Human splenic fibroblasts (HSFs) (ScienCell, California, USA) were maintained in Fibroblast Medium (ScienCell, USA) containing 2% fetal bovine serum (FBS), 1% fibroblast growth supplement (FGS), and 1% antibiotic solution (P/S) at 37°C in a humidified incubator with 5% CO₂. For better growth of HSFs, the culture dishes were precoated with poly-L-lysine (ScienCell, USA) to charged their surfaces, which increases cell adhesion. The precoated culture wares were left in a 37°C incubator overnight, rinsed twice with sterile water, and then added with 10 mL of the complete medium to culture the cells. The cell passage was conducted when the culture reached 95% confluency.

Gene sequences of pvtrag23 and pvtrag26 were obtained from the PlasmoDB website (https://plasmodb.org/plasmo/app; accession no. PVX_101515 and PVX_112660). Both sequences were derived from the Sal-1 strain of P. vivax; pvtrag23 and pvtrag26 were generated by DNA synthesis, and the gene fragments were then inserted into the pET30a expression vector (Tianlin Bio, Wuxi, China). This vector was added with thioredoxin and His-tag at the N- and C-terminal ends to enable easier purification and immunodetection using polyclonal antibodies against His-tag. Escherichia coli (E. coli) BL21 (DE3) pLysS cells (TransGen Biotech, Beijing, China) were used as the expression host. The proteins rPvTRAg23 and rPvTRAg26 were purified using a Ni-Sepharose column under nondenaturing conditions by a biotechnology company (YouLong Bio, Shanghai, China).

HSFs (5 × 10⁶ cells) were cultured in a 37°C incubator until the density reached 80%, pretreated with or without rPvTRAg23 and rPvTRAg26 for 48 h. Then the cells were harvested and resuspended in protein lysis buffer to extract protein. All samples were boiled and separated on a 10% SDS-PAGE, transferred to a polyvinylidene difluoride (PVDF) membrane, and blocked in blocking buffer (1 × TBST, 5% milk powder) for 2 h at room temperature. The blot was washed and incubated with primary antibodies, such as His (1:1000, Abcam, Cambridge, USA), collagen I (1:1000, Abcam, Cambridge, USA), vimentin (1:1000, Abcam, Cambridge, USA), NF-κBp65 (1:1000, CST, Danvers, USA), phospho-NF-κBp65 (1:1000, CST, Danvers, USA), FAK (1:1000, CST, Danvers, USA), phospho-FAK (1:1000, CST, Danvers, USA), p38 MAPK (1:1000, CST, Danvers, USA), phospho-p38 MAPK (1:1000, CST, Danvers, USA), and GAPDH (1:1000, Abcam, Cambridge, USA) at 4°C overnight. The blots were washed and incubated for 1 h with secondary antibody conjugated to HRP (1:5000, Abcam, Cambridge, USA). Bands were visualized by ECL detection kit (NCM, Suzhou, China), and protein expression was quantified by Image J (National Institutes of Health, MD, USA).

Five BALB/c female mice, six weeks old, were immunized with formulations containing 50 μg recombinant vimentin protein. Mice were injected abdominally with 200 μL of the formulation three times, with an interval of two weeks between each injection. The formulation consisted of protein and Freunds adjuvants (Sigma, Darmstadt, Germany). The proteins were emulsified in a constant volume of adjuvant and then injected into mice. Complete Freunds adjuvant was used for the first injection and incomplete Freunds adjuvant was used for the next two injections. Mice were euthanatized 14 days after the last immunization and serum samples were collected. Meanwhile, a total of 1.5 mg rPvTRAg23 protein was used to immunize rabbits (YouLong Bio, Shanghai, China) for the production of polyclonal antibodies.

The schizont stage-rich parasites of P. vivax were collected from malaria patients in Thailand. The slides smeared parasite infected blood were fixed with 4% paraformaldehyde for 5 min, dried, and blocked with PBS containing 5% non-fat milk at 37°C for 30 min (26). Then, slides were incubated with rabbit anti-PvTRAg23 antisera (1:50) and mouse anti-VIR antibody (1:100) for 2 h, washed three times with PBS and air-dried. Anti-VIR antibody was donated by Dr. Louis H. Miller (Laboratory of Malaria and Vector Research, National Institute of Allergy and Infectious Diseases, NIH). After primary antibody reactions, the samples were then treated with Alexa Fluor 488-conjugated donkey anti-rabbit IgG (1:500, Invitrogen, CA, USA), Alexa Fluor 568-conjugated goat anti-mouse IgG (1:500, Invitrogen CA, USA), and 4, 6-diamidino-2-phenylindole (DAPI) (1:1000, Invitrogen, CA, USA). The slides were then sealed with neutral resin, and visualized with confocal laser-scanning microscopy (Zeiss, Jena, Germany) under oil immersion. Images were edited using Adobe Photoshop CS5 (Adobe Systems, San Jose, CA, USA).

