Transcriptome analysis of a novel maize bsd C4 mutant using RNA-seq

ABSTRACT

The C4 plants like maize are at an advantage because they exhibit higher carbon conversion efficiency than C3 during photosynthesis. Using the evaluation of photosynthetic phenotypes and subcellular structure, and high-quality transcriptome analysis for four types of leaves from a bundle sheath defective maize mutant (bsd) and 501 wild line, the key target genes, important transcription factors, and specific pathways were obtained, which may regulate the C-concentrating mechanisms and antioxidant protection of the photosynthetic system, gibberellin signaling, ribosome editing, glycolysis, and chlorophyll biosynthesis. Based on these target genes, a novel network with photosynthetic transformation efficiency with oxidative decarboxylation and ribosome regulation was filtered innovatively by Cytoscape, which adds to our understanding for high-efficiency C-fixation and its genetic improvement in C3 and C4 plants.

Graphical abstract

The cover illustration

Based on the comparative analysis for photosynthetic phenotypes and genes expression of leaves from maize 501 and bsd, the target genes, transcription factors, pathways, and an innovative model for high-photosynthetic transformation efficiency were filtered, which add to understanding for high-efficiency C-fixation in C3 and C4 plants.

Introduction

As the area of available arable land decreases and the human population increases, much interest is being focused on crop plants exhibiting C4 carbon fixation as a potential for a second green revolution. The reason being that the carbon conversion efficiency of C4 plants during photosynthesis is superior to that of C3 plants.¹ Although C4 photosynthesis has been known about for quite some time, knowledge about gene regulation and expression for this trait remains incomplete. In this study, we explored the typical photosynthetic metabolic patterns for an important C4 crop plant, maize.

The photosynthetic efficiency of C4 plants originates from CO₂ concentration in the ribulose 1,5-bisphosphate carboxylase (Rubisco) via a biochemical carbon-concentrating mechanism distributed between mesophyll cells (MC) and bundle sheath cells (BSC).² In maize, BSC and MC have different photosynthetic enzyme systems and relatively large differences in chloroplast structure, development and function. And both employ phosphoenolpyruvate carboxylase (PEPC) to incorporate bicarbonate into phosphoenolpyruvate (PEP), yielding the four-carbon molecule oxaloacetate (OAA). Next, in Rubisco, OAA is converted to either malate by reduction or aspartate with transamination.³ Meanwhile, decarboxylation of the now organic carbon to release the CO₂ in Rubisco, either (1) by NADP-dependent malic enzyme (ME) that decarboxylates malate to pyruvate in chloroplasts, or (2) by NAD-ME decarboxylation of malate to pyruvate in mitochondria and phosphoenolpyruvate carboxykinase (PEP-CK) decarboxylation of OAA to PEP in the cytosol. The resulting C3 acid is then transported back to the site of PEPC as PEP in the case of PEP-CK-based decarboxylation, or as pyruvate or alanine for NAD-ME and NADP-ME. In the chloroplasts, pyruvate is recycled to PEP by the action of pyruvate phosphate dikinase (PPDK) with the reaction products pyrophosphate and AMP recycled by pyrophosphorylase (PPase) and AMP kinase.⁴ The C4 pathway requires NADP-ME and ribulose bisphosphate carboxylase (RuBPCase) activity in the BSC, and NADP-malate dehydrogenase (NADP-MDH), PEPC and PPDK function in the MC.⁵

In maize, two layers of BSCs surround the vascular bundle (as the Kranz anatomy), with the large inner BSCs having increased organelle number and lacking grana; this contrasts with C3 BSCs whose chloroplasts have fully developed grana.⁶ As a consequence of linear electron transfer in the bundle sheath chloroplasts, oxygen is produced and leading to higher photorespiration.⁷ Next-generation sequencing technologies (RNA-seq) have advanced the study of the photosynthetic mechanism in C4 plants. Comparative transcriptomic analysis of five developmental stages in Kranz and non-Kranz leaves of maize has identified that the transcription factors SHR/SCR are preferentially expressed in BSCs and control BSC fate and function, and that the auxin response factors ARF2 and Golden2-like2 (GLK2) function to establish and maintain the C4 leaf in Flaveria species.⁸ In addition, genes for cell wall-associated kinase, several transporters in the nodulin family of proteins involved in C4 metabolites and BSC wall development have been identified.⁹ These efforts would be aided by an improved understanding of the regulatory networks underlying C4 and C3 photosynthesis, of which still little is currently known.¹⁰

In this study, a novel bundle sheath defective maize mutant (bsd) and 501 wild line were used¹¹ to transcriptome analysis, which can add to our understanding of genes expression and photosynthetic metabolic pathways in the chloroplast differentiation within the BSC between C3 and C4 plants.

