Postharvest Biology and Technology 185 (2022) 111806

Contents lists available at ScienceDirect

Postharvest Biology and Technology

journal homepage: www.elsevier.com/locate/postharvbio

Transcriptome and hormone analyses reveals differences in physiological

age of ′ Hass′ avocado fruit
Ignacia Hernández a, Virgilio Uarrota a, Claudia Fuentealba a, Diego Paredes b,
Bruno G. Defilippi c, Reinaldo Campos-Vargas d, Gerardo Nuñez a, Esther Carrera e,
Claudio Meneses f, Maarten Hertog g, *, Romina Pedreschi a, *
a
Facultad de Ciencias Agronómicas y de Los Alimentos, Escuela de Agronomía, Pontificia Universidad Católica de Valparaíso, Chile
b
Departamento de Ingeniería Matemática (DIM) & Centro de Investigación en Ingeniería Matemática (CI2MA), Universidad de Concepción, Chile
c
Unidad de Postcosecha, Instituto de Investigaciones Agropecuarias INIA, La Platina, Santiago, Chile
d
Centro de Estudios Postcosecha, Facultad de Ciencias Agrarias, Universidad de Chile, Chile
e
Instituto de Biología Molecular y Celular de Plantas, Universidad Politécnica de Valencia-Consejo Superior de Investigaciones Científicas, Valencia, Spain
f
Centro de Biotecnología Vegetal, Facultad de Ciencias de la Vida, Universidad Andrés Bello, Santiago, Chile
g
Department of Biosystems, Katholieke Universiteit Leuven, Belgium

A R T I C L E I N F O A B S T R A C T

Keywords: The objective of this study was to identify transcripts or hormone-based biomarkers to define the physiological
Heterogeneity age of ′ Hass′ avocado fruit and to elucidate the changes at the level of metabolic pathways and their regulation.
Firmness ′
Hass′ avocado fruit from orchards in different agroclimatic zones were collected during two harvest periods.
Hormones
Fruit were stored for 30 d under controlled atmosphere and regular air conditions and then transferred to shelf-
Maturation
life conditions at 20 ◦ C. The physiological age as represented by the initial state of a hypothetical enzyme system
Transcripts
Persea Americana (E0) of each fruit was obtained through a mechanistic softening model for Chilean ′ Hass′ avocado previously
developed. Fruit from three different E0 ranges (low, intermediate and high) were selected for transcriptome and
hormone analyses. Sequencing data were processed by partial least squares regression analysis, which revealed
46 genes correlated to E0. Different metabolic pathways were over expressed between low and high E0 fruit. Low
E0 fruit showed overexpression of genes related to DNA replication, auxin transport, cell wall remodeling,
gibberellin synthesis, brassinosteroids and flavonols. On the other hand, fruit with high E0 revealed genes related
to ethylene and abscisic acid biosynthesis and related responses and phenylpropanoid biosynthesis. Likewise,
targeted hormone analysis revealed higher concentrations of active gibberellins (GA1 and GA4) and jasmonic acid
for low E0 fruit and for high E0 fruit higher concentrations of abscisic acid, salicylic acid, dihydrozeatin, indole
acetic acid and trans-zeatin, only the latter two being significant for this phenotype. This study reveals the
relationship between transcripts and hormones during fruit maturation that is key to evaluate the physiological
age of ′ Hass′ avocado fruit.

1. Introduction 2021). The low water resources have reduced the production area
planted with avocado cv. Hass by 20 % over recent years (Oficina de
The cultivation of avocado cv. Hass in Chile is of significant eco­ Estudios y Política Agrarias (ODEPA), 2021). Thus, competition with
nomic importance, thanks to the large planted area, reaching 30,000 ha neighbouring exporting countries that present more favourable hydro­
in 2020 (Oficina de Estudios y Política Agrarias (ODEPA), 2021). During logical conditions solely based on production volume is not possible but
the 2019–2020 season, 168,000 ton fruit was produced, of which 72 % requires differentiation towards a high-quality product.
was exported; mainly to Europe, the United States, China and Argentina. The main challenge to provide the market with a high-quality
Over 70 % of the ′ Hass′ avocado production in Chile is concentrated in product (in terms of colour, firmness, days to reach eating quality) is
the Valparaíso Region (Oficina de Estudios y Política Agrarias (ODEPA), the high heterogeneity evidenced postharvest (Hernández et al., 2017;

* Corresponding authors.
E-mail addresses: maarten.hertog@kuleuven.be (M. Hertog), romina.pedreschi@pucv.cl (R. Pedreschi).

https://doi.org/10.1016/j.postharvbio.2021.111806
Received 29 September 2021; Received in revised form 28 November 2021; Accepted 29 November 2021
Available online 7 December 2021
0925-5214/© 2021 Elsevier B.V. All rights reserved.
I. Hernández et al. Postharvest Biology and Technology 185 (2022) 111806

Pedreschi et al., 2019). Although the Chilean ′ Hass′ avocado industry RNA-seq allows the creation of entire transcriptome profiles,
relies on a minimum 23 % dry matter as commercial harvest index, providing a much more extensive coverage of the metabolic pathways
previous studies have demonstrated that neither dry matter content nor and networks involved in ′ Hass′ avocado at its different physiological
at-harvest firmness are good indicators of the physiological age of the stages (as indicated by E0), especially when complemented with a tar­
fruit and thus of its ripening behaviour (Rivera et al., 2017; Pedreschi geted hormone analysis. The present study aims: (i) to propose early
et al., 2014; Hernández et al., 2021). Previous research has developed biomarkers of physiological age at the level of transcripts and hormones
kinetic models capable of using simplified physiological concepts to and (ii) to elucidate the differences at the level of metabolic pathways
predict the loss of firmness and ripening behaviour of different batches and their regulation between avocado cv. Hass fruit with different
of avocado cv. Hass (Ochoa-Ascencioa et al., 2009; Gwanpua et al., physiological age (as indicated by E0).
2018; Hernández et al., 2021) as to early segregate batches of fruit into
fast and slow ripening. Recently, Hernández et al. (2021) developed a 2. Material and methods
mechanistic model based on the premises of Ochoa-Ascencioa et al.
(2009) but incorporated real measurements of non-destructive firmness 2.1. Conditions of sampling, fruit storage and biopsy sampling
at harvest, in addition to correlating polar metabolites content at harvest
with the model parameter E0 reflecting the physiological or biological The plant material, sampling and storage conditions used correspond
age of the fruit. The parameter E of the mechanistic model developed is to those detailed in Hernández et al. (2021). Briefly, two hundred ′ Hass′
an enzyme complex that encompasses all the enzymes and their in­ avocado fruit from 12 orchards from three different agroclimatic zones
teractions involved in the softening process, describing a complex pro­ were used. The sampling considered early harvest fruit (≥ 23 % - 26 %
cess, such as loss of firmness, in a simplified manner. Therefore E0 dry matter content) and middle harvest fruit (> 26 % to 30 % dry matter
(enzyme complex at time zero) will represent the initial state of the content). One hundred fruit from each batch were stored in controlled
complex at harvest and whether it is higher or lower will have a direct atmosphere (CA) conditions of 4 kPa O2 and 6 kPa CO2 at 5 ◦ C for 30 d.
effect on the softening process (rate and heterogeneity). In addition, The other 100 remaining fruit from each orchard were stored in regular
Monte Carlo simulations, revealed that segregation of fruit based on the air (RA) at 5 ◦ C for 30 d. After storage in RA or CA, fruit were brought to
E0 value reflecting their physiological age would have a tremendous shelf-life conditions at 20 ◦ C until each fruit reached the ready-to-eat
positive effect on improved firmness retention for distant markets (low stage (4− 8 N). Biopsy sampling conditions and non-destructive firm­
E0 – premium fruit vs high E0 mainstream fruit) and on reducing ness measurements of each evaluated fruit were performed as described
ripening heterogeneity. The GC–MS polar metabolite profile revealed by Hernández et al. (2021).
potential as a biochemical phenotyping technique to assess the physio­
logical age of the ′ Hass′ avocado fruit at an early stage, as did previous
2.2. Mechanistic model used - estimation of physiological age at fruit level
research by García et al. (2018 and 2019) where they used metab­
olomics to find early biomarkers of the browning of freshly cut lettuce.
The estimated E0 values for each fruit analyzed were obtained from
However, other omics platforms such as transcriptomics through RNA
the mechanistic model developed by Hernández et al. (2021). The
sequencing allow the profiling of the entire transcriptome, providing a
experimental data were analyzed using the mechanistic model based on
much broader coverage compared to the metabolome. For instance, the
ordinary differential equations (ODE), to describe the softening of the
study reported by Nielson et al. (2017) successfully used gene expression ′
Hass′ avocado from different agroclimatic zones and harvests (early and
profiles with selection of a set of genes as predictive biomarkers of
middle). The model presents a generic approach based on batches and
cold-induced sweetening in potatoes. Therefore, due to the complexity
another specific based on individual fruit. The physiological basis of the
of the maturation process of the ′ Hass′ avocado due to the multiple in­
model used is based on a simplified representation of the participation of
teractions that exist among different levels of cellular control, tran­
an enzyme complex (E0) in autocatalytic processes, including the
scriptomics in conjunction with targeted hormone analysis could bring
exponential increase in the activity of E0 during ripening and on the
more solid early correlations with E0 (reflecting the physiological age of
action of E0 on firmness retention.
the fruit).
The estimation of the E0 parameter at the batch and individual fruit
Due to the climacteric nature of ′ Hass′ avocado most research has
level is further detailed in the study by Hernández et al. (2021). Briefly,
focused on the role of ethylene during fruit ripening (Adato and Gazit,
the enzyme complex (E in arbitrary units) is responsible for the break­
1977; Jeong and Huber, 2004; Kumar et al., 2014) and only recently the
down of firmness (F in N):
role of abscisic acid (ABA) has been reported (Meyer et al., 2017). In
addition, Vincent et al. (2020) have recently reported a complex hor­ kf
F + E ̅̅̅̅̅̅̅→ E (1)
monal interplay during ripening of avocado revealing the participation
of other hormones such as abscisic acid (ABA), jasmonates, gibberellins In a simplified way, the autocatalytic process of the enzyme complex
and auxins. In addition, Uarrota et al. (2019) reported large proteome is explained from a limited inactive precursor resource (Epre):
differences at harvest associated with differences in physiological age
(2)
ke
affecting ripening behavior. Ripening is a high complex process that Epre + E ̅̅̅̅̅̅̅̅̅→ 2∙E with rate constant ke (in d − 1)
involves the activation and interaction of several metabolic pathways, Then, to explain the changes in firmness and in the enzyme complex,
gene expression and regulatory mechanisms. Therefore, a tran­ three ordinary differential equations are elaborated that can be derived
scriptomics approach could provide an additional platform to search for from Eq.s (1) and (2).
early biomarkers of physiological age of ′ Hass′ avocado. The nuclear
genome of P. americana var. drymifolia and P. americana var. Hass has dF
= − kf ∙ E ∙ (F − Ffix)
been sequenced relatively recently (Rendón-Anaya et al., 2019). De dt
novo transcriptomics has been used to study lipid biosynthesis during dE
fruit development of avocado cv. Hass (Vergara-Pulgar et al., 2019) = ke ∙ E ∙ Epre
dt
revealing potential biomarkers of fruit development. To our knowledge,
previous biomarker studies on fruit maturity of ′ Hass′ avocado relied on dEpre
= − ke ∙ E ∙ Epre (3)
proteomics or metabolomics platforms (Fuentealba et al., 2017; Pedre­ dt
schi et al., 2019; Uarrota et al., 2019; Hernández et al., 2021) but did not
With the following initial at-harvest values (t = 0 d):
explore the potential of transcriptomics to identify marker genes
correlating with the physiological age of the fruit at commercial harvest. E(0) = E0

