This Article Abstract Full Text (PDF) Supplemental Data All Versions of this Article: 140/4/1255    most recent pp.105.076059v1 Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Related articles in Plant Physiol. Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Pischke, M. S. Articles by Sussman, M. R. Search for Related Content PubMed PubMed Citation Articles by Pischke, M. S. Articles by Sussman, M. R. Agricola Articles by Pischke, M. S. Articles by Sussman, M. R.

DEVELOPMENT AND HORMONE ACTION

A Transcriptome-Based Characterization of Habituation in Plant Tissue Culture^1,[W]

University of Wisconsin Biotechnology Center and Department of Biochemistry, Madison, Wisconsin 53706

	ABSTRACT

TOP ABSTRACT RESULTS DISCUSSION CONCLUSION MATERIALS AND METHODS LITERATURE CITED

 
For the last 50 years, scientists have recognized that varying ratios of the plant hormones cytokinin and auxin induce plant cells to form particular tissues: undifferentiated calli, shoot structures, root structures, or a whole plant. Proliferation of undifferentiated callus tissue, greening, and the formation of shoot structures are all cytokinin-dependent processes. Habituation refers to a naturally occurring phenomenon whereby callus cultures, upon continued passage, lose their requirement for cytokinin. Earlier studies of calli with a higher-than-normal cytokinin content indicate that overproduction of cytokinin by the culture tissues is a possible explanation for this acquired cytokinin independence. A transcriptome-based analysis of a well established habituated Arabidopsis (Arabidopsis thaliana) cell culture line was undertaken, to explore genome-wide expression changes underlying the phenomenon of habituation. Increased levels of expression of the cytokinin receptor CRE1, as well as altered levels of expression of several other genes involved in cytokinin signaling, indicated that naturally acquired deregulation of cytokinin-signaling components could play a previously unrecognized role in habituation. Up-regulation of several cytokinin oxidases, down-regulation of several known cytokinin-inducible genes, and a lack of regulation of the cytokinin synthases indicated that increases in hormone concentration may not be required for habituation. In addition, up-regulation of the homeodomain transcription factor FWA, transposon-related elements, and several DNA- and chromatin-modifying enzymes indicated that epigenetic changes contribute to the acquisition of cytokinin habituation.

Habituation is a mitotically transmissible character (Meins, 1989, and refs. therein). That is, new callus cultures derived from habituated tissues retain the cytokinin-independent state. To date, several loci (Habituated leaf [Hl]) conferring habituation on tobacco (Nicotiana tabacum) leaf tissues have been identified (Hl-1, Hl-2, and Hl-3), two of which were reported to be meiotically transmissible (Meins and Foster, 1986; Meins, 1989; Meins and Seldran, 1994; Meins and Thomas, 2003). Interestingly, this meiotic transmission of habituation is reversible (Meins and Thomas, 2003). One explanation for the reversibility of the heritable, habituated state is that the genetic alteration at the Hl-2 locus, for example, is due to DNA modification rather than mutation. Indeed, variations in DNA methylation levels have been documented during the tissue culture process, as well as among plants regenerated from tissue culture and their progeny (Planckaert and Walbot, 1989; Phillips et al., 1994; Kaeppler et al., 2000; Fojtova et al., 2003; Hao et al., 2004; Koukalova et al., 2005).

Presently, experimental evidence for the mechanism of habituation is scant. Whereas ectopic expression of cytokinin-signaling components has been shown to artificially confer a habituated state on plant tissues (e.g. CYTOKININ INDEPENDENT 1 [CKI1; Kakimoto, 1996], ARABIDOPSIS RESPONSE REGULATOR 1 [ARR1; Sakai et al., 2001], ARR2 [Hwang and Sheen, 2001], and ARR4 [Osakabe et al., 2002]), overproduction of cytokinin by the plant tissues has been a predominant explanation for acquired cytokinin independence (Murashige and Skoog, 1962; Meins, 1989; Nogué et al., 2000; Catterou et al., 2002; Sun et al., 2003). One approach toward identifying the mechanism of habituation is a comparison of gene expression in habituated and nonhabituated cell cultures. If overproduction of cytokinin is the mechanism for habituation, then this could be reflected by increased expression of genes encoding cytokinin synthases (IPTs), decreased expression of genes encoding cytokinin-degrading enzymes (CKXs), and increased expression of known cytokinin-inducible genes (KNAT1, CYCD3, CAB1, NIA1, Type-A ARRs) in habituated calli. Likewise, if epigenetic modifications play a role in habituation, then this may be reflected by alterations in the expression levels of genes encoding enzymes involved in DNA methylation and/or histone modification.

This study explored the mechanism of habituation in the well established T87 Arabidopsis (Arabidopsis thaliana) cell culture line (Axelos et al., 1992). A transcriptome-based analysis of habituated and nonhabituated plant cell cultures revealed differential expression of more than 800 genes, which included a 19-fold up-regulation of the transcript encoding the cytokinin receptor CRE1. A concomitant overexpression of the CRE1 protein was verified using an isotope-assisted mass spectrometric method called AQUA (an acronym for absolute quantification of proteins) developed by Gerber et al. (2003). Phenotypic and transcriptomic differences between habituated and nonhabituated cell cultures, as well as potential mechanisms for the phenomenon of habituation, are discussed.

	RESULTS

TOP ABSTRACT RESULTS DISCUSSION CONCLUSION MATERIALS AND METHODS LITERATURE CITED

 

Phenotypic Differences between Habituated and Nonhabituated Callus Tissue

On solid media, callus tissue that has been freshly induced from Arabidopsis root segments (also called FC for freshly derived callus, or nonhabituated callus tissue) has an obvious requirement for exogenous cytokinin for maximal growth (Figs. 1 and 2 ). Habituated callus tissue derived over several passages onto solid media from the Arabidopsis T87 cell line (also T87), on the other hand, does not require cytokinin for maximal growth. In fact, proliferation of habituated callus tissues is inhibited by exogenously applied cytokinin (Figs. 1 and 2). T87 cell cultures are rapidly dividing with respect to freshly derived callus cultures and have a visibly different morphology. For example, cells within the T87 callus clumps are more easily teased apart and have a shiny appearance.

