The poplar hybrid P. simonii × P. nigra (hereafter referred to as poplar) was used for all studies. Poplar seedlings were grown for 1 month under a 16 h day/8 h night cycle at 25°C in a tissue culture vessel containing MH medium, pH 5.8, plus 2% sucrose and 0.7% agar. Subsequently, healthy poplar seedlings were transferred to a mixture of soil and vermiculite (1:1) and cultivated in a greenhouse under a 16 h day (25°C)/8 h night (22°C) cycle for 5 months. Leaves from the fourth internode to the sixteenth internode, stem chlorenchyma (Sc) and stem vascular tissue (Sv) from the sixth internode to the eighteenth internode, and root tissues were harvested from 20 independent 5-month-old poplar plants 4 h after the onset of illumination (Figure 1).

Twelve additional 5-month-old poplar plants grown in a greenhouse were sprayed with a low concentration (2.5 and 5 μM) of TSA (Sigma, St. Louis, MO, United States) for 2 days to study the correlation between histone acetylation and gene expression. All tissues were immediately frozen in liquid N₂ and completely ground using the 6870 Freezer Mill (SPEX SamplePrep, Metuchen, NJ, United States).

Free-hand cross-sections, 25–30 μm thick, of the 12th internode of fresh stems were stained with 0.1% toluidine blue (Sigma, St. Louis, MO, United States). Sections were observed under an OLYMPUS microscope (OLYMPUS, Japan) equipped with a computer-assisted digital camera MODEL ARTCAM-1400MI-WOM (ARTRAY, Japan).

Chlorophyll content was measured using a previously published method (Ni et al., 2009). 300 mg of leaf, Sc and Sv tissues from 5-month-old poplar were ground in a mortar and pestle under liquid nitrogen. The ground tissue was transferred to 15 ml Falcon tubes, 5 ml 80% acetone was added, and samples were then mixed in the dark for 30 min. Subsequently, the tubes were centrifuged at 3,000 rpm at 4°C for 15 min, and the supernatant was transferred to a new tube. These procedures were repeated twice. A spectrophotometer (MAPADA, Shanghai, China) was used to measure to the absorbance due to chlorophyll at 663 and 645 nm. Chlorophyll content was determined for at least six independent samples per tissue.

Total RNA was extracted from 100 mg of powdered leaves, Sc, Sv, and roots of poplar using the pBIOZOL Plant Total RNA Extraction Reagent according to the manufacturer’s instructions (BioFlux, Tokyo, Japan). The concentration of RNA was determined using the Nanodrop 2000 (Thermo Fisher, Waltham, MA, United States), and the sample was resolved on a 1.2% agarose gel to check the integrity of the RNA. To remove genomic DNA contamination, 1 μg of RNA was digested using the DNase I digestion Reagent Kit (Invitrogen, Carlsbad, CA, United States). Total cDNA was synthesized using the M-MLV Reverse Transcriptase Reagent Kit (Invitrogen, Carlsbad, CA, United States) and an oligo(dT) primer. A working sample of cDNA was prepared by diluting five-fold with sterile water and storing at -20°C.

The genomic sequences of poplar genes homologous to maize C₄ photosynthesis enzyme genes CA, PPDK, PCK, and PEPC were obtained from the poplar sequence database¹ (Supplemental Table S1). Full-length sequences of these genes were amplified from cDNA synthesized from poplar leaf or root RNA with KOD polymerase (NEB) according to the manufacture’s protocol using the primers listed in Supplemental Table S2. PCR products were purified with the TIANgel mini purification kit (TIAN GEN, Beijing, China), cloned into the pEasy-blunt cloning vector (Trans Gene, Beijing, China), and sequenced. The upstream sequences (about 2 kb in length) corresponding to the promoters of these genes were amplified from poplar genomic DNA using the primers listed in Supplemental Table S3.

