Gene expression associated with N-induced shifts in resource allocation in poplar

Authors

  • J. E. K. COOKE,

    1. School of Forest Resources and Conservation, and Program in Plant Molecular and Cellular Biology, University of Florida, PO Box 110410 Gainesville, FL 32611, USA and
    Search for more papers by this author
  • K. A. BROWN,

    1. School of Forest Resources and Conservation, and Program in Plant Molecular and Cellular Biology, University of Florida, PO Box 110410 Gainesville, FL 32611, USA and
    Search for more papers by this author
  • R. WU,

    1. Department of Statistics, University of Florida, PO Box 110339, Gainesville FL 32611, USA
    Search for more papers by this author
  • J. M. DAVIS

    Corresponding author
    1. School of Forest Resources and Conservation, and Program in Plant Molecular and Cellular Biology, University of Florida, PO Box 110410 Gainesville, FL 32611, USA and
    Search for more papers by this author

  • *

    Present address: Vanderbilt University, Department of Cancer Biology, 771 Preston Research Building, Nashville TN 37232, USA.

    Present address: Centre de Recherche en Biologie Forestière, Université Laval, Québec QC G1K 7P4, Canada.

Correspondence: J.M. Davis, Fax: +1 352 846 1277; e-mail: jmdavis@ufl.edu

ABSTRACT

Surprisingly little is known about molecular mechanisms by which nitrogen (N) availability acts to modulate the growth of forest trees. To address this issue, differential display was used in conjunction with filter-based arrays to identify 52 partial cDNA clones that were significantly regulated within days in response to limiting or luxuriant levels of NH4NO3 fertilization in Populus trichocarpa Torr. & Gray × deltoides Bartr. ex Marsh. A subset of these cDNAs also demonstrated shifts in expression patterns in stem-girdled trees, a manipulative physiology technique that disrupts phloem transport. Stem girdling also induced changes in glutamine and asparagine pools which were correlated with the observed changes in expression profiles for these genes. The identity of these genes provides insight into biochemical processes that are altered by N availability in poplar. Carbon–nitrogen interactions appear to figure prominently in the N-response. The gene expression data suggest that N availability modulates the partitioning of C and N resources into metabolic fates that have the potential to alter both wood quality and quantity, including synthesis of vegetative storage proteins, cell wall components, and terpenoids.

INTRODUCTION

Nitrogen has dramatic effects on plant growth and architecture. These changes in growth are a manifestation of changes in the expression of a myriad of genes that modulate resource utilization. Together, these changes in gene expression can influence how C and N resources are allocated amongst plant parts to determine plant proportions, and how these resources are partitioned into biochemical pathways that ultimately determine the characteristics of a tissue (e.g. Scheible et al. 1997a; Nielsen et al. 1998).

Comparatively little is known about the effect of N availability on gene expression in perennial species, and how N-associated changes in gene expression mediate changes in growth. However, perennial species presumably contend with considerable variation in N availability over the course of the plant's lifespan, and this is likely reflected in the overall size and form of the plant. Poplars are an ideal system to study the effects of N on the growth of a perennial species because they have a remarkable ability to take up and utilize N from their surroundings (Min et al. 2000). Furthermore, poplars display great plasticity in growth and architecture in response to changing N availability (e.g. Liu & Dickmann 1992; Heilman & Xie 1993; Coleman, Dickson & Isebrands 1998).

The N resources not only provide N moieties to biosynthetic pathways; inorganic and organic N compounds can also mediate changes in gene expression (Stitt 1999; Coruzzi & Zhou 2001). Recent genomic surveys of nitrate-responsive gene expression patterns in Arabidopsis (Wang et al. 2000) and tomato (Wang, Garvin & Kochian 2001) indicate that many genes respond rapidly to changes in N. The N-responsive genes identified in these and other studies encode for a diverse array of proteins involved in many cellular processes, including metabolism and signal transduction.

Phloem transport is an important means of allocating nutrients, including N compounds, among plant parts, and also provides a vital conduit for communication between tissues (Oparka & Turgeon 1999). Poplars are a good system for studying the role of phloem transport and phloem-transmissible compounds in the N-response. Stem phloem is relatively accessible in the bark of poplars and other trees, and makes up a sizeable percentage of the cells constituting young bark (Telewski, Aloni & Sauter 1996). In addition, patterns of transport for both C and N compounds have been well studied in poplar (reviewed in Dickson 1989). Glutamine (gln) is the major organic N transport compound in poplar (Dickson 1979; Sauter & van Cleve 1992), and is preferentially transported up the stem to developing leaves via a xylem-to-phloem transfer facilitated by ray cells (Dickson, Vogelmann & Larson 1985). Gln modulates the expression of a number of genes in herbaceous plants (Sivasankar, Rothstein & Oaks 1997; Lam, Hsieh & Coruzzi 1998; Oliveira & Coruzzi 1999; Rawat et al. 1999). Recently, the promoter of a poplar gene encoding bark storage protein (bspA) was demonstrated to be transcriptionally regulated by gln (Zhu & Coleman 2001a, 2001b).

In this study, our objective was to identify genes whose expression patterns were correlated with N-induced shifts in resource allocation. We used differential display in conjunction with filter arrays to identify 52 cDNAs whose corresponding transcript levels are significantly modulated by nitrogen availability in Populus trichocarpa Torr. & Gray × deltoides Bartr. ex Marsh. Stem-girdling experiments indicate that expression of a subset of these genes is also altered by phloem-transmissible compounds. Changes in levels of the phloem-transmissible amino acids gln and asparagine (asn) are correlated with the girdling-induced changes in gene expression patterns. The identities of these genes, together with their expression profiles, suggest N-modulated processes involved in the allocation and partitioning of resources that potentially impact both wood quality and wood quantity.

