Multiplex CRISPR-Editing of Wood for Sustainable Fiber Production

Abstract

The domestication of forest trees for a more sustainable fiber bioeconomy has long been hindered by the complexity and plasticity of lignin, a biopolymer in wood that is recalcitrant to chemical and enzymatic degradation. Here, we show that multiplex CRISPR-editing enables precise woody-feedstock design for combinatorial improvement in lignin composition and wood properties. By assessing every possible combination of 69,123 multigenic-editing strategies for 21 genes in the lignin pathway, we deduced 7 unique genome-editing strategies targeting the concurrent alteration of up to 6 genes, and produced 163 edited poplar variants. CRISPR-editing increased the wood carbohydrate-to-lignin ratio to 239% of wildtype, leading to more efficient pulping for fibers. The edited wood alleviates a major fiber production bottleneck, bringing unprecedented operational efficiencies, bioeconomic opportunities, and environmental benefits.

One-Sentence Summary:

Multiplex CRISPR-editing enables precise woody-feedstock design to enhance fiber production efficiency and sustainability.

Wood is the most abundant carbon biomass on earth and the major source of sustainable green fibers (1). Globally, 550 gigatons of carbon are stored as wood, representing 57% of the biogenic carbon sink (1). The biomass supplies over 170 megatons of virgin fibers annually (2) to meet the growing demand for renewable tissue, paper, packaging, textile, and other fiber products, including structural materials (3). Despite the importance of wood fibers, its production has remained largely limited to undomesticated forest trees with often sub-optimal wood properties that hamper production efficiency. The propensity of wood for efficient isolation of cellulosic fibers is largely determined by the content and composition of lignin (4–6), one of three major components of wood (7). Lignin is a phenolic polymer that crosslinks with cellulose and hemicelluloses in the secondary cell walls of vascular plants (8, 9). The polymer in angiosperm wood is formed by free radical polymerization of three major monolignol precursors, 4-coumaryl, coniferyl, and sinapyl alcohols, which form the p-hydroxyphenyl (H), guaiacyl (G), and syringyl (S) units in lignin (10–12). The monolignols are biosynthesized from phenylalanine through a series of enzymatic reactions in a metabolic grid consisting of at least 11 enzyme families and 24 metabolites (13). The pathway is mediated by hundreds of regulatory influences, encompassing transcriptional (14), translational (15), and post-translational regulations (16, 17), enzyme-enzyme interactions (18–20), and metabolic regulations (21, 22). Over 5 decades of research have extensively investigated the individual components of lignin biosynthesis and determined the effects of their perturbation on lignin content and composition in diverse plant species (6, 13, 23). However, these efforts have predominantly focused on the modification of single genes or gene families, whereas the combinatorial effects of multigenic perturbations remained elusive. Here, we show that strategic multiplex CRISPR-editing of monolignol biosynthetic genes improves wood properties beyond the editing of single genes or gene families, and debottlenecks a key operational constraint in industrial pulp mills. The improvements substantially increase fiber production capacity while reducing the global warming potential of pulp mills, leading to a more sustainable and efficient fiber bioeconomy.

Gene targets for multiplex genome editing were identified using our established predictive model for monolignol biosynthesis (15, 22, 24). The model predicts the transduction of quantitative relationships from gene transcript abundances to absolute enzyme abundances, pathway metabolic fluxes, and 25 wood chemical and physical properties. Using the predictive model, we explored 69,123 sets of multigenic editing strategies to reveal the extent to which individual wood properties can be modified through multiplex editing of monolignol genes (Fig. 1A and B). The 69,123 strategies encompass 48,831 gene combinations that represent all possible permutations of loss-of-function editing of up to six genes, and 20,292 gene combinations that target the combinatorial editing of up to three genes and overexpression of one gene (dataset S1A). For each gene combination, the model estimated the corresponding abundances of monolignol enzymes, pathway metabolic fluxes, and 25 wood properties. Targeting multiple genes expanded the range of phenotypic variation attainable by genome editing compared to single gene edits (Fig. 1C and dataset S1B). For example, multigenic editing could reduce lignin content to 50.7% of the wildtype level, whereas single gene edits only reduced lignin content to 61.3% (Fig. 1C and dataset S1B).

