Nuclear Transformation There are a variety of established transformation meth­ods to integrate heterologous DNA into the nuclear genome including particle bombardment (Debuchy et al. 1989; Kindle et al. 1989; Mayfield and Kindle 1990), agitation of cell-wall-deficient strains with glass beads (Kindle 1990), electropor­ation (Shimogawara et al. 1998), agitation with silicon carbide whiskers (Dunahay 1993) and biologically mediated gene transfer by Agrobacterium tumefaciens (Kumar et al. 2004). In C. reinhardtii, transformation of the nuclear genome occurs by random insertion through non-homologous end joining (Tam and Lefebvre 1993) or by using linear DNA that promotes the insertion of multiple copies in one locus (Cerutti et al. 1997b). Phenotypic and genetic screening of transformants can minimise undesirable non-target effects of the random insertion of a transgene that can lead to disrupted genes or regulatory elements. In return, this effect is used to study genes of unknown function using high-throughput insertional mutagenesis (Dent et al. 2005). In C. reinhardtii, targeted gene integration through homologous recombination (HR) using single-stranded transforming DNA (Zorin et al. 2005,

2009) is possible, but only at low frequencies to date. However, high rates of homologous recombination have been reported in another green algal species Nannochloropsis (Kilian et al. 2011), showing promise for reverse genetics and targeted gene knockouts.

Random and Insertional Mutagenesis The creation of mutants of an organism by the use of irradiation or chemical mutagenesis is the classic approach pioneered by Morgan in Drosophila and used for a century to study the effect of significant alteration in the behaviour of a gene, typically by partial or total deletion. These methods produce base pair changes leading to a range of disturbances including altered amino acid sequence, small deletions, truncations, frameshifts and splicing defects; the resultant mutants include temperature-sensitive and dominant mutants as well as functional knockouts sometimes giving rise to complex phenotypes. In the last 3 decades, the insertion of foreign transgenes (usually carrying a marker to allow selection and identification) has also been well established, as described above. Insertion mutants usually have knocked out or disabled genes, though insertion into promoters can lead to alterations in expression levels and splicing. Gene inactivation is largely on a random basis (Zhang et al. 2014) across the genome (at least where the chromatin structure is sufficiently open) which allows the unbiased identification of genes relevant to biological processes and has been very useful for research into biological mechanisms and physiology in Chla — mydomonas and other algae. The difficulty is usually that a specific phenotypic screen is needed to identify relevant genes. Lethal mutations will not be identified, nor will mutants in genes which are redundant or which show no overt phenotype under the conditions of the screen. Of the estimated * 15,000 genes in the Chla- mydomonas genome, only a few hundred have been reported in the literature, and many genes known to be important are not represented in collections of mutants. A good example is the multigene family of LHC genes. An insertion mutant in a single LHC gene will usually not produce any easily measurable phenotype, due to compensation by other LHCs, while the highly specific physiological or genetic tests needed to demonstrate the loss of a specific LHC gene are not suitable for screening assays. Finally, each transformation produces only a few hundred colo­nies, with very likely a biased set of genes being affected. This makes it a major project to uniquely identify mutants of each gene in an alga. Fortunately, this is being performed at Stanford University, and the resulting collection of mutants will be an invaluable research resource for the Chlamydomonas community (Zhang et al. 2014). However, it is unlikely that this resource will be duplicated for every species of algae of research or commercial interest.

Homologous Recombination (HR) in the Nucleus Homologous recombination, the recombination between homologous DNA sequences, is essential for eukaryotes to repair DNA double-strand breaks and introduce genetic diversity during cell division, and two main pathways (‘double-strand break repair’ and ‘synthesis — dependent strand annealing’) have been proposed (Sung and Klein 2006). In plants and algae, nuclear located RecA homologues show high similarity to the prokary­otic RecA genes which suggests an endosymbiotic transfer from mitochondria and chloroplasts to the nucleus of ancestral eukaryotes (Lin et al. 2006). Although the introduction of foreign DNA into the nucleus occurs predominantly via random insertional mutagenesis, successful targeted homologous recombination has been reported in several algal species such as C. reinhardtii (Gumpel et al. 1994; Sodeinde and Kindle 1993; Zorin et al. 2009), Nannochloropsis sp. (Kilian et al.

2011) and Cyanidioschyzon merolae 10D (Minoda et al. 2004) with as little as 230- bp DNA sequence homology in the haploid cell (Gumpel et al. 1994). It has been demonstrated that the introduction of single-stranded repair DNA leads to a more than 100-fold reduction of non-homologous DNA integration in comparison with double-stranded DNA (Zorin et al. 2005). Attempts to increase the low frequency of homologous recombination in plants by the over-expression of well-characterised enzymes involved in homologous recombination such as the recA and ruvC pro­teins have been reported to increase homologous recombination and double-strand break repairs; however, these reports suggest that foreign gene targeting is not improved (Reiss et al. 1996, 2000; Shalev et al. 1999).

