Resource depletion and growing piles of waste

The continuous rapid growth of the human population is causing a consequent rise in food demand (Springmann et al. (2018)). As a result of this, for the past 50 years, the amount of food produced has tripled (FAO (2017a)). In 2017, the Food and Agriculture Organization (FAO) of the United Nations estimated that agriculture needs to increase the annual production of food from 8.4 billion tonnes to almost 13.5 billion tonnes by 2050 to feed the expected population of 9.7 billion people (FAO(2017b)). Meeting this staggering food demand generates several sweeping and inescapable challenges of which two relate to this project: 1) finding the necessary space and resources to increase food production and 2) managing the simultaneously increasing piles of agricultural waste (Springmann et al. (2018), Bhatt et al. (2021)).

Figure 2. Graph representing the increasing food demand. The world population (blue) is growing, and as such, so will the food demand increase.

Finding the necessary space and resources to increase food production is a huge challenge; Land for food production and available resources are scarce (Wada et al.(2010), Curtis et al. (2018), Famiglietti (2014), Laurance et al. (2014)), meaning that 50% of the additional food required in 2050 will have to be produced from land already undergoing cultivation (FAO (2017a), FAO(2017b)). This also means that we have to be extremely careful with how we use our resources, to avoid additional pressure on the environment. In other words, it is clear that we need to get the most out of all of the resources we have.

Evergrowing piles of agricultural waste are an unavoidable side effect of the increasing demand and production of food. Cereals, like wheat, maize, and rice, are the most common crops on earth (Leff et al. 2004) and their production also generates tonnes of inedible plant material. The annual global production of crop residues is estimated to be 2802 million tonnes (Sinha et al. (2021), Zabed et al.(2016), of which rice straws, wheat straws and corn stover respectively contribute 731, 354.31 and 128.02 million tonnes to this production (Sarkar et al.(2012), Pattanaik et al.(2019). These waste types are examples of lignocellulosic waste, a difficult-to-utilize plant material!

While some lignocellulosic waste is reused for things such as animal feed, substantial amounts are also burned or dumped in open landfills. In China, the world's largest producer of crop residues, 20-30% of the crop residues are burned. These open burnings cause a release of toxic particles and greenhouse gasses, causing severe adverse effects on human health and the environment (Chen et al. (2019)). Importantly, these burned crop residues are an example of a resource that is currently wasted by simply burning it, instead of utilizing them fully.

Figure 3. The annual global production of crop residues. Over 2000 million tonnes of crop residues are produced, highlighting the huge resource that lignocellulosic waste represents.

In other words, burning and dumping agricultural waste, rather than reusing it, adds pressure to our environment and contributes to the depletion of resources without adding any value to the economy. For these reasons, it is crucial to manage and recycle these wastes to combat the depletion of our resources on Earth, thus promoting a bio-based circular economy.

Current solutions and challenges

Lignocellulosic waste is commonly burned to gain heat and energy, or sometimes used for animal feed. Currently, some research groups are also working on using microbes to produce biofuels from lignocellulosic waste, in the form of bioethanol. However, considering the potential value of bio-based products, as exemplified in the bio-based economy's value pyramid (figure 4), it is clear that this way of using carbon and energy sources is inefficient (Bosman et al. (2016)). Rather, the carbon and energy source should enter the pyramid at the top, aiding in the production of high-value products such as medicine, and then trickle down through the pyramid, and only in the last step be converted to heat or fuel.

Figure 4. The biobased economy's value pyramid. This pyramid exemplifies how resources for bio-based production should be utilized, as it should prioritize the items at the top, creating value compounds, and trickle down to less valuable compounds. Currently, lignocellulosic waste enters at the lowest two stages, which is not an ideal utilization of the resource.

To ensure the most efficient use of Earth's resources, we can not accept the usage of sugar and foodstuff (such as cereals) for production of bioethanol or other molecules obtained through fermentation. We can not allow the amount of land currently occupied solely to produce bioethanol, as it taps into resources needed for food production. Rather, we need to find a way to ensure sustainable production of desirable organic molecules through microbial fermentation of lignocellulosic waste! However, aiming to use lignocellulosic waste in microbial fermentation raises new complicated challenges.