View larger version (118K): [in this window] [in a new window]   Figure 1. The effect of cytokinin on the growth and morphology of habituated and nonhabituated callus cultures. A, Representative habituated callus culture maintained in the presence of cytokinin (benzyl adenine) for 3 weeks. B, Representative habituated callus culture maintained in the absence of cytokinin for 3 weeks. C, Representative nonhabituated callus culture maintained in the presence of cytokinin for 6 weeks. D, Representative nonhabituated callus culture maintained in the absence of cytokinin for 6 weeks.

View larger version (28K): [in this window] [in a new window]   Figure 2. The effect of cytokinin on the proliferative capacities of habituated and nonhabituated callus cultures. Calli were cultured in the absence (black) or presence (magenta) of benzyl adenine. A, The fresh weight of callus cultures derived from nonhabituated (horizontal stripes) and habituated (diagonal stripes) tissues, after 6 and 3 weeks in culture, respectively. B, The fold-change in surface area of callus cultures derived from nonhabituated (horizontal stripes) and habituated (diagonal stripes) tissues, after 6 and 3 weeks in culture, respectively. At least 10 calli were measured for each data point. Error bars represent the SD.

Full-genome transcriptome analyses of habituated callus cultures grown in the presence or absence of cytokinin (T87 + BA, T87 – BA), as well as freshly derived (nonhabituated) callus cultures grown in the presence or absence of cytokinin (FC + BA, FC – BA), were carried out on the Arabidopsis 60mer microarray (NimbleGen Systems). Robust multiarray averaging and log₂ transformation were applied to the data before analysis. A very strong positive correlation was seen for all possible pairwise comparisons of signal intensities across approximately 28,000 genes, within each set of four biological replicates, hereafter referred to as "groups" (T87 + BA, T87 – BA, FC + BA, and FC – BA; Table I ), demonstrating the reproducibility of the technology.

For subsequent characterization of differential gene expression in this study, we chose to focus on the genes identified as significantly differentially expressed by the SAM method (Tusher et al., 2001). A comparison between the groups FC + BA and FC – BA identified 467 up-regulated genes (Supplemental Table I) and 23 down-regulated genes (Supplemental Table II). None of the 23 genes identified as significantly down-regulated by cytokinin in this study has been identified as a cytokinin down-regulated gene in the previous studies we queried (Crowell et al., 1990; D'Agostino et al., 2000; Hoth et al., 2003; Rashotte et al., 2003; Yamada et al., 2004; Kiba et al., 2005; Rashotte et al., 2005). On the other hand, 24 of the genes identified as cytokinin up-regulated in this study have been identified as cytokinin up-regulated genes in other studies as well (Supplemental Table I). Previous studies have identified cytokinin-regulated genes after hormone application to seedlings or tissue at intervals ranging from 10 min to 24 h. The overlap of 24 genes between this study and several previous studies represents genes whose expression remains up-regulated in response to cytokinin over an extended time period, i.e. on the order of months rather than minutes or hours. Not surprisingly, the Type-A response regulators (RRs), which are rapidly and transiently induced by cytokinins (Brandstatter and Kieber, 1998; Imamura et al., 1998; D'Agostino et al., 2000), were not identified as differentially expressed in nonhabituated calli. The previously identified cytokinin-regulated genes also identified in this study include several enzymes involved in cell wall modification (pectinacetylesterase, expansin, glycosyl hydrolases, and a pectin methylesterase inhibitor), nutrient and carbon acquisition (sulfate transporter, nitrate reductase, Suc transporter, Rubisco subunit), and light harvesting (chlorophyll a/b-binding proteins; Supplemental Table I).

Although only 5% (24/467) of the genes found to be up-regulated in nonhabituated calli in response to cytokinin were previously identified as cytokinin up-regulated, there is a much greater overlap (39%, 182/467) in the types of gene families identified among the cytokinin-regulated transcript studies (Supplemental Table I). Among the genes found up-regulated by cytokinin in nonhabituated calli were several genes involved in light harvesting and photosynthesis, cell wall modification, amino acid and protein synthesis and transport, and nutrient and carbon acquisition. These findings correlate well with several known biological roles for cytokinins in planta, including promotion of greening (Stetler and Laetsch, 1965; Daniell and Rebeiz, 1982), cell division (Miller et al., 1956), and protein synthesis (Maaß and Klämbt, 1977), and serving as a nutrient sink (Roitsch and Ehneß, 2000). There were also a number of genes identified that potentially highlight cross talk between cytokinin-signaling and other signaling pathways, including auxin transport and signaling (At4g29080, At2g01420, At1g19840, At2g46690, At2g39730, At1g22630, At2g33830, At1g67700), calcium signaling (At4g33000, At1g25230, At5g54130), brassinosteroid synthesis (At4g13290, At1g78490, At4g13770, At2g34490, At3g26330, At3g03470, At2g22330), GA₃ synthesis (At1g06640, At5g59530), and ethylene responses (At5g07580, At5g61590).

A Comparison between the Groups T87 + BA and T87 – BA

A comparison between the groups T87 + BA and T87 – BA identified zero up-regulated genes and 36 down-regulated genes (Supplemental Table III). There was no overlap between genes down-regulated in the FC ± BA comparison and the T87 ± BA comparison, indicating that T87 cells respond to the presence of cytokinin differently than nonhabituated cells. Previous studies have shown that T87 cells can respond to transient cytokinin application by up-regulation of cytokinin primary response genes, similar to wild-type Arabidopsis seedlings (Yamada et al., 2004). This study, however, shows that at the level of the transcriptome, T87 calli and nonhabituated calli respond very differently to the presence or absence of exogenous cytokinin in the plant culture media. In fact, 19% (7/36) of the genes significantly down-regulated in habituated calli, in response to cytokinin, were found to be up-regulated in nonhabituated calli, in response to cytokinin (Supplemental Table III). Furthermore, an additional 53% (19/36) of these genes belong to gene families, some members of which were also identified as cytokinin up-regulated in nonhabituated calli or in previous studies (Supplemental Table III). Thus, while T87 cells may transiently respond to cytokinin in a physiologically relevant fashion, in the long term habituated T87 cells seem to respond to cytokinin in a manner quite opposite to nonhabituated cells.

