To identify genes influencing traits linked to biomass growth in Populus, we generated and screened under controlled greenhouse conditions a population of 2,700 activation-tagged lines (see Material and Methods). We focused on traits that are linked to biomass growth ( Figure 1A ). A total of 761 lines exhibited significant effect on at least one trait, categorized as phenotypic mutant ( Figure 1B ). The various mutations had an approximately evenly distributed impact over the 10 studied traits. The number of internodes showing the greatest impact among all traits ( Figure 1A ). In contrast, traits such as diameter at leaf plastochron index (LPI) 20, green density, dry weight at the base and dry leaf weight were least affected ( Figure 1A ). More than half of the mutants (402) were affected in only one trait ( Figure 1B ). In many cases, mutations impacted simultaneously as many as 5-8 traits ( Figure 1B ).

To explore potential correlation between traits, pair-wise phenotypic correlations were calculated across the entire experiment ( Table 1 ). As expected, strong correlations were evident between closely related traits, such as green and dry density, and between total dry biomass of the whole stem and the stem base. Interestingly, less intuitive correlations were also observed. For instance, dry leaf biomass showed significant and positive correlation with the stem base diameter and dry stem biomass. Additionally, dry stem biomass correlated positively with internode number and the stem base diameter. These correlations may suggest underlaying morphological or physiological connections that are not apparent and require further investigations.

We characterized T-DNA insertions ( Supplementary Table S1 ) in 40 mutant lines ( Table 2 ). Chromosome 1 had the highest number of insertions (7), followed by 6 insertions on chromosome 6, and 5 insertions on each chromosome 10 and 14 ( Figure 2A ). A significant correlation between chromosome size and the number of insertions suggests random insertion of the tag ( Figure 2B ). Most (37.5%/15) of the insertions were located within the 10 Kbp 5′/3′ regions proximal to genes, while 17.5 % (7) in exons and 12.5 % (5) in introns ( Figure 2C ). Additionally, 22.5 % (9) were positioned in the 5’- or 3’-untranslated gene regions (UTR) ( Figure 2C ).

For nine mutants, we proceeded with further characterization of the expression of the genes near to the insertion sites ( Table 3 ). These mutants were selected based on the magnitude of their impact on the affected trait, the number of traits affected, and the novelty of the gene function ( Table 3 ). Since the tag is often inserted near multiple genes, and thus potentially impacts more than one, we studied the expression of all potential candidates. In most cases (8 out of 9 lines), we found only one of the nearby genes activated ( Table 3 ). However, in only one line (A630-7), both genes proximal to the tag showed activation.

We characterized the expression of the activated genes in various organs, including the apex, leaf, stem, and root of wild type (WT-717) plants ( Figure 3 ). These genes exhibited diverse tissue specific expression patterns, suggesting varied influence on biomass traits. For example, PtXaTreH.14G052500, PtXaTreH.14G052600 and PtXaAlbH.08G010800 were predominantly expressed in the apex. PtXaAlbH.10G086200 and PtXaTreH.15G090200 were abundant in the root, whereas PtXaAlbH.05G161200 was mainly expressed in both the apex and roots. PtXaTreH.06G072700 was expressed in all studied tissues.

One of the lines selected for recapitulation experiments was A630-7, chosen for its simultaneous impacts on multiple traits (five in total, as shown in Table 2 ). Most importantly, the mutation significantly enhanced stem dry weight, a crucial aspect of biomass production. Additionally, the line exhibited increased leaf dry biomass, which, as previously mentioned, is positively correlated with stem dry weight ( Table 1 ). This prompted our interest in identifying the gene responsible for these changes. Position and expression analyses indicated the upregulation of two genes near the tag insertion site ( Table 3 ). One gene showed the highest homology to Arabidopsis's Fasciclin 10 (FLA10), which we named PtaFLA10. The other gene exhibited the highest sequence homology to Growth Regulating Factor 9 (GRF9) transcription factor from Arabidopsis, which was called PtaGRF9. It was unclear which of the two genes was responsible for the phenotypic changes. We thus produced overexpression constructs for both genes and transformed them into transgenic plants. Numerous independent events were regenerated with the PtaFLA overexpression construct (oe-PtaFLA10). However, despite several transformations attempts, we could only recover four transgenic plants with PtaGRF9 overexpression construct (oe-PtaGRF9), suggesting that PtaGRF interferes with the regeneration process.

