Natural Competence
As mentioned in our description, Synechococcus elongatus UTEX 2973 is no longer naturally competent, presumably due to a point mutation in the pilN gene ( Li et al., 2018 ), which means that when genetically engineering this organism other ways to introduce exogenous DNA have to be taken into consideration. This is mainly done through electroporation or conjugation - especially triparental conjugation (Yu et al., 2015) . Triparental conjugation into the UTEX 2973 strain is typically performed with two E. coli HB101 strains, one harboring the pRL443 and one harboring the pRL623 plasmid. The latter strain is then again transformed with the plasmid of interest, the prior is used as the conjugal strain - both have to be incubated together with the cyanobacteria for the conjugation to take place (Wendt et al., 2016)
To overcome this time-consuming process, we planned to reintroduce natural competence into our strain. It was already shown, that this can be done by integrating an intact copy of the pilN gene into one of the neutral sites (Li et al., 2018) , though this technique is not ideal: you have to add an antibiotic cassette in order to keep selective pressure on the bacteria, so that they integrate the new gene into all the chromosome copies. This antibiotic resistance will persist in the strain, meaning that when further engineering the organism later on, this one resistance can not be used e.g. in vectors for transient expression - a huge downside. Furthermore, one of the neutral sites has to be used, resulting in a strain that has less neutral sites available for further introduction of genes.

Although we did not prefer this method, we still tried it, as we were not sure, if our other approach would prove to be successful. We also used extensive bioinformatic tools to identify new integration sites in UTEX 2973, which can be used if one were to reintroduce natural competence in the above mentioned way. Additionally, we came up with a plan to revert the point mutation in the pilN gene with a CRISPR/Cas12a system.

This approach is promising, as the integration of the new pilN copy only enabled a low efficiency of natural transformation, which might be due to the point mutation negatively affecting expression of the pil0 and pilQgenes laying downstream of pilN (Li et al., 2018 ; Barten and Lill, 1995) . As CRISPR/Cas12a allows accurate targeting of genetic sequences, we designed a crRNA leading the Cas12a protein to the pilN locus. The repair template was taken from the S. elongatus PCC 7042 genome, where the gene is still intact, allowing the cell to repair the cut introduced by Cas12a accordingly, reversing the point mutation, which leads to an intact copy of pilN again - a more elegant approach than simply inserting a second copy of the gene. As our own CRISPR system was still in building at that point, we had to rely on pSL2680, a replicating base vector for constructing CRISPR/Cas12a editing plasmids by Ungerer and Pakrasi, 2016 .
We followed their protocol (available here on Addgene) , annealing oligos to construct the crRNA. Small overhangs were added to enable the ligation into the AarI-digested vector, where a lacZ cassette was replaced, which allowed for blue/white screening of recombinant colonies. This cloning step alone took approximately a week. Additionally, the repair template had to be constructed by PCR with added overhangs for the following Gibson reaction. As stated, it was taken from the S. elongatus PCC 7942 genome. It was designed in such a way that the point mutation inside the UTEX 2973 genome was part of the PAM sequence for Cas12a, meaning that the repair template did not include the PAM and would not be cut by the enzyme.

After the successful Gibson assembly of crRNA and repair template into the Cas12a carrying vector, nearly two weeks had passed, indicating that working with this vector can be quite tedious and time consuming. This is one of the many reasons why we chose to implement such a CRISPR system into our MoClo based toolbox. While building this system we made sure to directly prove it by using it to reverse this point mutation, making sure that we tackle this crucial goal through multiple approaches.

CRISPR gene editing)
CRISPR/Cas systems are powerful tools that have gained a lot of popularity in the recent years. As they can be used for a wide array of applications - like the integration of whole genes, alteration of single nucleotides, knock-outs of whole genetic regions, as well as the use of the DNA-binding property in a multitude of applications through so called deadCas systems, where the Cas protein does not exhibit nuclease activity (Hsuet al., 2014) - we were eager to implement such a system into our own toolbox. Diving into the literature we noticed many different systems are available,the most commonly used one being CRISPR/Cas9, and we began to wonder which of them we should use.
In our description we presented CRISPR/Cas9 and CRISPR/Cas12a, showing the differences of these two systems. Looking deeper into CRISPR/Cas12a we noticed a few advantages that finally led us to choose it as our preferred system. As the sgRNA used as a guide for Cas9 is usually ~100nt long, chemical synthesis is more complex and expensive in comparison to the ~43nt needed for the Cas12a guiding crRNA (Swarts and Jinek, 2018) - an unpleasant fact, especially for iGEM teams that do not have many resources available to them, but the main reasons we chose Cas12a are others. Multiplexed gene editing is one of the key features of these CRISPR/Cas systems, but how to actually apply it differs:
For Cas9 each sgRNA is in need of its own promoter, which means that they have to be expressed from different vectors or a multi cassette vector ( X. Ma et al., 2015 ; Z. Zhang et al., 2016 ). In contrary, multiplexed genome editing with Cas12a can be achieved simply by expressing all of the needed guide RNAs in one transcriptional unit, where they are then processed into different crRNAs by Cas12a (Kim, et al., 2016; Nishimasu et al., 2017) . This is a huge advantage of Cas12a. Furthermore, CRISPR/Cas9 was shown to be toxic in cyanobacteria (Wendt et al., 2016) , which is one of the foremost reasons CRISPR technologies have not been widely applied in cyanobacteria - the usage of Cas12a though, does not seem to have the same toxicity Ungerer and Pakrasi, 2016 , making it the ideal candidate for the Green Extension of the Marburg Collection.
The actual implementation of the CRISPR/Cas12a system into our toolbox necessitated a well thought out plan. The design of our CRISPR/Cas12a system was mainly affected by the fact that we wanted to have a convenient and rapid tool for genomic manipulation. The lvl 0 part (What are parts? Read more about it click here!) of the Cas12a protein was created via PCR amplification from the plasmid pSL2680, but special overhangs were added in order to clone the PCR product into a lvl 0 acceptor vector. The part was introduced as a coding sequence (CDS) part in the MoClo standard to be included in the Green Expansion of the Marburg Collection. The lvl 1 part of the Cas12a protein was equipped with a rather weak promoter so that the toxicity caused by overproduction of the endonuclease could be kept low. The parts used for lvl 1 assembly were: pMC0_1_03 + pMC0_2_03 + pMC0_3_07 + pMC0_4_33 + pMC0_5_07 + pMC0_6_17. For the construction of the crRNA part the design of the plasmid pSL2680 was mainly maintained, but the lacZ cassette was replaced by a GFP cassette to enable easier screening of crRNA assembly and to reduce expenses for X-Gal/IPTG. It was constructed as a part reaching from the RBS site to the end of the terminator site. As the whole system is built for modular cloning in PhytoBrick syntax, it is possible to freely exchange the parts around the Cas12a and crRNA parts - in this way the amount of crRNA/Cas12a can be controlled by choosing promoters with different strengths. Our initial plan was to synthesize the crRNA with the desired overhangs, but as the sequence contains multiple direct repeats, it was not possible for providers to synthesize this construct, which is why we split it into four different parts that then had to be assembled. For this assembly the four parts were first cloned into the pJET1.2/blunt vector by Thermo Scientific and then digested with BsaI while the acceptor vector was digested with BsmBI. In this way the final vector still contains BsaI recognition sites, so that it can be used in a level 1 assembly Golden Gate reaction. The cloning of the level 2 part with this crRNA part was done by ending with a ligation step to make sure the GFP dropout remains in the vector.