Materials and methods

Plant materials and phenotypic evaluation

A bsd with the line 501 genetic background was used to developing an F₂ by crossing with wild line 501 (2013–2015, Baoding, 38.87°N, 115.47°E). Among, bsd was acquired from line 501 embryogenic callus culture. The phenotype of bsd leaves is stable by the verification in a population at several environments. And its novel locus was mapped between umc1603 and umc1395.¹¹ An F₂ consisting ~210 plants (38.87°N, 115.47°E; Baoding, BD, 2016) were planted by a randomized block design with ~15 plants in a row (5.0 m x 0.6 m). Growth environment (temperature, humidity, etc.) and management conditions (N – 40 kg/667 m², 2–3 times of irrigation, etc.) are consistent with others common maize at the fixed experimental site of Hebei Agricultural University. In the F₂, bsd with pale green leaves (or mosaic leaves of pale green and wild leaves) and 501 individuals by SSR identification were isolated. In the stage of maize bellmouth, four types of leaves were cut to the evaluation of photosynthetic phenotypes and subcellular structure, and RNA sequencing.

The chlorophyll content in leaves was determined spectrophotometrically in acetone extracts. The leaves (1.0 g) from bsd and 501 were selected. The solution containing 80% acetone and 100% ethanol (v/v = 1:1) with 50 mL final volume at room temperature was performed to the extraction of the chlorophyll as previously described.¹² The content of chlorophyll a (chl__a) and b (chl__b) was assessed using a DU-800 spectrophotometer (Beckman, USA), and were calculated:

chl__{a or b} = (n₁ × OD_665 nm – n₂ × OD_649 nm) × v/(1000 × w), where n₁ = 13.9 and n₂ = 6.9 for chl__a, n₁ = 24.9 and n₂ = 7.3 for chl__b, chl__total = chl__a + chl__b; OD, v and w are optical density value, total volume and fresh weight of extract, respectively.¹² For leaves from bsd and 501 chlorophyll fluorescence was measured by an integrating fluorescence fluorometer (LI-6400, Lincoln, USA). After dark-adaptation of samples for 40 min, Fs (fluorescence instable state), Fm′(maximum fluorescence under light), Fm (maximum fluorescence in darkness), Fo′ (minimum fluorescence under light) and Fo (minimum fluorescence in darkness) were measured. The maximal photochemical efficiency and excitation energy capture efficiency of PSII were calculated by Fv/Fm = (Fm − Fo)/Fm and Fv′/Fm′ = (Fm′ − Fo′)/Fm′, respectively.¹³ No less than 10 leaves for all each variety with three biological repeats were analyzed.

The leaves of 501 and bsd were selected for transmission electron microscopy. The leaf pieces (2 × 4 mm) were transferred into 5% (v/v) glutaraldehyde solution overnight at 4°C. After rinsing for soaked leaf pieces with 0.1 M PBS buffer (pH 7.2), they were fixed with 1% (w/v) osmium tetroxide in 100 mM PBS buffer at 4°C. The samples were rinsed three times with PBS buffer and once with ddH₂O, then dehydrated through 30% (v/v) and 50% (v/v) ethanol for 15 min. Samples were then treated with saturated uranyl acetate in 70% (v/v) ethanol overnight. The samples were rinsed with 70% (v/v) ethanol, then dehydrated in 90% (v/v) and 100% ethanol for 15 min each. Samples were then treated in sequence with acetone:ethanol (1:3, v/v; 1:1, v/v; 3:1, v/v) and 100% acetone for 20 min each. Next, samples were then infiltrated with a graded acetone/Epon/Spurr’s epoxy resin and polymerized at 70°C for 24 h, and the polymerized blocks were sectioned on an LKB-V ultramicrotome (Leica, Germany). Samples for transmission electron microscopy were sectioned at 70 nm thickness using a 35° ultra diatome diamond knife, the thin sections were collected on 200 mesh copper thin bar grids, and viewed in an H600 transmission electron microscope (Hitachi, Japan).

