Prunus Knotted-like Genes: Genome-Wide Analysis, Transcriptional Response to Cytokinin in Micropropagation, and Rootstock Transformation

Knotted1-like homeobox (KNOX) transcription factors are involved in plant development, playing complex roles in aerial organs. As Prunus species include important fruit tree crops of Italy, an exhaustive investigation of KNOX genes was performed using genomic and RNA-seq meta-analyses. Micropropagation is an essential technology for rootstock multiplication; hence, we investigated KNOX transcriptional behavior upon increasing 6-benzylaminopurine (BA) doses and the effects on GF677 propagules. Moreover, gene function in Prunus spp. was assessed by Gisela 6 rootstock transformation using fluorescence and peach KNOX transgenes. Based on ten Prunus spp., KNOX proteins fit into I-II-M classes named after Arabidopsis. Gene number, class member distribution, and chromosome positions were maintained, and exceptions supported the diversification of Prunus from Cerasus subgenera, and that of Armeniaca from the other sections within Prunus. Cytokinin (CK) cis-elements occurred in peach and almond KNOX promoters, suggesting a BA regulatory role in GF677 shoot multiplication as confirmed by KNOX expression variation dependent on dose, time, and interaction. The tripled BA concentration exacerbated stress, altered CK perception genes, and modified KNOX transcriptions, which are proposed to concur in in vitro anomalies. Finally, Gisela 6 transformation efficiency varied (2.6–0.6%) with the genetic construct, with 35S:GFP being more stable than 35S:KNOPE1 lines, which showed leaf modification typical of KNOX overexpression.


Introduction
In all eukaryote genomes, TALE transcription factors are typified by a homeodomain (HD) with a three amino acid loop extension between helices I and II. Plant TALEs encompass the KNOTTED-like (KNOX) and BELL-like (BELL) factors that form heterodimers and cooperate in organ development [1]. The Arabidopsis KNOX (KNAT) classification was framed on structural and expression features into class I (HD identity >73% vs. maize Kn1 and meristem expression; STM, BP/KNAT1, KNAT2, and KNAT6), class II (one intron within the ELK domain and widespread transcription: KNAT3, KNAT4, KNAT5, and KNAT7), and class M (devoid of HD, KNATM), conventionally named KNOXI, KNOXII, and KNOXM. This grouping has been suitable to cluster KNOX from several plant species [2]. Synoptically, class I genes work to maintain meristematic identity and are associated with cell proliferation. Referring to epigeous organs, they have several functions, such as shoot apical meristem formation and development (STM) and meristem maintenance (KNAT6), leaf shape diversity [3], carpel identity (KNAT2), stem inflorescence stress at the molecular level can reveal the clone's health status and provide information for solutions to mitigate and/or avoid aberrations. Contextually, the KNOXs are known to play roles in aerial organ development by cross talking with hormones and CKs (see above), and the study of their response to CK variation is expected to give insight into causes affecting propagule performance and subsequent acclimatization capacity. Relatedly, the expanding technologies of genome modification in Prunus are tightly dependent on MP [30,31]. Specifically, peach (the reckoned Prunus model species) is among the most recalcitrant to transformation, mainly due to inefficient regeneration, and thus is difficult to subject to gene function surveys and biotech-based breeding [32], while genetic engineering of several Prunus rootstocks has been feasible [33]. Finally, the fine control of TF to improve crop traits has been applied in trees [34] and desired in Prunus spp. [35].
This work aimed at providing a genome-wide computational and updated characterization of KNOX in Prunus fruit trees (PRUNOX), focusing on CK regulatory elements, followed by a survey on the PRUNOX transcription variation in response to CK during rootstock shoot multiplication. Gene transfer (GFP) was first assayed using the Gisela 6 rootstock system technology. Subsequently, the overexpression of peach KNOPE1, which is known to affect leaf shape in trees, was used to explore the gene's effects in aerial organs and its usefulness as a phenotypic marker of transformation.

