The study of the gravity response of roots and shoots is of great significance when exploring the polarity of plants and the development of the forest industry.
Forests provide the most important natural raw material for industry and the environment 
. Trees show obvious orientational growth in their stems and roots, and gravity is one of the most important environmental signals during this plant orientational growth 
. Plants can reorient their growth after sensing gravitational stimuli to maintain the optimal angle of their organs. Although the first research of tropism was performed approximately 300 years ago, the fundamental aspects of the underlying mechanisms are still unclear 
Until now, researchers divided the response to gravity into four steps: (i) the sensing of gravity; (ii) the transduction of the gravity signal; (iii) the generating, maintaining, and transmitting of the auxin gradient; and (iv) the response of the organ 
. The roots manifest positive gravitropism, while the shoots manifest negative gravitropism 
. In the first step, statocytes, types of cells that are located in the root columella and the shoot endodermal, have been identified as being able to sense gravity 
. Amyloplasts are enriched in these cells 
. With amyloplast sedimentation, an auxin gradient is generated 
, which finally promotes the downward curvature of roots and the upward curvature of shoots. Several genes have been verified as being able to generate an auxin gradient 
. The most studied genes in this step are PINs
, which encode auxin efflux carriers 
Previous studies have found that the localization pattern of PIN proteins is consistent with the direction of auxin transport, and their uneven polarity distribution on the plasma membrane is associated with a change in the auxin concentration gradient 
. In Arabidopsis
, a total of eight PIN proteins were detected, and at least five of them were found to be related to gravitropism. In the roots, PIN3 and PIN7 were identified in the lower side of the root cap columella cells, initiating the differential flow of auxin toward the lower flank of the root 
. The AUXIN-BINDING PROTEIN 1 (ABP1) and the TRANSPORT-INHIBITOR-RESISTANT 1 (TIR1)/AUXIN SIGNALING F-BOX (AFB) receptors can bind to auxin and activate or suppress cell expansion on the upper and lower sides of the roots and shoots 
Previous studies have also identified that SNARES can affect the gravitropism of shoots 
. SNARES are named after SNAP (soluble NSF attachment protein) receptors and are small proteins that mediate vesicle fusion 
. However, the manner in which gravity signals transduce to auxin gradient-related genes is still unclear.
(Koidz) is widely distributed across the world, especially in China 
. Additionally, willows have the characteristics of easy rooting and germination. Thus, they are an ideal model system for studying the response to gravity.
2. Overview of the RNA Sequencing Data
Gravity may affect many traits of the roots and shoots of plants. Twenty days after being hydroponically cultured, the roots showed positive gravitropism, while the shoots showed negative gravitropism (Figure 1a). Additionally, the shoots were much smaller and the roots much shorter under the inverted condition than under the normal condition. To characterize the role of the response of active genes to gravity in the roots and shoots, deep-sequencing libraries were generated using the total RNAs extracted from the roots and shoots under normal and inverted conditions. After trimming off the adapter sequences and removing the low-quality reads, we obtained 19,434,675–22,891,810 clean reads for the 12 libraries, with a single read length of 90 bp and a Q30 percentage (percentage of the sequences with sequencing error rates lower than 0.1%) over 94%. The clean reads were then mapped onto the reference genome of S. matsudana using HISAT2. In total, 35,585 (61.52% of the 57,841 gene models in the reference genome) genes were identified as being expressed in at least one library. Moreover, 5715 and 4525 genes were identified as only expressed in the shoots and roots, respectively. While considering the effect of gravity, 1111 and 951 genes were specially expressed under the inverted condition in the roots and shoots, respectively.
Figure 1. The roots and shoots showed different responses to gravity: (a) the phenotypes of the roots and shoots under normal and inverted conditions; (b) Venn analysis of the differentially expressed genes (DEGs) in the roots and shoots.; (c) Venn analysis of the up- and down-regulated DEGs in the roots and shoots.