2
I. Hernández et al. Postharvest Biology and Technology 185 (2022) 111806

F(0) = F0 used for the partial least squares regression (PLS-R) including mixOmics
(Rohart et al., 2017), mdatools (Kucheryavskiy, 2020), pls (Mevik et al.,
Epre(0) = Etot - E0 (4) 2020), caret (Kuhn, 2020) and tidyverse (Wickham et al., 2019).
Candidate genes for PLS-R were selected in DaMiRseq R package (Chiesa
In these equations it is considered that avocados lose their firmness & Piacentini, 2020). A series of functions enables data cleaning by
until edible ripeness around 4–8 N (Ffix in N). Etot is in arbitrary units. filtering genomic features and samples. Gene filtering was done by
The rate constants kf and ke are assumed to be temperature dependent setting up the minimum number of read counts permitted across samples
following Arrhenius’s law. and by removing hyper variants (i.e., those genes that present anoma­
lous read counts) by comparing to the mean value across the samples
2.3. At harvest transcriptomics analysis: RNA extraction and library and finally filtered by low correlation.
construction Based on the results obtained in the differential expression analysis,
the differences were compared at the level of metabolic pathways using
A total of 36 biopsies were selected from three different physiological the most contrasting E0 categories only (high vs low). To search for
age ranges (in terms of E0) and included a wide range of dry matter genetic functions and pathways overrepresented in the DEG lists, genetic
contents (from 22.3 up to 32.8 %). These three ranges corresponded to: enrichment analyzes were performed using the Genetic Ontology (GO)
12 samples for the lowest range (E0 < 5), 12 samples for the interme­ database with AgriGO v2.0 (Du et al., 2010) and the reconstruction of
diate range (5 < E0 <10) and 12 samples for the highest range (E0 > 10). the metabolic pathways was performed using the Kyoto Encyclopedia of
Total RNA was extracted from 100 mg of frozen tissue using a Spec­ Database of Genes and Genomes (KEGG) (Kanehisa et al., 2021).
trum™ Plant Total RNA kit (Sigma-Aldrich, St. Luis, USA) following the
manufacturer’s instructions and stored at − 80 ◦ C. The quantity and 2.5. Validation of RNA-seq by qRT-PCR analysis
purity of RNA were evaluated with a Qubit®2.0 fluorometer (Invi­
trogen™, Carlsbad, CA, USA) using a Qubit™ RNA BR assay kit. RNA To perform a technical validation of DEG analysis, primers for F-ACP
integrity and concentration were assessed by capillary electrophoresis housekeeping and for 6 candidate genes PIN1, TRN1, NAC072, EXPA8,
using an automated CE Fragment Analyzer™ system (Agilent Technol­ XTH5 and BR6ox1 were designed using the software Primer 3Plus
ogies, Santa Clara, CA, USA) with the RNA kit DNF-471− 0500 (15 nt). (https://www.bioinformatics.nl/cgi-bin/primer3plus/primer3plus.cgi/
The RNA quality number (RQN value) was used to identify the integrity ) based on the RNA sequences aligned. First, from the same input used to
of the RNA. RNA samples with an RQN value beyond 7.0 were used for construct the libraries, total RNA was treated with DNase I (Fermentas,
the following steps. Total RNA-seq libraries were prepared according to Thermo Fisher Scientific, Waltham, MA, USA) according to the standard
the TruSeq Stranded Total RNA Kit (Illumina, San Diego, CA, USA) protocol. The first strand cDNA was obtained by reverse transcription
following the manufacturer’s instructions. The concentration of the li­ using the Superscript II RT system (Invitrogen, Carlsbad, CA, USA). The
braries was determined with a Qubit®2.0 fluorometer (InvitrogenTM, cDNA concentration was obtained by measuring absorbance at 260 nm.
Carlsbad, CA, USA) using a Qubit™ dsDNA BR assay kit and the size and Each cDNA sample was diluted to 20 ng μL− 1 before being used in qRT-
integrity of the library was evaluated with capillary electrophoresis PCR assays. The qRT-PCR assays were performed in a AriaMx real-time
using the Automated CE Fragment Analyzer™ (Agilent Technologies, PCR system (Agilent Technologies, Santa Clara, CA, USA) with KAPPA
Santa Clara, CA, USA) with DNF-474− 0500 HS NGS Fragment Kit. The SYBR® Fast suitable for qPCR (Sigma-aldrich, Saint Louis, MO, USA) to
constructed libraries were sequenced using Macrogen sequencing ser­ measure the DNA product derived from RNA.
vices (Seoul, Korea) in paired end mode on a HiSeq4000 sequencer.
2.6. Hormone analysis
2.4. RNA data analysis
A total of 18 samples of avocado cv. Hass were analyzed at harvest
For total RNA differential expression analysis, a quality check was from the two most contrasting E0 groups: 9 samples corresponded to the
first performed on the raw data files with FASTQC software (Andrews, lowest E0 < 5 group and the other 9 samples to the highest E0 > 10
2010) to assess the most appropriate read quality filtering and clipping. group. The quantification of abscisic acid (ABA), jasmonic acid (JA),
The following criteria were used with Flexbar (Dodt et al., 2012): (1) gibberellins (GA1 and GA4), auxin as indole acetic acid (IAA), cytokinins
remove adapter sequences; (2) eliminate reads with a quality score less (trans-zeatin tZ and dihydrozeatin DHZ) and salicylic acid (SA) were
than 30; and (3) eliminate reads with a length of more than 60 nucle­ identified and quantified by UHPLC–ESI–MS/MS following the protocol
otides. The STAR aligner software (Dobin et al., 2013) was used to align of Seo et al. (2011). After extraction, samples were analyzed in a
the filtered reads against Persea americana var. drymifolia genome v3.0. Q-Extractive mass spectrometer applying electrospray ionization (ESI)
For each library, the featureCounts software from the Rsubread package followed by selected ion monitoring. The hormones were quantified
(Liao et al., 2019) was applied to assign expression values to each based on internal deuterated standards by construction of calibration
uniquely aligned fragment. Differential gene expression analysis was curves. Each hormone is expressed in μg kg− 1 DW (dry weight).
performed using the Bioconductor R edgeR package (Robinson et al.,
2010). Differentially expressed genes (DEG) were selected with a false 3. Results and discussion
discovery rate less than 0.05 and a log 2-fold change (FC) larger than 1.0
or smaller than -1.0. 3.1. At harvest-relevant transcripts and their correlation with
Differential expression data were subjected to a statistical analysis physiological age: in the search of early biomarkers of the physiological age
with a multivariate approach to find potential physiological age bio­ of the fruit
markers for the different harvests (early and middle). Before multivar­
iate analysis, gene count data was pre-processed by filtering The thirty-six samples subjected to transcriptome analysis corre­
unexpressed genes, inconsistent genes, and then normalized using the sponded to 12 different orchards in three agroclimatic zones (interior,
"Variance Stabilizing Transformation (VST)" method. Samples were intermediate and coast) and two harvests (early and middle). These
checked for quality and adjusted to eliminate putative conformity fac­ samples represent the high biological variability of the typical Chilean
tors from the expression data. The analyzes were carried out in the R ′
Hass′ avocado production (Hernández et al., 2021). Due to the signifi­
software (R Core Team, 2021), for the preprocessing stage the DaMiRseq cant number of genes found in the sequencing of the samples (22,929
package was used (Chiesa et al., 2018). Factoextra (Kassambara and genes), the samples were processed using an unsupervised approach
Mundt, 2020) was used to perform the PCA and a mix of packages were through principal component analysis (PCA), to reduce the

