A Structural Bridge Between Kingdoms: How Collagen‐Derived Peptides Influence Plant Stress and Growth Pathways

ABSTRACT

Collagen‐derived protein hydrolysates (CDPH) are widely used as plant biostimulants primarily due to their content of bioactive oligopeptides. When applied to hydroponically grown Solanum lycopersicum plants, CDPH significantly promoted root development, particularly by increasing the number and length of lateral roots. To gain insight into the underlying molecular mechanisms, we hypothesized that plants may possess proteins capable of interacting with collagen‐like peptides. To test this, we conducted a comprehensive homology search of the Arabidopsis thaliana proteome using a Hidden Markov model‐based approach built from three human collagen‐binding proteins (CBPs) and 14 known collagen‐binding domains (CBDs). After filtering, 10 Arabidopsis proteins emerged as putative candidates with the potential to bind collagen. Notably, the highest homology was observed for a matrix metalloproteinase, At5‐MMP, showing 44% identity with its human counterpart HsMMP1, and for AtSERPIN1, which displayed the strongest e‐value match to HsSERPINH1 (22% identity). Both plant proteins are functionally associated with responses to abiotic and biotic stresses, a feature that mirrors the known physiological effects of CDPH‐based biostimulants. These findings support the hypothesis that plants possess proteins capable of recognizing collagen‐like structures, offering a plausible molecular basis for the activity of CDPH‐based biostimulants and paving the way for future biochemical validation.

1. Introduction

Biostimulants represent a heterogeneous class of products capable of modulating plant physiological processes, offering promising avenues for enhancing growth, development, and stress resilience (du Jardin ²⁰¹⁵). Among these, protein hydrolysates (PHs) are a prominent category (Moreno‐Hernández et al. ²⁰²²). Produced from the physical, chemical, or enzymatic breakdown of plant and animal by‐products and wastes, PHs are rich in oligopeptides and free amino acids (Cavani et al. 2003; Pituello et al. ²⁰²²). The beneficial effects of PHs on plants are well documented in the literature (Malécange et al. 2023). Among the animal‐derived PHs recognized as effective plant biostimulants are those obtained from keratin (chicken feathers), hemoglobin (erythrocytes), and collagen (bovine hair, epithelial tissues; also available as gelatin) (Cristiano et al. 2018; Tejada et al. ²⁰¹⁸; Wilson et al. ²⁰¹⁸). Although their plant growth‐promoting effects—such as enhanced crop growth, biomass accumulation, improved nutrient uptake, and greater abiotic stress tolerance—are well established, the molecular mechanisms underlying these effects remain largely unexplored. Moreover, the fundamental biological question of how plants perceive and respond to the bioactive molecules in these plant‐conditioning compounds remains unresolved. Notably, previous studies have demonstrated that a specific type of PH, collagen‐derived protein hydrolysate (CDPH), promotes maize root growth by modulating the expression of genes involved in cell wall organization, transport, stress response, and hormone metabolism (Ertani et al. 2013; Santi et al. ²⁰¹⁷) and enhances tolerance to abiotic and nutritional stresses (Ambrosini et al. 2021). CDPH fractionation revealed a heterogeneous peptide composition (200–6500 Da) (Ambrosini et al. 2022) and circular dichroism analysis indicated that CDPH spectra closely match type II polyproline helices (PPII), suggesting the presence of peptides that retain this secondary structure (Ambrosini et al. 2021).

Collagens, the most abundant proteins in mammals, comprise approximately 30% of total protein mass and play critical roles in cellular and tissue development, including processes such as cell growth, differentiation, and migration (Ricard‐Blum ²⁰¹¹). The extracellular matrix contains 28 types of collagens, further diversified by isoforms and supramolecular organizations (Mienaltowski and Birk 2014). Other functional diversity arises from proteolytic cleavage, which releases bioactive peptides, and from cryptic functional sites exposed by conformational changes (Ricard‐Blum ²⁰¹¹). All collagens share at least one domain composed of three polypeptides folded into a triple helix, featuring a repetitive [GXY]_n motif, where X and Y are often proline (Pro) and 4‐hydroxyproline (Hyp) (Jariwala et al. 2022). Extensive research in humans has identified numerous collagen‐binding proteins (CBPs) and, in some cases, experimentally characterized collagen‐binding domains (CBDs). A review by Elango et al. (2022) highlights the remarkable diversity of proteins interacting with collagens, including receptor tyrosine kinases, integrins, immunoglobulin‐like receptors, and leukocyte receptor complexes (LRCs). Collagen fragments, from small fibrils to large triple‐helix segments, interact with receptors in various ways, eliciting distinct physiological responses. These findings suggest, as noted by Elango et al. (2022), an absence of a consistent molecular pattern in collagen‐mediated responses.

