The Cascade Recording System

To verify the feasibility of our cascade recording system, we established two series of parts, recorder with a barcode (stgRNA-barcode-cassette series) and recorder without barcode (stgRNA-cassette series). We co-transformed the plasmids with pCas plasmid into Escherichia coli DH5α, and performed several validations and measurements.

Pilot experiment: validations of the knockout effectiveness

We firstly preformed a preliminary concentration gradient experiment on stgRNA-barcode-cassette 1 (BBa_K4630100). After cultivating 22hrs for L-Arabinose and 5hrs for IPTG induction, we diluted the media and spread plate to halt induction and get single colonies (Tbl 1).

The relevant sequence was amplified and sequenced (Fig 1a), confirming a successful knock-out. Large-scale screening revealed that Condition 2 exhibited the highest efficiency, with efficiencies of 70%, 10%, and 38.1% for Condition 2, 3, and 4, respectively (Fig 1b). Non-induction controls substantiated that induction is the prerequisite for recording (Fig 1c).

Concentration matrix and time-gradient test

To get a comprehensive view of the stable relationship between the inducer and the functional consequence, we perform two rounds of concentration matrix (Tbl 2).

The results are documented from aspect of the growth of bacteria, electrophoresis result, sequencing result and the correlation between the two readouts.

The first concentration matrix

It's noted that the induction of arabinose inhibited the growth of bacteria strikingly (Fig 2), and the Lac promoter exhibited significant leakage expression (Tbl 3, Fig 16b). Despite the missing data due to failure of sequencing, E6, E5, C2, B5 performed better (Tbl 4, Fig 3a). However, a quality test based on electrophoresis provided parallel data for randomly picked groups, and the two data access showed a significant correlation, with paired t-test P = 0.7602, no significant difference (Tbl 5, Tbl 6, Fig 4). Given the substantially larger amount of data from the electrophoresis test for E6 (N = 20) compared to sequencing (N = 4), we adjusted the knock-out ratio of E6 to 60%.

The second matrix and statistical analysis

In the second matrix, growth inhibition and leakage expression still occur (Fig 5a, b). Also, an integrated heatmap is plotted based on the mean value of the two matrices (Fig 5c), and the variation pattern of the two matrices showed some kind of correlation. Paired t test result of the two matrices showed a significant difference, with P = 0.0054(**) and mean of differences = -0.2071(Fig 5d). Two-way ANOVA of the second matrix and the average matrix indicate that the IPTG concentration is the main variation factor (Tbl 7).

Time-gradient test on stgRNA-barcode-cassette 1

We performed a time-gradient test to provide guidance for inducing multi-level knockout. The experimental results showed a bell-shaped induction time-knockout ratio curve (Tbl 8, Fig 6). The increasing trend ends on dot 4h and the extension of IPTG induction time does not help the editing afterwards.

Multiple cassette recorder

Firstly, we concluded from the previous statistic data that the multi-level recombination might occur properly in most of the cases.

Indeed, in the matrix tests, we meet several cases of unexpected knock-out (Tbl 9, Fig 7). Theoretically, the larger the homologous arm, the higher the recombination ratio. Interestingly, there are two sets of homologous arms intrinsically flanking the DSB target site. However, the preferred unexpected recombination occurred at the low ratio, 0.028, with a shorter homologous arm. The result demonstrates that the effect of distance is larger than the effect of homologous arm, and both of them are weak. We have more confidence that the change of cassette is sequential.

Then, employing the condition with the highest editing efficiency from the previous experiment (2g/L L-Arabinose and 3g/L IPTG, for 22hrs and 5hrs respectively) with stgRNA-cassette (1+2) (Fig 8a), stgRNA-cassette (1+2+3) (Fig 8b), and stgRNA-barcode-cassette (1+2) (Fig 8c), we achieved successful self-targeted knockout in all sets.

Time-gradient test on stgRNA-cassette (1+2+3)

Consequently, we set up another time-gradient within 2 hours on stgRNA-cassette (1+2+3), to verify the time reliance of the multi-level knock-out (Tbl 10). All of them achieve full knock-out, at a ratio of 100% (Fig 9). However, the time-gradient within 2 hours did not really set up because of inconsistency with the first time-gradient and the small data amount of both. In addition, the Plac leakage was observed once again in the time-gradient test.

Optimization of plasmid pCas

1. Elimination of the sgRNA on pCas

In order to prevent the reduction of activity due to genome knockout in Escherichia coli where the pCas expression system was transformed, we made optimizations to pCas by eliminating the sgRNA segment on the plasmid.

We eliminated the essential sgRNA segment responsible for genome knockout, and subsequently performed PCR amplification of the long sgRNA fragment using specific primers (Fig 1a). Our DNA product was then circularized through Golden Gate Assembly and transformed into competent E. coli DH5α cells. Subsequently, we conducted verification by selecting multiple positive transformants, and both electrophoresis and sequencing analysis confirmed the successful removal of sgRNA (Fig 1b-d).