Protein Features
A phylogenetic tree was built ( Figure 1A) using KNOX proteins from primary transcripts of ten Prunus spp. diploid fruit trees (PRUNOX), and the class I, II, and M members of Arabidopsis thaliana were used as an outgroup. Manual curation was mandatory to achieve highly confident coding sequences (Table S1), and the resulting phyletic groups were named referring to Arabidopsis classes. Among class I, a PRUNOX group, here named extra group, fell between the BP-like and KNAT2/6-like clades without having specific Arabidopsis counterparts. Overall, PRUNOXI and II maintained distinctive HD traits (identity > 70 and 55-58% vs. that of maize Kn1 HD; Table S1), class-specific motifs of MEINOX (KNOX1 and KNOX2) and class II-specific residues between the MEINOX and ELK [21,36]. The MEINOX of class M proteins, devoid of HD, was more similar to that of class I than II (30 and 26%). Residue conservation grade was investigated ( Figure 1B) in all PRUNOXs, using peach KNOX proteins (KNOPEs) as reference. Briefly, MEINOX, ELK, and HD maintained high identity levels (90-100%) within each type of PRUNOX of the examined species, while a grade of variability mostly occurred in the N-and C-termini (Table S1). Overall, molecular weights and isoelectric points showed very modest variations among the proteins of a fixed group, except for KNOPE2.1 and KNOPEM2 ( Figure 1B, Tables S1 and S2). Further analysis was focused on regions within variable stretches (conservation score ≤ 0.75) and bearing amino acid replacements (missense substitution at the DNA level), which turned out to be similar at the physical and chemical levels (Table S2). Moreover, divergent substitutions were classified and predicted to be tolerated (SIFT score ≥ 0.05), except for one observed within the extra group members.

Genomic Features
As for Prunus diploid genomes, 11 PRUNOX members were found in all species, except for P. avium which had 12 (Table S1). Moreover, PRUNOXIs were more numerous than PRUNOXIIs and PRUNOXMs and organized in a 6-3-2 module, except for P. avium which had a 6-4-2 one. The PRUNOXs were scattered on scaffolds/linkage groups (Table 1, Figure 2); hereafter, we provide details for the species (P. armeniaca, P. avium, P. dulcis, P. mira, P. mume, P. persica, P. salicina) that were fully assembled into eight chromosomes (Chrs).    Comparative schematic maps of KNOX genome distribution occurring in species of the Armeniaca section (e.g., P. mume, Pm) vs. that in other Prunus spp. (e.g., P. persica, Pp). The gene KNOPE2.1 is located in the unplaced scaffold1397 (Scf1397) in P. mume but in Chr 7 in P. armeniaca. The scale on the left represents chromosome lengths in megabases (Mb).
As for PRUNOXI, the two-copy STM-like genes recurred on Chrs 3 and 4; the one-copy BP-like members were on Chrs 2 (Armeniaca section) and 1 (all the other); the two-copy KNAT2/6-like genes were on Chrs 2 and 7 (Armeniaca sect.) and Chrs 1 and 5 (all the others). The one-copy genes of the class I extra group resided on Chr 6, except that of P. mume (Chr 1). As for PRUNOXII, the 1-3-copy KNAT3/4/5-like genes were on Chrs 2 and 8 (Armeniaca sect.) and on Chrs 1 and 7 (all the others); the one-copy KNAT7-like members were on Chr 7 (Armeniaca sect.) and Chr 5 (all the others). Finally, the 1-2-copy KNATM genes lay on Chr 2 (Armeniaca sect.) or on Chrs 1 and 5 (all the others).
Synoptically, the PRUNOX intron-exon (IN-EX) organization was mostly conserved as in Arabidopsis, except for the extra group and the class M members (Figure 3)  PRUNOX genomic organization was further analyzed for colinearity, duplication events, nonsynonymous/synonymous substitutions of colinear KNOX, and evolution time analysis using six Prunus spp. genomes (Table S1). Hereafter, we only report major outcomes that classify the Prunus multicopy orthologs to Arabidopsis STM, KNAT2/6, KNAT3/4, and KNATM genes as segmental duplications, that is, long stretches of duplicated sequences with high identity. Further manual alignments highlighted significant differences between introns (length/identity) in STM-like 1 and STM-like 2 gene groups (not shown); hence, transposon-mediated duplication-that is, gene fragments embedded into DNA transposons-might have taken place. Similarly, KNATM gene shuffling recalled transduplication. Retroduplication events (retrocopied intronless genes bearing a poly-A tail) were not addressed.
We looked for hormone-related cis elements sited in the KNOX promoters (1500 bp upstream the ATG) of peach and almond considering that they are the parents of GF677, a rootstock used for CK assays in this work. KNOX genes included a variable number of motifs for ABA, AUX, GAs, and ethylene (Table S3). Here, manual and bioinformatic tools deepened the search for CK motifs, which recurred in all PRUNOX members with similar abundance and positions in both species (Figure 4, Table S4). The PRUNOXI harbored 6-17 binding sites of 5-8 bp. The PRUNOXII hosted 7-18 motifs, mostly 5 bp long (>80%), PRUNOX genomic organization was further analyzed for colinearity, duplication events, nonsynonymous/synonymous substitutions of colinear KNOX, and evolution time analysis using six Prunus spp. genomes (Table S1). Hereafter, we only report major outcomes that classify the Prunus multicopy orthologs to Arabidopsis STM, KNAT2/6, KNAT3/4, and KNATM genes as segmental duplications, that is, long stretches of duplicated sequences with high identity. Further manual alignments highlighted significant differences between introns (length/identity) in STM-like 1 and STM-like 2 gene groups (not shown); hence, transposon-mediated duplication-that is, gene fragments embedded into DNA transposons-might have taken place. Similarly, KNATM gene shuffling recalled transduplication. Retroduplication events (retrocopied intronless genes bearing a poly-A tail) were not addressed.
We looked for hormone-related cis elements sited in the KNOX promoters (1500 bp upstream the ATG) of peach and almond considering that they are the parents of GF677, a rootstock used for CK assays in this work. KNOX genes included a variable number of motifs for ABA, AUX, GAs, and ethylene (Table S3). Here, manual and bioinformatic tools deepened the search for CK motifs, which recurred in all PRUNOX members with similar abundance and positions in both species ( Figure 4, Table S4). The PRUNOXI harbored 6-17 binding sites of 5-8 bp. The PRUNOXII hosted 7-18 motifs, mostly 5 bp long (>80%), while the PRUNOXM contained more than 14 elements.  . Cytokinin binding motifs residing on PRUNOX genes' promoters of P. persica and P. dulcis (left and right panels). The score accounts (covering 1.5 kb region before the start codon in Table S4) refer to the experimentally determined extended (ECRM; blue) and core (CRM; light blue) motifs [38], as well as octameric sequences (yellow/orange/brown) enriched in CK-responsive promoters [39].

