Full Paper
Draft genome sequence of bitter gourd
(Momordica charantia), a vegetable and
medicinal plant in tropical and
subtropical regions
Naoya Urasaki
1
, Hiroki Takagi
2
, Satoshi Natsume
2
, Aiko Uemura
2
,
Naoki Taniai
1
, Norimichi Miyagi
1
, Mai Fukushima
3
, Shouta Suzuki
3
,
Kazuhiko Tarora
1
, Moritoshi Tamaki
1
, Moriaki Sakamoto
1
,
Ryohei Terauchi
2
, and Hideo Matsumura
3,
*
1
Okinawa Prefectural Agricultural Research Center, Itoman, Okinawa 901-0336, Japan,
2
Iwate Biotechnology
Research Center, Kitakami, Iwate 024-0003, Japan, and
3
Shinshu university, Ueda, Nagano 386-8567, Japan
*To whom correspondence should be addressed. Tel. þ81 268 21 5801. Fax. þ81 268 21 5810.
Email: hideoma@shinshu-u.ac.jp
Edited by Dr. Satoshi Tabata
Received 15 May 2016; Accepted 9 October 2016
Abstract
Bitter gourd (Momordica charantia) is an important vegetable and medicinal plant in tropical
and subtropical regions globally. In this study, the draft genome sequence of a monoecious bit-
ter gourd inbred line, OHB3-1, was analyzed. Through Illumina sequencing and de novo assem-
bly, scaffolds of 285.5 Mb in length were generated, corresponding to 84% of the estimated
genome size of bitter gourd (339 Mb). In this draft genome sequence, 45,859 protein-coding
gene loci were identified, and transposable elements accounted for 15.3% of the whole ge-
nome. According to synteny mapping and phylogenetic analysis of conserved genes, bitter
gourd was more related to watermelon (Citrullus lanatus) than to cucumber (Cucumis sativus)
or melon (C. melo). Using RAD-seq analysis, 1507 marker loci were genotyped in an F
2
progeny
of two bitter gourd lines, resulting in an improved linkage map, comprising 11 linkage groups.
By anchoring RAD tag markers, 255 scaffolds were assigned to the linkage map. Comparative
analysis of genome sequences and predicted genes determined that putative trypsin-inhibitor
and ribosome-inactivating genes were distinctive in the bitter gourd genome. These genes
could characterize the bitter gourd as a medicinal plant.
Key words: Cucurbitaceae, Momordica charantia, bitter gourd, draft genome, de novo sequencing
1. Introduction
Bitter gourd (Momordica charantia,2n ¼ 2x ¼ 22
1
) is a dicot vine
species belonging to the family Cucurbitaceae originating in tropical
Asia. Bitter gourd, also known as African cucumber, bitter
cucumber, bitter melon, balsam pear, or karela in the region,
2,3
is
characterized by its warty-skinned fruit and is widely cultivated in
tropical and subtropical regions of the world. The flesh of bitter
gourd fruit tastes bitter owing to the presence of the cucurbitacin-
V
C
The Author 2016. Published by Oxford University Press on behalf of Kazusa DNA Research Institute.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits
unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited.
51
Dna Research, 2017, 24(1), 51–58
doi: 10.1093/dnares/dsw047
Advance Access Publication Date: 17 December 2016
Full Paper
Downloaded from https://academic.oup.com/dnaresearch/article/24/1/51/2709597 by guest on 21 August 2022
like alkaloid, momordicine, and triterpene glycosides. Bitter gourd
fruit are rich in vitamin C and phenolic compounds with antioxidant
activity.
4–7
Additionally, leaf decoction of bitter gourd is used in tra-
ditional medicine for the treatment of stomach pain, anemia, ma-
laria, coughs, and fever.
8
Recently, several studies have shown its
antidiabetic effect in vitro and in vivo.
