GE Healthcare). For hGH detection, the primary antibody was chicken polyclonal antibody to human growth hormone (Abcam product ab27223) at 1:5000 dilution and the secondary was donkey anti-chicken IgY IRDye 800CW (LI-COR) at 1:5000, or goat anti-chicken IgY HRP-linked antibody (Abcam product ab97135) at 1:5000 for ECL. For RbcL detection, the primary antibody was anti-RbcL produced in rabbit (a gift from John Gray, Cambridge University, UK) at 1:5000 followed by secondary antibody Goat anti-Rabbit IgG (H&L), Dylight™ 800 Conjugated (Thermo Scientific product 35571) at 1:10,000, or HRP-linked Goat Anti-Rabbit IgG H&L (Abcam product ab6721) at 1:5000 for ECL.
For the detection of proteins in E. coli, 5 ml of an overnight LB culture was harvested by centrifugation (10 min at 4500 g) and resuspended in X/4 ml solution A (0.8 M Tris-HCl pH 8.3, 0.2 M sorbitol, 1 % β-mercaptoethanol) where X is the absorbance of the culture at 600 nm. SDS was added to a final concentration of 1 %. Samples were boiled for 5 min and passed through a 0.5-mm needle then centrifuged (2 min at 21,000 g), and 25 μl of supernatant was loaded. Gel and hGH immunodetection conditions using the ODYSSEY system were as described above. For both the Chlamydomonas and E. coli analysis, pure recombinant hGH (Abcam product 51232) was used as a reference.
Nb2-11 cell proliferation assay for hGH activity
Nb2-11 cells (European Collection of Cell Cultures) were grown in suspension culture in Fischer’s medium supplemented with 10 % fetal bovine serum, 10 % horse serum, 2 mM l-glutamine, 0.075 % sodium bicarbonate, and 0.05 mM β-mercaptoethanol, at 37 °C in an atmosphere containing 5 % CO2. Cells were starved for 24 h by centrifugation (750 g for 5 min) to remove the culture medium, followed by washing and resuspending the cells in supplemented Fischer’s medium excluding fetal bovine serum. The cells were seeded in a 12-well plate at a concentration of approximately 1 × 105 cells/ml. Recombinant hGH (Abcam) was added to 1 ml of cells in a volume of 5 μl to give a final concentration of 1, 10, or 100 ng/ml. Phosphate buffer (50 mM, pH 7.4) was added to cells as a control and was used to dilute the hGH. Samples were analyzed in triplicate. Cell proliferation was monitored every 24 h using the CASY® Cell Counter and Analyzer Model TT (Innovatis, Bielefeld, Germany), with cells >8.03 μm being defined as viable. To test C. reinhardtii extracts, transformant TN-hGH, and the negative control TN-C, cells were cultured in 500-ml TAP medium to a concentration of 2 × 106 cells/ml, broken by three cycles of freezing in liquid nitrogen and thawing, centrifuged at 9000 g for 30 min, followed by ultracentrifugation of the supernatant at 100,000 g for 60 min. The hGH content in the TN-hGH extract was quantified by Western blotting as above and a volume containing ~100 ng hGH used in the proliferation assay, together with an equivalent amount of TN-C extract, or phosphate buffer as controls. The experiment was repeated in triplicate with proliferation measured after 96 h.
The Genbank accession number of the synthetic hGH used in the study is KU179444.
The ΔpsbH strain TN72 for reliable, marker-free selection of transformant lines
Our previous research has shown that disruption of the chloroplast gene psbH, which encodes the 9 kDa phosphoprotein subunit of photosystem II, results in a complete destabilization of this complex. As a result, psbH mutants are blocked in photosynthetic function and, thus, have an acetate-dependent phenotype (O’Connor et al. 1998). These mutants can be restored to phototrophy following biolistic transformation with a plasmid carrying wild-type psbH, with selection on minimal medium (Bateman and Purton 2000). The psbH gene can therefore be used as an effective selectable marker for C. reinhardtii, as has been established for other photosynthetic genes including atpB, tscA, and rbcL (Boynton et al. 1988; Goldschmidt-Clermont 1991; Chen and Melis 2013).