3
I. Hernández et al. Postharvest Biology and Technology 185 (2022) 111806

dimensionality of the experimental data. PCAs were carried out

considering both harvests together and each harvest independently
(Fig. 1). When both harvests were analyzed together, no clear separation
of the samples could be observed by PCA (Fig. 1a). Samples from the
early harvest (Fig. 1b) presented a more pronounced separation between
the three categories of E0 (low, medium and high) as compared to the
samples of middle harvest (Fig. 1c). This is evidenced by the larger
amount of explained variation by the first two PCs (38.92 % and 16.57 %
respectively). Previous studies have reported early harvest ′ Hass′ avo­
cado to be the most problematic in terms of ripening heterogeneity
(Fuentealba et al., 2017; Hernández et al., 2017 and Uarrota et al.,
2019). Part of this is related to the fact that commercial maturity (dry
matter content ≥ 23 %) is determined based on few fruit only (10–20
fruit). Thus, early harvest fruit within a single batch can display very
different physiological/biological age including fruit that might not
even have reached physiological maturity, then a sufficient large sample
size is crucial to make the correct harvesting decision.
One of the objectives of this work was to identify early tran­
scriptomic biomarkers of physiological age that correlate with the esti­
mated parameter (E0) provided by the model developed by Hernández
et al. (2021) for each individual fruit. The total of genes sequenced (22, Fig. 2. Principal Partial Least Squared Regression (PLSR) of samples with
929 genes) was used to perform partial least squares regression analysis different biological age (E0) corresponding to early and middle harvests based
(PLS-R) as described in the materials and methods section. This model on relevant transcripts with E0. Partial least squares regression biplot (PLS-R)
where relevant transcripts of early and middle harvests are the X variables and
was able to explain with the first two latent variables 30.49 % and 21.04
the biological age E0 corresponds to the response of Y variable. Full annotation
% of the X variance and 45.37 % and 18.02 % of the Y variance (Fig. 2).
of these 46 genes is provided in Supplementary Table 4.
Model cross-validation (CV) (i.e., the ability to select the correct number
of components) was done by dividing the data into segments and the
(2021), which used the GC–MS polar metabolite profile and explained
validation results presented by the Root Mean Squared Error of Pre­
only 8.81 % and 12.82 % of the X variance and 17.35 % and 16.85 % of
diction (RMSEP) with two cross-validation estimates: the ordinary CV
the Y variance, so our transcriptomic analysis proved to be more robust
estimate, and a bias-corrected CV estimate, obtaining a total accumu­
to search for early-stage biomarkers of the physiological age of ′ Hass′
lated value of R2 = 0.634 and Q2 = 0.395, respectively with the two
avocado. In an earlier study by Vergara-Pulgar et al. (2019) using
selected first components. Feature selection analysis revealed 46
RNA-seq, the authors reported genes related to lipid biosynthesis and the
candidate genes and partial least squares regression (PLS-R) showed that
development of ′ Hass′ avocado fruit, as potential candidate biomarkers
25 displayed a positive correlation with E0 and 21 displayed a negative
to monitor fruit development and harvest index, however, these candi­
correlation with E0. Other previous studies (Neilson et al., 2017; García
dates were not validated. The RNA-seq technique creates profiles of the
et al., 2018, 2019) have used the same multivariate approach to find
entire transcriptome clearly providing wider information as compared
early biomarkers of specific quality traits that occurs after the storage
to other omics technologies addressing the proteome or the
period. The gene expression or metabolite profiles developed in these
metabolome.
investigations were established as predictive biomarkers, so that later
the physiological basis behind these biomarkers can be clarified. In these
investigations, as in the current study, the number of biomarkers is quite 3.2. Sequencing and mapping of the ′ Hass′ avocado transcriptome of
high, which could make the prediction process difficult. To reduce the different physiological/biological ages
high number of biomarkers, Neilson et al. (2017) and García et al.
(2018) proposed the use of a multiple regression analysis, obtaining a To know the dynamics of the ′ Hass′ avocado transcriptome of
model with a high R2 (0.92) and with a fine selection of biomarkers. Our different physiological/biological ages (as reflected by their E0), RNA
approach based on partial least squares regression (PLS-R) and feature libraries for 3 different E0 ranges (low, medium, and high) were built.
selection could be further optimized in subsequent studies in order to The sequencing of the 36 samples produced an average of 61,703,640
reduce even more the number of transcriptomic biomarkers (46 genes) reads (SRA codes https://www.ncbi.nlm.nih.gov/bioproject/PRJ
that predict the biological age of the ′ Hass′ avocado. NA754775/) for each sample and after performing the quality filtering
The partial least squares regression analysis, in addition to revealing of each sample, the number of readings remained on average at 99 %
interesting potential biomarkers, the variance explained in this analysis (Supplementary Table 1). As mentioned above, the principal component
was higher than the PLS-R analysis performed by Hernández et al. analysis (PCA) showed a better separation of the E0 categories for early

Fig. 1. Principal Component Analysis (PCA) of

samples with different biological age (E0) cor­
responding to early and middle harvests
together and independently analyzed. (A) Score
plot of the PCA analysis of the 3 different ranges
of E0, low (E0 < 5), medium (5 > E0 < 10) and
high (E0 > 10) for the early and middle harvest
fruit (dry matter content of ≥ 23 – 26 % and >
26 % respectively) (B) Score plot of the PCA
analysis of the 3 different ranges of E0, low (E0
< 5), medium (5 > E0 < 10) and high (E0 > 10)
for the early harvest fruit (dry matter content of
23 – 26 %); (C) Score plot of PCA of the 3
different ranges of E0, low (E0 < 5), medium (5 > E0 < 10) and high (E0 > 10) for middle harvest fruit (dry matter content > 26 %).

4
I. Hernández et al. Postharvest Biology and Technology 185 (2022) 111806

harvest fruit (Fig. 1b), confirming that differences in physiological was performed and after selection only 46 candidate genes remained
maturity are more problematic for early harvest fruit characterized by a with a high amount of explained variance. To gain insight into the
lower dry matter content (≥ 23–26 % DM). Based on these results, only biological interpretation of the differences between the most contrasting
early season fruit was included in the differential expression analysis of low E0 and high E0 categories, differential expression analysis only
the next section. considering these two categories were used and revealed 1806 differ­
In addition, Fig. 1 considered the three E0 categories (low, medium entially expressed genes. Both approaches shared 13 genes (Supple­
and high) corresponding to E0 ranges (low, E0 < 5; medium, 5 > E0 < 10; mentary Table 3). These 13 common genes displayed lower log FC (-4.04
high, E0 > 10), also revealed that only marked separated groups were to 2.33) than the 26 genes selected by differential expression analysis
obtained between low and high E0 samples. Therefore, to biologically (Table 1). For the biological interpretation contrasting the high and low
interpret the differentially expressed genes, we will now focus on the E0 categories, only the 26 genes selected by the differential expression
two most contrasting classes of E0, the low and high ranges observed for analysis were considered.
early harvest being fruit that were also the most clearly differentiated in
the PCA analysis (Fig. 3a), and in this way to be able to estimate the
variability of global gene expression between these more extreme cat­ 3.3. Evaluation of expression differences between high vs low E0 fruit
egories. This PCA analysis (Fig. 3a) revealed an evident separation be­ samples over-expressed metabolic pathways
tween the low E0 and high E0 samples and explained 55.01 % and 11.33
% of the total variance with the first two components. Avocado cv. Hass does not ripen while attached to the tree, thus
Although the partial least squares regression analysis revealed 46 within a tree, fruit of very different maturities co-exist and differences in
candidate genes related to the biological age (in terms of E0) of the ′ Hass′ physiological age among fruit only become evident after harvest
avocado fruit, these biomarkers are not capable to provide a physio­ (Hernández et al., 2016). Our attempt to search for biomarkers of bio­
logical/biological explanation related to the differences in biological age logical age at harvest is challenging since the transition from growth to
of the fruit since they were determined considering the three categories maturation is quite discrete involving shifts in phytohormones profiles
and the actual E0 estimates from the model. Thus, we decided to perform to stop fruit expansion and promote ripening (Forlani et al., 2019; Fenn
a differential expression analysis between the most contrasting E0 cat­ and Giovannoni, 2021). The differences in pathways over expressed
egories. Results are displayed in Fig. 3b based on a Pearson correlation between the low vs high E0 samples nicely represent these changes from
that shows a high similarity between the samples of the same category growth to maturation as can be seen in Fig. 5. The low E0 category,
(low and high E0 categories) being consistent with the PCA analysis revealed higher expression of genes related to DNA replication (POL2A,
displayed in Fig. 3a. In both cases (PCA analysis and correlation matrix), FC = -2.2; N.N, FC = -2.6; POLA2, FC = -3), auxin transport (PIN4, FC =
a homogeneity is shown between the samples of each E0 category, but -2.4; PIN1, FC = - 2.7; TRN2, FC = -3.7; TRN1, FC = -4.2), cell wall
there are significant changes in the expression of certain genes that remodeling (EXPA1, FC = -2.2; EXPA8, FC = -2.3; XTH33, FC = -3.0;
compose the transcriptome. A total of 1806 genes with a false discovery XTH5, FC = -4.3; EXPA8, FC = -3.3; N.N, FC = -4.3; N.N, FC-3,7),
rate (FDR) < 0.05 and log fold changes (FC) > |1| (Supplementary synthesis of gibberellins (GA3ox4, FC = -5.5), synthesis of brassinoste­
Table 2) were differentially expressed between the contrasting E0 cate­ roids (BR6OX1, FC = -2.3) and flavonols (F3H, FC = -5.4). During fruit
gories (low and high). These genes were used for genetic ontology (GO) growth, the processes of cell division and expansion occur simulta­
analysis and were associated with different GO terms. After this GO neously (Inzé and Veylder, 2006). For these two processes to take place,
analysis, it was possible to observe two groups of labeled GO terms for the DNA replication process must occur beforehand. Results of this study
each E0 category (low and high) that have a close biological relationship revealed higher expression of genes related to this process (POL2A, N.N,
with each category (Fig. 4). Subsequently, from the 1806 genes, a se­ POLA2) in the low E0 samples, thus indicative that the low E0 category
lection of 26 genes were further used for the biological interpretation of samples are still in the stage of full cell division and expansion. Avocado
the differences. These genes were selected for presenting a marked dif­ growth is different from the growth of other fleshy fruits since cell di­
ference in their expression between the E0 categories (low and high) vision continues over a relatively long development period, decreasing
with log fold changes ranging from -5.5 to 4.3 and an interesting bio­ as the fruit reaches maturity (Cowan et al., 2001). The samples
logical functionality (Table 1). belonging to the low E0 category (E0 < 5) in addition presented a higher
As previously described, to find and select potential early E0 bio­ expression of genes related to metabolic pathways related to fruit
markers, partial least squares regression analysis (PLS-R) considering growth including cell division and expansion (Fig. 5) both regulated by
the estimated values from the model of the 36 E0 samples and all genes auxin and gibberellins and with input from other hormones including
brassinosteroids (Fenn and Giovannoni, 2021). The BR6OX1

