Living Colors in the Gray Mold Pathogen Botrytis cinerea: Codon-Optimized Genes Encoding Green Fluorescent Protein and mCherry, Which Exhibit Bright Fluorescence

Abstract

The green fluorescent protein (GFP) and its variants have been widely used in modern biology as reporters that allow a variety of live-cell imaging techniques. So far, GFP has rarely been used in the gray mold fungus Botrytis cinerea because of low fluorescence intensity. The codon usage of B. cinerea genes strongly deviates from that of commonly used GFP-encoding genes and reveals a lower GC content than other fungi. In this study, we report the development and use of a codon-optimized version of the B. cinerea enhanced GFP (eGFP)-encoding gene (Bcgfp) for improved expression in B. cinerea. Both the codon optimization and, to a smaller extent, the insertion of an intron resulted in higher mRNA levels and increased fluorescence. Bcgfp was used for localization of nuclei in germinating spores and for visualizing host penetration. We further demonstrate the use of promoter-Bcgfp fusions for quantitative evaluation of various toxic compounds as inducers of the atrB gene encoding an ABC-type drug efflux transporter of B. cinerea. In addition, a codon-optimized mCherry-encoding gene was constructed which yielded bright red fluorescence in B. cinerea.

INTRODUCTION

The green fluorescent protein (GFP), originally isolated from the jellyfish Aequorea victoria, has been developed into a widely used reporter system, allowing the observation of a variety of cellular and molecular events in living cells. While the original jellyfish GFP yielded only weak or no fluorescence when expressed in other organisms, several amino acid substitutions in the chromophore region have been discovered which lead to improved fluorescence yields. Most of the commonly used GFP derivatives carry the S65T substitution, which leads to a red-shifted excitation maximum and strongly increased fluorescence (5, 19). A further improvement was the use of synthetic GFP (sGFP)-encoding genes, with a codon usage adapted to the host organisms (5, 18). Probably the most widely used GFP variant is enhanced GFP (eGFP), with the substitutions F64L and S65T in the chromophore region, encoded by a gene with a codon usage optimized for expression in human cells (7, 8, 18, 56). The eGFP-encoding gene has been demonstrated to be well expressed in various eukaryotes, including fungi (30). In basidiomycetes, satisfying fluorescence has been found to require the presence of an intron in the GFP-encoding gene (2, 31).

In filamentous fungi, expression of GFP has been first reported for Ustilago maydis (48), Aspergillus nidulans (13), Cochliobolus heterostrophus (32), Magnaporthe oryzae (26), and Colletotrichum spp. (12). GFP has been used for a number of live-imaging applications with plant pathogens, such as invasion of host plant tissue (Fusarium graminearum [23] and M. oryzae [25]), the dynamics of nuclear movement and division during spore germination (M. oryzae [53], Colletotrichum gloeosporioides [35], and Fusarium oxysporum [41]), translocation of regulatory protein kinases from the cytoplasm to nuclei during development and in response to light and nutrient signals (M. oryzae [1], Colletotrichum trifolii [4], and Mycosphaerella graminicola [38]), and stage-specific expression of effector proteins in the host plants (U. maydis [11] and M. oryzae [33]).

Botrytis cinerea is a necrotrophic fungus that attacks more than 200 host plants, causing great damage to a variety of economically important fruits, vegetables, and ornamental flowers (55). In the last years, significant progress has been made in understanding the molecular mechanisms of infection and in the identification of genes that contribute to the pathogenicity of B. cinerea (52). A few reports describe the use of GFP in B. cinerea. Transformants carrying a pls1-egfp fusion were shown to express GFP fluorescence in conidia during germination and host cell penetration (17). Other papers describe expression of eGFP in B. cinerea using constructs with strong constitutive promoters, but only low or inhomogeneous fluorescence in the hyphae was documented (28, 37, 39). In our laboratory, eGFP expression constructs have also been found to result in weak fluorescence, hardly useful for live-cell imaging purposes.

The nonrandom distribution of synonymous codons, known as codon bias, has been previously shown to be correlated to gene expression. Highly expressed genes tend to contain especially high numbers of favored codons that correspond to the most highly expressed tRNAs (21, 24). In contrast, if a gene contains many rarely used codons, the availability of corresponding tRNAs might become limiting, leading to a slowing down of translation (20). This phenomenon has been confirmed by the observation that expression of heterologous proteins is sometimes strongly increased by improving the codon usage of the encoding gene (8, 15, 51).

In this report, we describe the construction of codon-optimized, intron-containing genes encoding eGFP and mCherry and show that their expression in B. cinerea leads to bright green and red fluorescence. We demonstrate the use of GFP for microscopic observation of growth in vitro and during host invasion, for labeling of nuclei, and for quantitative evaluation of induction of a promoter by a variety of natural and synthetic chemicals.

MATERIALS AND METHODS

Nucleic acid manipulations and sequence data analysis.