New expression vectors allow a perfect fusion of the GOI to the 5′UTR and rapid establishment of chloroplast homoplasmy
In order to test different transcription and translation efficiencies, four different versions of the vector were constructed that each have the same multiple cloning site for GOI insertion followed by the 3′UTR from the endogenous rbcL gene, but differ in the endogenous promoter/5′UTR element used to drive expression, namely that from psaA exon 1 (vector pSRSapI), atpA (pASapI), psbA (pPSapI), and chlL (pCSapI) as detailed in Table S2. Each vector can be used in combination with the TN72 line to target any GOI to the neutral locus within the psbH-trnE intergenic region, using restoration of psbH as selection as illustrated in Fig. 1.
Since the C. reinhardtii plastome is polyploid with ~40 copies per cell under phototrophic conditions and ~100 copies under mixotrophic conditions (Lau et al. 2000), an important early step in the analysis of putative transformant lines is to confirm that the GOI has inserted into the plastome and that all copies carry the GOI—i.e., the lines are homoplasmic (Purton 2007). We have devised a PCR strategy involving three primers (indicated in Fig. 1, with details in Table S1) to confirm both integration and homoplasmy in a single step. Primers F1 and R1 generate a 0.88-kb PCR product from the untransformed TN72 plastome, whereas in the transplastomic genome, a larger band (e.g., 1.13 kb for pSRSapI transformants) is generated by primers F1 and R2, with primer R2 designed to the promoter/5′UTR element used in each vector and primer F1 outside of the left-hand region of homology carried on each vector (thereby preventing amplification from any free plasmid or plasmid integrated into the nuclear genome). Heteroplasmic lines containing some untransformed plastome copies would be expected to give rise to both PCR products.
The pSRSapI vector allows efficient expression in both C. reinhardtii and E. coli
A functional hGH is produced in the C. reinhardtii chloroplast
To determine whether the hormone is biologically active, a proliferation assay was carried out using the Nb2-11 rat lymphoma cell line. The replication of Nb2-11 cells in suspension culture is specifically stimulated by lactogenic hormones, and by hGH which is known to have lactogenic activity (Tanaka et al. 1980; Bulatov et al. 1996). This was confirmed using the commercial hGH which showed a dose-dependent effect of the hormone on the proliferation of the cell line (Fig. S5). This proliferation was also seen when a cell extract from TN-hGH containing approximately 100 ng of hGH was added to the Nb2-11 cells, thereby confirming its biological activity. As shown in Fig. 5b, an increase in cell number of approximately eightfold was obtained after 96 h using the TN-hGH extract when compared to an equivalent extract from the control strain TN-C, or to buffer only.
Currently, the industrial biotechnology sector is almost exclusively based around the use of heterotrophic platforms (bacteria, yeasts, mammalian, and insect cells) for the biosynthesis of pharmaceutical proteins, bioactive metabolites, or other high-value products (Murphy 2012). Nevertheless, the ever-increasing growth of the global bioeconomy and the need for sustainable alternatives to petrochemical-based products is catalyzing interest in the exploitation of alternative cell factories, including photosynthetic microalgae such as C. reinhardtii (Wijffels et al. 2013). Recent successes in the biosynthesis of recombinant therapeutics in the C. reinhardtii chloroplast have underlined the potential of this alga as a commercial platform (Rasala and Mayfield 2015), and highlight the need for further development of new tools for rapid and reliable strain engineering (Scaife et al. 2015).