This study investigates whether collagen‐derived peptides can be recognized or bound by plant proteins. Specifically, we hypothesize that plants may possess putative collagen‐interacting proteins with structural similarities to known human collagen‐binding proteins (CBPs) and collagen‐binding domains (CBDs). Through computational analysis and predictive modeling, we aim to identify and characterize candidate collagen‐interacting proteins in two species chosen for their complementary value: Arabidopsis thaliana for its well‐annotated genome, facilitating in silico protein identification and structural prediction, and Solanum lycopersicum for its agronomic relevance and responsiveness to biostimulants.

2. Material and Methods

2.1. Hydroponics Growth and Biostimulant Treatment

Solanum lycopersicum seeds (Micro‐Tom) were sterilized and laid down in 8 g L⁻¹ agar plates placed in the growth chamber to germinate for 5 days (120 μE m⁻² s⁻¹ average light intensity;16 h light/8 h dark photoperiod; 25°C). Similarly sized seedlings, with a root of approximately 3 cm, were transferred to 600‐mL pots containing a nutrient solution with the following composition: 400 μM CaSO₄, 200 μM K₂SO₄, 200 μM KNO₃, 175 μM KH₂PO₄, 100 μM MgSO₄, 50 μM Na‐Fe‐EDTA, 25 μM (NH₄)H₂PO₄, 5 μM KCl, 2.5 μM H₃BO₃, 0.2 μM MnSO₄, 0.2 μM ZnSO₄, 0.05 μM CuSO₄, 0.05 μM (NH₄)₆Mo₇O₂₄, pH 6.0. Each pot accommodated 20 seedlings. Seedlings were treated with the CDPH at a concentration of 1.4 mg N L⁻¹, or with an equivalent amount of N supplied as (NH₄)H₂PO₄. The CDPH employed has been thoroughly characterized in previous publications (Santi et al. 2017; Ambrosini et al. ²⁰²¹). The treatments were applied when plants were transferred to the hydroponic system (t ₀), 5 days after t ₀, and 7 days after t ₀. Plants were sampled an hour after the last treatment. Root parameters were analyzed with WinRHIZO software (EPSON V850 Pro, WinRHIZO‐ Pro2021a‐ Regent Instruments Inc.). Data were analyzed applying a two‐tailed Welch's t‐test (p > 0.05).

2.2. Protein Homology Assessment

The software HHPred (Meier and Söding 2015) was employed to assess the homology between human CBPs or CBDs and Arabidopsis thaliana proteins (default parameters). In particular, the Hidden Markov Models‐based program, HHsearch, within HHpred was used with default parameters. Here, we list the PDB IDs of the analyzed entries and the scientific work that determined the functional CBD, if present; otherwise, the CBP is indicated: Osteoclast associated immunoglobulin‐like receptor (OSCAR), 5CJB_1 (Haywood et al. 2016); Platelet glycoprotein VI (GP6), 2GI7_1 (Feitsma et al. 2022); Leukocyte‐associated immunoglobulin‐like receptor (LAIR1), 3KGR_1 (Brondijk et al. 2010); Integrin alpha‐2 (ITGA2), 1DZI_1 (Emsley et al. 2000); Discoidin domain‐containing receptor 2 (DDR2), 2WUH (Carafoli et al. 2009); Megakaryocyte and platelet inhibitory receptor G6b (MPIG6B), CBP: 6R0X (Vögtle et al. 2019); C‐type mannose receptor 2 (MRC2), 5AO5_1 (Paracuellos et al. 2015); Mannan‐binding lectin serine protease 1 (MASP1), 3DEM (Nan et al. 2017); Complement C1s subcomponent (C1S), 4LOR_1 (Girija et al. 2013); Serine proteinase inhibitors (SERPINH1), 4AU2 (Widmer et al. 2012); Secreted protein acidic and rich in cysteine (SPARC), 2V53_1 (Hohenester et al. 2008); von Willebrand factor (vWF), 1ATZ_1 (Brondijk et al. 2012); Matrix metalloproteinase‐1 (MMP‐1), 4AUO (Manka et al. 2012); Kelch repeat and BTB domain‐containing protein 4 (KBTBD4), PDB structure not present, Uniprot ID: Q9NVX7; Fibronectin (I), FN1, 3EJH_1 (Erat et al. 2009). For two entries, ITGA11 and DDR1, the CBDs were determined by similarity via BlastP from ITGA2 and DDR2, respectively.

