In ornamental crops such as Hydrangea, reduction of shoot length and increased branching contribute to the production of high-quality, compact, and heavily flowered plants that meet markets’ and consumers’ criteria (Boumaza et al., 2010; Megersa et al., 2018). These criteria are classically obtained after repetitive chemical treatments using plant growth regulators (PGRs), especially growth retardants (Dilta et al., 2015; Rademacher, 2015; Jamwal et al., 2018). Gibberellin biosynthesis inhibitors, such as daminozide, are the most widely used growth retardants in plant production, as they account for approximately 40% of the global PGR market (Rademacher, 2017). However, several studies revealed the negative impacts of these chemicals on both humans and the environment, which has led public authorities to progressively phase out growth retardant use in horticultural industries (Andersen and Andersen, 2000; Sorensen and Danielsen, 2006; Bergstrand, 2017). This regulation strongly impacts the ways ornamental plants are produced today (Carvalho et al., 2008). Luckily, numerous studies have shown that environmental factors have an impact on plant compactness (Huché-Thélier et al., 2011; Morel et al., 2012; Alem et al., 2015; Li-Marchetti et al., 2015; Demotes-Mainard et al., 2016; Huché-Thélier et al., 2016; Bergstrand, 2017; Crespel et al., 2018; Runkle et al., 2019), and these responses could be exploited to develop non-chemical-based dwarfing methods. This strategy is underway and has brought some success in several species, but it remains an urgent challenge to overcome for most crop productions.

Mechanical stimulation (MS) is one of these alternative methods (Börnke and Rocksch, 2018), and it has been inspired by the thigmomorphogenetic effects of winds on plants (Jaffe, 1973). Wind sways impose recurrent mechanical loads on plants, which can be uni- or multi-directional, of variable intensities and durations, depending on several factors such as location and topography (de Langre, 2008; Gardiner et al., 2016). These characteristics are different from the long-lasting unidirectional mechanical loads imposed by gravity on leaned stems, with these phenomena already being well documented (Moulia and Fournier, 2009; Ramos et al., 2012; Gardiner et al., 2014). Thigmomorphogenesis under wind sways is often characterized by a reduction of shoot elongation together with an enhancement of diametrical growth (Neel and Harris, 1971; Jaffe and Forbes, 1993). This leads to the development of smaller plants with larger stems more capable of withstanding strong winds and maintaining their vertical position (Bonnesoeur et al., 2016; Gardiner et al., 2016). Often, this is accompanied by an increased branching and a reduction of the global leaf area (Vernieri et al., 2003; Morel et al., 2012). All these characteristics meet in fact the goals of the ornamental industry (Andersen and Andersen, 2000). In addition, MS sometimes induces a delay in flowering time, as observed in Arabidopsis thaliana (Lange and Lange, 2015), Helianthus annuus (Goodman and Ennos, 1996), and Brachypodium distachyon (Gladala-Kostarz et al., 2020).

Several techniques were developed to mimic wind effects on plants for either research purposes or plant production: shaking plants under air streams or using a vibrating table, bending stems by hand, or through recurrent passing of a material, such as a solid bar or with a softer material. So far, these techniques were applied to a few ornamental species in production, such as Salvia splendens (Vernieri et al., 2003; Mensuali-Sodi et al., 2006), Callistephus chinensis (Autio et al., 1994), Petunia x atkinsiana (Jędrzejuk et al., 2020), and Impatiens capensis (Anten et al., 2009). In rosebush, for example, MS performed using a solid bar effectively reduced stem elongation by up to 23% in some varieties (Crespel et al., 2018). Nevertheless, thigmomorphogenetic responses strongly depend on plant genotype and the plant’s herbaceous, climbing, or woody growth habit (Börnke and Rocksch, 2018). They also depend on the mechanical load intensities, frequencies, and directions (de Langre, 2008; Niez et al., 2020).

Mechanisms behind wind-induced thigmomorphogenesis have been less investigated compared to gravimorphic mechanisms (Niez et al., 2020). To this day, investigations have been only dealing with very few species, mainly Pinus taeda (Telewski and Jaffe, 1986; Dranski et al., 2018), Abies fraseri (Telewski, 1989), poplars and their hybrids (Kern et al., 2005; Fluch et al., 2008; Niez et al., 2019; Niez et al., 2020), A. thaliana (Braam and Davis, 1990; Antosiewicz et al., 1997; Johnson et al., 1998; Paul-Victor and Rowe, 2011; Zhdanov et al., 2021), Solanum lycopersicum (Gartner, 1994; Cipollini and Redman, 1999; Coutand et al., 2000), and more recently B. distachyon (Gladala-Kostarz et al., 2020; Coomey et al., 2021). Thus, research needs to broaden its range to a larger plant panel in order to decipher the entire machinery behind the responses to this mechanical stimulus.