A Comparison between the Groups T87 – BA and FC + BA

A comparison between the groups T87 – BA and FC + BA identified 440 up-regulated genes (Supplemental Table IV) and 405 down-regulated genes (Supplemental Table V). Thirty-two genes previously identified as cytokinin up-regulated and 31 genes previously identified as cytokinin down-regulated were also identified as up- and down-regulated, respectively, in habituated cells maintained in the absence of cytokinin (Supplemental Tables IV and V). The up- and down-regulated genes identified by SAM for the comparisons FC + BA versus FC – BA and T87 – BA versus FC + BA were categorized by biological process, using the gene ontology tool available on The Arabidopsis Information Resource (TAIR) Web site (http://www.arabidopsis.org/tools/bulk/go/index.jsp; Table IV ). Several of the biological process categories, into which the 490 genes with altered expression in nonhabituated calli fall, reveal a bias toward either up- or down-regulation of gene expression. For example, genes whose products are involved in cell organization and biogenesis, electron transport or energy pathways, and transport are more likely to be up-regulated than down-regulated. Perhaps this tendency toward up-regulation of genes important for energy and nutrient acquisition, and cell biogenesis, reflects the actively growing and dividing nature of nonhabituated callus tissue in the presence of cytokinin. Differentially expressed genes involved in nucleic acid and protein metabolism tend to be up-regulated in healthy habituated calli with respect to healthy nonhabituated calli, while genes involved in responding to environmental stimuli tend to be down-regulated. These tendencies may reflect that the T87 cell line has been selected for rapid proliferation, independent of cell division-promoting substances. Interpreted in this light, genes important for generating the nucleic acids and proteins necessary for rapid growth would tend to be up-regulated, while genes involved in sensing external growth and division cues would tend to be down-regulated.

Genes Involved in Cytokinin Responses

Cytokinin signaling in Arabidopsis occurs through a multistep His-to-Asp (His-Asp) phosphorelay (for review, see Hutchison and Kieber, 2002; Kakimoto, 2003). The key components of this signal transduction system are the receptor HKs, His phosphotransfer proteins (HPts), and RRs. Because the SAM method for identifying differentially expressed genes was quite conservative, we chose to take a closer look at the expression levels of several genes encoding His-Asp-signaling components. To summarize the 2-fold gene expression differences in His-Asp-signaling components between habituated and nonhabituated callus cultures (Table V ), the expression of the cytokinin receptor CRE1 was highly up-regulated, while that of AHK3 was moderately down-regulated. Expression of the HPt (positive regulator of cytokinin signaling) AHP1 was highly down-regulated, while expression of AHP5 was moderately down-regulated. Two Type-A RRs, ARR7 and ARR15 (negative regulators of cytokinin signaling), were moderately up-regulated. Thus, while the cytokinin receptor CRE1 was highly up-regulated in the habituated T87 cell line, two HPt proteins that may serve to propagate the cytokinin signal initiated by CRE1 in planta were down-regulated. Furthermore, two RR proteins that may serve to repress the CRE1-initiated His-Asp phosphorelay in planta were up-regulated. These changes may reflect a down-regulation of specific cytokinin-mediated responses in habituated cells. For example, some cytokinin responses were present in the habituated calli (proliferation and greening), whereas other cytokinin responses were lost (shoot induction).

Other Phytohormone-Related Changes

Several genes involved in hormone biosynthesis were identified by SAM as significantly down-regulated in habituated calli (ethylene synthesis, At2g19590 and At1g12010; GA₃ synthesis, At1g06640, At2g25450, At1g14120, At1g14130, and At2g30840; brassinosteroid synthesis, At2g30490, At1g78490, At2g34490, At1g13080, At2g27000, At2g22330, and At1g12740). The expression of additional genes involved in responses to the plant hormone ethylene was explored further in habituated calli. As shown in Table V, expression of the ethylene receptors ETR2 and ERS1 was down-regulated in habituated calli (–4.0-fold and –2.2-fold, respectively). Likewise, expression of several additional ethylene-signaling components was moderately down-regulated in habituated calli (CTR1, –1.8-fold; EIN3, –1.8-fold; EIL1, –1.8-fold; ERF1, –1.9-fold; data not shown). Overlaps in the cytokinin- and ethylene-response pathways have been demonstrated. For example, the inhibitory effect of cytokinin on etiolated hypocotyl elongation, as well as root elongation, has been linked to cytokinin-induced ethylene production (Cary et al., 1995). In addition, the Type-B RR ARR2 has been identified as a transcription factor (Sakai et al., 2000; Lohrmann et al., 2001) that acts as a positive regulator of both cytokinin signaling (Hwang and Sheen, 2001) and ethylene signaling (Hass et al., 2004).

The first committed step in ethylene biosynthesis is catalyzed by 1-aminocyclopropane-1-carboxylic acid synthase (Yang and Hoffman, 1984). While expression of some 1-aminocyclopropane-1-carboxylic acid synthase homologs was up-regulated in habituated calli (ACS2, 9.6-fold; ACS6, 2.1-fold; ACS7, 2.7-fold; ACS8, 2.1-fold), expression of others was down-regulated (At1g05010, –2.5-fold; At1g12010, –9.1-fold; At2g19590, –16.7-fold; data not shown). Cytokinin-induced ethylene production, however, occurs primarily through ACS5 (Vogel et al., 1998), which is unaltered in habituated calli (data not shown). Thus, while our data do not support a role for cytokinin-mediated ethylene production, the down-regulation of several ethylene receptors and signaling components may reflect a negative interaction between ethylene- and cytokinin-signaling pathways in habituated callus cultures.