9–11
Therefore, these properties
have given the plant a high medicinal value and made it the subject
of recent scientific research. Similar to other Cucurbitaceae crops,
bitter gourd is a monoecious plant species. However, some gynoe-
cious lines have been found,
4
providing useful genetic resources (as
maternal plants) in breeding programs for the production of F
1
hy-
brids. Matsumura et al.
12
succeeded in genetically mapping the locus
responsible for gynoecy and identified restriction-associated DNA
tag sequencing (RAD-seq) markers linked to the locus. In Cucumis
spp., sex determination has been well studied, and ethylene has been
shown to play a key role in its regulation. Genes encoding
aminocyclopropane-1-carboxylic acid (ACC) synthase have been
shown to be responsible for gynoecy, unisexual flower development,
and andromonoecy in cucumber or melon.
13–15
Through silver ni-
trate mediated inhibition of ethylene, production of female flowers in
the gynoecious bitter gourd was obstructed in favor of bisexual flow-
ers, indicating the possible involvement of ethylene in the sex deter-
mination. However, because of the limited genome sequence
information in comparison to cucumber,
16
melon,
17
and water-
melon,
18
the genes underlying sex determination in bitter gourd are
yet to be identified.
In this study, we determined the whole genome sequence of bitter
gourd, which was generated through the Illumina next-generation se-
quencing platform followed by de novo assembly. Ab initio gene pre-
diction and annotation of predicted genes were also carried out.
Based on these assembled genome sequences and gene prediction, the
bitter gourd genome was compared with known genome sequences
of other Cucurbitaceae species. Additionally, through RAD-seq anal-
ysis, a linkage map was constructed onto which the assembled scaf-
folds were assigned. These results provided a basis for gene
identification and DNA marker development in bitter gourd, and a
platform for studying evolution in Cucurbitaceae species.
2. Materials and methods
2.1. Plant materials and DNA preparation
A monoecious inbred line OHB3-1 developed by the Okinawa
Prefectural Agricultural Research Center was used for de novo se-
quencing of the bitter gourd genome. Genomic DNA was extracted
from young leaves using a NucleoSpin Plant II kit (Macherey-Nagel)
according to the manufacturer’s instructions. For RAD-seq analysis,
two parental bitter gourd lines, OHB61-5 and OHB95-1A, and their
F
2
progeny
12
were used as materials.
2.2. Library preparation and sequencing
Sequencing libraries were prepared from genomic DNA for Illumina
MiSeq and HiSeq2500 platforms. A short insert (330 bp) paired-end
(PE) library was constructed using a TruSeq DNA PCR-Free LT
Sample Prep Kit (Illumina), which reduced PCR amplification bias in
library preparation. Mate-paired (MP) libraries with various insert
sizes (2, 4, 6, and 8 kbp) were constructed using the Nextera Mate
Pair Sample Prep Kit (Illumina). The PE library was sequenced using
MiSeq (2 230 bp) and the four MP libraries were sequenced using
HiSeq2500 (2 100 bp).
2.3. Sequence assembly
Sequence reads in fastq files from MiSeq and HiSeq2500 were
quality-filtered by FASTX-Toolkit version 0.0.13 (http://hannonlab.
cshl.edu/fastx_toolkit/). For de novo assembly, sequence reads with a
PHRED quality score of 30 comprising of 90% of the reads were
extracted. After adaptor trimming and removal of reads with inap-
propriate insert sizes in MP libraries using an in-house pipeline of
scripts, qualified reads (Supplementary Table S3) were applied to de
novo assembly using ALLPATHS-LG assembler version R49856
19
with setting PLOIDY ¼ 2 and HAPLOIDIFY ¼ True. All the con-
structed scaffold sequences were aligned with each other using
BLASTN, and perfectly identical scaffolds to others in entire se-
quences, were excluded as duplicated scaffolds. In the remaining
scaffolds showing similarity to others, when the observed frequency
of mismatch and indel sites per 1,000 bases was less than 1 in both
aligned scaffold sequences, they were presumed to be allelic.