DNA isolation and manipulation were performed according to established protocols. Synthesis of the codon-improved versions of B. cinerea eGFP (Bcgfp1) and mCherry (Bcche1) was performed by GenScript (Piscataway, NY). Both sequences, containing a Kozak sequence for efficient translation, were cloned into in the pUC57 cloning vector via EcoRV. SmaI and EcoRI sites were included during synthesis of the genes for generation of the expression constructs (see below). The 51-bp intron placed into the sequences of Bcgfp1 and Bcche1 was derived from the second intron in the predicted B. cinerea gene XP_001553520. For synthesis of Bcgfp2, cDNA of dormant spores of the Bcgfp1-expressing strain was prepared and used as a template for PCR, using the primer pair GFP2-fw and GFP2-rev (Table 1). For constitutive expression of the fluorescent proteins, the coding region of uidA in the plasmid poliGUS-Hyg5 (where GUS is β-glucuronidase) (27) was replaced by an SmaI/EcoRI fragment of the following fragments: (i) a 730-bp egfp fragment, (ii) a 781-bp Bcgfp1 fragment (codon optimized, with intron), (iii) a 730-bp Bcgfp2 fragment (codon optimized, no intron), and (iv) a 772-bp Bcche1 fragment (codon optimized, with intron). The plasmids containing these expression constructs were linearized with KpnI and transformed as previously described (10). For transformation, the B. cinerea strain B05.HYG-3 developed by Noda et al. (36) was used. B05.HYG-3 was constructed by transforming strain B05.10 with a plasmid containing a phleomycin resistance cassette and a 5′-truncated copy of the hygromycin resistance cassette originally derived from pLOB1 (10, 36). The site of plasmid integration in the genome of B05.HYG-3 is unknown. Site-directed integration of constructs of choice can be achieved with the vector pBS.HYG-5, containing a corresponding 3′-truncated hygromycin resistance cassette, which overlaps with the 5′-truncated cassette of B05.HYG-3 (36). For expression of Bcgfp1 under the control of the atrB promoter, a fragment covering 803 bp upstream of the start codon of B. cinerea atrB was amplified with the primers AtrBPr-803-Bam and AtrBPr-803-Sma and used to replace the oliC promoter fragment of the Bcgfp1-expressing plasmid. The plasmid was linearized with KpnI and transformed into B. cinerea strain B05.HYG-3 (36), resulting in strain Bc-atrB-GFP. Hygromycin-resistant transformants were checked for complete genomic integration of the constructs by PCR using the primer pairs T7/niad_rev (2,077 bp) and niad_fw/hph-TAG (2,225 bp) (data not shown). To target eGFP to the nucleus, a 538-bp PCR fragment encoding 171 N-terminal amino acids (except for the first 28 amino acids) of Mrr1 (27) was amplified from B05.10 genomic DNA with the primer pair Mrr1-NLS-a/Mrr1-NLS-b (where NLS is nuclear localization signal). The fragment was digested with NcoI and SmaI and cloned upstream of Bcgfp1, resulting in a construct encoding an Mrr1-NLS-GFP fusion protein, which was transformed into B. cinerea strain B05.HYG-3 (10).

Table 1.

Oligonucleotides used in this study

Primer pair	Sequence 5′→3′	Application
T7	GTAATACGACTCACTATAGGGC	PCR screening
niad_rev	GCCGCAAGCTTCAGATAGATACAGGCATTGG
niad_fw	CGGCGAATTCGAGGTTTTAAGTAACTGAGAGGTG	PCR screening
hph-TAG	TCGGATCCCTATTCCTTTGCCCTCGG
GFP2_fw	CTACCCGGGACCATGGTTTCCAAGGGTGAGGAGCTTTTCACTGGCGTCGTTC	Bcgfp2 construction; probe
GFP2_rev	GCCGAATTCCTATTTGTAAAGTTCATCCATTCCC
eGFP_fw	CTACCCGGGATTATGGTGAGCAAGGGCGAGGAG	B. cinerea egfp construction; probe
eGFP_rev	GACGAATTCTTACTTGTACAGCTCGTCCATGCCG
GFP_RT_fw	GCCGAGGTAAAGTTCGAG	qRT-PCR
GFP_RT_rev	GCTTGTCGGCCATGATA
Act_RT_fw	TCTGTCTTGGGTCTTGAGAG	qRT-PCR
Act_RT_rev	GGTGCAAGAGCAGTGATTTC
EF_RT_fw	ATGCTATCGACCCTCCTTCC	qRT-PCR
EF_RT_rev	GTTGAAACCGACGTTGTCAC
AtrBPr-803-Bam	GAGGATCCGGAGCAATGCAGCAACCAAC	atrB-Bcgfp1 construction
AtrBPr-803-Sma	CGCCCGGGTGATGGCAATTGAAGTATTGATG
Mrr1-NLS-a	AAACCCGGGACCATGGCATCAGCATCAGCATCAGCA	mrr1-NLS-Bcgfp1 construction
Mrr1-NLS-b	TAACCCGGGTGTCCAAAATCCATCATGAGTGTA
atrB-RT-fw	GCACTTGTGGCGAGTATCTATC	qRT-PCR
atrB-RT-rev	TGCATCCCTCCATCCATAGC

For genomic Southern hybridization, 10 μg of total B. cinerea DNA was digested with XbaI, which cuts once upstream of the oliC promoter. As probes, a PCR product mixture of 741 bp of the egfp and Bcgfp2 coding regions was used (primers for egfp, eGFP_fw and eGFP_rev; for gfp2, GFP2_fw and GFP2_rev) (Table 1). Probe radioactive labeling and hybridization were performed as described previously (10).

Expression analysis.

For analysis of gfp expression in B. cinerea, two petri dishes (9 cm) were inoculated with 2 × 10⁶ conidia in 22 ml of Gamborg minimal medium (3g liter⁻¹ Gamborg B5 basal medium [Biochemie BV, Haarlem, Netherlands], 10 mM KH₂PO₄, 10 mM fructose, pH 5.5) without shaking for 15 h at 20°C in the dark. The hyphae were removed from the surfaces with a tissue cell scraper, centrifuged for 5 min at 4,000 rpm at 4°C, and washed with 20 ml of ice-cold water. The pellet was transferred into a mortar containing liquid nitrogen and sea sand for grinding. Fungal RNA was isolated with a Nucleo Spin RNA Plant Kit (Macherey-Nagel, Düren, Germany). One microgram of RNA was reverse transcribed into first-strand cDNA with oligo(dT) primers using a Verso cDNA Kit (Thermo Fisher Scientific, Epsom, Surrey, United Kingdom). Quantitative reverse transcription-PCR (qRT-PCR) was performed using an MyIQ Real Time PCR Cycler (Bio-Rad, Munich, Germany). Transcript levels of gfp were normalized against the expression levels of the reference genes encoding elongation factor 1α and actin and are shown as relative gfp expression (29).

For atrB expression analysis in the wild-type strain B05.10, 1 × 10⁶ conidia ml⁻¹ were incubated in an Erlenmeyer flask containing 50 ml of NY medium (20g liter⁻¹ malt extract, 2g liter⁻¹ yeast extract, pH 5.5) and shaken at 150 rpm and 20°C. After 60 min, the inducer was added at a final concentration of 20 μg ml⁻¹, and shaking continued for 30 min. Conidia treated with 2% ethanol served as controls. The spores were harvested by centrifugation for 5 min at 4,000 rpm at 4°C and washed once with ice-cold water. The spore pellet was transferred into a mortar containing liquid nitrogen. RNA preparation, cDNA synthesis, and quantitative RT-PCR were performed as described above. Transcript levels of atrB were shown relative to the levels of induced spores.