While a few genes involved in auxin signaling (auxin-responsive proteins At3g15540, At1g15580, and At3g62100) were identified by SAM as up-regulated in habituated calli, others were identified as down-regulated (auxin-induced proteins At1g19840 and At1g72430, auxin-regulated protein At2g33830, and auxin receptor TIR1 homolog At1g12820). The interaction of cytokinin- and auxin-response pathways has been well documented in the organogenesis of plant cell cultures (Skoog and Miller, 1957; Steward, 1970), as well as in the regulation of apical bud dominance (Wickson and Thimann, 1958). Several auxin-related signaling genes, including TIR1, ABP1, Aux/IAAs, SAURs, ARFs, and GH3s, were analyzed for 2-fold changes in gene expression in habituated versus nonhabituated callus cultures.

While the expression of the putative auxin receptor ABP1 was not altered in habituated calli, the expression of the auxin receptor TIR1, as well as several TIR1 homologs, was down-regulated in habituated calli (TIR1, –2.1-fold; At1g12820, –3.8-fold; At3g26810, –2.0-fold; At3g62980, –2.1-fold; At4g03190, –2.6-fold). Out of the 29 Aux/IAAs analyzed, three were up-regulated in habituated calli (IAA5, 7.1-fold; IAA19, 7.9-fold; IAA30, 5.5-fold), and six were down-regulated (IAA2, –5.9-fold; IAA9, –1.9-fold; IAA13, –2.1-fold; IAA16, –3.3-fold; IAA27, –2.4-fold; IAA28, –2.1-fold). Out of the 70 SAURs analyzed, three were up-regulated (At4g34750, 2.5-fold; At4g34760, 1.9-fold; At5g53590, 3.3-fold) and five were down-regulated (At1g19840, –16.7-fold; At2g45210, –4.0-fold; At2g46690, –2.5-fold; At4g36110, –2.0-fold; At4g38840, –2.9-fold) in habituated calli. Out of the 23 ARFs analyzed, two were down-regulated in habituated calli (ARF9, –2.2-fold; ARF2, –3.2-fold). Out of the 20 putative GH3s analyzed, one was up-regulated (At5g13320, 4.2-fold) and two were down-regulated (At1g28130, –1.9-fold; At4g37390, –2.1-fold) in habituated calli. Hence, no general trends in the alteration of auxin-signaling-related components were seen in habituated calli compared to nonhabituated calli, highlighting the complexity between cytokinin-signaling and auxin-signaling interactions in planta.

Cell Division-Related Changes

Several cell cycle-related proteins (At3g53230, At2g38620, At1g44110, At2g07690, At3g44620), nucleosome components (Histone H2A, At3g20670 and At1g51060; Histone H2B, At3g53650 and At3g09480; Histone H3.2, At1g75600; Histone H4, At3g45930 and At3g46320), and protein and nucleic acid synthesis genes (At2g39590, At3g28500, At3g16780, At1g44900, At3g23890, At4g21710, At2g24050, At3g54490, At4g29090, At3g49000, At2g18720, At2g39820) were identified as up-regulated in habituated calli by the SAM method. Up-regulation of these genes indicates that processes normally up-regulated by cytokinins in nonhabituated tissues, such as DNA replication, protein synthesis, and cell division, are constitutively up-regulated in the habituated T87 cell line.

We chose to take a closer look at the expression levels of several cyclins, cyclin-dependent kinases, and histones in habituated calli with reference to nonhabituated callus (Tables VI and VII ). Twelve of 31 cyclins or cyclin homologs (identified using nucleotide BLAST searches against the Arabidopsis genome with identified cyclins) were up-regulated by 2-fold or more in habituated calli, while three were down-regulated by 2-fold or more. Six of 14 cyclin-dependent kinases (CDKs) or CDK homologs (identified using nucleotide BLAST searches against the Arabidopsis genome with identified CDKs) were identified as up-regulated. In terms of nucleosome components (as defined by the Plant Chromatin Database; http://www.chromdb.org/), 7/13 Histone H2A, 8/11 Histone H2B, 9/14 Histone H3, and 6/8 Histone H4 genes were up-regulated in habituated callus cultures. On the other hand, one of three linker Histone H1 genes was down-regulated. While differential expression for many of these histone and cell cycle-related genes was also seen in nonhabituated callus cultures (Tables VI and VII), the transcriptome analysis of these two different tissue types revealed both overlapping and distinct expression changes among gene family members. This result indicates that the accelerated proliferation of habituated T87 cell cultures, with reference to nonhabituated cell cultures, does not simply result from expressing genes normally involved in callus proliferation to a higher degree.

Other Interesting Changes

While SAM identified only one transcription factor as up-regulated in habituated calli (WRKY family, At2g03340), 15 were identified as down-regulated (AP2 domain family, At1g78080, At1g22190, and At1g13260; Bhlh related, At2g22770; bZIP family, At1g13600, At2g18160, At1g75390, and At1g77920; heat shock family, At1g46264; myb family, At1g74840, At1g14350, and At1g19000; scarecrow-like, At1g21450 and At1g07530; TCP family, At1g35560; WRKY family, At1g29280). While members of each of these gene families have been identified as cytokinin regulated in other studies, few genes themselves have been identified in these studies. The WRKY-family transcription factor found up-regulated in habituated calli was also found up-regulated by cytokinin in a previous study (Hoth et al., 2003). In contrast, the heat shock transcription factor (Hoth et al., 2003) and Bhlh-related transcription factor (Kiba et al., 2005) found down-regulated in habituated calli were previously identified as cytokinin up-regulated. The fact that we see differential expression of a variety of transcription factors between habituated and nonhabituated callus cultures is not surprising, given the large number of differentially expressed genes between the two tissue types.

