Proteomic analysis of the secretome of HepG2 cells indicates differential proteolytic processing after infection with dengue virus

    Proteomic analysis of the secretome of HepG2 cells indicates differential proteolytic processing after infection with dengue virus Marjolly B. Caruso, Monique R.O. Trugilho, Andr´e S. Teixeira-Ferreira, Jonas Perales, Andrea T. Da Poian, Russolina B. Zingali PII: DOI: Reference:

S1874-3919(16)30307-4 doi: 10.1016/j.jprot.2016.07.011 JPROT 2624

To appear in:

Journal of Proteomics

Received date: Revised date: Accepted date:

25 January 2016 27 June 2016 13 July 2016

Please cite this article as: Caruso Marjolly B., Trugilho Monique R.O., Teixeira-Ferreira Andr´e S., Perales Jonas, Da Poian Andrea T., Zingali Russolina B., Proteomic analysis of the secretome of HepG2 cells indicates differential proteolytic processing after infection with dengue virus, Journal of Proteomics (2016), doi: 10.1016/j.jprot.2016.07.011

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Proteomic analysis of the secretome of HepG2 cells indicates differential proteolytic processing after infection with dengue virus.

RI P

T

Marjolly B. Carusoa, b, Monique R.O. Trugilhoa, c, d, Luiza M. Higaa, b, ¤, André S. TeixeiraFerreiraa, c, Jonas Peralesa, c, Andrea T. Da Poiana, b, and Russolina B. Zingalia, b, * a

Rede Proteômica do Rio de Janeiro, Brazil Programa de Biologia Estrutural, Instituto de Bioquímica Médica Leopoldo de Meis, Universidade Federal do Rio de Janeiro, Rio de Janeiro, RJ 21941-902, Brazil c Laboratório de Toxinologia, Departamento de Fisiologia e Farmacodinâmica, Instituto Oswaldo Cruz, Fundação Oswaldo Cruz, Rio de Janeiro, RJ 21045-900, Brazil

SC

b

d

MA

NU

Current address: Centro de Desenvolvimento Tecnológico em Saúde (CDTS), Fundação Oswaldo Cruz, Rio de Janeiro, RJ 21045-900, Brazil ¤ Current address: Laboratório de Virologia Molecular, Departamento de Genética, Instituto de Biologia, Universidade Federal do Rio de Janeiro

ED

* Corresponding author at: Unidade de Espectrometria de Massas e Proteômica (UEMP), Instituto de Bioquímica Médica Leopoldo de Meis, Universidade Federal do Rio de Janeiro - UFRJ, Rio de Janeiro, Brazil. Tel/Fax.: +55 21 39386782. E-mail addresses: [email protected] (R.B. Zingali)

PT

ABSTRACT

CE

Secretome analysis can be described as a subset of proteomics studies consisting in the analysis of the molecules secreted by cells or tissues. Dengue virus (DENV)

AC

infection can lead to a broad spectrum of clinical manifestations, with the severe forms of the disease characterized by hemostasis abnormalities and liver injury. The hepatocytes are a relevant site of viral replication and a major source of plasma proteins. Until now, we had limited information on the small molecules secreted by hepatic cells after infection by DENV. In the present study, we analysed a fraction of the secretome of mock- and DENV-infected hepatic cells (HepG2 cells) containing molecules with less than 10 kDa, using different proteomic approaches. We identified 175 proteins, with 57 detected only in the samples from mock-infected cells, 59 only in samples from DENV-infected cells, and 59 in both conditions. Most of the peptides identified were derived from proteins larger than 10 kDa, suggesting a proteolytic processing of the secreted molecules. Using in silico analysis, we predicted consistent differences between the proteolytic processing occurring in mock and DENV-infected

ACCEPTED MANUSCRIPT samples, raising, for the first time, the hypothesis that differential proteolysis of secreted

T

molecules would be involved in the pathogenesis of dengue.

RI P

Key words: dengue virus, hepatic cells, proteome, low molecular mass proteins,

SC

prediction of proteolytic enzymes, secretome.

Biological Significance

NU

Since the liver, one of the targets of DENV infection, is responsible for producing molecules involved in distinct biological processes, the identification of proteins and

MA

peptides secreted by hepatocytes after infection would help to a better understanding of the physiopathology of dengue. Proteomic analyses of molecules with less than 10 kDa secreted by HepG2 cells after infection with DENV revealed differential proteolytic

PT

ED

processing as an effect of DENV infection.

1. Introduction

CE

Following the genomic revolution, proteomics rose as a promising approach to give a much wider comprehension of an unlimited number of biological phenomena [1].

AC

However, as time went by in the early 2000s, proteomics began to be used mostly for hypothesis-free exploratory studies, making many proteomic studies to become reports of lists of proteins expressed in a cell, tissue or organism without introducing the novelty that has been expected [2]. But even in this scenario, with the hypothesis-driven studies generating results of much higher impact, sometimes the exploratory studies reveal completely unexpected findings. This is the case of the work described here. In a previous work, we used an exploratory proteomic approach to analyze the proteins secreted by hepatic cells infected with dengue virus (DENV) [3]. Dengue is endemic in over 100 countries and is considered the most important arthropod-borne disease nowadays, with almost 4 billion people living at risk areas for infection [4-6]. Brazil is one of the most affected countries in Latin America, reporting more than 1,500,000 cases in 2015 with almost 900 deaths [7]. Although the majority of dengue

ACCEPTED MANUSCRIPT cases are asymptomatic, infection may cause a severe life-threatening disease, characterized by several hemostatic disorders [8-13]. Since the liver produces most of the plasmatic proteins, including the coagulation factors and several inflammatory

T

cytokines involved in the control of vascular permeability, the evaluation of the panel of

RI P

proteins and peptides secreted by the hepatocytes is of great interest. Additionally, the liver is recognized as a major site of DENV replication [14-17].

SC

Analyzing the secretome fraction of DENV-infected hepatocytes containing the proteins with molecular masses higher than 10 kDa, we observed the presence of

NU

several differentially secreted proteins, such as peptidyl-prolyl isomerases A and B, tissue inhibitor of metalloproteinases 1 and 2 and superoxide dismutase, macrophage

MA

inhibitory factor migration and α-enolase, all of them are potentially involved in the dengue pathogenesis [3]. This study also revealed a surprising finding: the presence of α-enolase, a classic cytoplasmic glycolytic enzyme, among the most abundant secreted

ED

proteins by infected cells. This finding allowed us to show enolase as a molecule with unconventional secretion pathway that, being a plasminogen activator when secreted,

PT

may contribute to hemostatic disorders observed in dengue pathogenesis [18].