Histological investigations revealed that within the woody stems of Pinus, Abies, and Populus plants that have been subjected to non-static swaying, due to transient symmetrical or asymmetrical unidirectional bendings, a particular wood, called flexure wood (Telewski, 1989), develops itself in the bent stems, improving its tolerance to mechanical loads and maintaining verticality while diminishing mechanical failures (Telewski, 2016; Roignant et al., 2018; Niklas and Telewski, 2022). When unidirectional mechanical load is applied, flexure wood sometimes develops asymmetrically, displaying an elliptical shape on stem cross-sections in the direction of bending (Telewski, 1989; Telewski, 2016; Roignant et al., 2018). However, this does not occur in all species; therefore, these opposite responses need to be further investigated (Telewski, 1989). In poplar, which is the most investigated woody angiosperm on this subject, significant reductions of vessel lumen areas, vessel diameters, and vessel frequencies in response to repeated bi-directional flexures (backward and forward) were observed in the flexure wood, and the development of wood fibers with no cellulosic G-layer was also reported (Kern et al., 2005). Interestingly, in this same species, unilateral bending caused radically different responses. In the stretched side of the stem, differentiation of up to one-third of wood fibers with a cellulosic G-layer occurred, while in the compressed side of the stem, there is almost no observed development of wood fibers containing such G-layer (Roignant et al., 2018). These results illustrate how plants may respond differently and adapt according to the characteristics of the mechanical load they encounter, in particular the contrast between symmetrical and asymmetrical bendings.

At the molecular level, there is very little information about the mechanisms involved in the perception and responses of the stem to recurring mechanical loads mimicking wind impacts. Most molecular data were about hybrid poplar after one or two asymmetrical bendings (Martin et al., 2010; Pomiès et al., 2017). In correlation with the histological impact of MS on stems, several cell wall genes were identified as responsive within a few hours after stimulation. For example, upregulation of the closest homologs of touch gene TCH4 occurs after one single or two asymmetrical bendings of the poplar stem (Martin et al., 2010). TCH4 is a member of the XET/XTH family encoding xyloglucan endotransglycosylase/hydrolases involved in the control of cell wall plasticity (loosening or thickening) in secondary xylem cells of Arabidopsis (Xu et al., 1995; Kushwah et al., 2020). This upregulation correlates well with an accumulation of TCH4-encoded XET protein in the epicotyl pith parenchyma cells, hypocotyl epidermis, and primary xylem in Arabidopsis subjected to 4 days of continuous blowing (Antosiewicz et al., 1997). Other genes encoding parietal proteins, such as CELLULOSE SYNTHASE and FASCICLIN-LIKE ARABINOGALACTAN PROTEINS, and some PECTINASES, including POLYGALACTURONASES and PECTIN ESTERASES, were also upregulated after a single asymmetrical flexure in poplar stem, while some such as the cell wall expansion gene EXPANSIN B3 were downregulated (Pomiès et al., 2017). In the same study (Pomiès et al., 2017), the upregulation of genes involved in lignin biosynthesis such as CINNAMOYL ALCOHOL DEHYDROGENASE and S-ADENOSYL-L-METHIONINE SYNTHETASE were also reported. Indeed, the latter is involved in the synthesis of S-adenosyl-l-methionine, an important methyl donor during the biosynthesis of both G- and S-type lignin units (Li et al., 2022). In addition, repression of the expression of the CELL DIVISION CYCLE gene within the first 30 minutes following stimulation was measured and correlated with the transient cessation of cambial growth observed during the first hours after one bending prior to the resumption of secondary growth (Pomiès et al., 2017).