The structural model of At5‐MMP was carried out by using the program SWISS‐MODEL (https://swissmodel.expasy.org/, Waterhouse et al. ²⁰¹⁸). The template used was the CBD of HsMMP1 [PDB accession code: 4AUO.1pdb modifying A200E to match the residue present in the active site (Chung et al. 2004)]. AtSERPIN1 was modeled on the canine SERPINH1 protein (PDB accession code: 3ZHA), which is almost identical (99%) to the human HsSERPINH1 (alignment not shown) and, in this case, was co‐crystallized with the collagen triple helix and therefore more informative.

The list of Uniprot IDs and the GRAMENE database accession for the Solanum lycopersicum orthologs presented in Table 2 was retrieved by the TAIR database accession when present; alternatively, they were identified by the UniProt BLAST function employing Arabidopsis thaliana identified proteins as queries.

TABLE 2.

List of Solanum lycopersicum putative orthologs for CBDs and CBPs.

Reference CBD/CBP	Arabidopsis thaliana	Solanum lycopersicum
MMP‐1	At5‐MMP, matrixin metalloproteinase Q9ZUJ5, At1g59970	SlMMP1, matrix metalloproteinase I7JCM3, Solyc04g005050
SERPINH	AtSERPIN1, serine protease inhibitor A0A1P8API2, At1g47710	Serpin domain‐containing protein A0A3Q7GAG7, Solyc04g079440.3
SPARC	AtCP1, Ca2+‐binding protein 1 Q9FDX6, At5g49480	EF‐hand domain‐containing protein A0A3Q7EET5, Solyc01g058720.3
MRC	C‐type LecRLK, lectin receptor‐like kinase Q9C823, At1g52310	Protein kinase domain‐containing protein A0A3Q7J058, Solyc02g068370.3.1
ITGA2, ITGA11, vWF	AtMUP24.2 Q9FF49, At5g60710	Uncharacterized protein A0A3Q7J058, Solyc11g069600.2
ITGA2, ITGA11, vWF	AtSEC24A Q9SFU0, At3g07100	Protein transport protein Sec24‐like A0A3Q7F833, Solyc02g082220.3
ITGA2, vWF	Inter‐alpha‐trypsin inhibitor heavy chain‐like protein A0A1P8ANU9, At1g72500	VWFA domain‐containing protein A0A3Q7FW48, Solyc03g122270.3
DDR1	AtFUC1, alpha‐L‐fucosidase 1 Q8GW72, At2g28100	Alpha‐L‐fucosidase A0A3Q7J033, Solyc11g069000.2.1
ITGA2, ITGA11, vWF	AtRPN10, 26S proteasome non‐ATPase regulatory subunit 4 homolog P55034, At4g38630	26S proteasome regulatory subunit RPN10 A0A3Q7F6N7, Solyc02g083710.3
ITGA2, ITGA11, vWF	AtBABAM1, BRISC and BRCA1‐A complex member 1 O82638, At4g32960	BRISC and BRCA1‐A complex member 1 A0A3Q7JEC8, Solyc12g095920.2

Note: For each protein, the UniProt ID is listed. For Arabidopsis thaliana , the gene ID refers to the TAIR database accession, while for Solanum lycopersicum , it refers to the Gramene database accession. Solanum lycopersicum orthologs were retrieved by the TAIR database accession when present; alternatively, they were identified by the UniProt BLAST function employing Arabidopsis thaliana identified proteins as queries.