3. The Differentially Expressed Genes between the Roots and Shoots under Normal and Inverted Conditions
The DEGs under normal and inverted conditions in the roots and shoots were then identified using a threshold FDR ≤ 0.05 and the absolute value of log2-fold change ≥ 1, respectively. A total of 412 and 668 DEGs were identified in the roots and shoots, respectively (Figure 1b). Under normal conditions, most of the DEGs showed high expression levels in the roots (326/412) and low expression levels in the shoots (446/668) (Figure 1c). According to the Venn diagram (Figure 1c), nine and three DEGs were identified, respectively, as being up- and down-regulated in both the roots and shoots under the inverted condition. Moreover, 48 DEGs were identified as showing the opposite trend of expression in the shoots and roots. Interestingly, all of these 48 DEGs showed up-regulation in the shoots and down-regulation in the roots. Among the 412 and 668 DEGs, only 60 were identified in both the roots and the shoots. This result means that the mechanisms of the response to gravity in the roots and shoots may differ.
4. Construction of Gravity-Related Gene Expression Networks in the Roots and Shoots
To further understand the difference between the gravity response of the roots and shoots, enrichment analysis of their DEGs was performed. As shown in Figure 2a, most of the DEGs were enriched in transcription according to the eggnog function classification. The DEGs were also selected for GO enrichment analysis. For molecular functioning, DNA binding, and transcription factor activity, sequence-specific DNA binding was enriched. The nucleus and the plasma membrane were enriched in the cellular component. For biological processes, oxidation reduction processes and the regulation of transcription were identified as being enriched.
Figure 2. (a) GO and (b) KEGG analysis of the differentially expressed genes (DEGs).
We then compared the KEGG class of up- and down-regulated DEGs between the roots and shoots. The up-regulated genes were enriched in the carbon metabolism and nitrogen metabolism in the roots, while enriched in the amino sugar and nucleotide sugar metabolism in the shoots. Down-regulated genes were enriched in the galactose metabolism and DNA replication in the roots and shoots, respectively (Figure 2b). According to the GO enrichment analysis, the enriched items in the cellular component of the roots and shoots were detected. For the DEGs in the shoots, the up-regulated genes were mostly enriched in the plasma membrane, cell wall, and extracellular region, while the down-regulated genes were mostly enriched in the nucleus. In contrast, the up-regulated DEGs in roots were mostly enriched in the nucleus. Although the most enriched item of the down-regulated DEGs in roots was the nucleus, the plasma membrane was also significantly enriched (Figure 3a).
Figure 3. GO and KOG enrichment analysis of the up- and down-regulated differentially expressed genes (DEGs) in the roots and shoots: (a) cellular component enrichment of the up-regulated and down-regulated DEGs in the roots and shoots; (b) KOG enrichment analysis of the up-and down-regulated DEGs in the roots and shoots.
The KOG function classification analysis also showed that the signal transduction mechanism and transcription were enriched at high levels in both the shoots and the roots (Figure 3b). We then compared the DEGs in the signal transduction mechanism in the roots and shoots. Interestingly, all of these genes in the roots were up-regulated, annotated as serine/threonine protein kinases (STPKs) and Wall-associated receptor kinases (WAKs). In the shoots, the genes encoding the STPKs and MAPKs were down-regulated, while the genes encoding the lectin receptor-like kinases (LecRLKs) were up-regulated. These results indicate that the roots’ response to gravity by the STRKs regulated the network, while the shoots’ response to gravity mainly through the LecRLKs regulated the network. Regarding transcription, opposite trends were also observed in the roots and shoots. Nearly all of the transcripts were up-regulated in the roots but down-regulated in the shoots, and more than half of the transcripts were AP2/ERFs, indicating that ethylene may be more important in the roots than in the shoots.