RNA extraction, RNA-seq library sequencing, reads mapping and genes annotation

Total RNA was isolated from 1.0 g of leaves from each sample using a Tiangen reagent kit (Tiangen, Beijing, China). And total RNA with high quality was used to construct an RNA-seq library. The synthetic cDNA with the principle of specific binding of oligo(dt)-rich magnetic beads to the poly(A) region of the mRNA was ligated to Illumina paired-end Solexa adapters and sequencing through an Illumina HiSeq 2500 (Illumina, USA); The sequencing raw reads were processed by the sliding scale window method. Reads with low quality (≤ Q20) or short fragments (≤35 bp) were cleared and ambiguous base N was removed. A quality assessment on data with ≥0.80 query coverage and ≤10⁻¹⁰ (e-value) was collected by FastQC v0.10.1 (http://www.bioinfor matics.babraham.ac.uk/). The high-quality data were then mapped to the maize B73 RefGen_v4 using bowtie 2–2.1.0 (http://ftp.maizesequence.org/). Using the software Cufflinks, normalized genes expression levels with unit as reads per kilobase of exon model per million mapped reads (RPKM) were determined.¹⁴

The reference genes were submitted and aligned to the KEGG Automated Annotation Server (http://www.genome.jp/). Second, the obtained data K number of the KEGG orthology group with the highest score was used for corresponding to the relevant pathways. Finally, pathways were linked to graphical pathway maps and the annotated query genes on the maps. The transcript abundance differences between the samples were calculated based on the ratio of the RPKM. For differentially expressed transcripts, genes set enrichment analysis of the KEGG and gene ontology (GO) database was used for all the annotated query transcripts. The hypergeometric test with the threshold of the p value in branches of KEGG or GO classification and then the false discovery rate (FDR) was used to computer for significance of difference at 0.01 level.¹⁵

Differentially expressed transcript/gene (DET/Gene) analysis

A total of the transcripts for each sample were calculated through pairwise comparison of FPKM with p < .01, which were used to filters or constructing for DETs/Genes (│log₂^ratio │≥1), GO and KEGG using the DESeq R package (v1.10.1).¹⁶ The DETs/Genes were also analyzed by heatmap clustering using Origin Pro 2018 (https://www.calemeam.com/). In our study, DETs/Genes with a fold change of FPKM (│log₂^ratio │≥1 or ≥3) in comparison between either two samples were identified.

Verification of expression level by qRT-PCR

To verify the expression level of the DETs/Genes in the leaves, a 100-μL volume was mixed to synthesize 1^st-strand cDNA using a cDNA Archive Kit. The expression levels were detected by an SYBR Green quantitative real-time polymerase chain reaction (qRT-PCR) kit with three biological replicates in an ABI Prism 7000 Sequence Detection System (Life Technologies, USA), and the an average of the replicates was calculated to the Ct with the endogenous control as actin (NM_001155179). Relative expression values were calculated by employing the ΔΔCt method. And normalization of the expression levels (RPKM) was calculated by ∣log₂^RPKM∣.¹⁷ The significance of differences between groups was assessed by GraphPad Prism 8 (https://www.graphpad.com/) with one-way analysis of variance. The primers were listed in Supplementary Table S2. The synthetic primers and analytical kit were obtained from TaKaRa Biotech (Dalian, China).

Network construction for target DET/Genes

To construct a network of target DETs/Genes with upstream and downstream relationships, Cytoscape software, STRING database (https://string-db.org/) were used. They can be visualized to integrating for biomolecular interaction networks with high-throughput expression data and other molecular states. In our study, we screened target DETs/Genes with most significant difference levels to construction of the network.

Results and discussion

Evaluation of the bsd mutant

The markers umc1603 and umc1395 were used on the F₂ to detect bsd (pale green leaves or mosaic leaves) and wild-type individuals, which can be observed from seedling stage onward (Figure 1a). And contents of chl__a, chl__b and chl__total in seedling leaves are significant difference between 501 and bsd (p < .01; Figure 1a); To assess the cause of the reduced pigment in bsd leaves, we measured the chloroplast numbers in MC and BSC, which showed no significant difference in MC (p > .05) and a significant difference in BSC (p < .01) in bsd by comparison with 501 (Figure 1a); Since chl__a and chl__b are core components of the photosystems and protein light-harvesting complexes in the photosystems, and the fluorescence parameters of the photosystems were relevant. The PSII fluorescence parameters in bsd and 501 leaves were determined. The data indicated lower maximal photochemical efficiency (Fv/Fm) and lower excitation energy capture efficiency of PSII reaction centers (Fv′/Fm′) for efficient electron transport in the bsd leaves compared with 501 leaves (Figure 1a). Taken together, these data show that bsd exhibit a lower efficiency of PSII photochemistry or electron transport than 501 line. Transmission electron microscopy was used to observe subcellular chloroplasts in leaf. In BSC of bsd and 501 the Kranz anatomy was well developed and clearly visible (Figure 1a), which is closely related to functional C4 physiology.¹⁸ Thus, the bsd is suitable for research for efficiency of photosynthesis of C3 and C4 plants and the regulation of key genes by RNA-seq (Figure 1b).

Figure 1.