Analysis of GFP fluorescence.

For fluorescence microscopy, either an Axio Observer.A1 (equipped with a Semrock GFP-A Basic filter set, with excitation at 469 nm and emission at 525 nm, or a Semrock TXRED-A Basic-000 filter set, with excitation at 559 nm and emission at 630 nm [Zeiss, Jena, Germany]) (for experiments shown in Fig. 2A, B, and F and 5A), or an Keyence BZ-8000 compact fluorescence microscope, with BZ-Analyser software (Keyence, Osaka, Japan) (for Fig. 2C and D and 4A to F) was used. For the analysis of germinating spores, vegetative hyphae, and infection cushions, conidia were inoculated in 30-μl droplets containing Gamborg B5 basal salt mixture with 10 mM glucose and incubated for the indicated times at 20°C in the dark. The confocal microscopy image of Arabidopsis leaf cells infected with a Bcgfp1-expressing B. cinerea strain was recorded on a TCS-SP5 confocal microscope (Leica, Bensheim, Germany) as described previously (11). Leaf inoculation with B. cinerea was performed as described previously (43). For fluorescence microscopy, egfp transformant number 1 and Bcgfp1 transformant number 1 were used (see Fig. 2). With the Axio Observer microscope, fluorescence micrographs were taken using a Canon Powershot 9 (Canon Inc., Tokyo, Japan) camera, using the same settings (aperture, 4.8; ISO80), with the exposure times indicated in the legends to Fig. 2, 4, and 5. With the Keyence microscope, camera settings similar to those used with the Canon camera were employed (aperture, 4.8; ISO200), with manually set exposure times.

Fig. 2.

Fluorescence microscopy of GFP-expressing B. cinerea strains. (A) Mixture of germlings (at 6 h postinfection) of an egfp-expressing strain showing weak GFP fluorescence and strain B05.HYG-3 (white arrowheads) showing almost undetectable autofluorescence. Exposure time, 4 s. (B) Mixture of a strongly Bcgfp1-expressing strain showing bright fluorescence and a weakly fluorescent egfp-expressing strain (white arrowheads). The gray arrowhead indicates a germling which is not clearly egfp of the Bcgfp1-expressing strain. Exposure time, 4 s. (C and D) GFP fluorescence of 16-h-old vegetative mycelium forming infection cushion-like structures of Bcgfp1-expressing (C; exposure time, 1 s) and egfp-expressing (D; exposure time, 4 s) strains. (E) Stack of confocal microscopic images showing invasion of Arabidopsis leaf tissue by B. cinerea expressing Bcgfp1 (24 h postinfection). The epidermal cell initially attacked by the B. cinerea hyphae has died and shows cell wall autofluorescence. Chloroplasts of mesophyll cells show red fluorescence. (F) B. cinerea germlings (6 h postinfection) expressing Mrr1-NLS-GFP, showing nuclear GFP fluorescence. Scale bars, 20 μm.

Fig. 4.

Fluorescence microscopy of Bcche1- and Bcgfp1-expressing B. cinerea strains. (A) Bright-field picture of Bcche1-expressing germinated conidia (6 h postinfection). (B) Red fluorescence picture of panel A (exposure time, 4 s). (C to F) Mycelium of a mixture of two strains, one expressing Bcche1 and the other Bcgfp1 (24 h postinfection). The infection cushion-like structure showing both colors results from hyphal fusion between the two strains. (C) Bright-field image. (D) Red fluorescence (exposure time, 4 s). (E) Green fluorescence (exposure time, 1 s). (F) Overlay of images from panels D and E. Scale bars, 20 μm.

Fig. 5.

Quantitative evaluation of drug-induced expression of a B. cinerea strain expressing Bc-atrB-GFP. (A) Fluorescence micrographs of conidia after 2 h of incubation with different inducers (20 μg ml⁻¹). A no-inducer control (2% ethanol; —) is shown in the first frame. Exposure time, 1 s. (B) GFP fluorescence levels determined by flow cytometry, after treatment with drugs at concentrations of 0.2, 2, and 20 μg ml⁻¹. The dotted line indicates the average fluorescence of noninduced cells (33 units). Standard deviations of three independent experiments are shown. Significant differences of values from those of noninduced cells are indicated: n.s., not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001. (C) Fold atrB transcript induction after a 30-min incubation of wild-type strain B05.10 with different drugs (20 μg ml⁻¹), measured by quantitative RT-PCR. Transcript levels were normalized against the expression of actin and elongation factor 1α. Values are shown in relation to conidia treated with 2% ethanol. Data are means of two experiments with two technical duplicates each, with standard deviations. Car, carbendazim; Cyp, cyprodinil; Flu, fludioxonil; Ipr, iprodione; Azo, azoxystrobin; Tri, trifloxystrobin; Bos, boscalid; Fen, fenhexamid; Bit, bitertanol; Teb, tebuconazole; Oxp, oxpoconazole; Pro, prochloraz; Tol, tolnaftate; Cam, camalexin; Eug, eugenol; Cyc, cycloheximide.

To study induction of atrB by drugs, conidia of B. cinerea strain Bc-atrB-GFP were harvested from agar plates, washed with sterile water, suspended to a concentration of 1 × 10⁶ conidia ml⁻¹ in an Erlenmeyer flask containing 10 ml of NY, and shaken at 150 rpm and 20°C. After 60 min, the inducer was added, and shaking continued for another 2 h. The flasks were placed on ice, and the suspension was diluted 1:10 in water. To avoid aggregation of conidia, the diluted suspension was vortexed before counting was begun. An acquisition dot plot was generated for each sample to estimate the particle size and to ensure that single conidia were passing the laser beam. The relative fluorescence intensity of 10,000 spores per sample was determined, at a speed of approximately 100 conidia per second, in a FACSCalibur flow cytometer (BD Biosciences, San Jose, CA), using the BD CellQuest Pro software.