Most common microorganisms used in biomanufacturing struggle to utilize lignocellulose as their carbon source due to its complex structure. To make the polymers in lignocellulose more accessible to microbes, pretreatment is required. Sadly, the pretreatment of lignocellulosic waste produces a range of toxic compounds. One of the most harmful of these is furfural, which greatly hampers the growth of microorganisms at even low concentrations. Currently, researchers at companies such as Novozymes have tried to overcome this by adapting their microbial host through Adaptive Laboratory Evolution (ALE), which improves the tolerance with varying success, as discussed with Carsten Hjort from Novozymes. However, to the best of our knowledge, no industrially relevant filamentous fungi have been adapted to tolerate the toxic compounds in pretreated lignocellulosic waste, even though these microbes show great potential in growing on lignocellulosic waste and are widely used for production of medicine, organic acids, enzymes and bulk chemicals (Cairns et al. (2018)).

Our solution: mold capable of detecting and converting furfural

We aimed to generate an Aspergillus niger strain capable of detecting and converting furfural to harmless compounds! We hypothesized that our fungal cell factory would be able to produce enzymes to convert furfural into a less toxic compound, or utilize it as a carbon source. This would allow the engineered fungus to grow more efficiently on lignocellulosic waste. To reduce the metabolic burden of producing these furfural converting enzymes, we set out to develop a furfural detecting biosensor. The biosensor should then activate the expression of furfural converting enzymes only in the presence of furfural.

Figure 5. Simplified overview of our idea

Strategies to engineer a furfural biosensor

As we wanted to only activate the furfural-converting enzymes in the presence of furfural, we needed a regulator protein capable of detecting furfural. However, no such furfural detecting regulator has been identified and tested so far. Some transcription factors have been shown to be involved in up-regulating the expression of genes involved in the furfural stress response in Saccharomyces cerevisiae, but little to nothing is known about their mechanisms of action (Li et al. (2021)). For example, it is unknown whether they bind directly to furfural, are phosphorylated by another furfural-sensing protein, or what nucleotides comprise their DNA-binding site. This missing information complicates using these transcription factors in new contexts of genetically engineered systems. We therefore decided to develop our own biosensor using two different strategies:

1. Rationally modifying a synthetic transcription factor proven to sense benzoic acid, a furfural-like compound, in yeast (Castaño-Cerezo et al. (2020)).
2. Identifying and using native promoters within fungi that are naturally activated by furfural.

The two different strategies are exemplified in figure 6.

Figure 6. Overview of our two approaches to developing a biosensor capable of detecting furfural. On the left is the native promoter track, in which we tried to find possible promoters native to fungi, that could detect furfural. On the right is our synthetic expression system, created by combining several different modules that we have predicted to bind furfural using Alphafold2 and Rosetta docking software.

Furfural detection using modified synthetic transcription factors

We searched for transcription factors sensing other compounds with chemical structures similar to that of furfural. With such a transcription factor as a starting point, we anticipated it would be possible to tune it to be able to sense furfural, by changing the amino acids in its binding pocket.

Castaño-Cerezo et al. (2020) made a synthetic transcription factor, sBAD, capable of sensing benzoic acid and benzoic acid derivatives in S. cerevisiae. This transcription factor is composed of three parts: the sensing domain HbaR., the DNA binding domain LexA, and the transactivation domain B112. The structures of benzoic acid and furfural contain similarities; They are both composed of a ring structure with a small functional group containing double-bound oxygen.

Figure 7. Benzoic acid (left) and furfural (right). These are structurally quite similar to one another, thus we attempted to rationally modify an sTF detecting benzoic acid to detect furfural instead.

As furfural is a slightly smaller molecule than benzoic acid, we expected that it would be able to fit into the binding pocket of sBAD to activate it. But would it bind similarly to the same amino acids as benzoic acid and activate sBAD despite the structural differences between furfural and benzoic acid? Also, the transcription factor should ideally only respond to furfural and not benzoic acid. We needed a strategy to modify sBAD aiming to change its ligand-binding capacities.