For mitochondrial (Mt) or chloroplast (Cp) genome sequences, PE
and MP reads were aligned to 80 and 697 Mt and Cp reference ge-
nome sequences in the NCBI database (Supplementary Tables S1
and S2) using Burrows-Wheeler Aligner (BWA) version 0.6.1 with
default parameters, respectively. Mapped reads to Mt or Cp refer-
ence sequences were extracted from the original fastq files, and ap-
plied to the assembly using ALLPATHS-LG as described above.
2.4. Gene prediction and annotation
Gene prediction analysis in the bitter gourd scaffold sequences was
carried out using ab initio prediction by FGENESH software ver
3.1.1 (Softberry)
20
based on Hidden Markov Model (HMM)-based
gene prediction (Supplementary method). For annotating predicted
genes, encoded protein sequences were applied to the BLASTP search
against the non-redundant (NR) protein database in NCBI (ftp://ftp.
ncbi.nlm.nih.gov/blast/db/) and UniProtKB/Swiss-Prot database
(http://www.uniprot.org), respectively. As domain searches of
encoded proteins of predicted genes, all the amino acid sequences
were applied to InterProscan version 5.19-58.0 (https://www.ebi.ac.
uk/interpro/) with default settings (Supplementary method).
Transposable elements in the predicted genes were identified using
TransposonPSI (http://transposonpsi.sourceforge.net/), and only the
top hits against individual library searches with default settings were
employed for their annotation.
2.5. Comparative analysis of genomes among
Cucurbitaceae species
Comparison of the bitter gourd genome and other Cucurbitaceae ge-
nomes was performed by mapping OHB3-1 scaffold sequences to cu-
cumber (cucumber_ChineseLong_v2_genome, http://www.icugi.org/
cgi-bin/ICuGI/index.cgi),
16
melon (CM3.5.1_pseudomol, https://mel
onomics.net/),
17
and watermelon (WCG_chromosome_v1, http://
www.icugi.org/cgi-bin/ICuGI/index.cgi)
18
genome sequences using
SyMap 4.2 (http://www.agcol.arizona.edu/software/symap/).
21
2.6. Conserved genes among Cucurbitaceae species
and unique genes in bitter gourd genome
By comparing a list of anchors analyzed by the SyMap program,
genes showing conserved synteny among all four Cucurbitaceae spe-
cies were identified, and applied to phylogenetic analysis using
Aminosan
22
and RAxML
23
as described in Supplementary method.
To identify genes showing unique structures in the bitter gourd ge-
nome, anchor gene lists in the SyMap analysis were compared in all
52 The draft genome sequence of bitter gourd
Downloaded from https://academic.oup.com/dnaresearch/article/24/1/51/2709597 by guest on 21 August 2022
four Cucurbitaceae species. Unanchored genes to any predicted genes
in melon, cucumber, or watermelon genomes, were found. Reversely,
conserved (syntenic) genes among melon, cucumber, and watermelon
genomes, but not in the bitter gourd genome, were also found.
Functional annotation of these selected genes was determined by do-
main searches of encoded protein sequences using InterProScan as
described above.
2.7. RAD-seq analysis
RAD-seq analysis was performed as described previously.
12
Briefly,
genomic DNA was digested with AseI restriction endonuclease, and
a biotinylated adapter, harboring index sequences, was ligated to the
digested DNA fragments. The adapter ligated genomic DNA frag-
ments were then digested with NlaIII restriction endonuclease.
Biotinylated fragments were collected using streptavidin-coated mag-
netic beads (Dynabeads M270, Thermofisher), and the additional
adapter was ligated to the end of the fragments on the magnetic
beads. These adapter-ligated fragments on the beads were amplified
by PCR. The PCR products were then sequenced using the
HiSeq2500 system. From the sequence reads, 80 bp sequences in-
cluding AseI-recognition sites were extracted as RAD-seq tags. Tag
extraction and counting was carried out using CLC Genomics
Workbench software (Qiagen).