MATERIALS AND METHODS

Plant material and cultivation

Rooted cuttings of Populus trichocarpa Torr. & Gray × deltoides Bartr. ex Marsh hybrid UCC-1 (provided by Union Camp Corporation, Savannah GA, USA in 1997) were grown as described in Lawrence et al. (1997). The plants used in the differential display experiments were grown in growth chambers (Conviron Model E15; Controlled Environments Ltd, Winnipeg, MB, Canada) under a 12 h photoperiod at 25 °C. All other experiments were carried out in greenhouses maintained at 20–35 °C under natural lighting with day lengths of 12–14 h. Plants were fertilized every 3 d with Hocking's complete nutrient solution (Hocking 1971) supplemented with 5 mm ammonium nitrate (molarity calculated with respect to NH4NO3) prior to experimentation.

N-availability experiments

Plants that were approximately 60–80 cm tall were ranked according to height. Each cohort was then divided equally and individuals of a cohort assigned randomly between treatments such that there were at least three plants per treatment in each experiment. At day 0, the plants were fertilized daily with Hocking's solution supplemented with 0 mm, 2 mm, or 50 mm NH4NO3. Harvested tissues were quick frozen in liquid nitrogen and stored at −80 °C until analysed. Leaves were numbered according to leaf plastochron index (LPI; Larson & Isebrands 1971), with LPI 0 being approximately 2 cm long with a one-half expanded lamina. Shoot tip samples included the shoot apex, all leaves up to and including LPI 1, and the adjoining stem segment. Tissue samples for differential display and RNA analysis included source leaves collected at approximately one-third of the total plant height (generally, LPI 15 and LPI 16 were harvested), stems (sampled at internodes between LPI 14 and LPI 17), and total root mass.

Stem-girdling experiments

Plants at least 60 cm in height were assigned to treatments as described above. Daily fertilization with Hocking's solution plus 2 mm NH4NO3 was started 2 weeks prior to girdling and continued through the duration of the experiment. Girdling was performed at day 0 by mechanically removing a 1 cm strip of bark just above the midpoint of the stem, which corresponded to between LPI 8 and LPI 11, depending on plant height. Source leaves were present both above and below the girdle. A single internode of stem section was harvested one internode away from the girdle. A region corresponding to the girdle was marked on the stems of control plants so that comparable tissues could be harvested. This control non-girdled tissue is referred to in the text as a ‘girdle-equivalent’. Stem sections were separated into wood and bark for subsequent analyses. These are referred to as xylem and phloem, respectively, since these are the predominant tissues in the wood and bark at this stage of development. Harvests were carried out at the midpoint of the day.

Carbon and nitrogen analyses

Total C and N content of tissues was determined using a NCS 2500 automatic elemental analyser (CE Instruments, ThermoQuest Italia SpA, Milan, Italy). The instrument was calibrated with a pine needle standard (National Institute for Standards and Testing, Gaithersburg, MD, USA).

Photosynthesis measurements

CO2 exchange was measured with a portable infrared gas analyser (Li 6400P photosynthesis system; Li-Cor Inc., Lincoln, NE, USA). Photosynthetically active radiance was maintained at 1000 µmol s−1, airflow rate at 500 µmol s−1 and leaf temperature between 27 and 30 °C. The chamber CO2 was adjusted to 400 µmol mol−1.

Amino acid analyses

Amino acids extracts were prepared according to King & Gifford (1997). Samples were derivatized with phenylisothiocyanate on an ABI420 derivatizer and analysed with an ABI172 high-performance liquid chromatograph (HPLC; Applied Biosystems, Foster City, CA, USA).

RNA extraction and differential display

Total RNA was extracted using a CTAB method (Chang, Puryear & Cairney 1993). All steps of the differential display [differential display reverse transcriptase (DDRT)-polymerase chain reaction (PCR)] process were conducted essentially according to the manufacturer's instructions (GenHunter, Nashville, TN, USA). Briefly, reverse transcription was carried out with DNase-treated RNA using MMLV-RT from RNAimage (GenHunter) or Gibco-BRL (Gaithersburg, MD, USA) with identical results. PCR was performed with factorial combinations of the three anchored primers and 31 random primers (no. 9 to 39), for a total of 93 reactions (GenHunter); 74 KBq α-[33P]dATP (NEN, Boston, MA, USA) was included in each 10 µL reaction. The PCR products were separated in 6% denaturing polyacrylamide gels (HR1000 gel; Genomyx, Foster City, CA, USA) subjected to electrophoresis under 60 W constant power, and visualized via autoradiography. Differentially displayed products were recovered, re-amplified by PCR, and ligated into pGEM-T (Promega, Madison, WI, USA) for transformation into competent E. coli DH5α (Gibco BRL). Sequencing was carried out by the Uni-versity of Florida DNA Sequencing Core Laboratory (Gainesville, FL, USA). Sequences were analysed using BLAST utilities (Altschul et al. 1997) at NCBI (http:www.ncbi.nlm.nih.govBLAST).

Array Construction cDNA inserts were PCR-amplified from plasmids using SP6 and T7 primers and purified with Qiaex resin (Qiagen, Valencia, CA, USA). PCR products were checked on agarose gels to ensure appropriate size, quantity and integrity. Loading dye was added to the cDNA to facilitate spotting.

The filter arrays were produced using a procedure modified from Desprez et al. (1998). cDNAs were arrayed onto Hybond N+ membranes using a 96-pin colony replicator together with an alignment template that allowed for a total of 384 spots per membrane (V & P Scientific, San Diego, CA, USA). Each spot contained approximately 4 ng DNA. After arraying the cDNAs, the membranes were incubated DNA side up on 1.5 m NaCl/0.5 m NaOH for 10 min, followed by 0.5 m Tris·HCl (pH 7.4)/0.5 m NaOH for 5 min. Membranes were rinsed in 2× NaCl/sodium citrate buffer (SSC)/0.1% (w/v) sodium dodecyl sulphate (SDS) before air-drying.

Each cDNA was arrayed as a quartet; namely a group of four spots. Each quartet was replicated at random three times on a membrane. A total of four membranes were used to accommodate all cDNAs. Two cDNAs (pni79 and pot178), shown by Northern blot analysis to be expressed at relatively constant levels in all tissues under different nitrogen regimes, were included on every membrane for comparison purposes. In addition, a cDNA (pni171–4) known to be a low abundance transcript based on Northern analysis was included on each blot.