When plants are subjected to recurrent mechanical stimulation, attenuation of the thigmomorphogenic responses is often observed, and this phenomenon may prevent unnecessary reduction of growth while still allowing plants to resist (Martin et al., 2010; Pomiès et al., 2017). For example, in poplar, 96% of the early mechanoresponsive genes showed reduced response to a second bending applied 24 h after the first one (Pomiès et al., 2017). Despite the importance of understanding plant’s long-term responses to MS under windy natural environments and in horticultural crops, knowledge of the mechanisms behind such desensitization reactions is still in its infancy (Brenya et al., 2022). Only two recent studies report molecular control after long-term recurrent mechanical stimulations of stems. First, in hybrid poplar, a link was established between the cell wall modifications in flexure wood after 8 weeks of recurrent asymmetrical bending of the stem and the expression of cell wall formation genes in stems (in particular FASCICLIN-LIKE ARABINOGALACTAN PROTEINS) (Roignant et al., 2018). Second, in B. distachyon, a monocotyledon where no cambial secondary growth occurs, a recent report links the inactivation of gibberellin responses with SECONDARY WALL INTERACTING bZIP transcription factors and secondary wall thickening in roots after 3 weeks of daily mechanical bending of the plant (Coomey et al., 2021).

With the aim to broaden knowledge and evaluate the genericity of reported plant responses to recurrent long-term mechanical loads and desensitization, we investigated the responses of Hydrangea macrophylla, an important ornamental bush species, making it the only woody angiosperm species after poplar to be examined from plant to molecular levels after wind-mimicking MS. We focused on histological changes through analyses of stem cross-sections and quantitative expressions of nine MS-responding genes that were already described in literature (Martin et al., 2010; Pomiès et al., 2017) as related to cell wall and cell growth. We also observed for the first time that XYLOGEN PROTEIN 1 (XYP1), a gene encoding an arabinogalactan protein that promotes xylem cell differentiation (Motose et al., 2004; Ma et al., 2014), is responsive to MS. We carried out our analysis on two portions of a stem, allowing the simultaneous comparison of MS responses in primary and secondary growing internodes along the same stem. We carried out measurements 3 weeks of recurrent daily MS treatment using two frequencies of MS in order to evaluate quantitative growth and desensitization responses according to the number of daily MS. Finally, in a perspective of replacement and/or reduction of growth retardants use in horticulture, we compared the efficiencies of MS and daminozide, as it is the most widely used PGR for plant compactness, and we discussed the potential benefit of MS in H. macrophylla production.

Little is known about the physiological and molecular processes induced by long-term, recurrent mechanical stimulation, as well as the mechanisms of desensitization. Also, very few plant species have been investigated on these subjects. In this paper, we studied for the first time the responses of the bush H. macrophylla to MS and deciphered some histological and molecular changes after 2 and 3 weeks of daily light (1B) or heavy (12B) MS treatments. In addition, we compared responses to MS and chemical plant growth regulator daminozide. When H. macrophylla plants were subjected to repetitive symmetrical MS, plants responded to the mechanical loads through a reduction of stem elongation, increased basal diametrical growth, and reduction of leaf expansion ( Table 1 , Figure 1 ). Since the same number of internodes was produced as in control plants, it appears that the reduction in stem height after MS was not due to a repressive effect of the mechanical load on SAM organogenic activity. Rather, stem bending led to a repression of internode elongation, both in those present before treatment and in the most distal internodes that emerged during treatment. According to Coutand et al. (2000), MS impacts not only the stimulated internodes but also the neighboring internodes. These authors demonstrated that the transient bending of the basal part of the tomato stem led to a long-distance effect on the elongating internodes, inducing a growth cessation. In our treatment, where the entire stem was bent by the plastic curtain passing over the plant, each internode was mechanically stimulated, suggesting that the reduced growth of each internode may have been due to direct stimulation. This does not preclude that long-distance signaling after MS between H. macrophylla basal and upper internodes may take place and should be investigated further.

Our treatment induced a strong increase in the diametrical growth of H. macrophylla stems, up to 98% with the 12B treatment. For this response, not all parts of the stem were affected, but only the most basal internode 1 that had achieved primary growth at the beginning of the treatment. This is consistent with the same differential responses observed along the stem of bell pepper plants when transiently and repetitively bilaterally bent (Graham and Wheeler, 2017). In A. fraseri (Telewski, 1989) or hybrid poplar (Populus tremula × Populus alba, clone INRA 717-1B4) (Roignant et al., 2018), repetitive unilateral flexures caused an ovalization of the stem due to an asymmetrical diametrical growth along the rays subjected to the highest deformation stimuli. In H. macrophylla, no asymmetrical growth was observed on bent stems after recurrent bilateral stimulation, even though asymmetrical growth can be produced in this species as illustrated by our experiment imposing permanent unilateral flexure ( Supplementary Material ). Such symmetrical growth of the stem after bilateral MS was also observed in poplar (Kern et al., 2005) and bell pepper stems (Graham and Wheeler, 2017) and suggests that the backward stimulation somehow interacts with the mechanosensitive responses of the forward stimulation.