RNA Interference (RNAi) In the absence of tools for precise genome manipula­tion, RNAi-mediated knock-down of gene expression enables the creation of highly specific research mutants without the need for phenotypic screening or selection. RNAi techniques also enable the study of reduced levels of gene expression where total ablation would be lethal. The phenomenon of RNA interference is produced by the action of specific cellular machinery [Dicer and Argonaute (AGO) proteins (Casas-Mollano et al. 2008; Cerutti and Casas-Mollano 2006)] on mRNAs, guided by microRNAs which are naturally occurring small double-stranded RNAs (dsR — NAs) in a process called mRNA cleavage (Bartel 2004). RNAi represents the use of this natural process for experimental ends by the artificial provision of small RNA molecules designed to interfere with the expression of a target gene. MicroRNAs (miRNAs) have also been discovered in C. reinhardtii (Molnar et al. 2009; Zhao et al. 2007) enabling a highly specific genetic tool (Moellering and Benning 2010; Molnar et al. 2009; Schmollinger et al. 2010; Zhao et al. 2007). Difficulties in achieving direct nuclear gene knockout via homologous recombination for C. reinhardtii (Nelson and Lefebvre 1995) led to RNAi becoming a widely used method to accomplish post-transcriptional gene silencing for gene function dis­covery (knock-down approaches via reverse genetics) and metabolic engineering (Fuhrmann et al. 2001; Molnar et al. 2009; Rohr et al. 2004; Schroda et al. 1999; Zhao et al. 2009). Artificial microRNAs have also been engineered to create functional knock-downs of several nuclear genes in Dunaliella salina (Sun et al. 2008) and in the diatom Phaeodactylum tricornutum (De Riso et al. 2009). This method could potentially be used for algal genetic engineering for biofuel pro­duction (Cerutti et al. 2011; Grossman 2000; Wilson and Lefebvre 2004). The weakness of RNAi is the need to maintain the expression of the RNAi construct, typically requiring ongoing selective pressure, for example with antibiotics.

Genome Editing The ability to precisely edit the genome enables the specific deletion or mutation of genes and regulatory regions, with the resultant ability to create targeted mutations for study or industrial applications. As the resultant mutations are permanent, the stability problems inherent in RNAi knock-down constructs are eliminated. Genome editing also offers the prospect of precise mutants lacking foreign DNA, which consequently are technically non-GMO organisms. Several systems of genome editing have been developed in recent years, including zinc finger nucleases (ZFN), transcription activator-like effector nucleases (TALENs), meganuclease and the CRISPR/Cas system. Both ZFN (Sizova et al. 2013) and TALENS (Gao et al. 2014) have been used for genome editing in Chlamydomonas. The drawback of these systems is the substantial investment of time and effort required. In contrast, the CRISPR/Cas system promises a simpler, more facile approach. CRISPR (clustered regularly interspaced short palindromic repeats) sequences are found in prokaryotes and are short repetitive DNA sequences, corresponding to part of the DNA of bacteriophage genomes. In con­junction with a specific nuclease (Cas), they form the basis of a prokaryotic ‘immune system’ to recognise and eliminate viral genomes. The CRISPR RNA acts as a guide for Cas-mediated cleavage of the viral DNA or provirus. For genome editing, first described in 2013 (Cho et al. 2013; Mali et al. 2013), the addition of a targeted ‘guide RNA’ with co-expression of the Cas9 nuclease enables precise and specific genome editing and has been rapidly adopted in many organisms including both animals (Hsu et al. 2014) and plants (Feng et al. 2013). Although first attempts to use CRISPR/Cas in algae have encountered difficulties, it is anticipated that these will be overcome in the near future, enabling rapid and flexible genome editing in algae.

Chloroplast Transformation The chloroplast is the site of photosynthesis and storage of the resultant starch and is also important for the production of fatty acids and photosynthesis-related pigments especially carotenoids. Transformation of the chloroplast requires transfer of the transforming DNA to the interior of the chlo — roplast, and consequently, particle bombardment (biolistics) using DNA-coated gold or tungsten particles is the most commonly used method for chloroplast transformation. Some particles penetrate the cell wall, plasma and chloroplast membrane and deposit the transforming DNA in the plastid, which can then inte­grate into the local genome through homologous recombination (Boynton et al. 1988). Flanking endogenous sequences that are homologous to the targeted inser­tion site can make chloroplast transformation events highly targeted to any region in the genome (Rasala et al. 2013) which in C. reinhardtii is a great advantage over nuclear transformation where homologous recombination occurs only at low effi­ciency (Sodeinde and Kindle 1993).