The photosynthetic evaluation for 501 and bsd, and RNA-seq process. (a) 501 and bsd from an F₂; ^{**, NS} the p- value < 0.01 and p- value > 0.05 (t-test). (b) the RNA-seq analysis process.

RNA-seq profiles along maize leaves from bsd and 501

We performed RNA-seq with an average of 71.2 million 92-bp reads per segment and a total of 300.7 million reads (Supplementary Table S1). A total of 90,148 unique transcripts were well aligned to B73RefGen_v4, well distributed on per chromosome, and displaying high integrity and accuracy, and displaying close correlation in four samples with the abundance variation ranges (Figure 2a–c; Table 1). A total of 4,446 and only 1,961 differential expressed transcripts (DETs) with comparing between either two samples (p < .01) were filtered in all either samples and all common each sample, respectively (Table 1, Figure 2d). Among, 155/273 and 2/4 (up-/down-) DETs were acquired at more than twofold and more than eightfold ratio (Figure 2e, Table 2), respectively. We concluded that six DETs could be used as target genes for the study of C3 and C4 transformation efficiency; they are closely related to carbon and energy supply of photosynthesis, GA signaling, photosynthesis reduction, and chlorophyll biosynthesis previously reported.^19–21

Figure 2.

The evaluation of RNA-seq quality and screening of target transcripts. (a) the mapping in B73RefGen_v4; (b) the ranges of transcripts expression; (c) the scatter plot for transcripts; (c) the heatmap with DETs at│log₂^ratio│≥ 1; (e–h) using ratio with DETs between paired samples at FDR < 0.01.

Table 1.

The enrichment analysis of DETs, GO, and KEGG.

Type (numbers)	W1/M1	W2/M2	W1/M2	W2/M1
Total transcripts	90,148	90,148	90,148	90,148
Unique transcriptsa	34,124/55,842	33,050/56,982	32,165/57,899	33,272/56,647
Total DET (p < .01)b	4,446	4,473	5,742	5,914
Up-/down DETb	2,689/1,757	2713/1,760	3467/2,275	3,420/2,494
Up-/down- DET (|log2ratio| >1)c	927/1,034	936/1,025	1,226/966	1,403/1,084
Total transcripts in GO (p < .01)	90,148	90,148	90,148	90,148
Up-/down-transcripts in GOd	5,028/4,696	3,212/4,807	4,688/6,600	6,332/6,065
Total GO	1,757	1,757	1,757	1,757
Up-/down- GO	138/155	91/176	117/203	124/135
Up-/down- GO (FDR <0.01)e	46/27	15/31	42/71	32/53
Total transcripts in KEGG (p < .01)	39,697	39,697	39,697	39,697
Up-/down-transcripts in KEGGd	1,943/2,445	1,655/2,438	2,172/3,011	2,543/3,199
Total KEGG	300	300	300	300
Up-/down- KEGG	49/37	46/36	53/41	53/32
Up-/down- KEGG (FDR <0.01)e	33/28	31/22	45/38	43/26

^a–c with mapping in B73RefGenV4 (https://www.maizegdb.org/), by comparing between corresponding two samples with p-value < 0.01, and with |log₂^ratio| >1, respectively; ^d,e by comparing between corresponding two samples with p-value < 0.01 and by comparison of the GO or KEGG and total GO or total KEGG with FDR < 0.01, respectively.

Table 2.

The screening target transcripts and network construction.