3. Results and Discussion

The treatment with CDPH significantly promoted root growth in Solanum lycopersicum seedlings (Figure 1). Treated seedlings developed roots that were 45% longer and 41% wider than those of control plants (Figure 1A,B), with a 68% increase in lateral root length (Figure 1C) and a higher number of lateral roots (Figure 1D). Overall, the root system of treated plants exhibited greater biomass than control plants, with fresh and dry root weights increasing by 33% and 47%, respectively. In contrast, shoot biomass was not significantly affected by CDPH application (Figure 1G,H). The stimulatory effect of CDPH on root growth was previously observed in maize, both in optimal conditions and under abiotic stress (hypoxia and water stress) (Santi et al. 2017; Ambrosini et al. ²⁰²¹).

FIGURE 1.

Solanum lycopersicum root and shoot parameters. (A) Total root length, (B) total surface area, (C) lateral root length (D) number of lateral roots, (E) root fresh weight, (F) dry weight, (G) shoot fresh weight and (H) shoot dry weight after 7 days of hydroponic growth with the addition of either inorganic N (CTR) or of the CDPH. Data, derived from three independent experiments, are expressed as mean ± S.E.M. (n = 76). Statistical method: Two‐tailed Welch's t‐test (p < 0.05).

In silico sequence analyses were conducted to find human CBPs (collagen‐binding proteins), and 17 entries were identified (Table S1). The list included three LRCs (OSCAR, GP6, LAIR1), two alpha integrins (ITGA2 and ITGA11), two DDRs (DDR1 and DDR2), MPIG6B, MRC2, MASP1, C1S, SERPINH1, SPARC, vWF, MMP‐1, KBTBD4, and FN1.

To improve the homology search, the CBDs (collagen‐binding domains) of the CBPs in the list were employed for the analysis when available. For MPIG6B, SERPINH1, and KBTBD4, a CBD was not found or was not clearly established; therefore, the whole protein sequence was used for the following analysis. The identified CBDs or CBPs were then used for an extensive homology search on Arabidopsis thaliana using the state‐of‐the‐art HHpred prediction server (Meier and Söding 2015). We focused on Arabidopsis thaliana because its protein databases are significantly more comprehensive and well‐annotated compared to those of other plant species. This allowed for a more reliable identification of putative CBPs and facilitated structural and functional predictions, providing a solid foundation for future validation in other plant models such as Solanum lycopersicum . Overall, 358 entries with an e value ≤ 0.001 were retrieved, heterogeneously distributed among the list of human CBPs (Table S2). The search with MPIG6B, MASP1, C1S, KBTBD4, FN1, and with the three LRCs did not reveal any statistically significant homologous protein. DDR1 and DDR2 both displayed the same plant homologue, the alpha‐L‐fucosidase FUC1, with slightly different e‐values (for DDR2 above the threshold). The MRC2 was also connected only to one plant homologue, a C‐type lectin domain‐containing protein. The entries that showed the higher number of outcomes were the CBP SERPINH1 (18 hits) and the CBDs MMP‐1, ITGA2, vWF, ITGA11, and SPARC (4, 63, 68, 70 and 133 hits respectively). After removing redundant entries, the final dataset included 213 proteins (Table S3), most of which (124 non‐redundant/161 total hits) were associated with calcium‐binding functions. These were grouped into four categories: calmodulin‐like proteins, copines, Ca‐binding EF‐hand proteins, and other Ca‐related proteins. Other groups represented included 18 Sec23/Sec24 transport proteins, 16 SERPINs (serine protease inhibitors), 13 Zinc finger proteins (C3HC4‐type RING finger), 7 RING proteins (domain ligase 2), and 4 matrixins (Table S3). For further analysis, we focused on 10 selected hits, ensuring that at least one plant homolog was represented for each of the eight human entries. Selection was based on biological criteria, requiring that the proteins are expressed in specific organs and cellular localizations where interactions with collagen‐derived oligopeptides are likely to occur. Based on the literature and the TAIR database, we focused on proteins expressed in roots and leaves. Both our data and previous studies (Ertani et al. 2009; Santi et al. ²⁰¹⁷; Ambrosini et al. ²⁰²¹) involving the addition of protein hydrolysates (PHs) to the nutrient solution have shown that roots are highly responsive to treatment. Additionally, evidence suggests that animal‐derived PHs can also exert biostimulant effects when applied as a foliar spray (Tejada et al. 2018). Concerning cellular localization, we selected proteins expressed in the extracellular matrix (ECM) or in the cytosol. In Table 1, we show the 10 proteins that have an identity percentage above 10% and that match the chosen biological.

