# Source paper archive

- Source title: Expansin couples salinity sensing to plant cell growth regulation
- Author(s): zhou, k., Cosgrove, D. J.
- Source landing URL: https://www.biorxiv.org/content/10.64898/2026.11.14.766679v1
- PDF URL: https://www.biorxiv.org/content/10.64898/2026.11.14.766679v1.full.pdf
- PDF SHA-256: 0a6931b427ee05907b971b4582173cf1eade636c56d6e2ad0b459e2bf8f339be
- 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/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity.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


**0B**
# **Title: Expansin couples salinity sensing to plant cell growth regulation**


**Authors:** Ke Zhou*, Daniel J. Cosgrove


**Affiliations:** Department of Biology, The Pennsylvania State University, University Park, PA
16802, USA


5 - Corresponding author. Email: kpz5105@psu.edu.


**Abstract:**


In living plant cells, salt affects cell wall architecture directly and triggers diverse cellular
responses that, in turn, feed back on wall organization and deposition. These intertwined effects
obscure whether salt directly alters wall mechanical properties and thereby influences turgor10 driven wall enlargement, the process underlying cell growth. In this study, we studied the tensile
mechanical behavior of isolated cell wall materials at physiologically relevant salt concentrations.
NaCl induced a small, reversible deformation without significantly altering the apparent axial
tensile stiffness of hypocotyl walls. However, salt directly and reversibly inhibited acid-induced
wall creep. Similar inhibition was observed in distinct cell wall types and in living Arabidopsis
15 hypocotyls. Assays combining α-expansin extracts with cellulose filter paper further supported
direct inhibition of wall-loosening activity by salt. Molecular dynamics simulations suggested a
mechanism in which Na⁺ electrostatically perturbs the conserved active-site Asp and its associated
hydrogen-bond network. Together, these findings indicate that salt directly inhibits acid growth by
suppressing α-expansin-mediated wall creep, highlighting expansin as a physicochemical
20 transducer linking environmental salinity to growth regulation.


**Introduction:**


Plant cell growth involves turgor-driven irreversible enlargement of primary cell walls, a process
that overcomes mechanical resistance imposed mainly by the noncovalent network of cellulose
microfibrils( _1_ ). To rapidly regulate this process, plants evolved a group of conserved, non25 enzymatic proteins known as expansin( _2-4_ ). Expansins induce cell wall extension (creep) driven
by turgor, potentially by disrupting the noncovalent binding between cellulose microfibrils, in a
pH-dependent manner( _4-6_ ). This expansin-induced creep at low pH forms the basis of the
canonical “acid growth” theory, whereby hormonal and environmental cues control growth
partially by modulating this process( _7, 8_ ). Soil salinity is one of the most pervasive abiotic stresses
30 in modern agriculture, severely restricting the growth and survival of terrestrial glycophytes,
including most crop species( _9-11_ ). Early studies reported that divalent and trivalent cations( _12_ ),
as well as high concentrations of NaCl( _13_ ), inhibit acid growth in isolated cucumber hypocotyl
walls. However, whether salt acts on wall polymers, expansin proteins, or both - and how these
interactions suppress wall extensions - remain unresolved. The physiological relevance of this
35 inhibition also remains unclear.


Salt can rapidly and reversibly inhibit plant cell expansion, while prolonged salt exposure can
cause irreversible growth suppression( _14_ ). Beyond immediate reduction in turgor, salinity is
perceived through multiple pathways, including GIPC-dependent signaling at plasma
membrane( _15_ ), the cytoplasmic SOS pathway( _16_ ), the secondary responses to oxidative stress( _14_ ),
40 and perturbation of pectin structure( _17_ ). These responses can modify the apoplastic environment


1


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


and transcriptionally govern cell-wall biosynthesis and deposition( _18_ ), thereby feeding back on
wall mechanics and extension. Of particular relevance is the direct disruption of Ca &#91;2+&#93; -pectate
crosslinking by salt, which softens the pectic matrix mechanically and triggers intracellular
signaling( _17, 19, 20_ ). However, pectin softening, whether assessed in isolated pectin systems or
45 by indentation of intact walls perpendicular to the growth axis, does not necessarily reflect changes
in the tensile mechanical properties relevant to sustained wall extension along that axis( _21_ ). This
distinction is consistent with recent work showing that noncovalent interactions between cellulose
microfibrils provide the principal resistance to tensile stress( _22_ ) and that sliding between cellulose
microfibrils underlies irreversible wall extension( _23_ ). It therefore remains unclear whether salt50 induced structural changes alter wall mechanical properties in ways that directly influence wall
extension and cell growth.


Here, we used complementary mechanical assays on isolated cucumber hypocotyl and maize
coleoptile walls and on pure cellulose systems to examine how salt affects tensile mechanical
properties independently of protoplast-mediated responses and changes in turgor( _4, 6_ ). We further
55 combined molecular dynamics simulations of cucumber α-expansin CsEXPA1 with creep assays
at different pH values to explore how cations might inhibit expansin activity through interactions
with the conserved active-site Asp and its associated hydrogen-bond network.


**Results**


**NaCl induces reversible hypocotyl deformation electrostatically**


60 After freeze-thawing and flattening to remove cellular processes and turgor, apical segments from
etiolated cucumber hypocotyls were clamped on both ends, soaked in an adjustable buffer, and
subjected to a uniaxial constant load that replaces the axial component of cell turgor( _24_ ) ( **Fig. 1A-**
**1B** ). This setup mimics hypocotyl elongation, restricting the key experimental variables to the
composition of the controllable buffer. We previously utilized this system to characterize
65 expansin’s cell wall creep activity( _4, 5, 12_ ). In neutral buffer (20 mM HEPES buffer, pH 7), the
hypocotyl underwent minimal, time-dependent, and irreversible elongation (creep, red in **Fig. 1C-**
**1D** ), which reflects the gradual, permanent yielding of the cell wall polymers as they slide past
one another( _22, 25_ ). Substituting the neutral buffer with an acidic buffer (50 mM MES, pH 5.2), a
pH commonly found in the apoplast of growing primary cell walls, mimicked the typical apoplastic
70 acidification that occurs in growing tissues( _26_ ), eliciting the characteristic acid-induced creep
process: an abrupt acceleration of wall extension followed by deceleration to a new, stable, yet
significantly higher rate( _4_ ) (blue in **Fig. 1C-1D** ). In contrast, adding NaCl to the neutral buffer
induced prompt but transient elongation without sustained effects ( **Fig. 1C-1E** ). This elongation
is concentration-dependent ( **Fig. 1F** ), and fully reversible, as washing out the salt caused shrinkage
75 that offset the initial elongation ( **Fig. 1G, 1H, and Fig. S1** ). Adding 100 mM sodium acetate
(NaAc, adjusted to pH 7) induced an identical, reversible deformation, suggesting that the
observed deformation may result from Na &#91;+&#93; rather than Cl &#91;- &#93; ( **Fig. S2** ). These results indicate that
environmental Na &#91;+&#93; does not cause the cell wall to undergo creep; instead, it results in an immediate
and reversible wall deformation, distinct from the sustained and irreversible acid-induced
80 extension.


Next, we investigated the effects of other monovalent (LiCl, KCl, CsCl, and NH 4 Cl) and divalent
(MgCl 2, CaCl 2, and BaCl 2 ) chloride salts. All tested monovalent cations induced similar reversible
deformations, except for LiCl, which caused a slightly weaker effect ( **Fig. 1I** ). Divalent cations


2


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


elicited a more pronounced but largely irreversible deformation ( **Fig. 1I** ), which could be partially
85 reversed by the chelating reagent EDTA (5 mM) ( **Fig. 1J and Fig. S3** ). This dependence on ionic
valence, together with the partial reversal by EDTA, suggests that electrostatic interactions are
largely responsible for the deformation.


Carboxyl groups, predominantly provided by the de-esterified polygalacturonic acid backbone of
pectin, and supplemented by structural proteins( _27_ ), constitute the major electrostatic binding sites
90 for metal ions in the primary cell wall. Among these electrostatic interactions, the Ca &#91;2+&#93; -bridged
pectate “egg-box” structure is a particularly specialized form that is a well-known target of salt( _17,_
_19_ ). However, our data show that Mg &#91;2+&#93;, which has a minimal capacity to form this specific, stable
configuration ( _28_ ), caused an similar deformation that is nonetheless partly reversible ( **Fig. 1I** ).
These observations therefore suggest that salt-induced hypocotyl deformation is dominated by
95 generic electrostatic interactions with anionic sites in the cell wall, instead of arising from the
specific Ca &#91;2+&#93; -pectate “egg-box” crosslinking interaction( _29_ ).