CE

Considering that approximately 50% of all human proteins have molecular masses lower than 26.5 kDa [19] or are small proteins or peptides produced by proteolytic processing of larger proteins [20], in the present study we investigated the small

AC

molecules (< 10 kDa) present on the secretome of hepatic cells after DENV infection. Again, the findings were surprising. Instead of identifying biologically active peptides that could be involved in the disease progression, the results revealed the presence of several fragments from proteins with known high molecular masses. This unexpected observation led us to formulate a new hypothesis: the occurrence of changes in the proteolytic processing of the secreted molecules. More interestingly, in silico analyses revealed that while in mock-infected sample a higher number of peptides result from cleavage by an enzyme similar to thermolysin, in DENV-infected samples we found a higher number of peptides cleaved by an enzyme similar to chymotrypsin and a decrease in the number of peptides resulted from the cleavage by an enzyme similar to trypsin, which could be explained by the presence of serine protease inhibitors, such as neuroserpin and inter-alpha trypsin inhibitor, only observed in the sample from DENV-

ACCEPTED MANUSCRIPT infected cells. The results altogether suggest a differential proteolytic processing during DENV

infection

in hepatocytes,

which may have

a direct

correlation

with

RI P

T

pathophysiological alterations that occurring in dengue.

2. Materials and methods

SC

2.1 Separation steps to obtain the low molecular mass samples HepG2 cells were grown in DMEM (Invitrogen Corporation, USA) supplemented

NU

with 10% fetal bovine serum (FBS). DENV serotype 2 (strain 16681) was propagated in the Aedes albopitcus C6/36 cell line (DENV MOI = 0.1 – 9dpi) [21]. HepG2 cells were

MA

mock-infected or infected with DENV at MOI 1. After 32 h, cells were extensively washed with PBS and incubated for further 16 h with serum-free DMEM (Invitrogen Corporation, USA). At 48 h post-infection, the conditioned medium containing secreted

ED

proteins was collected as described elsewhere [3]. In order to ensure that the experimental conditions used to prepare the samples did not cause cell death and

PT

disruption, MTT, Trypan Blue and LDH assays were performed [3]. Viability assays (MTT and Trypan Blue exclusion) showed less than 3% death cells, both for infected

CE

and mock-infected cells. Additionally, absence of cell disruption was confirmed by the negligible LDH activity in the conditioned media (1.35% and 1.23% of the total LDH

AC

activity, for mock-infected and infected cells, respectively). Four independent experiments were conducted to obtain suitable biological replicates (n = 4) for each condition. The conditioned media of the cells were concentrated and fractioned using a 10 kDa cutoff membrane [3]. The fractions containing proteins with molecular masses below 10 kDa were called ultrafiltrate (UF), as shown in figure 1. The UF from mock (M) and DENV-infected (DENV) cells were concentrated by lyophilization and submitted to a desalting step using a Sep-Pak Vac C18 cartridge (Water Corporation Sep-Pak – 1 cc/100 mg), according to the manufacturer’s recommendations, prior to analysis by MS.

2.2 Analysis of the digested molecules from the ultrafiltrate secretome of infected hepatic cells

ACCEPTED MANUSCRIPT 2.2.1 Separation and concentration of the ultrafiltrate samples prior to distinct enzymatic treatments The UF from the M and DENV conditions were filtered as describes in the item

RI P

T

separation steps (2.1) and concentrated using an Amicon® Ultra-4 Centrifugal Filter Unit (3 kDa), according to the manufacturer´s instructions (Merck Millipore). This step was necessary in order to remove contaminants observed in the standardization stage

SC

that interfered on the ionization steps of mass spectrometry analysis (data not shown). The samples above 3 kDa were quantified by a bicinchoninic acid assay [22] using BSA

NU

(Sigma) as a standard. Aliquots containing 10 µg of proteins from each condition were treated with 1 M urea and 10 mM DTT at 56C for 1 hour and 55 mM iodoacetamide at

MA

room temperature for 45 minutes. At the end of the carbamidomethylation step, each aliquot was submitted to a distinct enzymatic treatment, where the endoproteinases were prepared to a final concentration of 1 µg/µl: trypsin (1 µg), or Lys-C (0.5 µg), or

ED

trypsin (1 µg) followed by Lys-C (0.5 µg) overnight at 37C. The reactions were

PT

terminated upon the addition of 0.1% formic acid.

CE

2.2.2 Mass spectrometry analysis using an Orbitrap Velos The peptides obtained after the enzymatic treatments were desalted and

AC

solubilized in 20 µl of 0.1% formic acid, and 4 µl of the 5x diluted samples were loaded on a C18 trap column (20 mm x 100 μm, 200 Å, Magic, AQ matrix, 5 µm – Michrom Bioresources, USA), followed by the separation of the samples in a C18 column (100 mm x 75 μm, 200 Å, Magic, AQ matrix, 5 µm - Michrom Bioresources, USA) onto an EASY-nLC-II System (Thermo Scientific, USA) coupled to an Orbitrap Velos. The samples were eluted (0.2 μL/min) using a linear gradient (0 to 60%) of 100% acetonitrile containing 0.1% formic acid for 58 min. The nESI voltage was set at 1.9 kV without the influence of any gas, and the source temperature was 200 °C. The instrument control and data acquisition for the MS and MS/MS were conducted using Xcalibur software (2.0.7 version). The data acquisition was performed using alternating MS for an Orbitrap with a resolution of 60,000 (FWHM @ m/z 400) and using MS2 for a linear trap. A maximum of 10 ions was selected for MS fragmentation using CID. The fragmentation

ACCEPTED MANUSCRIPT had normalized collision energy of 35, and only precursors with a charge state of 2 were selected for MS/MS acquisitions using the CID mode. Precursors with a charge state of

RI P

T

1 and/or an intensity lower than 10.000 were excluded.

2.3 Analysis of the intact molecules from the ultrafiltrated secretome of infected

SC

hepatic cells 2.3.1 Desalting step in POROS R2

NU

The UF (50 µg) from the M and DENV conditions were completely dried using a speed vacuum concentrator system (Thermo Scientific Savant SDP111V) and

MA

solubilized in 1% trifluoroacetic acid. These samples were subjected to a desalting step using an in-house made column using POROS R2 C8-C18 resin (Applied Biosystems) at 10 mg/ml. The columns were packaged in 10 µl tips (Axygen) using pieces of C8 disc

ED

(3M Empore) as blockades. The activation solution used was 100% acetonitrile (10 µl), and the equilibration solution was 1% trifluoroacetic acid (two aliquots of 10 µl). The

PT

elution was performed using sequential solutions, where the first solution contained 20 µl of 0.1% trifluoroacetic acid /70% acetonitrile and the last solution contained 20 µl of

CE

100% acetonitrile. The eluted molecules were completely dried by a speed vacuum

AC

concentrator system prior to a strong cationic exchange separation step.