Probe preparation and array hybridization

Arrays were probed with first-strand cDNA synthesized from 8 or 10 µg total RNA using MMLV-RT (Gibco-BRL), with 40 pmol (4.44 MBq) α-[32P]dCTP, and dATP, dTTP and dGTP in excess. Labelled cDNA probes were separated from unincorporated nucleotides using Sephadex G50 columns (Amersham Pharmacia Biotech, Ithaca, NY, USA), and quantified by liquid scintillation (Beckman Coulter, Fullerton, CA, USA). Hybridization and washing were carried out at 65 °C using standard protocols described in Davis et al. (1991).

Array data acquisition and analysis

Spot intensities on the arrays were visualized with a phosphorimager and quantified using the manufacturer's ImageQuant software (Molecular Dynamics, Amersham Biosciences Corp., Piscataway, NJ, USA). Data were imported into Microsoft Excel (Microsoft Corp., Redmond, WA, USA) for further analyses.

Defective spots, including those with high local background, were removed from the analysis. To facilitate comparisons between blots, data were scaled by dividing each spot value by the standard deviation of all spots on a blot (xsi = xi/s), a slight modification of the scaling function used by Schenk et al. (2000). As an independent check, data were also scaled by dividing each spot value by the average value of the two ‘constitutive’ cDNAs included on every blot. There was an overall 85% agreement between the two scaling methods in the statistical analysis. However, because scaling with standard deviations is a more widely applicable and objective technique, the data in this paper are presented as xsi = xi/s. A low signal cut-off value for each blot was determined as the mean value plus one standard deviation (inline image + s) of the low-expressing cDNA described above. A cDNA that did not exhibit a mean value greater than this cut-off value in at least one treatment of the statistical analysis was considered below detectable limits.

Statistics

Statistical analyses of the array data and physiological data were essentially the same: t-tests, anova, and multiple comparisons of means (with least squared means option) were carried out using Excel and SAS (SAS Institute, Cary, NC, USA).

RNA blots

RNA blots were used to confirm expression patterns for several cDNAs. The cDNAs for labelling were prepared by PCR as described above, and radiolabelled with α-[32P]dCTP using random primer labelling (GibcoBRL). Formaldehyde-agarose gel electrophoresis of total RNA, as well as hybridization and washing of the RNA blots, was carried out essentially as described in Davis et al. (1991). The final wash [1% (w/v) SDS/1 mm EDTA/40 mm Na phosphate buffer (pH 7.4)] was performed at 55 or 65 °C.

RESULTS

Nitrogen availability affects resource allocation in poplar within days

In order to determine the time frame in which shifts in resource allocation occur in Populus trichocarpa × deltoides in response to N fertilization, we examined total N and C content of leaves, stems and roots over a 4 week time course treatment with limiting (0 mm), adequate (2 mm), or luxuriant (50 mm) NH4NO3. The mean height of the trees at day 0 was 74.7 ± 9.2 cm (mean ± SD). After 28 d of treatment with 0, 2 or 50 mm NH4NO3, the tree heights were 146.7 ± 10.9, 172.7 ± 14.8 and 168.3 ± 10.4 cm, respectively. The trees did not exhibit symptoms of deficiencies for other nutrients, such as P, K, or Mg over the course of the experiment (data not shown).

The trees showed perceptible differences in total N content after 7 d of treatment (Fig. 1a–c). The difference in N content of stems and roots grown with 0 versus 50 mm NH4NO3 was statistically significant at 7 d (Fig. 1b–c). There were significant differences in leaf N content at 14 d (Fig. 1a). These differences were even greater by 28 d of treatment. When expressed on a percentage dry weight basis, differences in N content between 0- and 50 mm NH4NO3-treated plants were statistically significant in leaves, stems and roots by 7 d of treatment (data not shown). By 28 d, there were 2.5-, 4.9- and 3.0-fold differences in percentage N content of leaves, stems and roots, respectively, from 0 and 50 mm NH4NO3-treated plants (data not shown).

Figure 1.

Increasing N availability is correlated with increased N and C allocation to leaves and stems, and a concomitant shift in the C : N ratio. Plants were treated with 0 mm (diagonally hatched bars) 2 mm (cross-hatched bars) or 50 mm (solid bars) NH4NO3 for up to 28 d. Total N in (a) leaves (b) stems and (c) roots. Total C in (d) leaves (e) stems and (f) roots. C : N ratio in (g) leaves (h) stems and (i) roots. Data are the means of three separate experiments, with at least two plants per treatment per day per experiment. anova and multiple comparison of means were used to determine significant differences between 0, 2 and 50 mm NH4NO3 treatments for each tissue for each day (α = 0.05). Means assigned the same letter within a grouping are not significantly different.

Total C allocated to roots did not differ significantly between treatments at 7, 14 or 28 d (Fig. 1f). Although the roots never stopped growing during these experiments, containment of roots by the pots may have limited root growth to some extent, which is an inherent limitation of pot studies. Total C allocated to leaves steadily increased in 2 and 50 mm NH4NO3-treated plants, but remained relatively constant in 0 mm NH4NO3-treated plants (Fig. 1d). Total C content of leaves from 2 and 50 mm NH4NO3-treated plants was significantly different from total C of leaves from 0 mm NH4NO3-treated plants by 14 d, and the differences were more pronounced at 28 d. Increased allocation of C resources to leaves under adequate and luxuriant N conditions was mainly due to increased size and number of leaves (data not shown). Differences in C allocation to stems of 0 versus 50 mm NH4NO3-treated plants were significant only by 28 d of treatment (Fig. 1e).

Increasing N availability caused a shift in the C : N ratio of all tissues (Fig. 1g–i), even though increasing N availability caused an increase in both N and C resources to leaves and stems. Differences in C : N ratios between treatments were statistically significant by 7 d of treatment. By 28 d of treatment, there were 2.5-, 5.2- and 3.9-fold differences between the C : N ratios of 0 and 50 mm NH4NO3-treated plants in leaves, stems and roots, respectively.