Total RNA was extracted from isolated HSFs (5 × 10⁶ cells) or pulverized spleen according to the manufacturer’s instructions (ES Science, Shanghai, China). The resulting total RNA was used for reverse transcription (RT). Genomic DNA was removed by gDNA Remover Mix, and cDNA was produced by 5 × HiFiScript RT MasterMix according to the manufacturer’s protocol (CWBIO, Beijing, China). Each sample was assayed in triplicate. RT-PCR was performed with 2 × TB Green Premix Ex Taq II (Takara, Japan) in an LightCycler480 II apparatus (Roche, USA). The thermal cycling conditions were as follows: initial denaturation at 95°C for 30 s, followed by 40 cycles of PCR at 95°C for 5 s and 60°C for 30 s, then 95°C for 5 s and 60°C for 1 min, followed by 50°C for 30 s. GAPDH was used as an internal control, and the ratio of each target gene was determined. The primer sequences are presented in Supplementary Table S1 .

Soluble collagen I levels in the cell supernatant were quantified by ELISA kit following the manufacturer’ s instructions (Abclonal, Wuhan, China). HSFs (5 × 10⁶ cells) were cultured in the presence of with rPvTRAg23 and rPvTRAg26 in 50 µg/mL for 48 h. Cell-free culture supernatant was collected and analyzed for collagen I content. Each sample was assayed in triplicate at 450 nm, and the collagen I concentration in supernatant was expressed in pg/mL.

HSFs (5 × 10⁶ cells) were harvested in clean EP tubes and stained with rabbit anti-His antibody (1:300, Abcam, Cambridge, USA) or rabbit anti-Vimentin antibody (1:300, Abcam, Cambridge, USA) at 4°C for 30 min. The stained HSFs were washed twice with 1 × PBS and then treated with Alexa Fluor 647-conjugated goat anti-rabbit IgG (1:2000, Abcam, Cambridge, USA) at 4°C for 1 h with light protection. The labeled HSFs were subjected to flow cytometry analysis using FACScan flow cytometer (Becton Dickinson, Franklin Lakes, NJ, USA). For cells that needed to block the vimentin receptors, HSFs were incubated with mouse anti-vimentin antisera (1:10) for 2 h after collection, washed twice with 1 × PBS, and performed the above experiments. The percentage of His-positive or vimentin-positive cells was determined using the FlowJo software package.

Five six-week-old wild BALB/c mice weighing 18–22 g were purchased from Cavens Laboratory Animal (Changzhou, China). Strains of Plasmodium yoelii 17XNL were obtained from the Jiangsu Institute of Parasitic Disease in Wuxi, China. Mice were randomly grouped and intraperitoneally injected with 1 × 10⁵ iRBCs in 200 μL of saline, prepared from frozen stock. Animal precautions and experimental processes were approved by Jiangnan University’s Animal Ethics Committee.

293T cells were maintained in DMEM supplemented with 10% FBS (Gemini Bio-Products), 100 U/mL penicillin, and 100 μg/mL streptomycin in a humidified atmosphere containing 5% CO2 at 37°C. For immunoprecipitation, 293T cells were transfected with Lipofectamine (Yeasen, Shanghai, China) according to the protocol provided by the manufacturer and cultivated for 24 h. Briefly, cells were grown on 6 cm plates, transfected with 5 μg pEGFP-C1-Vimentin eukaryotic plasmid and 5 μg pEGFP-C1-Vector for 20 min, then returned to DMEM for 24 h. The transfection efficiency was observed under a Nikon Inverted Fluorescent Microscope (Nikon, Tokyo, Japan). After the cells were lysed by NP40 cell lysate (Beyotime, Shanghai, China), the expression of vimentin was verified by western blot.