PtaFLA10 overexpression positively affected several phenotypic traits linked to biomass growth compared to WT-717 plants ( Figure 4 ). For instance, both the original mutant A630-7 line and oe-PtaFLA lines were about 9 % taller than the wild type. Additionally, the number of internodes increased by 15% in the A630-7, while the oe-PtaFLA lines exhibited an average increase of 20 % ( Figure 4 ). The oe-PtaFLA lines also displayed changes not observed in the original mutant, likely due to the strong overexpression. For example, there was a significant increase in the diameter at the 20^th internode and at the stem base ( Figure 4 ). Most notably, a 40 % increase in the dry stem biomass was observed in both the original A630-7 mutant line and oe-PtaFLA transgenics. Consistent with the strong correlation between leaf and stem dry weight ( Table 1 ), leaf dry weight was also increased in both the mutant and recapitulation transgenics.

In addition to enhanced biomass growth, both the original line and the overexpression transgenics showed a significant decrease in lignin content ( Table 4 ). The overexpression transgenics also exhibited a reduced S/G ratio, a change not observed on original line ( Table 4 ). No significant changes were measured in cell wall carbohydrates.

Overexpression of the other gene, PtaGRF9, did not result in significant phenotypic changes, suggesting that the observed mutant phenotypic characteristics are due to the upregulation of PtaFLA10 gene.

We also further characterized the A541-1 mutant line through recapitulation experiments. The activated gene in this line encoded a protein with high similarity to patatin, and thus named PtaPAT. PtaPAT overexpression led to an increase in stem diameter, but only in stems undergoing primary growth, specifically 5^th and 10^th internodes ( Figures 5A, B ; Table 2 ). Consistent with the original A541-1 mutant line, the oe-PtaPAT lines also exhibited an increased number of internodes ( Figure 5C ; Table 2 ). Additionally, overexpression of PtaPAT resulted in leaves with an uneven adaxial leaf surface, reminiscent of potato leaves ( Figures 6A–C ).

Given the distinct leaf surface observed in the oe-PtaPAT lines, we performed histological analysis ( Figures 6 D, E ). The oe-PtaPAT lines displayed a much denser palisade mesophyll cell layer, with a larger number of closely spaced cells compared to the WT plants ( Figures 6 D, E ).

Because of these unique changes in leaf structure, we wanted to study the performance of the oe-PtaPAT lines under drought conditions ( Figures 6A, F, G ). Wild-type and oe-PtaPAT lines were subjected to drought stress for 3 days and CO₂ assimilation and stomatal conductance were monitored during this period. On the 2^nd day of the stress treatment oe-PtaPAT lines were not wilted ( Figure 6A ) and showed higher photosynthetic activity ( Figure 6F) and stomatal conductance ( Figure 6G ) compared to the wild-type plants. These results indicate a possible and yet unknown role of the patatin gene in the response to the drought and create new opportunities for further gene manipulations to improve drought tolerance in plants.

The oe-PtaPAT lines also showed increases in lignin, cellulose, and S/G ratio, changes not observed in the original mutant line and likely attributable to the much higher level of overexpression ( Table 4 ).

A hybrid aspen clone, Populus tremula x Populus alba INRA 717-IB4 (referred to as WT or 717), was used in all experiments, including the transgenic manipulations. The activation tagging population was generated using a binary vector pSKI074 via an Agrobacterium-mediated procedure (Han et al., 2000). All putative transformants were PCR verified for the presence of neomycin phosphotransferase II (NPT), as a selectable marker (Yordanov et al., 2010). Only the NPT-positive transformant were used in further experiments. These verified transformant were propagated and maintained in vitro on ½ MS media with 20 g/l sucrose (Caisson), 0.1mg/l IBA (Sigma-Aldrich), vitamins (Han et al., 2000) solidified with 2.5 g/l Gelrite (Sigma) and 4 g/l Phytablend agar (Caisson), at 16/8 h day/night photoperiod (20 µmol m^-2s^-1).

For the greenhouse experiments, plants were first propagated and grown in vitro for four weeks on ½ MS solid media (as describe above). The rooted plantlets were then transferred to soil and gradually acclimated to greenhouse conditions. Once acclimated, uniformly developed plants were transplanted and grown in greenhouse for approximately 3 months as previously described (Rauschendorfer et al., 2020). The experiment was conducted in a completely randomized design with three replications.

