Lymphoid Organ Proteomes Identify Therapeutic Efficacy Biomarkers Following the Intracavitary Administration of Curcumin in a Highly Invasive Rat Model of Peritoneal Mesothelioma

This study aimed to identify the proteomic changes produced by curcumin treatment following stimulation of the host immune system in a rat model of malignant mesothelioma. We analyzed the proteomes of secondary lymphoid organs from four normal rats, four untreated tumor-bearing rats, and four tumor-bearing rats receiving repeated intraperitoneal administrations of curcumin. Cross-comparing proteome analyses of histological sections of the spleen from the three groups first identified a list of eighty-three biomarkers of interest, thirteen of which corresponded to proteins already reported in the literature and involved in the anticancer therapeutic effects of curcumin. In a second step, comparing these data with proteomic analyses of histological sections of mesenteric lymph nodes revealed eight common biomarkers showing a similar pattern of changes in both lymphoid organs. Additional findings included a partial reduction of the increase in spleen-circulating biomarkers, a decrease in C-reactive protein and complement C3 in the spleen and lymph nodes, and an increase in lymph node purine nucleoside phosphorylase previously associated with liver immunodeficiency. Our results suggest some protein abundance changes could be related to the systemic, distant non-target antitumor effects produced by this phytochemical.


Introduction
Curcumin belongs to a category of bioactive compounds with the propensity to target multiple signaling pathways that are crucial for tumor development [1]. This molecule provides a diversity of interactions explained by its binding to numerous proteins via a unique symmetrical structure combining a central keto-enol tautomer, flexible a,bunsaturated b-diketone linkers, and terminal o-methoxyphenolic groups [2]. In line with the concept of polypharmacology, its mechanism of action suggests that it modulates multiple sensitive nodes belonging to a network of interacting targets [3]. This new concept now offers a strategy to overcome drug resistance [4]. In the case of colon cancer, curcumin represents an epigenetic modulator, cancer stem cell suppressor, and potent autophagy modulator. Trials have demonstrated clinical evidence in prevention and treatment, including synergistic effects on the efficacy of current cancer drugs [5].
Curcumin's multifaceted effects have already contributed to addressing the chemoresistance and radioresistance commonly found in aggressive cancers such as glioblastoma (GBM) [6]. To circumvent its poor oral absorption following the inclusion of curcumin in the diet, other strategies have been proposed to treat experimental models of GBM or malignant mesothelioma (MM), e.g., intraperitoneal injection [6,7]. Based on proteomic analyses of the liver from curcumin-treated tumor-bearing rats vs. untreated tumor-bearing rats and normal rats, we previously identified quantitative changes in a set of biomarkers representing both determinants of liver colonization and therapeutic targets [8].
It has been reported that an important factor explaining curcumin's polypharmacology could be the diversity of its degradation products [9]. In our previous studies, successive intraperitoneal administrations of curcumin produced large areas of necrotic/apoptotic tumor cells while inducing an influx of activated CD8+ T cells at the periphery of residual tumors [7] and in the liver [8]. This immune response suggests some common features with the recently described "abscopal effect" (meaning an out-of-field systemic antitumor effect) observed in melanoma patients receiving radiation therapy [10]. Among lymphoid organs, emphasis was given a few years ago to the spleen's crucial role in regulating immune responses both locally and in the whole body [11]. Therefore, to give insight into the possible biomarkers associated with this effect in curcumin-treated rats, we first investigated the quantitative changes observed in the spleen proteome compared with normal and untreated tumor-bearing rats in our experimental mM model. Cross-comparing spleen proteomic analyses identified eighty-three biomarkers of interest, thirteen of which corresponded to proteins already reported in the literature and involved in curcumin's therapeutic effects against different cancers. In a second step, cross-comparing spleen data with mesenteric lymph node analyses revealed eight previously undescribed common biomarkers affected by curcumin treatment, which may be related to an abscopal effect.