Several genes involved in protein degradation, particularly members of the F-box protein family, were identified as either up-regulated (proteasome subunit, At3g22630; F-box protein family, At3g47030, At3g23970, At3g61340, At3g50710, At1g23770, At2g16450, At1g23780, At1g66290, At3g60710, At1g66310, At3g23950, At1g48400, At3g59190, At3g16590, At4g05470, At4g22390, and At3g47130; Kelch repeat-containing F-box protein family, At1g60570 and At4g04670) or down-regulated (F-box protein family, At1g22220, At1g67340, At1g21410, and At2g36090; Kelch repeat-containing F-box protein family, At1g15670, At1g23390, At1g26930, At1g30090, and At1g67480; ubiquitin-conjugating enzyme, At1g63800) in this study. Few reports regarding the involvement of protein degradation in cytokinin signaling have been made to date, and the results are conflicting (Smalle et al., 2002; Yamada et al., 2004). While the presence of both up- and down-regulation of several genes involved in protein degradation in cytokinin-habituated calli does not simplify the question of whether proteolysis plays an important role in regulating cytokinin signaling, these results do highlight specific genes to investigate further for cytokinin-related phenotypes.

Several calcium-binding proteins were identified as either up-regulated (At2g03450, At3g47480, At4g05520, At4g04695, At4g04720, At3g22910, At3g25600, At4g21820) or down-regulated (At2g18750, At1g25230, At1g76650), indicating a potential role for calcium signaling in cytokinin responses, as has been long expected though direct evidence has been lacking (for review, see Brault and Maldiney, 1999; Faure and Howell, 1999).

Evidence for Epigenetic Modifications in Habituated Callus Cultures

Notably, the expression of several transposon-related elements was up-regulated in habituated T87 callus cultures (At3g43563, At3g43862, At3g42253, At4g08680, At1g78095, At2g30640, At3g42806, At1g49090, At2g14230; Supplemental Table IV), as identified by SAM. A closer look revealed up-regulation of 2-fold or more for 37/485 transposon-related elements represented on the microarray (Table VIII ). Activation of transposons during the process of plant cell culture has been documented previously (for review see Kaeppler et al., 2000) and has also been shown to correlate with changes in DNA methylation patterns (for review see Bender, 2004).

Arabidopsis utilizes three classes of DNA methyltransferases that transfer a methyl group from S-adenosylmethionine to the C5 position of cytosine residues: methyltransferases (METs), chromomethylases (CMTs), and domains rearranged methylases (DRMs). The METs maintain CpG methylation (Cao and Jacobsen, 2002b; Tariq et al., 2003), CMTs maintain non-CpG methylation (Cao and Jacobsen, 2002b), and DRMs initiate de novo cytosine methylation at CpG, CpNpG, and asymmetric sites (Cao and Jacobsen, 2002a). DRMs also play a role in maintenance of methylation at CpNpG and asymmetric sites (Cao and Jacobsen, 2002b). The Arabidopsis genome encodes four putative METs. Of these, only MET1 was altered in habituated calli (3.5-fold up-regulation; Table IX ). The Arabidopsis genome also encodes four putative CMTs. Of these, both CMT1 and CMT3 were altered in habituated calli (5.5-fold and 3.2-fold induction, respectively; Table IX). Three putative DRMs are encoded by the Arabidopsis genome. Of these, expression of DRM1 was altered in habituated calli (2.1-fold up-regulation; Table IX). The Arabidopsis genome encodes four putative DNA glycosylases, which act to demethylate cytosine residues (for review, see Chan et al., 2005). One of these genes, DNG1, was down-regulated 2.0-fold in habituated calli.

Surprisingly, expression of the FWA gene (At4g25530) was up-regulated approximately 87-fold in habituated calli (Supplemental Table IV). FWA is a homeodomain-containing transcription factor that is important for the transition to flowering, as well as for floral meristem identity (Soppe et al., 2000). FWA expression is normally confined to the central cell of the female gametophyte and the endosperm (Kinoshita et al., 2003). Hypomethylation of the 5' region of FWA leads to ectopic expression and causes a delay in flowering (Soppe et al., 2000). This is an interesting case in which the methylated (silenced) state of FWA is the default state in all tissues, while endosperm-specific expression of the gene requires DNA demethylation. The regulation of FWA expression is accomplished, at least in part, by the DEMETER DNA glycosylase (Kinoshita et al., 2003). The DEMETER transcript was not altered in T87 calli with respect to nonhabituated calli (Table IX). Other genes for which differential expression has been detected based on promoter or gene methylation state (SUPERMAN, AT3G23130; PAI1, AT1G07780; PAI2, AT5G05590; PAI3, AT1G29410) were not differentially expressed in habituated calli (data not shown). This result indicates that the alterations in gene expression seen between habituated and nonhabituated callus cultures do not result simply from global hypomethylation of DNA.

Verification of Microarray Results by RT-PCR

Several genes whose expression was altered to varying degrees in habituated calli were chosen for verification of the microarray results. The results of reverse transcription (RT)-PCRs performed on serial dilutions prepared from habituated and freshly derived callus tissues were in agreement with the alterations in gene expression detected by the microarray analysis. For these experiments, cDNA aliquots were taken from the same samples used for hybridization to the microarray. As can be seen in Supplemental Figure 1, this agreement was seen for the direction of change, and was also generally seen for the magnitude of change, in gene expression. For example, the microarray analysis revealed that the expression of CRE1 was up-regulated by 19.6-fold in habituated calli (Table V). By RT-PCR analysis, the CRE1 transcript could barely be amplified from a 100-fold dilution of cDNA prepared from freshly derived calli, while this transcript could still be amplified from a 10,000-fold dilution of cDNA prepared from habituated calli (Supplemental Fig. 1). In contrast, microarray analysis revealed a reduction in AHP1 expression by 12.5-fold (Table V). By RT-PCR analysis, the AHP1 transcript could be amplified from a 10,000-fold dilution of cDNA prepared from freshly derived calli, while this transcript could be amplified to approximately the same degree from a 100-fold dilution of cDNA prepared from habituated calli (Supplemental Fig. 1). In addition, the microarray analysis revealed an up-regulation of FWA expression by about 86-fold, of AtHK1 expression by 3.2-fold, and of ARR5 expression by 1.8-fold (Table V; Supplemental Table IV). By RT-PCR analysis, the FWA transcript was undetectable even in undiluted cDNA prepared from freshly derived calli, while this transcript could be amplified from a 10,000-fold dilution of cDNA prepared from habituated calli. The AtHK1 transcript, on the other hand, could be amplified from a 100-fold dilution of cDNA prepared from freshly derived calli and from a 1,000-fold dilution of cDNA prepared from habituated calli. The ARR5 transcript could be amplified from a 10,000-fold dilution of cDNA prepared from both freshly derived and habituated calli, but the intensity of the amplified transcript was slightly less in freshly derived calli (Supplemental Fig. 1).

