Results
Our project aimed to engineer yeast strains with enhanced electron transfer capacity to serve as the main component of microbial fuel cells. Our strategy combined insights from published research, hardware design, computational modeling, and our own experimental data. In this section of the Wiki, we present the rationale behind our design strategy to modify yeast metabolism for improved extracellular electron transfer and to evaluate microbial fuel cell performance.
Test
To test the effect of our gene deletion on the growth, glycerol and ethanol production, a time course sample collection experiment was performed. Two colonies from each of the single gene deletion strains along with the DOM90 + Cas9 as a control strain were grown in CSM + 2% glucose. OD600 measurements were performed and samples for HPLC and NADH/NAD measurement assay were collected at the following time points: 0, 1.5, 3, 4.5, 6.5, 8.5, 10.5, 12, 24, and 48 hours.
Growth experiment and HPLC analysis
OD measurements indicated that most of the engineered strains exhibited impaired growth, with the slowest growth observed in the Δgpd1 and Δgpd2 strains (Figure 6). After 48h, OD600 in Δgpd1_7, Δgpd2_1, and Δgpd2_1 strains was about 3 times lower than in one of the controls. However, the Δnde1 and Δnde2 strains, as well as the Δgpd1_8 mutant, did not show such substantial differences compared to the control.
Figure 6. Time course of cell growth for engineered and control yeast strains over 48 hours. The absorbance of yeast cultures at every time point was measured at 600 nm to monitor cell growth.
HPLC analysis of glycerol production over a 48-hour time course demonstrated the impact of our gene deletions. As shown in the graph, both Δgpd1 strains exhibited complete suppression of glycerol accumulation. The Δgpd2 and Δnde2 strains accumulated glycerol more slowly during the initial stages of growth, but by the end of the experiment, their glycerol levels were similar to those of the control strain. In contrast, the Δnde1 mutant showed enhanced glycerol accumulation compared to the control, likely reflecting the need to reduce cytosolic NADH due to impaired mitochondrial uptake—directly indicating the intended effect of the mutation (Figure 7).
Figure 7. Glycerol production over time. Glycerol accumulation in the culture medium was measured during the growth of the parental DOM90 strain and deletion strains (Δgpd1, Δgpd2, Δnde1, Δnde2). Differences between strains demonstrates the role of the targeted genes in glycerol biosynthesis.
However, under our experimental conditions, complete suppression of glycerol biosynthesis had also an adverse effect on growth as it was revealed by both OD measurements and HPLC analysis. Consistent with their lower growth rates, Δgpd1_7, Δgpd2_1, and Δgpd2_1 strains exhibited reduced glucose consumption compared to the DOM0090 control. Glucose was not fully depleted from the culture medium even after 48 hours of growth, with approximately half of the initial glucose amount remaining (Figure 8).
Figure 8. Glucose consumption of engineered and control yeast strains over 48 hours measured by HPLC. The decrease in glucose concentration in the medium was measured by HPLC as the parental DOM90 strain and deletion strains (Δgpd1, Δgpd2, Δnde1, Δnde2) grew. The graph illustrates glucose uptake by each strain during different cell growth phases.
All the Δgpd1_7, Δgpd2_1, and Δgpd2_1 strains also exhibited decreased ethanol production during growth, whereas other deletion mutants showed ethanol synthesis rates similar to the DOM0090 control. However, differences were observed in ethanol consumption after 24 hours, with the DOM0090 strain displaying more intensive ethanol utilization (Figure 9).
Figure 9. Ethanol production in engineered and control yeast strains over 48 hours measured by HPLC.
Ethanol accumulation in the culture medium was measured throughout the growth of the parental DOM90 strain and deletion strains (Δgpd1, Δgpd2, Δnde1, Δnde2). The graph depicts the fermentative activity of each strain and the impact of specific gene deletions on ethanol production over time.
NAD/NADH assay
Since the ultimate goal of our gene deletions was to increase the cytosolic NADH pool in yeast cells, we measured NADH/NAD levels using Amplite® Colorimetric NAD/NADH Ratio Assay Kit. Samples for the assay were also collected during the growth experiment. The measurements were repeated twice. In parallel with the sample measurements, the standard curve for different NADH concentrations was built (Figure 10).