Taylor et al. (2016) successfully modified the transcription factor for LacI in E. coli to respond to new inducer molecules. One of the methods they tested relied on an adaption of the computational software tool called Rosetta (Jiang et al. (2008), Röthlisberger et al. (2008)). Inspired by Taylor et al. (2016), we used Rosetta to predict the docking of furfural into the sBAD binding site and how we could modify amino acids in the binding pocket of the HbaR domain to promote furfural-binding rather than binding of benzoic acid. This resulted in the design of 16 variants of HbaR.

Another gene that caught our attention was the Hmox1 encoding the human heme oxygenase I enzyme. Despite its relation to furfural being seemingly distant, Santhakumar et al. (2021) computationally investigated the binding affinities of heme oxygenase I and furfural. Remarkably, they found a relatively strong binding affinity to furfural, and we therefore made a variant of sBAD containing Hmox1 as its sensing domain.

Besides changing the sensing domain of sBAD, we needed to consider the impacts of moving this transcription factor from yeast to A. niger. Due to the molecular differences between yeast and A. niger, we did not expect the same promoter used by Castaño-Cerezo et al. (2020) to necessarily work in our strain. Similarly, transcription factors in filamentous fungi often contain certain nuclear localisation signals (NLS) and different activation domains compared to yeast (Lu et al. (2021)). We therefore generated our own version of the sBAD promoter, consisting of a minimal fungal promoter and six LexA binding sites (lexO). We also created several versions of sBAD containing the NLS called SV40 and the activation domain VP16, which are proven to work in A. niger (Rantasalo et al. 2018).

After testing various combinations of domains and mutations in our sTF, we constructed a functional synthetic expression system (SES) containing our synthetic promoter constitutively activated by the sTF, FunsTF05. This FunsTF05 contains the LexA binding domain, Hmox1, VP16, and the SV40 NLS. This highly modular SES provides the foundation of generating a furfural-detecting biosensor.

Read more about our SES system here. A Simplified version of our SES system can also seen in figure 8 below.

Figure 8. A simplified version of our SES system. Our TF consists of 3 different domains (in this picture LexA, Hmox1, VP16 and the NLS SV40 are pictured), which should theoretically bind furfural and then activate transcription of mCherry, an RFP gene.

Furfural detection using native promoters

A simple biosensor, which would require no rational engineering, would be a promoter native to A. niger that is naturally activated in the presence of furfural. Prior to the study of the regulatory mechanisms, we hypothesized that we can control the expression of the enzymes by using furfural-induced native promoters. Ideally, this promoter should be specifically activated by furfural and not other weak acids or stressors. Such promoters could potentially be identified by designing growth experiments consisting of exposing A. niger to furfural and other weak acids and subsequently extracting the RNA for transcriptome analysis. From this transcriptome dataset, it would be possible to extract the promoters in front of genes that are upregulated in response to furfural only. No such dataset has been published for A. niger. However, Li et al. (2021) published a transcriptomic dataset of Saccharomyces cerevisiae exposed to acetic acid and furfural.

Figure 9. Volcano plot made from data published by Li et al. (2021). The data was generated by exposing S. cerevisiae to furfural, acetic acid, or both. Two formate dehydrogenase genes (scFDH1 and scFDH2) are upregulated significantly in response to furfural but not in response to acetic acid. When both furfural and acetic acid are present, scFDH1 is mostly upregulated. This indicates that the expression of scFDH1 is induced by furfural.

By further investigating the yeast transcriptome dataset generated by Li et al. (2021), we identified two genes, scFDH1 and scFDH2, that were highly up-regulated in response to furfural but not in response to acetic acid, indicating that these promoters are active specifically when furfural is present. These two genes encode formate dehydrogenases (Sanda et al. (2011), Hasunuma et al. (2011)). Remarkably, we found a homolog of FDH1 in A. niger (anFDH) with 61% identity of the amino acid sequence (NCBI accession code: XP_001396580.2). We decided to try to use the native promoter regulating this gene as a furfural biosensor and verify if it would respond to furfural using qPCR. To find other furfural activated promoters, we attempted to generate a transcriptomic dataset similar to that made by Li et al. (2021), this was however not possible for us to do, due to delivery issues regarding samples. Nonetheless, we developed a bioinformatic pipeline capable of searching for binding sites common for significantly up- or down-regulated genes in transcriptomic data. This pipeline can be used to search for possible transcription factors involved in regulating furfural activated promoters.