Several of these genes were also chosen for quantitative RT-PCR analysis (qPCR) on cDNA prepared both from RNA isolated from the same samples used for hybridization to the microarray and from RNA isolated from separate habituated and nonhabituated callus tissues handled the same way as those used for microarray hybridization (Table X ). Fold-changes in transcript abundance between habituated and nonhabituated callus cultures were calculated based on average count numbers normalized to the abundance of an Actin2 control transcript (for variations in control gene expression, see Table XI ). In all cases, the direction of differential expression (i.e. up- or down-regulation) calculated for each transcript based on microarray analysis or qPCR was the same, and in most cases (CRE1, AHK2, AHK3, AtHK1, AHP1) the fold-changes calculated by both methods were numerically very close to one another (Table X). While there were a small number of discrepancies between the two methods (namely, for the TMK3 and FWA transcripts), whether one favors the qPCR or the microarray data does not change our conclusions. Thus, based on these two independent PCR-based validation methods, which were performed with different gene-specific primer pairs, we felt confident that the fold-changes calculated by microarray analysis accurately reflected genome-wide transcriptome-based changes between habituated and nonhabituated callus cultures.

The completion of the Arabidopsis genome sequence, coupled with the development of DNA microarray technologies, has made it possible to analyze genome-wide mRNA expression patterns (i.e. the transcriptome) within whole plants (Rashotte et al., 2003; Bergmann et al., 2004), specific plant tissues (Che et al., 2002; Himanen et al., 2004), and even particular cells within a tissue type (Birnbaum et al., 2003). It is the changes in protein expression rather than the changes in mRNA expression, however, that truly reflect how a plant's growth and development are regulated in response to various internal and external cues. Recent studies comparing transcriptome and proteome profiles in yeast (Saccharomyces cerevisiae; Gygi et al., 1999) and human synovial tissue (Lorenz et al., 2003) have indicated that in many instances transcript levels are not good predictors of the corresponding protein levels. For this reason, we sought to quantify the level of CRE1 protein expression in habituated callus tissue.

A method has been developed enabling the absolute quantification (AQUA) of a protein from a complex mixture, by directly comparing the relative levels of a native tryptic peptide from the protein of interest with that of a known quantity of synthetic, isotopically labeled peptide standard (Gerber et al., 2003). In this case, the protein of interest was CRE1, the complex mixture was a total membrane protein fraction isolated from habituated or nonhabituated callus tissues, and the synthetic peptide was SS[L^*]PENPTVEER, where L^* refers to an isotopically labeled Leu residue ([¹³C₆,¹⁵N]Leu). This synthetic peptide corresponds to a unique tryptic peptide within the putative extracellular, ligand-binding domain of CRE1.

A schematic of the protocol used for CRE1 quantitation by the AQUA method is outlined in Figure 3 . Total membrane protein fractions were independently isolated from habituated and nonhabituated callus tissues, trypsinized, spiked with the isotopically labeled CRE1 internal standard peptide, and introduced into a triple-quadrupole mass spectrometer by electrospray ionization after online reversed-phase HPLC separation. The y₉ (2+) fragment ion from the parent peptide (SSLPENPTVEER) was chosen for selected reaction monitoring, on the basis of its strong signal. Representative extracted ion chromatograms of the CRE1-selected fragment ion, from habituated (blue) and nonhabituated (magenta) callus cultures, are presented in Figure 4 . Based on the ratio of the area under the curve for the native and synthetic CRE1 fragment ions, as well as the known initial quantity of synthetic CRE1 peptide added to the callus protein mixture, the quantity of CRE1 in 45 µg total membrane protein was determined. For T87 calli, the quantity of CRE1 protein was 0.477 ± 0.041 pmol (n = 5). For freshly derived calli, the quantity of CRE1 protein was 0.0247 ± 0.0026 pmol (n = 5). These differences correspond to an approximately 19-fold increase in CRE1 protein levels of habituated calli with respect to nonhabituated calli (0.477 pmol/0.0247 pmol). Thus, the approximately 19-fold induction of CRE1 mRNA expression in habituated callus tissue corresponds to an approximately 19-fold induction of CRE1 protein expression.

View larger version (25K): [in this window] [in a new window]   Figure 3. Schematic of the CRE1 protein quantification protocol.

View larger version (7K): [in this window] [in a new window]   Figure 4. Selected reaction monitoring reveals a 19-fold increase in CRE1 expression in habituated (T87) calli relative to nonhabituated (Fresh callus) calli. Representative extracted ion chromatograms of the selected fragment ion, from the CRE1-derived peptide, are shown. Blue, Habituated callus culture; magenta, nonhabituated callus culture.

To explore the hypothesis that a mutation is responsible for overexpression of CRE1 in habituated tissues, the promoter and coding regions for three genes, CRE1, TMK3, and AHP1, were selected for sequencing from genomic DNA prepared from habituated callus cultures. TMK3 was chosen for analysis because it is the gene immediately downstream of CRE1 and is also up-regulated in habituated calli (6.2-fold). AHP1 was chosen for analysis because it is a representative down-regulated gene in habituated calli (Table V). Comparison of the AHP1 and TMK3 promoter and coding region sequences from T87 cells to that generated by the Arabidopsis Genome Initiative (2000) revealed no base changes specific to habituated calli. Sequence analysis of the CRE1 coding region revealed one putative base change specific to habituated T87 callus cultures: T181C within exon 5. This nucleotide change corresponds to an amino acid change, F453L, within the HK domain of the CRE1 protein sequence (Fig. 5 ).