Figure 10. Calibration curve constructed from absorbance measurements at 460 nm of standard NADH solutions. The results of the second measurement were used to build the curve.
However, the sample measurements did not yield meaningful results. In some cases, the absorbance measured for total NAD⁺/NADH samples was lower than that of NADH alone (Table 3). We therefore concluded that these results are unreliable and that further optimization of the protocol is required.
Table 3. NAD/NADH ratio measurements using Amplite® Colorimetric NAD/NADH Ratio Assay Kit.
Although the NAD/NADH measurements did not provide any reliable data, we can still conclude that the cytosolic NADH pool in our engineered yeast strains increased, as evidenced by the suppression of glycerol production.
To evaluate the performance of our engineered yeast strains as biocatalysts for electron transfer in MFCs, we designed custom-built hardware and tested multiple yeast strains under these conditions. The results of testing our engineered strains in the MFC hardware showed that the Δgpd1 strain performed slightly better than the DOM90 control as a catalyst for electron transfer.
Modeling
Microbial fuel cell (MFC) design has traditionally relied on trial-and-error strain screening, lacking a predictive framework to link genetics with power output. Here, we introduce a mechanistic multi-scale model that addresses this gap. Cellular NADH flux models predict electron surplus resulting from specific gene deletions, biofilm kinetics translate single-cell metabolism into collective current dynamics, and reactor optimization maximizes power across biological and electrochemical timescales.
The key innovation is the integration of redox balance constraints at the molecular level with population dynamics and electrochemical processes. By measuring only five strains, our framework can predict the performance of novel genotypes and validate these predictions using bootstrap-based uncertainty quantification. We identified statistically significant electron surplus, demonstrating that the model accurately captures real metabolic constraints.
This approach transforms MFCs from an empirical biotechnology into a predictive bioengineering platform, providing a genotype-to-watts pipeline for systematic optimization
Learn
Our experiments lead to several important conclusions. Firstly, the CRISPR - Cas9 system was successfully implemented in DOM90 yeast, as confirmed by colony PCR verification and sequencing of single- and double-gene deletions. This establishes a robust platform for genome editing in our strains and provides a foundation for more efficient and complex modifications.
Second, our experiments with engineered strains revealed measurable physiological changes following gene deletions. HPLC analysis showed altered ethanol and glycerol production, as well as reduced glucose consumption in the engineered strains compared to the DOM90 control. The reductions in ethanol and glycerol production suggest a possible increase in the intracellular NADH pool.
However, NAD/NADH ratio measurements did not yield reliable data suitable for quantification. The measurements were repeated twice. In the first attempt, no meaningful results were obtained; only the standard curve samples were measured, showing a low correlation between concentration and absorbance. To optimize the protocol, we reviewed the literature and modified the sample handling.
When ordering the kit, we were aware that it might not be optimal for yeast cells, as it was primarily designed for mammalian and bacterial systems. Although the kit documentation mentioned potential applicability to plant cells with tough cell walls, it only suggested that it could possibly be used for yeast. Nevertheless, we proceeded with the measurements, and the second attempt produced somewhat improved results. We hypothesize that the main limitation of the assay is the robust yeast cell wall. Considering the high instability of NADH and NAD during handling [19], it is likely that a significant fraction of these metabolites was degraded during cell lysis before the addition of the measurement reagents. Before the second measurement, we added glass beads to the samples along with lysis buffer and used a FastPrep-24 homogenizer prior to incubation at 37 °C, which resulted in a significant improvement in the measurements. The improvement observed in the second attempt suggests that there is still room for further optimization. Another potential approach would be to employ protocols specifically developed for measuring the NADH/NAD ratio in yeast cells [20].
Based on our laboratory data and findings from the scientific literature, we developed models to guide further optimization of MFC engineering. Gene deletions often impose a fitness cost, which can increase when multiple genes are deleted. Our model allows us to predict the consequences of different gene deletion combinations, including their impact on cell viability, growth, and potential benefits for MFC performance. It is important to note that growth conditions during the time-course experiments differ significantly from those within MFC hardware, which involve anaerobic conditions, limited space, and constraints on electron transfer. Understanding these differences is essential for accurately translating experimental findings into improved MFC designs.