**NaCl insignificantly affects tensile mechanical properties of hypocotyl walls**


Under constant load, salt induced reversible hypocotyl deformation but did not promote sustained,
irreversible extension (creep). To complement these time-dependent measurements, we performed
100 constant-rate tensile tests on freeze-thawed hypocotyls using the same initial gauge length (5 mm)
as in the creep assays. Specimens were stretched at a constant crosshead speed of 3 mm/min while
force and extension were recorded simultaneously. The resulting force-strain curves characterize
the tensile mechanical response of the specimens by showing how force varies with strain. These
curves were similar in the presence and absence of 100 mM NaCl in neutral HEPES buffer ( **Fig.**
105 **S4A)** . Their local slopes, which provide estimates of apparent tensile stiffness at a given strain,
were also comparable over the common strain range ( **Fig. S4B** ). Together, these results suggest
that NaCl may not substantially alter the apparent axial tensile stiffness of hypocotyl walls under
the conditions tested.


**NaCl reversibly inhibits acid-induced hypocotyl elongation** _**in vitro**_


110 We then applied NaCl (10-100mM), spanning mild to severe soil salinity in nature( _9_ ), to hypocotyl
segments undergoing sustained acid-induced elongation. Adding salt caused a pronounced,
concentration-dependent suppression of the elongation rate ( **Fig. 2A-2B** ). The inhibition was fully
reversed upon washing out NaCl ( **Fig. 2A**, **2C** ), indicating that NaCl acts as a potent yet reversible
inhibitor of acid-induced elongation. We tested additional mono- and divalent chloride salts, which
115 produced responses that closely paralleled their effects on hypocotyl deformation in neutral buffer:
all monovalent cations (35 mM) induced almost identical and reversible inhibition, with LiCl again
an exception, showing a slightly attenuated effect ( **Fig. 2D** ); divalent salts (MgCl 2, CaCl 2, and
BaCl 2 ) at equimolar chloride concentrations (17.5 mM) caused a more potent inhibition that was
largely irreversible; MgCl 2 caused an inhibition comparable in potency to that of other divalent
120 cations, yet this effect was partially reversible ( **Fig. 2E** ); and EDTA partially reversed the
irreversible inhibition caused by Ca &#91;2+&#93; ( **Fig. 2F-2G** ). Likewise, these results are consistent with a
predominant contribution of electrostatic interactions, with little evidence for a significant role of
Ca &#91;2+&#93; -pectate crosslinking for the inhibition.


**NaCl may directly suppress cellulose microfibril-loosening by α-expansin**


3


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


125 Acid-induced elongation in cucumber hypocotyls is mediated by the pH-dependent action of
expansins which may be extracted from the wall by NaCl( _30_ ). Notably, the first-isolated expansins,
CsEXPA1 and CsEXPA2, were characterized from this NaCl-eluted protein extract( _4, 5_ ). In this
study, we heat-inactivated wall specimens and then reconstituted acid-induced elongation by
addition of NaCl-eluted wall proteins from cucumber hypocotyls ( **Fig. 3A** ). The reconstituted acid130 induced elongation is amplified by the removal of Na &#91;+&#93; and NH 4 &#91;+&#93; present in the extract ( see
"Materials and Methods")( _4, 5_ ). NaCl (added at 90 min) potently inhibited this reconstituted
elongation, and this inhibition was reversed upon salt washout (at 120 min) ( **Fig. 3A-3C** ). These
results show that NaCl reversibly suppresses the reconstituted wall-loosening response, consistent
with a direct, reversible inhibition of α-expansin-mediated loosening rather than irreversible
135 damage or loss.


Given the difficulties of purifying EXPA proteins and the compositional complexity of native plant
cell walls, we further minimized the system by replacing the cucumber hypocotyl with pure
cellulose filter paper strips. Under constant tensile stress in an aqueous environment, the
mechanical integrity of this material depends on noncovalent binding between cellulose fibers,
140 analogous to those that stabilize the native cell wall ( _31_ ). Previous work has established that
exogenous α-expansin weakens these interactions, leading to unstable extension and mechanical
failure of the filter paper( _5_ ). With a 12-gram load, hydrated filter paper strips ( 2.1 mm x 5 mm,
Whatman 3# pure cellulose square filter paper) in acidic buffer (with and without 100 mM NaCl,
but without added protein) exhibited negligible creep, without failing within the maximum
145 observation period of 200 minutes) ( **Fig. 3D, S5A-S5B** ). The addition of N aCl-eluted cell wall
proteins from cucumber hypocotyls, however, markedly accelerated filter paper extension, driving
the strips into accelerating creep followed by failure. Additional 100 mM NaCl substantially
suppressed the accelerated creep ( **Fig. 3D, S5B** ), prolonging the time-to-failure ( **Fig. 3E, S6** ).
These findings in a pure-cellulose system therefore indicate that NaCl directly inhibits α-expansin150 mediated disruption of load-bearing noncovalent interactions between cellulose microfibrils.


**The inhibition by salt is ubiquitous** _**in vitro**_ **and** _**in planta**_


Grass primary walls differ from dicot walls in matrix composition( _32_ ). In grasses the coleoptile
has become the standard system for studying rapid growth and its hormonal regulation( _33-35_ ). It
is a fast-elongating, hollow sheath surrounding the shoot apex and protecting the plumule; acid155 induced elongation of coleoptile walls _in vitro_ requires α-expansin activity( _6_ ). In the following
experiments we substituted freeze-thawed maize coleoptiles for hypocotyls in our elongation assay
and used this system to compare the effect of NaCl on freeze-thawed maize coleoptile walls
undergoing acid-induced elongation. As shown in **Fig. 4A** and **4B**, salt suppressed acid-induced
growth in coleoptile walls, suggested this NaCl effect may be a conserved feature of plant cell
160 walls.


In living plant tissues, cell growth is governed by turgor-driven wall extension. Beyond reducing
turgor, salt acts on walls via multiple ways, including direct disruption of Ca²⁺-pectate
crosslinks( _17, 19, 20_ ), transcriptional remodeling wall biosynthesis and deposition( _18_ ), and
alteration of apoplastic environment( _36_ ). To decouple the direct impact of salt on cell wall
165 extension from these multifaceted and interacting processes, we incubated live Arabidopsis
hypocotyls in 50 mM MES buffer (pH 5.2) to stabilize apoplastic pH. The observation period was
limited to 30 minutes to minimize transcriptional effects. Application of 20 mM NaCl markedly
but reversibly suppressed the elongation ( **Fig. 4C** ), whereas 40 mM mannitol, which imposes a


4


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


comparable osmotic stress, had negligible effect ( **Fig. 4C** ). These _in vivo_ results indicate that salt
170 triggers rapid and reversible growth arrest beyond turgor reduction, consistent with salt-induced
inhibition on α-expansin-mediated wall creep.


**A conserved H102-T11-D104 hydrogen-bond network in CsEXPA1**


To investigate how salt interferes with α-expansin-induced wall creep, we performed molecular
dynamics (MD) simulations using CsEXPA1, an ortholog of Arabidopsis EXPA8 and a major
175 active expansin in NaCl-eluted cell-wall protein extracts from cucumber hypocotyls _(4, 5)_ .
Modeling by AlphaFold (AF-Q39625-F1-v6( _37, 38_ )) shows a canonical expansin structure
consisting of an N-terminal domain-1 with structural features resembling Glycosyl Hydrolase
Family-45 (GH45) and a C-terminal domain-2 with features of carbohydrate-binding module-63
(CBM63)( _39-43_ )( **Fig. 5A** ). In GH45 enzymes, the conserved Asp of HFD motif acts as the
180 catalytic proton donor, protonating the glycosidic oxygen by its side chain carboxyl group to
promote β-1,4-glycosidic bond cleavage, and may subsequently be re-protonated through a
conserved hydrogen-bond network involving a His-Thr-Asp proton relay( _44_ )( **Fig. 5B** ). The
corresponding Asp is critical for wall-creep activity in both bacterial expansins( _41_ ) and plant
EXPAs( _45_ ). Notably, although expansins lack hydrolytic activity and the complete proton185 acceptor/proton-relay machinery of GH45 enzymes( _2_ ), canonical EXPAs retain the conserved Thr
and HFD motif corresponding to T11 and H102-F103-D104 in CsEXPA1( _45, 46_ )( **Fig. 5C** and **Fig.**
**S7** ).