Fig. 3. Principal component analysis (PCA) of only the low (E0 < 5) and high (E0 > 10) ranges of early harvest fruit (dry matter content of ≥ 23 – 26 %). (a) Score
plot; (b) Heat map of similarity based on Pearson’s correlation between low E0 and high E0 samples.

5
I. Hernández et al. Postharvest Biology and Technology 185 (2022) 111806

Fig. 4. Annotation and relevant GO terms of

the transcriptome of high E0 versus low E0
samples, indicating the relevance of the Gen­
eRatio and the adjusted P of each relevant gene
belonging to each category. GO analysis
included all differentially expressed genes ob­
tained comparing high E0 samples and low E0
samples with FDR < 0.05 and logFC > |1|
(1806 genes). Of them, only 1316 genes have a
GO term associated. These 1316 genes were
divided in genes overrepresented in high E0
samples (497 genes) and low E0 samples (819
genes). The Gene Ratio is the size of the dot of
each GO category representing the number of
genes from our input list that match the GO
term.

Table 1
Selected candidate genes for high and low biological age. The count values of E0 high and E0 low are normalized.
GeneID *Low E0 *High E0 logFC GeneName Description Pathway association

Peam|DryGeneModel_21343 2 25 4.3 ACS1 ACC synthase 1 Ethylene biosynthesis

Peam|DryGeneModel_8209 3 22 3.3 ERF110 ethylene response factor 110 Ethylene response
Peam|DryGeneModel_5803 12 94 3.3 CAD9 cinnamyl alcohol dehydrogenase 9 Phenylpropanoid biosynthesis
Peam|DryGeneModel_8327 18 108 2.7 ZEP zeaxanthin epoxidase, chloroplastic-like* Abscisic acid biosynthesis
Peam|DryGeneModel_1776 381 1900 2.6 NAC72 NAC domain-containing protein 72* Transcription factor
Peam|DryGeneModel_11291 244 48 − 2.1 F3H flavanone 3-hydroxylase Flavonoid biosynthesis
Peam|DryGeneModel_15100 178 33 − 2.2 POL2A DNA polymerase epsilon catalytic subunit DNA replication complex
Peam|DryGeneModel_12712 7 1 − 2.2 EXPA1 expansin A1 Cell wall remodelation
Peam|DryGeneModel_11039 110 19 − 2.3 BR6OX1 brassinosteroid-6-oxidase 1 Brassinosteroid biosynthesis
Peam|DryGeneModel_19523 4126 675 − 2.3 EXPA8 expansin A8 Cell wall remodelation
Peam|DryGeneModel_5971 18 3 − 2.4 PIN4 Auxin efflux carrier family protein Auxin efflux transport
Peam|DryGeneModel_19482 68 9 − 2.6 — DNA primases DNA replication complex
Peam|DryGeneModel_13952 95 11 − 2.7 PIN1 Auxin efflux carrier family protein Auxin efflux transport
Peam|DryGeneModel_21461 50 6 − 2.7 E2F3 DP-E2F-like 1 Transcription factor
Peam|DryGeneModel_17215 180 19 − 3.0 POLA2 DNA polymerase alpha 2 DNA replication complex
Peam|DryGeneModel_4356 132 13 − 3.0 XTH33 Xyloglucan endotransglucosylase/hydrolase 33* Cell wall remodelation
Peam|DryGeneModel_1282 74 6 − 3.3 EXPA8 expansin A8 Cell wall remodelation
Peam|DryGeneModel_11760 57 3 − 3.7 E2F3 E2F transcription factor 3 Transcription factor
Peam|DryGeneModel_64 105 7 − 3.7 — Pectin lyase-like superfamily protein Cell wall remodelation
Peam|DryGeneModel_22355 222 14 − 3.7 TRN2 protein TORNADO 2* Auxin efflux transport
Peam|DryGeneModel_374 585 26 − 4.2 TRN1 protein TORNADO 1* Auxin efflux transport
Peam|DryGeneModel_11347 572 23 − 4.3 XTH5 Xyloglucanendotransglucosylase/hydrolase 5 Cell wall remodelation
Peam|DryGeneModel_11106 102 4 − 4.3 — Pectate lyase family protein Cell wall remodelation
Peam|DryGeneModel_1401 15 1 − 4.3 CYCD4;1 CYCLIN D4;1 Cell cycle
Peam|DryGeneModel_6064 51 1 − 5.4 F3H flavanone 3-hydroxylase Flavonoid biosynthesis
Peam|DryGeneModel_13115 26 0 − 5.5 GA3ox4 gibberellin 3-beta-dioxygenase 4* Gibberellin biosynthesis
*
Normalized counts (FPKM).

(brassinosteroid-6-oxidase 1) gene showed higher expression in the low and GA20ox has been reported (McAtee et al., 2013; Kumar et al., 2014).
E0 samples compared to the high E0 samples (FC = -2.3) and participates Fenn and Giovannoni (2021) indicates that GA promotes cell expansion
in the metabolic pathway of brassinosteroid biosynthesis. Brassinoste­ in a synergistic manner with auxins in a large number of fleshy fruits.
roids (BR) are plant hormones that participate in cell division and It is important to mention that target genes of auxin and GA include
elongation (Hu et al., 2000), presenting a close relationship with the cell wall remodeling enzymes including expansins and pectate lyases
functionality of expansins (EXPA5) (Park et al., 2010). In addition, many (McAtee et al., 2013; Kumar et al., 2014; Fenn and Giovannoni, 2021).
of the hormonal functions of auxins have been shown to act in synergy Our results showed many genes related to expansins and pectin lyases to
with the functions of brassinosteroids (Park et al., 2010). Our results, be higher expressed in low E0 samples. For instance, three genes
next to revealing a higher expression of BR6OX1 in the low E0 samples, (EXPA1, EXPA8, and EXPA8) related to expansin functionality displayed
also showed higher expression of genes involved in auxin transport fold changes of -2.2, -2.3 and -3.3 (higher expressed in low E0 samples),
(PIN1, FC = -2.7; TRN2, FC = -3.7; TRN1 (FC = -4.2) (Table 1). Another respectively (Table 1). Expansins are capable of loosening cell walls in a
gene that was observed higher expressed in fruit with low E0 was related non-enzymatic but pH-dependent manner (Marowa et al., 2016). This
to gibberellin biosynthesis (GA3ox4, FC = -5.5). The function of gib­ relationship is attributed to the ability of auxins to stimulate the syn­
berellins (GA) is to promote cell elongation and / or division (Xu et al., thesis of proton pumps that causes the acidification of the apoplast and
2016). According to Chapman et al. (2012) gibberellin biosynthesis is therefore promote the activity of expansins (Majda and Robert. 2018).
necessary for the normal auxin response; therefore, it could be that IAA The plasma membrane hyperpolarization process caused by stimulation
acts interdependently with gibberellin pathways to regulate the of H + -ATPase proton pumps has also been reported to be regulated by
expression of growth-associated genes during cell expansion. In tomato, auxin-inducible SMALL AUXIN UP-RNA (SAUR) proteins (Spartz et al.,
during the onset of fruit cell expansion, increased expression of GA3ox 2014), this gene was found in our study among the 1088 genes