Using DDRT-PCR and arrays to identify N-associated genes

DDRT-PCR was carried out with RNA from plants treated daily for 10 d with either 0 or 50 mm NH4NO3, namely during the window in which shifts in allocation were occurring. RNA from shoot tips, mature leaves and stems was used in the DDRT-PCR reactions. Ninety-three primer combinations were used for differential display, suggesting that approximately one-third of the transcriptome was sampled (RNAimage; Liang & Pardee 1997). Altogether, 97 unique DD cDNA products were successfully amplified, cloned and sequenced. These cDNAs were designated pni (poplar nitrogen-responsive), based on their putative regulation. These pni clones, as well as additional cDNAs from our collection relevant to N utilization and metabolic processes (designated po or pot), were used for analysis of mRNA abundance with filter arrays. Since pni34, pni67 and pni207 were 98% identical to bsp, encoding bark storage protein (BSP), and exhibited an expression pattern identical to that of bsp in Northern blot analysis (data not shown), the full-length bsp was used on the arrays instead of the partial cDNAs.

Arrays were probed with first-strand cDNA synthesized from mRNA extracted from shoot tips, leaves, stems or roots of plants treated with either 0 or 50 mm NH4NO3 daily for 14 d. Transcript abundance for each of the 116 cDNA clones in each of the different tissues from 0 versus 50 mm NH4NO3-treated plants were statistically compared using t-tests. Because of the variation associated with measuring transcript abundance from experiment to experiment, a relatively permissive α = 0.1 was chosen as the rejection level. Altogether, 52 of the 116 cDNA clones analysed by filter arrays demonstrated significantly different transcript levels in 0 versus 50 mm NH4NO3-treated plants in at least one tissue (Table 1; Fig. 2). This included 49 pni clones identified by DDRT-PCR. Ten of the 116 cDNA clones displayed transcript abundances below the threshold set for reliable detection in all tissues (see Materials and methods for details). Of the 52 cDNAs showing a significant N response, eight pni clones show no similarity to GenBank sequences. Another 17 pni clones show similarities to genes or ESTs with unknown function.

Table 1.  Nitrogen-responsive cDNA clones identified by differential display in Populus trichocarpa × deltoides
CloneLength (bp)Accession noRelated sequenceE value
  1. The length of the 3′ cloned partial cDNA is given in base pairs. Sequence similarities based on BLASTN (ESTs) or BLASTX (others) (NCBI). The GenBank accession number of the sequence demonstrating the highest similarity is indicated in parentheses.

cDNAs with putative function
bsp1153CAA49669Populus bark storage protein
win41112AAA16342Populus vegetative storage protein
pni288 934AF330050Populus win4 vegetative storage protein (AAA16342)3 × 10−51
pni69 695BU791162Lycopersicon ADP-glc pyrophosphorylase large subunit (U85497)3 × 10−73
pni88 271BU791222Lycopersicon osmotin-like protein (L76632)3 × 10−4
pni95 201BU791224Arabidopsis similar to protein phosphatase 2C (ABI1-like) (AAD46006)2 × 10−19
pni102 387BU791128Medicago histone H3 (U09460)1 × 10−52
pni107 537BU791129Solanum soluble starch synthase (P93568)2 × 10−28
pni122 835BU791131Arabidopsis similar to bHLH regulatory proteins (AL023094)3 × 10−46
pni134 430BU791225Brassica pollen coat protein similar to cold-induced kin1 (BAB10133)7 × 10−7
pni145 592BU791172Arabidopsis similar to chloroplast 31 kDa RNA-binding protein (CAB78028)8 × 10−40
pni1501250BU791137Arabidopsis glutathione-S-transferase (U70672)3 × 10−37
pni159 705BU791139Citrus thiazole biosynthetic enzyme (Z82983)2 × 10−84
pni164 267BU791176Nicotiana kinesin-like calmodulin-binding protein2 × 10−22
pni183 731BU791143Arabidopsis similar to recA and LRR proteins (AAG51016)3 × 10−29
pni212 583BU791147GlycineSE60 sulfur-rich protein similar to proteinase inhibitor II (Z18359)5 × 10−19
pni220 505BU791148Arabidopsis ABC transporter-like protein (BAB10074)2 × 10−39
pni236 830BU791150Arabidopsis similar to protein disulfide isomerase (AAD46003)4 × 10−35
pni240 540BU791152LavateraLTCO11 similar to GASA, gibberellin-regulated proteins (AF00784)9 × 10−24
pni263 771BU791154Gossypium sesquiterpene cyclase (U88318)1 × 10−42
pni275 259BU791195Arabidopsis similar to ser hydroxymethyltransferase (AAG40343)5 × 10−10
pni278 776BU791156Arabidopsis similar to vacuolar sorting receptor (AAF98196)3 × 10−31
pni282 492BU791157Lycopersicon histone H1 (AF253416)5 × 10−6
pni287 461BU791158Malus proline rich protein (T17107)2 × 10−20
pni289 409BU925847Pyrus Rubisco ss (D00572)9 × 10−38
po109 318AY166668Populus photosystem Q(B) protein (P36491)3 × 10−56
po164 424AY166669Fragaria hydroxyproline rich protein (AAD01800)4 × 10−14
pot171 536AY161276Populus Caffeoyl-CoA 3-O-methyltransferase (AJ224895)3 × 10−75
cDNAs with unknown function
pni22 617BU791149No similarity
pni27 471BU791155Populus cambial region EST A047P47 U (AI163751)3 × 10−4
pni28 409BU791204Arabidopsis unknown protein (AAF27106)2 × 10−20
pni48 323BU791211No similarity
pni68 160BU791218No similarity
pni120 503BU791130Arabidopsis unknown protein (AAK59650)2 × 10−26
pni123 535BU791165No similarity
pni125 358BU791166Populus cambial region EST A010P08 U (AI161945)1 × 10−130
pni130 571BU791133No similarity
pni135 248BU791169No similarity
pni138 335BU791170Populus leaf EST CO74P73 U (BI072388)5 × 10−58
pni143 189BU791171No similarity
pni151 212BU791173Populus cambial region EST A031P41 U (AI163072)1 × 10−5
pni171–1 298BU791178Populus developing bud cDNA bd8–1 (BF299456)1 × 10−83
pni175 243BU791181Arabidopsis unknown protein (AAD21437)1 × 10−4
pni177 269BU791182Arabidopsis unknown protein (AAF21213)3 × 10−22
pni180 481BU791141Arabidopsis unknown protein (BAB11541)2 × 10−19
pni182 534BU791141Populus cambial region EST A028P59 U (AI63003)1 × 10−56
pni184 508BU791144Arabidopsis unknown protein (CAB86048)5 × 10−23
pni187 482BU791145Populus leaf EST C062P01 U (BI071672)7 × 10−64
pni218 402BU925848Cicer tissue specific protein (X97455)6 × 10−6
pni242 398BU791190Arabidopsis unknown protein (AAK59486)6 × 10−17
pni246 595BU791153Arabidopsis unknown protein (AAG52577)4 × 10−49
pni279 426BU791199Arabidopsis unknown protein (AAD32907)3 × 10−29
pni285 375BU791201No similarity
pni286 300BU791203Populus leaf EST CO74P73 U (BI072388)4 × 10−52
Figure 2.