2.8. Reference mapping of RAD-seq tags
Tag sequences showing more than 20 counts in either parent line
(OHB61-5 or OHB95-1A) were employed in further analysis. These
tag sequences (80 bp) were mapped to the scaffold sequences of
OHB3-1 as ‘reference sequences’ using BWA version 0.6.1 in DDBJ
Read Annotation Pipeline (https://p.ddbj.nig.ac.jp/pipeline/).
Procedures for detection of polymorphic or heterozygous loci were
described in Supplementary method.
2.9. Linkage map development
An RAD-seq analysis of 97 F
2
plants derived from OHB61-5 and
OHB95-1A was carried out as described above. Based on analyzed
RAD-seq data in individual F
2
plants, genotypes of bi-allelic tags as
co-dominant markers were determined following a previously de-
scribed method.
12
Genotyping procedure and a linkage map con-
struction using JoinMap4.1 (Kyazma)
24
were described in
Supplmentary method.
2.10. Comparative analysis of orthologous and
paralogous genes
Homologues of genes for putative trypsin inhibitor, ribosome inacti-
vating protein, ACC synthase and CmWip1 were identified by
BLAST searches against predicted genes in melon, cucumber and wa-
termelon genome. Sequence alignment and phylogenetic analysis was
performed using MEGA7.0.18.
25
Detail of analysis was described in
Supplementary method.
2.11. RT-PCR analysis
Total RNA was extracted from flower buds of bitter gourd plant,
and expression of sex determination-related genes was analyzed by
RT-PCR as described in Supplementary method.
3. Results and discussion
3.1. Sequencing and assembly of the bitter gourd
genome
In the current study, a monoecious inbred line (OHB3-1) of bitter
gourd was used for genome sequencing using the Illumina platform.
Paired-end (PE) and mate-pair (MP; with 2, 4, 6, and 8 kbp inserts)
libraries were constructed from genomic DNA and sequenced using
the Illumina MiSeq or HiSeq2500 DNA sequencer. For PE library
development, PCR amplification was avoided and long read se-
quencing (2230 bp) was carried out. The total length of the ana-
lyzed sequence reads amounted to over 37 Gb (Supplementary
Table S3), which was equivalent to approximately 110 times that
of the estimated genome size (339 Mb) of bitter gourd,
26
represent-
ing a sufficient quantity of sequence reads for whole genome assem-
bly. Using these sequence reads, scaffolds were constructed using
the ALLPATHS-LG assembler.
19
Using BLAST analysis of the as-
sembled scaffolds of each other, six pairs of putative allelic scaf-
folds (scaffold_617 and 614, 950, and 911, 988 and 901, 699 and
700, 690 and 691, 657 and 604), which contained mismatch and/
or indel sites, were found. However, since it was difficult to dis-
criminate between allelic and paralogous sequences in this study,
they were included in the draft genome sequence data as indepen-
dent scaffolds in this study. The total length of the assembled scaf-
folds was 285.5 Mb, which comprised 1,029 scaffolds (Table 1),
corresponding to approximately 84% of the previously estimated
genome size.
26
The N50 value of these scaffolds was 1.1 Mb, and
the longest scaffold sized was over 7 Mb (Table 1). According to
previous studies, coverage (%) and N50 values of assembled se-
quences were 66% and 1.1 Mb in cucumber,
16
83% and 4.7 Mb in
melon,
17
and 83% and 2.3 Mb in watermelon,
18
respectively. The
present genome assembly of bitter gourd is comparable to the as-
sembly of other cucurbits genomes. However, 15% of the genome
was undetermined in the present sequencing analysis. It is possible
that the redundant regions of the genome, such as sequences encod-
ing multiple copies of repeats or transposons, interfered with accu-
rate assembly, resulting in shorter assembled scaffolds than the
actual complete genome. Among the scaffolds of the OHB3-1 ge-
nome, sequences of 94,148 ambiguous degenerate bases (0.03%)
were present, possibly owing to heterozygous loci or assembly of
redundant regions (data not shown).