Transcriptomic Features
We further addressed Prunus genome-wide transcription in aerial organs ( Figure 5), using publicly available RNA-seq data (Table S5), and only fruit and leaves recurred in all species. The PRUNOXIs (vertical blue bar), regardless of developmental stage, shared low expression (blue z-score values) in fruit and leaves; however, their high transcription levels (orange values) characterized meristem-rich organs such as buds and stems (hosting vegetative/floral meristems and cambium). They were all also active in phloem tissues in tested species. Most of the PRUNOXIIs (vertical orange bar) showed opposite patterns to PRUNOXIs in leaves, though exceptions recurred for KNOPE4 orthologs, while behaviors in fruits, buds, and phloem varied among members and with species. The trends of class M genes often recalled those of class I. . Cytokinin binding motifs residing on PRUNOX genes' promoters of P. persica and P. dulcis (left and right panels). The score accounts (covering 1.5 kb region before the start codon in Table S4) refer to the experimentally determined extended (ECRM; blue) and core (CRM; light blue) motifs [38], as well as octameric sequences (yellow/orange/brown) enriched in CK-responsive promoters [39].

Transcriptomic Features
We further addressed Prunus genome-wide transcription in aerial organs ( Figure 5), using publicly available RNA-seq data (Table S5), and only fruit and leaves recurred in all species. The PRUNOXIs (vertical blue bar), regardless of developmental stage, shared low expression (blue z-score values) in fruit and leaves; however, their high transcription levels (orange values) characterized meristem-rich organs such as buds and stems (hosting vegetative/floral meristems and cambium). They were all also active in phloem tissues in tested species. Most of the PRUNOXIIs (vertical orange bar) showed opposite patterns to PRUNOXIs in leaves, though exceptions recurred for KNOPE4 orthologs, while behaviors in fruits, buds, and phloem varied among members and with species. The trends of class M genes often recalled those of class I.

Pheno-Histological Features
GF677 rootstock microshoots were grown on media containing 1.7 and 5.1 µM BA. Effects of CK treatment were evidenced by the formation of higher numbers of leaflets and side shoots than controls 10 days post-treatment (dpt), while stem length was unvaried ( Figure 6A,F,K, and Table 2). In precocious histological analyses (3 dpt), the shoot apical meristem (SAM) and leaf axillary buds (LABs) of controls did not show anomalies ( Figure 6B-D), and stems adjacent to the medium did not produce callus ( Figure 6E). In microshoots treated with 1.7 µM BA, SAM and LABs ( Figure 6G,H) were similar to controls; some LABs started elongation ( Figure 6I), and callus occurred at the stem basis ( Figure 6J). At 5.1 µM BA, SAM was unaffected ( Figure 6L), though several microshoots bore LABs with normal ( Figure 6M) or swollen ( Figure 6O) morphology, or buds bulging from stem inner layers ( Figure 6N), and callus at the stem basis (not shown).