Quantitative real-time polymerase chain reaction is a reliable technique for quantifying gene expression, and requires stable reference genes for data normalization (Xiao et al., 2015). However, no single reference gene has stable expression under all experimental conditions (Podevin et al., 2012). Therefore, it is necessary to choose a suitable reference gene which is expressed at a relatively stable level across the conditions being tested.

In our study, five housekeeping genes, ACTIN2, TUBLIN, UBIQUITIN (UBQ), Elongation Factor 1a (EF1a), and 18S ribosomal RNA (18S rRNA) were tested for suitability as reference genes for gene expression studies in four vegetative tissues, including leaves, Sc, Sv, and roots, as well as in TSA-treated P. simonii × P. nigra. These gene sequences were obtained from the Populus trichocarpa genome annotation v3.0². QRT-PCR was performed using the primers listed in Supplemental Table S4 to evaluate the expression variation of these candidates across tissues. The average cycle threshold (Cq) values of candidate genes ranged from 10.68 to 22.76 in the four tissues (Supplemental Figure S2A). Except for 18S rRNA, all genes showed low variability in Cq value between leaves, Sc, Sv, and roots.

The average expression stability value (M-value) is a parameter used by the geNorm software program to identify the best reference genes. The lower the M-value is, the more stable the gene expression (Bustin et al., 2009). We also used geNorm software to select the most stable reference genes (Dong et al., 2016). Based on geNorm analysis, ACTIN2 and EF1a were the most stable genes among the five candidate reference genes (Supplemental Figure S2B). In a previous study, PtACTIN2 was used as reference gene for diverse tissues of 1-year-old P. trichocarpa, including differentiated or mature xylem and phloem (Yu et al., 2013). Therefore, we chose to use ACTIN2 as a reference gene in our study.

We carried out ChIP as previously described (Bowler et al., 2004; Gendrel et al., 2005; Li et al., 2014) with the following modifications: 1.5 g of leaves, Sc, Sv, and roots from 5-month-old poplar plants grown in a greenhouse were harvested separately and cross-linked with 1% (wt/wol) formaldehyde for 15 min at 4°C. The purified chromatin was sheared to 0.3–0.7 kb fragments by sonicating under cooling for 4 min (4 s-on, 10 s-off) at 15% amplitude using a Vibra-cell VCX-505 sonicator (Sonics, Newtown, CT, United States). The sheared chromatin was diluted with ChIP dilution buffer and pre-cleared with 100 μl of protein A agarose (Millipore, Billerica, MA, United States) (Bowler et al., 2004; Gendrel et al., 2005).

For each sample, 1 ml of diluted pre-cleared chromatin was used for immunoprecipitation, and a 10 μl aliquot was used to quantify the amount of input. To immunoprecipitate chromatin containing histone modifications, 100 μl of protein A agarose and 5 μl anti-acetyl H3K9 (07-352, Millipore, Billerica, MA, United States) or 5 μl anti-acetyl H4K5 (07-327, Millipore, Billerica, MA, United States) were added to the diluted pre-cleared chromatin. Inmunocomplexes were washed with buffer in the following order: salt buffer, LiCl buffer, and TE buffer (Li et al., 2014). After discarding TE buffer, the immunoprecipitated chromatin was eluted with elution buffer (Li et al., 2014). After formaldehyde cross-linking, ChIP-DNA was purified by phenol/chloroform/isoamyl alcohol extraction followed by ethanol precipitation (Li et al., 2014).

Quantitative real-time polymerase chain reaction was performed to profile the expression patterns of reference genes, poplar homologs of C₄ photosynthetic enzyme genes and the changes in expression of these genes in TSA-treated poplar. Specific primers were designed using the online Integrated DNA Technologies software³ and are listed in Supplemental Tables S4, S5. QRT-PCR reactions were performed using SYBR Premix Ex Taq (Takara, Shiga, Japan) on a LightCycler 480 system (Roche, Basel, Sweden). Two microliters of diluted cDNA sample was used as template in an amplification reaction volume of 10 μl. The amplification program consisted of 30 s of initial denaturation at 95°C, followed by 40 cycles of 10 s at 95°C, 20 s at 60°C and 20 s at 72°C, and ended with a final extension step at 72°C for 20 s. Three replicated reactions per sample were done. For the analysis of the expression of poplar homologs of C₄ photosynthetic enzyme genes, all samples were normalized to the reference gene PsnACTIN2. The final relative expression was calculated using the formula F = 2^-ΔΔCt (Chen et al., 2014).

