# Source paper archive

- Source title: Mapping the Plant Extracellular Proteome: From Fraction-Specific Secretion Patterns to Experimental Guidelines
- Author(s): Cerny, M., Wang, P.
- Source landing URL: https://www.biorxiv.org/content/10.64898/2026.11.16.766913v1
- PDF URL: https://www.biorxiv.org/content/10.64898/2026.11.16.766913v1.full.pdf
- PDF SHA-256: dd22dd527b7a0217fec477dfac8c5319d3c8285831ddd966005225755f870b6b
- Converter: pymupdf4llm 0.0.27
- Limits: Automated PDF text extraction; reading order, equations, tables and formatting may be incomplete or incorrect. Images are omitted, and no OCR was performed. Markup and links are neutralized for safe downloading. Consult the original PDF for the authoritative version.

---

&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)

# Title page


**Mapping the plant extracellular proteome: from fraction-specific secretion patterns to**

**experimental guidelines**


**Authors: Pengxi Wang** **&#91;1&#93;** **, Martin** Č **erný** **&#91;1*&#93;**


1 Department of Molecular Biology and Radiobiology, Faculty of AgriSciences, Mendel


University in Brno, 613 00 Brno, Czech Republic;


Pengxi Wang penxiwang@gmail.com


***Correspondence:** penxiwang@gmail.com; martincerny83@gmail.com; martincerny@mendelu.cz


**Tables: 2**


**Figures: 4**


1


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


**Mapping the plant extracellular proteome: from fraction-specific secretion patterns to**


**experimental guidelines**


**Abstract**

Proteins operating beyond the plasma membrane are central to how plants build and remodel their


cell walls, acquire and redistribute nutrients, coordinate developmental processes, and communicate


between cells and tissues. Moreover, the reach of this extracellular proteome extends beyond the

plant itself, with root and seed exudates shaping the rhizosphere and spermosphere by providing

molecular signals and resources that influence microbial communities even before pathogens or


symbionts establish direct contact with the plant surface. The extracellular space is therefore a

dynamic biological interface connecting processes within the plant to interactions with the soil,


microbiome, and broader environment. Despite this central role, the plant extracellular proteome


remains fragmented across experimental approaches. Here, we review current techniques for the

collection and purification of major plant extracellular fractions and use a uniform prediction


pipeline to assess representative publicly available datasets. This analysis reveals consistent


fraction-specific patterns in predicted secretion-class composition, with putative

unconventional-secretion candidates exceeding canonical candidates in most fractions. We then


present a decision tree linking biological questions to appropriate collection and fractionation

strategies, together with complementary experimental criteria for assessing extracellular origin that

extend beyond sequence-based prediction. Together, these resources provide a framework for


interpreting existing extracellular proteomic datasets and for designing future studies that account for


the biological distinctness of individual extracellular fractions. They are intended both as an entry

point to plant extracellular proteomics and as a reference framework for improving the biological

interpretation and comparability of studies across this diverse extracellular landscape.


Keywords: Plant extracellular proteome; sample collection; separation and purification; apoplast;


extracellular vesicles; proteomics; method selection


2


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


**Introduction**

The release of proteins into the extracellular environment enables organisms to interact with


neighboring cells and their surroundings. Animals, plants, fungi and unicellular eukaryotes share an

endoplasmic reticulum-Golgi secretory pathway, although their extracellular environments and

additional export routes differ; bacteria release proteins through distinct secretion systems and


membrane vesicles (Wang et al., 2018; Toyofuku et al., 2023). Plants additionally secrete proteins


through unconventional routes that bypass or repurpose the canonical pathway (Ding et al., 2014).

The plant cell wall forms a further interface between the plasma membrane and the environment:

proteins recovered beyond the membrane may remain soluble in the apoplast, associate with the wall


matrix, or occur within extracellular vesicles (EVs) (Rutter and Innes, 2017).


Plant extracellular proteins participate in cell wall remodeling, nutrient transport, intercellular


communication, growth and development, and responses to biotic and abiotic stress (Gao et al., 2024;

Chen et al., 2025a ). Signaling peptides, proteases and cell wall-modifying enzymes further mediate


extracellular immunity and defense against pathogens (Jian et al., 2024). For example, leaf apoplast


proteomics has revealed changes in extracellular protein composition during cold acclimation in

jojoba (Zheng et al., 2026) and _Ammopiptanthus mongolicus_ (Zheng et al., 2023), and


pattern-triggered immunity alters the abundance of defense-related proteins in the _Arabidopsis_ leaf

apoplast (Chen et al., 2025).


Proteins can reach these extracellular samples through several routes: canonical secretion,


unconventional secretion, developmentally programmed cell death, and leakage from damaged cells.

The protein composition of sequentially collected xylem sap fractions may vary following stem
cutting (Kopecká and Č erný, 2024). Developmentally regulated cell death in the Arabidopsis root


cap may contribute intracellular proteins to extracellular samples (Wang et al., 2024). Such

observations show that the sample collected and the fraction ultimately analyzed are not necessarily


equivalent: apoplastic washing fluid, wall-bound extracts, EV preparations and exudates each


recover different protein populations (Fig. 1). We therefore use extracellular proteome for the

proteins recovered from a given sample, and secretome for proteins actively released from cells


through secretion pathways. These terms are not necessarily interchangeable because extracellular


samples may also contain proteins released through cellular damage or leakage.


Recent advances in plant extracellular vesicle markers, isolation methods, and characterization

practices, together with the increasing availability of public proteomics datasets, have created

opportunities for comparative analysis (Perez-Riverol et al., 2022; Thieron et al., 2024; Rodríguez de


Lope et al., 2025). Nevertheless, the interpretation of extracellular proteomic data depends on both


the starting sample and the fraction ultimately analyzed. Unfractionated leaf apoplastic washing fluid,

its soluble fraction, and an EV-enriched fraction cannot be treated as equivalent samples. Moreover,

most proteins identified in an Arabidopsis leaf EV preparation lacked a predicted classical signal


peptide, illustrating why signal-peptide prediction alone cannot establish how proteins reached the

extracellular space (Rutter and Innes, 2017). Differences in sample sources, collection and isolation


procedures, and protein secretion and localization prediction tools complicate comparisons across


3


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


studies. Whether common patterns in predicted protein classes emerge across datasets obtained from


different plant materials and isolation procedures therefore remains an important question.


This review compares methods for collecting plant extracellular samples and separating their

protein-containing fractions. The numbers and relative abundances of predicted protein classes are

then examined across nine proteomic datasets. Based on these findings and published


methodological evidence, a decision tree is developed to guide method selection. We then discuss


proteomics-based strategies for evaluating extracellular association, quality control, and the limits of

inferring protein release mechanisms.


**Fig. 1. Workflow for extracellular proteomic profiling in plants.**


**1. Biological origins of proteins recovered from plant extracellular samples**


**Secretory routes across kingdoms and in plants**


Protein secretion occurs across all kingdoms of life through both conserved and lineage-specific

mechanisms. Animals, plants, fungi, and other unicellular eukaryotes share the conventional


endoplasmic reticulum (ER)-Golgi secretory pathway but also employ unconventional protein


secretion (UPS) routes and release extracellular vesicles, although their extracellular environments

and additional export routes differ (De Vallée et al., 2023; Pope and Prade, 2026). Plant and fungal

cell walls also affect protein recovery. Bacteria instead use Sec/Tat translocation pathways,


specialized secretion systems, and membrane vesicles to export proteins (Toyofuku et al., 2023).

In plants, the canonical secretory pathway typically begins with an N-terminal signal peptide


directing a protein into the ER, followed by trafficking through the Golgi apparatus and trans-Golgi


network toward the cell surface. UPS encompasses routes that deviate from this pathway. Four UPS


4


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


routes have been proposed: Golgi bypass (type I), vacuole-mediated release (type II), multivesicular


body (MVB)-mediated release (type III), and exocyst-positive organelle (EXPO)-mediated release

(type IV) (Ding et al., 2014). Importantly, these routes describe potential transport mechanisms


rather than protein classes that can be assigned from sequence alone. The absence of a predicted

signal peptide therefore does not establish that a protein was released through UPS (Robinson et al.,

2016).


MVBs and EXPOs illustrate the diversity of vesicle-associated protein export in plants. MVBs


contain intraluminal vesicles that can be released into the apoplast following fusion with the plasma

membrane. In _Arabidopsis_, EVs have been implicated in the transfer of small RNAs to the fungal

pathogen _Botrytis cinerea_ (Cai et al., 2018). More recent work showed that TET8- and


PEN1-marked EVs form largely distinct populations whose release responds to biotic stress (Koch et

al., 2025). The functional importance of EV protein cargo was further demonstrated in rice, where


EV-associated defense proteins were transferred into _Rhizoctonia solani_ and reduced disease


susceptibility (Huang et al., 2025). EXPO, first characterized as a double-membrane organelle

involved in cytosol-to-cell-wall export, represents another vesicle-associated route (Wang et al.,


2010). Recent structural and functional analyses identified Exo70E2-positive EVs derived from


EXPO and linked this pathway to plant defense (Gao et al., 2026). Together, these studies

demonstrate the diversity of plant EV populations and their possible biological origins. However,


identifying a protein in an EV-enriched fraction does not establish its individual export route,

particularly because EV preparations may also contain co-isolated non-vesicular material.