Given the proton-donating role of the homologous Asp in GH45 enzymes, we first sought to
determine the protonation state of the D104 side-chain carboxyl group. We therefore characterized
190 the pH-dependent protonation behavior of titratable residues in CsEXPA1 by constant-pH replicaexchange molecular dynamics (pH-REMD)( _47_ ). Conformations were sampled by MD in explicit
solvent, and discrete protonation-state changes were attempted at fixed intervals using Monte
Carlo moves evaluated with a generalized Born (GB) implicit-solvent model( _47_ ). Simulations
spanned pH2.8-7.0 in 0.6-unit increments. The modified protocol reproduced experimentally
195 determined p _K_ ₐ values for titratable residues in hen egg-white lysozyme (HEWL)( _47, 48_ )( **Table.**
**S1** ). We next analyzed the titration behavior of Asp, Glu and His residues in domain-1 of
CsEXPA1 ( **Table. S2** ). D104 and the neighboring H102 exhibited unusually smooth titration
curves that deviated markedly from Henderson-Hasselbalch behavior ( **Fig. 5D** ). T11, H102, and
D104 formed a dynamic hydrogen-bond network through their hydroxyl, imidazole, and carboxyl
200 groups, respectively ( **Fig. 5E** ). The T11A substitution, which removes hydroxyl group from the
Thr, shifted the protonation equilibria of both H102 and D104 towards their neutral states ( **Fig.**
**S8A-S8B** ), consistent with a role for T11 in stabilizing their charged states within the hydrophobic
groove ( **Fig. 5A** ). Most H102-D104 hydrogen bonds were charge-assisted, forming between
positively charged H102 (HIP⁺) and negatively charged D104 (ASP &#91;-&#93; ) ( **Fig. 5F** ). Consistent with
205 these strong interactions, fixing the protonation state of either residue altered the titration behavior
of the other, indicating strong protonation coupling ( **Fig. S8C-S8D** ). Together, these simulations
suggest that, as in GH45 enzymes, the active-site Asp D104 in domain-1 of CsEXPA1 lies within
an H102-T11-D104 hydrogen-bond network. This network may modulate the protonation state of
the D104 side chain, which could be important for CsEXPA1 activity.


210 **Potential mechanisms for salt-induced inhibition**


5


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


Including 100 mM NaCl in the explicit solvent of our pH-REMD simulations markedly altered the
titration behavior of H102 and D104, bringing both curves into closer agreement with the standard
Henderson-Hasselbalch relation ( **Fig. 5G-5H** and **Table. S3** ), whereas the protonation behavior of
most other Asp, Glu and His residues remained largely unchanged ( **Fig. S9** and **Table. S3** ). Most
215 H102-D104 hydrogen bonds were charge-assisted contacts between protonated H102 (HIP⁺) and
deprotonated D104 (ASP &#91;-&#93; ) ( **Fig. 5F** ). NaCl reduced salt-bridge occupancy within this HIP⁺-ASP &#91;- &#93;

state ( **Fig. 5I** ), consistent with perturbation of electrostatic coupling between the two residues by
the added ions( _49_ ). In parallel, overall T11-D104 hydrogen-bond occupancy ( **Fig. S10** ). Together,
these changes suggest that NaCl may reorganize the H102-T11-D104 hydrogen-bond network, in
220 part through electrostatic screening of H102-D104 interactions, potentially influencing CsEXPA1
activity.


In our pH-REMD simulations, protonation-state changes were evaluated using a GB model that
excluded explicit ions( _47_ ) and therefore may not fully capture the coupling between Na⁺
coordination, D104 protonation and H102-T11-D104 hydrogen-bond network. We therefore
225 performed conventional MD simulations in explicit solvent containing 100 mM NaCl, with D104
fixed as ASP &#91;-&#93; and H102 as HIP⁺, HID⁰ or HIE⁰. Na⁺ transiently coordinated D104 when H102 was
neutral, with a preference for the HIE⁰ state, whereas protonated H102 (HIP⁺) strongly suppressed
this coordination ( **Fig. 5J-5K** and **Fig. S11A** ). A role for cation coordination in CsEXPA1
inhibition was further supported by our findings with divalent cations. In simulations of the HIE⁰230 ASP &#91;- &#93; state in explicit solvent containing 50 mM CaCl₂, Ca²⁺ coordination to the D104 carboxylate
persisted after initial contact for the remainder of the trajectories ( **Fig. S11B** ). This sustained
coordination is consistent with the stronger, less reversible inhibition of hypocotyl elongation by
divalent cations ( **Fig. 2E** ) and the partial reversal of CaCl₂ inhibition by EDTA chelation ( **Fig. 2D**
**2G** ). Together, these findings suggest that cation coordination to the D104 side-chain carboxylate,
235 which could alter the local electrostatic environment or perturb D104’s interactions with cellulose
or neighboring residues, may contribute to salt inhibition of CsEXPA1.


Na⁺ coordination to the D104 side-chain carboxylate was strongly suppressed when H102 was
positively charged (HIP⁺) ( **Fig. 5J-5K** ). We therefore predicted that the higher fraction of H102 in
the HIP⁺ state at lower pH ( **Fig. 5D** ) would electrostatically disfavor Na⁺ coordination and thereby
240 reduce Na⁺-mediated inhibition of CsEXPA1. Consistent with this prediction, inhibition of acidinduced hypocotyl elongation by 100 mM NaCl weakened progressively as pH decreased ( **Fig.**
**5M** ).


Together, these results support a model in which Na &#91;+&#93; reversibly perturbs the H102-T11-D104
hydrogen-bond network and transiently coordinates with the active-site Asp D104, thereby inhibits
245 its wall-loosening activity.


**Discussion:**


Cell wall-localized expansins can be activated by apoplastic acidification, facilitating wall creep
and turgor-driven cell expansion( _4_ ). This pH-dependent process underpins acid-growth theory and
provides a conceptual framework for understanding how hormonal and environmental signals
250 regulate plant cell growth( _3, 25, 32_ ). This mechanism contributes to both diffuse growth _(4)_ and
tip growth( _45_ ). Our findings suggest that expansins, the central effectors of ‘acid growth’, can also
function as physicochemical sensors of salinity. Salt-induced inhibition of their wall-loosening
activity directly translates salt exposure into rapid, reversible growth inhibition. This direct


6


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


physicochemical mechanism could operate independently of canonical signaling pathways, such
255 as FERONIA-mediated cell wall integrity signaling and GIPC-dependent salt sensing that elicit
Ca &#91;2+&#93; signals( _15, 17_ ). Moreover, it also offers a mechanistic explanation for long-standing
observations of the immediate and reversible inhibition of plant cell growth by salinity( _50, 51_ ).


Ca²⁺-pectate crosslinking is a specific electrostatic interaction: Ca &#91;2+&#93; ions bridge de-esterified
polygalacturonic acid chains to form the characteristic “egg-box” configuration( _52, 53_ ). High
260 salinity can disrupt these Ca²⁺-mediated bridges, thereby softening the pectic matrix( _17, 54_ ). In
our study, however, NaCl neither induced sustained wall creep nor significantly altered the
apparent axial tensile stiffness of hypocotyl walls. Instead, our results suggest that NaCl suppresses
wall expansion through direct, rapid, and reversible inhibition of expansin activity. These
observations are consistent with our previous findings that softening of pectic wall components
265 and growth-associated loosening are not necessarily coupled processes ( _21, 55_ ). We therefore
propose that salinity-induced changes in pectate crosslinking may influence cell growth primarily
through cell wall integrity signaling and downstream responses( _17_ ), rather than through direct
mechanical effects on wall expansion.


Although salinity imposes fundamental physicochemical constraints on plant cell growth,
270 halophytes have adapted to chronically saline environments( _56_ ), and monocotyledonous
seagrasses even thrive in full-strength seawater( _57_ ). Our findings raise the question of how plants
circumvent this physicochemical constraint. Possible strategies include increasing EXPA
abundance, consistent with reports that EXPA overexpression enhances salt tolerance( _18_ ); further
acidifying the apoplast to alleviate inhibition; or deploying salt-adapted EXPA variants or other
275 expansin family members. The unexpectedly broad functional diversity recently revealed within
the expansin superfamily( _45_ ) suggests that such specialization is plausible. Defining these
adaptive mechanisms could inform the rational breeding of crops with enhanced salt tolerance.