					Log2ratio				RPKM
Type	№	Gene ID	Chra	Description	W1/M1	W2/M2	W1/M2	W2/M1	W1/W2/M2/M1	p-valueb
DETs	1	Zm00001d049957	4	Kaurene synthase 6	−5.8	−4.1	−6.7	−3.2	0.2/1.3/22.2/11.9	1.48E-02
	2	Zm00001d026377	10	Organic cation/carnitine transporter 4	−4.1	−5.2	−3.3	−6.0	3.9/1.0/37.0/64.9	4.75E-06
	3	Zm00001d013907	5	Nonspecific lipid-transfer protein (nsLTPs)	−4.7	−4.5	−3.3	−5.9	4.6/2.0/46.2/121.4	2.04E-07
	4	Zm00001d022518	7	Thioredoxin-like 1–1 chloroplastic	−4.9	−3.8	−4.5	−4.2	0.7/1.2/16.3/21.7	2.21E-03
	5	Zm00001d029031	1	Cytochrome P450 CYP714B3	3.7	5.5	5.9	3.3	18.4/13.8/0.3/1.4	1.67E-03
	6	Zm00001d001820	2	Protochlorophyllide reductase1	3.6	5.3	5.2	3.8	7.6/8.5/0.2/0.6	2.59E-02
GO:0005622 (Cellular component intracellular)	1	Zm00001d006377	2	Asparagine–tRNA ligase chloroplastic	−1.6	−1.5	−2.0	−1.1	13.8/19.7/54.6/42.3	1.99E-02
	2	Zm00001d034724	1	50 S ribosomal protein L18 (RPL18) chloroplastic	−1.8	−1.4	−1.1	−2.1	65.1/53.8/143.7/229.1	1.58E-05
	3	Zm00001d038686	6	Unknown	−1.6	−2.0	−1.9	−1.7	13.5/12.1/49.7/40.0	1.81E-03
	4	Zm00001d030593	1	G-patch domain containing protein	−1.1	−1.6	−1.1	−1.5	20.2/15.1/44.5/43.3	2.07E-02
	5	Zm00001d004052	2	60 S ribosomal protein L35	−1.3	−1.6	−1.5	−1.3	85.1/80.5/236.9/204.9	8.06E-10
	6	Zm00001d007154	2	60 S ribosomal protein L22-2	−1.1	−1.2	−1.0	−1.3	57.2/49.5/114.9/125.0	7.53E-04
	7	Zm00001d007819	2	40 S ribosomal protein S15-4	−1.2	−1.6	−1.4	−1.3	20.9/18.8/55.6/46.9	8.43E-03
	8	Zm00001d036565	6	40 S ribosomal protein S14-3	−1.1	−1.5	−1.3	−1.3	69.9/59.7/166.3/147.1	8.55E-06
	9	Zm00001d029124	1	40 S ribosomal protein S21	−1.0	−1.3	−1.1	−1.2	55.0/47.9/116.2/110.9	2.78E-04
	10	Zm00001d024571	10	60 S ribosomal protein L34	−1.1	−1.2	−1.1	−1.1	65.9/61.8/138.8/136.8	6.79E-05
	11	Zm00001d031429	1	PfkB-like carbohydrate kinase family protein	−1.1	−1.5	−1.5	−1.1	36.8/35.8/104.3/78.4	1.15E-03
	12	Zm00001d010030	8	60 S ribosomal protein L29-1	−1.1	−1.2	−1.1	−1.1	114.9/113.1/254.6/240.5	1.25E-08
	13	Zm00001d049552	4	60 S ribosomal protein L34-3	−1.0	−1.2	−1.1	−1.1	36.0/35.1/78.7/74.5	2.26E-03
	14	Zm00001d016549	5	60 S ribosomal protein L35-1	−1.0	−1.4	−1.4	−1.1	117.2/112.0/301.5/237.6	5.98E-08
	15	Zm00001d007815	2	40 S ribosomal protein S15-4	−1.1	−1.3	−1.4	−1.1	47.3/49.4/120.5/104.3	2.13E-04
GO:0055114 (oxidation-reduction process)	1	Zm00001d005279	2	Peroxidase 73	−3.4	−2.6	−3.8	−2.2	1.4/3.2/19.9/15.0	2.09E-02
	2	Zm00001d014136	5	Ent-cassadiene C2-hydroxylase	−2.1	−1.8	−1.8	−2.1	9.3/9.8/33.5/41.0	3.10E-03
	3	Zm00001d034338	1	Deoxyribodipyrimidine photolyase	−2.2	−1.4	−1.7	−2.0	12.5/15.1/39.8/59.4	8.44E-03
	4	Zm00001d029853	1	Proline oxidase	−1.5	−1.5	−1.3	−1.8	17.2/14.5/42.1/49.6	1.07E-02
	5	Zm00001d043174	3	Cytochrome P450 98A3	−1.5	−1.7	−1.6	−1.6	14.3/13.1/42.6/39.2	3.62E-03
	6	Zm00001d031970	1	NAD(P)-linked oxidoreductase superfamily protein	−1.5	−1.6	−1.5	−1.6	11.2/10.8/32.8/31.8	8.80E-03
	7	Zm00001d024633	10	Monodehydroascorbate reductase 5 (MDAR5) mitochondrial	1.4	1.4	1.5	1.3	18.6/16.7/6.6/6.9	3.80E-02
	8	Zm00001d000060	10	NAD(P)-linked oxidoreductase superfamily protein	1.1	1.5	1.3	1.4	23.5/27.5/9.6/10.7	2.14E-02
	9	Zm00001d037333	6	Ketose-bisphosphate aldolase class-II family protein	1.4	1.3	1.1	1.5	18.2/20.1/8.3/6.9	3.46E-02
	10	Zm00001d005818	2	Aldehyde dehydrogenase family 7 (ALDH7) member B4	2.2	1.4	1.9	1.7	129.7/92.8/34.9/27.9	3.21E-09