**Other origins of proteins in collected samples**

In addition to canonical and unconventional secretion, proteins may enter extracellular samples

through natural cell detachment, developmentally programmed cell death, or cellular damage during


sample collection and processing (Charura et al., 2024). Root border cells naturally detach from the

root cap and contribute intracellular and extracellular material to the surrounding environment


(Ropitaux et al., 2020; Kranawetter and Sumner, 2025). Distinct from cell detachment, programmed


cell death can also release cellular contents. Studies in _Arabidopsis_ have shown that root cap cell

death and the subsequent clearance of cellular remains are regulated developmental processes


associated with microbial colonization (Charura et al., 2024; Wang et al., 2024). Developmental


timing is also relevant to seed samples. In barley, aleurone cells secrete hydrolytic enzymes during

germination and subsequently undergo hormonally regulated programmed cell death (Domínguez


and Cejudo, 2014). These processes may contribute intracellular proteins to extracellular samples,

although their contribution to individual proteomic datasets requires independent evidence.


Sample collection and processing may introduce intracellular proteins through cellular damage.


Tissue homogenization, stem cutting for xylem sap collection, leaf infiltration, and handling of
germinating seeds may disrupt cells and release their contents. Kopecká and Č erný (2024) examined

sequentially collected xylem sap fractions, highlighting the importance of collection procedures


when interpreting recovered proteins. However, proteins predicted to have intracellular localization

cannot automatically be attributed to sampling damage, as they may also arise from regulated release,


developmental processes, or incompletely characterized export pathways. Protein origin must


5


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


therefore be interpreted in the context of sample collection, tissue integrity, and independent


localization or functional evidence.


**2. Collection of plant samples for extracellular proteomics**

The choice of sampling method determines which extracellular proteins can be collected before any


protein extraction or purification begins. Four targets guide sampling: soluble apoplastic proteins,


cell wall-bound proteins, EV-associated proteins, and proteins released into the environment

surrounding the plant (Rutter and Innes, 2017). These targets are not separate anatomical spaces. For

example, apoplastic washing fluid (AWF) contains soluble proteins and may also contain cell wall


proteins and EVs, whereas a root exudate can contain free proteins, vesicles, and material released

with root-cap cells (Kranawetter and Sumner, 2025).


**Soluble apoplastic proteins**


Infiltration followed by low-speed centrifugation is the principal approach for collecting leaf AWF.


Water or a defined buffer is introduced into the intercellular space by vacuum or syringe infiltration;

fluid is then recovered from the tissue by centrifugation. Recent Arabidopsis studies have used this


approach to profile the leaf apoplast during immune responses and to examine the effects of

collection conditions on protein recovery and cytosolic leakage ( Chen et al., 2025a ; Chen, Karinshak

and Vierstra, 2025b ). Syringe infiltration is useful for individual leaves or small experiments, while


vacuum infiltration can process more tissue in parallel. Gravity-based collection and water


displacement have also been applied to leaves or stems, but are less established across species

(Pechanova et al., 2010; Wu et al., 2014).


Methods should be compared in terms of fluid recovery per unit tissue mass and the composition of

the recovered material, rather than assuming that any method yields an inherently pure extracellular


sample. Infiltration dilutes the native apoplastic fluid, while excessive infiltration pressure or


centrifugal force may damage cells and introduce intracellular material into the recovered fluid.

Tissue age, buffer composition, infiltration time, and centrifugal force should therefore be reported


to facilitate comparisons across studies. AWF is a starting sample rather than an isolated


soluble-protein fraction, because soluble proteins and EVs can be separated during subsequent

processing (Rutter and Innes, 2017; Eldahshoury et al., 2024). For example, soluble apoplastic


proteins were recovered as the supernatant after ultracentrifugation of the total apoplastic extract at

200,000 × g for 20 min at 4 °C (Lee et al., 2025).


**Cell wall bound proteins**


Cell wall bound proteins require a different sampling decision because many remain associated with

the wall after soluble AWF has been removed. One route begins with intact leaves, stems, or roots


and exposes accessible wall proteins to an infiltration solution. The other begins with harvested

tissue that will undergo disruption and preparation of a wall-enriched fraction. A comparison of


sugarcane stem methods showed that non-destructive and destructive approaches recovered different


6


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


sets of cell wall proteins (Calderan-Rodrigues et al., 2016). Recent studies have also sampled roots


and leaves for cell wall proteomics under stress conditions (Chen et al., 2022).


For cell wall-associated proteins, sampling records should specify the organ, developmental stage,

fresh mass, and tissue integrity before collection. Unlike AWF collection, preparation of a

wall-enriched fraction typically involves tissue disruption followed by separation of cell wall


material. Salt extraction may then be applied either to intact tissue or to a prepared wall fraction,


depending on the experimental workflow. Distinguishing these approaches is important because

proteins extracted from accessible wall surfaces and those recovered from wall-enriched

preparations may represent different protein populations (Albenne et al., 2013).


**EV associated proteins**


Plant EV studies first collect a fluid in which vesicles may be present; the vesicles are then isolated

from that fluid. Leaf AWF is the best developed starting material and has been used for EV


proteomics in Arabidopsis (Rodríguez de Lope et al., 2025). Other reported sources include root


exudates, conditioned culture medium, and phloem sap. EVs have been recovered from tomato root

exudates, Arabidopsis cell cultures, and melon phloem sap collected after stem incision (De Palma et


al., 2020; Yugay et al., 2023; Sánchez-López et al., 2024).

The source fluid and its collection procedure should be specified before describing the EV isolation


method. AWF, root exudates, culture medium, and vascular sap represent different extracellular


environments and may contain distinct non-vesicular components. Tissue homogenates and

expressed fruit juices require particular caution because mechanical disruption can release cellular

debris and intracellular vesicles, as well as generate vesicle-like particles during processing (Huang


et al., 2025; Yang et al., 2025). Consequently, the origin of vesicles recovered from disrupted tissues

cannot be determined by size alone. The choice of starting material and collection procedure


therefore influences the composition and biological interpretation of the isolated vesicle population.


**Proteins released outside the plant and other accessible fluids**


Root and seed exudates are collected from the surrounding medium without extracting an internal

tissue fraction. Root exudates can be obtained from plants grown in hydroponic systems or


transferred to a defined collection solution. This approach permits timed sampling, but plant transfer,

collection-medium composition, sampling duration, and microbial activity can influence the

recovered material (McLaughlin et al., 2025). In Arabidopsis, activation of stress-related


mitogen-activated protein kinases altered the proteins detected in root exudates (Strehmel et al.,


2017). Root-tip mucilage can also be collected separately from bulk root exudates; a wheat study

analyzed mucilage and soluble root-secreted proteins as distinct preparations (Staudinger et al.,

2022). Seed exudates are collected from the surrounding medium during seed imbibition. Their


composition may vary with seed integrity, imbibition duration, germination stage, and collection

volume, making these parameters important to report (Tiddia et al., 2023).


7


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


Xylem and phloem sap also yield proteins, although they are internal transport fluids before


collection (Carella et al., 2016). Xylem sap can be collected from a cut stem under root pressure or

with an applied-pressure method; recent proteomic work has used xylem sap to examine root–shoot

responses to flg22 (Kopecká and Č erný, 2024). Phloem sampling includes EDTA-facilitated

exudation from excised leaves, incision-derived sap in suitable species, and aphid stylectomy

(Broussard et al., 2023). An Arabidopsis proteomic study used EDTA-facilitated collection, while a


melon study obtained phloem sap by stem incision (Sánchez-López et al., 2024; Moreno-Sanguino et


al., 2026). Sap emerging from a wound is thus a valid sampling route, but it may contain material

released from damaged cells and adjacent tissues (Broussard et al., 2023). Wound exudate should

specify the tissue cut, the fluid collected, and the collection interval rather than imply a single


biological source.


Conditioned medium from suspension-cell cultures provides another accessible source. Cells can be


removed from the medium after a defined culture or treatment period, as demonstrated in secretome

studies of sorghum cells under heat stress (Ngcala et al., 2020) and Pinus nigra embryogenic cultures


(Pernis et al., 2026). Culture conditions and cell viability should be documented to assess potential


intracellular contamination. However, culture-medium proteomes may not fully represent the

extracellular environment of intact plant organs.


Specialized plant surfaces offer further sampling routes. Exudate from a wet stigma can be collected

as visible droplets or by gentle surface sampling; proteomic work on lily and olive used these


approaches to study the stigma secretion itself, rather than a whole stigma extract (Rejón et al.,


2013). Floral and extrafloral nectar can be collected from nectaries with a capillary or syringe, and

both have supported protein identification (Silva et al., 2020; Nascimento et al., 2018). Guttation

droplets are accessible at leaf margins, but current sampling studies mainly address their metabolites


and physiology; their value for broad extracellular protein profiling remains less established

(Routaboul et al., 2024). These surface fluids should be sampled at a defined developmental stage


and protected from pollen, microbes, and environmental deposits when the aim is to study plant


proteins.


**3. Separation and purification of plant extracellular proteins**


The samples described in Section 2 are starting materials for protein preparation. Their proteins may


be dissolved in a collected fluid, associated with cell-wall material, or carried by EVs. Each form

requires different processing. In particular, concentrating an entire AWF sample does not separate its

soluble proteins from vesicle-associated proteins. The intended fraction must therefore be defined


before a purification method is selected.