6
I. Hernández et al. Postharvest Biology and Technology 185 (2022) 111806

Fig. 5. Reconstruction of metabolic pathways

expressed at different biological ages (E0).
Pathways within the green dashed line: genes
expressed in the low E0 category. Pathways
within the red dashed line: genes expressed in
the high E0 category. (A) Brassinosteroids
metabolic pathway, (B) Metabolic pathway of
auxin transport, (C) Metabolic route of DNA
replication, (D) Gibberellin metabolic pathway,
(E) Flavanol metabolic pathway, (F) Phenyl­
propanoid metabolic pathway, (G) Isoprenoid
metabolic pathway, (H) ABA metabolic
pathway, (I) Ethylene metabolic pathway, (J)
Metabolic pathway of cell wall remodeling. (For
interpretation of the references to colour in the
Figure, the reader is referred to the web version
of this article).

displaying differential expression (Supplementary Table 2). Although biosynthesis and response to abscisic acid (ZEP, FC = 2.7), ethylene
the genes found with the highest expression in the low E0 fruit were biosynthesis and response (ACS, FC = 4.3 and ERF110, FC = 3.3) and
related to auxin transport rather than their functionality, these actions phenylpropanoid biosynthesis (CAD9, FC = 3.3). These results coincide
are related to each other. In addition to a higher expression of expansin with the higher potential of high E0 fruit to trigger the different meta­
genes in low E0 samples, an overexpression of xyloglucan endo­ bolic pathways activated during ripening. The high E0 category repre­
transglycosylase (XTH33, FC = -3.3) and XTH5 (FC = -4.3) were also sents mature fruit which has the competence to ripen but has not yet
found (Table 1). These are also cell wall remodeling enzymes involved in started the ripening process (McAtee et al., 2013; Kumar et al., 2014).
cell expansion. Pectin lyase genes (FC=-3.7) were also higher expressed Auxin and cytokinins seem to be the key regulators of fruit maturation
in the E0 samples. Remodeling enzymes such as expansins, endo­ (McAtee et al., 2013) based on tomato mutant studies displaying the
transglycosylase and pectinases have been reported to participate in non-ripening phenotype that maintained higher levels of auxins and
phytohormone mediated cell expansion, since fruit growth require cytokinins at the breaker stage compared to the wild type fruit (Rolle
loosening of the cell wall matrix for the deposition of new cell wall and Chism, 1989) and on apple where suppression of a rin-like
components (Sánchez-Rodríguez et al., 2010; Cosgrove, 2016). MADs-box gene resulted in high auxin concentrations during matura­
Genes related to flavonoid biosynthesis (F3H, FC = -2.1 and F3H, FC tion and fruit did not ripen (Ireland et al., 2013). Our results only
= -5.4) were also higher expressed in the low E0 fruit category. Although revealed genes related to auxin transport but as indicated in the next
it has been reported by Xoca-Orozco et al. (2019) that in general the section on the targeted hormone analysis, auxin and cytokinin concen­
greatest biosynthesis of phenolic compounds such as flavonoids or trations were higher in high E0 samples. ABA together with ethylene
phenylpropanoids occurs when the avocado fruit reaches physiological play a crucial role in inducing ripening (Zhang et al., 2009). Our results
maturity. The content of phenolics in the avocado mesocarp varies showed higher expression of genes related to the biosynthesis (ACS1,
considerably depending on the levels of abiotic and biotic stress to FC = 4.3) and perception (ERF110, FC = 3.3) of ethylene and ABA (ZEP,
which this fruit was subjected, regardless its size and maturity (Tru­ FC = 2.7) in the high E0 samples. Even though, the role of ethylene and
jillo-Mayol et al., 2020). According to the study carried out by Figueroa ABA during ripening of climacteric fruit has been extensively reported,
et al. (2018) the main flavanols in avocado are rutin, isorhamnetin, studies on the role of both hormones during fruit maturation are still
narirutin and quercetin, acting as powerful free radical stabilizers. scarce (Kumar et al., 2014).
Campos et al. (2020) has reported that the profile and content of Avocado, being a climacteric fruit, is characterized by presenting an
phenolic compounds increases and changes as the fruit transits from increase in respiration rate and ethylene production at the beginning of
maturity to edible ripeness, finding sixteen new phenolic compounds, of ripening (Seymour et al., 1993). Previous studies on differences at
which p-coumaric acid, caffeic acid and their derivatives are the most harvest displaying different ripening behavior could not associate these
important. Research work carried out by Di Stefano et al. (2017) and differences to endogenous ethylene levels but to other metabolites such
Hurtado-Fernández et al. (2011) reported the same trend on the evo­ as amino acids (Pedreschi et al., 2014; Fuentealba et al., 2017; Uarrota
lution of phenolic compounds in avocado but with some differences in et al., 2019). Up to now, the information on the interaction of hormonal
the reported compounds. This may be due to factors, such as genetics signals and the changes in the primary metabolism during the ripening
(avocado cultivars), climatic conditions or management conditions of of ′ Hass′ avocado is quite limited (Pedreschi et al., 2019). Despite this,
the crop. In the recent research by Hernández et al. (2021) one of the 17 ethylene has been widely studied as the hormone responsible for the
metabolites with the highest correlation to the E0 parameter was quinic ripening process, but other hormones can act sequentially and / or
acid, showing a positive correlation. Quinic acid is an intermediate in synergistically with ethylene in controlling fruit ripening. ABA has been
the shikimic acid pathway (Dewick, 2002), the main pathway in the shown to interact with ethylene production, improving its production in
biosynthesis of flavonoids. The low E0 category presented higher various climacteric fruits, although the way the interaction occurs
expression of a gene (F3H), this enzyme is part of the flavonoid currently is not entirely clear (Meyer et al., 2017). In a study carried out
biosynthesis pathway (Table 1). on ′ Granny Smith′ apples, it was observed that 1–2 months before the
The over-expressed metabolic pathways for the high E0 category are commercial harvest begins, a noticeable increase in 1-
displayed in Fig. 5 and are mostly stress-related genes such as aminocyclopropane-1-carboxylic acid (ACC) and in parallel the increase

7
I. Hernández et al. Postharvest Biology and Technology 185 (2022) 111806

in ABA and endogenous ethylene occurs (Lara and Vendrell, 2000). This relevant from commercial maturity, during room and/or cold storage
could indicate that the increase in ACC synthesis may be the point that and ripening of avocado mesocarp (cv. Bacon).
signals the transition from immature preclimacteric to mature pre­ The results were analyzed using a t-test with a significance level of
climacteric fruit and if endogenous ABA is involved in ACC synthesis it p < 0.05. Only IAA and tZ were significant between the high E0 and low
could be used as a maturity marker in ′ Granny Smith′ apples (Lara and E0 categories, with higher concentrations of both hormones in the high
Vendrell, 2000). Due to the high heterogeneity reported in ′ Hass′ avo­ E0 category (Fig. 7). Previous studies have reported that auxin and cy­
cado (Pedreschi et al., 2014; Hernández et al., 2017; Fuentealba et al., tokinins are key regulators of fruit maturation (McAtee et al., 2013).
2017) it is very likely that fruit with a high E0 are in a more advanced Physiologically ready fruit maintains high concentrations of both auxin
stage of preclimaterium maturity and for that reason have a higher and cytokinins during fruit maturation, but fruit does not ripen (Ireland
content of ABA and ethylene. et al., 2013). Thus, our results point to high E0 samples having reached
In addition to the genes related to ABA and ethylene (ZEP, ACS, and maturity and having the potential to ripen.
ERF110), also genes pertaining to the phenylpropanoids machinery For the other six hormones, no significant differences were observed
(CAD9, FC = 3.3) were over-expressed in the high E0 category. These between the low and high E0 categories although showing trends of
metabolites are being synthesized when the avocado reaches physio­ increasing (ABA, DHZ and SA) and decreasing (JA, GA1 and GA4) levels
logical maturity with metabolites such as epicatechin and persin with increasing levels of E0. Our results at the hormone level agree with
becoming more relevant. These metabolites decrease as ripening pro­ the results found at the gene level, where genes related to ethylene
gresses in the peel (Xoca-Orozco et al., 2019); however, Campos et al. biosynthesis and perception and ABA were higher expressed in the high
(2020) have recently reported that as ripening progresses some other E0 category. The current work did, to our knowledge for the first time,
phenolics are synthesized in the mesocarp. study the involvement of hormones other than auxin and cytokinins
(that remain high during maturation) during maturation of ′ Hass′ avo­
3.4. Internal validation of candidate genes as potential early biomarkers cado fruit. It has been reported jasmonic acid acting antagonistically to
of biological age ABA in non-climacteric fruit (Garrido-Bigotes et al., 2018), our results at
maturation seem to reveal similar observations of a negative correlation
In order to validate the genes proposed as candidate biomarkers of between ABA and JA, with high E0 samples displaying a trend towards
biological age (as reflected by their E0) that presented a marked differ­ lower concentrations of JA.
ential expression between the high E0 and low E0 categories, an internal Several works on different climacteric fruits have reported the action
validation was also carried out by quantitative real-time PCR (qRT- of ABA and ethylene in relation to fruit ripening (Forlani et al., 2019;
PCR). Meyer et al., 2017). For avocado fruit ripening, most of the work has
Five genes (PIN1, TRN1, EXPA8, XTH5 and BR6OX1) that presented focused on ethylene and to a minor extent on ABA (Meyer et al., 2017).
higher expression in the low E0 category and one gene (NAC072) that Despite the many studies on the ripening of climacteric fruits, the signs
presented a higher expression in the high E0 category were selected. of the onset of ripening remain unknown, but the idea that the ripening
Results were completely consistent with the transcriptomic data. Pear­ of the fruit is regulated by the interaction of a set of hormonal factors
son’s correlations (R > 0.77) showed that there is a positive correlation becomes more documented (Giovannoni et al., 2017).
between the two methods, validating the transcriptome analysis (Fig. 6). Auxins and gibberellins promote fruit softening by activating en­
zymes that promote cell expansion, observing a clear increase in their
concentration as the avocado fruit ripens (Majda and Robert, 2018;
3.5. Hormone analysis and E0 Guzmán et al., 2021). Vincent et al. (2020) reported increases of auxin
and GA1 content during avocado softening at 25 ◦ C, as well as cytokinin
Hormone analysis was performed on 18 randomly selected biopsies accumulation (iP and trans-zeatin riboside). Jasmonic and salicylic acid
that belong to the high and low E0 categories of biological age. The are known to actively participate in the plant immune system
hormone quantification (expressed on a dry weight basis as μg kg− 1) of (Xoca-Orozco et al., 2019). Additionally, jasmonic acid could be
eight relevant hormones included an auxin – indole acetic acid (IAA), an involved in ripening, as its precursor 12-oxo-phytodienoic acid (OPDA)
active gibberellin (GA1 and GA4), abscisic acid (ABA), jasmonic acid increases during ripening or over-ripening. However, the study of Vin­
(JA), salicylic acid (SA) and the cytokinins trans-zeatin (tZ) and dihy­ cent et al. (2020) did not report a direct correlation with free or con­
drozeatin (DHZ). Selection of these eight hormones was based on the jugated jasmonates in avocado fruit during ripening. Therefore, the role
results of Vincent et al. (2020) that reported these hormones to be