The His-tagged rPvTRAg23 protein was immobilized on Ni-NTA agarose (QIAGEN, Germany) to generate an affinity column. Briefly, 200 μg of purified His-tagged rPvTRAg23 was added to 500 μL of Ni-NTA-agarose, which was washed in cold 1 × PBS twice and 1 mL of equilibration buffer (50 mM NaH₂PO₄, 300 mM NaCl, 20 mM imidazole, adjust pH to 8.0 by adding NaOH) and incubated overnight at 4°C in a silent mixer. The Ni-NTA-agarose/rPvTRAg23 protein columns were washed in cold equilibration buffer and added with 200 μg of HSF membrane proteins, which were extracted using the Mem-PER Plus Membrane Protein Extractiont Kit (Thermo, Waltham, MA, USA) and incubated at 4°C overnight. After washing the beads with equilibration buffer three times, the proteins were eluted with 500 μL of the elution buffer (50 mM NaH₂PO₄, 300 mM NaCl, 250 mM imidazole, adjust pH to 8.0 by adding NaOH) and size fractioned by sodium dodecyl sulfate polyacrylamide gel electrophoresis.

To evaluate the interaction between PvTRAg23 and vimentin, His-pulldown assays were carried out. rPvTRAg23 protein (200 μg) were incubated with 500 μL anti-His Ni-NTA-agarose (QIAGEN, Germany) for 3 h at 4°C to immobilize on beads. The beads were collected using a magnetic separator and incubated separately with 500 μL lysate of 293T transfected pEGFP-C1-vimentin overnight at 4°C. After washing the beads with 20 volumes of equilibration buffer three times, the proteins were eluted with 500 μL of the elution buffer. Take 100 μL eluent and resuspended in 25 μL of 5 × SDS reducing loading buffer and boiled for 8 min. Supernatant was used for western blot analysis.

The rPvTRAg23-binding proteins were visualized by silver staining using the Fast Silver Stain Kit (Beyotime, Shanghai, China) according to the manufacturer’s instructions. The protein bands different from rPvTRAg23 in silver-stained gel were excised and cut into 1 mm pieces. Proteins were identified by mass spectrometry (Shanghai Applied Protein Technology, Shanghai, China).

For proteomics of the spleens, the infected and non-infected mouse spleens were extracted under sterile conditions. The spleens were ground thoroughly, lysed in lysis buffer, inactivated by autoclave sterilization, and processed for quantitative proteomic analysis (Jingjie PTM BioLab, Hangzhou, China).

Collagen fiber specially type I in the spleen was assessed by Sirius Red tissue staining on paraffin-embedded sections. After the removal of paraffin with xylene, different concentrations of ethanol (100%, 90%, 80%, and 70%), and water, the tissue sections were incubated in Sirius Red reagent (Solarbio, Beijing, China) for 1 h, followed by washing with 0.5% acidic water, and sealed with neutral resin. All the stained sections were observed under ordinary optical microscope (OLYMPUS, Tokyo, Japan).

Data with normal distribution from more than two groups were compared using one‐way ANOVA, while differences between groups were analyzed by Student’s t-test. All statistical analyses were performed using GraphPad Prism 8 software (GraphPad, California, USA). P-value<0.05 was considered statistically significant for differences (*P<0.05, **P<0.01, ***P<0.001).

The full-length cDNA of the PVX_101515 gene encodes the PvTRAg23 protein of 321 amino acids (aa) with molecular weight of 39.951 kDa and a predicted pI of 9.51, which contains a transmembrane region at residues 36–55 aa and a special tryptophan-rich domain residue at 95-311 aa. To elucidate the role of PvTRAg23, we expressed partial PVX_101515 cDNA (encoding the tryptophan-rich domain) in E. coli. The recombinant PvTRAg23 (rPvTRAg23) of 36 kDa was purified to verify its role. Another recombinant protein rPvTRAg26 with high antigenicity and immunogenicity (27) was expressed for comparison ( Figure 1A ). Studies on PvTRAgs have shown that many members of PvTRAgs are localized on the erythrocyte membrane. They could be exported through parasitophorous vacuolar (PV) or other ways to interact with host cells (26–29). To determine the location of rPvTRAg23, co-localization was carried out using anti-VIR antibody. VIR has previously been shown to be an export protein that could be exported outside the membrane of iRBCs and interact with host cells (19). Results of IFA assay on infected erythrocytes showed that rPvTRAg23 signal (green) partially merged with VIR (red) during the early or trophozoite stage of the parasite ( Figure 1B ), indicating that rPvTRAg23 could be exported to the surface of iRBC and might interact with host cells.