A nitrogen-response roadmap in poplar. Filters arrayed with cDNAs were probed with 32P-labelled first-strand cDNA synthesized from 10 µg total RNA extracted from tissues of plants treated with 0 or 50 mm NH4NO3 for 16 d. t-tests were used to statistically compare scaled transcript abundance values in 0 versus 50 mm NH4NO3 for each tissue for each cDNA. The cDNAs showing significantly greater transcript abundance with 0 mm NH4NO3 treatment than with 50 mm NH4NO3 treatment were considered to be low-N induced, and are shown in yellow (P = 0.01–0.1) or red (P < 0.01). The cDNAs showing significantly greater transcript abundance in 50 mm NH4NO3 treatment than with 0 mm NH4NO3 treatment are considered to be high-N induced, and are shown in light green (P = 0.01–0.1) or dark green (P < 0.01). Light grey indicates no significant difference in transcript abundance was detected between treatments. Dark grey indicates that the mean transcript abundance in both 0 and 50 mm NH4NO3 treatments was below the mean + SD transcript abundance for a low-abundance reference cDNA. Individual cDNAs were replicated three times.

Northern blots were used to confirm the expression patterns for 24 pni clones, representing both significantly and not significantly responsive cDNAs (data not shown). In all cases, the array expression patterns were confirmed using Northern blots. In five of the 24 comparisons, differences between 0 and 50 mm NH4NO3 treatments observed with Northern blots were not found to be statistically significant in the array data, although the transcript abundance patterns were consistent. This was mainly due to variance associated with the array data.

Twenty of the 26 cDNAs that displayed a significant response to N levels in stems (Fig. 2) also showed significant differences (P < 0.01) in transcript abundance in xylem versus phloem. Of these, 12 showed higher expression in phloem, whereas the remainder showed higher expression in xylem. A subset of these cDNAs is depicted in Fig. 3.

Figure 3.

Several stem N-responsive cDNAs are preferentially expressed in xylem or phloem. Trees were treated for 3 weeks with 2 mm NH4NO3. Filters were probed with 32P-labelled cDNA synthesized from 8 µg total RNA. Transcript abundance is expressed as the log10 values of inline image = 10(Σxi/s)/n (see Materials and methods). Individual cDNAs were replicated three times. Plotted values for xylem (cross-hatched bars) and phloem (solid bars) are means ± SD. Values to the right of each bar pair indicate the fold difference in expression of a cDNA between xylem and phloem.

Girdling as a tool to disrupt phloem transport

Removing a strip of bark from the circumference of a tree stem can be used to disrupt phloem transport (e.g. Day & DeJong 1990; Sauter & Neumann 1994) (Fig. 4a). Xylem-to-phloem transfer of gln has been shown to occur along the entire stem length in poplar, with a higher proportion occurring between LPI 8 and 13 (Dickson et al. 1985). Accordingly, we positioned the girdle between LPI 8 and LPI 11. Photosynthesis was used as an indicator of plant vigour in the girdling experiments. Photosynthetic rates of leaves above or below the girdle or girdle-equivalent did not show significant changes over the duration of the experiment (data not shown).

Figure 4.

Disrupting phloem transport by stem girdling causes an increase in percentage N below the girdle. (a) A schematic representation of stem girdling, providing a legend used for the bar fills in b-d. (b) Xylem percentage N ratios of girdled plants to control plants, sampled above (striped bars) or below (grey bars) the girdle or girdle-equivalent. (c) Phloem percentage N ratios of girdled plants to control plants, sampled above (striped bars) or below (grey bars) the girdle or girdle-equivalent. (d) C : N ratios at 6 d after girdling of control tissues above (diagonally-hatched bars) or below (open bars) the girdle-equivalent; girdled tissues above (cross-hatched bars) or below (solid bars) the girdle. Results are the means of three separate experiments, with at least two plants per day per experiment. anova and multiple comparison of means were used to determine significant differences (α = 0.05). Means assigned the same letter are not significantly different. In (b), (c), lowercase letters are used to denote significance between days within tissues sampled either above or below the girdle. Uppercase letters are used to denote significance between sampling position of a tissue within a day. In (d), letters indicate significance within tissue.

Although girdling did not significantly affect percentage N (α = 0.05) above the girdle or girdle-equivalent, it stimulated a significant increase below the girdle within 3 d (Fig. 4). By 6 d, there was a 2.4-fold and 1.8-fold greater percentage N content in girdled stem xylem (Fig. 4b) and phloem (Fig. 4c), respectively, relative to control tissues below the girdle or girdle-equivalent. This shifted the C : N ratio by approximately two-fold in both xylem and phloem tissues (Fig. 4d).