Histological Features of the Immune Response in the Spleen and in Residual Tumors of Curcumin-Treated Rats
Three groups of four male rats at four months of age (to mimic the physiological conditions of mM carcinogenesis in adult humans) were used for the study, the experimental design being summarized in Figure 1A. The first group (G1) consisted of normal rats. The second (G2) and third groups (G3) corresponded to rats given a single injection of 3 × 10 6 M5-T1 tumor cells (tumor challenge) at day 0. G1 and G2 rats received four successive injections of DMSO (50% in saline, i.p., as the vehicle), while G3 rats had four injections of curcumin (1.5 mg/kg in 50% DMSO plus saline) at days 7, 9, 11, and 14. These 8 rats were euthanized at day 16, and the tumors were collected (for treated rats, mostly residual tumors present in the peritoneal cavity). For all twelve rats, spleen and mesenteric lymph nodes were additionally sampled. For both the averaged total tumor mass and spleen mass, a significant decrease was observed in treated (G3) vs. untreated (G2) rats. General histological views of tumor masses (G2) and residual tumors (G3) collected in each rat are shown in Supplementary Figure S1.
In previous studies, we observed an immune response directed against M5-T1 tumor cells in G3 rats following multiple i.p. curcumin injections, which could be explained by the fact that numerous tumor cells experienced a necrosis/apoptosis process in vivo [7]. Therefore, the decrease in spleen mass observed in G3 vs. G2 rats led us to investigate the differential histological features of the spleen between the three groups ( Figure 1B). General views of histological sections of the spleen after HPS staining first revealed a decreased contrast between the white and red pulps of G2 rats, as compared to G1 and G3 rats. In parallel, numerous tumor cells attached to the spleen capsule were specifically present in G2 rats. In a second step, high-magnification views of the white pulp in G3 rats showed the presence of a specific subpopulation of immune cells in the lymphoid sheath surrounding the central artery, which was not observed in G1 and G2 rats. Subsequently, a comparative analysis of the spleen from the G2 and G3 groups after immunostaining with anti-CD8 monoclonal antibody revealed this population corresponded to the CD8+ T cells specifically observed in G3 ( Figure 1B, bottom rank, yellow arrows), which were larger and more intensely stained than their G2 counterparts. In a third step, high-magnification views of these areas confirmed that G2 rats were specifically characterized by smaller, more heterogeneous, and irregularly shaped cells, while the lymphoid sheath cells in G3 rats were nicely individualized ( Figure 1C, top rank). Examination of red pulps also revealed the same specific feature ( Figure 1C, bottom rank, arrow). Finally, residual tumor masses collected in the peritoneal cavity of G3 rats were also characterized by massive infiltration of their surrounding environment and connective tissue separating muscle fibers with activated lymphocytes presenting the same morphology (as previously observed in the liver [8]), a finding never obtained in G2 rats ( Figure 1D).
surrounding the central artery, which was not observed in G1 and G2 rats. Subsequently, a comparative analysis of the spleen from the G2 and G3 groups after immunostaining with anti-CD8 monoclonal antibody revealed this population corresponded to the CD8+ T cells specifically observed in G3 ( Figure 1B, bottom rank, yellow arrows), which were larger and more intensely stained than their G2 counterparts. In a third step, high-magnification views of these areas confirmed that G2 rats were specifically characterized by smaller, more heterogeneous, and irregularly shaped cells, while the lymphoid sheath cells in G3 rats were nicely individualized ( Figure 1C, top rank). Examination of red pulps also revealed the same specific feature ( Figure 1C, bottom rank, arrow). Finally, residual tumor masses collected in the peritoneal cavity of G3 rats were also characterized by massive infiltration of their surrounding environment and connective tissue separating muscle fibers with activated lymphocytes presenting the same morphology (as previously ob-  (G3 group). Top rank, general views of the spleen from one representative rat of each group (HPS staining, ×15, the scale bar represents 1 mm). Middle rank, anti CD8+ IHC staining, comparison of G3 (right) vs. G2 rat (left), ×50, the scale bar represents 500 µm. Bottom rank, magnification of a lymphoid nodule, anti CD8+ IHC staining, showing a significant concentration of CD8+ T cells into the lymphoid sheath surrounding the central artery (right), and corresponding HPS staining (left) revealing the presence of bigger, densely stained lymphocytes in the corresponding area (yellow arrows), ×200, the scale bar represents 100 µm. (C) Highmagnification views (HPS staining) of lymphoid sheaths (top rank) and red pulps (bottom rank, the yellow arrow points to a small cluster of large lymphocytes), the scale bar represents 25 µm. (D) Comparison of histological sections (HPS staining) of M5-T1 tumors, G3 (curcumin-treated rat, top rank) vs. G2 group (untreated rat, bottom rank). Top, general view (left, the scale bar represents 1 mm), and magnifications of two areas illustrated by the yellow rectangles showing clusters of immune cells infiltrating the connective tissue separating muscle fibers (insert, the scale bar represents 50 µm) and the external part of the tumor (right, the scale bar represents 25 µm). Bottom, general view (the scale bar represents 1 mm), and magnification of a large area (the scale bar represents 100 µm) showing numerous tumor cells invading the muscle free of any lymphocyte cluster.