2. pCasop exhibited superior growth

Then we proceeded to prove the genome knockout capability. Upon induction, we observed that the bacteria transformed with pCasop exhibited superior growth compared to those transformed with pCas (Fig 2a-e). Through specific primers, we successfully verified that the poxb gene's genome did not knockout in the case of pCasop (Fig 2f).

3. Integrated the functional parts of pCas into the recorder

Finally, to make more contribution to synthetic biology, we integrated the functional parts of pCas, the Cas9 and Lambda system, into the plasmid of the recorder device (fig 3).

The prepared Cas, Lambda and backbone DNA sample were used as positive control, as the paired samples differ at several 10 bp.

EL222 Directed Evolution

1. BL21 with pET-28a-[EL222 new-mRFP1] can express mRFP1 when induced by blue light.

Part 1: gene synthesis

We synthesized two gene sequences, blue light sensitive protein EL222(BBa_K2332004), and EL222 controlled promoter pEL222(BBa_K2332002). Terminators and mRFP1 are acquired using PCR from pET-28a plasmid and iGEM 2019 contribute plate3 5G BBa_J04450, respectively (Fig 1). Subsequently, they are cloned in pET-28a vector(Fig 2) and expressed in E. coli BL21(DE3).

Part 2: Blue light induction pre-experiment

Bacteria are spread on LB plates with kanamycin and first cultured under 37℃ for 10 hours, then exposed to 400 Lux blue light around 475nm for 20 and 22 hours with half of the plates covered. Blue lights are controlled using Arduino Mega 2560. On the slides we made, the pET-EIN-mRFP-Golden plasmid transformant exhibits stronger red fluorescence compared to the negative control after 20 hours induction (Fig 3).

Part 3: measurement of mRFP1 expression level

We use spectrophotometry to measure the expression level of mRFP1 at protein level (Fig 4). We measured their absorbance at 540nm and 600nm, which present for mRFP1 and bacteria concentration respectively. The OD540/OD600 at a specific time for a sample culture was determined after subtracting from each of the technical triplicate readings of the negative control cultures (fluorescence free) at the same time.

Part 4:Bacteria with different pEL222 expressed different amounts of mRFP after induction

We acquired different sequence of pEL222 by error-prone PCR and incorporated the generated sequences into pET-28a-[EL222-mRFP1] to replace the original sequence of pEL222. The plasmid was transformed in BL21(DE3) directly. We sequenced the PCR products (Fig 6) to confirm the success of error-prone PCR.

To ensure the expression of EL222 protein, We performed the pre-experiment first. We used IPTG to induce the expression of mRFP in pET-EIN-mRFP-Golden plasmid instead of the blue light induction. After 22 hours induction of IPTG, Colonies with different pEL222 show differencet expression of mRFP1, resulting in disparate colors (Fig 7). This work demonstrates that different pEL222 has different sensitivity to the inducer.

Moreover, we measured the sensitivity and leakage of different pEL222 by debugging different blue light induction times (Fig 8). We cultured all the bacteria in PCR tubes and measure the expression level of mRFP following the same steps in part 3.

sgRNA Screening

1. Construcrion of pUC57-N20s-gRNA+HA

N20 sequence can be loaded onto pUC57-N20s-gRNA+HA. The plasmid synthesis was executed in collaboration with Atantares. Employing Golden Gate Assembly, each plasmid can be endowed with a distinct N20 sequence. The assembly process was successfully executed in conformity with our design, demonstrating the feasibility of the concept of BREM.

2. Knockout for N20s sequence

pUC57-N20s-gRNA+HA with pRed_cas9_△poxb300 in E.coli DH5α can achieve N20s fragment knockout.Following a 24-hour induction with 2g/L arabinose, we determined the edited colonies via colony PCR. Our analyses revealed successful N20s sequence deletions for some of the tested N20 sequences, exemplified by NO. 11, 13, 14, and 15 (Fig 2), demonstrating the feasibility of the knock out system.

3. Construction of pUC57-EGFP-N20s-HA+gRNA

pUC57-EGFP-N20s-HA+gRNA is constructed by Gibson Assembly and Restriction Cloning.

Application

In order to validate the functionality of our recorder in various scenarios, we designed to replace the Pel222 in the stgRNA-cassette system with Pphsa, PluxR, and PcaU activated promoter, which are responsive to biomarker of IBD, biomarker of PA infection, and product of polyethylene terephthalate degradation, respectively. Therefore editing can be initiated by the promoters mentioned above. We synthesized the required DNA fragments in collaboration with biotech companies but received only the sequence for the PluxR system before wiki freeze. We conducted induction experiments with a concentration gradient and time gradient for the PluxR system, however, we were unable to obtain the necessary experimental data due to lack of time.