Enzymes converting furfural

A few organisms have been reported to be able to convert furfural into less toxic compounds, however, very few enzymes responsible for these reactions have been identified (Koopman et al. (2010)). From our literature research, we chose to investigate if three different genes, Arz_7774 (Wang et al. (2015)), hmfH (Koopman et al. (2010)), and fucO (Wang et al. (2011)), reported to be involved in furfural conversion would work in our A. niger strain and improve its growth in media containing furfural.

Arz_7774 encodes an NAD-dependent aldehyde dehydrogenase from the creosote consuming filamentous fungi Amorphotheca resinae ZN1 (Wang et al. (2015)). This strain was isolated from pretreated corn stover containing furfural, where it grew notably faster than other microbes (Zhang et al. (2010)). Interestingly, it could grow using furfural as its sole carbon source. Additionally, Ran et al. found that glucose was not consumed until the concentration of furfural had decreased to low levels ((Zhang et al. (2010)), Ran et al. (2014)). Later, the aldehyde dehydrogenase encoded by Arz_7774 was suggested to be involved in the furfural degradation pathway in A. resinae by converting furfural into furoic acid, which is much less toxic (Wang et al. (2015)).

hmfH and fucO are both bacterial genes. hmfH encodes a putative FAD-dependent oxidoreductase from the soil bacterium Cupriavidus basilensis HMF14 (Koopman et al. (2010)). This enzyme has been shown to oxidize furfural into furoic acid as well. Lastly, the gene fucO encodes an NADH-dependent propanediol oxidoreductase and comes from one of the most used microbes in industry: E. coli. Overexpression of this gene increased the furfural tolerance in strains producing ethanol or lactate from xylose (Wang et al. (2011)).

Figure 10. The two furfural converting enzymes we engineered into our strains. We succesfully managed to engineer two (Arz_7774 and hmfH) out of the three furfural converting enzymes into A. niger, to help it in circumventing the harmful effect of furfural.