	11	Zm00001d037346	6	Chromatin structure-remodeling complex protein SYD	1.5	1.9	1.7	1.7	13.8/16.2/4.3/4.8	3.96E-02
	12	Zm00001d006193	2	Cytochrome P450 family 78 subfamily	2.6	1.7	1.5	2.8	758.3/866.5/275.8/122.1	1.30E-67
	13	Zm00001d029031	1	Cytochrome P450 CYP714B3	3.7	5.5	5.9	3.3	18.4/13.8/0.3/1.4	1.67E-03
	14	Zm00001d034986	6	Cytochrome P450 71A26	4.7	1.3	1.7	4.3	960.5/725.0/290.2/36.5	2.54E-58
ko03010 (Ribosome)	1	Zm00001d024895	10	Embryonic flower 1-like protein	−1.4	−1.3	−1.6	−1.1	10.7/13.4/32.9/28.5	9.75E-07
	2	Zm00001d004052	2	60 S ribosomal protein L35	−1.3	−1.6	−1.5	−1.3	85.1/80.5/236.9/204.9	5.98E-08
	3	Zm00001d051662	4	60 S ribosomal protein L31	−1.8	−1.9	−1.0	−2.6	12.7/7.2/26.3/43.9	6.79E-05
	4	Zm00001d024571	10	60 S ribosomal protein L34	−1.1	−1.2	−1.1	−1.1	65.9/61.8/138.8/136.8	1.58E-05
	5	Zm00001d016549	1	60 S ribosomal protein L35-1	−1.0	−1.4	−1.4	−1.1	117.2/112.0/301.5/237.6	1.15E-03
	6	Zm00001d007819	2	40 S ribosomal protein S15-4	−1.2	−1.6	−1.4	−1.3	20.9/18.8/55.6/46.9	2.78E-04
	7	Zm00001d013282	5	Ribosomal protein S27 c	−1.0	−1.7	−1.4	−1.4	89.8/70.7/232.5/185.4	2.26E-03
	8	Zm00001d031429	1	PfkB-like carbohydrate kinase family protein	−1.1	−1.5	−1.5	−1.1	36.8/35.8/104.3/78.4	1.25E-08
	9	Zm00001d029124	1	40 S ribosomal protein S21	−1.0	−1.3	−1.1	−1.2	55.0/47.9/116.2/110.9	8.43E-03
	10	Zm00001d034724	1	50 S ribosomal protein L18 chloroplastic	−1.8	−1.4	−1.1	−2.1	65.1/53.8/143.7/229.1	2.13E-04
	11	Zm00001d010030	8	60 S ribosomal protein L29-1	−1.1	−1.2	−1.1	−1.1	114.9/113.1/254.6/240.5	7.53E-04
	12	Zm00001d049552	4	60 S ribosomal protein L34-3	−1.0	−1.2	−1.1	−1.1	36.0/35.1/78.7/74.5	8.06E-10
	13	Zm00001d007815	2	40 S ribosomal protein S15-4	−1.1	−1.3	−1.4	−1.1	47.3/49.4/120.5/104.3	1.29E-02
	14	Zm00001d038686	6	Unknown	−1.6	−2.0	−1.9	−1.7	13.5/12.1/49.7/40.0	8.55E-06
	15	Zm00001d022462	7	60 S ribosomal protein L18-3	−1.1	−1.8	−1.8	−1.1	21.7/21.6/73.1/46.8	1.81E-03
ko00010(Glycolysis/Gluconeogenesis)	1	Zm00001d045326	9	ATP-dependent 6-phosphofructokinase 3	2.1	1.3	1.5	1.9	29.2/24.3/10.2/6.6	7.00E-03
	2	Zm00001d005818	2	Aldehyde dehydrogenase family 7 (ALDH7) member B4	2.2	1.4	1.9	1.7	129.7/92.8/34.9/27.9	3.21E-09
	3	Zm00001d031659	1	Pyruvate dehydrogenase2	1.5	1.7	1.6	1.6	28.0/29.4/9.4/9.9	2.03E-03
TFc	1	Zm00001d040393	3	BSD domain-containing protein (bsd2)	2.7	1.8	－	－	194.4/135.3/36.9/30.5	1.60E-03
	2	Zm00001d020834	7	bHLH-transcription factor 106 (bHLH106)	0.8	1.9	－	－	9.4/8.6/2.2/5.1	2.36E-04
Network pathway	1	Zm00001d037333	6	Ketose-bisphosphate aldolase class-II family protein	1.4	1.3	－	－	18.2/20.1/8.3/6.9	3.46E-02
	2	Zm00001d024633	10	Monodehydroascorbate reductase 5 (MDAR5, mito)	1.4	1.4	－	－	18.6/16.7/6.6/6.9	3.80E-02
	3	Zm00001d029853	1	Proline oxidase	−1.5	−1.5	－	－	17.2/14.5/42.1/49.6	1.07E-02
	4	Zm00001d031659	1	Pyruvate dehydrogenase2	1.5	1.7	－	－	28.0/29.4/9.4/9.9	2.03E-03
	5	Zm00001d034724	1	50 S ribosomal protein L18 chloroplastic	−1.8	−1.4	－	－	65.1/53.8/143.7/229.1	1.58E-05
	6	Zm00001d016549	1	60 S ribosomal protein L35-1	−1.0	−1.4	－	－	117.2/112.0/301.5/237.6	1.15E-03
	7	Zm00001d022462	7	60 S ribosomal protein L18-3	−1.1	−1.8	－	－	21.7/21.6/73.1/46.8	1.81E-03
	8	Zm00001d036565	6	40 S ribosomal protein S14-3	−1.1	−1.5	－	－	69.9/59.7/166.3/147.1	8.55E-06