2.3.2 Strong cationic exchange separation The dried samples were solubilized in 100 µl of a solution containing 5 mM KH2PO4/ 25% acetonitrile (pH 3), and the molecules were fractionated using strong cationic exchange Ultra-Micro SpinColumns (Harvard Apparatus). The columns were hydrated in the same solution and centrifuged at 110 g for 2 min. The elution was performed using 150 µl of KCl at increasing concentrations (75, 150 and 300 mM) followed by a centrifugation step at 110 g for 2 min. The eluted fractions were dried by a speed vacuum concentrator system and submitted to a new desalination step using POROS R2 columns. Then, the samples were subjected to mass spectrometry analysis by an LTQ Orbitrap XL.

ACCEPTED MANUSCRIPT

2.3.3 Mass spectrometry analysis using the LTQ Orbitrap XL

T

The intact samples were desalted and loaded onto an EASY-nLC-II System as

RI P

previously described for the Orbitrap Velos, but in this case coupled to LTQ Orbitrap XL. The nESI voltage was set to 1.9 kV without the influence of any gas, and the source temperature was 200 °C. The instrument control and data acquisition for the MS and

SC

MS/MS were conducted using Xcalibur software (2.0.7 version). The data acquisition was performed using alternating MS for an Orbitrap with a resolution of 60,000 (FWHM

NU

@ m/z 400) and using MS2 for a linear trap. A maximum of 10 ions was selected for MS fragmentation using collision-induced dissociation (CID), and a maximum of 3 ions with

MA

a higher intensity was fragmented using electron transfer dissociation (ETD). Both fragmentations had normalized collision energy of 35. Only precursors with a charge

ED

state of 2 were selected for MS/MS acquisitions using the CID mode, and only precursors with charge states of 3, 4 and 5 were selected for MS/MS acquisitions in the ETD mode. Precursors with a charge state of 1 and/or an intensity lower than 10.000

CE

PT

were excluded.

2.4 Computational analysis

AC

2.4.1 Protein identification and quantification The spectra obtained from the Orbitrap Velos (experimental duplicates) and from LTQ Orbitrap XL analyses (experimental duplicates) were compared with the theoretical spectra using the “PatternLab for Proteomics” software v 3.2 [23] for protein identification. All mass spectrometry data generated for this study is made available at http://max.ioc.fiocruz.br/mtrugilho/CarusoMB2016/. Peptide sequence matching (PSM) was performed using the Comet algorithm [24] against the protein-centric human, from Homo sapiens database (NeXtProt - 20,126 proteins entries) with proteins entries downloaded on January 2015 plus a FASTA file containing Dengue virus dengue virus database (Uniprot - GeneBank taxon number 14). A target-reverse decoy strategy was employed. The search parameters used for the UF samples prior distinct enzymatic

ACCEPTED MANUSCRIPT treatments were as follows: digestion with the endoproteinases trypsin and LysC, spectra fixed modification of cysteine carbamidomethylation and as a variable

T

modification the oxidation of methionine. Alternatively, for the analysis of the UF intact

RI P

molecules, the search parameters were: no enzyme, spectra fixed modification of cysteine carbamidomethylation and as a variable modification the oxidation of methionine. The Comet search engine considered a precursor mass tolerance of 10

SC

ppm for MS spectra and 500 ppm for MS/MS. All identification results are reported with the maximum parsimony principle [25], which is the minimal number of proteins or

NU

peptides that can be explained when the same peptide sequence is shared by all identified peptides, and less than 1% FDR, both peptide and protein level, by

MA

PatternLab’s Search Engine Processor (SEPro) module. Spectral counting (SC) for estimation of protein copy number was accomplished using the normalized spectral

ED

abundance factor (NSAF) [26]. Differentially abundant molecules were pinpointed using PatternLab’s TFold module with a Benjamini–Hochberg q-value of 0.05 [27]. Proteins

PT

uniquely identified in one condition (mock or DENV-infected HepG2 cells) were

module.

CE

pinpointed according to PatternLabs Approximate Area Proportional Venn Diagram

AC

2.4.2 Prediction analysis of proteolytic process Since the results showed the presence of proteolytic fragments of proteins with a molecular mass higher than 10 kDa, we decided to evaluate intrinsic proteolysis in the samples. In order to predict possible differences in the cleavage sites of the molecules present in the ultrafiltrates, we conducted an in silico analysis using the PeptideCutterExPASy program. The program predicts possible sites of cleavage after a simulation of digestion with different types of enzymes. We use the FASTA sequences from the proteins mapped of the identified peptides on intact sample. The sequences of the peptides obtained from the analysis and the FASTA sequence of the equivalent protein were manually compared. The N-terminal sites where the sequence of intact peptides

ACCEPTED MANUSCRIPT and the FASTA sequence for each protein were aligned were manually compared and

T

grouped according to the type of enzyme that would be able to act on that site.

RI P

3. Results and discussion

The pathophysiological alterations that occur during severe dengue cases seem

SC

to involve directly or indirectly vascular, hemostatic and hepatic disturbances. Several reports have described hepatic cells as one of the sites for virus replication during

NU

dengue infection [28-30]. Since the liver is responsible for the synthesis of numerous molecules, such as plasma proteins and coagulation factors, as cited earlier, the study of the proteins secreted by hepatic cells is important to better understand dengue

MA

pathogenesis. Previously, our group performed a study of the secretome of DENVinfected HepG2 cells with a focus on the molecules with masses higher than 10 kDa [3].

ED

However, until now we had limited information on the small molecules secreted during infection, which would contain biologically active peptides involved in disease

PT

progression. Therefore, we decide to analyze the fraction containing molecules with less than 10 kDa present in that secretome, which we named as ultrafiltrate (UF), as

CE

depicted in the diagram shown in Figure 1. To confirm the presence of only small molecules in our samples, we initially performed a MS analysis (data not shown), which

AC

indicates that the majority of molecules from mock-infected and DENV-infected samples have molecular masses below the cutoff used in the separation step, with the exception of one protein with a molecular mass of approximately 24.7 kDa, found in both samples. This initial analysis also revealed similar populations of ions in both samples (data not shown), but no natural peptide was identified. One important limitation of the analysis of intact molecules by mass spectrometry is the difficulty in equally ionizing all the molecules contained in the mixture for further identification. Therefore, the identification of the molecules present in the UF fraction was performed using two complementary mass spectrometry approaches known as bottom-up and top-down analyses. To ensure that the majority of the polypeptides contained in the UFs were identified, we characterized these molecules after enzymatic digestion, the bottom-up

ACCEPTED MANUSCRIPT strategy. Several studies have reported that the diversification of enzymatic treatments improves the amount and diversity of peptides obtained and identified [31]. Therefore, these UFs were subjected to three distinct enzymatic treatments with the

T

endoproteinases trypsin, Lys-C and Lys-C followed by trypsin. The peptides generated

RI P

from each digestion were analyzed by LTQ mass spectrometer using an EASY-nLC-II System (Thermo Scientific, USA) coupled to an Orbitrap Velos. The results obtained

SC

reinforced the relevance of enzymatic diversification since some molecules only had their sequences identified after a specific enzymatic treatment. The number of

NU

molecules identified after each enzymatic treatment is shown as a Venn diagram for the samples from mock-infected (Figure 2A) and DENV-infected (Figure 2B) cells. The

MA

combined result of all enzymatic treatments allowed the identification of 135 proteins. Of these molecules, 39 were detected only in the mock condition, 58 only in the DENVinfected condition, and 38 were identified in both conditions (Figure 3A).