Protein interaction networks were constructed in the roots and shoots. Furthermore, nine and 12 hub genes were detected in the roots and shoots, respectively. In the roots, all nine hub genes were up-regulated under the inverted condition. Moreover, four of the nine hub genes were annotated as Kip-related proteins (KRPs), which are cyclin-dependent kinase (CDK) inhibitors (CKIs) and can negatively regulate cell division 
. In the shoots, only two hub genes were up-regulated under the inverted condition, annotated as tyrosine kinases. Meanwhile, eight of the 10 down-regulated hub genes were annotated as being associated with DNA replication, recombination, and repair, such as minichromosomal maintenance (MCM) proteins and DNA polymerases. These results indicate that gravity may inhibit cell division in both roots and shoots in different ways. In the roots, cell division inhibitors (KRPs
) were up-regulated, while DNA helicase MCMs were down-regulated in the shoots.
5. Analysis of the Gravity-Induced Alternative Splicing Genes in S. matsudana
We then analyzed the alternative splicing genes in the roots and shoots. As a result, a total of 185 and 114 genes were identified in the roots and shoots, respectively. Most of the gene alternative splicing events skipped exons, with only the genes in the roots being identified as mutually exclusive exons. Fourteen genes were identified as having alternative splicing events in both the roots and the shoots. Among these skipped exon genes, only 19 showed the existence of significant differences under the normal and inverted conditions. However, most of these 19 genes were annotated as uncharacterized proteins. According to the annotation, two genes were treated as gravity-related alternative splicing genes, encoding the soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) protein and the APETALA 2/ethylene-responsive factor protein (AP2/ERF). SNARE was annotated as being related to the transport of auxin on the plasma membrane, while the AP2/ERF protein is known to transduce the signal of ethylene and regulate the expression of downstream genes in the nucleus. These results indicate that auxin may play an important role in the negative geotropism of the shoots, while the geotropism of the roots may be mainly related to ethylene, which is consistent with the results above.
The protein sequences of these two genes were then analyzed. As shown in Figure 4a, the Snapin/Pallidin domain had changed in SNARE, and the first AP2 domain had changed in AP2/ERF. Under the inverted condition, SNARE and AP2 lost a domain in the roots and shoots, respectively. The alternative splicing of these two genes may result in a loss or change of their function.
Figure 4. Alternative splicing-induced protein domain analysis: (a) alternative splicing caused the AP2 domain to change in AP2/ERF; (b) alternative splicing caused the loss of the Snapin/Pallidin domain in SNARE. The skipped exons are marked in red.
6. Expression Profiling of the Gravity-Related Genes
The relationships between these two genes and the DEGs were then analyzed, and the correlations between the alternative splicing genes and the DEGs were measured. As expected, a total of 369 DEGs in the roots showed significant correlations with AP2/ERF. Moreover, AP2/ERFs were identified as showing opposite expression trends in the roots and shoots. In the shoots, the correlations between the expression levels of the DEGs and SNARE were not significant. The function of SNARE is to help transport auxin on the plasmalemma, which may not affect the expression level of the DEGs directly. According to the annotation enrichment analysis, correlation analysis, and the hub genes of the protein interaction networks, the LecRLKs, SNAREs, AP2/ERFs, and MCMs in the shoots and the STPKs, WAKs, AP2/ERFs, and KRPs in roots were further verified by qRT-PCR analysis (Figure 5). AP2/ERFs were significantly up-regulated in the roots and significantly down-regulated in the shoots (Figure 5a). The LecRLKs, which were highly expressed in the shoots and showed a much higher expression level under the inverted condition, were barely expressed in the roots (Figure 5b). In the roots, STPK and WAK1 were highly expressed under the inverted condition (Figure 5c,d). We also examined the expression level of one SNARE gene, SYP121. As shown in Figure 5e, it was highly reduced in the shoots under the inverted condition. Two hub genes (KRP in the roots and MCM in the shoots) were also detected (Figure 5f,g), consistent with the transcriptome.
Figure 5. The expression trends of the gravity responses-related genes in the roots and shoots. The qRT–PCR analysis of the transcript levels of the candidate genes in the roots and shoots under the two conditions. The Y-axis represents the relative expression level, while the X-axis represents the different cultivars. The error bars denote the standard error (SE). The expression level was significantly different at * p < 0.05 and ** p < 0.01.