^aIn B73RefGen_v4 (https://www.maizegdb.org/).

^bThe max p-value from comparison between corresponding two samples.

^cTranscription factors.

Using qRT-PCR the RNA levels of the six genes were analyzed. Results indicated that the transcription patterns were consistent with RPKM levels (Figure 3, Supplementary Table S2).

Figure 3.

The comparison of expression levels for target genes between qRT-PCR and RNA-seq. qRT-PCR on three biological replicates (n = 3); normalization for RNA-seq expression (RPKM) by│log₂^RPKM│; ^{**, NS} the p- value < 0.01 and p- value > 0.05 (t-test).

Components of architectural modifications in C3 and C4 plants

GO analysis for the 90,148 unique transcripts associated ~35% to molecular function, ~26% to biological processes, and ~14% to cellular components in total 1,757 GO terms, which were mainly involved in ribosomes and their translation, electron carrier activity, and energy metabolism (Supplementary Table S3). More than 15 and only two GO terms (GO:0005622 and GO:0055114) with comparing between all either paired samples and all common each sample (FDR < 0.01) were filtered (Figure 2f, Table 1), and we identified 30 DETs/genes at more than twofold (or less than 0.5) (Table 2), which are related to the C-concentrating mechanisms (CCM) and antioxidant protection of the photosynthetic system, GA signaling, and ribosome metabolism.

Notably, expression levels of MDAR (as Zm00001d024633) and ALDH (as Zm00001d005818) are low in bsd (Table 2), which compromises the photooxidation system causing a reduction in efficiency of C-fixation.²² Five DETs/genes were identified that regulate the efficiency of C-fixation for NAD-ME complexes and DNA damage in C4-BSC.²³Five DETs/genes were signal regulators of C3 and C4 plants, which regulate the efficiency of C-fixation by plant hormones (GA).¹⁹ In C4 plants, bicarbonate is typically captured by phosphoenolpyruvate (PEP) carboxylase in MC from where they diffuse to BSC, resulting in a higher concentration of CO₂ as a key for high-efficiency C-fixation around Rubisco.²⁴ Eleven DETs/genes as ribosomal proteins were obtained, which have been displayed for important functions to protein structure, expression, post-transcriptional regulation and translation in plant development, and regulation for subunits of photosystem I and II, ATPase and cytochrome b6 f complexes, and the Rubisco in chloroplasts.^25,26 Therefore, we suggest that DETs/genes as ribosomal proteins be used as key target genes for the study of C3 and C4 transformation efficiency.

Using the KEGG analysis, a total of 168 KEGG were constructed (Table 1). And more than 22 and two KEGG (ko03010 and ko0010) from comparing between all either paired samples, and common KEGG and total KEGG (FDR < 0.01) were filtered (Figure 2f, Table 1), respectively, which are related to ribosome and glycolysis/gluconeogenesis. And 18 DETs with ratio at more than twofold (or less than 0.5) were filtered (Table 2), which are closely related to regulation for glycolate and/or glyoxylate oxidation in chloroplasts and ribosome metabolism. Among, four DETs/genes as regulators of glyoxylate pathway were obtained (Table 2), and glyoxalate is a photorespiration product from the regulation of pyruvic acid and Rubisco, and under anaerobic conditions, fructokinase phosphate is involved in the regulation of carbohydrate catabolism.²⁷ Twelve DETs/genes as ribosomal proteins were obtained again (Table 2), indicating that ribosomal protein genes play an important role in the study of C3 and C4 transformation efficiency.