For ChIP assays, 2 μl ChIP-DNA sample was used as template in an amplification reaction volume of 10 μl. Primer positions and sequences are listed in Supplemental Figures S3, S4 and Table S6. The amplification program consisted of 30 s of initial denaturation at 95°C, followed by 45 cycles of 10 s at 95°C, 20 s at 60°C and 20 s at 72°C. Relative enrichment of H3K9ac and H4K5ac in the promoters of the PsnCA, PsnPPDK, PsnPCK, and PsnPEPC genes was calculated by normalizing to the value of these marks in the promoter of the PsnACTIN2 housekeeping gene. Relative enrichment values are based on the average of three PCR reactions for each sample.

Total proteins from leaves, Sc, Sv, and roots of control and TSA-treated poplar were extracted using a phenol (Sigma, St. Louis, MO, United States) extraction procedure as described previously (Hurkman and Tanaka, 1986; Wang et al., 2011). Protein concentrations were measured using the Bio-Rad Protein Assay (Bio-Rad, United States). Thirty micrograms of protein was separated by 15% SDS-PAGE gel and transferred by electroblotting (180 mA for 3 h) to a PVDF membrane (Millipore, Billerica, MA, United States). Membranes were blocked with 5% skim milk and probed with anti-H3K9ac (07-352, Millipore, Billerica, MA, United States) and anti-H4K5ac (07-327, Millipore, Billerica, MA, United States) antibodies to detect histone marks. Polyclonal anti-actin (EASYBIO, Beijing, China) antibody was used as a control for protein loading. Fluorescence goat anti-rabbit (GAR) antibody (Odyssey, United States) was used as the secondary antibody. Membranes were digitized using a two-photo far infrared scanner (Odyssey, United States) and analyzed with Image Studio software⁴. Western blotting experiments were repeated three times for each protein sample.

The P. simonii × P. nigra genome has not been sequenced. Therefore, we used the known protein sequences of maize C₄-CA (GRMZM2G121878), C₄-PPDK (GRMZM2G306345), C₄-PCK (GRMZM2G001696), and C₄-PEPC (GRMZM2G083841) as queries to blast against the P. trichocarpa genome database¹. Using primers specific to the P. trichocarpa sequences and cDNA from P. simonii × P. nigra leaves or roots, we cloned nine P. simonii × P. nigra genes homologous to the maize C₄-CA, C₄-PPDK, C₄-PCK, and C₄-PEPC genes (Supplemental Table S1). Sequence alignments showed that the PsnCA1 and PsnCA2 proteins shares 91.6% identity, and both are different from PsnCA3. Compared with PsnCA3, the N-terminal ends of both the PsnCA1 and PsnCA2 proteins are shorter by 76 amino acids and both C-terminal ends are 8 amino acids shorter (Supplemental Figure S1A). We cloned two transcripts (named PsnPPDK1-1 and PsnPPDK1-2) for a single PsnPPDK gene, and the protein sequences share 88.08% identity (Supplemental Figure S1B). Note that these transcripts are transcribed from different promoters; therefore, for the purposes of our analysis, we treat them as independently transcribed genes. There are two PsnPCK genes, the protein sequences of which share 82.76% identity (Supplemental Figure S1C). In addition, we also identified two PsnPEPC genes, the proteins sequences of which share 94.32% identity (Supplemental Figure S1D).

