Mycosporine-like Amino Acids (MAAs) are a family of UV-absorbing natural products found in diverse organisms, especially in those that are exposed to extensive solar radiation such as cyanobacteria in ocean. Demand for novel,natural sunscreens is growing due to the potential negative effects on human health (endocrine disruption et.) and environmental (coral reef bleaching et.) of synthetic sunscreens. The diversity of MAAs allows them to absorb different wavelengths of UV radiation. The current method to obtain MAAs is through extraction from MAA-containing organisms such as nori; however, this method is extremely inefficient and costly. This year, LINKS_China attempted and succeeded in establishing and optimizing a series of metabolic pathways in S. cerevisiae to produce four molecules as the novel sunscreen. We aimed our experiments at three aspects. Firstly, we increased the production of S7P and 4DG, the precursors of most MAAs, by deleting endogenous TAl1 and introdcing exogenous xylose-utilizing pathway . Secondly, we combined different AGL and Alal ligases from three organisms to produce shinorine and porphyra-334 from 4DG, then choose and optimize the most effective combination. Lastly, we produced palythine and gadusol based on our previous genetic circuit. All gene expression is ultimately performed in the S. cerevisiae genome, a more stable and efficient expression than plasmid expression. In order to provide a more secure protection for our skin, we produced hSOD-linker-catalase in order to break down reactive oxygen species(ROS) caused by UV radiation that bypasses the barrier of MAAs.
Sedoheptulose 7-phosphate(S7P) is an intermediate of the pentose phosphate pathway, one of the alternative biosynthetic pathways of MAAs. Naturally, S7P is converted into saccharides for the later glycolytic pathway in S. cerevisiae. For more S7P to be utilized by the MAAs pathway, we decided to block the flow to glycolytic pathway by removing the endogenous TAL1. We employ the gene editing methods studied by Apel et. al. to construct a new Cas9-sgRNA plasmid pCRCT-TAL1 for the deletion of gene TAL1 (Fig.2A). By performing colony PCR (Fig.2B) and sequencing(Fig.2C), we found that we successfully removed 32 bp in TAL1's coding sequence. The deletion of the fragments will result in a frameshift mutation that renders TAL1 non-functional. After the removal of pCRCT plasmids from the cell, we obtained our first engineered strain: SC.L1.
Another method to increase the S7P levels is the introduction of xylose-utilizing pathway that provides an alternative carbon source for the pentose phosphate pathway. Xyl1, Xyl2, and Xyl3 genes from Scheffersomyces stipitis proved to be successfully heterologous expressed for efficiently xylose-fermenting in S. Cerevisiae. We used yeast toolkit (Lee, 2015) to assemble the Level 1 plasmid concluding the promoter, coding sequence, and terminator. Xyl1, Xyl2, and Xyl3 are under the control of promoters pTDH3, pPGK1, and pTEF2 respectively which have a high and stable transcription efficiency (Apel, et. al, 2016).Then we used PCR amplification to obtain the DNA fragment Xyl1, Xly2, Xly3 from Level1 plasmid, and two homology arm(LA and RA)from yeast genome(Fig.3B). The five fragments are transformed into the SC.L1 strain along with a pCRCT-His3 plasmid that will cut open the yeast genome at the His 3 position. The recombinant strains were identified by pcr and sequencing(Fig.3C and D), and the pCRCT plasmid was discarded using URA nutrient deficient medium. We name the strain with Xyl 1, 2, 3 inserted at His3 position SC.L2.
The recombinant strains SC.L2 ensured that S7P was sufficient for subsequent production.The next engineering is to convert S7P to 4DG efficiently. We selected the two genes from cyanobacteria N. punctiforme encoding DDGS(NpR5600) and OMT(NpR5599) for homologous recombination into SC.L2 genome. We used the strong promoters pTDH3 for DDGS and pPGK1 for OMT and constructed Level 1 plasmids. We chosed position 308 from chromosome III (Apel et. al., 2016) for genome recombinant. Using similar methods, We transformed pCRCT-308 plasmid, the DNA fragments of homologous arms, DDGS, and OMT into SC.L2. The recombinant strains were identified by PCR and sequencing(Fig.4C and D) , SC.L3 was obtained.