Spleen section from a P. vivax-infected patient showed a large number of malaria parasites and hemozoin (30). Meanwhile, after Plasmodium infection, destruction of the segregated lymphoid areas of the white pulp and remodeling of the red pulp were found in the host spleens (6), suggesting that the clearance function of the spleen might be affected during this period. Collagen I secreted by SF is an important component in maintaining spleen struction and functions (10, 11), while VIR protein, expressed by the spleen-dependent gene of P. vivax, has been proved to interact with HSFs to achieve immune evasion (17). Thus, as PvTRAg23 was also expressed by spleen-dependent gene, it might play a role in this process. To investigate the function of PvTRAg23 on HSFs, the cells were cultured with different concentrations of recombinant protein for 48 h, and the levels of collagen I in cells were detected. With the increase of rPvTRAg23 concentration, the protein expression of collagen I in HSFs decreased, while control rPvTRAg26-stimulated cells had no significant effect on it ( Figure 2A ), showing that rPvTRAg23 affected the collagen I content in HSFs specifically and in a concentration-dependent manner. Similar decreases in the mRNA levels of collagen I alpha 1 chain (COL1A1) and collagen I alpha 2 chain (COL1A2) were observed in response to protein stimulation ( Figure 2B ). Given that collagen I is the main component of ECM, the level of collagen I in the supernatant of HSF culture medium was measured after 48 h of protein stimulation to verify whether there is a change in collagen outside the cells. The ELISA results showed that the collagen I level in the supernatant of rPvTRAg23-stimulated HSFs culture medium decreased significantly compared with that in untreated group ( Figure 2C ). These data infered that rPvTRAg23 reduced the content of collagen I production by HSFs and thus affected its extracellular secretion.

Many effects of Plasmodium on host cells are mediated by binding (31–33). As PvTRAg23 was an export protein, we analyzed its interaction with HSF. Flow cytometry analysis showed a high expression rate of His on HSFs in a dose dependent way of rPvTRAg23 simulation; meanwhile, no difference was found in the rPvTRAg26-stimulated HSFs group ( Figure 3A ). This finding implied that rPvTRAg23 could specifically bind to HSFs. Thus, we expect to identify receptors for binding and explore whether it relates to collagen I reduction. The comparison of rPvTRAg23, HSF membrane protein, and the eluate after co-incubation showed differential bands when visualized by silver stain ( Figure 3B ). The strips were cut off and analyzed by mass spectrum. Among the identified candidate proteins, vimentin yielded the highest score ( Supplementary Table S2 ), suggesting that it was likely to be the binding receptor of rPvTRAg23 and HSFs. Hence, 293T cells were transfected with pEGFP-C1-vimentin eukaryotic plasmid and the His-pulldown experiment was performed after verifying the expression of vimentin. As shown in Figure 3C , obvious interactions could be observed between rPvTRAg23 and vimentin. So we blocked the vimentin receptors on HSFs by anti-vimentin antisera prior to protein stimulation and performed the same test again. The flow cytometry results showed that the His expression rate decreased almost as much as that of the unstimulated group, similar to the western blot data ( Figure 3D ). Hence, vimentin was indeed a binding receptor between rPvTRAg23 and HSFs. Furthermore, HSFs were stimulated with rPvTRAg23 after blocking vimentin receptors, and collagen I level was detected. The content of collagen I in the cells decreased regardless of whether the blocking of the vimentin receptors ( Figure 3E ). In contrast to expectations, the binding of rPvTRAg23 to HSFs vimentin did not affect the reduction of collagen I in the cells. As such, we need to identify its mediated pathway from other perspectives.

Previous studies showed that NF-κBp65, FAK, and P38 MAPK signaling pathways can regulate collagen changes in cells and thus cause different reactions in the body (34–36). Thus, we explored whether rPvTRAg23 decreases the collagen I expression in HSFs through these signaling pathways. Here, HSFs were stimulated with rPvTRAg23, and each signal path state in different time periods was detected. The western blot results showed that the signal intensities of the three pathways increased with prolonged stimulus time ( Figure 4A ). This finding suggested that rPvTRAg23 activated all the detected signaling pathways of HSFs. To determine which pathway mediates the reduction of collagen I in HSFs, we treated the cells with corresponding inhibitors, namely, Bay 11-7082, SB203580, and TAE226. After 1 h of pretreatment, the HSFs were stimulated with rPvTRAg23 and continued to culture for 48 h. The decline in the collagen I level was significantly improved after the inhibition of the NF-κBp65 pathway compared with the control group, while the level continued to decrease even after the inhibition of other pathways ( Figure 4B ). Hence, the NF-κBp65, FAK, and P38 MAPK pathways of HSFs were activated in response to rPvTRAg23 stimulation, but only the NF-κBp65 pathway mediated intracellular collagen reduction.

