Shoot branching is an important agronomic trait that determines plant architecture and affects crop productivity (Shen et al., 2019). Molecular signals from the shoot apical meristem (SAM) create a hormonal environment that integrates with the expression of axillary bud-specific repressors such as BRANCHED1 (BRC1) to inhibit axillary shoot formation (Wang et al., 2019 and references therein). The signal is eliminated by topping (SAM removal), enabling the formation of new shoots (suckers) from axillary buds (Figure 1a). In tobacco (Nicotiana tabacum), topping is necessary to enhance leaf development/maturation before harvesting, but sucker growth after topping is undesirable because it reallocates resources to axillary buds, reducing yield and quality of the main leaves. Sucker growth can be inhibited by fatty alcohols, flumetralin or maleic hydrazide, but chemical control is time-consuming and expensive, and the chemicals may persist after leaf processing due to environmental variability (Bailey et al., 2019). Tobacco plants with delayed axillary bud initiation or shoot growth would therefore significantly improve harvest and/or product quality, as previously shown for other species (e.g., Groot et al., 1994).

To analyse the transcriptomes of axillary meristems/buds from tobacco before and after topping, we grew plants in the greenhouse for 8 weeks and took seven samples (n = 3) including young leaf, SAM and axillary meristems/buds before and 2, 6, 24 and 72 h after topping, from plants with 8–10 fully expanded leaves. RNA was extracted for RNA-Seq analysis on an Illumina HiSeq 2000 device (100-bp single reads, at least 30 million reads per sample), and we identified 17 candidate genes that were deregulated in axillary buds post-topping (Figure 1b). Expression was validated by qPCR, and the six most promising genes were selected for further analysis. The corresponding promoters were analysed in the axillary meristem of a commercial dark tobacco before and after topping using a GUS reporter assay. Four of the promoters showed nonspecific activity, but gusA expression driven by promoters P#1_2.5kb and P#15_2.5kb was limited to the axillary meristem, with P#15 showing the more restricted spatial domain (Figure 1c). P#15_2.5kb::GUS activity was stable even 7 days post-topping, but GUS activity in the P#1_2.5kb::GUS transgenic lines declined shortly after topping. P#1_2.5kb sequence analysis revealed the presence of the sugar-repressible element TTATCCA (Tatematsu et al., 2005) at positions –2401 to –2407. Shortening P#1_2.5kb to 2.4 kb (P#1_2.4kb) did not change its axillary meristem specificity but prolonged its activity, so that GUS staining was still detected 10 days post-topping in P#1_2.4kb::GUS transgenic plants (Figure 1c).

Gene#1 encoded a BRC1 homolog, and silencing enhanced sucker growth even before topping, as reported in other species, whereas strong overexpression driven by the constitutive CaMV35S promoter was lethal, allowing the regeneration of only one transgenic line with severely stunted growth (data not shown). Transgenic lines with weak gene#1 expression showed sucker development comparable to wild-type controls, and the expression of gene#1 driven by P#1_2.4kb only slightly reduced sucker growth (data not shown), probably reflecting endogenous regulation and/or a positive regulator of bud formation such as NtBRC2 (Ding et al., 2020). Gene#15 encoded a vicilin-like protein, and neither RNAi nor constitutive overexpression generated a notable phenotype (data not shown).

To selectively inhibit axillary bud initiation and subsequent sucker growth, we expressed the cytotoxic ribonuclease barnase from the bacterium Bacillus amyloliquefaciens under the control of the axillary bud-specific promoters P#15_2.5kb and P#1_2.4kb in order to ablate the cells responsible for sucker formation. We initially generated 11 P#15_2.5kb::barnase transgenic plants, seven of which did not develop axillary bud primordia during vegetative growth. Next, we topped two lines (L10 and L11) and no axillary bud primordia were visible even 1 week post-topping (Figure 1d). These plants showed a normal phenotype, but axillary bud initiation was delayed by at least 3 weeks, resulting in fewer and shorter suckers with a weight reduction of 50%–79% even 4 weeks post-topping compared with wild-type controls (Figure 1d). Seeds from the remaining lines failed to germinate on selective medium, indicating that the P#15_2.5kb promoter (and thus barnase) is also active during seed formation/germination. We analysed the P#15_2.5kb::GUS transgenic lines again and confirmed weak GUS staining in the seeds (Figure 1d).