**Proteins in extracellular fluids and exudates**

Extracellular fluids and exudates can be processed to recover proteins from the collected sample or

to isolate a soluble-protein fraction after particle removal. AWF, root and seed exudates, xylem sap,


and cell-culture medium often require concentration before proteomic analysis. Centrifugal


ultrafiltration concentrates material above a selected molecular-weight cutoff and permits buffer


8


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


exchange. For example, _Arabidopsis_ AWF was concentrated using a 3-kDa filter before protein


quantification and peptide preparation ( Chen et al., 2025a ). In cotton fibers, the total apoplastic

fraction was separated by ultracentrifugation at 200,000 × g for 20 min into a soluble supernatant


and a particle-enriched pellet (Lee et al., 2025). When soluble proteins are the target, particles

should be removed before concentration, because ultrafiltration alone does not separate soluble

proteins from particles (Eldahshoury et al., 2024).


Once the target fluid fraction has been selected, lyophilization and protein precipitation provide

alternative methods for recovering proteins from the whole collected fluid or its soluble supernatant.

Arabidopsis root-exudate medium was filtered and lyophilized before protein digestion (Strehmel et


al., 2017). Proteins from cell-free sorghum suspension-culture medium were recovered by acetone

precipitation, followed by pellet solubilization (Ngcala et al., 2020). Xylem sap was lyophilized,


washed with organic solvents, and processed for protein digestion and peptide desalting (Kopecká

and Č erný, 2024).


These preparation methods can influence the proteins ultimately detected. During ultrafiltration,


small proteins and peptides may pass into the filtrate, depending on the molecular weight cutoff. For

example, a 10-kDa filter was used to obtain separate phloem-exudate fractions enriched in larger


proteins and smaller proteins or peptides, and both fractions were analyzed (Moreno-Sanguino et al.,

2026). Precipitation and solvent washing may also result in protein loss. Therefore, studies should

report both the preparation method and the fraction used for proteomic analysis, because different


procedures may recover different proteins.


**Cell wall associated proteins**

Recovering proteins retained in the cell wall matrix generally requires extraction beyond fluid

concentration. A common approach first enriches a cell wall fraction from disrupted tissue and


washes it to reduce soluble material. Proteins are then released from the isolated wall fraction with


salt solutions. CaCl � and LiCl were used to extract proteins from isolated cell wall fractions of

Brachypodium distachyon (Douché et al., 2021). Sequential CaCl � and LiCl extraction was used to


recover proteins held by weak or ionic interactions in Marchantia polymorpha (Kolkas et al., 2022).


The resulting extract is thus a defined salt extractable wall protein fraction, not the complete wall

proteome. Proteins that remain tightly attached may need a different approach, such as digestion of


the isolated wall fraction (Kolkas et al., 2022). Because cell wall preparation involves tissue

disruption, intracellular carryover remains an interpretation issue even after extensive washing. Cell

wall proteomics reviews likewise emphasize how extraction and wall preparation affect the


recovered proteins (Calderan-Rodrigues et al., 2019).


**EV associated proteins**


For studies of EVs present outside intact plant cells, the source of the vesicles matters as much as the

separation method. AWF or conditioned medium can contain vesicles released into an extracellular


fluid (Rutter and Innes, 2017; Yugay et al., 2023). Tissue homogenates can also yield membrane


particles, but their isolation does not establish that those particles existed extracellularly before


9


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


disruption. These preparations should not be treated as equivalent when interpreting an EV proteome


(Thieron et al., 2024; Rodríguez de Lope et al., 2025).


Plant AWF studies commonly begin by removing cells and debris, followed by differential

centrifugation to obtain EV-enriched fractions (Rutter and Innes, 2017; Gao et al., 2024). Successive

centrifugation steps remove different particle populations, while the final high-speed pellet may


contain EVs together with other sedimentable material (Huang et al., 2021; Welsh et al., 2024).


Successive centrifugation at 2,000, 10,000, and 25,000 × g was used before small EVs were enriched

from _Nicotiana benthamiana_ AWF at 100,000 × g (Eldahshoury et al., 2024). EV fractions obtained

from _Arabidopsis_ AWF at different centrifugal forces were also compared and further separated


using density gradients (Huang et al., 2021). The resulting pellet should therefore be described as an

EV-enriched fraction rather than a pure EV preparation (Welsh et al., 2024).


Further purification can reduce overlap between EVs and other components of the starting fluid

(Thieron et al., 2024). Density-gradient centrifugation separates particles by buoyant density,


whereas size-exclusion chromatography (SEC) separates larger particles from smaller soluble


components as they pass through a porous column (Liangsupree et al., 2021). Ultrafiltration can

concentrate a particle-containing fluid before further separation, but does not alone establish that the


retained particles are EVs. Other particles and macromolecules may also be retained, and membrane

clogging can affect recovery (Liangsupree et al., 2021). Ultrafiltration followed by SEC was used to

prepare EV fractions from _Arabidopsis_ and _Brassica oleracea_ AWF for proteomic marker analysis


(Rodríguez de Lope et al., 2025). These methods therefore serve different functions within a


separation workflow, and their performance should be considered in relation to the starting material

and the fraction selected for analysis (Thieron et al., 2024; Welsh et al., 2024).


Other approaches, including polymer precipitation, aqueous two-phase separation, immunoaffinity

capture, and microfluidic isolation, may also be considered for EV recovery (Weng et al., 2016;


Kırba ş et al., 2019; Jia et al., 2022). Polymer precipitation may co-recover non-vesicular


macromolecules, whereas immunoaffinity capture selects vesicles bearing the targeted surface

marker (Liangsupree et al., 2021; Jia et al., 2022). Their applicability to plant extracellular fluids


requires evidence from comparable samples (Thieron et al., 2024). Results obtained from animal


biofluids or disrupted plant tissues cannot, by themselves, establish their performance for EVs

collected from plant AWF or conditioned medium.


After an EV-enriched fraction has been obtained, its proteins can be extracted for proteomic analysis

(Gao et al., 2024). Detection in that fraction establishes association with the preparation, but does


not distinguish proteins enclosed within vesicles from those attached to vesicle surfaces or


co-isolated during separation (Welsh et al., 2024). Particle measurements, microscopy, protein

markers, and cargo-protection assays provide complementary evidence for characterizing the

preparation and assessing protein localization. Plant EV marker analysis provides a basis for


evaluating EV-enriched proteomes (Rodríguez de Lope et al., 2025), while reporting guidance

emphasizes distinguishing what each separation and characterization step demonstrates from what


remains uncertain (Welsh et al., 2024).


10


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


**4. Comparative analysis of plant extracellular protein fractions**


We reanalyzed nine plant extracellular proteomic datasets using six prediction tools: SignalP 6.0,


DeepTMHMM 1.0, LSPpred, OutCyte, DeepLoc 2.1 and ApoplastP (Table 1). The datasets included

suspension-cell culture medium, seed and root exudates, leaf apoplastic washing fluid (AWF), xylem

and phloem sap, an AWF-derived EV fraction, a soluble cotton fibre apoplastic fraction, and a


cell-wall protein preparation. Proteins were assigned to four prediction categories using the same


criteria in all nine datasets: canonical secretion candidate (C), unconventional secretion candidate

(U), intracellular-like (I), and other/unresolved (O). Canonical candidates required a predicted signal

peptide without an additional transmembrane helix. Among proteins with neither feature,


unconventional candidates required a high-confidence LSPpred "LSP" prediction; intracellular-like

candidates required LSPpred "Non-LSP", OutCyte "Intracellular", a DeepLoc prediction other than


"Extracellular", and ApoplastP "Non-apoplastic". Proteins meeting none of these criteria were


assigned to the other/unresolved class (Table 2; Fig. 2). Because the original studies used different

collection and fractionation procedures, we compared the protein compositions of the fractions


analyzed, rather than treating all nine datasets as equivalent samples of the extracellular proteome.


**Table 1. Bioinformatic tools and resources relevant to plant extracellular protein analysis**

|Tool /resource|Function|Website|Status|Reference|
|---|---|---|---|---|
|Phobius|Signal peptides and transmembrane topology|https://phobius.sbc.su.se/|Active|Kall et al.,2004|
|iPSORT|N-terminal protein sorting signals|https:/ipsort.hgc.jp/|Available|Bannai et al.,2002|
|Plant-mPLoc|Plant protein subcellular localization|https:/www.csbio.sjtu.edu.cn/bioinf/plant-muli/|Active|Chou and Shen,2010|
|Predotar|N-terminal organelle targeting sequences|https:/urgi.versailles.inra.fr/Tools/Predotar|Available|Smallet al.,2004|
|PTS1 predictor|Peroxisomal targeting signal 1|https://mendel.imp.ac.at/pts1/|Active|Neuberger et al.,2003|
|TargetP2.0|N-terminal secretory and organelle targeting signals|https:/services.healthtech.dtu.dk/services/TargetP-2.0/|Active|Almagro Armenteros et al.,2019|
|WoLF PSORT|Eukaryotic protein subcellular localization|https://wolfpsort.hgc.jp/|Available|Horton et al.,2007|
|Plant Cell Marker|Plant cell marker database|https:/www.tobaccodb.org/pcmdb/|Active|Jin et al.,2022|
|SUBA5|Arabidopsis subcellular localization database|https://suba.live/|Active|Hooper et al.,2022|
|cropPAL2020|Crop protein subcellular localization resource|https://croppal.org/|Active|Hooper et al.,2020|
|LSPpred|Plant leaderless secretory protein prediction|http:/1sppred.Ispdb.org/|Available|Lonsdale et al.,2023|
|Cell wall genomics|Plant cellwall resource links|https://cellwall.genomics.purdue.edu/inkslindex.html|Available|Website only|
|CAZy|Carbohydrate-active enzyme families and functions|https:/ww.cazy.org/|Active|Drula et al.,2022|
|DeepTMHMM 1.0|Transmembrane protein topology|https:/services.healthtech.dtu.dk/services/DeepTMHMM-1.0/|Active|Hallgren et al.,2022|
|ApoplastP|Apoplastic localization of plant proteins|https://apoplastp.csiro.au/|Available|Sperschneider et al.,2018|
|WallProtDB-2|Plant cell wall proteome database|https:/www.polebio.Irsv.ups-tlse.frallProtDB/|Available|San Clemente et al.,2022|
|SignalP 6.0|Signal peptides and cleavage stes|https://services.healthtech.dtu.dk/services/SignalP-6.0/|Active|Teufel et al.,2022|
|DeepLoc 2.1|Subcellular localization and membrane association|https://services.healthtech.dtu.dk/services/DeepLoc-2.1/|Active|Odum et al.,2024|
|OutCyte|Putative unconventional protein secretion|https:/www.outcyte.com/|Active|Zhao et al.,2019|
|BUSCA|Integrated subcellular localization prediction|http://busca.biocomp.unibo.it|Active|Savojardo et al.,2018|
|PROSITE|Protein families,domains and functional-site motifs|https://prosite.expasy.org/|Active|Sigrist et al.,2026|