ED

To complement the information regarding the complexity of the polypeptides present in the UFs, we conducted a complementary study of the samples using the top-

PT

down mass spectrometry strategy. In this approach, the molecules were directly analyzed by mass spectrometry without previous digestion. Before the analysis, the

CE

intact samples were desalinated, and the eluted fraction was applied to a nESI LTQ Orbitrap XL mass spectrometer (Thermo Scientific, USA). Since the intact molecules

AC

have intrinsic charge, are larger than peptides produced by enzymatic digestion, and have some limitations regarding the ionization process, we use two complementary fragmentation methodologies, CID (selecting precursors with charge state of +2) and ETD (selecting precursors with multiples charge state such as +3, +4 and +5). This analysis resulted in the identification of 53 molecules. Among them, 18 were detected only in the mock condition, 8 in the DENV-infected condition and 26 were common to both conditions (Figure 3B). By combining all the data obtained by the top-down and bottom-up strategies, we were able to identify a total of 175 proteins, with 57 molecules detected in the mockinfected condition, 59 in the DENV-infected condition and 59 observed in both conditions (Figure 3C). Also, a label free quantification analysis based on the spectral counting (SC) was used to estimate the protein copy number in the PatternLabs

ACCEPTED MANUSCRIPT software. The results were plotted as an Approximate Area Proportional Venn Diagrams in Figure 3 revealing that only 12 molecules were differentially expressed between the studied conditions (considering a q-value of 0.05 using PatternLab’s TFold module with

T

a Benjamini–Hochberg). Among these molecules, 4 were observed after enzymatic

RI P

treatment analyses and 8 in the intact samples (Table S2a and S2b). It is interesting to mention that among the 175 identified polypeptides, 33 were only detected during the

SC

analysis of intact molecules on LTQ Orbitrap XL mass spectrometer. Twelve of them were detected only in the mock condition, 5 only in the DENV-infected condition and 16

NU

were common to both samples. These data reinforce the relevance of applying complementary strategies to obtain a more complete panel of information about the

MA

molecules present in the UFs of the studied conditions. A compilation of all the identified molecules, disregarding the strategy adopted, can be found in the Supplementary Data (Table S1b).

ED

According to the literature, the most challenging event in the study of the secretome is avoiding the action of proteases on secreted proteins because of the

PT

difficulty in distinguishing natural peptides from those originating from protein degradation as a result of artifactual proteolysis [32]. Most natural bioactive peptides

CE

present in body fluids are found at reduced concentrations; thus, fragments originating from more abundant proteins would hamper the identification of these natural peptides

AC

[33-34]. Therefore, the use of protease inhibitors is imperative to lessen the unspecific proteolytic action and to allow the identification of real natural peptides. However, the characterization of the molecules present on the UF showed that even after adding protease inhibitors during sample preparation, the majority of polypeptides (49 out of 52 molecules) found in the intact samples had theoretical molecular masses higher than 10 kDa. Surprisingly, instead of active peptides, the results revealed the presence of several fragments from proteins with known high molecular masses. This unexpected observation led us to formulate the hypothesis of proteolytic processing upon the secreted molecules. As mentioned before, previous data obtained from MS analysis (data not shown) revealed the presence of only one protein with molecular mass higher than 10 kDa in the samples, therefore suggesting that most of the peptides identified correspond to

ACCEPTED MANUSCRIPT fragments of proteolytically processed proteins. Importantly, UF analyses also revealed that some of the peptides derived from proteins with theoretical masses higher than 10 kDa were present in only one of the conditions studied (mock-infected or DENV-

T

infected), even though those proteins were observed in both conditions in the sample

RI P

containing proteins above 10 kDa [3], as represented in Figure 4, suggesting a differential proteolysis between infected and uninfected cells during secretion

SC

processes. Another evidence that differential proteolysis occur is demonstrated in Figure 5, which shows that distinct peptides from fibrinogen were detected in samples

NU

from mock-infected or DENV-infected cells.

Taken together, the observation of peptides originated from secreted proteins

MA

with high molecular mass, some of them identified in the previous analysis of the secretome [3], the identification of proteases and proteases inhibitors differentially secreted by DENV-infected cells throughout the same study, and the presence of

ED

fragments from proteins described as inserted in the membrane of extracellular vesicles [35] in this study, support the hypothesis of the differential proteolytic processing during

PT

DENV infection. However, it is important to mention that since the viability tests revealed negligible values for cell death and/or lysis, the presence of these molecules in

CE

the extracellular medium would be a result of either a differential extracellular proteolysis or an intracellular proteolysis followed by secretion of the resulting peptides

AC

by unconventional mechanisms. Thus, in order to verify whether the identified polypeptides were classified as secreted molecules, we compared our data with those obtained in other secretome studies [3, 36-45], and also searched for their subcellular location using the Uniprot database. These analyses revealed that 62% of the molecules identified in our study were classified as secreted. While the remaining molecules were described as present in the membrane of extracellular vesicles (11%) [32] or as intracellular proteins (components of intracellular organelles as well as cytosolic or nuclear proteins – 27%). This data confirms that most of identified peptides came from secreted proteins. To predict differences in the cleavage site by distinct proteases in the intact samples, we conducted an analysis using the PeptideCutter program considering the sequences exclusive to each condition. This analysis predicts possible sites of

ACCEPTED MANUSCRIPT cleavages after simulation of the digestion process using different types of enzymes, and allowed us to find out differences between the samples (Figure 6). In the mockinfected condition, we noticed a higher chance of protein cleavage by enzymes with

T

actions similar to thermolysin (which cleaves between aliphatic amino acids with

RI P

hydrophobic side chain, such as isoleucine, leucine, valine and alanine) and trypsin (which cleaves between basic amino acids, such as arginine and lysine). Conversely, in

SC

the DENV-infected condition, the peptides presented a higher chance to have been cleaved by enzymes with actions similar to chymotrypsin (which cleaved between

NU

aromatic amino acids with hydrophobic side chains, such as tyrosine, and phenylalanine - with high specificity -, or leucine, methionine and histidine – with low specificity). This

MA

variation would occur due to differential proteolytic processing, explaining the differences among the sequences and the number of peptides obtained from the mockinfected condition (37 exclusive sequences) and the DENV-infected condition (23

ED

exclusive sequences). It is worth mentioning that the action of more than one enzyme upon the same protein could generate very small peptides, which would explain the

PT

reduced number of identifications, since the length of the sequence can hinder the reliability to identify these molecules.