Homologous Recombination in the Chloroplast In the chloroplast, homologous recombination is mediated by an efficient RecA-type system which, due to its homology to the Eschericia coli RecA system, is suggested to be related to the cyanobacterial ancestors of chloroplasts (Cerutti et al. 1992, 1995; Inouye et al. 2008; Nakazato et al. 2003). In cyanobacteria, the RecA system is essential for cell viability especially under DNA damaging conditions (Jones 2014; Matsuoka et al. 2001). DNA repair is also believed to be the main function for homologous recombination in the chloroplast (Cerutti et al. 1995; Rowan et al. 2010). Several different models exist for the precise mechanism of homologous recombination and are not presented in detail here; however, the main steps of prokaryotic homologous recombination include initial strand breakage, formation of an enzyme complex to unwind the double-stranded DNA to form a single strand, followed by a mechanism called ‘strand invasion’ which searches for similar sequences on a homologous DNA fragment for pairing. This is then followed by DNA synthesis in accordance with the new template strand and resolution of the structure (Amundsen et al. 2007; Smith 2012).

Genome Stability in the Chloroplast With 50-100 genome copies, the chloroplast is highly polyploid, and apart from a few exceptions, each plastome shows a tet — rapartite organisation containing two inverted repeat sequences that are mirror images of one another, separated by a large and a small single-copy unit. Though absent in present-day cyanobacteria and not essential for the general chloroplast genome function, the inverted repeat sequences show properties which suggest their involvement in gene maintenance and increased genome stability (Goulding et al.

1996) . Chloroplast genomes can undergo homologous recombination between the inverted repeat sequences as several studies have shown two populations of plast — omes in the same organism differing only in the single-copy sequence orientation (Aldrich et al. 1985; Palmer 1983; Stein et al. 1986). Furthermore, it has been observed that the inverted repeat regions accumulate nucleotide substitution muta­tions 2.3 times more slowly than the single-copy regions (Perry and Wolfe 2002; Ravi et al. 2008; Shaw et al. 2007). It was also demonstrated that DNA rearrange­ments occur more frequently when a large inverted repeat sequence is lost (Palmer and Thompson 1982). This suggests that homologous recombination between plastomes and inverted repeats contributes to this increased genome stability.

Manipulation of the chloroplast genome via homologous recombination offers the potential for exact gene deletions and insertions. Generally, chloroplast trans­formation vectors are E. coli plasmid derivatives carrying the foreign DNA flanked by DNA sequences (>400 bp each) which are homologous to the target region of the plastid DNA (Bock 2001; Hager and Bock 2000). Due to the high ploidy of the plastid genome and the fact that initially only a single plastome copy is transformed, the resulting phenotype may be weak in primary transformants and it is important to establish an efficient and suitable selection system or strategy to identify and enrich cells containing transformed ptDNA copies, which may easily revert to wild type (Hager and Bock 2000; Rasala et al. 2013). Once the genome is homoplastomic and thus lacking template for further undesired homologous recombination events, the selectable marker can be removed (Day and Goldschmidt-Clermont 2011), though selective pressure against the transgene can still exist.

Mitochondrial Transformation Respiration and photosynthesis are coupled pro­cesses, and mitochondrial mutations are known to affect photosynthesis (Schonfeld et al. 2004) as well as many other aspects of cellular metabolism. Although in Chlamydomonas homologous recombination in mitochondria DNA is only detected after crosses between different mating types in mitotic zygotes (Remacle et al.

2012) , it demonstrates that the cellular machinery is available for recombination — based foreign gene integration. Although the first transformation of the mito­chondrion was published in 1993 (Randolph-Anderson et al. 1993) using biolistic bombardment, reports of mitochondrial transformation are rare and initially limited to the restoration of the wild-type mitochondria genome in mutant strains. In 2006, Remacle et al. reported the first and, to date, the only modification of the mito­chondrial genome in vivo in a photosynthetically active organism. The work pre­sented the introduction of a nucleotide substitution in the cob gene, conferring resistance to myxothiazol, and an internal deletion in nd4 (Remacle et al. 2006). Homologous recombination was facilitated with as little as 28-bp homology between the introduced and endogenous DNA (Remacle et al. 2006).

Selection Techniques Following Transformation The identification of chloroplast and nuclear transformants can be achieved through the rescue of mutants impaired in endogenous cellular functions (non-photosynthetic, flagellar or auxotrophic mutants) or the introduction of antibiotic and herbicide resistances by point mutation of endogenous genes or the expression of heterologous resistance genes. A range of auxotrophic mutants have been developed (including mutations in arginine biosynthesis, nitrate reductase, nicotinamide biosynthesis and thiamine biosynthesis (Debuchy et al. 1989; Ferris 1995; Kindle et al. 1989; Rochaix and Vandillewijn 1982)) as well as mutants in photosynthetic capability or flagellar motility (Diener et al. 1990; Mayfield and Kindle 1990; Mitchell and Kang 1991; Smart and Selman 1993).