The higher diametrical growth of the most basal internode 1 of stimulated H. macrophylla stem was caused by overall greater development of all tissues with the exception of pith tissues, with its proportion being lower than in control plants ( Figure 3B ). Interestingly, this is different from bean stems, where successive bendings stimulated the development of the pith tissues, allowing the hollow stem of the bean to become more solid and therefore more resistant to MS (Biro et al., 1980). Our results also differ from observation in A. thaliana, where mechanical loads on inflorescence stems using a similar mechanical treatment as on Hydrangea caused little variation in total stem diameter but were associated with an increase in cortical tissues together with a decrease in lignified and pith tissues (Paul-Victor and Rowe, 2011). These opposite responses may illustrate the different strategies of plants under mechanical constraints according to their growth habits (Niklas and Telewski, 2022): herbaceous plants, such as A. thaliana and beans with a short and flexible stem growth strategy, and woody species, such as poplar tree and Hydrangea shrub with a short and rigid stem growth strategy for maintaining upright axes in conditions of severe mechanical perturbations (Paul-Victor and Rowe, 2011).

Interestingly, when permanent static flexure of the H. macrophylla stem was imposed, the upper internodes, which were still elongating, displayed a stronger asymmetrical development of the cortical tissues than of vascular tissues ( Supplementary Material ). On the contrary, no asymmetrical development of cortical tissues was observed in the basal internode ( Supplementary Material ). This suggests that the herbaceous upper part of the H. macrophylla stem reacts to the permanent gravimorphic strain imposed in this experiment through the accumulation of cellulosic tissues rather than through lignified tissues, probably because the expansion of cellulosic cell walls still allows cell and organ elongation. This result also points out that for the same stimulus in the same species, the variability of responses of an organ depends on its developmental stage and needs to be examined carefully.

Our histological analysis demonstrated that the H. macrophylla strategy to increase stem stiffness under successive bilateral mechanical constraints is to promote secondary xylogenesis together with a strong thickening of the walls of secondary xylem fibers, parenchyma, and vessels in the basal internode ( Figures 2A , 3A , 4C, F ). Only one type of flexure wood developed in the stem cross-sections of Hydrangea after bilateral stimulation. On the contrary, in poplar, after successive unilateral mechanical loads, tensile flexure wood with fibers containing a cellulosic G-layer accumulates on the stretched side of the stem, while such fibers were not observed in the compressive flexure wood developing on the other side (Roignant et al., 2018). In Hydrangea, only wood fibers without a G-layer formed in the stem after MS. One hypothesis is that under bilateral bending, with both sides of the stem being repetitively stretched and compressed, a unique outcome for developing fibers occurs in Hydrangea. This likely corresponds to the best developmental response that allows the Hydrangea stem to appropriately withstand recurrent bilateral bendings. This could also imply that the signaling pathway after the stretching stimulus is somehow repressed by the compressive stimulus that follows or, as suggested by Roignant et al. (2018), switched on and off repetitively. Further investigations should bring new information on the dialog between these two developmental pathways and their interactions.

Our molecular data bring first knowledge on the molecular paths controlled by mechanical loads in Hydrangea ( Figure 5 ). Our results show that HmFLA11 is upregulated in basal internode 1 of H. macrophylla stem upon bending. HmFLA11 has a high similarity with AtFLA11 ( Supplementary Material ), a gene suspected to take part in cell surface sensing of mechanical stimulation, the upregulation of which leads to secondary cell wall development (MacMillan et al., 2010; Roignant et al., 2018; Ma et al., 2022a; Ma et al., 2022b; Ma et al., 2023) as we observed in this internode ( Figure 4 ). In basal internode 1, the increased development of secondary xylem by MS is also linked to the upregulation of the XYLOGEN PROTEIN 1 (HmXYP1), an arabinogalactan protein that promotes xylem cell differentiation (Motose et al., 2004; Ma et al., 2014). Recently, both XYP-LIKE PROTEIN, XYLP1 and XYLP2, were shown to promote lignin biosynthesis during Capsicum annuum stem development in response to drought and cold stresses (Zhang et al., 2022). Additionally, the strong accumulation of lignified cells in this secondary wood after MS may be associated with the higher transcription levels of the major lignin synthesis gene HmCAD1 observed after bendings. Downregulation of HmEXPB3 and HmXTH33, which both promote cell wall extensibility (Yamaguchi et al., 2023), may also contribute, together with the accumulation of lignin, to increasing stem stiffness in the basal part of the bent Hydrangea stem.