To better understand the potential functions of poplar genes homologous to C₄ photosynthetic enzyme genes, we profiled their expression patterns in leaves, Sc (includes epidermis and cortex), Sv (includes phloem fiber, phloem, cambium and xylem), and roots (Figures 1, 2). Based on the average cycle threshold (Cq) value and the average expression stability value (M-value), we chose to use ACTIN2 as a reference gene in this study (Supplemental Figure S2). Among the CA genes, both PsnCA1 and PsnCA2 were highly expressed in roots and Sc (Figures 2A,B). However, PsnCA3 was most highly expressed in leaves (Figure 2C). Surprisingly, the level of PsnCA3 transcript was nearly 15,000-fold higher than PsnCA1 and PsnCA2 in leaves (Figure 2D). The PPDK and PCK gene pairs showed similar differences in expression pattern. For example, both PsnPPDK1-1 and PsnPCK1 were highly expressed in leaves, whereas PsnPPDK1-2 and PsnPCK2 were highly expressed in Sc and Sv (Figures 2E–J). Unlike the CA, PPDK, and PCK genes, neither PEPC gene was most highly expressed in leaves. The abundance of both transcripts was high in Sc, but the expression level of PsnPEPC1 in roots was much higher than PsnPEPC2(Figures 2K–M).

Based on the presence or absence of tissue chlorophyll (Figure 1 and Supplemental Figure S3) and gene-specific expression patterns (Figure 2), we generally classified these nine genes into two clusters. Genes in one cluster, including PsnCA3, PsnPPDK1-1, PsnPPDK1-2, PsnPCK1, PsnPCK2, and PsnPEPC2, are highly expressed in photosynthetic tissues (chlorophyll content in the leaves and Sc is 1.42 and 0.32 mg/g, respectively). Genes in the second cluster, including PsnCA1, PsnCA2, and PsnPEPC1, are highly expressed in non-photosynthetic tissues (chlorophyll content in Sv and roots is 0.002 and 0 mg/g).

Previous studies have found that for many plant loci, histone modification within the promoter and gene body is involved in the tissue-specific regulation of gene expression (Cai et al., 2003; Heintzman et al., 2009; Zhou et al., 2010; Ong and Corces, 2011; Heimann et al., 2013), and histone acetylation has been extensively correlated with transcriptional activation (Wolffe and Hayes, 1999; Turner, 2000). Therefore, we asked whether histone acetylation modification is correlated with the tissue-dependent expression of the PsnCA, PsnPPDK, PsnPCK, and PsnPEPC genes. We analyzed the levels of H3K9ac and H4K5ac marks in the promoters of these genes using a ChIP assay. Promoter sequences upstream of the transcription initiation site (TIS) of these genes (∼2 kb) were divided into three regions, including the distal region (P1), middle region (P2) and proximal region (P3) (Supplemental Figure S4). Immunoprecipitation efficiency of the PsnACTIN2 gene promoter was used to correct for variation in the amount of chromatin prepared from leaves, Sc, Sv, and roots (Supplemental Figure S5).

As shown in Figure 3, for almost genes, the level of H3K9ac in the P3 region (close to the TIS) was higher than in the P1 and P2 regions. In photosynthetic tissues, the highest enrichment of H3K9ac in the P3 region of PsnCA3, PsnPPDK1-1, and PsnPCK1 was detected in leaves (Figures 3A–C). This correlates well with the greater transcript abundance of these genes in leaves (Figures 2C,E,H). In Sc, relatively high enrichment of H3K9ac was observed in the promoters of the PsnPPDK1-2, PsnPCK2, and PsnPEPC2 genes, which are highly expressed in this tissue (Figures 3D–F). In the non-photosynthetic root tissue, we found that H3K9ac was strongly enriched in the promoters of PsnCA1, PsnCA2, and PsnPEPC1 which correlates with the high level of expression of these genes in roots (Figures 3G–I). We also found enrichment of H3K9ac in the P1 and P2 regions of these genes, but there was no obvious regular changes in both two regions between tissues (Figures 3A–I).