CE

Protein degradation may be triggered by different stimuli, such as oxidative stress, aging, pathogens and alterations in environment conditions [46-48]. It is possible

AC

that the DENV infection promotes alterations in cellular proteolytic processes modifying the activity or the expression of some enzymes and/or inhibitors. In our earlier work, we detected different molecules potentially involved in proteolytic processes, such as proteasome alpha-4, cathepsins (D, L and X), PCSK9 and TIMP-1 both in mock and infected conditions; carboxypeptidase A3, LON peptidase (N-terminal and ring finger 3) and serine protease (Htra isoform 1) only in mock condition; and serine (or cysteine) proteinase inhibitor clade I (neuroserpin) member 1 and TIMP-2 only in DENV-infected condition [3]. Furthermore, we observed that the concentration of TIMP-1 (a metalloproteinase inhibitor) was lower in the secretome of DENV-infected cells compared to that of mock-infected cells [3]. In the present work, we identified other proteases and protease inhibitors, such as inter-alpha-trypsin inhibitor (heavy chain H2)

ACCEPTED MANUSCRIPT and ADAM10 only in mock condition; and TFPI, neurotrypsin and trypsin only in DENVinfected condition. Inter-alpha trypsin inhibitor is a serine protease inhibitor; the peptide identified in

RI P

T

the UF is located in its propeptide region [49-50] (Figure 7a). The sequence identified in ADAM10, the disintegrin and metalloproteinase domain-containing protein 10, is located in the extracellular region of the protein [51-52] (Figure 7b). TFPI (tissue factor pathway

SC

inhibitor) is a protease inhibitor formed by three domains (K1, K2 and K3), where K1 and K2 inhibit the complex coagulator factor VIIa-tissue factor and coagulator factor Xa,

NU

respectively, preventing the extrinsic pathway activation in the coagulation process, and K3 domain, from which the peptide identified belongs, had no protease inhibition

MA

function [53] (Figure 7c). Neurotrypsin is a serine protease and the sequence identified is located in the zymogen activation region [54] (Figure 7d). Finally, a peptide from human trypsin was also identified. However, it was not possible to determine from which

ED

protein species of trypsin it came from since it is a conservative sequence [55]. The fact that all these proteases and protease inhibitors have theoretical molecular masses

PT

higher than 10 kDa, but were only identified in the UF, indicates that these molecules underwent a proteolytic processing where only the peptides generated after this event

CE

were identified. This processing would hinder the original function of these molecules or help the conversion of zymogens to the active form, leading to changes in the panel of

AC

peptides produced by the cell. An explanation for the absence of detection of these molecules in our previous study could be related to the lower sensitivity of the experimental procedures used, which would have underestimated the amount and variety of the identified proteases and protease inhibitors. Although data collected during this work allowed us to observe a proteolytic processing upon molecules presented in the secretome, complementary studies are necessary for a better understanding of the pathophysiological relevance of this phenomenon. It is noteworthy that the identification of these proteases and protease inhibitors complements our previous work, and taken together, the results of both studies indicate alterations in the variety and amount of these molecules. For instance, among the molecules only identified in the mock intact samples, we found a peptide originated from proSAAS, also known as proprotein convertase subtilisin/kexin type 1

ACCEPTED MANUSCRIPT inhibitor. This peptide corresponds to the KEP segment that is normally generated during processing of proSAAS by protein convertases 1/3 (PC1/3) and PC2 followed by carboxypeptidase E [56] (Figure 8), indicating that these proteases were active in mock-

RI P

T

infected cells but not in the DENV-infected cells.

Through the analysis using the PeptideCutter program, we observed in the mockinfected sample a higher number of peptides probably resulting from cleavage by an

SC

enzyme similar to thermolysin (Figure 6), which acts between aliphatic amino acids with hydrophobic side chain. When we analyzed the proteases only present in the secretome

NU

from mock-infected cells, we found two enzymes, LON peptidase and Htra [3], both described in other organisms as acting at hydrophobic residues [57-58]. Conversely, in

MA

the DENV-infected sample, we observed a reduction on the number of peptides resulted from the cleavage by an enzyme similar to trypsin (Figure 6). This phenomenon could be explained by the presence of serine protease inhibitors, such as neuroserpin and

ED

inter-alpha trypsin inhibitor, only observed in the sample from DENV-infected cells. We could also hypothesize the existence of a chymase-like enzyme in the secretome of the

PT

DENV-infected cells, since we noticed a higher number of peptides cleaved by an enzyme similar to chymotrypsin (cleaves of aromatic amino acids with hydrophobic

CE

lateral chains) [59]. Consequently, the infection would modulate the action of certain enzymes on secreted proteins, leading to different proteolytic processing and the

AC

generation of a peptide panel distinct from the non-infected condition.

4. Conclusion The compilation of the results obtained in this study, including mass spectrometry analyses combined with in silico studies for prediction of cleavage sites for proteolytic activity, together with the results from our previous secretome study [3] allowed us to identify alterations in the proteolytic processing of the secreted proteins. This could be explained by infection-induced differences in the expression of proteases or protease inhibitors, as well as the activation of different enzyme families triggered by specific stimuli or by differential regulation of their inhibitors.

ACCEPTED MANUSCRIPT Conflict of interest statement

T

The authors declare that they have no conflicts of interest.

RI P

Acknowledgements

This work was supported by Fundação Carlos Chagas Filho de Amparo à Pesquisa do

SC

Estado do Rio de Janeiro (FAPERJ 103.163/2011, FAPERJ E-26/110.092/2013, FAPERJ E-26/201.167/2014), Conselho Nacional de Desenvolvimento Científico e

NU

Tecnológico (MCT/CNPq 476448/2008-5, (MCT/CNPq 550114/2010-6 and MCT/CNPq 306669/2013-7) and Plataformas Tecnologicas, Fiocruz. MBC was a recipient of a

MA

Ph.D. fellowship from CAPES and MROT was a recipient of a post-doctoral fellowship from CAPES/Faperj. The authors thanks the professor Igor C. Almeida, from Department of Biological Sciences, Border Biomedical Research Center, University of

ED

Texas at El Paso, and the PhD Paulo C. Carvalho, from Laboratory for Proteomics and Protein Engineering, Carlos Chagas Institute, Fiocruz, Paraná, for the fruitful

PT

discussions and help, and also the designer Pedro Bigler, from Centro de Desenvolvimento Tecnológico em Saúde (CDTS), for preparing high resolution figures

AC

References

CE

for publication, and for creating the graphical abstract.