Spleen Biomarkers of Tumor Progression and Curcumin-Induced Immune Response
We detected 1483 proteins in each spleen sample analyzed by SWATH-MS. The number of proteins exhibiting significant abundance differences (p < 0.05) in G2 vs. G1, G3 vs. G2, and G3 vs. G1 were 285, 192, and 250, respectively. Cross-comparing the three files led to a list of 83 proteins (Table 1) common to the 285 proteins found in G2 vs. G1 and the 192 proteins found in G3 vs. G2, of which 60 proteins were excluded from the 250 proteins found in G3 vs. G1 (satisfying the condition p < 0.05).

Spleen Biomarkers of Tumor Progression and Curcumin-Induced Immune Response
We detected 1483 proteins in each spleen sample analyzed by SWATH-MS. The number of proteins exhibiting significant abundance differences (p < 0.05) in G2 vs. G1, G3 vs. G2, and G3 vs. G1 were 285, 192, and 250, respectively. Cross-comparing the three files led to a list of 83 proteins (Table 1) common to the 285 proteins found in G2 vs. G1 and the 192 proteins found in G3 vs. G2, of which 60 proteins were excluded from the 250 proteins found in G3 vs. G1 (satisfying the condition p < 0.05). Table 1. List of spleen proteins exhibiting significant abundance changes (p < 0.05) common to the two comparisons: group 2 (G2, untreated tumor-bearing rats) vs. group 1 (G1, normal rats), and group 3 (G3, tumor-bearing rats treated with curcumin) vs. group 2. Non-significant differences for G3 vs. G1 = ns (p > 0.05). Gene names (in italics) are given for Rattus norvegicus. ↑ increase, ↓ decrease.

AACS
Hk3 Ifitm3 Interferon-induced transmembrane protein 3 Prkar1a cAMP-dependent protein kinase type I-alpha regulatory subunit Lrrc59 Leucine-rich repeat-containing protein 59 Mapk3 Mitogen-activated protein kinase 3 Myl6 Myosin light polypeptide 6 ↑ p = 0.0052 Pacsin2 Protein kinase C and casein substrate in neurons 2 protein Pdcd4 Programmed cell death protein 4 P4hb Pdlim5 PDZ and LIM domain protein 5 ↑ p = 0.03056 Phldb1 Pleckstrin homolog-like domain family B member 1 Psmb10 Proteasome subunit beta type-10 Hnrnpa2b1 Heterogeneous nuclear ribonucleoprotein s A2/B1 Tagln2 Uso1 General vesicular transport factor p115 Wipf1 WAS/WASL-interacting protein family member 1 Arf3 Alad Delta-aminolevulinic acid dehydratase Nap1l1 Nucleosome assembly protein 1-like 1 Hspb1 Heat shock protein beta-1 ↑ p = 9.81 × 10 −5 Copper transport protein ATOX1 Band 3 anion transport protein ↓ p = 0.04023 Gys2 Hspe1 10 kDa heat shock protein, mitochondrial Among the first group of proteins, 55 exhibited an increase (including 4 ribosomal proteins) and 5 a decrease linked to tumor progression, which returned to normal values after curcumin treatment (Table 1). A second group of 14 proteins was also common to G2 vs. G1 and G3 vs. G2 (still evolving in opposite directions), while showing significant abundance changes (p < 0.05) between G3 and G1, including 12 presenting an increase and 2 a decrease. Finally, a third group of nine proteins presented abundance changes evolving in the same direction in G2 vs. G1 and G3 vs. G2 (with p < 0.05 between G3 and G1), including two showing an increase while seven a decrease (Table 1). Figure 2A shows the main subcellular and extracellular locations of the 83 proteins listed in Table 1. A gene ontology analysis of upregulated proteins, conducted