TABLE 1.

List of putative plant orthologues for CBDs and CBPs. Arabidopsis thaliana proteins found by homology via HHPred using as a query CBDs (ITGA2, ITGA11, DDR1, DDR2, OSCAR, GP6, LAIR, MRC, MASP, C1S, SPARC, vWF, MMP‐1, FN1) or CBPs (MPIG6B, SERPINH1, KBTBD4).

Reference CBD/CBP	HHPred predicted homologues in Arabidopsis thaliana	Tissue expression	Subcellular localization	Identity
MMP‐1	At5‐MMP, matrixin metalloproteinase	Roots, leaves, stem (to a lower extent in flowers) (Maidment et al. 1999)	Cell membrane (lipid‐anchor) (Maidment et al. 1999)	44%
SERPINH	AtSERPIN1, serine protease inhibitor	Ubiquitary (Lampl et al. 2013)	Cytosol (Lampl et al. 2013; Asqui et al. 2018)	22%
SPARC	AtCP1, Ca2+‐binding protein 1	Roots and flowers (Jang et al. 1998)	Cytosol (Jang et al. 1998)	20%
MRC	C‐type LecRLK, lectin receptor‐like kinase	Ubiquitary*	Cell membrane (single‐pass type I membrane protein)	19%
ITGA2, ITGA11, vWF	AtMUP24.2	Ubiquitary*	Plasma membrane**	18%
ITGA2, ITGA11, vWF	AtSEC24A	Mainly expressed in pollen, leaves, inflorescences, roots and stems, and, to a lower extent, in cotyledons, petioles and hypocotyls (Sato and Nakano 2007)	Cytoplasmic vesicle COPII, ER membrane, cytosol (Sato and Nakano 2007)	17%
ITGA2, vWF	Inter‐alpha‐trypsin inhibitor heavy chain‐like protein	Ubiquitary*	Plasma membrane**	17%
DDR1	AtFUC1, alpha‐L‐fucosidase 1	Ubiquitary*	ECM (de la Torre et al. 2002)	14%
ITGA2, ITGA11, vWF	AtRPN10, 26S proteasome non‐ATPase regulatory subunit 4 homolog	Ubiquitous, highest expression in flowers (Marshall et al. 2019)	Nucleus, plasma membrane, proteasome complex (Marshall et al. 2019)	10%
ITGA2, ITGA11, vWF	AtBABAM1, BRISC and BRCA1‐A complex member 1	Ubiquitary*	Cytosol, nucleus**	10%

Note: Only hits with an e value ≤ 0.001 were considered. When a plant protein was retrieved from multiple queries, then the identity percentage refers to the hit with a higher percentage of identity (and it is indicated in bold). Asterisks indicate that the tissue expression or subcellular localization were retrieved respectively from the database TAIR (*) and UniProt (**), while for the other entries we reported information present in the literature.

The data obtained for Arabidopsis thaliana were used to identify homologous proteins in Solanum lycopersicum (Table 2). These results suggest that homology can be established between mammalian and plant proteins. Since HsMMP‐1 and HsSERPINH1 in mammals and their homologues At5‐MMP and AtSERPIN1 in plants showed the highest identity percentages among the retrieved proteins, we chose to focus our discussion on these pairs. HsMMP‐1 is a matrix metalloproteinase (zinc‐containing endopeptidase) secreted in the ECM by different cell types (i.e., fibroblasts, keratinocytes, endothelial cells, macrophages and hepatocytes) and it can recognize and cleave collagen I, II, III, VII, VIII, X, gelatin (a mixture of peptides derived from the disruption of fibrillar collagen cross‐linkages between the polypeptide chains and a partial breakage of polypeptide bonds) as well as aggrecan, versican, perlecan, casein, nidogen, serpins, and tenascin‐C (McCawley and Matrisian ²⁰⁰¹; Chang ²⁰²³). The HsMMP‐1 gene encodes for a 469 AAs protein that comprises a signal peptide, a prodomain, a catalytic domain (where collagen fibrils are cleaved), and a hemopexin‐like C‐terminal domain (crucial to unwind the fibrillar structure exposing the cleavage site to the catalytic domain) (Bertini et al. 2012). The interaction with the collagen triple helix involves both the hemopexin and the catalytic domains, is temperature‐dependent, and was confirmed by biochemical assays employing short collagen model peptides of 27 AAs (Manka et al. 2012).