For fluorometric quantification of GFP fluorescence, conidia were harvested and washed twice with sterile water. The spore pellets were ground in a mortar containing liquid nitrogen and sea sand. The frozen powder was transferred to a 1.5-ml reaction tube, 500 μl of extraction buffer (50 mM sodium phosphate buffer, pH 7.0, 10 mM EDTA, 0.1% [wt/vol] Triton X-100, 0.1% [wt/vol] N-lauryl sarcosine, 10 mM β-mercaptoethanol) was added, and the suspension was mixed shortly by vortexing. After 5 min of centrifugation at 13,000 rpm and 4°C, the supernatant was used for total protein quantification using Bradford reagent. The emission of 200 μg of total protein extracts was measured at 508 nm (excitation, 488 nm) with a spectrofluorometer (FP6500; Jasco, Gross-Umstadt, Germany). For calculation of GFP concentrations, fluorescein (absorption maximum at 492 to 494 nm; emission maximum at 520 to 521 nm; extinction coefficient, 78,000) was used as a reference compound (46).

Bioinformatic analyses.

Codon usage and GC content of fungal genes (Table 2) were taken from the Codon Usage Database (http://www.kazusa.or.jp/codon/) for the following fungi (number of coding sequences [CDS]; number of codons): Ustilago maydis (188; 122,561), Neurospora crassa (3,953; 2,048,035), Magnaporthe oryzae (= M. grisea) (207; 109,046), Gibberella zeae (220; 102,185), Colletotrichum lagenarium (24; 17,292), Cochliobolus heterostrophus (111; 154,643), Emericella nidulans (746; 443,998), Aspergillus fumigatus (674; 347,031), Sclerotinia sclerotiorum (36; 18,291), and Botryotinia fuckeliana (= Botrytis cinerea) (149; 128,023).

Table 2.