with DAVID Bioinformatics Resources 6.8 (https://david.ncifcrf.gov) (accessed on 15 July 2021) revealed two main charts: (1). proteins involved in cell-cell adhesion (with a p value of 5.5 × 10 −3 ), and (2). Extracellular Matrix (ECM) proteins (with a p value of 8.5 × 10 −4 ). Downregulated proteins did not provide any information due to their restricted number. Moreover, a functional analysis of known and predicted protein-protein interactions for the 55 upregulated proteins, using the STRING Database (https://string-db.org) (accessed on 15 July 2021), showed 31 were associated in a single network, with 2 others being associated separately while 22 presented no interactions ( Figure 2B).
We next analyzed which of the eighty-three proteins listed in Table 1 were already reported as being associated with curcumin's therapeutic effects in the literature. We found thirteen proteins reported in PubMed, three of which were cited more than four times over a 25-year period and encoded by Ahr, Pdcd4, and Rhoa ( Table 2). The quantitative changes we observed in the spleen for these three proteins are illustrated in Figure 3A. For PDCD4, immunohistochemical staining of the spleen also revealed the presence of clusters of densely stained PDCD4+ cells in the lymphoid sheaths surrounding the central arteries of the splenic nodules in G2 rats, which were rarely found in G3 rats ( Figure 3B,C).  Table 1. Each protein is identified by the gene (in italics) encoding its expression. Names in black and bold represent the first group of proteins for which a p value < 0.05 was observed in both G2 vs. G1 and G3 vs. G2, with abundance changes evolving in opposite directions, respectively (arrows indicating changes observed in G2 vs. G1, ↑ for increase, ↓ for decrease), while differences were not significant in G3 vs. G1 (p value > 0.05). Names between brackets correspond to the second group of proteins satisfying the same conditions in G2 vs. G1 and G3 vs. G2, while showing significant differences in G3 vs. G1 (p value < 0.05), written in brown for increase and blue for decrease. Proteins of the third group (changes evolving in the same directions in both G2 vs. G1 and G3 vs. We next analyzed which of the eighty-three proteins listed in Table 1 were already reported as being associated with curcumin's therapeutic effects in the literature. We found thirteen proteins reported in PubMed, three of which were cited more than four  Table 1. Each protein is identified by the gene (in italics) encoding its expression. Names in black and bold represent the first group of proteins for which a p value < 0.05 was observed in both G2 vs. G1 and G3 vs. G2, with abundance changes evolving in opposite directions, respectively (arrows indicating changes observed in G2 vs. G1, ↑ for increase, ↓ for decrease), while differences were not significant in G3 vs. G1 (p value > 0.05). Names between brackets correspond to the second group of proteins satisfying the same conditions in G2 vs. G1 and G3 vs. G2, while showing significant differences in G3 vs. G1 (p value < 0.05), written in brown for increase and blue for decrease. Proteins of the third group (changes evolving in the same directions in both G2 vs. G1 and G3 vs.   Table 1) reported in the literature as being associated with curcumin's therapeutic effects.

