result in that miR169c was between 33 and 82%
of the reads in each condition, with large variability between the three biological replicates.
This result was believed by the authors of the
study to be an experimental artifact due to the
lack of this expression in the other experiments,
the only partial replication of the expression in
the qPCR follow-up, and the current reputation
of the SOLiD5500 sequencer. With this one
sequence ignored and the dataset renormalized,
we can accurately see the responses of other
miRNAs to the various conditions. There were
large increases in miR166 expression under the
influence of cold and kinetin and miR168 in the
heat and sucrose conditions. The meristem regulating 396 familiesy doubled expression in
response to the heat, ABA, and copper stimuli.
Finally, miR156, which maintains the juvenile,
neotenous life cycle of the duckweed family,
decreased over fourfold in response to sucrose,
which was the condition responsible for 13 of the
19 instances of differential miRNA expression
indicating that the mixotrophic lifestyle often
used in laboratory experiments is quite different
from duckweed grown in an outdoor setting.
Accurate miRNA annotation is quite difficult,
since miRNAs are vastly outnumbered by similarly sized siRNAs in the genome, and even the
more stringent miRNA prediction programs
supply tens or hundreds of false predictions. An
analysis in 2014 suggested that 75% of the land
plant miRNA families in miRBase are questionable, especially those with only a single member
(Taylor et al. 2014). In an attempt to manage the
large number of submissions and false positives
coming in, miRBase has established criteria for
its high confidence miRNAs that analyze the
structure of the hairpin, the read distribution
along it, and the miR, and miR* read count. For
plants in miRBase release 21, there are currently
6942 hairpins in 2408 distinct miRNA families,
with only 587 from 227 families (9.7%) making
the high confidence cutoff (Griffiths-Jones 2006).
As an attempt to preserve miRNA annotation
confidence, 21 of the leading minds of the field
wrote the plant miRNA annotation criteria in
2008 that has since been updated by two of them
thanks to new information and sequencing
capabilities (Meyers et al. 2008; Axtell and
Meyers 2018). The plant miRNA annotation
criteria are generally more stringent than the high
confidence criteria, except for the latter’s
requirement of 10 miR* reads, since plant
miRNA biogenesis is quite specific. Both
miRNA studies in Spirodela annotated miRNAs
based on homology according to the 2008 criteria, with most of these being well-conserved,
high-confidence miRNA families. The conserved
miRNAs with family names above 535 are relatively likely to be based off of lower confidence
annotations in previous reports. The novel miRNAs from strain 9509 were predicted in 2017
using cutoffs very similar to the 2018 criteria,
demonstrating a high degree of confidence, while
those predicted in the LT5a and 7498 study had a
lower degree of confidence. These authors
reviewed all their data, with the revised criteria
finding that 30 of the 47 hairpin structures met
the current standards.
In addition to applying the stringent structural
and read distribution filters above, the authors of
the 2018 study verified miRNAs through a
method called degradome sequencing where
uncapped mRNAs are sequenced and aligned to
miRNA target sites to measure evidence of precise miRNA cleavage above random mRNA
degradation. There were several methods available at the time, and the authors chose the
GMUCT2.0 library for its read length and minimal PCR amplification and the sPARTA program for its accuracy in analysis of the
degradome data (Kakrana et al. 2014; Willmann
et al. 2014). Biological triplicate libraries of the
same eight conditions observed in the
miRNA-sequencing study were sequenced on the
Illumina NextSeq 500, yielding 911 million total
reads. When running the sPARTA program, the
Spirodela 7498 gene models were extended
150nt upstream and 250nt downstream, since
many of the degradome reads were from the
UTRs of the mRNAs. The sequencing verified
activity of 66 miRNAs on 149 targets. For the 42
conserved miRNAs, the targets were mainly the
transcription factor families reported in other
plant species. While these essential developmental transcription factors mostly made up the
16 Small RNAs in Duckweeds
161
of the reads in each condition, with large variability between the three biological replicates.