Fig. 1. Identification of multiplex CRISPR-editing strategies to improve wood fiber traits.

(A) Three-dimensional scatterplot of predicted lignin content, S/G ratio, and tree growth (stem diameter) for 69,123 genome-editing strategies. (B) Venn diagram of the 69,123 strategies showing the relationships between predicted lignin content, S/G ratio, C/L ratio, and tree growth (height). (C) Distribution of lignin contents in single-gene and multigene editing strategies. (D) The total number of target genes in the selected 367 strategies. (E) The frequency of each monolignol gene in the 367 strategies. (F-I) Global sensitivity analysis of wildtype (a), single-gene editing (b–d), and multigene editing (e–k) strategies for lignin content (F), S/G ratio (G), C/L ratio (H), and tree growth (I).

Wood with low lignin content and high syringyl to guaiacyl (S/G) ratio is ideal for fiber production (25). We mined all 69,123 strategies to identify gene combinations that are predicted to reduce lignin content by at least 15%, increase carbohydrate to lignin (C/L) ratio by at least 200%, and have a higher S/G ratio than the wildtype. The C/L ratio is an indicator of the potential maximum cellulosic yield for wood fiber (13). The strategies must also have predicted growth characteristics (e.g., tree height) that are comparable to (>75%) or exceed wildtype controls. Of the 69,123 strategies, only 347 (0.5%) matched the aforementioned criteria in lignin content, C/L and S/G ratios, and tree growth (Fig. 1B and dataset S1C), highlighting the need to mine such strategies to practically test the most promising combinations. All 347 strategies target at least two monolignol genes, and 99.7% of the strategies target at least 3 genes (Fig. 1D and dataset S1B). The number of target genes affirms the need for a multigenic approach to improve fiber traits in Populus trichocarpa. Lignin gene families most frequently appeared in these 347 strategies are C3H, CCoAOMT, AldOMT, PAL, C4H, and CAD (Fig. 1E). We then performed sensitivity analysis for these gene families to examine the robustness of fiber trait improvements in response to varying transcript abundances of non-targeted monolignol genes (Fig. 1F–I, fig. S1A–E, dataset S2 and S3), and selected seven strategies for editing in P. trichocarpa. The seven strategies (table S1) encompassed various numbers of target genes (from 3 to 6) and were selected based on the extent and robustness of predicted improvement in fiber traits (reduced lignin, increased S/G and C/L ratios, and good growth) (dataset S2 and S3). The selected strategies showed a predicted reduction in lignin content by up to 36% compared to wildtype (Fig. 1F and dataset S3). S/G and C/L ratios increased up to 248% and 215%, respectively (Fig. 1G and H, dataset S3), with no change in tree height (Fig. 1I and dataset S3).

To test the seven strategies in planta for their modulation of wood properties for fiber production, a multiplex CRISPR construct (26) was assembled for each strategy and delivered in P. trichocarpa using Agrobacterium tumefaciens (27) (fig. S2–S4, dataset S4, table S2 and S3). We generated 163 independent lines of multiplex CRISPR-edited P. trichocarpa, including 75 lines that target the concurrent editing of 3 monolignol genes, 18 lines that target 4 genes, 37 lines that target 5 genes, and 33 lines that target 6 genes. The edited lines exhibited varying extent of loss-of-function mutations of the target genes (Fig. 2A, fig. S5, dataset S5). Biallelic loss-of-function editing of all target genes was obtained in P. trichocarpa for strategies that target the concurrent editing of 3 or 4 genes (Fig. 2A, fig. S5, dataset S5). For strategies that target 5 or 6 genes, none of the 67 edited lines showed complete loss-of-function editing of both alleles in all target genes. Nonetheless, we identified several lines (e.g., J-7 and K-28) that harbor substantial editing of most target genes (fig. S5 and dataset S5). The frequency of biallelic edits varied between target genes, ranging from 86% for PtrPAL4 and PtrPAL5, to 7% for PtrCCoAOMT2 (Fig. 2B). Most edits consisted of small INDELs of 1 to 3 nt in the immediate vicinity of the targeted cleavage site (Fig. 2C and D, fig S6). However, deletions of up to 19 nt were also occasionally observed (fig. S6). No off-target edits were detected in our edited lines (table S4). We used RNA-seq to evaluate two potential xylem-expressing off-target genomic loci in 10 edited P. trichocarpa lines (dataset S6). The absence of detectable off-target editing suggests that genome editing is highly specific in P. trichocarpa. The broad variation in the editing profiles of target genes in our 163 edited lines created genetic diversity in monolignol biosynthesis not present in nature (Fig. 2E–H). Such genetic diversity enables an exploration of how multigenic pathway perturbations can combinatorially regulate wood formation and wood utilization, as well as enhance genetic diversity.