Barnase expression driven by the P#1_2.4kb promoter also strongly inhibited sucker growth in T₀ plants, and seven of an initial population of 21 plants produced no or only a few axillary bud primordia during vegetative growth (Figure 1e). Again, two lines (L55 and L57) were topped, revealing that sucker growth was delayed by at least 1 week. In contrast to the P#15_2.5kb::barnase plants, seeds from the P#1_2.4kb::barnase plants germinated, allowing comprehensive analysis of the T₁ generation. Ten L7 and nine L23 plants were compared to 10 vector control plants, and again, no axillary bud primordia formed during vegetative growth in the transgenic plants (Figure 1e). After topping, axillary bud initiation in L7 plants was delayed by 2 weeks. Four weeks post-topping, all L7 plants formed fewer than eight, short suckers (vector control average ~16) with an average weight reduction of ~65% (Figure 1e). Axillary bud initiation was completely abolished in seven L23 plants even 4 weeks post-topping, and the other two plants produced fewer and shorter suckers with a reduced weight (Figure 1e). However, L23 plants were also smaller with thinner leaves than vector controls.

Finally, we conducted a field study (Southern Piedmont AREC, Blackstone, Virginia) with 20 offspring each from three T₁ parents of line L7. The T₂ plants were cultivated from April to September 2019, and a comparative analysis of suckers 2 weeks post-topping revealed that 43% of the transgenic plants and the 10 wild-type controls formed >10 suckers (Figure 1f). However, 45% of the transgenic plants (27 of 60) formed ≤5 suckers and ~12% (7 of 60) formed 6–10 suckers, confirming the stability of the trait even under field conditions.

In summary, we found that the axillary meristem-specific expression of barnase significantly delays and reduces sucker number, length and weight in tobacco after topping, adding to the body of knowledge on the control of axillary branching in plants. For tobacco field cultivation, our results may help to reduce the use of chemicals and the laborious work required for sucker control.

芽分枝是决定植物结构并影响作物生产力的重要农艺性状（Shen等， 2019 ）。来自芽顶端分生组织 (SAM) 的分子信号创建了一个激素环境，该环境与 BRANCHED1 (BRC1) 等腋芽特异性阻遏物的表达相结合，以抑制腋芽形成（Wang等人， 2019 年及其中的参考文献）。通过打顶（SAM 去除）消除信号，从而使腋芽形成新芽（吸盘）（图 1a）。在烟草（ Nicotiana tabacum ）中，在收获前打顶对于促进叶片发育/成熟是必要的，但打顶后的吸芽生长是不可取的，因为它会将资源重新分配给腋芽，从而降低主叶的产量和质量。脂肪醇、氟节胺或马来酰肼可以抑制出芽生长，但化学控制既耗时又昂贵，而且由于环境变化，这些化学物质可能在叶子加工后持续存在（Bailey等人， 2019 ）。因此，具有延迟的腋芽萌发或枝条生长的烟草植物将显着提高收获和/或产品质量，如之前针对其他物种所显示的(例如，Groot等人， 1994 )。

为了分析打顶前后烟草腋生分生组织/芽的转录组，我们在温室中种植植物8周，并采集了7个样本（ n = 3），包括打顶前的幼叶、SAM和腋生分生组织/芽，以及2、6、打顶后 24 小时和 72 小时，来自具有 8-10 个完全展开叶子的植物。在 Illumina HiSeq 2000 设备上提取 RNA 进行 RNA 测序分析（100 bp 单次读取，每个样本至少 3000 万次读取），我们鉴定了 17 个在打顶后腋芽中失调的候选基因（图 1b）。通过 qPCR 验证表达，并选择六个最有希望的基因进行进一步分析。使用 GUS 报告基因分析在打顶前后商业深色烟草的腋生分生组织中分析相应的启动子。其中四个启动子显示出非特异性活性，但由启动子 P#1 _2.5kb和 P#15 _2.5kb驱动的gusA表达仅限于腋生分生组织，其中 P#15 显示出更受限制的空间域（图 1c）。即使打顶后 7 天，P#15 _2.5kb ::GUS 活性仍保持稳定，但 P#1 _2.5kb :: GUS转基因品系中的 GUS 活性在打顶后不久就下降。 P#1 _2.5kb序列分析显示，在 –2401 至 –2407 位点存在糖抑制元件 TTATCCA（Tatematsu等， 2005 ）。将 P#1 _2.5kb缩短至 2.4 kb (P#1 _2.4kb ) 不会改变其腋生分生组织特异性，但会延长其活性，因此在 P#1 _2.4kb打顶后 10 天仍可检测到 GUS 染色 :: GUS转基因植物（图1c）。