Codon usage of fungal genes, of egfp, and of Bcgfp

Parameter	Value for indicated organism or genea
	U. maydis	N. crassa	M. oryzae	G. zeae	C. lagenarium	C. heterostrophus	E. nidulans	A. fumigatus	S. sclerotiorum	B. cinerea	egfp	Bcgfp
GC content (%)	56.5	56.1	56.3	51.1	58.2	51.7	53.2	54.2	47.6	46.6	61.5	44.8
Codon usage (%)
Ala
GCG	19.8	19.9	19.3	11.5	16.4	15.7	20.7	21.6	8.3	12.0	0.0	0.0
GCA	19.5	14.5	18.1	19.6	11.3	27.3	19.4	17.7	27.4	29.6	0.0	37.5 (3)
GCT	26.7	24.3	19.9	35.6	24.7	29.8	28.0	26.4	38.2	34.2	0.0	25 (2)
GCC	33.9	41.4	42.8	33.4	47.7	27.3	31.8	34.3	26.0	24.1	100 (8)	37.5 (3)
Arg
AGG	9.4	19.1	18.5	10.2	13.2	11.0	10.5	9.8	10.8	10.2	0.0	0.0
AGA	8.2	12.8	9.3	14.3	12.1	13.2	10.3	12.2	27.9	27.1	0.0	33.3 (2)
CGG	6.4	13.8	13.9	9.0	10.0	11.4	16.6	18.2	4.4	6.8	0.0	0.0
CGA	17.9	11.4	12.8	23.1	10.4	18.4	16.3	15.6	19.8	21.1	0.0	0.0
CGT	22.9	14.4	14.6	19.0	17.5	18.6	18.5	18.2	23.2	20.2	100 (6)	66.7 (4)
CGC	35.2	28.5	30.9	24.3	36.7	27.5	27.8	26.1	13.8	14.6	0.0	0.0
Asn
AAT	26.9	27.7	23.8	30.0	18.8	38.5	37.4	38.1	43.4	51.9	0.0	38.5 (5)
AAC	73.1	72.4	76.2	70.0	81.2	61.5	62.6	61.9	56.6	48.1	100 (13)	61.5 (8)
Asp
GAT	39.5	42.4	35.6	46.9	28.7	47.4	46.7	46.5	69.2	66.5	10.0 (2)	55.5 (10)
GAC	60.5	57.6	64.4	53.1	71.3	52.6	53.3	53.5	30.8	33.5	90 (16)	44.5 (8)
Cys
TGT	31.1	30.3	27.0	39.4	22.5	40.3	36.8	37.7	46.7	53.1	0.0	0.0
TGC	68.9	69.7	73.0	60.6	77.5	59.7	63.2	62.3	53.3	46.9	100 (2)	100 (2)
Glu
GAG	62.1	65.5	70.2	63.0	75.4	54.8	58.9	61.8	46.9	45.3	93.8 (15)	62.5 (10)
GAA	37.9	34.5	29.8	37.0	24.6	45.2	41.1	38.2	53.1	54.7	6.2 (1)	37.5 (6)
Gln
CAG	64.4	60.6	64.0	57.2	69.8	53.5	62.5	64.2	28.2	32.3	100 (8)	12.5 (1)
CAA	35.6	39.4	36.0	42.8	30.2	46.5	37.5	35.8	71.8	67.7	0.0	87.5 (7)
Gly
GGG	6.2	15.2	12.7	8.3	7.3	10.9	15.2	14.9	6.0	10.1	13.6 (3)	0.0
GGA	18.0	18.9	18.0	26.3	16.7	24.3	21.1	20.7	34.3	33.4	0.0	40.9 (9)
GGT	31.6	25.5	23.4	33.5	26.6	28.7	28.0	27.0	43.7	35.7	0.0	36.4 (8)
GGC	44.1	40.4	45.8	31.9	49.3	36.2	35.7	37.4	16.1	20.9	86.4 (19)	22.7 (5)
His
CAT	37.9	39.0	31.8	47.9	24.9	45.7	45.2	46.9	50.2	58.2	0.0	44.4 (4)
CAC	62.1	61.0	68.2	52.1	75.1	54.3	54.8	53.1	49.8	41.8	100 (9)	55.6 (5)
Ile
ATA	4.1	9.2	13.0	11.0	6.0	15.2	11.0	9.2	10.4	15.6	0.0	0.0
ATT	26.2	31.4	30.5	36.2	30.7	38.3	36.6	34.5	38.3	44.8	0.0	33.3 (4)
ATC	69.8	59.4	56.4	52.8	63.3	46.5	52.3	56.3	51.3	39.6	100 (12)	66.7 (8)
Leu
TTG	15.5	18.0	15.3	17.7	13.8	17.7	15.2	17.2	23.5	23.2	0.0	33.3 (7)
TTA	1.6	3.3	3.7	3.0	1.4	5.8	5.7	4.5	11.4	11.4	0.0	0.0
CTG	22.7	22.0	28.9	18.2	28.3	19.5	22.5	28.1	7.7	10.7	85.7 (18)	0.0
CTA	5.2	7.2	6.4	8.8	4.3	11.2	9.1	7.2	6.5	9.5	0.0	0.0
CTT	16.0	17.2	15.2	24.9	14.7	20.8	21.3	17.0	25.4	24.4	0.0	28.6 (6)
CTC	39.0	32.3	30.3	27.3	37.5	25.0	26.2	26.0	25.5	20.7	14.3 (3)	38.1 (8)
Lys
AAG	72.7	77.6	75.3	75.9	85.5	64.6	66.9	70.7	62.7	54.3	95.0 (19)	70 (14)
AAA	27.3	22.5	24.7	24.1	14.5	35.4	33.1	29.3	37.3	45.7	5.0 (1)	30 (6)
Phe
TTT	40.9	34.8	38.4	41.9	27.3	41.1	34.9	33.5	30.8	41.8	0.0	33.3 (4)
TTC	59.1	65.2	61.6	58.1	72.7	58.9	65.1	66.5	69.2	58.2	100 (12)	66.7 (8)
Pro
CCG	22.7	22.6	26.1	11.0	25.0	17.3	24.3	22.3	9.1	11.4	0.0	0.0
CCA	20.3	19.2	21.1	26.3	10.9	30.9	20.9	19.5	42.3	39.1	0.0	60 (6)
CCT	27.1	23.4	21.0	33.1	21.7	27.5	27.8	28.0	34.6	31.6	0.0	40 (4)
CCC	29.9	34.8	31.8	29.6	42.4	24.3	26.9	30.1	14.0	17.9	100 (10)	0.0
Ser
AGT	7.7	10.6	8.3	11.6	6.7	13.4	11.1	11.5	13.7	16.3	0.0	10 (1)
AGC	21.3	21.3	24.9	19.4	23.8	19.0	19.4	19.2	12.8	12.9	70.0 (7)	10 (1)
TCG	30.6	17.7	22.7	11.7	20.3	15.8	16.9	18.2	9.9	12.0	0.0	0.0
TCA	9.9	11.3	12.5	16.5	7.3	16.8	13.5	12.1	14.7	19.4	0.0	20 (2)
TCT	13.1	14.6	13.6	24.1	15.8	18.9	18.5	16.5	22.5	22.8	0.0	30 (3)
TCC	17.4	24.4	18.0	16.7	26.1	16.2	20.5	22.4	26.4	16.6	30.0 (3)	30 (3)
Thr
ACG	27.0	22.5	25.0	13.9	26.5	20.2	20.6	20.7	9.0	12.4	0.0	0.0
ACA	18.0	17.9	19.6	24.6	12.6	27.5	21.9	20.0	19.6	28.8	0.0	31.3 (5)
ACT	17.4	18.6	17.4	28.9	16.8	25.3	24.2	23.2	33.9	31.6	6.3 (1)	37.5 (6)
ACC	37.5	41.1	38.0	32.5	44.1	27.0	33.3	36.0	37.5	27.1	93.8 (15)	31.3 (5)
Tyr
TAT	25.9	32.7	27.6	33.6	18.6	38.4	37.6	38.2	45.7	49.3	9.9 (1)	45.5 (5)
TAC	74.1	67.3	72.4	66.4	81.4	61.6	62.4	61.8	54.3	50.7	90.1 (10)	54.5 (6)
Val
GTG	27.6	26.0	25.8	17.3	20.4	22.3	23.4	27.1	11.2	15.6	72.2 (13)	0.0
GTA	10.2	9.1	8.9	10.5	6.1	16.0	9.5	8.7	12.0	16.2	5.6 (1)	5.6 (1)
GTT	18.1	23.2	21.8	32.4	21.0	25.9	28.4	23.3	38.4	36.7	0.0	38.9 (7)
GTC	44.1	41.7	43.5	39.7	52.5	35.8	38.7	40.8	38.3	31.5	22.2 (4)	55.6 (10)

^aNumbers in parentheses indicate codon frequencies in eGFP-encoding genes. Bold numbers indicate codons mainly used in egfp but underrepresented in B. cinerea and S. sclerotiorum compared to other fungi. GC content and codon frequencies of fungal coding sequences were taken from http://www.kazusa.or.jp/codon/.

Nucleotide sequence accession numbers.

The DNA sequences of Bcgfp1, Bcgfp2, and Bcche1 have been submitted to the GenBank database under accession numbers HQ423138 (Bcopt-egfp1), HQ423139 (Bcopt-egfp2), and HQ423140 (Bcche1). Accession numbers of other genes reported in this paper are as follows: egfp, CVU55763 (nucleotides 613 to 1410 [4]); sgfp, EF090408; B. cinerea atrB, AJ006217; and B. cinerea mrr1, BC1G_13059.1 (http://urgi.versailles.inra.fr/index.php/urgi/Species/Botrytis/Sequences-Databases).

RESULTS

Codon optimization of gfp results in increased fluorescence levels.