Measurements of the height and basal diameter of the stem, counting of the internodes of each plant were performed regularly to analyze the growth characteristics of poplar plants. A slide digital caliper was used to determine the basal diameter of each plant above the pot surface. For harvest the above-ground part of each plant was separated into leaves and stems. The fifth leaf from the top of the plant was used to determine the leaf area. Digital images were taken with Nicon Coolpix camera. Leaves and stems were air dried, and their dry weight was measured until was unchanged. We measured green wood density (Den-g, g/cm³), basic/dry wood density (Den-d, g/cm³), and moisture content (MC, %) of wood samples collected at the base (15 cm from the soil surface) of the stem. For each sample, we determine the green disc mass and green volume, using water displacement. The stem sections were kiln-dried at 105°C and again weighed. Den-g and Den-d were estimated for each tree as sample mass (g) / disc green volume (cm³).MC will be estimated as [(Den-g – Den-d) / Den-g] x 100.

Two-month-old greenhouse-grown plants subjected to drought stress, were used to measure photosynthesis. Net photosynthetic rate was measured using LI-6400XT portable photosynthesis system (Li-Cor Inc., Lincoln, NE, USA). The measurements were conducted during mid- and late- morning (usually 09:00 – 11:30am) on uniformly sunny days. Leaves of each genotype (four biological replicates/genotype) were measured under the following conditions: 400 μmol s ^-1 flow rate, 400 μmol mol ^-1 reference CO ₂ concentration and photosynthetic photon flux density (PPFD) of 1500 μmol m ^-2 s ^-1, provided by a red–blue light source (6400-02B). Relative humidity was maintained between 50 % and 75 % inside the chamber (RH_S_%). The control temperature was set at 30°C.

Recovery of sequence flanking the insertion site of the activation tag was performed as previously described (Yordanov et al., 2010). The isolated DNA fragments were positioned in the poplar 717 genome using BLAST searches in the Phythozome v13 (http://www.phythozome.net/poplar.php) database and proximal genes to the insertion site identified. Expression of the flanking genes was studied using RT-PCR with gene specific primers ( Supplementary Table S2 ) and Ubiquitin (Ubq) gene (Wei et al., 2013) as a loading control.

The open reading frame of candidate genes were amplified using the following primers ( Supplementary Table S3 ). The amplified product was then cloned into pDONR221 vector (ThermoFisher Scientific) using the BP Gateway system (Invitrogen), sequence validated and transferred into the overexpression pK7WG2 vector (Karimi et al., 2002) using the LR Gateway system (Invitrogen). The binary vectors were transformed into Agrobacterium strain AGL1 (Lazo et al., 1991) and transformed into the hybrid aspen clone, Populus tremula x Populus alba INRA 717-IB4 as previously described (Han et al., 2000).

Total RNA was extracted as previously described (Busov et al., 2003). Reverse transcription was performed on 1 μg of DNAase I-treated total RNA in a final reaction volume of 20 μl using an MMuLV (Moloney Murine Leukemia Virus) reverse transcriptase (ThermoFisher Scientific) following the manufacturer's protocol. Quantitative RT-PCR (qRT-PCR) was performed using the StepOnePlus Real Time System (Applied Biosystems) with the Maxima SYBR Green detection system (Thermo Fisher). Each PCR reaction contained 1× Maxima SYBR Green qPCR master mix, 0.1 μM of each forward and reverse primer (Eurofins MWG Operon), 1 μl of 10 × diluted cDNA solution and nuclease-free water. The final volume of each PCR reaction was 20 μl. The qRT-PCR cycling stages consisted of initial denaturation at 95°C for 10 min, followed by 40 cycles of 95°C for 15 s and 60°C for 1 min, and a final melting curve stage of 95°C for 15 s, 60°C for 1 min and 95°C for 15 s. qRT-PCR was performed with three biological and two technical replicates for each sample. Relative gene expression was calculated as previously described (Livak and Schmittgen, 2001; Schmittgen and Livak, 2008; Tsai et al., 2006; Yordanov et al., 2017). The Ubq gene expression was used as a loading control (Yordanov et al., 2010). All primers used in the gene expression analyses are shown in Supplementary Table S2 . Primers were designed using the Primer-BLAST web resource at NCBI (National Center for Biotechnology Information; http://www.ncbi.nlm.nih.gov/BLAST).

Wood samples were milled to a 20-mesh using a Wiley mill. Approximately 4 mg of milled wood sample was measured and loaded into metal cups, which were then placed into an auto-sampler tray. The cell wall composition was analyzed by studying pyrolysis vapors produced using a commercially available molecular beam mass spectrometer (PyMBMS) designed specifically for biomass analysis as previously described (Sykes et al., 2009).