Fig. 6. Correlation between qRT-PCR and

RNAseq gene expression. The figure shows six
candidate genes for biological age phenotype
identified in this study: (A) NAC072, (B) PIN1,
(C) TRN1, (D) EXPA8, (E) XTH5 (F) BR6ox1.
The left Y axis shows the RNAseq expression
values (blue continuous line), while the right Y
axis shows the qRT-PCR relative expression
values normalized to Pam F-ACP gene (grey
bars), the X axis indicates the biological age
categories (E0 low and E0 high). R indicates the
Pearson correlation coefficient between RNA­
seq and qRT-PCR expression values. RNAseq
expression values are represented by the blue
continuous line and qRT-PCR relative expres­
sion values are represented by grey bars. (For
interpretation of the references to colour in the
Figure, the reader is referred to the web version
of this article).

8
I. Hernández et al. Postharvest Biology and Technology 185 (2022) 111806

Fig. 7. Quantification of hormones (μg hormone kg− 1 DW) at harvest of samples with extreme ranges of biological age (E0), low (E0 <5) and high (E0> 10). Data
were analyzed by t-test with a significance (*) p < 0.05 was performed.

OPDA could play during the ripening process is not very clear. Finally, Data availability
there are no recent studies clearly indicating the role of cytokinins
during avocado ripening. Also, the most recent study of Vincent et al. The authors are unable or have chosen not to specify which data has
(2020) only observed a sporadic increase in cytokinins (including iso­ been used.
pentenyl adenine (2-iP) and its precursor isopentenyl adenosine (IPA),
as well as trans-zeatin riboside (ZR) during ripening). Although more CRediT authorship contribution statement
studies are needed to elucidate the behavior of cytokinins during avo­
cado fruit ripening, the current study revealed a significantly higher Ignacia Hernández (IH): Methodology implementation, experiment
content of cytokinin (trans-zeatin) in high E0 fruit compared to the low execution, data collection, data analysis and modelling and interpreta­
E0 fruit which could be an important lead for future research on the tion and manuscript writing. Virgilio Uarrota (VU): Statistical data
activity of this hormone in avocado fruit. It might be relevant to indicate analysis and interpretation and manuscript editing/revision. Diego
that fruit maturity has been reached and the fruit has developed all the Paredes (DP): Methodology implementation, critical review on
required machinery needed to trigger ripening. modelling and manuscript editing/revision. Claudia Fuentealba (CF):
Methodology implementation and manuscript editing/writing. Bruno
4. Conclusions Defilippi (BD): Methodology implementation, data analysis and inter­
pretation and manuscript editing/revision. Reinaldo Campos-Vargas
Using the RNA-seq technique it was possible to provide robust in­ (RC): Methodology implementation, data analysis and interpretation
formation in the search for possible biological age markers of ′ Hass′ and manuscript editing/revision. Gerardo Nuñez (GN): transcriptomics
avocado. Through partial least squares regression analysis (PLS-R), a analysis, pPCR analysis, data analysis and manuscript editing/revision.
total of 46 candidate genes were obtained, which showed a significant Esther Carrera (EC): targeted hormone analysis, manuscript editing/
correlation with biological age (in terms of E0). This indicates that the revision. Claudio Meneses (CM): Methodology implementation, data
transcriptomics platform can provide robust information to find bio­ modelling and analysis and interpretation and manuscript editing/
markers of E0, compared to other omics platforms such as metabolomics revision. Maarten Hertog (MH): Conception and design, methodology
or targeted hormone analysis. In addition, it was observed as in previous implementation, data interpretation and manuscript writing/editing/
works, that early harvest batches continue to present greater problems revision. Romina Pedreschi (RP): Conception and design, methodol­
in terms of heterogeneity, as greater differences in biological ages were ogy implementation, data interpretation, funding acquisition and
observed. With respect to the biological age (in terms of E0) of the manuscript writing/editing/revision.
samples, a clear difference in the metabolic pathways expressed in
samples with high E0 and low E0 could be observed. Through differential
expression analysis, it was observed that low E0 fruit showed a higher Declaration of Competing Interest
expression of genes associated with DNA replication together with a
higher auxin transport and increased synthesis of gibberellins, cell wall The authors declare no conflicts of interest.
remodeling and flavonols. On the other hand, high E0 fruit showed a
higher expression of ABA, ethylene and phenylpropanoid synthesis.
Acknowledgements
Likewise, targeted hormone analysis revealed higher concentrations of
GA1, GA4 and JA in the low E0 phenotype and higher but not significant
This study was financially supported by Fondecyt No 1180303 and
concentrations of ABA, DHZ and SA in high E0 and significant for trans-
PCI Redes Bio001 grants from ANID (Chile). The authors wish to thank
zeatin and IAA in this phenotype. This is the first study in ′ Hass′ avocado
the ′ Hass′ Avocado Committee of Chile and the different orchards that
to reveal the complex hormonal interplay during maturation, where the
granted access for sampling the fruit and to VRIEA-PUCV grant 039.426.
transition from fruit growth to fruit maturation is key to assess biological
age of the fruit potentially enabling improved postharvest management
of the commercial batches. Appendix A. Supplementary data

Supplementary material related to this article can be found, in the

online version, at doi:https://doi.org/10.1016/j.postharvbio.2021.111
806.