In order to explain the weak expression of the commonly used eGFP-encoding gene (egfp) in B. cinerea, its codon usage was compared with the average codon usage in B. cinerea and other fungi (Table 2). The codon distribution of egfp, reported to be adapted to humans, is strongly biased, with the majority of amino acids being encoded by single codons. Almost all of these codons have either G or C as the third base. In B. cinerea, many of the codons used in egfp are used with low frequencies, for example, GAC (aspartate), CTG (leucine), CCC (proline), and GTG (valine). Both the codon usage and the GC content of coding sequences in B. cinerea and the closely related white mold fungus Sclerotinia sclerotiorum are similar to each other but significantly different from other filamentous fungi. The other fungi shown in Table 2 have higher GC contents and higher frequencies of codons used in egfp. We therefore designed an adapted version of egfp, called Bcgfp, with codon usage similar to that of B. cinerea (Table 2). The GC content of Bcgfp was 44.9%, which is similar to the average GC content (46.6%) of B. cinerea but significantly lower than the GC content of egfp (61.5%). For synthesis of Bcgfp, 152 (63.3%) of the codons in egfp were exchanged. The codon-adapted gene was synthesized in one version (Bcgfp1) containing a 51-bp intron between the sixth and seventh codons. Starting with RNA from a B. cinerea transformant carrying Bcgfp1, a second version (Bcgfp2) containing no intron was obtained by RT-PCR. To compare their performance in B. cinerea, egfp, Bcgfp1, and Bcgfp2 were cloned into an expression cassette with the Aspergillus nidulans oliC promoter and the B. cinerea niaD terminator for strong constitutive expression. For cloning, the vector pBS.HYG-5 was used, which allows directed integration of the constructs via recombination between overlapping fragments of the 5′- and 3′-truncated hygromycin phosphotransferase coding regions in the B. cinerea recipient strain B05.HYG-3 (36). After transformation, three transformants were obtained for Bcgfp1, and five transformants each were chosen for egfp and Bcgfp2 for further analysis. Complete integration of the different eGFP expression constructs in the recipient strain B05.HYG-3 was confirmed by PCR (data not shown). To estimate the number of integrations in the individual transformants, genomic Southern hybridization was performed (Fig. 1). Depending on the transformant, between one and four hybridization bands were observed. Because the chosen restriction enzyme, XbaI, cuts only at the 5′ end of the oliC promoter but not elsewhere in the egfp (Bcgfp1 and Bcgfp2) expression constructs or within the hygromycin cassette used for selection, each hybridization band probably represents one copy of the construct. The different sizes of the hybridization bands indicate different kinds of integration events. Although hygromycin-resistant transformants of strain B05.HYG-3 can be generated only by targeted recombination, additional ectopic insertion(s) of the gfp expression cassette also seem to have occurred in some transformants. Taken together, the hybridization data indicate that one (in transformants Bcgfp1-1, Bcgfp2-1, Bcgfp2-8), two, or up to four integrations (in Bcgfp2-10) of the gfp genes had occurred.

Fig. 1.

Genomic Southern hybridization of GFP-expressing transformants. Numbers below indicate different independent transformants. Ten micrograms of total DNA was digested with XbaI and hybridized with a probe mixture of egfp and Bcgfp2 DNA. DNA marker bands are indicated.

The transformants were analyzed by fluorescence microscopy. With extended illumination, the control strain B05.HYG-3 showed very weak autofluorescence while an egfp-expressing strain showed significant GFP expression (Fig. 2A). However, the fluorescence of the egfp-expressing strain was much lower than a Bcgfp1-expressing transformant which showed strong overexposure under these conditions (Fig. 2B). When exposure time was reduced to 0.3 s, the fluorescence of Bcgfp1-expressing strains remained clearly visible, while the weak fluorescence of egfp strains became undetectable (data not shown). The difference between the strains became even more pronounced after extended illumination because the fluorescence of egfp strains showed more rapid bleaching. The Bcgfp1 transformants showed overall bright fluorescence in all growth stages although fluorescence levels appeared to be more variable in 24-h-old vegetative mycelium (Fig. 2C). Bcgfp2 transformants also showed bright fluorescence but at a level somewhat lower and more variable than in Bcgfp1 transformants (data not shown). While fluorescence of hyphae of egfp strains was generally rather weak, considerable fluorescence was observed in clusters of hyphae resembling infection cushions (6) (Fig. 2D). Confocal microscopy of the infection process of a Bcgfp1 transformant on an Arabidopsis leaf demonstrated that the improved fluorescence allows detailed studies of B. cinerea host invasion (Fig. 2E). To confirm the suitability of Bcgfp1 for organelle labeling, it was fused to a fragment encoding the N-terminal 171 amino acids of the transcription factor Mrr1 (27) which contains putative nuclear localization signals (NLSs). The Mrr1-NLS-GFP-expressing transformants showed high GFP fluorescence in nuclei, which makes them useful for studies on nuclear migration and division (Fig. 2F).

To quantify the visual differences of GFP fluorescence in the different transformants, we performed fluorometric measurements of cell extracts from freshly harvested conidia (Fig. 3A). Comparison of the average fluorescence levels revealed fluorescence levels 12.1-fold higher for Bcgfp1 strains and 6.3-fold higher for Bcgfp2 strains than for egfp strains. In yeast and some filamentous fungi, increased protein levels resulting from an optimized codon usage have been found to be correlated with increased transcript levels (8, 51). We therefore compared the gfp mRNA levels in the transformants (Fig. 3B). While only low gfp transcript levels were observed in all tested egfp transformants, they were on average 9.4-fold higher in Bcgfp1 transformants. The Bcgfp2 strains without the intron showed more variable gfp expression levels, i.e., on average 5.7-fold higher than those of egfp strains.

Fig. 3.

Transcript and protein quantification of B. cinerea strains expressing Bcgfp1 and Bcgfp2 and egfp. (A) Fluorescence of GFP expressing B. cinerea strains from freshly harvested conidia. Fluorescein was used as a reference compound for GFP quantification. One fluorescence unit corresponds to 1 pmol of fluorescein ml⁻¹ mg⁻¹ of protein. (B) gfp transcript levels of GFP-expressing B. cinerea strains, measured by quantitative RT-PCR. Values are shown in relation to the expression of actin and elongation factor 1α. Data are means of two experiments with two technical duplicates each, with standard deviations. Numbers below bars indicate different independent transformants.

Expression of codon-optimized mCherry in B. cinerea.