To verify the previous engineering on yeast strains works, four strains(Wild Type, SC.L1, SC.L2, and SC.L3) were cultivated in four different media containing a different ratio of glucose to xylose. We found that in the medium containing only glucose(2%, 20 g/L), the growth curves of the 4 strains are roughly identical(Fig.5A), implying that our modification had no impact on the ability of cells to use glucose for growth when glucose is abundant. In the medium containing 1% glucose and 1% xylose, SC. L2 and SC.L3 containing Xyl1/2/3 showed a slight growth advantage compared with WT and SC.L1, which suggests xylose may be used as a potential carbon source for cell growth in yeast. As the concentration of glucose decreased and xylose increased further, the cell growth of all four strains is inhibited, especially SC.L3 and SC.L4(Fig. 5C and D). This result indicates most xylose is converted into S7P for MAAs production instead of being used for growth when xylose is abound and the xylose-utilizing pathway is introduced. Furthermore, the conversion of xylose into S7P requires the usage of ATP and NADPH, which might also affect growth. Therefore, we concluded that we need to supply xylose to generate S7P and maintain certain levels of glucose as the carbon source for cell growth. We chose the media containing 1% glucose and 1% xylose for later yeast fermentation.
Production
After S7P is converted to 4DG by DDGS and OMT, we set producing shinorine and porphyra-334 as our primary goal. These two substances are produced by ATP-grasp ligase (AGL) and D-Ala-D-Ala ligase (ALAL) by two enzymatic steps. First, 4DG is converted to mycosporine-glycine(MG) by conjugating glycine to 4DG under the action of AGL. And then, another amino acid is attached to MG by AGL to produce shinorine or porphyra-334, L-serine for shinorine and L-threonine for porphyra-334.
Due to the existence of multiple different types of AGL and ALAL with different efficiency and amino acid preference in nature, we selected ligases from three different marine organisms: Nostoc punctiform(Np5598 and Np5597), Nostoc linckia(NlmysC and NImysD) and Actinosynnema mirum(Am4257 and Am4256), and expected to create nine combinations of AGL-AlaL. We used the promoter pTDH3 for AGL and pPGK1 for AGL. We used Lee's yeast toolkit to produce Level1 plasmids containing only one gene(AGL or AGL), and used Golden Gate assembly on the basis of Level1 to construct Level2 plasmid containing both AGL and AGL. Finally, we transformed nine Level2 plasmids containing AGL and AlaL into SCL3,respectively, yielding SC.L5 series.
We cultivated the L5 yeast using a SC-Ura culture medium with 1% glucose and 1% xylose and tested the absorption spectrum of the supernatant of the fermentation broth. Other than the Am4257-Am4256 combination yeast which failed to grow, there is an obvious absorption peak of 334 nm for the other 8 combinations. We found that the absorption peak of AGL Np5598 series was significantly higher than that of NlmysC and Am4257. In ALAL series, NlmysD has the highest absorption peak, followed by Am4256 and Np5597. We selected the 6 samples with higher peak and lysed them to preform OD scanning on the lysate. The scanning results after lysis further performed that Np5598 is the most efficient AGL and NlmysD is the most efficient ALAL. We think we have found an optimal combination of these two enzymes and have successfully produced MAAs. However, the lysis method we used was not completely effective, which resulted in unstable yield per lysis. We use the supernatant as a preliminary yield assessment in subsequent experiments.