Table 1. Summary of assembly results in OHB3-1 genome
sequence
Nuclear genome
Scaffold number 1,029
Total length (bp) 285,543,823
N50 (bp) 1,100,631
Maximum length (bp) 7,185,522
GC content (%) 36.4
Mitochondrial genome
Scaffold number 1
Total length (bp) 312,781
GC content (%) 41.1
Chloroplast genome
Scaffold number 3
Total length (bp) 140,659
Maximum length (bp) 131,815
GC content (%) 35.8
53N. Urasaki et al.
Downloaded from https://academic.oup.com/dnaresearch/article/24/1/51/2709597 by guest on 21 August 2022
Based on 80 mitochondrial and 697 chloroplast reference genome
sequences (Supplementary Tables S1 and S2), scaffolds for organelle
genome were developed. Scaffold length of the mitochondrial ge-
nome was 312,781 bp, and the total length of three scaffolds of the
chloroplast genome was 40,659 bp (Table 1). Through the BLAST
search, the scaffold sequence of the bitter gourd mitochondrial
genome showed high similarity to the watermelon mitochondrial ge-
nome sequence (Supplementary Table S4). In the assembled chloro-
plast genome, scaffold1, scaffold2, and scaffold3 showed the highest
similarity to the plastid or chloroplast genome of five-leaf ginseng
(Gynostemma pentaphyllum) or cucumber, bottle gourd (Lagenaria
siceraria), and melon, respectively (Supplementary Table S4).
3.2. Gene prediction and transposon exploration
Genes in the scaffold sequence of OHB3-1 were inferred by an ab ini-
tio prediction using the FGENESH program.
20
In total, 45,859
protein-coding genes were found as predicted genes in the OHB3-1
scaffold sequence (Table 2, Supplementary Table S5). Their average
number of CDS (coding sequences) per a predicted gene was 4.41
and 8,512 genes constituted only a single CDS. Length of encoded
protein in these predicted genes was 331 a.a on average.
Transcription start sites and polyadenylation sites in the predicted
genes were also found in 45,267 and 45,799 genes, respectively.
Gene content in the bitter gourd scaffolds was more than that in the
other sequenced Cucurbitaceae genomes (26,682 in cucumber,
16
27,427 in melon,
17
and 23,440 in watermelon
18
). This is possibly be-
cause transcript information, such as EST (expressed sequence tag)
data, was also incorporated in the gene prediction in other
Cucurbitaceae genomes. Annotation of predicted genes was per-
formed by a BLASTP search of their encoded protein sequences
against non-redundant (NR) protein and UniProt database
(Supplementary Table S6). Consequently, encoded proteins of
34,986 and 25,348 predicted genes showed a similarity to the se-
quences in NR and UniProt databases, respectively. Most of them
(25,268 proteins) showed hits to both the NR and UniProt data-
bases, whereas 80 proteins showed only hits to sequences in the
UniProt, but not the NR database. In these predicted genes, 8,839
genes encoded putative transposons as determined through
TransposonPSI analysis (Supplementary Table S7, Table 2).
Sequences of these putative transposons (43,834 kb) accounted for
15.3% of the total scaffolds of the OHB3-1 genome. The majority
(65%) of them belonged to the long terminal repeat (LTR) retro-
transposons, Ty1/copia or Ty3/gypsy. The Ty3/gypsy type was the
most abundant (35.6%), covering 5.5% of the total genome.
Considering class II transposons (DNA transposons), the CACTA
family was the most abundant (24.7%), comprising 3.8% of the to-
tal scaffold. For annotating predicted genes, domain searching was
also carried out by InterProscan. Domain search results of their
encoded protein sequences against Pfam, SMART, ProDom, and
PRINTS databases were indicated in Supplementary Table S8. In to-
tal, putative encoded proteins for 24,183 genes had any conserved
domains (Table 2). Of the unannotated genes by BLAST, conserved
domains were found in 23 predicted genes. Consequently, BLAST
and conserved domain searching resulted in the annotation of
36,086 predicted genes (75% of the predicted genes) in total.