[1] Kenyon GL, DeMarini DM, Fuchs E, Galas DJ, Kirsch JF, Leyh TS, Moos WH, Petsko GA, Ringe D, Rubin GM, Sheahan LC; National Research Council Steering Committee. Defining the mandate of proteomics in the post-genomics era: workshop report. Mol Cell Proteomics. 2002 Oct;1(10):763-80. [2] Marchetti-Deschmann M, Allmaier G. Mass spectrometry - One of the pillars of proteomics. J Proteomics. 2011 Jun 10;74(7):915-9. [3] Higa LM, Caruso MB, Canellas F, Soares MR, Oliveira-Carvalho AL, Chapeaurouge DA, Almeida PM, Perales J, Zingali RB, Da Poian AT. Secretome of HepG2 cells infected with dengue virus: Implications for pathogenesis. Biochim Biophys Acta. 2008; 1784(11):1607-16 [4] Brady OJ, Gething PW, Bhatt S, Messina JP, Brownstein JS, Hoen AG, Moyes CL, Farlow AW, Scott TW, Hay SI. Refining the global spatial limits of dengue virus transmission by evidence-based consensus. PLoS Negl Trop Dis. 2012; 6(8):e1760. [5] WHO: World Health Organization; Dengue: Guidelines for Diagnosis, Treatment, Prevention and Control: New Edition. WHO Guidelines Approved by the Guidelines Review Committee. 2009.

ACCEPTED MANUSCRIPT [6] Guzman MG, Halstead SB, Artsob H, Buchy P, Farrar J, Gubler DJ, Hunsperger E, Kroeger A, Margolis HS, Martínez E, Nathan MB, Pelegrino JL, Simmons C, Yoksan S, Peeling RW. Dengue: a continuing global threat. Nat Rev Microbiol. 2010; 8(12 Suppl):S7-16 [7] Secretaria de Vigilância em Saúde − Ministério da Saúde, Boletim Epidemiológico, 2015: 46 (n 44):1-9

RI P

T

[8] Chuansumrit A, Chaiyaratana W. Hemostatic derangement in dengue hemorrhagic fever. Thromb Res. 2014; 133(1):10-6.

SC

[9] Chuang YC, Lin YS, Liu CC, Liu HS, Liao SH, Shi MD, Lei HY, Yeh TM Factors contributing to the disturbance of coagulation and fibrinolysis in dengue virus infection. J Formos Med Assoc. 2013; 112(1):12-7.

NU

[10] Djamiatun K, Faradz SM, Setiati TE, Netea MG, van der Ven AJ, Dolmans WM. Increase of plasminogen activator inhibitor-1 and decrease of transforming growth factor-b1 in children with dengue hemorrhagic fever in Indonesia. J Trop Pediatr. 2011; 57(6):424-32.

MA

[11] Mairuhu AT, Mac Gillavry MR, Setiati TE, Soemantri A, ten Cate H, Brandjes DP, van Gorp EC. Is clinical outcome of dengue-virus infections influenced by coagulation and fibrinolysis? A critical review of the evidence. Lancet Infect Dis. 2003; 3(1):33-41.

ED

[12] Albuquerque LM, Trugilho MR, Chapeaurouge A, Jurgilas PB, Bozza PT, Bozza FA, Perales J, Neves-Ferreira AG. Two dimensional difference gel electrophoresis (DiGE) analysis of plasmas from dengue fever patients. J Proteome Res. 2009; 8(12):5431-41.

PT

[13] de Azeredo EL, Monteiro RQ, de-Oliveira Pinto LM. Thrombocytopenia in Dengue: Interrelationship between Virus and the Imbalance between Coagulation and Fibrinolysis and Inflammatory Mediators. Mediators Inflamm. 2015; 2015:313842

CE

[14] Seneviratne SL, Malavige GN, de Silva HJ. Pathogenesis of liver involvement during dengue viral infections. Trans R Soc Trop Med Hyg. 2006; 100(7):608-14. [15] Limonta D, Capó V, Torres G, Pérez AB, Guzmán MG. Apoptosis in tissue from fatal dengue shock syndrome. Journal of Clinical Virology 2007; 40(1):50-54.

AC

[16] Pagliari C, Quaresma JA, Fernandes ER, Stegun FW, Brasil RA, de Andrade HF Jr, Barros V, Vasconcelos PF, Duarte MI. Immunopathogenesis of dengue hemorrhagic fever: Contribution to the study of human liver lesions. J Med Virol. 2014; 86(7):1193-7 [17] Jessie K, Fong MY, Devi S, Lam SK, Wong KT. Localization of dengue virus in naturally infected human tissues, by immunohistochemistry and in situ hybridization. J Infect Dis. 2004; 189(8):1411-8. [18] Higa LM, Curi BM, Aguiar RS, Cardoso CC, De Lorenzi AG, Sena SL, Zingali RB, Da Poian AT. Modulation of α-enolase post-translational modifications by dengue virus: increased secretion of the basic isoforms in infected hepatic cells.PLoS One. 2014 Aug 29;9(8):e88314. [19] Bäckvall H & Lehtiö J. Methods in Molecular Biology 1023. The low molecular weight proteome Methods and Protocols. Human Press (Springer Science/Business Media New York) 201329 [20] Schutz Y. Protein turnover, ureagenesis and gluconeogenesis. Int J Vitam Nutr Res. 2011; 81(23):101-7. [21] Neves-Souza PC, Azeredo EL, Zagne SM, Valls-de-Souza R, Reis SR, Cerqueira DI, Nogueira RM, Kubelka CF. Inducible nitric oxide synthase (iNOS) expression in monocytes during acute Dengue Fever in patients and during in vitro infection.BMC Infect Dis. 2005 Aug 18;5:64.

ACCEPTED MANUSCRIPT [22] Smith PK, Krohn RI, Hermanson GT, Mallia AK, Gartner FH, Provenzano MD, Fujimoto EK, Goeke NM, Olson BJ, Klenk DC. Measurement of protein using bicinchoninic acid. Anal Biochem. 1985 Oct;150(1):76-85. Erratum in: Anal Biochem 1987 May 15;163(1):279.