The comparison revealed two major patterns and distinct dominant protein classes across the nine


datasets. First, unconventional candidates exceeded canonical candidates in both protein number and


relative abundance in seven datasets, whereas suspension-cell culture medium showed the opposite

pattern and root exudates exhibited similar proportions of the two predicted classes (Fig. 2J, K).

Second, the combined proportion of canonical and unconventional candidates was higher in relative


abundance than in protein number in seven datasets. Seed exudates and the cell-wall preparation

were exceptions, where these two classes accounted for 51.6% versus 45.3% and 26.7% versus 14.9%


of protein number and relative abundance, respectively (Fig. 2L). Finally, suspension-cell culture


medium, the soluble cotton-fibre apoplastic fraction and the cell-wall protein preparation showed the


11


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


highest proportions of canonical, unconventional and intracellular-like candidates, respectively.


Canonical candidates accounted for 28.1% of proteins and 39.0% of relative abundance in

suspension-cell culture medium, unconventional candidates for 40.3% and 47.2% in the soluble


cotton-fibre apoplastic fraction, and intracellular-like proteins for 50.1% and 65.2% in the cell-wall

preparation (Fig. 2J, K).


**Table 2. Plant extracellular protein sampling and class composition**































|Sample/dataset|Sam (Sp ele tc ioo nlle 2c )tion&lt;br&gt;c|(SeS ce tp ioa nr a 3t )i ;o an / ap lyu zr eif di c fa rt ai co tn&lt;br&gt;n ion|N|Canonical&lt;br&gt;(C)|Putat (i Uve UPS&lt;br&gt;)|Intracellular-like&lt;br&gt;(1)|Other&lt;br&gt;(0)|
|---|---|---|---|---|---|---|---|
|**Suspension-cell**&lt;br&gt;**culture** **medium**&lt;br&gt;**PXD021536**|Sorghum suspension-culture&lt;br&gt; medium|Fiter out cells;acetone-precipitate medium&lt;br&gt;proteins.&lt;br&gt;**Analyzed:**culture-medium extract.|270|76&lt;br&gt;28.1% | 39.0%|23&lt;br&gt;8.5% | 1.5%|12&lt;br&gt;4.4% | 1.1%|159&lt;br&gt;  58.9% | 58.4%|
|**Seed** **exudates**&lt;br&gt;**PXD081462**|10 barley seeds in 1mL water;&lt;br&gt; collect liquid after 1-4h&lt;br&gt; imbibition|Lyophilize and process using the authors'&lt;br&gt;methanol-based procedure.&lt;br&gt;**Analyzed:**seed-exudate proteins.|849|83&lt;br&gt;9.8% | 19.8%|355&lt;br&gt;41.8% | 25.5%|170&lt;br&gt;20.0% | 13.0%|241&lt;br&gt;  28.4% | 41.7%|
|**Root** **exudates**&lt;br&gt;**PXD006328**|Liquid collected around&lt;br&gt; hydroponic Arabidopsis roots|Filter (0.45μm),lyophilize,and prepare&lt;br&gt;proteins.&lt;br&gt;**Analyzed:**root-exudate proteins.|154|33&lt;br&gt;21.4% | 24.7%|34&lt;br&gt;22.1% | 22.4%|22&lt;br&gt;14.3% | 10.6%|65&lt;br&gt;  42.2% | 42.4%|
|**Leaf** **AWF**&lt;br&gt;**PXD060654**|Arabidopsis leaf AWF;vacuum&lt;br&gt; infiltration and low-speed&lt;br&gt; centrifugation|Concentrate with a 3-kDa filter.&lt;br&gt;**Analyzed:**AWF proteins,without EV/soluble&lt;br&gt;separation.|2,913|332&lt;br&gt;11.4%117.6%|657&lt;br&gt;22.6% | 33.1%|635&lt;br&gt;21.8% | 6.5%|1,289&lt;br&gt; 44.2%142.8%|
|**Xylem** **sap**&lt;br&gt;**PXD053406**|Sap collected in successive&lt;br&gt; fractions after stem cutting|Flash-freeze,lyophilize,solvent-wash,and&lt;br&gt;solubilize.&lt;br&gt;**Analyzed:**xylem-sap proteins.|3,222|233&lt;br&gt;7.2% | 8.8%|708&lt;br&gt;22.0% | 38.7%|1,094&lt;br&gt;34.0% | 24.2%|1,187&lt;br&gt;  36.8% | 28.3%|
|**AWF-derived** **EVs**&lt;br&gt;**PXD058098**|Arabidopsis leaf AWF|Clarify,ultrafilter,separate by SEC,and extract&lt;br&gt;EV proteins.&lt;br&gt;**Analyzed:**EV-enriched fraction.|1,145|112&lt;br&gt;9.8% | 4.9%|356&lt;br&gt;31.1% | 41.2%|211&lt;br&gt;18.4% | 5.6%|466&lt;br&gt; 40.7%148.3%|
|**Solubleapoplastic**&lt;br&gt;**fraction**&lt;br&gt;**PXD051721**|9-DPA Gossypium hirsutum&lt;br&gt; fibers;total apoplastic wash&lt;br&gt; fluid|Centrifuge at 200,000×g for 20 min;collect the&lt;br&gt;supernatant and acetone-precipitate proteins&lt;br&gt;**Analyzed:**APO_s200 soluble fraction.|340|51&lt;br&gt;15.0% | 22.0%|137&lt;br&gt;40.3% | 47.2%|52&lt;br&gt;15.3% | 9.8%|100&lt;br&gt;  29.4% | 21.0%|
|**Phloem** **exudate**&lt;br&gt;**PXD065325**|Exudate collected from cut&lt;br&gt; Arabidopsis leaves|Separate with a 10-kDa fiter.&lt;br&gt;**Analyzed:**combined lists from retained fraction&lt;br&gt;A and fitrate C.|546|43&lt;br&gt;7.9% | 7.5%|187&lt;br&gt;34.2% | 39.1%|127&lt;br&gt;23.3% | 18.5%|189&lt;br&gt; 34.6% | 34.9%|
|Cell-wall protein&lt;br&gt;PXD012761|Sugarcane leaf or internode&lt;br&gt; tissue|Extract cellwall roteins.&lt;br&gt;**Analyzed:**wall-enriched protein preparation|359|18&lt;br&gt;5.0% | 2.3%|78&lt;br&gt;21.7% | 12.6%|180&lt;br&gt;50.1% | 65.2%|83&lt;br&gt;  23.1% | 19.9%|
|Clasication:C=SignalP-posive,no  adifional  TM;U=no  signal  peptide  or  TM  and  high-cnfience  LSPpred-positive;I=stringent  intacelar-ike  rule;O=remaining  proteins.|Clasication:C=SignalP-posive,no  adifional  TM;U=no  signal  peptide  or  TM  and  high-cnfience  LSPpred-positive;I=stringent  intacelar-ike  rule;O=remaining  proteins.|Clasication:C=SignalP-posive,no  adifional  TM;U=no  signal  peptide  or  TM  and  high-cnfience  LSPpred-positive;I=stringent  intacelar-ike  rule;O=remaining  proteins.|Clasication:C=SignalP-posive,no  adifional  TM;U=no  signal  peptide  or  TM  and  high-cnfience  LSPpred-positive;I=stringent  intacelar-ike  rule;O=remaining  proteins.|Clasication:C=SignalP-posive,no  adifional  TM;U=no  signal  peptide  or  TM  and  high-cnfience  LSPpred-positive;I=stringent  intacelar-ike  rule;O=remaining  proteins.|Clasication:C=SignalP-posive,no  adifional  TM;U=no  signal  peptide  or  TM  and  high-cnfience  LSPpred-positive;I=stringent  intacelar-ike  rule;O=remaining  proteins.|Clasication:C=SignalP-posive,no  adifional  TM;U=no  signal  peptide  or  TM  and  high-cnfience  LSPpred-positive;I=stringent  intacelar-ike  rule;O=remaining  proteins.|Clasication:C=SignalP-posive,no  adifional  TM;U=no  signal  peptide  or  TM  and  high-cnfience  LSPpred-positive;I=stringent  intacelar-ike  rule;O=remaining  proteins.|
|&lt;br&gt;Nand percentages were recalculated from the protein-level worksheets after accession-level deduplication.U is a prediction;I is not a measured contamination rate.AWF=apoplastic|&lt;br&gt;Nand percentages were recalculated from the protein-level worksheets after accession-level deduplication.U is a prediction;I is not a measured contamination rate.AWF=apoplastic|&lt;br&gt;Nand percentages were recalculated from the protein-level worksheets after accession-level deduplication.U is a prediction;I is not a measured contamination rate.AWF=apoplastic|&lt;br&gt;Nand percentages were recalculated from the protein-level worksheets after accession-level deduplication.U is a prediction;I is not a measured contamination rate.AWF=apoplastic|&lt;br&gt;Nand percentages were recalculated from the protein-level worksheets after accession-level deduplication.U is a prediction;I is not a measured contamination rate.AWF=apoplastic|&lt;br&gt;Nand percentages were recalculated from the protein-level worksheets after accession-level deduplication.U is a prediction;I is not a measured contamination rate.AWF=apoplastic|&lt;br&gt;Nand percentages were recalculated from the protein-level worksheets after accession-level deduplication.U is a prediction;I is not a measured contamination rate.AWF=apoplastic|&lt;br&gt;Nand percentages were recalculated from the protein-level worksheets after accession-level deduplication.U is a prediction;I is not a measured contamination rate.AWF=apoplastic|
|&lt;br&gt;Inclusion required≥2 unique peptides where available;root exudates used≥1.EV peptide counts were unavailable,so the reported EV list was retained without peptide-count filtering.|&lt;br&gt;Inclusion required≥2 unique peptides where available;root exudates used≥1.EV peptide counts were unavailable,so the reported EV list was retained without peptide-count filtering.|&lt;br&gt;Inclusion required≥2 unique peptides where available;root exudates used≥1.EV peptide counts were unavailable,so the reported EV list was retained without peptide-count filtering.|&lt;br&gt;Inclusion required≥2 unique peptides where available;root exudates used≥1.EV peptide counts were unavailable,so the reported EV list was retained without peptide-count filtering.|&lt;br&gt;Inclusion required≥2 unique peptides where available;root exudates used≥1.EV peptide counts were unavailable,so the reported EV list was retained without peptide-count filtering.|&lt;br&gt;Inclusion required≥2 unique peptides where available;root exudates used≥1.EV peptide counts were unavailable,so the reported EV list was retained without peptide-count filtering.|&lt;br&gt;Inclusion required≥2 unique peptides where available;root exudates used≥1.EV peptide counts were unavailable,so the reported EV list was retained without peptide-count filtering.|&lt;br&gt;Inclusion required≥2 unique peptides where available;root exudates used≥1.EV peptide counts were unavailable,so the reported EV list was retained without peptide-count filtering.|