Our molecular data involved one time point, i.e., 48 h after the last bending of a 2-week treatment. Further transcriptomic studies in H. macrophylla will allow us to obtain more knowledge about different actor expressions within a broader time course after flexure. Nevertheless, the late molecular regulations observed 48 h after the last stimulation indicate the long-lasting impact of mechanical stimuli on the molecular machinery. Considering that stems received up to 120 stimulations in 2 weeks and still responded at the transcriptional level to the stimulus, these results point to limited desensitization of H. macrophylla stem under our conditions, if not non-existent. This is also illustrated by the constant reduction in stem elongation observed during each of the treatments during 3 weeks. Higher thigmomorphogenic responses under 12B as compared to 1B bending point out as well to a dose response in H. macrophylla. This dose response concerns primary and secondary growths of the stem as well as leaf expansion. The relationship between frequencies of mechanical bending and magnitude of growth responses was reported in several species such as Phaseolus vulgaris (Jaffe et al., 1980), Ulmus americana (Telewski and Pruyn, 1998), and S. lycopersicum (Coutand and Moulia, 2000). In U. americana, seedlings were significantly shorter with the increasing number of bendings after 3 weeks of experiment. Conversely, no significant differences were observed in diameter growth in the responses to different bending doses, and such responses may indicate an accommodation process of U. americana seedlings that is not observed in H. macrophylla.

Desensitization to MS is an important response that needs to be evaluated prior to using MS in crop production if an efficient and prolonged dwarfing effect is the aim. We established that 12 bendings per day for five consecutive days per week did not induce such desensitization in H. macrophylla, suggesting that at least within this frame, such treatment could be used in crop production. The harmless 12B treatment was two-thirds as effective as the hazardous daminozide treatment in increasing plant compactness ( Table 1 ). In contrast to PGR treatment, MS induced a supplemental and important increase of diametrical growth in the basal part of the stem, mainly resulting from greater developments of vascular and cortical tissues ( Figure 2B ). This MS response will benefit plant quality by increasing plant robustness, especially during transplantation and transportation. Additionally, no negative impact of MS was observed on rooting, time to flower, and inflorescence size, which are important criteria for ornamental crop production ( Figure 1 ). Thus, our results indicate that MS has strong potential to replace, all or in part, chemical dwarfing molecules in H. macrophylla production.

A summary of the thigmomorphogenic effects after long-term, recurrent, symmetrical MS on H. macrophylla stems is presented in Figure 6 with a comparison of other species. For some developmental responses, H. macrophylla matches certain species or is opposite to others, independent of their growth habits (ligneous vs. herbaceous). Moreover, not one single pattern of responses was observed for all parameters. For example, H. macrophylla responses to MS were similar to those of Brassica napus for reduction of leaf area but opposite to B. napus for root development and time to flower (Cipollini, 1999). Similarly, H. macrophylla shares the same responses to MS with Acacia koa for stem elongation and diametrical growth but not for root development. This suggests a thin regulation of mechanosensitive responses at the species level and confirms that overall MS impacts on a species, or even a variety (Jędrzejuk et al., 2020), cannot be predicted until assayed. Fewer species were studied for histological responses to recurrent symmetrical bendings. Figure 6 shows also similar and divergent histological responses in comparison to H. macrophylla across species. At the molecular level, only the grass B. distachyon has been studied in terms of responses to recurrent, long-term symmetrical bending of the stem, but only root transcriptome modifications were assessed (Coomey et al., 2021). In the roots of this species, several cell wall genes were also regulated after MS loading. Evidence between secondary cell wall thickening after MS and gibberellin inactivation was also provided in B. distachyon. Further investigations in H. macrophylla should therefore examine the hormonal responses in bent stems after recurrent symmetrical MS to complete this scheme.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/ Supplementary Material.

BL-N: Formal analysis, Investigation, Methodology, Project administration, Validation, Visualization, Writing – original draft, Writing – review & editing. HR: Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Validation, Writing – review & editing. NB: Investigation, Writing – review & editing. LH-T: Conceptualization, Supervision, Validation, Writing – review & editing. VG: Conceptualization, Supervision, Validation, Writing – review & editing. NL: Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Validation, Writing – original draft, Writing – review & editing.