Figure 5.
Prunus genome-wide transcription analysis of PRUNOX genes active in aerial organs. The expression profiles refer to 10 species through heat maps of the z-scores, where orange and blue indicate higher and lower expression, respectively. fl., flower.

Pheno-Histological Features
GF677 rootstock microshoots were grown on media containing 1.7 and 5.1 µM BA. Effects of CK treatment were evidenced by the formation of higher numbers of leaflets and side shoots than controls 10 days post-treatment (dpt), while stem length was unvaried ( Figure 6A,F,K, and Table 2). In precocious histological analyses (3 dpt), the shoot apical meristem (SAM) and leaf axillary buds (LABs) of controls did not show anomalies (Figure 6B-D), and stems adjacent to the medium did not produce callus ( Figure 6E). In microshoots treated with 1.7 µM BA, SAM and LABs ( Figure 6G,H) were similar to controls; some LABs started elongation ( Figure 6I), and callus occurred at the stem basis ( Figure 6J). At 5.1 µM BA, SAM was unaffected ( Figure 6L), though several microshoots bore LABs with normal ( Figure 6M) or swollen ( Figure 6O) morphology, or buds bulging from stem inner layers ( Figure 6N), and callus at the stem basis (not shown).

PRUNOX Expression Patterns
Transcription of peach catalase (PpCAT) and CK-responsive (CKR) genes was monitored to mark events of oxidative stress and CK perception after BA treatment. CKRs were selected among Arabidopsis orthologs with ascertained behavior among the CK oxidases (PpCKX), responsive regulators (PpARR), and histidine kinases (PpHK). A further choice was based on genes acting in aerial organs of Rosaceae spp. and harboring CK-responsive motifs in promoters ( Figure S1). A statistical analysis of the effects of BA dose (D), time (T), and interaction effect (DxT) was carried out to assess the significance of the influence on transcriptional responses of markers and PRUNOX and BELL genes ( Table 3). As for PRUNOXI, BA concentration had significant effects at different levels on several genes tested. Similarly, time affected the expression of most class I KNOX genes (except for KNOPE6). DxT occurred for all class I genes except for STMlike1.