Moreover, because AlaL has a preference for different amino acids to produce shinorine and porphyra-334, we need to further confirm the type of MAA in the supernatant. To do so, we used HPLC and MS technology. Due to the lack of standard samples of these MAAs in the market, we extracted MAAs from nori samples using methods we learnt through our human practice activities and used the extractives as standard. We found that the metabolite of Np5598-NlmysD mainly has the absorption spectrum of porphyra-334 according to HPLC, and the resthe ult of MS also proved that mainly porphyra-334 exist in the metabolite (m/q = 347). This shows that NlmysD has a strong selective preference toward the amino acid Threonine, and thus will mainly produce porphyra-334 if both Threonine and Serine are present in the environment. This is the first successful case of producing mainly 334 only. Np5597, on the other hand, shows shinorine's absorption peak, meaning it has preference toward serine. Therefore, we decided to use Np5598-NlmysD for porphyra-334 production and Np5598-Np5597 for shinorine production.
Optimization
In order to increase MAA production, we need to further optimize the SC.L3 strains. Nqm1 has similar functions as TAL1 in shunting S7P into glycolytic pathway. To increase the S7P pool futher, we decided to remove this gene and simultaneously insert an extra copy of DDGS-OMT. It's shown that gene expression can be enhanced by using multiple promoters for increasing maas production (Yang et. al., 2018). We used pTDH3 to express OMT and pPGK1 to express DDGS, making the total transcription rate of the two copie roughly equal. We inserted DNA fragments of LA, OMT, DDGS, RA and the pCRCT-Nqm1 plasmid into SC.L3, and after PCR, DNA sequencing, selection of recombinant colonies, we removed the pCRCT-Nqm1 plasmid, obtaining SC.L6.
To test the result of the optimization, we transformed the Level2 plasmid Np5598-Np5597 and Np5598-NlmysD into L3 or L6 and named the new strains SC.L5 series and SC.L7 series. After fermentation for 72 hours, we tested the OD absorption of broth's supernatant from the yeast before and after optimization (i.e., L5 and L7 respectively). We found that there is a significant increase in the production of MAA. For porphyra-334, output increased by 91.8% in L7-Np5598-NlmysD compared to L5-Np5598-NlmysD, while for shiorine L7-Np5598-Np5597 increased by 70.9% compared to L5-Np5598-Np5597.
Considering the instability and lower efficiency of using plasmid expresssion, we decided to insert Np5598-NlmysD and Np5598-Np5597, the most efficient combinations to produce porphyra-334 and shinorine respectively (Figure 9 & 10) into SC.L6's genome at chromosome I, position 106. We named the new strains SC.L8 series. After fermentation for 72 hours in YPD medium, we compared the OD 334 absorption of the broth's supernatant between SC.L7 series and SC.L8 series. We found that the OD value of L8 series(for both Np5598-NlmysD and Np5598-Np5597) with genome recombinant, showed a steady increase of approximately 20% compared with that of L7 series with plasmid expression. Finally, the nori extract with a higher purity of shinorine and porphyra-334 is used as the standard sample to obtain rough yield data of SC.L8 series: the shinorine’s yield reached 258.5mg/L in L8-Np5598-Np5597, the porphyra-334’s yield reached 270.7mg/L in L8-Np5598-NlmysD.
These results reveal that we were able to increase the production of shinorine and porphyra-334 to a great extent by introducing another copy of DDGS-OMT at position Nqm1 and inserting the genes into S. cerevisiae's genome.
Production of palythine
After obtaining shinorine and porphyra-334, we could produce the MAA palythine by removing a carboxyl group from shinorine. Thus, we added pTEF2-NlmysH-tSSA1 to the Np5598-Np5597 combination to obtain the genetic circuit for producing palythine . We constructed the plasmid and transformed it into SC.L3 or SC.L6, obtaining L3:Np5598-Np5597-NlmysH and L6: Np5598-Np5597-NlmysH. Again, we fermented the two strains for 72 hours in SC-Ura culture medium with 1% glucose and 1% xylose. The absorption spectrum of the supernatant broth was tested, and results show that there is an obvious absorption peak at 320 nm. We also compared the production in L3:Np5598-Np5597-NlmysH and L6:Np5598-Np5597-NlmysH using OD 320 value, which shows that L6 strains are much more productive than L3 strains. The OD 320 value of L6 is 2.62 times of L3 strains, whereas L3 strains barely had any increase in UV absorption at 320 nm. This shows that we have achieved palythine production, and also shows that our previous optimization of L3 is successful.