References

• Babu S. et al. Exploring agricultural waste biomass for energy, food and feed production and pollution mitigation: A review. Bioresource technology (2022), 360
• Bhatt R. et al. Input Use Efficiency for Food and Environmental Security. Springer (2021).
• Bosman R. et al. Transition Governance towards a Bioeconomy: A Comparison of Finland and The Netherlands. Sustainability 8, 10 (2016), p. 1017
• Cairns, T. C., et al. How a fungus shapes biotechnology: 100 years of Aspergillus niger research. Fungal biology and biotechnology, 5, 13 (2018)
• Castaño-Cerezo S. et al. Development of a biosensor for detection of benzoic acid derivatives in Saccharomyces cerevisiae. Frontiers in bioengineering and biotechnology 7 (2020), p. 372.
• Chen J. et al. To burn or retain crop residues on croplands? An integrated analysis of crop residue management in China. Science of The Total Environment 662 (2019), p. 141–150.
• Curtis, Philip G., et al. Classifying drivers of global forest loss. Science 361.6407 (2018): pp. 1108-1111.
• Famiglietti, James S. The global groundwater crisis. Nature Climate Change 4.11 (2014): pp. 945-948.
• Food and Agriculture Organization of the United Nations (FAO), 2017a - The future of food and agriculture: Trends and challenges. https://www.fao.org/3/i6583e/i6583e.pdf Accessed: 11-09-2022.
• Food and Agriculture Organization of the United Nations (FAO), 2017b - Strategic work of FAO for sustainable food and agriculture. https://www.fao.org/3/i6488e/i6488e.pdf Accessed: 11-09-2022.
• Hasunuma T. et al. Efficient fermentation of xylose to ethanol at high formic acid concentrations by metabolically engineered Saccharomyces cerevisiae. Applied microbiology and biotechnology 90, 3 (2011), p. 997–1004.
• Jiang L. et al. De novo computational design of retro-aldol enzymes. Science 319, 5868 (2008), p. 1387–1391.
• Koopman F. et al. Identification and characterization of the furfural and 5-(hydroxymethyl) furfural degradation pathways of Cupriavidus basilensis HMF14. Proceedings of the National Academy of Sciences 107, 11 (2010), p. 4919–4924.
• Laurance, William F., Jeffrey Sayer, and Kenneth G. Cassman. Agricultural expansion and its impacts on tropical nature. Trends in ecology & evolution 29.2 (2014): pp. 107-116.
• Leff B. et al. Geographic distribution of major crops across the world. Global biogeochemical cycles 18, 1 (2004).
• Li B. et al. Improving acetic acid and furfural resistance of xylose-fermenting Saccharomyces cerevisiae strains by regulating novel transcription factors revealed via comparative transcriptomic analysis. Applied and environmental microbiology 87, 10 (2021)
• Lu J. et al. Types of nuclear localization signals and mechanisms of protein import into the nucleus. Cell Communicating an Signaling 19,60 (2021)
• Pattanaik L. et al. Biofuels from agricultural wastes. Second and third generation of feedstocks. Elsevier (2019) p. 103–142.
• Ran H. et al. Analysis of biodegradation performance of furfural and 5-hydroxymethylfurfural by Amorphotheca resinae ZN1. Biotechnology for biofuels 7, 1 (2014), p. 1–12.
• Rantasalo A. et al. A universal gene expression system for fungi. Nucleic Acids Research 46, 18 (2018)
• Röthlisberger D. et al. Kemp elimination catalysts by computational enzyme design. Nature 453, 7192 (2008), p. 190–1
• Taylor N. D. et al. Engineering an allosteric transcription factor to respond to new ligands. Nature methods 13, 2 (2016), p. 177–183.
• Wada Y. et al. Global depletion of groundwater resources. Geophysical research letters 37, 20 (2010).
• Wang X. et al. Transcriptional analysis of Amorphotheca Resinae ZN1 on biological degradation of furfural and 5-hydroxymethylfurfural derived from lignocellulose pretreatment. Biotechnology for biofuels 8, 1 (2015), p. 1–13.
• Wang et al. Increased furfural tolerance due to overexpression of NADH-dependent oxidoreductase FucO in Escherichia coli strains engineered for the production of ethanol and lactate. Applied and environmental microbiology 77, 15 (2011), p. 5132– 5140.
• Zabed H. et al. Fuel ethanol production from lignocellulosic biomass: an overview on feedstocks and technological approaches. Renewable and Sustainable Energy Reviews 66 (2016), p. 751–774.
• Zhang J. et al. Biodetoxification of toxins generated from lignocellulose pretreatment using a newly isolated fungus, Amorphotheca resinae ZN1, and the consequent ethanol fermentation. Biotechnology for Biofuels 3, 1 (2010), p. 1–15.
• Sanda T. et al. Repeated-batch fermentation of lignocellulosic hydrolysate to ethanol using a hybrid Saccharomyces cerevisiae strain metabolically engineered for tolerance to acetic and formic acids. Bioresource technology 102.17 (2011), p. 7917–7924.
• Santhakumar P. et al. Molecular docking analysis of furfural and isoginkgetin with heme oxygenase I and PPARγ. Bioinformation 17,2 (2021), p. 356.
• Sarkar N. et al. Bioethanol production from agricultural wastes: an overview. Renewable energy 37, 1 (2012), p. 19–27.
• Sinha A. K. et al. Agricultural waste management policies and programme for environment and nutritional security. Input Use Efficiency for Food and Environmental Security. Springer, (2021), p. 627–664.
• Springmann M. et al. Options for keeping the food system within environmental limits. Nature 562, 7728 (2018), p. 519–525.