The recipient strain and vectors described here provide a low-cost and simple system for creating a transgenic line within a few weeks. Transformation of the TN72 strain is achieved by agitation with glass beads rather than the more technically complex particle bombardment method, and the strong selection for restoration of photosynthesis means that stable, homoplasmic lines are obtained without repeated rounds of selection. Importantly, the system guarantees that every selected colony contains the GOI since the deletion of the essential PSII gene in TN72 prevents spontaneous mutation to phototrophic growth, and the deletion of the genomic region between psbH and the site of GOI insertion ensures that restoration of psbH is always accompanied by insertion of the GOI. Our finding that the psaA promoter/5′UTR gives high levels of expression in the chloroplast is in agreement with the findings of Michelet et al. (2011), who reported that the same psaA element gave higher levels of their two recombinant proteins than elements from atpA and psbA. This identifies pSRSapI as the best of our expression vectors. Furthermore, the observation that the psaA promoter/5′UTR element used in pSRSapI is able to drive expression in E. coli provides a useful tool for testing the synthesis and function of the recombinant protein prior to generation of C. reinhardtii transformants. We speculate that the difference we see in E. coli expression levels for the different chloroplast gene elements is, in part, due to differences in the efficiency of translation initiation in E. coli. Of the four genes, the psaA sequence immediately upstream of the translational start most closely matches the E. coli Shine-Dalgarno consensus sequence for ribosome binding (Fig. S6). In contrast, a weaker match is seen for chlL and psbA, and no match is seen in the atpA 5′UTR. Furthermore, efficient transcription from the psaA promoter probably also contributes to the overall expression in E. coli. The 5′ end of the psaA exon 1 mRNA was determined by Choquet et al. (1988), and Turmel et al. (1995) predicted that a TATAAT sequence within the psaA exon 1 promoter region corresponded to the –10 element; this exact consensus sequence is in fact preceded by TGn, correlating to the consensus extended –10 element that is thought to contribute to promoter strength (Hook-Barnard and Hinton 2007). Promoter prediction software, softberry BPROM, indicates a possible –35 element (TTGTAA) upstream of the –10 element. These data predict that the psaA exon 1 promoter would give strong transcription initiation in E. coli. While expression in E. coli is a desirable feature of pSRSapI, it can also present problems when cloning genes that are deleterious to the bacterium. To address this, we have recently developed a modified version of the vector carrying an engineered chloroplast tRNA gene. The cloning of synthetic genes carrying internal UGA stop codons results in premature translation termination in E. coli, but in the chloroplast, the tRNA recognizes the UGA as a tryptophan codon (Young and Purton 2016).
As a test of our system, we demonstrated the functional expression of a native bacterial gene (ereB) and of a codon-optimized synthetic gene encoding human growth hormone (hGH). Recombinant hGH is used to treat growth hormone deficiency, Turner syndrome, and several other conditions (Kirk 2012). Glycosylation of hGH is not required for biological activity, and therefore recombinant production in prokaryotic expression systems such as E. coli is preferred (Sockolosky and Szoka 2013). The hormone has also been expressed successfully in the tobacco chloroplast where it was shown to fold correctly, forming the two disulfide bonds within the 22-kDa molecule (Staub et al. 2000). This is in contrast to expression in the E. coli cytoplasm where hGH tends to accumulate within inclusion bodies that require solubilization and refolding (Sockolosky and Szoka 2013). The C. reinhardtii chloroplast also possesses the machinery for the correct folding of recombinant proteins and the formation of multiple disulfide bonds (Tran et al. 2013b), and our finding that hGH is recovered in a soluble and active form further supports this. However, it should be noted that previous studies have indicated that the presence of the disulfide bonds in hGH is not specifically required for biological activity (Kim et al. 2013). Further improvement in the yield of hGH could be achieved by introducing into the transgenic line the nuclear mutation raa-L121G. This mutation blocks the trans-splicing of psaA-exon1 mRNA and has been shown to greatly increase expression of transgenes fused to the psaA promoter/5′UTR (Michelet et al. 2011). Additionally, Ferreira-Camargo et al. (2015) have reported that the inclusion of selenocystamine in the growth medium improves the accumulation of proteins containing disulfide bonds and could therefore also increase hGH yield.
This research was supported by funding from the UK’s Biotechnology and Biological Sciences Research Council (BBSRC grant BB/1007660) and by European Council funding through the Framework Program 7 project GIAVAP (KBBE 2010-3: GA266401). TW was supported by a scholarship from the Royal Thai Government. We thank Andrea Townsend-Nicholson (UCL) for the assistance with the cell proliferation assay and Patrice Courvalin (Institut Pasteur, Paris) for the gift of the pAT72 plasmid.
Compliance with ethical standards
Conflict of interest
The authors declare that they have no competing interests.
Open Access This 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.