3.3. Similarity of genome sequences in bitter gourd
with Cucurbitaceae species
A comparison of the bitter gourd genome with those of other
Cucurbitaceae crops was performed by synteny mapping of the
OHB3-1 scaffolds (285.5 Mb) against pseudomolecule sequences of
cucumber, melon and watermelon using the SyMap 4.2 program.
21
In this analysis, genome or scaffold sequences of two species were
aligned and ‘anchors’, which allowed the connection of two ge-
nomes, were determined by filtering based on annotated gene (pre-
dicted gene) information. Synteny blocks were defined as regions
consisting of more than seven anchors between two species. Synteny
blocks against the bitter gourd scaffolds covered 80–90% of each cu-
curbit genome sequence (Supplementary Table S9). In the water-
melon genome, a few synteny blocks of a long length (>10 Mb) were
observed, whereas only short (<1 Mb) fragmented blocks were
mapped in the melon and cucumber genomes (Supplementary Table
S9, Supplementary Figs S1–S3), implying relative structural similari-
ties between the bitter gourd and watermelon. In the SyMap analysis,
conserved genes connected between the bitter gourd scaffold and
other cucurbit genomes were identified as anchors. In the bitter
gourd predicted genes, 16,820, 16,063, and 16,083 genes were de-
fined as anchors (Supplementary Table S10), which corresponded to
genes in the watermelon, melon, and cucumber genome, respectively,
and 14,775 loci presumed to be conserved in all compared
Cucurbitaceae species. Of the genes at these loci, multiplicated (re-
dundant) genes in each genome or scaffold sequences were elimi-
nated, and 69 loci were defined as unique in each cucurbit genome
and conserved among all four species (Supplementary Table S11).
Based on the alignment of encoded amino acid sequences of these
orthologous genes at each locus (Supplementary Table S12), phylo-
genetic relationships were analyzed by RAxML as described in
Supplementary method. According to the constructed phylogenetic
tree (Supplementary Fig. S4), bitter gourd was related to watermelon,
rather than Cucumis spp., but it was evolutionary distant from other
species. Previous studies suggested bitter gourd was more closely re-
lated to watermelon than to cucumber or melon, according to the in-
ternal transcribed spacer regions of nuclear ribosomal RNA genes
27
or sequences of chloroplast genes.
28
Our results of synteny mapping
Table 2. Summary of predicted genes in bitter gourd (OHB3-1) scaffold sequence
Total predicted genes
a
Average length (aa) of encoded protein
a
Annotated genes
b
Putative transposable elements
c
BLAST (NR) BLAST (UniProt) InteProScan
45,859 358 34,986 25,348 24,183 8,839
a
Prediction of protein-coding genes and their translated sequences were conducted by FGENESH.
b
Encoded amino acid sequences of the predicted genes were applied to BLASTP searching against non-redundant protein database in NCBI and UniProtKB/
Swiss-Prot database or InterProscan analysis for conserved domain searching.
c
Transposable elements in the predicted genes were surveyed by TranposonPSI.
54 The draft genome sequence of bitter gourd
Downloaded from https://academic.oup.com/dnaresearch/article/24/1/51/2709597 by guest on 21 August 2022
and phylogenetic analysis seemed to support these results, but further
information of genome sequences in more Cucurbitaceae species is
necessary to elucidate their phylogenetic relationships precisely.
3.4. Unique gene finding in the bitter gourd genome
The synteny mapping analysis by using SyMap allowed to identify
unique genes and gene orders in bitter gourd scaffolds. By comparing
anchor gene lists (Supplementary Table S10), 3,158 annotated genes
in the bitter gourd scaffolds did not correspond to any genes in other
cucurbits genomes. Reversely, 2,468 genes were conserved in the ge-
nome of three cucurbit species but absent in the bitter gourd scaf-
folds. Comparing the functional annotation of these characteristic
genes (uniquely present or absent) in bitter gourd genome, two gene
classes were distinguished. Predicted genes encoding putative trypsin
inhibitor-like proteins were more frequently observed in the bitter
gourd genome than other cucurbits genomes (Supplementary Table
S13), and 29 genes encoding trypsin inhibitor-like proteins were clus-
tered in the non-syntenic scaffold regions to the other three
Cucurbitaceae genomes (Supplementary Table S14). No known con-
served domain was found in encoded proteins of 10 genes
(Supplementary Table S14), but showed a sequence similarity to cu-
cumber trypsin inhibitor-like proteins by the BLAST search. Of these
genes, three genes in scaffold_32 (MOMC32_28g, MOMC32_34g,
and MOMC32_37g) encoded identical or homologous proteins to
Mch-1 or Mch-2,
29
which were isolated from bitter gourd seeds.