9
I. Hernández et al. Postharvest Biology and Technology 185 (2022) 111806

References Hu, Y., Bao, F., Li, J., 2000. Promotive effect of brassinosteroids on cell division involves
a distinct CycD3-induction pathway in Arabidopsis. Plant J. 24, 693–701. https://
doi.org/10.1046/j.1365-313x.2000.00915.x.
Adato, I., Gazit, S., 1977. Changes in the initiation of climacteric ethylene in harvested
Hurtado-Fernández, E., Carrasco-Pancorbo, A., Fernández-Gutiérrez, A., 2011. Profiling
avocado fruits during their development. J. Sci. Food Agric. 28, 240–242. https://
LC-DAD-ESI-TOF MS method for the determination of phenolic metabolites from
doi.org/10.1002/jsfa.2740280303.
avocado (Persea americana). J. Agric. Food Chem. 59 (6), 2255–2267. https://doi.
Andrews, A., 2010. FastQC: A Quality Control Tool for High Throughput Sequence Data.
org/10.1021/jf104276a.
Available online at: http://www.bioinformatics.babraham.ac.uk/projects/fastqc.
Inzé, D., Veylder, L.D., 2006. Cell cycle regulation in plant development. Annu. Rev.
Campos, D., Teran-Hilares, F., Chirinos, R., Aguilar-Galvez, A., García-Ríos, D., Pacheco-
Genet. 40, 77–105. https://doi.org/10.1146/annurev.genet.40.110405.090431.
Avalos, A., Pedreschi, R., 2020. Bioactive compounds and antioxidant activity from
Ireland, H.S., Yao, J.L., Tomes, S., Sutherland, P.W., Nieuwen-huizen, N., Gunaseelan, K.,
harvest to edible ripeness of avocado cv. Hass (Persea americana) throughout the
2013. Apple SEPALLATA1/2-like genes control fruit flesh development and ripening.
harvest seasons. Int. J. Food Sci. Technol. 55, 2208–2218. https://doi.org/10.1111/
Plant J. 73, 1044–1056. https://doi.org/10.1111/tpj.12094.
ijfs.14474.
Jeong, J., Huber, D.J., 2004. Suppression of avocado (Persea americana Mill.) fruit
Chapman, E.J., Greenham, K., Castillejo, C., Sartor, R., Bialy, A., Sun, T.P., Estelle, M.,
softening and changes in cell wall matrix polysaccharides and enzyme activities:
2012. Hypocotyl transcriptome reveals auxin regulation of growth-promoting genes
differential responses to 1-MCP and delayed ethylene application. J. Am. Soc. Hortic.
through GA-dependent and -independent pathways. PLoS One 7, e36210. https://
Sci. 129, 752–759. https://doi.org/10.21273/JASHS.129.5.0752.
doi.org/10.1371/journal.pone.0036210.
Kanehisa, M., Furumichi, M., Sato, Y., Ishiguro-Watanabe, M., Tanabe, M., 2021. KEGG:
Chiesa, M., Colombo, G.I., Piacentini, L., 2018. DaMiRseq - an R/Bioconductor package
integrating viruses and cellular organisms. Nucleic Acids Res. 49, D545–D551.
for data mining of RNA-Seq data: normalization, feature selection and classification.
https://doi.org/10.1093/nar/gkaa970.
Bioinformatics 34 (8), 1416–1418. https://doi.org/10.1093/bioinformatics/btx795.
Kassambara, A., Mundt, F., 2020. Factoextra: Extract and Visualize the Results of
Cosgrove, D., 2016. Plant cell wall extensibility: connecting plant cell growth with cell
Multivariate Data Analyses. R Package Version 1.0.7.. https://cloud.r-project.org/
wall structure, mechanics, and the action of wall-modifying enzymes. J. Exp. Bot. 67,
package=factoextra.
463–476. https://doi.org/10.1093/jxb/erv511.
Kucheryavskiy, S., 2020. “Mdatools - R Package for Chemometrics”._Chemometrics and
Cowan, A.K., Cripps, R.F., Richings, E.W., Taylor, N.J., 2001. Fruit size: Towards
Intelligent Laboratory Systems_, *198*. https://doi.org/10.1016/j.
understanding of the metabolic control of fruit growth using avocado as a model
chemolab.2020.103937.
system. Physiol. Plant. 111, 127–136. https://doi.org/10.1034/j.1399-
Kuhn, M., 2020. Caret: Classification and Regression Training. R Package Version 6.0-86.
3054.2001.1110201.x.
https://github.com/topepo/caret/.
Dewick, P.M., 2002. Medicinal Natural Products: a Biosynthetic Approach. John Wiley &
Kumar, R., Khurana, A., Sharma, A.K., 2014. Role of plant hormones and their interplay
Sons. https://doi.org/10.1002/9780470742761.
in development and ripening of fleshy fruit. J. Exp. Bot. 65, 4561–4575. https://doi.
Di Stefano, V., Avellone, G., Bongiorno, D., Indelicato, S., Massenti, R., Lo Bianco, R.,
org/10.1093/jxb/eru277.
2017. Quantitative evaluation of the phenolic profile in fruits of six avocado (Persea
Lara, I., Vendrell, M., 2000. Development of ethylene-synthesizing capacity in
americana) cultivars by ultra-high-performance liquid chromatography-heated
preclimacteric apples: interaction between abscisic acid and ethylene. J. Am. Soc.
electrospray-mass spectrometry. Int. J. Food Prop. 20, 1302–1312. https://doi.org/
Hortic. Sci. 125 (4), 505–512. https://doi.org/10.21273/JASHS.125.4.505.
10.1080/10942912.2016.1208225.
Liao, Y., Smyth, G.K., Shi, W., 2019. The R package Rsubread is easier, faster, cheaper
Dobin, A., Davis, C.A., Schlesinger, F., Drenkow, J., Zaleski, C., Jha, S., Batut, P.,
and better for alignment and quantification of RNA sequencing reads. Nucleic Acids
Chaisson, M., Gingeras, T.R., 2013. STAR: ultrafast universal RNA-seq aligner.
Res. 47, e47. https://doi.org/10.1093/nar/gkz114.
Bioinformatics 29, 15–21. https://doi.org/10.1093/bioinformatics/bts635.
Majda, M., Robert, S., 2018. The role of auxin in cell wall expansion. Int. J. Mol. Sci. 19,
Dodt, M., Roehr, J.T., Ahmed, R., Dieterich, C., 2012. Flexbar – flexible barcode and
951. https://doi.org/10.3390/ijms19040951.
adapter processing for next-generation sequencing platforms. Biology 1, 895–905.
Marowa, P., Ding, A., Kong, Y., 2016. Expansins: roles in plant growth and potential
https://doi.org/10.3390/biology1030895.
applications in crop improvement. Plant Cell Rep. 35 (5), 949–965. https://doi.org/
Du, Z., Zhou, X., Ling, Y., Zhang, Z., Su, Z., 2010. agriGO: a GO analysis toolkit for the
10.1007/s00299-016-1948-4.
agricultural community. Nucleic Acids Res. 38, W64–W70. https://doi.org/
McAtee, P., Karim, S., Schaffer, R., David, K., 2013. A dynamic interplay between
10.1093/nar/gkq310.
phytohormones is required for fruit development, maturation and ripening. Front.
Fenn, M., Giovannoni, J., 2021. Phytohormones in fruit development and maturation.
Plant Sci. 4, 79 https://doi.org/10.3389/fpls.2013.00079.
Plant J. 105, 446–458. https://doi.org/10.1111/tpj.15112.
Mevik, B.H., Wehrens, R., HovdeLiland, K., 2020. Pls: Partial Least Squares and Principal
Figueroa, J.G., Borrás-Linares, I., Lozano-Sánchez, J., Segura-Carretero, A., 2018.
Component Regression. R Package Version 2.7-3. https://github.com/khliland/pls.
Comprehensive characterization of phenolic and other polar compounds in the seed
Meyer, M.J., Chope, G.A., Terry, L.A., 2017. Investigation into the role of endogenous
and seed coat of avocado by HPLC-DAD-ESI-QTOF-MS. Food Res. Int. 105, 752–763.
abscisic acid during ripening of imported avocado cv. Hass. J. Sci. Food Agric. 97,
https://doi.org/10.1016/j.foodres.2017.11.082.
3656–3664. https://doi.org/10.1002/jsfa.8225.
Forlani, S., Masiero, S., Mizzotti, C., 2019. Fruit ripening: the role of hormones, cell wall
Neilson, J., Lagüe, M., Thomson, S., Aurousseau, F., Murphy, A.M., Bizimungu, B.,
modifications, and their relationship with pathogens. J. Exp. Bot. 70, 2993–3006.
Deveaux, V., Bègue, Y., Jacobs, J., Tai, H.H., 2017. Gene expression profiles
https://doi.org/10.1093/jxb/erz112, 11606.
predictive of cold-induced sweetening in potato. Funct. Integr. Genomics 17 (4),
Fuentealba, C., Hernández, I., Olaeta, J.A., Defilippi, B., Meneses, C., Campos-Vargas, R.,
459–476. https://doi.org/10.1007/s10142-017-0549-9.
Lurie, S., Carpentier, S., Pedreschi, R., 2017. New insights into the heterogeneous
Ochoa-Ascencioa, S., Hertog, M.L.A.T.M., Nicolaï, B.M., 2009. Modelling the transient
ripening in Hass avocado via LC MSMS proteomics. Postharvest Biol. Technol. 132,
effect of 1-MCP on ‘Hass’ avocado softening: a Mexican comparative study.
51–61. https://doi.org/10.1016/j.postharvbio.2017.06.001.
Postharvest Biol. Technol. 51, 62–72. https://doi.org/10.1016/j.
García, C.J., Gil, M.I., Tomas-Barberan, F.A., 2018. LC–MS untargeted metabolomics
postharvbio.2008.06.002.
reveals early biomarkers to predict browning of fresh-cut lettuce. Postharvest Biol.
Oficina de Estudios y Política Agrarias (ODEPA), 2021. Catastro Frutícola Cirén – Odepa.
Technol. 146, 9–17. https://doi.org/10.1016/j.postharvbio.2018.07.011.
Visited march 17, 2021. https://www.odepa.gob.cl/estadisticas-del-sector/catastr
García, C.J., Gil, M.I., Tomas-Barberan, F.A., 2019. Targeted metabolomics analysis and
os-fruticolas/catastro-fruticola-ciren-odepa.
identification of biomarkers for predicting Browning of fresh-cut lettuce. J. Agric.
Park, C.H., Kim, T.W., Son, S.H., Hwang, J.Y., Lee, S.C., Chang, S.C., Kim, S.H., Kim, S.
Food Chem. 67, 5908–5917. https://doi.org/10.1021/acs.jafc.9b01539.
W., Kim, S.K., 2010. Brassinosteroids control AtEXPA5 gene expression in
Garrido-Bigotes, A., Figueroa, C.R., Figueroa, P.M., 2018. Jasmonate metabolism and its
Arabidopsis thaliana. Phytochemistry 71 (4), 380–387. https://doi.org/10.1016/j.
relationship with abscisic acid during strawberry fruit development and ripening.
phytochem.2009.11.003.
J. Plant Growth Regul. 37, 101–113. https://doi.org/10.1007/s00344-017-9710-x.
Pedreschi, R., Muñoz, P., Robledo, P., Becerra, C., Defilippi, B., van Eekelen, H.,
Giovannoni, J., Nguyen, C., Ampofo, B., Zhong, S., Fei, Z., 2017. The epigenome and
Mumm, R., Westra, E., de Vos, R., 2014. Metabolomisc analysis of postharvest
transcriptional dynamics of fruit ripening. Annu. Rev. Plant Biol. 28 (68), 61–84.
ripening heterogeneity of Hass avocados. Postharvest Biol. Technol. 92, 172–179.
https://doi.org/10.1146/annurev-arplant-042916-040906.
https://doi.org/10.1016/j.postharvbio.2014.01.024.
Guzmán, Y., Pugliese, B., González, C.V., Travaglia, C., Bottini, R., Berli, F., 2021. Spray
Pedreschi, R., Uarrota, V., Fuentealba, C., Alvaro, J., Olmedo, P., Defilippi, B.,
with plant growth regulators at full bloom may improve quality for storage of’
Meneses, C., Campos-Vargas, R., 2019. Primary metabolism in avocado fruit. Front.
Superior Seedless’ table grapes by modifying the vascular system of the bunch.
Plant Sci. 10, 795 https://doi.org/10.3389/fpls.2019.00795.
Postharvest Biol. Technol. 176, 111522 https://doi.org/10.1016/j.
R Core Team, 2021. R: a Language and Environment for Statistical Computing.
postharvbio.2021.111522.
RFoundation for Statistical Computing, Vienna, Austria. URL. https://www.R-pro
Gwanpua, S.G., Qian, Z., East, A., 2018. Modelling ethylene regulated changes in ‘Hass’
ject.org/.
avocado quality. Postharvest Biol. Technol. 136, 12–22. https://doi.org/10.1016/j.
Rendón-Anaya, M., Ibarra-Laclette, E., Méndez Bravo, A., Lan, T., Zheng, C., Carretero-
postharvbio.2017.10.002.
Paulet, L., Perez-Torres, C.A., Chacón-López, A., Hernandez-Guzmán, G., Chang, T.
Hernández, I., Fuentealba, C., Olaeta, J.A., Poblete, C., Defilippi, B., González, M.,
H., Farr, K.M., Barbazuk, W.B., Chamala, S., Mutwil, M., Shivhare, D., Alvarez-
Campos-Vargas, R., Lurie, S., Pedreschi, R., 2017. Effects of heat shock and nitrogen
Ponce, D., Mitter, N., Hayward, A., Fletcher, S., Rozas, J., Sánchez-Gracia, A.,
shock pre-treatments on ripening heterogeneity of Hass avocados stored in
Kuhn, D., Barrientos-Priego, A.F., Salojärvi, J., Librado, P., Sankoff, D., Herrera-
controlled atmosphere. Sci. Hort. 225, 408–415. https://doi.org/10.1016/j.
Estrella, A., Albert, V.A., Herrera-Estrella, L., 2019. The avocado genome informs
scienta.2017.07.025.
deep angiosperm phylogeny, highlights introgressive hybridization, and reveals
Hernández, I., Uarrota, V., Paredes, D., Fuentealba, C., Defilippi, B.G., Campos-
pathogen-influenced gene space adaptation. Proc. Natl. Acad. Sci. 34, 17081–17089.
Vargas, R., Meneses, C., Hertog, M.L.A.T.M., Pedreschi, P., 2021. Can metabolites at
https://doi.org/10.1073/pnas.1822129116.
harvest be used as physiological markers for modelling the softening behaviour of
Rivera, S.A., Ferreyra, R., Robledo, P., Selles, G., Arpaia, M.L., Saavedra, J., Defilippi, B.
Chilean ′ Hass′ avocados destined to local and distant markets? Postharvest Biol.
G., 2017. Identification of preharvest factors determining postharvest ripening
Technol. 174, 111457 https://doi.org/10.1016/j.postharvbio.2020.111457.