The Arabidopsis thaliana homologue of HsMMP‐1, At5‐MMP, encodes a zinc‐dependent endopeptidase (Maidment et al. 1999). At5‐MMP is one of the MMPs identified in Arabidopsis thaliana , which function as active proteases with potentially overlapping but distinct roles in extracellular matrix (ECM) remodeling, degradation, and/or shedding of its components (Marino et al. 2014). These proteins share a basic structural organization, consisting of a signal peptide, a propeptide domain, and a catalytic domain (Flinn 2008). As with other plant MMPs, all five Arabidopsis thaliana MMPs lack a hemopexin‐like domain (Massova et al. 1998). However, in four of these proteins (AtMMP1, AtMMP2, AtMMP3, and AtMMP5), a putative C‐terminal transmembrane domain has been predicted (Maidment et al. 1999; Flinn ²⁰⁰⁸). For AtMMP2, AtMMP4, and AtMMP5, a glycosylphosphatidylinositol (GPI)‐anchored site has been predicted (Flinn 2008). At5‐MMP is constitutively expressed in leaves, buds, and siliques, with the highest expression levels observed in roots (Mishra et al. 2021). Functional characterization of the At5‐MMP knock‐out mutant (At5mmpKO) revealed reduced primary root growth, fewer lateral roots, increased stomatal conductance, and decreased water‐use efficiency under standard growth conditions compared to WT plants. Additionally, the mutants displayed impaired root‐to‐shoot auxin transport and reduced abscisic acid (ABA) accumulation in roots. Upon NaCl treatment, At5mmpKO plants exhibited heightened sensitivity to osmotic stress and an alteration in the composition of the rhizosphere bacterial community relative to WT (Mishra et al. 2021). At5‐MMP is involved in plant responses to biotic stress (Zhao et al. 2017). These authors reported that At5‐MMP expression is transiently induced at 16 h post‐inoculation (hpi) in Arabidopsis thaliana leaves following infection with Botrytis cinerea. Additionally, homozygous at5‐mmp mutants exhibited increased susceptibility to the fungus compared to wild‐type plants. Interestingly, the homologous Solanum lycopersicum protein encoded by Sl3‐MMP (Solyc04g005050), one of the five MMP genes identified in Solanum lycopersicum , has also been implicated in resistance to Botrytis cinerea and Pseudomonas syringae pv. tomato (Li et al. 2015). Given that At5‐MMP and its tomato homolog Sl3‐MMP are both implicated in responses to biotic stresses, particularly Botrytis cinerea, we propose that tomato MMPs may share functional roles with their Arabidopsis thaliana counterparts. In our study, these functions appear to be activated by CDPH application. Future investigations in tomato could clarify whether Sl3‐MMP contributes to collagen‐derived peptide perception, potentially linking this interaction to the enhanced root development we observed (Figure 1).

Despite increasing evidence of the roles played by plant MMPs in development and stress responses, their natural substrates remain largely unknown. In contrast to animal systems, where MMPs predominantly target ECM proteins (Pardo and Selman 2005), the identity of plant MMP substrates has yet to be clearly established. Plant MMPs exhibited in vitro proteolytic activity toward bovine myelin basic protein, gelatin, and synthetic peptides containing the scissile Gly‐Leu/Ile bond of collagen (McGeehan et al. ¹⁹⁹²; Maidment et al. ¹⁹⁹⁹; Delorme et al. ²⁰⁰⁰). This raises the intriguing possibility that plant MMPs may target atypical ECM‐like substrates, including collagen or collagen‐like glycoproteins introduced or produced in specific physiological or experimental contexts. To explore this hypothesis further, we performed a structural modeling analysis comparing At5‐MMP with human MMP‐1 (HsMMP‐1), a well‐characterized collagenase (Figure 2). The CBD of HsMMP‐1, spanning residues F81 to C259, was used as a backbone to model the corresponding region of At5‐MMP (Y143 to G319). Strikingly, the residues predicted to interact with the collagen fibril show a high degree of structural conservation between the two proteins. Most notably, the glutamic acid residue E200, which is critical for collagenolytic activity in HsMMP‐1 (Chung et al. 2004; Gorantla et al. ²⁰²⁴), is conserved in all five Arabidopsis thaliana MMP homologues. In At5‐MMP, the equivalent residue is E271. We further extended our homology modeling analysis to the Solanum lycopersicum homolog Sl3‐MMP. The key glutamic acid residue (E200 in HsMMP‐1) is also conserved in Sl3‐MMP. The resulting Solanum lycopersicum protein model, generated using the catalytic binding domain (CBD) of HsMMP‐1 as a structural template, showed a sequence identity of 47.27%.