Production of gadusol
Gadusol is a MAA-like molecule that is capable of absorbing UV radiation mostly in the form of UVB. To produce gadusol, we chose the genes EEVS and MTOX from Danio rerio, which converts 4DG into gadusol. We used promoters pTDH3 to express EEVS and pPGK1 to express MTOX and inserted these genes into L2 yeast at position 308 to obtain L4 strain. After testing the obtaining the absorption spectrum of the supernatant broth after 72 hours of fermentation, a slight absorption peak was observed at around 290 nm. To better observe the absorption spectrum to better determine the existance of the gadusol, we subtracted the curve of the negative control from L4's absorption curve (Figure 15B). The relative OD curve of L4 strain shows an obvious absorption peak at 290 nm, which suggests the production of gadusol.
After producing all planned UV-absorbing molecules - shinorine, porphyra-334, palythine, and gadusol, we compared the production and capabilities of our molecules against UV radiation. We concluded the absorption curves of the molecules expressed on plasmid vectors and compared our absorption curve (Figure 16.B) with the absorption curve from researches (Figure 16.A). We found that the absorption curve of all molecules had the same shape and location of peak compared to results from papers. Moreover, our MAAs also have a much higher absorbance value than Figure 16.A.
This year, LINKS_China successfully produced shinorine, porphyra-334, palythine, and gadusol at an efficient rate. We hope that these MAAs we produce could be implemented in sunscreens to better protect our environment and our skin. For more information, please see Proposed Implementation!
Considering the fact that not all UV radiation could be absorbed by MAAs and could still result in increased oxidative stress, we decided to break down ROS (Reactive Oxygen Species). Therefore, we decided to produce human Superoxide Dismutase 1 (hSOD1) and human catalase (hCatalase). hSOD1 possesses the ability to convert superoxide (O2-), the main form of ROS, into hydrogen peroxide. However, hydrogen peroxide is also harmful to our body, and catalase provides a perfect solution to this problem as it could break down hydrogen peroxide into water.
Thence, we decided to form a fusion protein, hSOD-linker-hCatalase (SHC) (Figure 16.A, B, C), in order to enhance the efficiency of the breakdown of ROS. We have successfully constructed the fusion protein by linking genes expressing the separate enzymes with a flexible linker and expressing them in E.Coli BL21. Successful production of hSOD1, hCatalase and SHC are achieved in concentrations of 270mg/L, 240mg/L and 400mg/L respectively. Furthermore, we discovered that the three proteins all possess high solubility (Figure 16.D), which suggests high efficiency to extract these proteins.
Under the same conditions, we set up assessments for hSOD, hCatalase and fusion protein hSOD-linker-hCAT. Quantitatively, our results showed that separate hSOD and hCatalase had activities of 3.00U/mg and 128.85U/g, respectively. In comparison, hSOD1 and hCAT in our fusion protein SHC produced activity levels of 8.04U/mg and 313.04U/g (Figure 17 & 18). Thence, our results proved that both proteins had enhanced activity when expressed with the linker connection, thereby proving that the fusion protein has improved performance.
Xantine Oxidase can catalyze the production of ROS by Xanthine (Figure 19.A), and by adding our fusion protein SHC into this system and reacting for 60 minutes, we discovered that the ROS and H2O2 produced by the reaction are all degraded (Figure 19.B). We discovered that SHC has enhanced activity to eliminate hydrogen peroxide compared to the single enzyme . By calculation, fusion protein SHC enhanced SOD activity by 168% and catalase activity by 143% compared with single protein. This confirmed our engineering hypothesis as the efficacy of the fusion protein is higher than the two enzymes working separately.