**4B** **References and Notes:**


1. E. Coen, D. J. Cosgrove, The mechanics of plant morphogenesis. _Science_ **379**, eade8055
280 (2023).
2. D. J. Cosgrove, Non-enzymatic action of expansins. _J Biol Chem_ **295**, 6782 (2020).
3. D. J. Cosgrove, Plant Cell Wall Loosening by Expansins. _Annu Rev Cell Dev Biol_ **40**,
329–352 (2024).
4. S. McQueen-Mason, D. M. Durachko, D. J. Cosgrove, A. New Collective, Two
285 endogenous proteins that induce cell wall extension in plants. _Plant Cell_ **4**, 1425–1433
(1992).
5. S. McQueen-Mason, D. J. Cosgrove, Disruption of hydrogen bonding between plant cell
wall polymers by proteins that induce wall extension. _Proc Natl Acad Sci U S A_ **91**,
6574–6578 (1994).
290 6. D. J. Cosgrove, Z. C. Li, Role of Expansin in Cell Enlargement of Oat Coleoptiles
(Analysis of Developmental Gradients and Photocontrol). _Plant Physiol_ **103**, 1321–1328
(1993).
7. G. Arsuffi, S. A. Braybrook, Acid growth: an ongoing trip. _J Exp Bot_ **69**, 137–146
(2018).
295 8. H. Y. Zeng _et al._, Origin and evolution of auxin-mediated acid growth. _Proc Natl Acad_
_Sci U S A_ **121**, e2412493121 (2024).
9. A. Läuchli, U. Lüttge, _Salinity: Environment - Plants - Molecules_ . (2002).


7


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


10. R. Munns, M. Tester, Mechanisms of salinity tolerance. _Annu Rev Plant Biol_ **59**, 651–
681 (2008).
300 11. Y. Yang, Y. Guo, Elucidating the molecular mechanisms mediating plant salt-stress
responses. _New Phytol_ **217**, 523–539 (2018).
12. D. J. Cosgrove, Characterization of long-term extension of isolated cell walls from
growing cucumber hypocotyls. _Planta_ **177**, 121–130 (1989).
13. M. A. Perini, I. N. Sin, G. A. Martinez, P. M. Civello, Measurement of expansin activity
305 and plant cell wall creep by using a commercial texture analyzer. _Electron J Biotechn_ **26**,
12–19 (2017).
14. E. van Zelm, Y. Zhang, C. Testerink, Salt Tolerance Mechanisms of Plants. _Annu Rev_
_Plant Biol_ **71**, 403–433 (2020).
15. Z. Jiang _et al._, Plant cell-surface GIPC sphingolipids sense salt to trigger Ca(2+) influx.
310 _Nature_ **572**, 341–346 (2019).
16. J. K. Zhu, Abiotic Stress Signaling and Responses in Plants. _Cell_ **167**, 313–324 (2016).
17. W. Feng _et al._, The FERONIA Receptor Kinase Maintains Cell-Wall Integrity during
Salt Stress through Ca(2+) Signaling. _Curr Biol_ **28**, 666–675 e665 (2018).
18. L. Colin _et al._, The cell biology of primary cell walls during salt stress. _Plant Cell_ **35**,
315 201–217 (2023).
19. B. Manunza, S. Deiana, M. Pintore, C. Gessa, Interaction of Ca2+ and Na+ ions with
polygalacturonate chains: a molecular dynamics study. _Glycoconj J_ **15**, 297–300 (1998).
20. L. Beauzamy, J. Derr, A. Boudaoud, Quantifying hydrostatic pressure in plant cells by
using indentation with an atomic force microscope. _Biophys J_ **108**, 2448–2456 (2015).
320 21. X. Wang, L. Wilson, D. J. Cosgrove, Pectin methylesterase selectively softens the onion
epidermal wall yet reduces acid-induced creep. _J Exp Bot_ **71**, 2629–2640 (2020).
22. Y. Zhang _et al._, Molecular insights into the complex mechanics of plant epidermal cell
walls. _Science_ **372**, 706–711 (2021).
23. C. Li, J. Yu, K. J. Hsia, D. J. Cosgrove, S. Zhang, Collective microfibril sliding underlies
325 plant cell wall creep. (2026).
24. D. J. Cosgrove, Loosening of plant cell walls by expansins. _Nature_ **407**, 321–326 (2000).
25. D. J. Cosgrove, Building an extensible cell wall. _Plant Physiol_ **189**, 1246–1277 (2022).
26. J. Wang _et al._, The apoplastic pH is a key determinant in the hypocotyl growth response
to auxin dosage and light. _Nat Plants_ **11**, 279–294 (2025).
330 27. M. Ellis, J. Egelund, C. J. Schultz, A. Bacic, Arabinogalactan-proteins: key regulators at
the cell surface? _Plant Physiol_ **153**, 403–419 (2010).
28. F. Munarin, M. C. Tanzi, P. Petrini, Advances in biomedical applications of pectin gels.
_Int J Biol Macromol_ **51**, 681–689 (2012).
29. H. Temple _et al._, Golgi-localized putative S-adenosyl methionine transporters required
335 for plant cell wall polysaccharide methylation. _Nat Plants_ **8**, 656–669 (2022).
30. D. J. Cosgrove, Characterization of Long-Term Extension of Isolated Cell Walls from
Growing Cucumber Hypocotyls. _Planta_ **177**, 121–130 (1989).
31. K. S. Salem, N. Barrios, H. Jameel, L. Pal, L. Lucia, Computational and experimental
insights into the molecular architecture of water-cellulose networks. _Matter-Us_ **6**, 1366–
340 1381 (2023).
32. D. J. Cosgrove, Structure and growth of plant cell walls. _Nat Rev Mol Cell Biol_, (2023).
33. M. J. Vesper, Use of a pH-response curve for growth to predict apparent wall pH in
elongating segments of maize coleoptiles and sunflower hypocotyls. _Planta_ **166**, 96–104
(1985).


8


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


345 34. R. M. Napier, M. A. Venis, M. A. Bolton, L. I. Richardson, G. W. Butcher, Preparation
and characterisation of monoclonal and polyclonal antibodies to maize membrane auxinbinding protein. _Planta_ **176**, 519–526 (1988).
35. A. M. Jones, M. A. Venis, Photoaffinity labeling of indole-3-acetic acid-binding proteins
in maize. _Proc Natl Acad Sci U S A_ **86**, 6153–6156 (1989).
350 36. C. M. Geilfus, R. Tenhaken, S. C. Carpentier, Transient alkalinization of the leaf apoplast
stiffens the cell wall during onset of chloride salinity in corn leaves. _J Biol Chem_ **292**,
18800–18813 (2017).
37. J. Fleming _et al._, AlphaFold Protein Structure Database and 3D-Beacons: New Data and
Capabilities. _J Mol Biol_ **437**, 168967 (2025).
355 38. J. Jumper _et al._, Highly accurate protein structure prediction with AlphaFold. _Nature_ **596**,
583–589 (2021).
39. V. S. Bharadwaj, B. C. Knott, J. Stahlberg, G. T. Beckham, M. F. Crowley, The
hydrolysis mechanism of a GH45 cellulase and its potential relation to lytic
transglycosylase and expansin function. _J Biol Chem_ **295**, 4477–4487 (2020).
360 40. F. Kerff _et al._, Crystal structure and activity of Bacillus subtilis YoaJ (EXLX1), a
bacterial expansin that promotes root colonization. _Proc Natl Acad Sci U S A_ **105**,
16876–16881 (2008).
41. N. Georgelis, A. Tabuchi, N. Nikolaidis, D. J. Cosgrove, Structure-function analysis of
the bacterial expansin EXLX1. _J Biol Chem_ **286**, 16814–16823 (2011).
365 42. N. K. Hepler, D. J. Cosgrove, Directed in vitro evolution of bacterial expansin BsEXLX1
for higher cellulose binding and its consequences for plant cell wall-loosening activities.
_FEBS Lett_ **593**, 2545–2555 (2019).
43. R. L. Silveira, M. S. Skaf, Molecular dynamics of the Bacillus subtilis expansin EXLX1:
interaction with substrates and structural basis of the lack of activity of mutants. _Phys_
370 _Chem Chem Phys_ **18**, 3510–3521 (2016).
44. A. Nakamura _et al._, "Newton's cradle" proton relay with amide-imidic acid
tautomerization in inverting cellulase visualized by neutron crystallography. _Sci Adv_ **1**,
e1500263 (2015).
45. K. Zhou, N. K. Hepler, M. Jia, D. J. Cosgrove, Functional divergence of plant expansins
375 as revealed by genetic complementation of root-hair tip growth in Arabidopsis thaliana.
_Proc Natl Acad Sci U S A_ **123**, e2623483123 (2026).
46. L. Okmane _et al._, Glucomannan and beta-glucan degradation by Mytilus edulis Cel45A:
Crystal structure and activity comparison with GH45 subfamily A, B and C. _Carbohydr_
_Polym_ **277**, 118771 (2022).
380 47. J. M. Swails, D. M. York, A. E. Roitberg, Constant pH Replica Exchange Molecular
Dynamics in Explicit Solvent Using Discrete Protonation States: Implementation,
Testing, and Validation. _J Chem Theory Comput_ **10**, 1341–1352 (2014).
48. H. Webb _et al._, Remeasuring HEWL pK(a) values by NMR spectroscopy: methods,
analysis, accuracy, and implications for theoretical pK(a) calculations. _Proteins_ **79**, 685–
385 702 (2011).
49. H. X. Zhou, X. Pang, Electrostatic Interactions in Protein Structure, Folding, Binding,
and Condensation. _Chem Rev_ **118**, 1691–1741 (2018).
50. P. M. Neumann, Rapid and reversible modifications of extension capacity of cell walls in
elongating maize leaf tissues responding to root addition and removal of NaCl. _Plant,_
390 _Cell &amp; Environment_ **16**, 1107–1114 (1993).
51. W. Fricke _et al._, The short-term growth response to salt of the developing barley leaf. _J_
_Exp Bot_ **57**, 1079–1095 (2006).