As shown in Figure 4, we found high levels of H4K5ac in the P3 region of almost target genes, which is similar to the levels of H3K9ac in that region. In terms of H4K5ac modification, we also found that strong enrichment in the P3 region of most genes was correlated with high transcript accumulation. For example, H4K5ac was highly enriched in the P3 region of the PsnCA3, PsnPPDK1-1, and PsnPCK1 genes in leaves (Figures 4A–C), and in the PsnCA1, PsnCA2, and PsnPEPC1 genes in roots (Figures 4G–I). For the PsnPCK2 and PsnPEPC2 genes, which are highly expressed in the photosynthetic tissue Sc, strong enrichment of H4K5ac in the P3 region was observed in both Sc and Sv (Figures 4E,F). However, we found the highest enrichment of H4K5ac in the P1 region of the PsnPPDK1-2 gene in Sc, not in the P3 region (Figure 4D).

The correlation between H3K9ac and H4K5ac levels and expression levels of the PsnCA, PsnPPDK, PsnPCK, and PsnPEPC genes indicates that histone acetylation modification may regulate the tissue-dependent expression of genes homologous to C₄ photosynthetic enzyme genes in C₃ woody plants.

Treatment with the HDAC inhibitor, TSA, results in the accumulation of acetylated histones in the genome (Bernstein et al., 2000). In order to further study the relationship between histone acetylation and tissue-dependent expression of the PsnCA, PsnPPDK, PsnPCK, and PsnPEPC genes, we applied low concentrations (2.5 and 5 μM) of TSA to 5-month-old poplar plants and measured levels of histone acetylation and gene expression in leaves, Sc, Sv and roots (Supplemental Figures S6, S7). As we expected, western blot analysis showed that application of TSA induced a slight increase in both H3K9ac and H4K5ac in leaves, Sc, Sv, and roots (Figure 5A and Supplemental Figure S7A). Moreover, qRT-PCR results showed that TSA altered the level of gene expression to different extents in different tissues (Supplemental Figures S7B–J). In general, TSA significantly increased the abundance of PsnCA3 and PsnPPDK1-1 transcripts in leaves (Figures 5B,C), PsnPPDK1-2, PsnPCK2, and PsnPEPC2 transcripts in Sc (Figures 5E–G), and PsnCA1, PsnCA2, and PsnPEPC1 transcripts in roots (Figures 5H–J) compared to untreated tissues. However, application of TSA significantly reduced the transcript level of PsnPCK1 in leaves (Figure 5D), suggesting that TSA probably has pleiotropic effects on other regulatory networks linked to this gene.

In order to further confirm that H3K9ac and H4K5ac regulate the tissue-dependent expression patterns of the PsnCA, PsnPPDK, PsnPCK, and PsnPEPC genes, we performed ChIP assays to detect the enrichment of H3K9ac and H4K5ac in the promoters of these genes in TSA-treated poplar. The accumulation of H3K9ac and H4K5ac was altered by TSA to different extents in the P3 regions of almost all genes detected in different tissues (Supplemental Figures S8, S9). In general, the enrichment of both H3K9ac and H4K5ac in the P3 regions of PsnCA3 in leaves, PsnPPDK1-2, PsnPCK2 and PsnPEPC2 in Sc, as well as PsnCA2 and PsnPEPC1 in roots was enhanced to different degrees (Figures 6A,D,E,F,H,I). Additionally, the enrichment of H3K9ac in the P3 regions of PsnPPDK1-1 in leaves and PsnCA1 in roots was significantly induced by TSA (Figures 6B,G). Moreover, the levels of H3K9ac and H4K5ac were in agreement with the transcript levels observed for these genes under TSA treatment (Figure 5). In contrast, no obvious change in H3K9ac or H4K5ac level was detected for PsnPCK1 (Figure 6C), indicating that in leaves H3K9ac and H4K5ac do not play important roles in the regulation of PsnPCK1 gene expression. Taken together, these results reveal that histone acetylation modification is correlated with the tissue-dependent expression of most poplars homologs of C₄ photosynthesis genes.