This result was believed by the authors of the
study to be an experimental artifact due to the
lack of this expression in the other experiments,
the only partial replication of the expression in
the qPCR follow-up, and the current reputation
of the SOLiD5500 sequencer. With this one
sequence ignored and the dataset renormalized,
we can accurately see the responses of other
miRNAs to the various conditions. There were
large increases in miR166 expression under the
influence of cold and kinetin and miR168 in the
heat and sucrose conditions. The meristem regulating 396 familiesy doubled expression in
response to the heat, ABA, and copper stimuli.
Finally, miR156, which maintains the juvenile,
neotenous life cycle of the duckweed family,
decreased over fourfold in response to sucrose,
which was the condition responsible for 13 of the
19 instances of differential miRNA expression
indicating that the mixotrophic lifestyle often
used in laboratory experiments is quite different
from duckweed grown in an outdoor setting.
Accurate miRNA annotation is quite difficult,
since miRNAs are vastly outnumbered by similarly sized siRNAs in the genome, and even the
more stringent miRNA prediction programs
supply tens or hundreds of false predictions. An
analysis in 2014 suggested that 75% of the land
plant miRNA families in miRBase are questionable, especially those with only a single member
(Taylor et al. 2014). In an attempt to manage the
large number of submissions and false positives
coming in, miRBase has established criteria for
its high confidence miRNAs that analyze the
structure of the hairpin, the read distribution
along it, and the miR, and miR* read count. For
plants in miRBase release 21, there are currently
6942 hairpins in 2408 distinct miRNA families,
with only 587 from 227 families (9.7%) making
the high confidence cutoff (Griffiths-Jones 2006).
As an attempt to preserve miRNA annotation
confidence, 21 of the leading minds of the field
wrote the plant miRNA annotation criteria in
2008 that has since been updated by two of them
thanks to new information and sequencing
capabilities (Meyers et al. 2008; Axtell and
Meyers 2018). The plant miRNA annotation
criteria are generally more stringent than the high
confidence criteria, except for the latter’s
requirement of 10 miR* reads, since plant
miRNA biogenesis is quite specific. Both
miRNA studies in Spirodela annotated miRNAs
based on homology according to the 2008 criteria, with most of these being well-conserved,
high-confidence miRNA families. The conserved
miRNAs with family names above 535 are relatively likely to be based off of lower confidence
annotations in previous reports. The novel miRNAs from strain 9509 were predicted in 2017
using cutoffs very similar to the 2018 criteria,
demonstrating a high degree of confidence, while
those predicted in the LT5a and 7498 study had a
lower degree of confidence. These authors
reviewed all their data, with the revised criteria
finding that 30 of the 47 hairpin structures met
the current standards.
In addition to applying the stringent structural
and read distribution filters above, the authors of
the 2018 study verified miRNAs through a
method called degradome sequencing where
uncapped mRNAs are sequenced and aligned to
miRNA target sites to measure evidence of precise miRNA cleavage above random mRNA
degradation. There were several methods available at the time, and the authors chose the
GMUCT2.0 library for its read length and minimal PCR amplification and the sPARTA program for its accuracy in analysis of the
degradome data (Kakrana et al. 2014; Willmann
et al. 2014). Biological triplicate libraries of the
same eight conditions observed in the
miRNA-sequencing study were sequenced on the
Illumina NextSeq 500, yielding 911 million total
reads. When running the sPARTA program, the
Spirodela 7498 gene models were extended
150nt upstream and 250nt downstream, since
many of the degradome reads were from the
UTRs of the mRNAs. The sequencing verified
activity of 66 miRNAs on 149 targets. For the 42
conserved miRNAs, the targets were mainly the
transcription factor families reported in other
plant species. While these essential developmental transcription factors mostly made up the
16 Small RNAs in Duckweeds
161