9


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


52. I. Braccini, S. Perez, Molecular basis of Ca2+-induced gelation in alginates and pectins:
the egg-box model revisited. _Biomacromolecules_ **2**, 1089–1096 (2001).
395 53. G. T. Grant, E. R. Morris, D. A. Rees, P. J. C. Smith, D. Thom, Biological interactions
between polysaccharides and divalent cations: The egg‐box model. _FEBS Letters_ **32**,
195–198 (1973).
54. T. A. Bell, J. L. Etchells, Influence of Salt (NaCl) on Pectinolytic Softening of
Cucumbersa. _Journal of Food Science_ **26**, 84–90 (1961).
400 55. T. Zhang, H. Tang, D. Vavylonis, D. J. Cosgrove, Disentangling loosening from
softening: insights into primary cell wall structure. _Plant J_ **100**, 1101–1117 (2019).
56. T. Taji _et al._, Comparative genomics in salt tolerance between Arabidopsis and
aRabidopsis-related halophyte salt cress using Arabidopsis microarray. _Plant Physiol_
**135**, 1697–1709 (2004).
405 57. X. Ma _et al._, Seagrass genomes reveal ancient polyploidy and adaptations to the marine
environment. _Nat Plants_ **10**, 240–255 (2024).
58. D. A. Case _et al._, AmberTools. _J Chem Inf Model_ **63**, 6183–6191 (2023).
59. D. A. Case _et al._, Recent Developments in Amber Biomolecular Simulations. _J Chem Inf_
_Model_ **65**, 7835–7843 (2025).


410 **Acknowledgments:** We thank Jingyi Yu and Chinghai Li for initial discussions on early
versions of this work, and C. Denise Okafor for helpful technical feedback on the molecular
dynamics methods and analysis.


**Funding:** United States Department of Energy DE-FG02-84ER13179 (D.J.C).


**Author contributions:** Conceptualization: KZ


415 Methodology: KZ


Investigation: KZ


Visualization: KZ


Funding acquisition: DJC


Project administration: KZ


420 Supervision: KZ, DJC


Writing – original draft: KZ


Writing – review &amp; editing: KZ, DJC


**Competing interests:** Authors declare that they have no competing interests.


**Data, code, and materials availability:** All data are available in the main text or the
425 supplementary materials.


10


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)

**1B**



**1B** **Fig. 1.** **Acid- and NaCl-induced hypocotyl elongation** _**in vitro**_ **.** ( **A** ) Schematic of the _in vitro_
hypocotyl elongation system. Actively elongating apical segments (0.5 cm) from three-day-old
cucumber seedling hypocotyls were freeze-thawed and flattened to remove cell sap, then clamped
430 at both ends and subjected to a 20 g uniaxial load. At cellular level, the axial component of turgor
(σ-axial), which normally provides the driving force for elongation, is thus replaced by the applied
load (f). ( **B** ) Recording of hypocotyl length every 30 s for determination of elongation and
calculation of elongation rate. ( **C** ) Time course of hypocotyl elongation and ( **D** ) elongation rate in
response to acid or NaCl. At 20 min (arrows), neutral buffer was replaced with either acidic buffer
435 or neutral buffer containing 100 mM NaCl. Untreated, red; acid buffer, blue; 100 mM NaCl, green.


11


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


( **E** ) Total elongation during 60 min after treatment; means ± SD; n= 21, 21, 21. ( **F** ) Total elongation
during 10 min in response to 50 mM, 100 mM, and 150 mM NaCl treatment in neutral buffer;
means ± SD; n=14, 14, 12. ( **G** ) Representative elongation-rate trace for sequential addition and
washout of 100 mM NaCl in neutral buffer. ( **H** ) Hypocotyl elongation after addition of 100 mM
440 NaCl and shrinkage after washout; means ± SD; n=14. ( **I** ) Hypocotyl elongation during the 10 min
after salt addition and shrinkage during the 10 min after washout (w/o) in response to 100 mM
monovalent chloride salts and 50 mM divalent chloride salts; n = 12 (LiCl), 14 (NaCl), 12 (KCl),
13 (CsCl), 15 (NH₄Cl), 16 (MgCl₂), 16 (CaCl₂), and 16 (BaCl₂); means ± SD; different letters
indicate statistically significant differences among treatments (P &lt; 0.05, Student’s t-test). ( **J** ) Time
445 course of elongation rate during sequential application of 50 mM CaCl 2, washout, and 5 mM
EDTA (arrows). The right y-axes of **C**, **E**, and **F** indicate the corresponding strain (%). Asterisks
indicate significant differences (t-test; ***, p &lt; 0.001; ****: p &lt; 0.0001).


12


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


450
**Fig. 2. Inhibition of sustained acid-induced hypocotyl elongation by mono- and divalent**
**chloride salts.** ( **A** ) Representative elongation-rate trace for a hypocotyl segment during sustained
acid-induced elongation in pH5.2 buffer, 35 mM NaCl treatment, and subsequent washout. Gray
shading, time windows for quantification of inhibition and recovery in Figure B and C. ( **B** )
455 Relative elongation rate during the inhibition phase at the indicated NaCl concentrations (10-100
mM), normalized to the steady acid-induced rate before salt addition: means ± SD; n = 8 (mock),
21 (10 mM), 7 (20 mM), 14 (35 mM), 7 (50 mM), and 7 (100 mM). ( **C** ) Relative elongation rate
during the recovery phase after washout of 35 mM NaCl compared with mock-treated controls;
means ± SD; n = 7 (mock), 11 (NaCl); t-test; ns, not significant, P&gt;0.05). ( **D, E** ) Relative
460 elongation rate during salt treatment (35 mM monovalent chloride salts, or 17.5 mM divalent
chloride salts) and after washout (w/o), normalized to the steady acid-induced rate before salt
addition (dashed line, 100%). Data are means ± SD; n = 12 (mock), 12 (NaCl), 14 (LiCl), 11 (KCl),
7 (CsCl), 8 (NH₄Cl), 12 (MgCl₂), 7 (CaCl₂), and 7 (BaCl₂); means ± SD; t-test; ****: p &lt; 0.0001;
ns, not significant, P&gt;0.05; different letters indicate statistically significant differences among


13


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


465 treatments (P &lt; 0.05, Student’s _t_ -test). ( **F** ) Representative elongation-rate trace for a hypocotyl
segment during sequential treatment with 17.5 mM CaCl₂, washout, and 5 mM EDTA; gray
shading marks inhibition, recovery, and chelation periods used for quantification. ( **G** ) Relative
elongation rates during the inhibition, recovery, and chelation phases in (F), normalized to the presalt acid-induced rate (dashed line, 100%); means ± SD; n=8.