View larger version (45K): [in this window] [in a new window]   Figure 5. Location of the CRE1 mutation in habituated calli. Schematic of the CRE1 protein domains was generated from the amino acid sequence at http://smart.embl-heidelberg.de/. Blue rectangles, Transmembrane domains. Green square/green box, HK domain. Green triangle/black underscoring, Catalytic domain. Blue pentagon/blue box, RR domain. The conserved His and Asp residues within the HK and RR domains, respectively, are depicted by a larger font size. The location of the amino acid change in habituated calli (F453L) is shown in red.

We generated several independent lines of habituated callus cultures by continued passage of a freshly derived, nonhabituated callus culture. A comparison of the cytokinin responses of root-derived callus cultures (ecotype Columbia [Col]) after four successive passages onto fresh media containing both auxin and cytokinin, followed by seven successive passages of the same callus cultures onto fresh media either containing or lacking cytokinin, is shown in Figure 6 . The Hab1 line is a good candidate for a habituated cell line because it proliferated and turned green in the absence of cytokinin but was inhibited by the presence of cytokinin, similar to the T87 cell line. The Hab3 line appeared to proliferate more rapidly in the presence of cytokinin, but still grew and had a greening response in the absence of cytokinin. The Hab4 line seemed to proliferate and turn green well in the absence of cytokinin, but was not particularly inhibited by the presence of cytokinin. Thus, while we see quite a bit of phenotypic variation, clearly we have been able to isolate callus lines that can be maintained in the absence of cytokinin and are thus cytokinin habituated. As Hab1 appeared to be the most promising cytokinin-habituated callus culture line, we used qPCR to analyze the expression levels of select genes, including CRE1, relative to nonhabituated callus cultures (Table X). While we did not see up-regulation of the cytokinin receptor CRE1 in the Hab1 line, we did see up-regulation of the cytokinin-signaling component AHP1 (8.4-fold), lending support to the idea that habituation occurs via aberrant expression of cytokinin-signaling components. Similar to the T87 cell line (Supplemental Table IV), we also saw up-regulation of FWA (42.4-fold) in the Hab1 line. Because it is well documented that up-regulation of FWA expression occurs via hypomethylation within the promoter and 5' region of the gene (Soppe et al., 2000; Cao and Jacobsen, 2002a; Kinoshita et al., 2003), this result lends support to the hypothesis that epigenetic changes have occurred in the Hab1 habituated callus line.

View larger version (35K): [in this window] [in a new window]   Figure 6. Isolation of habituated callus lines (Hab1–Hab4) from wild-type Col root segments after four successive passages onto media containing both auxin (3,000 ng/mL 1-NAA) and cytokinin (300 ng/mL BA), followed by seven successive passages of the same callus cultures onto media either containing or lacking cytokinin. A, Hab1 maintained in the absence of BA. B, Hab1 maintained in the presence of BA. C, Hab3 maintained in the absence of BA. D, Hab3 maintained in the presence of BA. E, Hab4 maintained in the absence of BA. F, Hab4 maintained in the presence of BA.

	DISCUSSION

TOP ABSTRACT RESULTS DISCUSSION CONCLUSION MATERIALS AND METHODS LITERATURE CITED

A growing body of evidence indicates that transduction of the cytokinin signal in Arabidopsis is carried out by means of a His-Asp phosphorelay, involving three key proteins: HKs, HPts, and RRs. Because habituated callus cultures no longer respond normally to cytokinin, we examined the expression of genes involved in cytokinin responses in habituated calli with respect to nonhabituated freshly derived calli (Table V). Strikingly, the expression level of the cytokinin receptor CRE1 was up-regulated 19-fold in habituated calli. Changes in the level of CRE1 expression have been reported, in response to increasing durations of cytokinin exposure, in Arabidopsis seedlings (approximately 4-fold induction after 24 h of cytokinin exposure [Rashotte et al., 2003]) as well as callus tissue (approximately 3-fold induction after 3 or 6 d on cytokinin-rich media [Che et al., 2002]). These results suggest that cytokinin sensitivity may be modulated through regulation of cytokinin-receptor production. Notably, the expression of CRE1 was not up-regulated in nonhabituated callus cultures maintained in the presence of cytokinin for 6 weeks (Table V).

Another member of the cytokinin-receptor family, AHK3, displayed altered expression in habituated calli. In this case, the cytokinin receptor was down-regulated 3.6-fold. Perhaps this down-regulation is a response to the overabundance of the CRE1 cytokinin receptor. AHK2 levels were down-regulated by only 1.7-fold in habituated calli. The AHK2 cytokinin receptor has been shown to play a very minor role in callus proliferation, as demonstrated by the wild-type callus induction of ahk2 mutant tissues, as well as by the severely compromised callus induction of cre1 ahk3 double mutant tissues (Higuchi et al., 2004). The CKI2 HK, the only HK in Arabidopsis lacking a putative receptor domain, was down-regulated 2.4-fold in habituated callus tissue. This gene was originally implicated in cytokinin signaling, as overexpression of CKI2 confers habituation on Arabidopsis callus cultures (Kakimoto, 1996). However, no subsequent evidence that CKI2 is involved in cytokinin signaling has been reported to date, as null mutations of CKI2 have no obvious cytokinin-related phenotype (Higuchi et al., 2004).