PRUNOX Expression Patterns
Transcription of peach catalase (PpCAT) and CK-responsive (CKR) genes was monitored to mark events of oxidative stress and CK perception after BA treatment. CKRs were selected among Arabidopsis orthologs with ascertained behavior among the CK oxidases (PpCKX), responsive regulators (PpARR), and histidine kinases (PpHK). A further choice was based on genes acting in aerial organs of Rosaceae spp. and harboring CK-responsive motifs in promoters ( Figure S1). A statistical analysis of the effects of BA dose (D), time (T), and interaction effect (DxT) was carried out to assess the significance of the influence on transcriptional responses of markers and PRUNOX and BELL genes ( Table 3). As for PRUNOXI, BA concentration had significant effects at different levels on several genes tested. Similarly, time affected the expression of most class I KNOX genes (except for KNOPE6). DxT occurred for all class I genes except for STMlike1.   Hereafter, we briefly describe transcription trends (Figure 7) in time at fixed BA dose followed by some highlights of KNOX members' peculiarities. As for stress markers ( Figure 7A,B), PpCAT1 but not PpCAT2 was upregulated with time with 1.7 µM BA. At higher BA, both PpCAT genes were at least 2-fold upregulated in time course, at higher levels at 24 than 72 h post-treatment (hpt). As for CKR markers ( Figure 7C,D), PpCKX6 expression was 2-fold higher at 24 hpt due to 1.7 µM BA, followed by restoration to control levels; PpARR12 and PpHK1 did not show significant response. Increasing BA to 5.1 µM caused a 4-fold increase in PpCKX6 transcription at 24 hpt followed by a level drop later on; the PpARR12 and PpHK1 genes were over 2-fold upregulated only at 72 hpt. All markers shared expression increase upon higher BA dose at 72 hpt, depicting a state of intense oxidative stress and modification of CK endogenous metabolism. Regarding PRUNOXI after treatment with 1.7 µM BA ( Figure 7E), all members showed higher transcription than controls (except for the unvaried KNOPE6). Onwards, all PRUNOX1 members decreased in expression to control levels, except for STMlike2, which was repressed. As for PRUNOXII ( Figure 7G), KNOPE3 showed comparable expression to controls with time, KNOPE4 was only slightly repressed at 24 hpt, and Regarding PRUNOXI after treatment with 1.7 µM BA ( Figure 7E), all members showed higher transcription than controls (except for the unvaried KNOPE6). Onwards, all PRUNOX1 members decreased in expression to control levels, except for STMlike2, which was repressed. As for PRUNOXII ( Figure 7G), KNOPE3 showed comparable expression to controls with time, KNOPE4 was only slightly repressed at 24 hpt, and KNOPE7 maintained upregulation that decreased in time course. Looking at the effects of higher BA concentration on PRUNOXI ( Figure 7F), transcript upregulation was significant only for STMlike1, KNOPE2, and KNOPE2.1 at 24 hpt; onwards, the latter two had mRNA levels similar to control, while STMlike1 was strongly repressed; the expression of KNOPE1, KNOPE6, and STMlike1 increased significantly from 24 to 72 hpt. The PRUNOXII ( Figure 7H) patterns were quite similar to those at low BA. Synoptically, at low BA, PRUNOXI shared the "up and down" regulation pattern with time, while member-specific diversified responses occurred at high BA from 24 to 72 hpt, suggesting that CK concentration alters regulatory mechanisms of PRUNOXI more intensely than those of PRUNOXII, these latter having a time-unvaried pattern under both BA dosages.
As for BELL responses to 1.7 µM BA ( Figure 7I), BEL1, BLH2, BLH5, and BLH6 were repressed at 24 hpt, BLH1 and BLH8 were triggered, and BLH3 was similar to control; afterward, BLH3 and BLH8 were significantly repressed while the other members' statuses were the same as those of controls. In media with higher BA ( Figure 7J), BEL1, BLH2, and BLH5 were downregulated, while BLH1 and BLH8 were upregulated and BLH3 and BLH6 levels were the same as those of controls at 24 hpt. Subsequently, the BEL1, BLH3, and BLH5 mRNA levels were the same as those of controls, while those of BLH1 and BLH6 were higher and those of BLH2 and BLH8 were lower. Comparing the patterns at different BA dosages, six out of seven members (BLH6) shared similar patterns at 24 hpt, while discordant trends occurred for BLH2, BLH3, and BLH6 at 72 hpt, suggesting that a CK increase alters BELL transcription in a complex way.

KNOPE1 Overexpression in Gisela 6 Rootstock
Regeneration of leaf explants followed a two-step procedure (details in Section 4) consisting of dark-liquid/light-solid on media RM1 and RM2 (Table S6); the latter is recommended for Gisela 6. After 6 weeks, regeneration frequency (percentage of explants forming novel shoots) was higher in the RM1/RM1 than in the RM2/RM2 (12.0 vs. 4.0 %), while the average shoot number per regenerating explant was similar (4.3 ± 0.5 vs. 4.5 ± 0.7). Consequently, we opted for RM1/RM1 to proceed with agro-infection ( Figure S2); shoots originating from selection media (HYG or PTT) underwent a second selection, and three 35S:GFP and three 35S:KNOPE1 putative transgenic lines were rescued and named primary transgenic clones (PTCs). Southern blot analysis pointed at independent events of T-DNA multiple insertion in Gisela 6 genomes ( Figure S3). Transformation frequency was 2.6% and 0.6%, respectively, for 35S:GFP and 35S:KNOPE1 events (Table S7). We subcultured the PTC and recovered six 35S:KNOPE1 plants. All 35S:GFP clones maintained stable expression over time ( Figure 9A-I, Table S7). Differently ( Figure 9S), the molecular analyses showed that the clones harbored the 35S:KNOPE1 transgene in the genome (genomic DNA PCR) but its message was undetected (RT-PCR), suggesting malfunctions (silencing), which we did not further address (e.g., detection of small interfering RNAs). Two plants showed phenotypes with altered margins in leaves ( Figures 9K-P and S2O), a distinctive trait of plants overexpressing KNOPE1/BP. Finally, one of the two also showed phenotype reversion later ( Figure 9Q).

KNOPE1 Overexpression in Gisela 6 Rootstock
Regeneration of leaf explants followed a two-step procedure (details in Sectio consisting of dark-liquid/light-solid on media RM1 and RM2 (Table S6); the latter is ommended for Gisela 6. After 6 weeks, regeneration frequency (percentage of expl forming novel shoots) was higher in the RM1/RM1 than in the RM2/RM2 (12. Figure S3; arrowheads, primers used to check for transgene integrity (black) and expression (red). (S) Upper panel, a check for p35S:KNOPE1:NOSt cassette integrity by PCR with gDNA. Six clones were rescued and analyzed; on the left, size of bands of DNA ladder in base pairs (bp); on the right, the amplicon size is specified. Mid panel, a check for KNOPE1 transgene expression in the six clones by RT-PCR using leaf blade RNA. Peach (Chiripa) KNOPE1-specific primers fell between the 5'UTR and the first exon (R). The constitutive RPII expression was assayed to check for correct retrotranscription and usage of equal cDNA amounts. The amplicon sizes are reported.