Fig. 2. Genetic diversity in multiplex CRISPR-edited P. trichocarpa.

(A) Percentage INDELs of each target gene in the 58 CRISPR-edited P. trichocarpa lines targeting the concurrent editing of 3 monolignol genes (PtrPAL2, PtrPAL4, and PtrPAL5). (B) Average percentage INDELs of the target monolignol genes in the 163 edited P. trichocarpa lines. (C-D) Most frequent CRISPR-Cas mediated mutations in PtrPAL2 (C) and PtrPAL4|5 (D). Stem cross-sections of wildtype (E) and CRISPR-edited H-2–1 tree (F), showing distinct red coloration of the xylem. (G) Six-months-old greenhouse-grown CRISPR-edited and wildtype P. trichocarpa. (H) Harvested stem segments from CRISPR-edited and wildtype P. trichocarpa.

Multiplex editing of monolignol genes significantly altered the chemical and physical properties of the wood (Fig. 3A–D, fig. S7, dataset S7–S10). Stem wood of 6-months-old wildtype P. trichocarpa contains 20.9% lignin and has a C/L ratio of 3.0 (dataset S7). In the edited wood, lignin content is reduced by up to 51.7% of wildtype, and the C/L ratio increased up to 239% (Fig. 3E and F, dataset S7). 2D-NMR (Fig. 3G–J, fig. S8–S10) revealed that multigenic strategies also modulated lignin composition, increasing the S/G ratio from 2.7 in the wildtype to as high as 4.0 in the edited lines (Fig. 3C and dataset S8). The most significant lignin reductions were observed in edited trees that harbor 4 to 6 gene edits, but strategies that targeted 3 genes also showed significant lignin reductions of up to 33% (Fig. 3E, and dataset S7). Wood density and elasticity were not significantly different between the wildtype and the genome-edited P. trichocarpa, except for two lines (E-9 and G-25), which had a reduced wood elasticity (fig. S7 and dataset S9). Several CRISPR-edited lines with significant reductions in lignin content showed little to no change in observable growth characteristics (Fig. 3D and K, fig. S7, dataset S10). A few CRISPR trees (e.g., H-4–1) even showed increased wood volume (up to 17% of wildtype) despite having lignin content reduced by up to 33% (Fig 3E and dataset S10).

Fig. 3. Phenotypic variation in CRISPR-edited and wildtype P. trichocarpa.

(A-D) Distribution of lignin content (A), C/L ratio (B), S/G ratio (C), and tree growth (height) (D) in CRISPR-edited (blue) and wildtype (red) P. trichocarpa. Lignin content (E) and C/L ratio (F) of selected CRISPR-edited lines with improved fiber traits. (G-J) 2D-NMR ¹³C-¹H (HSQC) correlation spectra (side chain and aromatic regions) of wildtype (G and H) and the G-27 edited-line (see fig. S8 for other edited lines). Contours in these regions were used to estimate the distribution of lignin interunit linkages and lignin composition, namely S/G/H ratios, as well as p-hydroxybenzoate (PB). (K) Quantitative relationship between lignin content and wood volume in CRISPR-edited and wildtype P. trichocarpa.