HPLC analysis of soluble amino acid pools demonstrated that stem girdling resulted in a statistically significant increase (α = 0.05) in total amino acids in both the xylem and phloem below the girdle compared to control plants (data not shown). Although there was a decrease in total amino acids in xylem and phloem above the girdle compared to control plants, this difference was not statistically significant (data not shown).

HPLC analyses also established that gln and asn are the major components of the soluble amino acid pool of both xylem and phloem for Populus trichocarpa × deltoides fertilized with 2 mm NH4NO3. Gln and asn made up 65 ± 7 and 19 ± 3% of the xylem amino acid pool, and 62 ± 4 and 23 ± 4% of the phloem amino acid pool of ungirdled plants (mean ± SD), respectively.

Gln and asn levels increased significantly in both xylem and phloem below the girdle in comparison with control plants (Fig. 5). A significant increase in gln in xylem below the girdle was detected by 1 d after girdling. Increases in gln in phloem below the girdle, and increases in asn in both xylem and phloem below the girdle were significant by 3 d after girdling. Gln and asn levels did not significantly change in xylem or phloem of the control plants.

Figure 5.

Disrupting phloem transport by stem girdling results in an accumulation of gln and asn below the girdle. (a) Gln levels in xylem. (b) Gln levels in phloem. (c) Asn levels in xylem. (d) Asn levels in phloem. Open symbols, control plants; closed symbols, girdled plants; circles, above the girdle or girdle-equivalent; squares, below the girdle or girdle-equivalent. Results are the means ± SD of three separate experiments.

The accumulation of gln and asn in tissues sampled below the girdle was significantly greater than the 2.8-fold (xylem) and 1.5-fold (phloem) increases observed in the pool of 17 other amino acids combined (excluding arginine, see below) by 6 d after girdling (data not shown). By 6 d after stem girdling, there was a 5.3- and 5.6-fold increase in gln and asn, respectively, in xylem below the girdle compared with its control. There was a 2.5- and 3.2-fold increase in gln and asn, respectively, in phloem below the girdle. Gln levels also decreased above the girdle in xylem (2.0-fold) and phloem (2.8-fold). Only the difference in phloem gln content was statistically significant at α = 0.05. Asn levels did not appreciably decrease above the girdle in either xylem or phloem.

Arginine also showed a statistically significant increase in both xylem and phloem below the girdle relative to tissues sampled from control plants below the girdle equivalent. However, arginine accounted for only 1–2% of the total soluble amino acid pool of tissues from control plants, and 2–3% of the total soluble amino acid pool of tissues sampled below the girdle (data not shown).

A subset of N-associated genes is also girdling responsive

Filter arrays were used to compare gene expression profiles in xylem and phloem of girdled versus control plants sampled both above and below the girdle or girdle-equivalent. We used 6 d girdled plants to maximize the probability of detecting statistically significant differences. The data were analysed by multifactor anova. cDNAs that showed a significant treatment × position or treatment × position × tissue interaction were further analysed using multiple comparison of means to determine which of the eight different samples displayed significantly different transcript abundance.

The cDNAs were considered positively N-responsive if they demonstrated increased transcript abundance below the girdle relative to above the girdle and to control plants. The cDNAs were considered negatively N-responsive if they demonstrated either increased transcript abundance above the girdle relative to below the girdle and to control plants, or decreased transcript abundance below the girdle relative to above the girdle and to control plants. cDNAs whose expression profiles showed differences mainly attributable to wounding (approximately equal induction in tissues above and below the girdle) or development (increased expression either above or below the girdle/girdle-equivalent in both girdled and ungirdled tissues) were removed from further analysis. Sixteen cDNAs identified as significantly girdling-responsive by the multifactor anova showed a pattern of expression that was congruous with their response to 0 versus 50 mm NH4NO3 fertilization. Expression patterns for 10 cDNAs are depicted in Fig. 6.

Figure 6.

N-responsive cDNAs that are also girdling-responsive. Xylem and phloem were sampled from above or below the girdle or girdle-equivalent at 6 d post-girdling. Filters arrayed with cDNAs were probed with 32P-labelled cDNA synthesized from 8 µg total RNA. Individual cDNAs were replicated three times. Transcript abundance values are expressed as inline image = 10(Σxi/s)/n (see Materials and methods). Control tissues above (open bars) or below (diagonally hatched bars) the girdle-equivalent; girdled tissues above (cross-hatched bars) or below (black bars) the girdle. Girdling-responsive cDNAs were identified by multifactor anova and multiple comparison of means. Individual cDNAs were replicated three times. Means assigned the same letter are not significantly different at α = 0.05.

As with the N-responsive cDNAs, expression patterns for several girdling-responsive cDNAs were confirmed by Northern blot analysis. The cDNAs that were not girdling-responsive by array analysis were also assessed to test for false negatives. In total, 14 cDNAs were examined by Northern blots; the patterns of expression for all the cDNAs were consistent between Northern blots and arrays (data not shown).

DISCUSSION

The N-response roadmap

Availability of inorganic N is often cited as a major factor limiting forest tree productivity (Dickson 1989). Many studies conducted at the whole plant level have demonstrated that N fertilization increases above-ground biomass of forest trees (e.g. Teskey, Gholz & Cropper 1994; Samuelson 1998). Despite the economic and ecological importance of forest trees, however, few tangible links have been made between production physiology and molecular physiology in forest tree species. Accordingly, we adopted a gene discovery strategy as our starting point to identify molecular mechanisms by which N availability impacts resource allocation and partitioning in poplar trees. The whole plant allocation data allowed us to determine that shifts in resource allocation in Populus trichocarpa × deltoides occur primarily between 7 and 14 d of treatment with either limiting or luxuriant levels of NH4NO3. To capture changes in gene expression linked to shifts in resource allocation priority, DDRT-PCR was carried out with RNA from plants treated daily for 10 d with either 0 or 50 mm NH4NO3.