PRUNOX
The complete KNOX catalog was retrieved from the genomes of 11 Prunus spp., and fine curation of gene structure was carried out for 35 out of 111 sequences. The phylogenetic tree of deduced proteins confirmed the separation into classes I, II, and M together with respective Arabidopsis orthologs [14]. Here, all 10 Prunus spp. had the KNATM branch, as already reported for Prunus spp. [14,21], which is a typical class of eudicots [2] and unfound in orchids [40]. A subgroup without Arabidopsis counterparts emerged in PRUNOXI, here named extra group (with peach KNOPE6 as reference), which is not unexpected considering the ample variation of KNOX number in dicot fruit trees. Specifically, the main diversity source of KNOPE6-like proteins resided in the N-and C-termini (not shown) as compared to the nearest members (e.g., KNAT2/6 and KNAT1). Moreover, KNOPE6-like gene structure differs from all the other PRUNOXIs in length and sequence (not position) of introns. Finally, all species maintained one KNOPE6-like gene (P. mume on Chr 1 and the other nine on Chr 6). These data suggest that KNOPE6-like may have specific roles in Prunus trees; one may regard fruit development as supported by KNOPE6 association with a QTL regulating several drupe characteristics [21]. As for protein variability, RNA-seq analyses indicate that coding sequences produce isoforms; for instance, both P. persica and P. mume harbor 11 KNOX genes encoding mRNAs for 17 and 12 proteins, implying splicing events. This was in agreement with a bioinformatic survey on P. mume TALEs [14]. For a given PRUNOX, amino acid variations (missense substitutions) among the Prunus spp. mainly occurred in N-and C-termini, but they were computed as similar and tolerated in most cases, hence supporting the conserved functions within the kind; however, substitutions potentially affecting functions were scored in KNOPE6 members.
The 10 Prunus spp. maintained the PRUNOX genomic organization in terms of the total number, class composition, and chromosome positions, with modest exceptions. This supports that the diversification of KNOX subfamilies took place [2] before the Prunus lineage diversification [41]. Referring to the Prunus subgenus, the shared PRUNOX colinearity brought out one group inclusive of Armeniaca section species (P. armeniaca and P. mume) and the other consisting of Persicae (P. persica and P. mira), Amigdalus (P. dulcis), and Prunus (P. salicina) sections. The separation was consistent with the genus evolutionary history and the high vicinity of P. mume and P. armeniaca compared to other Prunus species [42]. As for the Cerasus subgenus, P. avium hosted an additional copy of the class II KNOPE3-like gene, suggesting a specific role in the species of the deciduous corymbose group [42].
Several works report on KNOX responsiveness to CK in fruit tree organs [12,13] despite avoiding a computational search for CK-related motifs. Here, the investigation focused on peach and almond KNOX and showed that cis-elements of 5-8 bp occurred in all PRUNOX members abundantly and with conserved positions. Several other hormoneresponsive cis-elements were scored (ABA, AUX, GAs, and ethylene) consistently with other fruit tree works [12][13][14]. Finally, giving speculation on RNA-seq meta-analysis related to CK, the abundant PRUNOXI expression in organs characterized by meristem activity (buds and stems) is consistent with the class I roles in participating in maintaining cells at undifferentiated stages in a context of CK-AUX equilibria necessary for meristem development [43].