470


14


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


475 **Fig. 3. Salt-suppressed crude** **α-expansin-induced creep of hypocotyl segments and pure**
**cellulose filter paper strips.** ( **A** ) Representative elongation-rate trace of reconstituted acidinduced elongation and its salt-induced inhibition in a heat-inactivated hypocotyl segment. The
neutral buffer was replaced with acidic buffer at 20 min, followed by addition of crude α-expansin
extract in acidic buffer at 40 min. The extract was then washed out with acidic buffer at 60 min,
480 100 mM NaCl was added at 90 min, and washed out at 120 min (arrows). Gray shading indicates
the inhibition and recovery windows used for quantification. ( **B**, **C** ) Relative elongation rate during
the (B) inhibition and (C) recovery phases in mock and NaCl-treated samples, normalized to the
crude α-expansin-induced rate before NaCl addition. Means ± SD; n = 10 (mock), 13 (NaCl); ttest; ****, P &lt; 0.0001; ns, not significant, P&gt;0.05. ( **D** ) Representative elongation-rate traces during
485 creep of cellulosic filter paper strips. Treatments of crude α-expansin in the absence (red) or
presence (green) of 100 mM NaCl, and acidic buffer in the absence (blue) or presence (purple) of
100 mM NaCl. Strips were pre-equilibrated in acidic buffer containing the corresponding NaCl
concentration, and the arrow at 30 min indicates the time point of treatment. ( **E** ) Time from
treatments to strip failure in the absence (n = 12) or presence (n = 10) of 100 mM NaCl; means ±
490 SD; t-test; ****: p &lt; 0.0001.


15


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


**Fig. 4. Salt-suppressed acid-induced elongation of coleoptile and live Arabidopsis hypocotyls.**
495 **(A** ) Representative elongation-rate trace of freeze-thawed maize coleoptile during sustained acidinduced elongation in pH5.2 buffer, 35 mM NaCl treatment, and subsequent washout. Gray
shading, time windows for quantification of inhibition and recovery in Figure B. ( **B** ) Relative
elongation rate during the inhibition and recovery phases of maize coleoptiles, normalized to the
steady acid-induced rate before salt addition: means ± SD; n = 8 (mock), 8 (NaCl). ( **C** ) Relative
500 30-min elongation of living Arabidopsis hypocotyls. Three-day-old seedlings were treated with 20
mM NaCl or 40mM mannitol in 50mM MES buffer (pH 5.2). The same 11 hypocotyls were
measured during NaCl treatment and after washout; n = 10 for mannitol. Elongation was expressed
as a percentage of the 30-min length increase in the untreated control (100%; dashed line). Data
are means ± SD; t-tests for the comparisons indicated by brackets; ****, P &lt; 0.0001; ns, not
505 significant.


16


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


**Fig. 5. Molecular basis of pH- and salt-dependent regulation of CsEXPA1.** ( **A** ) Predicted
CsEXPA1 structure from Alphafold. Conserved aromatic residues in domain-2 (red),
510 hydrophobic/aromatic residues lining the domain-1 groove (orange), D104 (purple) and predicted
disulfide bonds (yellow) are shown. ( **B, C** ) Corresponding H-T-D residues in the GH45 enzyme


17


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


PcCel45A (B; H112, T16 and D114) and CsEXPA1 (C; H102, T11 and D104). His, Thr and Asp
are shown in cyan, orange and red, respectively. ( **D** ) pH-REMD titration curves of H102 (blue)
and D104 (green) without added NaCl. ( **E** ) Hydrogen-bond occupancies for H102-D104 (orange),
515 T11-H102 (green) and T11-D104 (blue). ( **F** ) Total H102-D104 hydrogen-bond occupancy
(orange) and the contribution from the HIP⁺-ASP &#91;-&#93; state (green). ( **G, H** ) Titration curves of H102
(G) and D104 (H) without added NaCl (green) or with 100 mM NaCl (purple). In D, G and H,
symbols indicate simulated protonated fractions; black dashed curves show HendersonHasselbalch reference profiles at the indicated p _K_ ₐ values. Horizontal dashed lines indicate a
520 protonated fraction of 0.5. ( **I** ) H102-D104 salt-bridge occupancy within the HIP⁺-ASP &#91;-&#93; state
without added NaCl (red) or with 100 mM NaCl (green). Analyses in D-I were derived from pHREMD simulations. ( **J, K** ) Na⁺ radial distribution functions (RDF; J) and cumulative coordination
numbers (CN; K) around the D104 carboxylate oxygens in conventional MD simulations
containing 100 mM NaCl. D104 was fixed as ASP⁻ and H102 as HIE⁰ (blue), HID⁰ (red) or HIP⁺
525 (green). r denotes the Na⁺-carboxylate O distance. Lines indicate means and shaded bands indicate
± SD across three independent 500-ns simulations. ( **L** ) pH-dependent inhibition of acid-induced
hypocotyl elongation by Na⁺. Freeze-thawed hypocotyl segments under a 20-g load were
transferred from neutral HEPES buffer to 20 mM sodium acetate buffer at the indicated pH,
followed by the same buffer supplemented with 80 mM NaCl. Relative inhibition was calculated
530 as in Fig.2A and 2B. Data are means ± SD; n = 8 per pH.

535


18


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


540 **Figure. S1** Absolute values of 10-minute hypocotyl elongation and shrinkage after adding and
washing out 100 mM NaCl in the neutral buffer; means ± SD; n = 14; t-test; ns (not significant),
p&gt;0.05.

545

550


**Figure. S2** 10-minute hypocotyl elongation and shrinkage during addition and washout of 100
mM NaCl or NaAc in neutral buffer. Data are means ± SD; n = 14 (NaCl) and 15 (NaAc); t-test;
555 ns, P &gt; 0.05.


19


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


560 **Figure. S3** 10-minute hypocotyl elongation and shrinkage during addition of 50 mM CaCl 2,
washout, and subsequent chelation with 5 mM EDTA in neutral buffer. Data are means ± SD; n =
16; t-test; ns (not significant), p&gt;0.05.

565


570 **Figure S4. Effects of NaCl on the tensile mechanical properties of hypocotyl walls.** (A) Forcestrain curves of freeze-thawed cucumber hypocotyl wall specimens stretched along the growth axis
in neutral HEPES buffer containing 0 (blue) or 100 mM NaCl (red). Specimens were stretched at
a constant crosshead speed of 3 mm/min with an initial gauge length of 5 mm. Extension was
referenced to the point at which the load first reached 0.1 gf. Lines show means, and shaded bands
575 indicate ± SD. ( **B** ) Apparent tensile stiffness estimated from local slopes of the force-strain curves
at 5%, 10%, 15%, 20%, 25%, 30%, 35%, and 40% strain. For each specimen, local slopes were
estimated by linear regression within approximately ± 1 percentage point of each target strain.
Slopes were expressed in gf per percentage point of strain. Symbols and error bars represent means
± SD. n = 19 control and 17 NaCl-treated hypocotyls.


20


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


580
**Figure S5.** Creep of filter paper strips. ( **A** ) Schematic diagram of a typical material creep curve,
which consists of three stages: primary (initial, characterized by a decreasing creep rate),
secondary (steady-state, with an approximately constant creep rate), and tertiary (accelerated,
leading to failure). ( **B** ) Representative extension curve of filter paper strips. Dry filter paper strips
585 were clamped at both ends, hydrated with acidic buffer, and their length was recorded immediately
thereafter. Crude EXPA extract solubilized in the pH 5.2 MES buffer without (green) and with
(blue) 100 mM NaCl was applied at 30 min.

590


**Figure S6.** Extension rate of cellulose filter paper strips upon application of crude α-expansin with
or without NaCl. Crude EXPA solubilized in acidic buffer without (red) or with (green) 100 mM
595 NaCl was applied at 30 min (arrow) to filter paper strips pre-hydrated in acidic buffer containing
the corresponding NaCl concentration. The y-axis shows extension rate; the sharp increase and
subsequent drop of the rate to zero indicate failure of the strip. The average time to failure after
expansin addition is summarized in Fig. 4e.