The spatial representation of genes whose expression profiles are altered in response to limiting versus luxuriant N can be thought of as a roadmap of the response of a woody plant to N availability at the transcriptional level. It needs to be stressed that because we sampled for changes in gene expression at 10–14 d post-treatment, many of the genes that exhibit differential gene expression are probably not regulated sensu stricto by inorganic or organic N compounds. For these genes, there are likely to be several degrees of separation between the perception of N compounds by the plant and the signal transduction networks that mediate their expression. Thus, within the set of genes on the N-response roadmap, some genes may be directly responsive to N compounds, whereas other genes are likely to be only indirectly responsive to N compounds. Nevertheless, differential expression of both subsets of genes coincides with changes in allocation of resources, and directly and indirectly N-responsive genes both have the potential to contribute to the overall changes in growth and architecture that are part of the global response to N availability.

The putative identities of genes shown to respond to limiting versus luxuriant N provide important clues to biochemical processes that may be altered by N availability in forest trees. At the same time, the number of genes on the roadmap for which there is no known function – 26 out of 52 or one-half of the genes – is noteworthy. Fourteen of these genes show a significant response to N in the woody tissues of the plant. Of these 14 genes, nine show no sequence similarity to genes with assigned function or to genes from the Arabidopsis database. Also noteworthy is that 33 of the 52 genes on the roadmap are up-regulated under limiting N conditions. Previous studies on N-responsive gene expression have tended to focus upon genes that are positively responsive to N.

Phloem-transmissible compounds alter the expression of some N-responsive genes

We used experimental manipulation of phloem transport as the next step in characterizing the response of cDNAs on the N-response roadmap to N availability in planta. Dickson et al. (1985) showed that gln is transported up the stem to developing leaves via a xylem-to-phloem transfer in poplars. We predicted that disrupting phloem transport by stem girdling would result in an accumulation of amino acids in stems below the girdle. Our data indicate that levels of both gln and asn are altered by stem girdling, mainly via an accumulation of gln and asn below the girdle. Thus, stem girdling is an experimental technique that can be used to manipulate the in planta N-status of trees over relatively short time frames.

The expression of a subset of stem-expressed genes on the N-response roadmap was found to be responsive to stem girdling. The expression patterns of these genes were correlated with altered gln and asn levels in the girdled plants, and were also consistent with their expression patterns in limiting versus luxuriant NH4NO3 applications. These data suggest that this subset of genes may be responding to altered N status of the tissues, and as such are putative N-responsive genes. We recognize that stem girdling alters pools of other phloem-transmissible compounds, and we cannot rule out the possibility at this point that these compounds may be affecting the observed changes in expression. Further studies, particularly with shorter time courses, will be essential to determine which of the genes that are responsive both to NH4NO3 application and to stem girdling are bona fide N-responsive genes. The remaining stem-expressed genes identified in the limiting versus luxuriant N availability study are either not girdling-responsive, or are girdling-responsive in a pattern inconsistent with tissue N status. The transcriptional activation of this subset of genes is probably not N-mediated, but rather a consequence of growth-associated parameters that result from increased N availability.

Promoter discovery and analysis of both directly and indirectly N-responsive genes will be integral components of a functional genomics approach to dissecting the N-response in poplar. Directly N-responsive genes ensuing from this work can serve as ‘anchors’ for studies of signal transduction networks in trees that are mediated by organic or inorganic N moieties. Cloning and analysis of directly N-responsive gene promoters may lead to the discovery and/or further definition of promoter motifs important in N-mediated signal transduction. Investigating the transcriptional activation of these promoters in trees may help us to understand how N moieties regulate N utilization strategies that are important for the perennial lifestyle, such as seasonal N cycling. Analysis of the bsp promoter in poplar has already yielded important insight into this process (Zhu & Coleman 2001a, b). No less important, promoter analysis of genes that are indirectly N-responsive may lead to the identification of motifs common to these promoters, and ultimately to the transcription factors that bind these motifs. In this way, these genes will help us to elucidate the far-reaching signal transduction ‘web’ that is initiated by N perception, helping to form a global picture of how N availability mediates plant growth and development. On a more applied level, N-responsive genes that exhibit tissue specificity make especially attractive targets for promoter discovery, since one of the constraints in woody plant transgenic research at present is the paucity of tissue-specific promoters with characterized triggers.

N availability alters expression of genes that potentially impact wood properties

In addition to identifying genes that may be useful for dissecting N-mediated signal transduction networks, the putative identities of genes that respond to limiting versus luxuriant N provide clues to cellular processes that are altered by N availability in forest trees. Based on these putative identities, we can make predictions about physiological events and metabolic pathways that may be altered in response to N availability in forest trees. This insight allows us to formulate specific hypotheses, so that we can begin to dissect the N-response at the biochemical level. In particular, we can explore how these changes at the biochemical level are affected by transcriptional regulation of the genes identified in this study.

By taking advantage of stem girdling as a means of altering the N status of poplars, we focused our attention on putative N-responsive genes in stems; that is, wood. Increased N fertilization has been correlated with altered wood properties such as reduced specific gravity (Blankenhorn et al. 1992), a utilitarian measure that reflects cell size, cell wall thickness, cell wall composition, and proportion of early wood to late wood (Koch 1972). However, we know very little about the mechanisms that link N availability to changes in the cellular structure or composition of wood. The putative identities of some of the girdling-responsive, N-associated cDNAs reveal hints as to how N availability may alter both wood quantity and wood quality.

Interactions between C and N metabolism and cross-talk between C- and N-signalling networks are emerging as important themes in how C and N resource availability affects growth and development (Koch 1997; Coruzzi & Zhou 2001). Putative identities of some of the girdling-responsive, N-associated genes suggest that limiting versus luxuriant N availability can shift the relative partitioning of C and N resources into C-intensive versus N-intensive biosynthetic pathways that may affect wood. We highlight a few examples that illustrate this concept, along with speculation as to how modulation of these genes by N availability may alter wood quantity or wood quality.