Classical antibiotic resistance markers include kanamycin [chloroplast: (Bateman and Purton 2000); nuclear: (Hasnain et al. 1985; Sizova et al. 2001)], spectinomycin/ streptomycin [chloroplast: (Goldschmidt-Clermont 1991); nuclear: (Cerutti et al. 1997a)], neamine/kanamycin and erythromycin [chloroplast: (Harris et al. 1989)] paromomycin and neomycin [nuclear: (Sizova et al. 2001)], cryptopleurine and emetine [nuclear: (Nelson et al. 1994)], zeocin and phleomycin [nuclear: (Stevens et al. 1996)] and hygromycin B [nuclear: (Berthold et al. 2002)]. Screening for resistance to herbicides, such as atrazine, which inhibits the function of photosystem II [chloroplast: (Erickson et al. 1984)], or Basta, leading to disruption of the chloroplast structure [chloroplast: (Cui et al. 2014)], is also possible. The use of antibiotic resistance markers is not problematic for research purposes, but is less desirable for industrial-scale production of vaccines or therapeutics, and for large — scale outdoor or agricultural production, antibiotic-free media is usually required for both sociopolitical and socio-economic reasons.

Recombinant Protein Expression Regardless of the energy return on investment (EROI) of microalgal biocrude, the low net value poses a significant problem for economic biocrude production from microalgae. One potential strategy to offset this is to exploit the fact that high-value products such as recombinant proteins are typically only a small fraction by weight of the biomass and could potentially be extracted prior to thermochemical processing of remaining biomass to biocrude. Consequently, an additional and lucrative revenue stream can arguably be obtained to subsidise the process without sacrificing overall biofuel productivity. The demand for recombinant proteins is growing with increasing population and bio­technological applications, and ultimately, the expression systems with the greatest cost-benefit are likely to dominate these markets. In this context, microalgal sys­tems offer significant advantages over traditional microbial fermentation or mam­malian cell culture systems. So-called molecular ‘pharming’ is the production of pharmaceutical proteins, therapy peptides, vaccine subunits, industrial enzymes and secondary metabolites or other compounds of interest in plants, and algae also offer commodity-scale opportunities. Proteins with biopharmaceutical or biotechnologi­cal relevance can be, e. g. monoclonal antibodies, vaccines, blood factors, hor­mones, growth factors and cytokines.

Nucleus Although the chloroplast is the expression system of choice for high protein expression levels, the nucleus as expression location has its advantages which have to be carefully considered in the context of each product. Despite the comparatively low expression rate of 1 % of TSP claimed by commercial vendors and the risk of gene silencing, nuclear expression tools enable the fusion of the desired product to a selection marker (to avoid silencing) with subsequent self­cleavage, as well as secretion of the expressed compound into the media (Lauersen et al. 2013; Rasala et al. 2012). So far, most proteins expressed from the nucleus have been reporter genes used to quantitatively measure protein expression levels as well as the localisation and accumulation (e. g. within the C. reinhardtii cell) (Rasala et al. 2013). Six fluorescent proteins (blue mTagBFP, cyan mCerulean, green CrGFP, yellow Venus, orange Tomato and red mCherry) were expressed from the nuclear genome to allow protein detection in whole cells by fluorescence microplate reader analysis, live-cell fluorescence microscopy and flow cytometry.

Chloroplast The production of recombinant proteins including reporters (e. g. GUS, luciferase or GFP) (Franklin et al. 2002b; Mayfield and Schultz 2004; Minko et al. 1999; Sakamoto et al. 1993), protein therapeutics (antibodies, hormones, growth factors or vaccines) (Franklin and Mayfield 2005; Manuell et al. 2007; Mayfield and Franklin 2005; Mayfield et al. 2003; Rasala et al. 2010; Surzycki et al. 2009) and industrial enzymes has established the chloroplast as a suitable production platform for a broad range for protein candidates. The potential for high expression levels [over 40 % of TSP (Surzycki et al. 2009)] and the ability to form complex proteins including disulphide bonds (Tran et al. 2009) allow for limited post­translational modification without interference caused by ‘gene silencing’ making the chloroplast the site of choice for protein expression. Furthermore, the pro­karyotic characteristics of the chloroplast genome such as gene organisation into operon-like structures (Holloway and Herrin 1998) potentially allow the expression of multiple transgenes from a single operon (transgene stacking) and with this the introduction of complete biochemical pathways. The additional development of a Gateway-compatible transformation system (Oey et al. 2014) allows the rapid production of multiple transformants. Success rates and costs for successfully establishing expression of new proteins in C. reinhardtii are estimated to be similar to those of yeast and mammalian cells (Rasala et al. 2013), but with potentially much lower production and purification costs.