600


21


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


**Figure S7. Sequence alignment of canonical Arabidopsis EXPA proteins and CsEXPA1.**
605 Conserved Thr, His and Asp residues (T-H-D) are marked with green stars. Conserved aromatic
residues in domain-2 and conserved hydrophobic/aromatic residues in domain-1 are boxed in


22


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


green and orange, respectively. The corresponding CsEXPA1 residues are labelled below the
alignment in matching colors. Matching green numbers beneath conserved cysteine residues
indicate predicted disulfide bonds.
610


615
**Figure S8. Protonation coupling within the H102-T11-D104 network of CsEXPA1.** ( **A, B** )
pH-REMD titration curves of H102 (A) and D104 (B) in wild-type (WT; blue) and T11A (green)
CsEXPA1. Arrows indicate shifts in the protonation transitions caused by T11A. ( **C** ) H102
titration with D104 fixed in its protonated, neutral state (ASH; purple) or deprotonated, negatively
620 charged state (ASP &#91;-&#93; ; orange). ( **D** ) D104 titration with H102 fixed as neutral HID &#91;0&#93; (orange) or HIE &#91;0&#93;
(blue) tautomers, or in the doubly protonated, positively charged HIP⁺ state (purple). HID &#91;0&#93; and
HIE &#91;0&#93; carry a proton on the N δ 1 and N ε 2 nitrogen atoms of the imidazole ring, respectively. Fixed
protonation states are labelled in colors matching the corresponding curves. WT curves (green) are
included for comparison in C and D. Symbols indicate simulated protonated fractions; dashed
625 horizontal lines indicate a protonated fraction of 0.5.


23


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


630 **Figure S9. Effects of NaCl on the titration behavior of additional residues in domain-1 of**
**CsEXPA1.** Titration curves of ( **A** ) H9, ( **B** ) D17, ( **C** ) E59, ( **D** ) D65 and ( **E** ) E107, without added
NaCl (green circles) or with 100 mM NaCl (purple triangles) in the explicit solvent. Points indicate
protonated fractions obtained from pH-REMD simulations across pH 2.8-7.0. Black dashed curves
show Henderson-Hasselbalch reference profiles at the indicated p _K_ ₐ values. Horizontal dashed
635 lines indicate a protonated fraction of 0.5.

640


24


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


**Figure S10. Effects of NaCl on T11-D104 hydrogen-bond occupancy in CsEXPA1.** Hydrogen645 bond occupancy between T11 and D104 was determined from pH-REMD simulations spanning
pH 2.8-7.0, without added NaCl (red open circles) or with 100 mM NaCl (green open triangles) in
the explicit solvent. Occupancy was calculated across all protonation states and expressed as a
percentage of the total sampled frames at each pH.


650


**Figure S11. Minimum cation distances to D104 carboxylate oxygens in fixed-protonation**
**conventional MD simulations.** Minimum distance between Na⁺ ( **A** ) or Ca²⁺ ( **B** ) and either D104
carboxylate oxygen in explicit solvent containing 100 mM NaCl (A) or 50 mM CaCl₂ (B). H102
and D104 were fixed as HIE⁰ and ASP &#91;-&#93;, respectively. Colored traces represent three independent
655 200-ns simulations. Dashed horizontal lines mark the 3.2 Å coordination cutoff.


25


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


**Table S1. pH-dependent protonation of titratable residues in hen egg-white lysozyme**
**(HEWL).** Values at each pH represent the fraction of each residue in its protonated state, as
determined by pH-REMD simulations. Apparent p _K_ ₐ values and Hill coefficients ( _n_ H ) were
660 obtained by fitting the titration curves to the generalized Henderson-Hasselbalch equation. The
p _K_ ₐ values of D48, D66, D87 and D119 were estimated by extrapolating the fitted curves below
the simulated pH range (2.8-7.0). MAE and RMSE denote the mean absolute error and root-meansquare error, respectively, between calculated and experimental p _K_ ₐ values. Experimental p _K_ ₐ
values (*) were taken from Webb et al. (2011).


665


**Table S2. pH-dependent protonation of titratable residues in domain-1 of CsEXPA1.** Values
at each pH represent the fraction of each residue in its protonated state, as determined by pH670 REMD simulations. Apparent p _K_ ₐ values and Hill coefficients ( _n_ H ) were obtained by fitting the
titration curves to the generalized Henderson-Hasselbalch equation. The p _K_ ₐ of E59 (1.50) was
estimated by extrapolating the fitted curve below the simulated pH range (2.8-7.0).

675


26


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


**Table S3. pH-dependent protonation of titratable residues in domain-1 of CsEXPA1 with**
680 **100 mM NaCl in the explicit solvent.** Values at each pH represent the fraction of each residue in
its protonated state, as determined by pH-REMD simulations. Apparent p _K_ ₐ values and Hill
coefficients ( _n_ H ) were obtained by fitting the titration curves to the generalized HendersonHasselbalch equation. The p _K_ ₐ of E59 (1.10) was estimated by extrapolating the fitted curve below
the simulated pH range (2.8-7.0).

685

**Materials and Methods**


**Collection of cucumber hypocotyls and maize coleoptiles and measurement of extension**


690 Cucumber (Cucumis sativus) seeds were sown on germination paper moistened with distilled
water and grown in darkness at 26 °C for 80 h, until seedlings reached 2.5-3 cm in length. Apical
hypocotyl segments were harvested and stored at -80 °C until use. After thawing, the cuticle was
gently abraded to facilitate buffer penetration. The freeze-thawed hypocotyls were placed between
two glass slides and compressed under a 750-g load for 20 min to remove residual liquid. A 0.5695 cm apical segment was then clamped at both ends and subjected to a 20-g constant load. Segment
length was recorded every 30 s, and extension rate were calculated as described previously ( _4_ ). To
inactivate endogenous α-expansin activity, compressed hypocotyls sandwiched between glass
slides were immersed in boiling distilled water for 60 s and then immediately cooled in fresh
distilled water.


700 Maize (Zea mays) seeds were sown on germination paper moistened with distilled water and
grown in darkness at 26 °C for 80 h, until coleoptiles reached 2.5-3 cm in length. Coleoptiles were
harvested, and the cuticle was gently abraded to facilitate buffer penetration. They were then stored
at -80 °C until used in extension assays.


**Tensile testing and stiffness analysis of hypocotyl walls**


705 Freeze-thawed cucumber hypocotyl wall specimens were incubated for 45 min in 20 mM HEPES
buffer (pH 7.0) containing 0 or 100 mM NaCl. Specimens were stretched along the growth axis
using an Instron 68SC-05 universal testing machine, with an initial gauge length of 5 mm and a
constant crosshead speed of 3 mm/min, to 50% strain or until specimen failure, whichever occurred
first. Force and extension were recorded simultaneously. Extension was zeroed at the first recorded
710 point at which the force reached or exceeded 0.1 gf. Percentage strain was calculated as,


27


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


**𝑆**



0

𝜀𝜀= 100 &#91;∆𝐿𝐿&#93;

𝐿𝐿


**𝑆**



0

_L_ 𝐿𝐿


**𝑆**



where _ΔL_ is extension from this reference point and _L_ _0_ = 5 mm.


**𝑆**



Local apparent axial tensile stiffness was estimated at 5% increments from 5% to 40% strain. For
each specimen, a fitting window comprised 101 consecutive force-strain data pairs: the recorded
715 point closest to the target strain and the 50 immediately preceding and 50 immediately following
points. Each window spanned approximately ±1 percentage point of strain. Force was fitted against
percentage strain by ordinary least-squares linear regression with an intercept, **𝑆**



, where is the fitted intercept. The local slope, _S_, was calculated as **𝑆**


101

𝑎𝑎 ∑ &#91;�&#93; )



_S_ **𝑆**


∑ 101 )

∑ 101

𝑖𝑖=1 𝑖𝑖 𝑖𝑖
(𝜀𝜀 −𝜀𝜀̅)(𝐹𝐹 −𝐹𝐹 &#91;�&#93;

&#91;2&#93;



𝐹𝐹= 𝑎𝑎+ 𝑆 _S_ **𝑆**



**𝑆**


∑ 101

𝑖𝑖 𝑖𝑖

𝑖𝑖=1 𝑖𝑖
(𝜀𝜀 −𝜀𝜀̅) &#91;2&#93;



**𝑆**


720



**𝑆**


∑

, where and are the mean percentage strain and force within the fitting window, respectively. &#91;2&#93;
Slopes were calculated separately for each specimen and expressed in gf per percentage point of
strain. 𝜀𝜀̅ 𝐹𝐹 &#91;�&#93;



**𝑆**

𝑆𝑆=



**𝑆**


**Crude α-expansin extraction**


725 Crude α-expansin extraction from cucumber hypocotyl has been detailed previously, and the cell
wall loosening activity from the extraction has been demonstrated to originate from α-expansin
( _4_ ). Briefly, fresh cucumber hypocotyls germinating in the dark at 26ºC for 80 hours were
homogenized, and the cell wall fragments were collected and washed. Then, crude α-expansin was
solubilized in 1 M NaCl and then precipitated with saturated (NH 4 ) 2 SO 4 . To assess the protein
730 content, the crude protein diluted in acidic buffer for the acid growth assay was centrifuged at
12000 rpm for 5 minutes at 4 °C, and the total protein concentration was immediately determined
to 360 μg/ml by measuring the absorbance at 280 nm.