10
I. Hernández et al. Postharvest Biology and Technology 185 (2022) 111806

behaviors in ‘Hass’ avocado under long term storage. Sci. Hort. 216, 29–37. https:// Uarrota, V.G., Fuentealba, C., Hernández, I., Defilippi, B.G., Meneses, C., Campos-
doi.org/10.1016/j.scienta.2016.12.024. Vargas, R., Lurie, S., Carpentier, S., Poblete-Echeverría, C., Pedreschi, R., 2019.
Robinson, M.D., McCarthy, D.J., Smyth, G.K., 2010. edgeR: a bioconductor package for Integration of proteomics and metabolomics data of early and middle season Hass
differential expression analysis of digital gene expression data. Bioinformatics 26, avocados under heat treatment. Food Chem. 289, 512–521. https://doi.org/
139–140. https://doi.org/10.1093/bioinformatics/btp616. 10.1016/j.foodchem.2019.03.090.
Rohart, F., Gautier, B., Singh, A., LeˆCao, K.A., 2017. mixOmics: An R Package for ‘omics Vergara-Pulgar, C., Rothkegel1, K., González-Agüero, M., Pedreschi, R., Campos-
Feature Selection and Multiple Data Integration. https://doi.org/10.1371/journal. Vargas, R., Defilippi, B., Meneses, C., 2019. De novo assembly of Persea americana cv.
pcbi.1005752. ‘Hass’ transcriptome during fruit development. BMC Genom. 20, 108. https://doi.
Rolle, R.S., Chism, G.W., 1989. Kinetic comparison of cytokinin nucleosidase activity org/10.1186/s12864-019-5486-7.
isolated from normally ripening and mutant tomato varieties. Plant Physiol. 91, Vincent, C., Mesa, T., Munné-Bosch, S., 2020. Hormonal interplay in the regulation of
148–150. https://doi.org/10.1104/pp.91.1.148. fruit ripening and cold acclimation in avocados. J. Plant Physiol. 251, 153225.
Sánchez-Rodríguez, C., Rubio-Somoza, I., Sobout, R., Persson, S., 2010. Phytohormones https://doi.org/10.1016/j.jplph.2020.153225.
and the cell wall in Arabidopsis during seedling growth. Trends Plant Sci. 15, Wickham, H., Averick, M., Bryan, J., Chang, W., McGowan, L.D., François, R.,
291–301. https://doi.org/10.1016/j.tplants.2010.03.002. Grolemund, G., Hayes, A., Henry, L., Hester, J., Kuhn, M., Pedersen, T.L., Miller, E.,
Seo, M., Jikumaru, Y., Kamiya, Y., 2011. Profiling of hormones and related metabolites Bache, S.M., Müller, K., Ooms, J., Robinson, D., Seidel, D.P., Spinu, V., Takahashi, K.,
in seed dormancy and germination studies. In: Kermode, Allison R. (Ed.), Seed Vaughan, D., Wilke, C., Woo, K., Yutani, H., 2019. Welcome to the tidyverse. J. Open
Dormancy: Methods and Protocols, Methods Mol. Biol, vol. 773, pp. 99–111. https:// Res. Softw. 4 (43), 1686. https://doi.org/10.21105/joss.01686.
doi.org/10.1007/978-1-61779-231-1_7. Xoca-Orozco, L.A., Aguilera-Aguirre, S., Vega-Arreguí, J., Acevedo-Hernández, G., Tovar-
Seymour, G.B., Taylor, J.E., Tucker, G.A. (Eds.), 1993. Biochemistry of Fruit Ripening. Pérez, E., Stoll, A., Herrera-Estrella, L., Chacón-López, A., 2019. Activation of the
Springer, Netherlands, p. 454. https://doi.org/10.1007/978-94-011-1584-1. phenylpropanoid biosynthesis pathway reveals a novel action mechanism of the
Spartz, A.K., Ren, H., Park, M.Y., Grandt, K.N., Lee, S.H., Murphy, A.S., Sussman, M.R., elicitor effect of chitosan on avocado fruit epicarp. Food Res. Int. 121, 586–592.
Overvoorde, P.J., Gray, W.M., 2014. SAUR inhibition of PP2C-D phosphatases https://doi.org/10.1016/j.foodres.2018.12.023.
activates plasma membrane H+-ATPases to promote cell expansion in Arabidopsis. Xu, Q., Krishnan, S., Merewitz, E., Xu, J., Huang, B., 2016. Gibberellin regulation and
Plant Cell 26, 2129–2142. https://doi.org/10.1007/978-94-011-1584-1. genetic variations in leaf elongation for tall fescue in association with differential
Trujillo-Mayol, I., Badillo-Muñoz, G., Céspedes-Acuña, C., Alarcón-Enos, J., 2020. The gene expression controlling cell expansion. Sci. Rep. 6, 30258. https://doi.org/
relationship between fruit size and phenolic and enzymatic composition of avocado 10.1038/srep30258.
byproducts (Persea americana mill.): the importance for biorefinery applications. Zhang, M., Yuan, B., Leng, P., 2009. The role of ABA in triggering ethylene biosynthesis
Horticulturae 6, 91. https://doi.org/10.3390/horticulturae6040091. and ripening of tomato fruit. J. Exp. Bot. 60, 1579–1588. https://doi.org/10.1093/
jxb/erp026.

11