基因 #1 编码 BRC1 同源物，甚至在打顶之前沉默也会增强吸芽的生长，正如其他物种所报告的那样，而由组成型 CaMV35S 启动子驱动的强烈过度表达是致命的，仅允许一个生长严重发育迟缓的转基因品系再生（数据未所示）。具有弱基因#1表达的转基因品系显示出与野生型对照相当的出芽发育，并且由P#1 _2.4kb驱动的基因#1的表达仅略微降低了出芽生长（数据未显示），可能反映了内源调节和/或芽形成的正调节因子，例如 NtBRC2 (Ding et al ., 2020 )。 Gene#15 编码一种豌豆球蛋白样蛋白，RNAi 和组成型过表达均未产生显着的表型（数据未显示）。

为了选择性抑制腋芽萌生和随后的吸芽生长，我们在腋芽特异性启动子 P#15 _2.5kb和 P#1 _2.4kb的控制下表达来自解淀粉芽孢杆菌的细胞毒性核糖核酸酶 Barnase，以消除负责的细胞。用于吸盘形成。我们最初生成了 11 株 P#15 _2.5kb ::芽孢杆菌RNA酶转基因植物，其中 7 株在营养生长过程中没有发育出腋芽原基。接下来，我们对两条线（L10 和 L11）进行打顶，即使在打顶后 1 周也看不到腋芽原基（图 1d）。这些植物表现出正常的表型，但与野生型对照相比，腋芽萌发延迟了至少 3 周，导致吸盘更少、更短，甚至在打顶后 4 周，重量也减少了 50%–79%（图 1d） ）。其余品系的种子未能在选择培养基上发芽，表明 P#15 _2.5kb启动子（以及芽孢杆菌RNA酶）在种子形成/发芽过程中也具有活性。我们再次分析了 P#15 _2.5kb :: GUS转基因品系，并确认了种子中较弱的 GUS 染色（图 1d）。

由 P#1 _2.4kb启动子驱动的芽孢杆菌RNA酶表达也强烈抑制 T ₀植物中的吸芽生长，并且 21 株植物的初始群体中的 7 株在营养生长过程中不产生或仅产生少量腋芽原基（图 1e）。再次，两条品系（L55 和 L57）被顶住，表明出芽生长延迟了至少 1 周。与 P#15 _2.5kb :: barnase植物相比，P#1 _2.4kb :: barnase植物的种子发芽，允许对 T ₁代进行全面分析。将十个 L7 和九个 L23 植物与 10 个载体对照植物进行比较，同样，在转基因植物的营养生长过程中没有形成腋芽原基（图 1e）。打顶后，L7 植株腋芽萌发延迟了 2 周。打顶后 4 周，所有 L7 植物形成少于 8 个短吸盘（病媒对照平均约 16 个），平均重量减少约 65%（图 1e）。即使在打顶后 4 周，七株 L23 植物中的腋芽萌发也完全消失，另外两株植物产生的吸盘更少、更短，重量也减轻（图 1e）。然而，L23 植物也比载体对照更小、叶子更薄。

最后，我们进行了一项实地研究（Southern Piedmont AREC，Blackstone，Virginia），其中来自 L7 系的 3 个 T ₁亲本的各 20 个后代。 T ₂植物于 2019 年 4 月至 9 月栽培，打顶后 2 周对吸芽的比较分析显示，43% 的转基因植物和 10 个野生型对照形成 > 10 个吸芽（图 1f）。然而，45% 的转基因植物（60 株中的 27 株）形成了 ≤5 个吸盘，约 12%（60 株中的 7 株）形成了 6-10 个吸盘，这证实了即使在田间条件下该性状的稳定性。

总之，我们发现芽孢杆菌RNA酶的腋生分生组织特异性表达显着延迟和减少打顶后烟草的吸芽数量、长度和重量，增加了控制植物腋生分枝的知识体系。对于烟田种植，我们的结果可能有助于减少化学品的使用和吸芽控制所需的繁重工作。