Vegetative storage proteins (VSPs) are one means that plants have evolved to store and recycle nutrients in response to changing N availability (Staswick 1994; Coleman 1997). VSPs play an especially prominent role in perennials, in which seasonal N cycling is an important process associated with overwintering (Coleman 1997). Up-regulation of the poplar vsps win4, bsp and pni288 by increased N availability and wounding has been well documented (Coleman, Pilar Bañados & Chen 1994; Lawrence et al. 1997; Lawrence et al. 2001). Zhu & Coleman (2001a) have demonstrated that the bspA promoter is responsive to exogenously applied gln. In this study, win4, bsp and pni288 all demonstrated a positive N-response, while exhibiting somewhat different spatial expression patterns. Lawrence et al. (2001) suggested that VSPs such as BSP, WIN4 and PNI288 act as molecular determinants of N sink strength within the plant, rather than passive consequences of luxuriant N levels within the cells. If this is true, then reducing individual VSP levels in the plant should alter resource allocation patterns in a manner consistent with their spatial expression pattern. This hypothesis is supported by preliminary analysis of bsp-down-regulated poplars: bsp is preferentially synthesized in stems, and transgenic plants showed reduced stem height and increased leaf area, with a correspondingly increased leaf:stem biomass ratio (G.D. Coleman, personal communication).

We identified four candidate N-responsive cDNAs that suggest limiting N conditions can increase the partitioning of both C and N resources into pathways contributing to cell wall synthesis, a C-intensive process. Lignin is a major component of cell walls, making up as much as 35% of the dry weight of wood (Baucher et al. 1998). Lignin imparts strength and hydrophobicity to the cell walls in secondary xylem, but causes discoloration and reduced brightness of pulp products (Whetten, MacKay & Sederoff 1998). pot171, which was negatively N-responsive in this study, encodes a protein similar to caffeoyl-CoA 3-O-methyltransferase (CCoAOMT) of the lignin biosynthetic pathway (Baucher et al. 1998). Experiments with transgenic plants indicate that CCoAOMT down-regulation can alter lignin content and composition (Meyermans et al. 2000; Zhong et al. 2000; Guo et al. 2001; Pinçon et al. 2001). Thus, modulation of pot171 expression represents a potential mechanism by which N availability might regulate the partitioning of resources into C-intensive pathways, and impact wood quality.

Methylation of lignin intermediates by CCoAOMT and other OMTs involves transfer of one-carbon units from S-adenosylmethionine (Vander Mijnsbrugge et al. 2000). Hanson & Roje (2001) estimate that in woody plants, lignin biosynthesis demands 10-fold more one-carbon units than all of primary metabolism combined. pni275 encodes a protein similar to serine hydroxymethyltransferase, one of the enzymes involved in S-adenosylmethionine biosynthesis. pni275 was negatively N-responsive in this study, suggesting that the availability of one-carbon units for metabolism, including lignin biosynthesis, may be reduced under luxuriant N conditions. The use of amino acids as a source of one-carbon units, a pathway that involves serine hydroxymethyltransferase, represents an intriguing point of intersection between C and N metabolism.

Some cell wall proteins may also respond to N availability. po164 and pni287 encode proteins similar to hydroxyproline-rich and proline-rich proteins, respectively, which are components of both primary and secondary cell walls (Loopstra 2000). Like pot171 and pni275, pni287 and po164 were up-regulated by limiting NH4NO3 fertilization. Whereas po164 was negatively N-responsive in the girdling experiments, pni287 was also responsive to the girdling-associated wounding (data not shown) Zhu & Coleman (2001a) have demonstrated that the bspA promoter is responsive to exogenously applied gln.

Terpenoids constitute a large and complex family of C-based secondary compounds (Trapp & Croteau 2001). Lipophilic compounds such as terpenoids can influence wood quality as well as wood processing (Martínez et al. 1999). pni263 shows strong sequence similarities to terpene cyclases (synthases). pni263 was negatively N-responsive in both the NH4NO3 application and girdling studies, suggesting that N availability might potentially decrease terpenoid concentration or alter terpenoid composition in poplar. Other studies with woody plants have shown only a weak correlation between N availability and terpenoid content (e.g. Lerdau et al. 1997; Powell & Raffa 1999; Lamontagne, Margolis & Bauce 2000). However, most of these studies have surveyed foliar terpenoids rather than stem terpenoids, and have been conducted under field conditions in which several environmental variables could potentially confound the results. More detailed studies are required to determine whether N-induced changes in terpenoid biosynthetic gene expression can alter terpenoid concentration or composition.

CONCLUSIONS

The gene discovery approach that we have described in this paper has allowed us to identify candidate N-responsive genes that play potential roles in allocating C and N resources to different plant parts and in partitioning those resources into different biosynthetic pathways. The putative identities of these candidate N-responsive genes suggest that C : N interactions might play an important role in the N-response of poplar. Some of these candidate N-responsive genes are putatively involved in the biosynthesis of vegetative storage proteins, cell wall components, and terpenoids. Each of these processes has the potential to alter the cellular composition of wood, thereby impacting wood properties. Most studies to date that have examined the effect of N availability on gene expression have focused on genes involved in primary metabolism. Our results have revealed candidate N-responsive genes implicated in secondary metabolism. This study provides us with the tools and testable hypotheses to further examine mechanisms by which N can influence wood quality and quantity. Such knowledge will be vital in understanding forest tree nutrient relations in natural and managed environments.

ACKNOWLEDGMENTS

Thank you to Edith Orozco, Christina Balangue, and David Nolletti for C/N analyses, and to Natalie Zinn for technical assistance. The authors also acknowledge the assistance of Scott McClung at the University of Florida Protein Chemistry Core Laboratory for amino acid analyses. This work was funded under Cooperative Agreement Number DE-FC07–97ID13529 to J.M.D. from the Department of Energy through the DOE/American Forest and Paper Association's Agenda 2020 program. J.E.K.C. is the recipient of an NSERC postdoctoral fellowship. This is Journal Series No. R-08928 of the Florida Agricultural Experimental Station.

Received 30 April 2002; received in revised form 12 November 2002; accepted for publication 17 November 2002

Ancillary