Candidate regulators of C4 photosynthesis

Transcriptional activators and repressors are likely to have crucial roles in the control of gene expression in C4 photosynthesis.^28,29 We analyzed the transcription factors (TFs) with the well-defined 2,719 maize TFs. A total of 288 TFs were filtered by comparing between all either paired samples (p < .01) (Figure 2h). And two TFs (bsd2/Zm00001d040393; bHLH106/Zm00001 d020834) were significantly different between 501 and bsd with ratio at│log₂^ratio │≥3, which could be used for the study of C3 and C4 transformation efficiency.

The constituents of regulation network for C3 and C4 plants

Using the acquired key DETs/Genes, Cytoscape and STRING (https://string-db.org/), a network with 23 nodes was visualized (Table 2; Supplementary Fig. S1), eight of which are implicated in ribosomal proteins (RP) genes and the regulation for gene expression of photorespiration (Figure 4, Table 2). And the enhanced expression of Ketose-BPA class-II family protein (BPA) and MDAR5 increased NAD(P) and pyruvate content, which the enhanced oxidative decarboxylation of L-proline (by POX and PDH2) increased photorespiration and decreased net photosynthetic yield in bsd (Figure 4). For plants photosynthesis, a mutual balance between C-fixation and antioxidant metabolites regulates net photosynthesis yield. BPA catalyzes fructose-1,6-diphosphate in the glycolysis pathway, and finally releases energy and NAD(P)H through pyruvate as an intermediate metabolite in mitochondria (EC:4.1.2.13).³⁰ MDAR5 can catalyze the transformation of NAD(P)H into NAD(P) to alleviate photosynthetic system damage in chloroplasts (EC:1.6.5.4).²² Proline oxidase exhibits pyrroline-5-carboxylate reductase activity, which catalyzes the oxidative decarboxylation of L-proline (EC:1.5.1.2). Pyruvate dehydrogenase 2 has pyruvate dehydrogenase activity, which catalyzes the oxidative decarboxylation of pyruvate (EC:1.2.4.1). Therefore, based on the function of the key DETs/Genes and their levels of expression (RPKM), we suggest that pyruvate can inhibit the oxidation of glycolate or glyoxalate to produce CO₂, and regulate photorespiration and net photosynthesis yield; Meanwhile, we identified four RP genes, which may provide further more comprehensive analysis and evidence for photorespiration and reduced of net photosynthetic yield in bsd (Figure 4), extending the understanding of C3 and C4 regulation. Among them, the RP L18 is a component of the ribosome’s large subunit and is essential for protein translation, cell division, differentiation, and selective messenger RNAs (mRNAs) translation as regulatory elements.^31,32 RPL35 and RPL18-3 also are important regulatory elements involved in translational elongation, the synthesis of proteins and binding mRNAs and translating the encoded message by selecting transfer RNA (tRNA) molecules,³³ and translation of RPS14 mRNA is strictly regulated by RNA editing.³⁴ Thus, we suggest RP genes may be candidates for providing more information on regulation in C3 and C4 plants.

Figure 4.

A novel model for the regulation of photosynthetic transformation efficiency.

Conclusion

Based on the photosynthetic phenotypes and RNA-seq of leaves from a novel maize mutant bsd with the bundle sheath defects, the target DETs/genes, important transcription factors, and pathways were filtered, which were involved in the regulation of the CCM and antioxidant protection of the photosynthetic system, GA signaling, ribosome metabolism, and glycolysis. Notably, a network for oxidative decarboxylation and ribosome regulation was constructed. These results add to our understanding for high-efficiency C-fixation in C4 plants, and more target genes for genetic improvement for high-photosynthetic efficiency in a C3 species.

Funding Statement

This work was supported by the Ministry of Science and Technology of the People's Republic of China(2017YFD0300901); National Natural Science Foundation of China(31071057).

Abbreviations

BSC

bundle sheath cells

bsd

bundle sheath defective

CCM

C-concentrating mechanism

chl_a

chlorophyll a

chl_b

chlorophyll b

DET

differential expressed transcript

FDR

false discovery rate

GA

gibberellin

GLK2

Golden2-like2

MC

mesophyll cell

qRT-PCR

quantitative real-time polymerase chain reaction

RP

ribosomal protein

RuBPCase

ribulose bisphosphate carboxylase

Data availability statement

All data included in this study are available upon request by contact with the corresponding author. RNA-seq was completed by Biomarker Technologies CO., LTD (Beijing, China) https://figshare.com/s/e3316bb9049dc23727c4.

Data deposition

The manuscript has no associated data to share or make the data underlying the study open.

Disclosure statement

The authors report no conflict of interest.

Supplementary materials

Supplemental data for this article can be accessed on the publisher’s website.