The proportion of unconventional candidates exceeded that of canonical candidates in both protein

number and relative abundance in seven of the nine datasets. A similar pattern was reported in sugar


beet leaf apoplastic fluid, where 38% of identified proteins were predicted to undergo


unconventional secretion, compared with 26% assigned to canonical secretion (Ceballos-Laita et al.,

2015). However, this relationship was not consistent across extracellular sample types. In oilseed


rape xylem sap, 42% of identified proteins were classified as canonical secretory proteins and 25%

as unconventional secretory proteins (Luo and Zhang, 2019), while a tomato xylem sap study

reported 74% predicted canonical secretory proteins and 9% predicted leaderless secretory proteins


(de Lamo et al., 2018). The suspension-cell culture medium likewise showed more canonical than


unconventional candidates, whereas root exudates contained similar proportions of the two classes.

Thus, although unconventional candidates predominated in seven datasets in our unified reanalysis,

published results from comparable extracellular fractions and the suspension-culture sample show


that this relationship is not universal. Differences in sample type, preparation and prediction criteria

should be considered when interpreting these comparisons. Together, these comparisons suggest that


the balance between predicted canonical and unconventional secretory protein classes reflects not


only the biology of secretion but also how the sample was obtained.


12


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


**Fig. 2. Plant extracellular protein sampling and predicted protein-class composition across**


**nine datasets.** (A – I) Collection routes and analyzed samples or fractions: leaf apoplastic washing


fluid (AWF; A), AWF used for extracellular vesicle (EV) isolation (B), soluble apoplastic fraction

obtained from AWF (C), root exudates (D), seed exudates (E), suspension-cell culture medium (F),

cell-wall protein preparation (G), xylem sap (H), and phloem-enriched fluid (I). (J, K) The


percentage of identified proteins (J) and the percentage of total protein abundance (K) assigned to

canonical secretion candidates (C), unconventional secretion candidates (U), intracellular-like


predictions (I), or unresolved proteins (O) within each dataset. Values within bars are rounded; n is


the number of identified proteins. (L) The percentage of identified proteins classified as C + U


13


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


plotted against their percentage of total protein abundance; the dashed line indicates equal


percentages. (M) The percentage of identified proteins classified as I plotted against the percentage

of total protein abundance classified as C + U; bubble area is proportional to the number of C + U


–
proteins. Letters in L and M refer to A I. The classes are predictions: the percentage classified as I is


not a measured contamination rate. Percentages describe each analyzed dataset, not protein yield


across methods. Exact values and dataset sources are given in Table 2.


Second, the combined proportion of canonical and unconventional candidates was higher in relative

abundance than in protein number in seven datasets. This indicates that proteins assigned to these


two predicted secretory classes had a higher mean relative abundance than proteins in the remaining

classes within those fractions. Previous analyses of plant apoplastic fluid have also shown that


protein-number and abundance-based profiles can differ substantially. In Arabidopsis leaf apoplastic


fluid, proteins assigned to the apoplast accounted for 29% of identified proteins but 70.3% of the

estimated abundance after adjustment for highly abundant Rubisco, whereas chloroplast-assigned


proteins accounted for 26.7% of protein number but 15.5% of abundance (Karimi et al., 2025). A


grapevine leaf apoplastic fluid study similarly found that pathogenesis-related proteins contributed

50.7% of the estimated protein content, illustrating how particular extracellular protein groups can


dominate the abundance profile (Delaunois et al., 2013). These studies support the importance of

considering protein abundance alongside protein number, although neither used our four-class

prediction framework and therefore neither independently establishes that the two predicted


secretion classes generally have a higher abundance proportion. In our comparison, seed exudates


and the cell-wall preparation showed the opposite relationship, with the two predicted secretion

classes accounting for 51.6% and 26.7% of protein number but only 45.3% and 14.9% of abundance,

respectively. The direction of this difference therefore depends on the protein composition of the


fraction analyzed and should not be treated as a universal feature of plant extracellular proteomes.


The highest proportions of canonical, unconventional and intracellular-like candidates were observed


in suspension-cell culture medium, the soluble cotton-fibre apoplastic fraction and the cell-wall

protein preparation, respectively. The high proportion of canonical candidates in suspension-cell


culture medium is consistent with an independent study of Medicago suspension cultures, in which


34 of 39 identified extracellular-medium proteins contained predicted secretion signals, although this

criterion is not identical to our canonical category (Kusumawati et al., 2008). The high proportion of


unconventional candidates in the cotton-fibre dataset should be interpreted specifically for the

soluble apoplastic fraction, as extracellular proteins differ in their association with the cell-wall

matrix and may be recovered differently by extraction procedures (Rose and Lee, 2010). In contrast,


the high proportion of intracellular-like proteins in the cell-wall preparation may partly reflect the


co-extraction of intracellular proteins: an independent study showed that improved wall purification

reduced predicted intracellular proteins and yielded a fraction in which 73% of identified proteins

were predicted to be secreted (Feiz et al., 2006). Nevertheless, an intracellular-like prediction alone


does not establish contamination.


Based on the target fractions, their preparation methods and the nine-dataset comparison, we propose


a decision tree for selecting plant extracellular protein sampling and fractionation methods (Fig. 3).


14


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


**Fig. 3. Decision tree for sampling and evaluating plant extracellular protein fractions.** The

research target determines one of four operational target fractions: (A) soluble apoplastic proteins,


(B) extracellular vesicle (EV)-associated proteins, (C) cell wall-associated proteins, or (D) proteins


recovered from exudates, vascular sap and culture medium. Each route links sample collection and

fraction preparation to relevant quality checks. The final check asks whether the controls support

enrichment of the intended fraction; weak support calls for better fractionation or a narrower


interpretation. AWF, apoplastic washing fluid; SEC, size-exclusion chromatography; QC, quality

control.