The same strategy of codon optimization and intron insertion as for GFP was followed for the mCherry red fluorescent protein (46). The resulting codon-optimized sequence, Bcche1, was also cloned between the oliC promoter and niaD terminator and transformed into B. cinerea B05.HYG-3 (36). Transformants showed red fluorescence in conidia, germlings, and mycelia (Fig. 4). When mixtures of Bcche1 and Bcgfp1 strains were cultivated, hyphae showing simultaneous expression of red and green fluorescence were occasionally observed, probably due to hyphal fusion (anastomosis) (Fig. 4C to F). Although the red fluorescence of Bcche1 transformants was easily detectable, it appeared less bright than the green fluorescence of Bcgfp1 transformants. The average exposure times required to generate pictures of comparable intensity were several times longer for mCherry than for GFP (data not shown).

Using GFP as a reporter for quantitative comparison of inducers of the ABC transporter-encoding gene atrB.

B. cinerea contains several ATP binding cassette (ABC) transporters that serve as drug efflux pumps. One of them, AtrB, has been shown to play an important role in the detoxification of chemically diverse phytoalexins, antibiotics, and fungicides (9, 27, 49). Usually, atrB is expressed at low basal levels, but it can be strongly induced by a variety of drugs (9). Recently, the transcription factor Mrr1 was found to be essential for expression of atrB (27). In order to compare the efficiency of different drugs in inducing atrB, we fused an atrB promoter fragment to the Bcgfp1 coding region and transformed the construct into B. cinerea strain B05.HYG-3, resulting in the atrB reporter strain B. cinerea Bc-atrB-GFP. Microscopic analysis of the resulting transformants revealed only very weak GFP fluorescence in the absence of inducing drugs, while drug-induced GFP expression was easily detected (Fig. 5A). Highest levels of expression were observed in the presence of the fungicides cyprodinil and fludioxonil and of the phytoalexin camalexin, which have previously been shown to induce atrB (9, 27, 49). In order to quantify the inducing activities of various drugs, we measured the relative fluorescence of drug-treated conidia by flow cytometry. As shown in Fig. 5B, a large variability in the inducing activities of the tested compounds was observed. The strongest activities were observed for cyprodinil (22-fold), followed by camalexin. Nine of the 16 compounds tested showed significant inducing activities at concentrations of either 2 or 20 μg ml⁻¹. Some compounds, such as cycloheximide and azoxystrobin, led to reduced fluorescence accumulation at increasing concentrations, probably due to toxic effects. To confirm the validity of the results obtained with the atrB-GFP reporter strain, drug-induced changes in atrB transcript levels were measured in the wild-type strain B05.10, using similar conditions as those used for fluorescence analyses. Highest atrB transcript levels were obtained in response to cyprodinil, and intermediate levels were found after treatment with fludioxonil and fenhexamid, whereas only weak induction (3.3-fold) by oxpoconazole was observed (Fig. 5C). Therefore, the transcript and fluorescence data are in good agreement with each other.

DISCUSSION

With this work, we have demonstrated the use of green and red fluorescent proteins as valuable tools for Botrytis molecular biology. Previously, genes encoding eGFP (containing the chromophore substitutions F64L, S65T, and the silent H231L substitution) and sGFP (containing the S65T substitution), both with optimized codon usage for humans and differing by only 3 nucleotides, have been reported to be expressed successfully in several ascomycetous fungi including B. cinerea and the closely related Sclerotinia sclerotiorum (17, 30). However, as shown previously and in this paper, the original egfp gene showed rather low fluorescence in B. cinerea despite being expressed under the control of the strong oliC promoter. Nevertheless, fluorescence varied between different transformants and different growth stages. For example, the egfp transformants showed rather strong GFP fluorescence in hyphae forming infection cushion-like structures, similar to those described by Choquer et al. (6) (Fig. 2D). Analysis of the codon usage in B. cinerea genes revealed that it strongly deviates from egfp. In egfp, the GC content (61.5%) is much higher than in B. cinerea (46.6%), and most amino acids are mainly or exclusively encoded by a single codon. Furthermore, only 30 of the 61 available codons encoding amino acids are used in egfp, and many of these are rarely used in B. cinerea, which has a lower GC content than other filamentous ascomycetes except S. sclerotiorum (Table 2). In other fungi in which egfp has been successfully used, the codons used in egfp are represented at significantly higher levels than in B. cinerea (Table 2).

The key for increasing fluorescence of GFP was the adaptation of the codon usage of egfp to that of B. cinerea. Species-specific codon optimization of GFP has been described for several microorganisms, including bacteria (42), the alga Chlamydomonas reinhardtii (14), and the yeast Candida albicans (8), but not yet for filamentous fungi. To evaluate the effects of codon optimization and insertion of an intron, GFP expression constructs with identical promoter and terminator sequences were transformed into B. cinerea strain B05.HYG-3 which allows site-directed integration of the constructs. In contrast to previous data (36), we found single, double, and, in one transformant, quadruple integrations of the GFP coding sequences. However, there was no evidence of a correlation between copy number and GFP expression.

Microscopic observations revealed great differences in GFP fluorescence between egfp and Bcgfp1- and Bcgfp2-expressing strains. While the fluorescence of egfp strains was clearly documented with extended light illumination, it appeared weak relative to the bright fluorescence of Bcgfp1 strains (Fig. 2B). Indeed, exposure times could be reduced approximately 10-fold to obtain micrographs of Bcgfp1 strains with brightness levels similar to those of egfp strains and almost no background fluorescence. For quantitative comparison of gfp expression, fluorometric measurements of protein extracts were performed. On average, the fluorescence levels of B. cinerea transformants expressing the codon-optimized gene Bcgfp2 were 6.3-fold higher than transformants expressing egfp. The presence of an intron in Bcgfp1 further increased GFP fluorescence almost 2-fold, reaching levels 12.1-fold-higher than those of egfp. Therefore, in addition to the dominant effect of the codon optimization, the intron also contributed to some extent to the high fluorescence levels in Bcgfp1 strains. However, this effect was less pronounced than in basidiomycetes, in which an intron is essential for visible gfp expression (2, 31). In Neurospora crassa, codon-optimization of a gene encoding luciferase led to a dramatic increase in luminescence, but insertion of an intron did not further improve expression (15).