PRUNOXI and -II Transcription Response to BA in Micropropagation
BA is routinely applied in MP; at 1.7 µM, increased numbers of leaves and lateral shoots, due to axillary bud activation, were expected [29]. The tripled dosage only raised the side shoot number, but histological sections showed the occurrence of budding from subcortical layers of the stem, of bud morphological abnormalities, and of abundant callus at the surface adjacent to the culture medium. At the molecular level, catalase genes (CAT) were upregulated by BA increase, while CK catabolism genes were affected by both concentration and time. These markers highlight a stress status associated with CK degradation as a response to hormone uptake. Indeed, at 72 hpt of BA high dose, the collective induction of CAT, CK oxidases, and CK receptors supported the stress exacerbation. These data are consistent with the concept that BA concentration variation is a multitype stress factor [44] and point at these markers as useful for the rapid monitoring of the physiological and health status of GF677 rootstock propagules.
CK responsive elements in promoters and transcriptional variation at 24 hpt support a rapid BA effect on PRUNOX, consistently with PRUNOXI triggering 3 h post-BA vascular uptake in stems [14]. Several experiments based on BA administration reveal KNOX message variation in long periods post-treatment [16,45]. These data pinpoint that responsiveness depends on supply methods, targeted tissue, and plant species. It is thought that KNOX genes respond to endogenous CK variations following BA applications since the BA root uptake in Arabidopsis seedlings did not cause KNOXI triggering, which instead occurred after endogenous CK levels were increased [46], and Arabidopsis KNOXs do not appear to be genes quickly responsive to BA [47]. In support, work on BA-induced caulogenesis from pine leaf showed that BA affected endogenous CK variation in the long term during which key KNOXI genes were responsive [48]. Relatedly, BA in the media can alter the inner CKs balance in GF677 [49]. In our system, CKR patterns after 72 hpt of high BA dosage may reflect endogenous CK variation (and hormone ratios) leading to altered KNOX profiles. Moreover, two gene regulatory aspects emerged from propagule responses: (a) PRUNOXIs are more sensitive to BA dose changes than PRUNOXIIs (Table 3), in particular STMlike2, KNOPE1, and KNOPE2.1; (b) all PRUNOXIs (except for KNOPE6) modulate expression in response to BA over time, and only KNOPE7 does so among PRUNOXIIs. This would suggest a kind of time-coordinated regulation between classes I and II over time [8] in response to CKs.
The propagule is a multiorgan system and KNOXs are multifunctional; therefore, establishing KNOX-specific (and/or undesired) effects on developmental changes due to BA increase requires additional experiments, here out of scope. However, considering the class I KNOX roles in maintaining cells in an undefined state and/or preventing cell expansion or lignification, the altered regulation could contribute to (a) the miniaturized state, typical of in vitro organs, or (b) tissue disorders that subtend hyperhydric traits [44]. Finally, it is known that KNOX and CK generate feedback-loop mechanisms of reciprocal control associated with cellular disorders [17], and these may contribute to the wide number of anomalies described for MP [50].
The tripled BA concentration did not affect stem height but did affect leaf and shoot numbers. This condition led to changes in PRUNOXI and BELL expression levels (and their correlations) at 72 hpt. However, in some cases, KNOX/BELL coexpression was preserved. Specifically, the STMlike1/KNOPE7/PpBLH8 coexpressed module included genes with orthologs that are involved in caulis development [6,51,52]. Hence, the unvaried gene module behavior may account for the unaffected stem trait. At 72 hpt of 5.1 µM, the increased PRUNOXI levels may also reflect the presence of more axillary meristems per propagule or de novo bud formation and shooting. Relatedly, it is known that almond STM-like transcription preceded the organization of adventitious meristems [53].