**Arabidopsis hypocotyl elongation assay**


Arabidopsis seedlings were grown in the dark at 22°C for 3 days. Etiolated hypocotyls were
735 immobilized on 0.5% agarose pads containing 20 mM MES buffer, pH 5.2, and imaged under a
microscope every 2 min. Hypocotyl length was quantified over 30 min before treatment, during
treatment, and after washout. For treatment, seedlings were transferred to 0.5% agarose pads
containing 20 mM MES, pH 5.2, supplemented with either 20 mM NaCl or 40 mM mannitol. For
recovery, seedlings were rinsed and transferred back to agarose pads containing 20 mM MES, pH
740 5.2, without added solute.3


**Molecular dynamics simulations**


The CsEXPA1 protein structure was obtained from the AlphaFold Protein Structure Database (AFQ39625-F1-v6)( _37, 38_ ). Molecular topologies were generated using tleap in AmberTools( _58_ ). For


28


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


pH-REMD simulations, the protein was parameterized with the Amber ff19SB force field, and
745 water with the OPC model; constant-pH-compatible parameters were loaded using leaprc.constph.
For fixed-protonation conventional MD simulations, the same ff19SB, and OPC parameter sets
were used. For all systems, OPC-matched Li-Merz ion parameters were loaded using
frcmod.ionslm_126_opc (Na &#91;+&#93; ) and frcmod.ionslm_1264_opc (Ca &#91;2+&#93; ). Prepared PDB structures
were loaded into tleap, and four predicted disulfide bonds were explicitly assigned between Cys26750 Cys54, Cys57-Cys130, Cys62-Cys69, and Cys82-Cys96. Systems were checked, solvated in a
truncated octahedral OPC water box with a minimum 12 Å buffer, and neutralized with Cl &#91;-&#93;
counterions using addIons2. Explicit Na⁺/Cl &#91;-&#93; or Ca²⁺/Cl &#91;-&#93; ions were added to produce final NaCl
and CaCl₂ concentrations of 100 mM and 50 mM, respectively.


All molecular dynamics simulations were performed with Amber22 ( _59_ ).


755 pH-REMD simulations followed the hybrid-solvent discrete-protonation protocol of Swails et al.
(2014)( _47_ ), with modifications described below. Protein conformations were sampled by MD in
explicit solvent, whereas discrete protonation-state changes were evaluated using Monte Carlo
(MC) moves with a generalized Born (GB) implicit-solvent model. Replicas spanned pH 2.8-7.0
in 0.6-unit increments. Each replica was first equilibrated for 2 ns under NVT conditions at its
760 assigned pH with constant-pH sampling enabled. Production simulations were then performed for
50 ns per replica, with exchange attempts between neighboring pH replicas every 1,000 MD steps.
Protonation-state updates were attempted every 100 MD steps using the hybrid-solvent constantpH implementation (icnstph = 2). During each update, explicit water and nonstructural ions were
temporarily excluded from the energy evaluation, and proposed protonation states were assessed
765 by MC using the GB model at fixed solute coordinates and the assigned pH. The implicit salt
concentration used in the GB calculations was set to 0.1M. Explicit solvent and ions were
subsequently restored. Following accepted protonation-state changes, the solvent and ions were
relaxed for 100 MD steps with the solute held fixed before full explicit-solvent MD resumed.
Simulations were carried out at 300 K using Langevin dynamics with a collision frequency of 2.0
770 ps⁻¹ and automatically assigned random seeds (ig = -1). Bonds involving hydrogen atoms were
constrained with SHAKE, allowing a 2-fs time step. Periodic boundary conditions were applied
with a 10 Å nonbonded cutoff. Coordinates were saved every 10 ps and restart files every 100 ps.
Convergence was assessed by comparing protonation fractions across successive 10-ns blocks. For
protocol validation, simulations of hen egg-white lysozyme (HEWL) were initiated from its crystal
775 structure obtained from the Protein Data Bank (PDB ID: 1AKI).


**&#91;𝑝&#93;**
**𝑝** **𝑝**

**𝑡** **𝑡**



𝑝 𝑝 = &#91;𝑁𝑁&#93; &#91;𝑝&#93; **&#91;𝑝&#93;** &#91;𝑝&#93;

𝑡 𝑡

**𝑝** **𝑝**
𝑓𝑓 **𝑡** **𝑡**

𝑁𝑁



For each titratable residue, the protonated fraction at each pH was calculated as:

**&#91;𝑝&#93;**
**𝑝** **𝑝**

**𝑡** **𝑡**



𝑡 𝑡

&#91;𝑝&#93; **&#91;𝑝&#93;** &#91;𝑝&#93;
**𝑝** **𝑝**

**𝑡** **𝑡** 𝑡𝑡
𝑁𝑁




**&#91;𝑝&#93;**
**𝑝** **𝑝**

**𝑡** **𝑡**

where _N_ _prot_ and _N_ _total_ denote the number of protonated samples and the total number of samples
collected at that pH, respectively. Protonated microstates were combined for Asp and Glu. For His,
780 the protonated fraction corresponded to HIP &#91;+&#93;, whereas HID⁰ and HIE⁰ together constituted the
neutral fraction. Apparent p _K_ ₐ values and Hill coefficients ( _n_ H ) were obtained by nonlinear leastsquares fitting to the generalized Henderson-Hasselbalch (Hill) equation:




**&#91;𝑝&#93;**
**𝑝** **𝑝**

**𝑡** **𝑡**


29


&#91;bioRxiv preprint doi: https://doi.org/10.64898/2026.11.14.766679; this version posted November 16, 2026. The copyright holder for this preprint&#93;(https://doi.org/10.64898/2026.11.14.766679)

(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
available under a &#91;Submitted Manuscript: Confidential&#93;(http://creativecommons.org/licenses/by/4.0/) &#91;CC-BY 4.0 International license.&#93;(http://creativecommons.org/licenses/by/4.0/)


1
𝑝 𝑝 (𝑝 ) = 1 + 10 &#91;𝑛𝑛&#93; &#91;𝐻𝐻&#93; &#91;(𝑝&#93; **&#91;𝑝&#93;** &#91;−𝑝𝑝𝐾𝐾&#93; &#91;𝑎𝑎&#93; &#91;)&#93;
**𝑝** **𝑝**
𝑓𝑓 **𝑝**


785 The fitted p _K_ ₐ corresponds to the pH at which _f_ prot = 0.5, and _n_ H describes the steepness of the
titration transition. A value of _n_ H = 1 gives standard Henderson-Hasselbalch behavior, whereas
values below or above 1 indicate broader or steeper transitions, respectively.


For conventional MD simulations, the protonation states of H102 and D104 were assigned as
indicated. All other titratable residues were kept in their default protonation states: Asp as ASP &#91;-&#93;,
790 Glu as GLU &#91;-&#93;, and His as HIE &#91;0&#93; . All quantitative analyses were based on three independent replicas.
Each replica was first equilibrated for 1 ns under NPT conditions and then subjected to 500 ns of
production MD. Replicas used time-dependent Langevin random seeds (ig=-1), and their initial
velocities were verified to be distinct. Production simulations were carried out under NPT
conditions using the Monte Carlo Barostat. Simulations were performed at 300 K with Langevin
795 dynamics and a collision frequency of 2.0 ps &#91;-&#93; ¹. Bonds involving hydrogen atoms were constrained
with SHAKE, allowing a 2-fs time step. Nonbonded interactions were calculated with a 10 Å cutoff
under periodic boundary conditions. Coordinates were saved every 10 ps, and restart files were
written every 100 ps. All trajectory analyses were performed using custom Python scripts with
NumPy and Amber NetCDF/topology files, applying periodic minimum-image corrections for
800 distance-based measurements.


Hydrogen bonds were identified using geometric criteria. A hydrogen bond was assigned when
the donor-acceptor distance was ≤ 3.2 Å and the donor-hydrogen-acceptor angle was ≥ 120°. Salt
bridges were defined by a minimum distance ≤ 4.0 Å between oppositely charged heavy atoms,
including H102 imidazolium nitrogens and D104 carboxylate oxygens in the HIP⁺-ASP &#91;-&#93; state.
805 Occupancies were calculated as the fraction of trajectory frames satisfying the corresponding
criterion, with periodic minimum-image corrections applied to all distance measurements.


30