The activation of the NF-κBp65 pathway is often accompanied by the increase in the levels of inflammatory cytokines, such as interleukin-1β (IL-1β), interleukin-6 (IL-6), and tumor necrosis factor-α (TNF-α) (37, 38). In this regard, we hypothesized that these cytokines might directly regulate collagen I expression in HSFs. The levels of relevant cytokines in HSFs after protein stimulation were measured. The rPvTRAg23-stimulated group showed significant increases in the levels of IL-1β, IL-6, and TNF-α compared with the other groups ( Figure 4C ). To verify the role of cytokines on HSFs, we added different concentrations of commercial cytokines IL-1β, IL-6, and TNF-α to cells and cultured them for 48 h. Collagen I level was subsequently measured. With increasing cytokine concentration, the collagen I protein expression was reduced ( Figure 4D ), similar to the mRNA level ( Figure 4E ). Hence, these cytokines inhibited collagen I expression in HSFs in a concentration-dependent manner. To further verify the relationship between protein regulation of collagen I and inflammatory cytokines, HSFs were stimulated by rPvTRAg23 with time gradient, and the changes of cytokines and collagen I were detected at the same time. The results showed that with the increase of stimulation time, the level of inflammatory cytokines increased while collagen I decreased, and the two were negatively correlated ( Figure 4F ), showed that the regulation of collagen I in HSFs by rPvTRAg23 had a certain relationship with inflammatory response. Overall, rPvTRAg23 stimulation activated the NF-κBp65 pathway in HSFs and thus increased the expression of inflammatory cytokines IL-1β, IL-6, and TNF-α, thereby inhibiting collagen I production.

Previous studies have shown that parasite deposition and circulatory changes in the spleen of mice infected with P. yoelii 17XNL are similar to those of patients with P. vivax (30, 39). Therefore, to investigate whether the collagen reduction occurred in the course of Plasmodium infection in vivo, we explored physiological ECM changes in mice models of infection with the P. yoelii 17XNL strain. Then, the spleens were analyzed by LC-MS/MS. The proteomic analysis showed that compared with non-infected mice, collagen VI, but not collagen I, was markedly reduced in the spleens of P. yoelii 17XNL-infected mice ( Supplementary Table S3 ). Collagen VI is a protein widely distributed in a variety of tissues and mediates the formation of fibronectin in the ECM of cultured fibroblasts (40). GO annotation analysis was performed to further understand the role of collagen VI. The result revealed that it was involved in cell and ECM composition, immune response, and cell adhesion ( Supplementary Table S4 ), and had similar functions to that of collagen I. To explore whether collagen I changes, we detected the collagen mRNA levels in the spleens of mice. Compared with the non-infected group, COL1A1 and COL1A2 in the spleen of P. yoelii 17XNL-infected mice had a downward trend, while the decline in collagen VI alpha 1 chain (COL6A1), collagen VI alpha 2 chain (COL6A2), and collagen VI alpha 4 chain (COL6A4) were more obvious ( Figure 5A ). The spleen sections of mice were stained with sirius red to visualize collagen I. The expression of collagen fiber in the white pulp of infected mice spleen was greatly reduced under ordinary optical microscope, accompanied by a large amount of malaria pigment deposition ( Figure 5B ). Overall, the collagen I level in the spleen of Plasmodium-infected mice was indeed reduced. Basing on the possible protein interaction between collagen I and collagen VI, we conducted STRING analysis (https://string-db.org/). The interaction maps of COL6A1, COL6A2, and COL6A4 showed that they interacted with COL1A1 and COL1A2 and co-regulated the components of ECM and the adhesion of cells ( Figure 5C ). Meanwhile, they were highly interactive with proline hydroxylase, thereby maintaining collagen structure stability and adhesion receptors, such as integrins.