**5. Establishing extracellular protein origin from proteomics data**

Proteomics studies have employed several strategies to estimate whether detected proteins genuinely


originate from the extracellular compartment. A straightforward strategy is comparative analysis

against a whole-tissue or total-protein reference, in which proteins enriched in the extracellular


fraction relative to the corresponding tissue extract are considered more likely to represent genuine

extracellular components (Fig.4A, B). For example, Kourelis et al. (2019) classified proteins

detected in _Nicotiana benthamiana_ apoplastic fluid as apoplastic when they were either exclusively


detected in the apoplastic fraction or strongly enriched relative to the total extract (log � FC ≥ 1.5, p


≤ 0.01), while proteins enriched in the total extract were classified as intracellular. A second

approach exploits the temporal dynamics of extracellular proteome collection (Fig. 4C, D). Proteins

released through tissue disruption may be expected to be most abundant during the initial collection


period, whereas proteins continuously present in the extracellular fluid may persist during

subsequent sampling. Evidence for such temporal changes in xylem sap protein composition dates


back to studies showing that individual protein isoforms can be restricted to the initial phase of stem


exudation while others persist throughout collection (Biles and Abeles, 1991). Consecutive xylem


15


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


sap fractions have also been analysed by proteomics, including our analysis of _Solanum tuberosum_


and _Hordeum vulgare_, which demonstrated an enrichment of proteins annotated as extracellular in
subsequent collection fractions (Kopecká and Č erný, 2024). A third approach, and one that can


extend these principles, is to examine protein abundance as a function of extraction severity (Fig. 4E,


F). The underlying rationale is that increasing extraction force should have a progressively greater

effect on proteins released through cellular disruption, whereas proteins already present in the


extracellular compartment should be less dependent on extraction severity. Experimental evidence

supports this premise: in maize roots, cytoplasmic contamination of apoplastic fluid was detectable

after centrifugation even at 200 × g, whereas material collected using filter strips lacked detectable


glucose-6-phosphate (Dragiši ć Maksimovi ć et al., 2014). Similarly, recent proteomic analysis of


_Arabidopsis_ apoplastic fluid showed that the extraction procedure substantially influenced the

resulting proteome and affected the extent of intracellular contamination (Chen, Karinshak and

Vierstra, 2025b). These observations provide a rationale for treating extraction severity not only as a


methodological variable but also as a potential analytical dimension. Systematic sampling across an

extraction-severity gradient could therefore distinguish proteins whose abundance remains relatively


stable from those showing a progressive increase with increasing extraction stress, providing an

additional criterion for assessing extracellular versus leakage-derived protein populations.


**Fig. 4. Estimating the extracellular origin of proteins from proteomics data.** (A-B) Ratio-based

classification using a complementary whole-tissue dataset. Enrichment is expressed as log2FC =

log2(abundance in extracellular fraction / abundance in whole tissue). The rationale is that proteins

abundant in whole tissue and not enriched in the extracellular fraction are more likely to be leakage


contaminants than genuine extracellular components. (A) Standard classification with a single fixed

log2FC threshold. (B) Adaptive classification in which the threshold depends on protein abundance.


In this example, the adaptive threshold was applied with an abundance penalty factor of c = 2.5 and a

lowered base threshold of log2FC &gt; 0.4. Only proteins above a minimum abundance of I � = 2

(log2iBAQ) were evaluated. (C, D) Kinetic classification from time-resolved sampling of a


continuously collected fluid (e.g., exudate or xylem sap). Genuinely extracellular proteins


accumulate throughout the collection period. Leaked proteins are released mainly at the earliest time


16


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


points, after damage, and then plateau. (E, F) Classification by extraction-stress titration. Parallel


samples are collected at increasing stress (higher centrifugal force, or higher vacuum for apoplastic

fluid). Constitutively extracellular proteins remain relatively independent of stress level. Leaked


proteins increase with stress as cell rupture increases. (D, F) Relative-abundance representations of

the data shown in C and E. Because relative abundances are constrained by the total protein pool,

accumulation of leaked intracellular proteins can cause genuine extracellular proteins to appear


reduced even when their absolute abundance remains unchanged. This compositional bias should


therefore be considered when interpreting relative proteomics data. All plotted values and thresholds

are illustrative.


**6. Challenges and future directions**


**Determining the origin of extracellular proteins**


Detection of a protein in an extracellular sample establishes its presence in the analyzed fraction, but


does not show how it reached that fraction. Proteins may be secreted, released during normal


development, or introduced through sampling-induced cell damage. These sources are particularly

difficult to distinguish in seed and root exudates. Developmentally programmed cell death in the


Arabidopsis root cap illustrates a possible source of released cellular contents, but does not establish

the origin of any protein in the datasets analyzed here (Wang et al., 2024). Proteins may also have

multiple localizations or appear in extracellular fractions through association with extracellular


proteins or structures. Accordingly, the intracellular-like category in Table 2 is a prediction category,


not a measured contamination rate. Assignment to the putative canonical or unconventional category

likewise does not establish a specific secretion route (Robinson et al., 2016).


**Quality control of extracellular fractions**

No single marker or threshold is suitable for every plant extracellular protein preparation. Quality


control should reflect the target fraction. For leaf AWF, tissue integrity and the activity of a panel of

intracellular marker enzymes should be assessed alongside extracellular enrichment; no single


enzyme establishes contamination. Cell viability should also be monitored when collecting


suspension-culture medium, because stress-induced loss of viability may contribute proteins from

damaged cells (Ngcala et al., 2020). For cell wall preparations, studies should document wall


enrichment and washing steps. EV preparations require particle characterization, protein markers,

and appropriate controls. Comparing an EV-enriched fraction with a matched soluble apoplastic

fraction can help determine whether a protein is enriched with vesicles (Rodríguez de Lope et al.,


2025).


For exudates and vascular sap, sequential collection can help identify proteins potentially associated

with sampling damage. In one xylem sap study, some proteins were abundant in the first collected


fraction but declined rapidly thereafter, whereas proteins predicted to be extracellular were enriched

in later fractions. The authors inferred that the early fraction was more affected by tissue damage at

the cut surface (Kopecká and Č erný, 2024). Guidance for plant EV studies recommends


complementary particle and molecular characterization and complete reporting of isolation steps


17


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


(Thieron et al., 2024; Welsh et al., 2024). Together, these measures assess enrichment and sample


quality, but do not independently determine each protein’s route of release.


**Limitations of cross-study comparisons and localization prediction**

The nine datasets analyzed here differ in species, tissue, developmental stage, treatment, sample


preparation, and proteomics workflow. Protein counts and abundance profiles describe differences


among the final analyzed fractions, but cannot isolate the effect of any single collection or

purification step. Nor can the proportion assigned to the intracellular-like category be converted into

an experimentally verified contamination rate. The “Other” category, comprising proteins not


assigned to the canonical, unconventional or intracellular-like categories, is substantial in some

datasets: it accounts for 58.5% of identified proteins in the suspension-culture dataset. Thus, the


current classification does not resolve every protein.


Localization prediction helps organize candidates, but may miss proteins with multiple localizations


or possible moonlighting functions, or proteins that enter an extracellular fraction through


association with extracellular structures. Metabolic enzymes such as MDH, GAPDH and enolase

illustrate this ambiguity: their activities are routinely used to estimate cytosolic contamination of


apoplastic fluids (Alexandersson et al., 2013), yet MDH also occurs in the plant apoplast as a
genuinely wall-bound enzyme rather than a cytoplasmic artefact (Hadži-Taškovi ć Šukalovi ć et al.,

2011), and GAPDH is an established moonlighting protein with functions beyond the cytosol (Cao et


al., 2025). Their detection in a proteome therefore warrants investigation but does not alone prove


sampling damage. A recent benchmark of human protein localization predictors identified multiple

localization as a continuing challenge; as its test set comprised human proteins, its performance

estimates cannot be transferred directly to plants (Wefers et al., 2026). Predictions should therefore


serve as supporting annotations, rather than as a quality-control gate in the method-selection decision

tree or as proof of a protein's extracellular localization.


**Practical priorities for future studies**


A useful next step is to compare fractions prepared from the same biological material under clearly


documented conditions. Retaining the starting sample and intermediate fractions, with suitable

quantitative measurements, would help assess protein recovery and enrichment at each step. For


exudates and vascular sap, studies should record collection time and, where feasible, collect

sequential fractions. They should also report the amount of starting material, collection volume,

protein recovery, and basis for abundance normalization. Normalization solely to total protein may


obscure changes in the overall amount of protein released. Changes in protein composition should


therefore be distinguished from changes in total release; comparison of absolute release amounts for

individual proteins requires an appropriate quantitative standard.


These records and indicators of tissue damage are especially important when a stress treatment may

alter both extracellular protein release and cell viability. Candidates of particular biological interest


can then be examined using independent localization or functional experiments suited to the system.


Such approaches would improve comparability across studies while preserving a necessary


18


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


distinction: a well-characterized extracellular fraction supports an association between a protein and


that fraction, but proteomic detection alone does not establish how the protein was released.


**Declaration of generative AI and AI-assisted technologies in the manuscript preparation**


**process**

During the preparation of this manuscript, the authors used Microsoft Copilot (M365 Copilot, GPT-5


based) to improve language and readability. After using this tool/service, the authors thoroughly


reviewed and edited the content as needed and take full responsibility for the final published version

of the article.


19


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


**References**

Ødum, M.T., et al., 2024. DeepLoc 2.1: multi-label membrane protein type prediction using

protein language models. Nucleic Acids Res, 52: W215-W220.

Albenne, C., Canut, H., Jamet, E., 2013. Plant cell wall proteomics: the leadership of

Arabidopsis thaliana. Front Plant Sci, 4: 111.

Alexandersson, E., Ali, A., Resjö, S., Andreasson, E., 2013. Plant secretome proteomics. front

plant sci 4:9

Almagro Armenteros, J.J., et al., 2019. Detecting sequence signals in targeting peptides using

deep learning. Life Sci Alliance, 2: e201900429.

Bannai, H., et al., 2002. Extensive feature detection of N-terminal protein sorting signals.

Bioinformatics, 18: 298-305.

Biles, C.L., Abeles, F.B., 1991. Xylem sap proteins. Plant Physiol, 96(2): 597-601.