FIGURE 2.

Structural alignment of At5‐MMP and HsMMP‐1 CBD. Ribbon structure of At5‐MMP (gold) modeled over the CBD of HsMMP‐1 (royal blue) in (A) the absence or (B) presence of a collagen fibril (red). The picture in (B) is zoomed in (C). HsMMP‐1 residue E200 is shown as a stick structure and clearly labeled in all figures. Zinc atoms are displayed in maroon. The image was created with UCSF ChimeraX: HsMMP‐1, F81 to C259 (PDB 4AUO); At5‐MMP Y143 to G319.

In addition to the plant MMPs, our comparative analysis highlighted another protein of interest: a plant serpin (AtSERPIN1), which displayed the second‐highest sequence identity (22%) to its human counterparts (HsSERPINH1). HsSERPINH1, also known as Hsp47 or CBP1, is a molecular chaperone localized in the lumen of the endoplasmic reticulum (ER) and is crucial for collagen maturation and secretion (Widmer et al. 2012) since it inhibits procollagen local unfolding and aggregation (Ito and Nagata 2019). While the mammal protein harbors an RDEL signaling peptide at the C‐terminus, which allows its return to the ER via the KDEL receptor (Widmer et al. 2012), the plant homolog does not possess such a sequence. Another difference between the two serpins is that AtSERPIN1 functions as an inhibitor of proteases, as most of the serpins in all kingdoms do, while HsSERPINH1 is a molecular chaperone that does not possess serine protease inhibitory activity (Hirayoshi et al. 1991). AtSERPIN1 is the best‐characterized plant serpin among the eight encoded by the Arabidopsis thaliana genome (Fluhr et al. 2012). It plays a key role in plant stress responses by interacting with various endogenous proteases. Notably, AtSERPIN1 can inactivate the cysteine protease Responsive to Dessication 21 (RD21), thereby modulating pathogen‐induced programmed cell death (Shindo et al. 2012; Lampl et al. ²⁰¹³). Besides, AtSERPIN1 inhibits the metacaspase AtMC1, a positive regulator of hypersensitive response‐associated cell death (Asqui et al. 2018). To date, no functional characterization of serpin proteins has been reported in Solanum lycopersicum . Regarding the interaction with collagen, numerous studies have identified key residues involved in binding by the human serpin HsSERPINH1, whereas no such information is currently available for its plant homolog. HsSERPINH1 has been shown to bind various collagen model peptides—even those as short as 15 amino acids—provided they form a triple helix, with binding stabilized by salt bridges and hydrophobic interactions (Ono et al. 2012; Widmer et al. ²⁰¹²). Notably, two residues critical for collagen binding in HsSERPINH1, D385 and H386, are conserved in the plant serpin AtSERPIN1, suggesting a possible evolutionary conservation of binding functionality (Widmer et al. 2012) (Figure 3).

FIGURE 3.

Structural alignment of AtSERPIN1 and HsSERPINH1 homodimer. Ribbon structure of AtSERPIN1 (violet) modeled over HsSERPINH1 (green) in (A) the absence or (B) presence of a collagen fibril (red). The picture in (B) is zoomed in (C). HsSERPINH1 residues D385 and H386 are shown as stick structures in all figures and clearly labeled in (C). The image was created with UCSF ChimeraX: HsSERPINH1, Chains Q and P, M35 to H425 (PDB 3ZHA); AtSERPIN1 V30 to H418.