Cyanobacterial shuttle vectors

Following our dream to create the most versatile, MoClo compatible shuttle vector for cyanobacteria we made sure to pay attention to detail. When creating new shuttle vectors, one of the most important points to consider is the replication element that is being used, mainly due to the wanted copy number and a phenomenon called plasmid incompatibility. Plasmids harboring the same replication or partitioning system can often not be stably maintained in a cell - they are incompatible (Novick and Hoppenstaedt, 1978) . With multiple different plasmids bearing the same replication elements, the replication machinery will randomly choose which plasmids to replicate, leading to one of the different plasmids being copied more frequently than the other (Thomas, 2014) . As we used the minimal replication element from the pANS plasmid of S. elongatus in our shuttle-vectors, in order to have a native origin of replication, we had to consider such plasmid incompatibilities and made sure to cure our strain of the endogenous pANS - which we could successfully prove.
The next step was the creation of our own, modular shuttle vector. For this we had to pay attention, as we had to fit it to the PhytoBrick standard in order to use it in our Green Extension of the Marburg Collection. This means that we had to remove some restriction enzyme cutting sites at multiple points in the sequence: In repB and repA - both a CDS of the minimal replication element needed for the vector - lay one BsaI recognition site each, which were removed by introducing a silent point mutation. This point mutation, in both cases, introduced a synonym codon for glutamic acid, so they should not cause any issues later on.
A BsmBI site was found within the non-coding sequence of the minimal replication element, meaning that this had to be changed with a more careful approach, as any change could have heavy influence on secondary structure and potentially impair the function. Due to this reason we made sure to try all possible variants of mutations to remove the recognition site of the restriction enzyme In order to assemble our desired part we synthesized different parts of it with the mutations we introduced. Due to the length and complexity of the sequence we had to divide the synthesis of the minimal replication element into three parts that later had to be fused together. Additionally, we wanted to implement a reporter for easy selection. We chose rfp, which was amplified out of the Lvl0_8_Amp/ColE1 part from last years Marburg Collection in addition with the ColE1 ori that can be found on it. This means that our vector does not just contain the cyanobacterial ori of our strain, but also a high copy origin for replication in E. coli (Gerhart et al., 2002) . As an antibiotic cassette we chose spectinomycin, which we also amplified by PCR, this time from pAM4787 (Chen et al., 2016) . Finally, those five fragments - the three parts of the minimal replication element, the ColE1 ori & rfp cassette and the spectinomycin resistance cassette (aadA) - were fused together in a Gibson reaction, resulting in BBa_K3228069 (sometimes also called lvl 1 ori), the first cyanobacterial shuttle vector for cloning lvl 1 constructs in a modular way. This part has two BsaI sites that flank the rfp cassette, so that this genetic element will be exchanged with the other parts in a lvl 1 Golden Gate reaction.

In addition to this, we created a second shuttle vector, this time for cloning lvl 2 constructs: This vector has mostly the same design as BBa_K3228069, but the rfp cassette is flanked by BsmBI sites, enabling the construction of lvl 2 vectors.

Furthermore this part bears a kanamycin resistance cassette instead of the spectinomycin resistance of the lvl 1 ori. This part was assembled in a four part Gibson reaction, as in addition to ColE1 and the rfp cassette also the kanamycin resistance cassette could be amplified via PCR, in this case from pYTK_0_84, a plasmid from the Dueber MoClo Yeast Toolkit, resulting in BBa_K3228089 (sometimes called lvl 2 ori). For all these cloning processes special overhangs had to be added for Gibson Assembly.