Transformation of Gisela 6 and KNOPE1 Phenotypes
In peach, stable [54] and/or transient [55] gene transfer has improved but remains a laborious, low-efficiency, and cultivar-specific process [54]. Alternatively, to assay technology efficiency and study KNOPE1 function, we employed the Gisela 6 transformation as used in different labs [56][57][58]. Regeneration efficiency is crucial for transformation; here we found that RM1 medium (regeneration frequency 12%, shoots per explant ca. 4%), based on QL and previously shown to be acceptable for the Montmorency cultivar [57], was also satisfactory for Gisela 6, despite the same authors recommending a specific one based on WPM. We may speculate that material origin and status may subtend the RM1 effect (e.g., different subclones of Gisela 6, hormonal treatments on material supplied by the company, and different in vitro growth conditions regarding light intensity). Finally, indirect organogenesis via callus prevailed, confirming that organogenesis-competent cells lay in wounded leaf mid-ribs [56]. The transformation frequency of Gisela 6 was reported to range from 0.5 to 3% [56][57][58], and here it consistently varied from 0.6 to 2.6%, depending on the vector type and transgene cassette. Two rounds of growth on the selection medium, molecular analyses, and phenotype selection (GFP functionality or leaf margin alteration for KNOPE1) were necessary to avoid technical escapes. However, malfunction of 35S:KNOPE1 was detected in clonal lines, together with phenotype reversion. Silencing events associated with transgenesis have been known for a long time [59], and here they were not further investigated (e.g., hypermethylation of transgene promoters, siRNA production due interference caused by the transgene vs. endogenous genes). The 35S:KNOPE1 plants bore leaf margin alteration similarly to KNOPE1 overexpression in Arabidopsis and several other simple leafed species overexpressing BP-like genes [17]. However, considering that KNOX overexpression causes pleiotropic and dramatic effects, the fine modulation of KNOXs by traditional approaches, such as the guide of time/tissue-specific or inducible promoters, or by novel genome editing strategies is envisaged [60].
As for genomic analyses, the reference assemblies and gene annotations were derived from the GDR database (www.rosaceae.org (accessed on 27 December 2022)), except for P. mume data from the NCBI repository. The keyword search of functional annotations used to identify PRUNOX genes was based on the term "KNOTTED" and terms related to KNOX domains (KNOX1, KNOX2, ELK, and HD). A similarity search was achieved using P. persica sequences as queries in each genome and using BLASTn (e-value cut-off ≤ 1 × 10 −15 ). Gene models and annotations were manually curated (Table S1). The Gene Structure Display Server 2.0 (http://gsds.gao-lab.org (accessed on 27 December 2022)) was used to draw gene structures and protein domain organization. The MCScanX [61] program was used to perform the colinearity analyses of KNOX proteins within each Prunus spp. genome at the chromosome assembly level, and duplication types were classified into segmental, tandem, proximal, and others. Subsequently, colinear KNOXs were assessed for nonsynonymous and synonymous substitutions (Ka and Ks values) using PAL2NAL web service (http://www.bork.embl.de/pal2nal (accessed on 27 December 2022)). The formula to calculate the duplication time of colinear KNOX was T = Ks/2λ (with λ = 6.56 × 10 −9 for dicots) [62]. The KNOX colinearity among Prunus spp. was determined by comparing chromosome map positions and graphed accordingly. The 1500 bp genomic sequences upstream the start codons of PRUNOX were submitted to the PLACE database for plant cis-acting regulatory element analysis. The program did not host CK binding motifs but was useful to detect other hormone-related motifs, which are listed in Table S3. Additionally, the Regulatory Sequence Analysis Tool 2022 (https://rsat.sb-roscoff.fr (accessed on 27 December 2022)) was run by using the same genomic sequences as above and CK motifs retrieved in [38,39] and reported in Figure 4 and Table S4.

Morpho-Histological Analyses
Phenotyping was carried out at 10 days post-treatment by photographing 20 propagules per treatment and timing (5 p/jar) and measuring stem height, number of leaves on the main axis, and number of side shoots using ImageJ software (https://imagej.nih (accessed on 27 December 2022)). Student's t-test was used to assess significant differences between treated and untreated samples at a given concentration.
Propagules (n ≥ 8) for each treatment and timing were immersed in ethanol (70% v/v) and stored at 4 • C. The dehydrated samples were embedded in Technovit 7100 (Heraeus Kulzer, Hanau, Germany), longitudinally sectioned at 8 µm with a Microm HM 350 SV microtome (Microm, Neuss, Germany), dried overnight at 40 • C, stained with 1% toluidine blue (w/v), dried and permanently mounted in Canada Balsam, and observed under light microscopy. Sections were digitally photographed using a Leica DMRB optical microscope equipped with a Leica DC 500 camera.

Shoot Regeneration
Regeneration tests were as follows: Leaf explants from healthy in vitro clones underwent a 24 h dark pretreatment in liquid medium by gentle shaking and then were transferred to agarized regeneration media (RM) under light. The combinations darkliquid/light-solid were on RM1/RM1 and RM2/RM2 media; RM2 was recommended for Gisela 6 [57]. RM1 (QL 3.38 g/L+ BA 3 mg/L + NAA 0.5 mg/L + sucrose 4% v/w) and RM2 (WPM + BA 2 mg/L + IBA 1 mg/L + sucrose 3% v/w) were at pH 4.8. Explants derived from shoots (length > 1 cm) from clones propagated on QLBI for 3-4 weeks. Leaves (length = 1.5-2 cm) were devoid of petioles and cut with three incisions perpendicular to the main vein (base, middle, and end of lamina). To test regeneration frequency, 50 explants (ca. 5 per Petri dish) were used for each treatment. The RM1/RM1 combination was satisfactory and used for transformation experiments.

Statistical Analyses
Data were analyzed by different statistical tests including Student's t-test and twoway ANOVA. Pearson correlations were calculated using the "rcorr" function in the R environment v3.4.3 (Core Team, Vienna, Austria, https://www.R-project.org (accessed on 27 December 2022)).