Kraft pulping is the dominant process for industrial wood fibers production (28–30). To understand the techno-economic impacts of CRISPR-edited wood on Kraft pulping, we used a Carolina Pulp and Paper (CPP) Mill model (fig. S11) constructed based on an existing industrial Brazilian pulp mill with an annual production of 1.24 M tons of pulp. Pulp yield and production efficiency can be significantly increased by multiplex editing of monolignol genes (Fig. 4A and B), as evidenced by micro-pulping wood with varying lignin content and composition. Reducing lignin content proportionally increases pulp yield and reduces the usage of pulping chemicals (Fig. 4A and B, fig. S12–S14, table S5–S7) that are undesirable from an environmental standpoint. Furthermore, low lignin wood reduces the solid content of black liquor, thereby debottlenecking the recovery boiler, arguably the most crucial and rate-limiting energetic component of pulp mills (Fig. 4C–E, fig. S15). The debottlenecking then enables an incremental production potential of the pulp mill by up to 40% (Fig. 4C–E, fig. S15). Additionally, increasing the C/L ratio in wood means less land is required to provide the same amount of cellulose (fig. S14). The benefits of increasing the S/G ratio encompassed the reduction of pulping chemicals and energy savings on chemical recovery (fig. S13 and S15). Remarkably, tremendous financial benefits can be obtained by reducing lignin from 28% to 16%, and increasing the S/G ratio from 2.8 to 6.0 (fig. S16A and S17A). Such changes are predicted to increase the net present value from $US 500 million to $US 2,245.9 million for pulp mills using natural gas as supplementary energy (fig. S16B), and from $US 1,128.6 million to $US 3,070.3 million for mills using biomass as supplementary energy (fig. S17B).

Fig. 4. Techno-economic analysis and global warming potential of CRISPR-edited wood in industrial Kraft pulping.

(A) Correlation between lignin content and pulp yield in CRISPR-edited and wildtype P. trichocarpa. (B) Paper disks produced from CRISPR-edited P. trichocarpa. (C) The impact of varying lignin content and S/G ratio on the incremental production potential of an industrial Kraft pulp mill. (D-E) Operational capacity of different components of an industrial Kraft pulp mill using wood with 28% lignin (D) or 16% lignin (E) at S/G ratio of 2.8. (F) Total life-cycle GWP of producing pulp from 1 BDMT log. The scenarios include varied S/G ratios (2.8 or 6), lignin content (28% or 16%), and production modes (constant or incremental).

Pulp and paper industry is a major contributor to greenhouse gas emissions. The U.S. Environmental Protection Agency reported 35 million metric tons of CO₂eq emission from pulp and paper industry in the U.S. in 2020 (31). Globally, the annual direct CO₂ emission from pulp and paper production reached 168 million metric tons (32). We assessed the impacts of CRISPR-edited wood on the carbon footprint of pulp production. Life-cycle Global Warming Potential (GWP) was estimated for pulp production from 1 bone dry metric ton (BDMT) of CRISPR-edited wood, compared with wildtype wood. For a pulp mill powered by natural gas, increasing the S/G ratio from 2.8 to 6.0 leads to a 34–35% reduction in the GWP per BDMT log processed (Fig. 4F, fig. S18–S22, table S8 and S9). Reducing lignin content from 28% to 16% decreases the GWP by 5–8% (Fig. 4F, fig. S18–S22, table S8 and S9). Altogether, incremental pulp production using edited wood harboring reduced lignin and increased S/G ratio could reduce GWP by up to 31% (Fig. 4F, fig. S18–S22, table S8 and S9), providing substantial benefits to environmental conservation and climate change mitigation. These results provide insight into the potential of genome editing technologies for tree breeding and illustrate how strategic multiplex editing to alter wood composition enables more sustainable fiber production with remarkable operational efficiencies, bioeconomic value creation, and tangible environmental benefits.