RI P

T

[23] Carvalho PC, Fischer JS, Xu T, Cociorva D, Balbuena TS, Valente RH, Perales J, Yates JR 3rd, Barbosa VC. Search engine processor: Filtering and organizing peptide spectrum matches. Proteomics. 2012; 12(7):944-9. [24] Eng JK, Jahan TA, Hoopmann MR. Comet: an open source tandem mass spectrometry sequence database search tool. Proteomics. 2013;13(1):22-4

SC

[25] Zhang B, Chambers MC, Tabb DL. Proteomic parsimony through bipartite graph analysis improves accuracy and transparency. J Proteome Res. 2007; 6(9):3549-3557

NU

[26] Zhang Y, Wen Z, Washburn MP, Florens L. Refinements to label free proteome quantitation: how to deal with peptides shared by multiple proteins. Anal Chem. 82 (2010) 2272–2281.

MA

[27] Carvalho PC, Iii JRY, Barbosa VC. Improving the TFold test for differential shotgun proteomics. Bioinormatics 28 (2012) 1652–1654. [28] Avirutnan P, Malasit B, Seliger B, Bhakdi S & Husmann M. Dengue virus infection of endothelial cells leads to chemokine production, complement activation, and apoptosis. Journal of Immunology. 1998; 161(11):6338-6346.

ED

[29] Bonner SM & O´Sullivan MA. Endothelial cell monolayers as a model system to investigate dengue shock syndrome. Journal of Virology Methods. 1998;71(2):159-167.

PT

[30] Póvoa TF, Alves AM, Oliveira CA, Nuovo GJ, Chagas VL, Paes MV. The pathology of severe dengue in multiple organs of human fatal cases: histopathology, ultrastructure andvirus replication. PLoS One. 2014; 15;9(4):e83386.

CE

[31] Tsiatsiani L, Heck AJ. Proteomics beyond trypsin. FEBS J. 2015; 282(14):2612-26.

AC

[32] Stastna M, Van Eyk JE. Secreted proteins as a fundamental source for biomarker discovery. Proteomics. 2012; 12(4-5):722-35. [33] Fricker LD. Neuropeptide-processing enzymes: applications for drug discovery. AAPS J. 2005; 7(2):E449-55. [34] Fricker LD, Sweedler JV. Fishing for the hidden peptidome in health and disease (drug abuse). AAPS J. 2010; 12(4):679-82 [35] Jeppesen DK, Nawrocki A, Jensen SG, Thorsen K, Whitehead B, Howard KA, Dyrskjøt L, Ørntoft TF, Larsen MR, Ostenfeld MS. Quantitative proteomics of fractionated membrane and lumen exosome proteins from isogenic metastatic and nonmetastatic bladder cancer cells reveal differential expression of EMT factors. Proteomics 2014 Mar;14(6):699-712. [36] Slany A, Haudek VJ, Zwickl H, Gundacker NC, Grusch M, Weiss TS, Seir K, Rodgarkia-Dara C, Hellerbrand C, Gerner C. Cell characterization by proteome profiling applied to primary hepatocytes and hepatocyte cell lines Hep-G2 and Hep-3B. J Proteome Res. 2010 Jan;9(1):6-21. [37] Yamashita R, Fujiwara Y, Ikari K, Hamada K, Otomo A, Yasuda K, Noda M, Kaburagi Y. Extracellular proteome of human hepatoma cell, HepG2 analyzed using two-dimensional liquid chromatography coupled with tandem mass spectrometry. Mol Cell Biochem. 2007 Apr;298(1-2):83-92. Epub 2006 Nov 16.

ACCEPTED MANUSCRIPT [38] Bruschi M, Santucci L, Ravera S, Candiano G, Bartolucci M, Calzia D, Lavarello C, Inglese E, Ramenghi LA, Petretto A, Ghiggeri GM, Panfoli I. Human urinary exosome proteome unveils its aerobic respiratory ability. J Proteomics. 2016 Mar 16;136:25-34.

T

[39] Mangrum JB, Martin EJ, Brophy DF, Hawkridge AM. Intact stable isotope labeled plasma proteins from the SILAC-labeled HepG2 secretome. Proteomics. 2015 Sep;15(18):3104-15.

RI P

[40] Srisomsap C, Sawangareetrakul P, Subhasitanont P, Chokchaichamnankit D, Chiablaem K, Bhudhisawasdi V, Wongkham S, Svasti J. Proteomic studies of cholangiocarcinoma and hepatocellular carcinoma cell secretomes. J Biomed Biotechnol. 2010;2010:437143.

SC

[41] Zhang H, Wu P, Chen F, Hao Y, Lao Y, Ren L, Sun L, Sun W, Wei H, Chan DW, Jiang Y, He F. SILAC-based quantitative proteomic analysis of secretome between activated and reverted hepatic stellatecells. Proteomics. 2014 Sep;14(17-18):1977-86.

MA

NU

[42] Chen CY, Chi LM, Chi HC, Tsai MM, Tsai CY, Tseng YH, Lin YH, Chen WJ, Huang YH, Lin KH. Stable isotope labeling with amino acids in cell culture (SILAC)-based quantitative proteomics study of athyroid hormone-regulated secretome in human hepatoma cells. Mol Cell Proteomics. 2012 Apr;11(4):M111.011270. [43] Lu Y, Li S, Ma L, Li Y, Zhang X, Peng Q, Mo C, Huang L, Qin X, Liu Y. Type conversion of secretomes in a 3D TAM2 and HCC cell co-culture system and functional importance of CXCL2 in HCC. Sci Rep. 2016 Apr 27;6:24558.

ED

[44] Villarreal L, Méndez O, Salvans C, Gregori J, Baselga J, Villanueva J. Unconventional secretion is a major contributor of cancer cell line secretomes. Mol Cell Proteomics. 2013 May;12(5):1046-60.

CE

PT

[45] Knebel B, Lehr S, Hartwig S, Haas J, Kaber G, Dicken HD, Susanto F, Bohne L, Jacob S, Nitzgen U, Passlack W, Muller-Wieland D, Kotzka J. Phosphorylation of sterol regulatory element-binding protein (SREBP)-1c by p38 kinases, ERK and JNK influences lipid metabolism and the secretome of human liver cell line HepG2. Arch Physiol Biochem. 2014 Dec;120(5):216-27. [46] Martinez-Vicente M, Sovak G, Cuervo AM. Protein degradation and aging. Exp Gerontol. 2005; 40(89):622-33.