Broussard, L., Abadie, C., Lalande, J., Limami, A.M., Lothier, J., Tcherkez, G., 2023. Phloem

sap composition: what have we learnt from metabolomics? Int J Mol Sci, 24(8):6917

Cai, Q., Qiao, L., Wang, M., et al., 2018. Plants send small RNAs in extracellular vesicles to

fungal pathogen to silence virulence genes. science 360, 1126-1129

Calderan-Rodrigues, M.J., et al., 2016. Cell wall proteome of sugarcane stems: comparison of a

destructive and a non-destructive extraction method showed differences in glycoside hydrolases

and peroxidases. BMC Plant Biol, 16: 14.

Calderan-Rodrigues, M.J., et al., 2019. Int J Mol Sci 20:1975

Cao, Y., Hong, J., Wang, H., Lin, M., Cai, Y., Liao, L., Li, X., Han, Y, 2025. Beyond glycolysis:

multifunctional roles of glyceraldehyde-3-phosphate dehydrogenases in plants. Hortic Res,

12(6): uhaf070.

Carella, P., Wilson, D.C., Kempthorne, C.J., Cameron, R.K., 2016. Vascular sap proteomics:

providing insight into long-distance signaling during stress. Front Plant Sci, 7: 651.

Ceballos-Laita, L., Gutierrez-Carbonell, E., Lattanzio, G., et al., 2015. Protein profile of Beta

vulgaris leaf apoplastic fluid and changes induced by Fe deficiency and Fe resupply. Front Plant

Sci, 6: 145.

Charura, N., Llamas, E., De Quattro, C., Vilchez, D., Nowack, M.K., Zuccaro, A, 2024. Root

cap cell corpse clearance limits microbial colonization in Arabidopsis thaliana. elife 13,

RP96266

Chen, H.-C., Newton, C.J., Diaz, G., et al., 2025a. Proteomic snapshot of pattern triggered

immunity in the Arabidopsis leaf apoplast. Plant J, 123(6): e70498.

Chen, K.-E., Karinshak, M., Vierstra, R.D., 2025b. Streamline protocol for Arabidopsis

apoplastic fluid isolation enables a detailed proteomic view of the plant extracellular space.

Plant Direct, 9: e70087.

Chen, S., et al., 2022. Quantitative proteomics analysis of tomato root cell wall proteins in

response to salt stress. frontiers in plant science 13, 1023388

Chou, K.-C., Shen, H.-B, 2010. Plant-mPLoc: A top-down strategy to augment the power for

predicting plant protein subcellular localization. PLoS ONE, 5: e11335.

de, Lamo.F.J., Constantin, M.E., Fresno, D.H., Boeren, S., Rep, M., Takken, FLW, 2018. Xylem

sap proteomics reveals distinct differences between R gene- and endophyte-mediated resistance

against Fusarium wilt disease in tomato. Front Microbiol, 9: 2977.


20


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


De Vall é e, A., Dupuy, J.-W., Moriscot, C., et al., 2023. Extracellular vesicles of the plant


pathogen Botrytis cinerea. journal of fungi 9, 495

De Palma, M., et al., 2020. Plant roots release small extracellular vesicles with antifungal

activity. Plants, 9: 1777.

Delaunois, B., Colby, T., Belloy, N., et al., 2013. Large-scale proteomic analysis of the

grapevine leaf apoplastic fluid reveals mainly stress-related proteins and cell wall modifying

enzymes. BMC Plant Biol, 13(1): 24.

Ding, Y., Robinson, D.G., Jiang, L, 2014. Unconventional protein secretion (UPS) pathways in

plants. current opinion in cell biology 29, 107-115


Dom í nguez, F., Cejudo, F.J., 2014. Programmed cell death (PCD): an essential process of cereal


seed development and germination. Front Plant Sci, 5: 366.


Douch é, T., et al., 2021. Cell wall proteomic datasets of stems and leaves of brachypodium


distachyon. Data Brief, 35: 106818.


Dragi š i ć, Maksimovi ć, J., J., Ž ivanovi ć, B., D., Maksimovi ć, V., M., Mojovi ć, M., D., Nikoli ć,


M., T., Vu č ini ć, Ž ., B., 2014. Filter strip as a method of choice for apoplastic fluid extraction


from maize roots. Plant Science, 223: 49-58.

Drula, E., et al., 2022. The carbohydrate-active enzyme database: functions and literature.

Nucleic Acids Res, 50: D571-D577.

Eldahshoury, M.K., et al., 2024. Isolation of small extracellular vesicles from the apoplastic

wash fluid of Nicotiana benthamiana leaves. Curr Protoc, 4: e70026.

Feiz, L., Irshad, M., Pont-Lezica, R.F., Canut, H., Jamet, E., 2006. Evaluation of cell wall

preparations for proteomics: a new procedure for purifying cell walls from Arabidopsis

hypocotyls. Plant Methods, 2: 10.

Gao, Y.Q., Su, Y., Chao, D.Y., 2024. Exploring the function of plant root diffusion barriers in

sealing and shielding for environmental adaptation. Nat Plants, 10: 1865-1874.

Gao, J., Li, Y., Zhang, S., et al., 2026. Structural and functional characterization of

EXPO-derived extracellular vesicles in plants. advanced science 13, e06163


Had ž i-Ta š kovi ć, Š ukalovi ć, V., Vuleti ć, M., Markovi ć, K., Vu č ini ć, Ž, 2011. Cell


wall-associated malate dehydrogenase activity from maize roots. Plant Science, 181(4):

465-470.

Hallgren, J., et al., 2022. DeepTMHMM predicts alpha and beta transmembrane proteins using

deep neural networks. biorxiv &#91;preprint&#93;

Hooper, C.M., et al., 2022. Subcellular localisation database for Arabidopsis proteins version 5

&#91;dataset&#93;

Hooper, C.M., et al., 2020. CropPAL for discovering divergence in protein subcellular location

in crops to support strategies for molecular crop breeding. Plant J, 104: 812-827.

Horton, P., et al., 2007. WoLF PSORT: protein localization predictor. Nucleic Acids Res, 35:

W585-W587.


21


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


Huang, D., Chen, J., Zhao, M., Shen, H., Jin, Q., Xiao, D., Peng, Z., Chen, T., Zhang, Y., Rao,

D., Liu, M., 2025. Plant-derived extracellular vesicles: composition, function and clinical

potential. J Transl Med, 23: 1065

Huang, Y., Li, W., Liu, T., et al., 2025. Rice extracellular vesicles send defense proteins into

fungus Rhizoctonia solani to reduce disease. developmental cell 60, 1168-1181.e6

Huang, Y., Wang, S., Cai, Q., Jin, H, 2021. Effective methods for isolation and purification of

extracellular vesicles from plants. J Integr Plant Biol, 63: 2020-2030.

Jia, Y., Yu, L., Ma, T., et al., 2022. Small extracellular vesicles isolation and separation: current

techniques, pending questions and clinical applications. Theranostics, 12(15): 6548.

Jian, Y., Liu, Z., He, P., et al., 2024. An emerging connected view: phytocytokines in regulating

stomatal, apoplastic, and vascular immunity. Curr Opin Plant Biol, 82: 102623.

Jin, J., et al., 2022. PCMDB: A curated and comprehensive resource of plant cell markers.

Nucleic Acids Res, 50: D1448-D1455.

Jung, Y.H., et al., 2008. J proteome res 7:5187


K ı rba ş, O.K., Bozkurt, B.T., Asutay, A.B., Mat, B., Ozdemir, B., Özt ü rko ğ lu, D., Ölmez, H., İş


lek, Z., Ş ahin, F., Ta ş l ı, P.N., 2019. Optimized isolation of extracellular vesicles from various


organic sources using aqueous two-phase system. Sci Rep, 9(1): 19159.

Käll, L., Krogh, A., Sonnhammer, E.L.L., 2004. A combined transmembrane topology and

signal peptide prediction method. J Mol Biol, 338: 1027-1036.

Karimi, H.Z., Chen, K.E., Karinshak, M., et al., 2025. Proteasomes accumulate in the plant

apoplast where they participate in microbe-associated molecular pattern (MAMP)-triggered

pathogen defense. Nat Commun, 16(1): 1634.

Koch, B.L., Rutter, B.D., Borniego, M.L., et al., 2025. Arabidopsis produces distinct

subpopulations of extracellular vesicles that respond differentially to biotic stress, altering

growth and infectivity of a fungal pathogen. journal of extracellular vesicles 14, e70090

Kolkas, H., et al., 2022. The cell wall proteome of Marchantia polymorpha reveals specificities

compared to those of flowering plants. Front Plant Sci, 12: 765846.


Kopeck á, R., Č erný, M., 2024. Xylem sap proteome analysis provides insight into root-shoot


communication in response to flg22. Plants, 13(14): 1983.

Kourelis, J., Kaschani, F., Grosse-Holz, F.M., Homma, F., Kaiser, M., van der Hoorn, R.A.L.,

2019. A homology-guided, genome-based proteome for improved proteomics in the alloploid

Nicotiana benthamiana. BMC Genomics, 20: 722.

Kranawetter, C., Sumner, L.W., 2025. Differential root zone secretions and the role of root

border cells in rhizosphere manipulation. Phytochem Rev, 24(6): 5639-5658.

Kusumawati, L., Imin, N., Djordjevic, M.A., 2008. Characterization of the secretome of

suspension cultures of Medicago species reveals proteins important for defense and

development. J Proteome Res, 7(10): 4508-4520.