Although the conserved domain as the trypsin inhibitor was not ob-
served in these proteins, it was shown that they had trypsin inhibi-
tion activity.
29
In another study, mcIRBP, corresponding to
MOMC1_984, was isolated from bitter gourd seeds as an insulin
receptor-binding protein, and its injection reduced blood glucose lev-
els in mice,
30
implying its possible application to anti-diabetic medi-
cation. Phylogenetic analysis of putative trypsin inhibitors in bitter
gourd showed that proteins with the I7 protease inhibitor and I13
protease inhibitor domains were separated, and several proteins
without conserved domains (Supplementary Table S14) were catego-
rized in the I7 protease inhibitor group. MOMC32_27, MOMC32_
28, MOMC32_34 (Mch-2), and MOMC32_37 (Mch-1) formed a
monophyletic group, close to I7 protease inhibitors (Fig. 1).
Additional notably unique genes in bitter gourd genome were ri-
bosomal inactivating protein (RIP) genes. RIP is known as a plant
toxin, which has N-glycosidase activity against adenine nucleotide in
ribosomal RNA.
31
Most RIPs were classified into two groups (type 1
and type 2).
31
Type 1 RIP is a monomeric protein encoding the N-
glycosidase activity domain (A-chain), and type 2 RIP consists of A-
chain and B-chain (lectin-like domain). Multiple copies of RIP genes
were observed in the Cucurbitaceae genome, but more paralogous
genes, including both genes encoding A-chain (18 genes) and B-chain
(8 genes), were present in the bitter gourd scaffolds than other cucur-
bit genomes (Supplementary Table S13). These bitter gourd RIP
genes were clustered in six scaffolds, which were non-syntenic re-
gions to other cucurbits genomes (Supplementary Table S15).
Biological functions of RIPs in plants were not always well eluci-
dated, but were possibly involved in the defense system against path-
ogenic fungi or bacteria via rRNA cleavage.
32,33
On the other side,
bitter gourd RIPs were well studied as a possible medically effective
ingredient.
34
. Alpha-momorcharin and MAP30 were type-1 RIPs iso-
lated from bitter gourd, and shown to have anti-viral and -tumor ac-
tivity in mammalian cells.
35,36
When these RIPs were used to treat
HIV-infected cells, inhibition of viral replication was observed
36
and
additionally, MAP30 also inhibited integrase acitivty of HIV and
Figure 1. Phylogenic analysis of putative trypsin inhibitor (A) and ribosome
inactivating protein (B) in bitter gourd. Based on 34 amino acid sequences of
trypsin inhibitor proteins and 18 amino acid sequences of ribosome inacti-
vating proteins of bitter gourd (Supplementary Table S12, S14, and S15),
phylogenetic tree for each protein family was constructed using the
Neighbor-Joining method by MEGA7.0.18. The percentage of replicate trees
in which the associated taxa clustered together in the bootstrap test (500 rep-
licates) are shown next to the branches. The tree is drawn to scale, with
branch lengths in the same units as those of the evolutionary distances used
to infer the phylogenetic tree. The evolutionary distances were computed us-
ing the p-distance method and are in the units of the number of amino acid
differences per site.
55N. Urasaki et al.
Downloaded from https://academic.oup.com/dnaresearch/article/24/1/51/2709597 by guest on 21 August 2022