AC

[47] Thepparit C, Khakpoor A, Khongwichit S, Wikan N, Fongsaran C, Chingsuwanrote P, Panraksa P, Smith DR. Dengue 2 infection of HepG2 liver cells results in endoplasmic reticulum stress and induction of multiple pathways of cell death. BMC Res Notes. 2013; 6:372. [48] Jheng JR, Ho JY, Horng JT. ER stress, autophagy, and RNA viruses. Front Microbiol. 2014; 5:388. [49] Ekeowa UI, Gooptu B, Belorgey D, Hägglöf P, Karlsson-Li S, Miranda E, Pérez J, MacLeod I, Kroger H, Marciniak SJ, Crowther DC, Lomas DA. alpha1-Antitrypsin deficiency,chronic obstructive pulmonary disease and the serpinopathies. Clin Sci (Lond). 2009; 116(12):837-50. [50] Crowther DC, Belorgey D, Miranda E, Kinghorn KJ, Sharp LK, Lomas DA. Practical genetics: alpha1-antitrypsin deficiency and the serpinopathies. Eur J Hum Genet. 2004; (3):167-72. [51] Seidah NG, Prat A. Precursor convertases in the secretory pathway, cytosol and extracellular milieu. Essays Biochem. 2002; 38:79-94. [52] Apte SS. A disintegrin-like and metalloprotease (reprolysin type) with thrombospondin type 1 motifs: the ADAMTS family. Int J Biochem Cell Biol. 2004; (6):981-5. [53] Maroney SA, Mast AE. New insights into the biology of tissue factor pathway inhibitor. J Thromb Haemost. 2015;Suppl 1:S200-7.

ACCEPTED MANUSCRIPT

[54] Belorgey D, Irving JA, Ekeowa UI, Freeke J, Roussel BD, Miranda E, Pérez J, Robinson CV, Marciniak SJ, Crowther DC, Michel CH, Lomas DA. Characterization of serpin polymers in vitro and in vivo. Methods. 2011; 53(3):255-66.

RI P

T

[55] Page MJ, Di Cera E. Serine peptidases: classification, structure and function. Cell Mol Life Sci. 2008; 65(7-8):1220-36. [56] Wardman JH, Fricker LD. ProSAAS-derived peptides are differentially processed and sorted in mouse brain and AtT-20 cells. PLoS One. 2014; 9(8):e104232.

SC

[57] Ondrovicová G, Liu T, Singh K, Tian B, Li H, Gakh O, Perecko D, Janata J, Granot Z, Orly J, Kutejová E, Suzuki CK. Cleavage site selection within a folded substrate by the ATP-dependent lon protease. J Biol Chem. 2005; 280(26):25103-10.

NU

[58] Cassone M, Gagne AL, Spruce LA, Seeholzer SH, Sebert ME. The HtrA protease from Streptococcus pneumoniae digests both denatured proteins and the competence-stimulating peptide. J Biol Chem. 2012; 287(46):38449-59

MA

[59] Caughey GH. Mast cell proteases as pharmacological targets. Eur J Pharmacol. 2015. pii: S00142999(15)00407-0.

ED

Figure legends

PT

Figure 1: A schematic overview of the strategies and methodologies applied to perform the mass spectrometry analysis of the samples obtained from the ultrafiltrate of the mock-infected and DENVinfected conditions.

AC

CE

Figure 2: The Venn diagrams constructed from the molecules identified in the mock-infected condition (A) and DENV-infected condition (B) after distinct enzymatic treatments (applying the bottom-up strategy) using the EASY-nLC-II System (Thermo Scientific, USA) coupled to Orbitrap Velos mass spectrometers. Endoproteinases: Lys-C, Tryp (trypsin) and Lys-C+Tryp (digestion by Lys-C followed by trypsin). The raw data was run using Comet program against the Uniprot (dengue virus) and NextProt (Homo sapiens) databases. Figure 3: The Venn diagrams constructed after molecule identification. The molecules only found in the mock infected condition (M) are shown in blue. The molecules only found in the DENV-infected condition (DENV) are shown in red, and the shadowed area represents the molecules common to both conditions. (A) The bottom-up strategy, (B) the top-down strategy and (C) a compilation of the molecules identified by both strategies using mass spectrometers (nESI LTQ Orbitrap XL and Orbitrap Velos). The raw data was run using Comet program against the Uniprot (dengue virus) and NextProt (Homo sapiens) databases. Figure 4: Molecules identified in both conditions during the secretome [3] analysis but only in one condition during the ultrafiltrate analysis. Figure 5: Alignment of sequences on Uniprot tool. Identified sequences on carboxy-terminal region from human fibrinogen (P02671-2 = Isoform 2 of Fibrinogen alpha chain, residues 601 to 644) among the conditions, indicating differential proteolysis. Sequences presents only in DENV-infected condition (red), mock-infected condition (blue), and common to both conditions (black). Figure 6: The analysis of the exclusive sequences presents in the mock-infected condition (blue) and the DENV-infected condition (red) by the PeptideCutter program. The tool was used to predict possible sites of cleavage after in silico digestion. All enzymes were considered in the analysis.

ACCEPTED MANUSCRIPT Figure 7: Scheme representing the regions of molecular processing, topology and domains from proteins found in the mock condition (a and b) and DENV condition (c and d). The region that corresponds to the localization of the identified peptides is highlighted in red.

AC

CE

PT

ED

MA

NU

SC

RI P

T

Figure 8: Schematic diagram of the proSAAS processing with emphasis in the generation of a proSAASderived peptide (KEP molecule). The region that corresponds to the localization of the identified peptide is highlighted in red.

NU

SC

RI P

T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

MA

Fig. 1

MA

NU

SC

RI P

T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

Fig. 2

CE AC

Fig. 3

PT

ED

MA

NU

SC

RI P

T

ACCEPTED MANUSCRIPT

RI P

T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

MA

NU

SC

Fig. 4

RI P

T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

MA

NU

SC

Fig. 5

MA

NU

SC

RI P

T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

Fig. 6

SC

RI P

T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

MA

NU

Fig. 7

MA

NU

SC

RI P

T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

Fig. 8

ACCEPTED MANUSCRIPT Biological Significance Since the liver, one of the targets of DENV infection, is responsible for producing molecules involved in distinct biological processes, the identification of proteins and

RI P

T

peptides secreted by hepatocytes after infection would help to a better understanding of the physiopathology of dengue. Proteomic analyses of molecules with less than 10 kDa secreted by HepG2 cells after infection with DENV revealed differential proteolytic

AC

CE

PT

ED

MA

NU

SC

processing as an effect of DENV infection.

NU

SC

RI P

T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

MA

Graphical abstract

ACCEPTED MANUSCRIPT Highlights 

Small molecules from the secretome of DENV-infected HepG2 cells were identified. Bottom-up

and

top-down

approaches allowed

on

protein

Differential proteolytic processing of the secreted molecules due to infection was

CE

PT

ED

MA

NU

SC

observed.

AC



RI P

identification

improvement

T