Expression of the HK AtHK1 was up-regulated 3.2-fold in habituated callus tissues. Up-regulation of AtHK1 in response to cytokinin has been reported for callus tissue, after 3 d on cytokinin-rich media (Che et al., 2002), as well as seedling tissue, after a 1.5 h exposure to cytokinin (D.J. Somers, personal communication). AtHK1 has been implicated in osmosensing, and the AtHK1 transcript is induced by both high (e.g. NaCl) and low (e.g. distilled water) osmolarity (Urao et al., 1999). It was recently shown that SLN1, a related osmosensor in yeast, responds specifically to a change in turgor (Reiser et al., 2003). T87 cell cultures are rapidly dividing with respect to freshly derived callus cultures and also have a distinct morphology. For example, cells within T87 callus cultures are more friable in texture and vitreous in appearance than cells within freshly derived callus cultures. Perhaps up-regulation of AtHK1 in the rapidly dividing T87 calli reflects changes in turgor and cell wall adhesion within the habituated cell line.

An analysis of the expression levels of RRs in habituated calli revealed up-regulation of the Type-A RRs ARR7 and ARR15. These proteins are primary cytokinin-response genes (D'Agostino et al., 2000) and act as negative regulators of cytokinin signaling (Hwang and Sheen, 2001; To et al., 2004). For example, several higher-order Type-A RR mutant combinations display a cytokinin-hypersensitive phenotype in callus culture and root elongation (To et al., 2004). Thus, inactivating these proteins in vivo relieves their repressive action on cytokinin signaling. Cytokinin induction of ARR15 in cre1 rosette leaves is wild type (Kiba et al., 2002; Nishimura et al., 2004). However, cytokinin induction of ARR15 is specifically impaired in root tissues of a CRE1 mutant (cre1-1; Kiba et al., 2002) and is completely abrogated in root tissues of the cre1 ahk2 ahk3 triple cytokinin-receptor mutant (Higuchi et al., 2004), indicating that this Type-A RR functions downstream of the CRE1 cytokinin receptor.

Whereas expression of the Type-B RRs in habituated calli was relatively unchanged (using a 2-fold-change cutoff), expression of two pseudo RR genes was altered in T87 calli. Expression of APRR2 was down-regulated 3.0-fold, while that of APRR5 was up-regulated 2.1-fold. The pseudo RRs bear homology to the Type-A and Type-B RRs, yet lack the conserved Asp residue required for propagation of a His-Asp phosphorelay. They fall into two broad categories (Makino et al., 2000): those possessing a C-terminal motif similar to that found in the CONSTANS gene product, including APRR5, and those lacking this C-terminal motif, including APRR2. This C-terminal motif seems to confer circadian regulation of transcription on the pseudo RRs (Yamamoto et al., 2003). Recently, reports that cytokinins, among other phytohormones, are involved in circadian rhythms have emerged (Hanano et al., 2005). Perhaps APRR5 represents a point of cross talk between cytokinin signaling and circadian rhythms.

Given that APRR2 belongs to a gene family that has been implicated in phytochrome-mediated circadian regulation yet is not itself regulated by light, it is intriguing that the APRR2 transcript is down-regulated in habituated callus tissue. APRR2 possesses a C-terminal protein domain that is homologous to the DNA-binding domain typical to Type-B RRs and thus may serve as a transcription factor (Makino et al., 2000). Although APRR2 cannot itself propagate a phosphorelay, protein interactions between RRs have been reported (Nakashima et al., 1991; Baikalov et al., 1998; Müller-Dieckmann et al., 1999). Perhaps APRR2 represents a point of cross talk between light- and cytokinin-mediated signaling pathways, as has been suggested for the Type-A RR ARR4. The ARR4 transcript accumulates in response to cytokinin (D'Agostino et al., 2000), while the protein accumulates in response to red light (Sweere et al., 2001). Furthermore, the ARR4 protein has been shown to bind to PHYTOCHROME B (PHYB), stabilizing this light receptor in its active form (Sweere et al., 2001). The fold-change of ARR4 expression in habituated calli compared to freshly derived calli was +1.9. Although PHYB levels were unaltered in habituated calli, one of the five red light receptors in Arabidopsis, PHYA, was down-regulated 3.6-fold in habituated callus cultures.

The HPts AHP1 and AHP5 were down-regulated in habituated calli by 12.5-fold and 1.9-fold, respectively. Overexpression of AHP2 confers cytokinin hypersensitivity on both root and hypocotyl tissues of transgenic plants (Suzuki et al., 2002), demonstrating a positive/stimulatory role for an HPt protein in cytokinin signaling. In addition, AHP1 and AHP2 have been shown to translocate from the cytoplasm to the nucleus in a cytokinin-dependent manner (Hwang and Sheen, 2001). Furthermore, coexpression of AHP1, 2, 3, or 5 along with CRE1, in a heterologous system, demonstrates that these HPt proteins are able to accept and titrate out the CRE1-initiated signal, albeit to varying degrees (Suzuki et al., 2001).

While we cannot rule out the possibility that T87 habituated cell line-derived callus cultures produce higher levels of cytokinin than nonhabituated callus cultures, the results of this study do not support this hypothesis for a number of reasons. First, several known cytokinin-inducible genes are down-regulated in habituated callus cultures rather than up-regulated (Table V). In addition, we found no up-regulation of cytokinin-producing enzymes or down-regulation of cytokinin-degrading enzymes. In fact, we saw an up-regulation of cytokinin-degrading enzymes (Table V). Furthermore, we found that a large percentage of genes up-regulated in nonhabituated callus cultures in response to the presence of cytokinin are actually down-regulated in habituated callus cultures (Supplemental Table III). Based on our results, it seems less likely that habituation is caused by an overproduction of cytokinins by the callus tissue and more likely that habituation is caused by altered expression of one or more cytokinin-signaling genes, for example, the cytokinin receptor CRE1.

We verified the 19-fold up-regulation of CRE1 in habituated callus cultures at both the mRNA and protein level. It is possible that overexpression of this cytokinin receptor alone is responsible for habituation in the T87 cell line. Overexpression of CRE1 may allow habituated T87 cells to sense very low endogenous levels of cytokinin. Other possible explanations for CRE1 overexpression-induced habituati