4. Conclusions

In conclusion, this study offers a pioneering molecular perspective on the interaction between plant proteins and collagen‐derived peptides used in biostimulant formulations. Through an in silico approach, we identified putative plant targets—including chaperones and protease inhibitors—that are closely associated with root development and with plant responses to both abiotic and biotic stresses. These molecular interactions align with the physiological effects previously observed upon application of the collagen‐derived biostimulant. Notably, since the collagen‐derived peptides retain the polyproline II (PPII) helical conformation of native collagen (Ambrosini et al. 2022)—a structural motif also found in plant cell wall extensins (van Holst and Varner ¹⁹⁸⁴; Shpak et al. ²⁰⁰¹; Herger et al. ²⁰¹⁹)—we hypothesize that this conformation may be specifically recognized by plant proteins such as MMPs and SERPINs. This structural mimicry could underlie the observed interactions and partially explain the biostimulant's ability to modulate stress‐related pathways. Overall, our findings provide novel insights into the potential mechanisms of action of collagen‐derived biostimulants and establish a framework for future functional studies. Functional validation using Arabidopsis thaliana mutants for At5‐MMP and AtSERPIN1 represents a logical next step to confirm our hypotheses. Although such analyses are beyond the scope of this Short Communication, they are currently in progress and will help identify the molecular components mediating plant–collagen peptide interactions. These studies are expected to clarify the molecular basis of CDPH activity and support the design of more targeted and effective biostimulant formulations for sustainable agricultural applications.

Author Contributions

Z.V., A.Z., and S.A. conceived and designed the research. S.A. and A.G. directed the experiments. S.A. and M.P. performed the experiments. S.A., A.G., M.P., and T.P. analyzed and interpreted the data. S.A. and A.G. wrote the manuscript. Z.V., A.Z., and T.P. revised the manuscript. All authors contributed to the article and approved the submitted version.

Supporting information

Table S1: List of human CBPs retrieved by in silico sequence analyses. Gene name(s), protein name, length in AA, identification of the CBD domain and its respective first and last AA were reported.

Table S2: Full list of putative plant orthologues for CBDs and CBPs. Arabidopsis thaliana proteins found by homology via HHPred using as a query CBDs (ITGA2, ITGA11, DDR1, DDR2, OSCAR, GP6, LAIR, MRC, MASP, C1S, SPARC, vWF, MMP‐1, FN1) or CBPs (MPIG6B, SERPINH1, KBTBD4). Only hits with an evalue ≤ 0.001 were considered. Ref prot: human query protein employed for the HHPred search; Hit: ID from NCBI or PDB databases of the Arabidopsis thaliana proteins retrieved; Name: name of the proteins retrieved; Probability: probability that the hit is homologous to the query; e‐value: value indicating the expected number of false positives in a database search that would achieve a score equal to or better than that of this sequence match; Score: value indicating the total score obtained for an homology alignment; SS: partial score from the secondary structure comparison; Aligned cols: total number of matched columns in the query–template alignment; Target length: length in amino acids of the hit sequence; Group: protein group to which the hit protein belongs to (when identified).

Table S3: Non‐redundant list of putative plant orthologues for CBDs and CBPs. Arabidopsis thaliana proteins found by homology via HHPred using as a query CBDs (ITGA2, ITGA11, DDR1, DDR2, OSCAR, GP6, LAIR, MRC, MASP, C1S, SPARC, vWF, MMP‐1, FN1) or CBPs (MPIG6B, SERPINH1, KBTBD4). Only hits with an e‐value ≤ 0.001 were considered. Ref prot: human query protein employed for the HHPred search; Hit: ID from NCBI or PDB databases of the Arabidopsis thaliana proteins retrieved; Name: name of the proteins retrieved; Probability: probability that the hit is homologous to the query; e‐value: value indicating the expected number of false positives in a database search that would achieve a score equal to or better than that of this sequence match; Score: value indicating the total score obtained for an homology alignment; SS: partial score from the secondary structure comparison; Aligned cols: total number of matched columns in the query–template alignment; Target length: length in amino acids of the hit sequence; Group: protein group to which the hit protein belongs to (when identified).

Ambrosini, S. , Giorgetti A., Peli M., Pandolfini T., Zamboni A., and Varanini Z.. 2025. “A Structural Bridge Between Kingdoms: How Collagen‐Derived Peptides Influence Plant Stress and Growth Pathways.” Physiologia Plantarum 177, no. 5: e70512. 10.1111/ppl.70512.

Data Availability Statement

The data that support the findings of this study are available in the Supporting Information of this article.