The increased fluorescence of the Bcgfp transformants was found to be strongly correlated with increased mRNA levels. The correlation was clearly observed for transformants expressing Bcgfp1 and egfp, while in Bcgfp2-expressing transformants transcript levels were more variable than fluorescence intensities. Analysis of more Bcgfp1 transformants might have revealed variations in gfp transcript levels similar to those in Bcgfp2 transformants. These data indicate that the effects of codon optimization are observed mainly on the level of gfp mRNA accumulation. There are numerous examples for increased protein yields resulting from codon adaptation of genes expressed in fungi, including bacterial toxins and enzymes (8, 40) and aequorin (34). In some cases, the increased protein yields were also found to be correlated with increased transcript accumulation in fungi (8, 51). Therefore, low stability of transcripts with nonadapted codon usage seems to be a major reason for their low level of expression in fungi. It is known that the stability of mRNA is affected by its translation. Transcripts can be destabilized by permanent or temporary translational arrest due to stop codons or a high content of rare codons, which leads to their degradation (3, 22). However, different mechanisms seem to be involved. In Aspergillus oryzae, poor expression of a mite allergen was found to be due to the addition of poly(A) tails to the mRNA within the coding region, leading to aberrant mRNAs that are rapidly degraded (51). Similar explanations have been provided for the poor expression of a plant α-galactosidase in Aspergillus awamori (16) and of a tetanus toxin fragment in Saccharomyces cerevisiae (40). In all of these cases, codon optimization resulted in increased GC content and the removal of the cryptic polyadenylation sites in the coding regions of the transcripts. In B. cinerea, however, codon optimization of egfp was achieved by lowering the GC content, which makes the possibility of premature polyadenylation unlikely. We therefore assume that the low egfp transcript levels in B. cinerea are a consequence of transcript destabilization in an unknown manner, which leads to low translation efficiency. The effects of codon optimization and intron insertion on gfp expression observed in our study certainly depend on the particular expression constructs used, and they might be different in other constructs. For example, in GFP fusion constructs with other genes, the effects of codon optimization and the intron might be less pronounced or even lost.

In addition to GFP, we have also constructed an optimized mCherry-encoding gene. mCherry was chosen as a versatile red fluorescent protein because it is monomeric, relatively photostable, and shows rapid folding of the chromophore (46, 47). The fluorescence levels of mCherry were found to be significantly lower than levels of GFP in transformants with similar constructs. This result is not fully explained by the fact that the brightness of mCherry is approximately 2-fold-lower than that of eGFP (47). Therefore, GFP was clearly the more sensitive reporter protein in this study. The availability of fluorescent proteins with different colors increases the experimental flexibility and allows the construction of strains with double-color reporters.

The high levels of fluorescence obtained with Bcgfp1 greatly increase the versatility of GFP as a live-imaging reporter in B. cinerea, including the detection of weakly expressed genes or proteins. This is particularly important for time-lapse imaging because extended high-power illumination leads not only to rapid fluorescence bleaching but also to the inhibition of development. We found that conidial germination of B. cinerea is especially sensitive to the blue light used for excitation of GFP (M. Leroch, unpublished observations). In order to study this and other developmental processes, both a sensitive reporter and high-quality microscopic equipment are required. Using codon optimization, the design of GFP variants for further live-imaging applications is feasible, for example, cyan fluorescent protein (CFP) and yellow fluorescent protein (YFP) for analyzing dynamic protein-protein interactions by fluorescence resonance energy transfer (57).

Taking advantage of the high sensitivity of detection of Bcgfp1 expression, we have constructed an atrB promoter-GFP reporter strain for quantitatively analyzing many toxic compounds for their inducing activity. With this reporter system, a large number of compounds could be tested in a short time, a large dynamic range of expression could be measured, and highly reproducible data were obtained. We found large variations in GFP accumulations of the reporter strain, indicating great differences in the gene-inducing activity of different compounds. The data were confirmed by quantitative measurements of atrB transcript levels in the laboratory wild-type strain B05.10, which also revealed strong and differential induction of atrB by cyprodinil, fludioxonil, and fenhexamid but not oxpoconazole. These data are similar to previous results on drug-mediated atrB induction obtained with hyphae grown overnight in shaking culture by Northern hybridization (45, 49, 54) but also revealed some differences. While induction by cyprodinil, fludioxonil, camalexin, eugenol, and tebuconazole has been previously described, we demonstrate for the first time atrB induction by fenhexamid, bitertanol, prochloraz, and (to a low extent) boscalid. On the other hand, there are also some quantitative differences between ours and previous data. For example, in contrast to the report by Vermeulen et al. (54), we found cyprodinil to be a much stronger inducer than fluodioxonil. These differences might be explained by the different experimental conditions and growth stages analyzed. Expression of atrB has been shown to be dependent on the activity of the zinc cluster transcription factor Mrr1 (27). We are investigating the molecular basis of the Mrr1-dependent induction of atrB by a variety of structurally different inducers. In Saccharomyces cerevisiae, the zinc cluster transcription factors PDR1 and PDR3, which control the expression of several drug efflux transporters, have been demonstrated to act as nuclear receptors that directly interact with the inducing drugs (50). If Mrr1 functions in a similar manner, the different atrB-inducing activities of the drugs tested in this study might reflect their different binding affinities to Mrr1. We are currently testing this hypothesis.

B. cinerea atrB has been shown to respond to a variety of toxic compounds of plant, fungal, bacterial, and synthetic origins (9, 44, 45). By using an atrB::GUS (β-glucuronidase) reporter strain, we have shown that atrB is specifically induced in germlings by the phytoalexin camalexin on the surface of Arabidopsis leaves (49). We have started to use the Bc-atrB-GFP reporter strain on different plant surfaces and have found evidence for atrB induction by different, as yet noncharacterized host (defense ?) compounds (M. Leroch, unpublished data). This strain will also be useful as a tool for studying atrB induction by toxins in the natural environment of B. cinerea, such as antibiotics released by epiphytic bacteria or fungicides in commercial fruit or vegetable fields (44).

As discussed above, the codon distribution of egfp strongly deviates from that in most organisms. The success of our approach leads us to suggest that codon optimization might be useful to improve the performance of fluorescent proteins in other fungi as well, in particular, those with low GC content.