Lee, Y., Rani, H., Mallery, E.L., Szymanski, D.B., 2025. A cell fractionation and quantitative

proteomics pipeline to enable functional analyses of cotton fiber development. Plant J, 121(4):

e17246.


22


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


Liangsupree, T., Multia, E., Riekkola, M.L., 2021. Modern isolation and separation techniques

for extracellular vesicles. J Chromatogr A, 1636: 461773.

Lonsdale, A., et al., 2023. LSPpred suite: tools for leaderless secretory protein prediction in

plants. Plants, 12: 1428.

Luo, J.S., Zhang, Z., 2019. Proteomic changes in the xylem sap of Brassica napus under

cadmium stress and functional validation. BMC Plant Biol, 19(1): 280.

McLaughlin, S., Himmighofen, P., Khan, S.A., Siffert, A., Robert, CAM, Sasse, J., 2025. Root

exudation: an in-depth experimental guide. Plant Cell Environ, 48(5): 3052-3065.

Moreno-Sanguino, I., et al., 2026. Phloem proteomics to identify small open reading frame

(sorf)-encoded peptides with a putative role in the control of flowering time in Arabidopsis.

Physiol Plant, 178: e70860.

Nascimento, K.S., et al., 2018. Common features between the proteomes of floral and

extrafloral nectar from the castor plant (Ricinus communis) and the proteomes of exudates from

carnivorous plants. Front Plant Sci, 9: 549.

Neuberger, G., et al., 2003. Prediction of peroxisomal targeting signal 1 containing proteins

from amino acid sequence. J Mol Biol, 328: 581-592.

Ngcala, M.G., et al., 2020. Heat stress triggers differential protein accumulation in the

extracellular matrix of sorghum cell suspension cultures. Proteomes, 8: 29.

Pechanova, O., et al., 2010. Apoplast proteome reveals that extracellular matrix contributes to

multistress response in poplar. BMC Genomics, 11: 674.

Perez-Riverol, Y., Bai, J., Bandla, C., et al., 2022. The PRIDE database resources in 2022: A hub

for mass spectrometry-based proteomics evidences. Nucleic Acids Res, 50(D1): D543-D552.


Pernis, M., Danchenko, M., Jurcik, J., Kovac, A., Klubicov á, K., 2026. Secretome profiling of


Pinus nigra embryogenic cultures reveals a freely secreted protein fraction dominated by cell

wall- and redox-related proteins. plant cell, tissue and organ culture 166, 86

Pope, R.E., Prade, R.A., 2026. Vesicle-driven endomembrane systems in fungi. Microbiol Mol

Biol Rev, 90: e00297-24


Rej ó n, J.D., et al., 2013. Proteomics profiling reveals novel proteins and functions of the plant


stigma exudate. J Exp Bot, 64: 5695-5705.

Robinson, D.G., Ding, Y., Jiang, L, 2016. Unconventional protein secretion in plants: A critical

assessment. Protoplasma, 253: 31-43.


Rodr í guez de Lope, M.M., S á nchez-Pajares, I.R., Herranz, E., et al., 2025. A compendium of


bona fide reference markers for genuine plant extracellular vesicles and their degree of

phylogenetic conservation. J Extracell Vesicles, 14(9): e70147.

Ropitaux, M., Bernard, S., Schapman, D., et al., 2020. Root border cells and mucilage

secretions of soybean, Glycine max (merr) L.: characterization and role in interactions with the

oomycete Phytophthora parasitica. cells 9, 2215

Rose, J.K., Lee, S.J., 2010. Straying off the highway: trafficking of secreted plant proteins and

complexity in the plant cell wall proteome. Plant Physiol, 153(2): 433-6.

Routaboul, J.-M., et al., 2024. Arabidopsis hydathodes are sites of auxin accumulation and

nutrient scavenging. the plant journal 120, 857-871


23


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


Rutter, B.D., Innes, R.W., 2017. Extracellular vesicles isolated from the leaf apoplast carry

stress-response proteins. Plant Physiol, 173(1): 728-741.


S á nchez-L ó pez, C.M., et al., 2024. Phloem sap from melon plants contains extracellular


vesicles that carry active proteasomes which increase in response to aphid infestation. J

Extracell Vesicles, 13: e12517.

San Clemente, H., et al., 2022. Plant cell wall proteomes: the core of conserved protein families

and the case of non-canonical proteins. Int J Mol Sci, 23: 4273.

Savojardo, C., et al., 2018. BUSCA: an integrative web server to predict subcellular localization

of proteins. Nucleic Acids Res, 46: W459-W466.

Sigrist, C.J.A., et al., 2026. The PROSITE database for protein families, domains, and sites.

Nucleic Acids Res, 54: D451-D458.

Silva, F.A., et al., 2020. LC-MS/MS based comparative proteomics of floral nectars reveal

different mechanisms involved in floral defense of Nicotiana spp., Petunia hybrida and Datura

stramonium. J Proteomics, 213: 103618.

Small, I., et al., 2004. Predotar: A tool for rapidly screening proteomes for N-terminal targeting

sequences. Proteomics, 4: 1581-1590.

Sperschneider, J., et al., 2018. ApoplastP: prediction of effectors and plant proteins in the

apoplast using machine learning. New Phytol, 217: 1764-1778.

Staudinger, C., et al., 2022. The wheat secreted root proteome: implications for phosphorus

mobilisation and biotic interactions. J Proteomics, 252: 104450.

Strehmel, N., et al., 2017. Stress-related mitogen-activated protein kinases stimulate the

accumulation of small molecules and proteins in Arabidopsis thaliana root exudates. Front Plant

Sci, 8: 1292.

Teufel, F., et al., 2022. SignalP 6.0 predicts all five types of signal peptides using protein

language models. nature biotechnology

Thieron, H., Krassini, L., Kwon, S., et al., 2024. Practical advice for extracellular vesicle

isolation in plant-microbe interactions: concerns, considerations, and conclusions. J Extracell

Vesicles, 13: e70022.

Tiddia, F., et al., 2023. The proteome of the imbibition spillage in Cicer arietinum L. current

plant biology, 35-36, 100310

Toyofuku, M., Schild, S., Kaparakis-Liaskos, M., Eberl, L., 2023. Composition and functions of

bacterial membrane vesicles. Nat Rev Microbiol, 21: 415-430.

Wang, J., Bollier, N., Andrade Buono, R., et al., 2024. A developmentally controlled cellular

decompartmentalization process executes programmed cell death in the Arabidopsis root cap.

Plant Cell, 36(4): 941-962.

Wang, X., Chung, K.P., Lin, W., Jiang, L., 2018. Protein secretion in plants: conventional and

unconventional pathways and new techniques. J Exp Bot, 69(1): 21-37.

Wang, J., Ding, Y., Wang, J., et al., 2010. EXPO, an exocyst-positive organelle distinct from

multivesicular endosomes and autophagosomes, mediates cytosol to cell wall exocytosis in

Arabidopsis and tobacco cells. the plant cell 22, 4009-4030

Wefers, Z., Gupta, A., Ahmed, N., Zhang, X., Lundberg, E, 2026. A comprehensive benchmark

of sequence-based subcellular localization predictors for human proteins. Nat Methods, 23(7):

1458-1469.


24


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.16.766913; this version posted November 17, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.16.766913)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made
&#91;available under a CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


Welsh, J.A., Goberdhan, D.C.I., O'Driscoll, L., et al., 2024. Minimal information for studies of

extracellular vesicles (MISEV2023): from basic to advanced approaches. J Extracell Vesicles,

13(2): e12404.

Weng, Y., Sui, Z., Shan, Y., Hu, Y., Chen, Y., Zhang, L., Zhang, Y., 2016. Effective isolation of

exosomes with polyethylene glycol from cell culture supernatant for in-depth proteome

profiling. Analyst, 141(15): 4640-6.

Wu, J., Lee, D.Y., Wang, Y., Kim, S.T., Baek, S., Kim, S.G., Kang, K.Y., 2014. Protein profiles

secreted from phylloplane of rice leaves free from cytosolic proteins: application to study

rice-Magnaporthe oryzae interactions. physiological and molecular plant pathology 88, 28-35

Yang, J., Wang, Y., Fu, Y., Wang, S., Luo, Y., Cui, X., Sun, C., Ji, J., Mo, J., Wang, Y., Liu, Y.,

Hu, X., Chen, X., Jia, Y., Luo, Y., 2025. Plant extracellular vesicles: a promising bionic

medicine platform for disease treatment and drug delivery. Interdiscip Med, 3: e20240101

Yugay, Y., et al., 2023. Isolation and characterization of extracellular vesicles from Arabidopsis

thaliana cell culture and investigation of the specificities of their biogenesis. Plants, 12: 3604.

Zhao, L., et al., 2019. OutCyte: A novel tool for predicting unconventional protein secretion. Sci

Rep, 9: 19448.

Zheng, L., Liu, Q., Wu, R., et al., 2023. The alteration of proteins and metabolites in leaf

apoplast and the related gene expression associated with the adaptation of Ammopiptanthus

mongolicus to winter freezing stress. Int J Biol Macromol, 240: 124479.

Zheng, L., Li, B., Liu, Q., Richardson, J.E., Zhou, Y., Gao, F., 2026. Proteomic analysis of leaf

apoplast reveals that a jasmonate-regulated CHIA participates in the response to cold and

drought stress in jojoba. Hortic Plant J, 12(7): 1697-1718.


25