Common and Specific Biomarkers of Curcumin-Induced Effects in Spleen and Lymph Nodes
We next determined which of the eighty-three biomarkers identified in Table 1 exhibited similar profiles in another lymphoid organ located at distance from the initial place of M5-T1 tumor development, the omentum, which has an anatomical relationship with the spleen. Therefore, we investigated the proteome of mesenteric lymph nodes, as these are numerous large masses allowing proteomic analyses. Representative examples of the histological features of lymph nodes from each group of rats are given in Supplementary Figure S2. We detected 1498 proteins in each lymph node sample analyzed by SWATH-MS. The number of proteins exhibiting significant abundance changes (p < 0.05) in G2 vs. G1, G3 vs. G2, and G3 vs. G1 were 415, 256, and 188, respectively.
In a first step, cross-comparing the proteins not common to these lists and Table 1 led us to define a repertoire of specific spleen biomarkers. The proteins exhibiting the largest abundance changes in this organ are illustrated in Figure 4.

Common and Specific Biomarkers of Curcumin-Induced Effects in Spleen and Lymph Nodes
We next determined which of the eighty-three biomarkers identified in Table 1 exhibited similar profiles in another lymphoid organ located at distance from the initial place of M5-T1 tumor development, the omentum, which has an anatomical relationship with the spleen. Therefore, we investigated the proteome of mesenteric lymph nodes, as these are numerous large masses allowing proteomic analyses. Representative examples of the histological features of lymph nodes from each group of rats are given in supplementary Figure S2. We detected 1498 proteins in each lymph node sample analyzed by SWATH-MS. The number of proteins exhibiting significant abundance changes (p < 0.05) in G2 vs. G1, G3 vs. G2, and G3 vs. G1 were 415, 256, and 188, respectively.
In a first step, cross-comparing the proteins not common to these lists and Table 1 led us to define a repertoire of specific spleen biomarkers. The proteins exhibiting the largest abundance changes in this organ are illustrated in Figure 4. In a second step, eight proteins showing similar profiles of abundance changes in spleen and mesenteric lymph nodes (common to the lists G2 vs. G1 and G3 vs. G2 above and to Table 1) were identified. Seven of these proteins were excluded from the 188 proteins found in G3 vs. G1 (satisfying the condition p > 0.05), encoded by Fubp1, Hnrnpa2b1, Hsd17b10, Mybbp1a, Psmb10, Purb, and Wipf1. The protein encoded by Impa1 was the only In a second step, eight proteins showing similar profiles of abundance changes in spleen and mesenteric lymph nodes (common to the lists G2 vs. G1 and G3 vs. G2 above and to Table 1) were identified. Seven of these proteins were excluded from the 188 proteins found in G3 vs. G1 (satisfying the condition p > 0.05), encoded by Fubp1, Hnrnpa2b1, Hsd17b10, Mybbp1a, Psmb10, Purb, and Wipf1. The protein encoded by Impa1 was the only one satisfying the condition p < 0.05 for the comparison G3 vs. G1. Two additional proteins showed tendencies for the same, encoded by Cpne1 and Mri1 genes (Supplementary Figure S3). Figure 5 shows the comparison of abundance changes for both spleen and lymph nodes for these eight proteins.  Figure S3). Figure 5 shows the comparison of abundance changes for both spleen and lymph nodes for these eight proteins. In a third step, the analysis of specific biomarkers of curcumin-induced effects in lymph nodes identified a single protein, purine nucleoside phosphorylase, encoded by the Pnp gene ( Figure 6A). This protein exhibited the same pattern of abundance changes as previously observed in the liver of the same rats [8]. Moreover, immunohistochemical staining of lymph nodes revealed the presence of higher stained PNPH+ cells in the In a third step, the analysis of specific biomarkers of curcumin-induced effects in lymph nodes identified a single protein, purine nucleoside phosphorylase, encoded by the Pnp gene ( Figure 6A). This protein exhibited the same pattern of abundance changes as previously observed in the liver of the same rats [8]. Moreover, immunohistochemical staining of lymph nodes revealed the presence of higher stained PNPH+ cells in the medullary sinus in G3 vs. G2 rats ( Figure 6B), and of PNPH+ cells in the two parts of the cortex of lymphatic nodules in G3 rats, which were absent in G2 rats ( Figure 6C). medullary sinus in G3 vs. G2 rats ( Figure 6B), and of PNPH+ cells in the two parts of the cortex of lymphatic nodules in G3 rats, which were absent in G2 rats ( Figure 6C).

Tumor Progression and Curcumin-Induced Effects on Plasma Biomarkers
Among the second group of biomarkers listed in Table 1, in addition to complement C9, for which the increase observed in the spleen and induced by tumor progression was

Tumor Progression and Curcumin-Induced Effects on Plasma Biomarkers
Among the second group of biomarkers listed in Table 1, in addition to complement C9, for which the increase observed in the spleen and induced by tumor progression was completely reversed following curcumin treatment (Figure 4), seven other plasma proteins exhibited a rise in abundance that was partially reversed in G3 rats ( Figure 7A). These included five positive acute-phase proteins (encoded by Cp, C3, Crp, Fga, and Fgb genes) for which elevated plasma levels are generally associated with inflammation. The two others were hemopexin (encoded by Hpx) involved in the scavenging of heme, and fibronectin (encoded by Fn1), a component of the extracellular matrix playing a major role in cell adhesion and migration. In lymph nodes, the only protein showing an evolution comparable to that observed in the spleen was the C-reactive protein ( Figure 7B), a decreased level in complement C3 also being induced after curcumin treatment. completely reversed following curcumin treatment (Figure 4), seven other plasma proteins exhibited a rise in abundance that was partially reversed in G3 rats ( Figure 7A). These included five positive acute-phase proteins (encoded by Cp, C3, Crp, Fga, and Fgb genes) for which elevated plasma levels are generally associated with inflammation. The two others were hemopexin (encoded by Hpx) involved in the scavenging of heme, and fibronectin (encoded by Fn1), a component of the extracellular matrix playing a major role in cell adhesion and migration. In lymph nodes, the only protein showing an evolution comparable to that observed in the spleen was the C-reactive protein ( Figure 7B), a decreased level in complement C3 also being induced after curcumin treatment. . p values of the comparison between groups are indicated at the top of the bars. Red (light red representing the partial decrease between "Tumor + curcumin" = G3 vs. "Tumor" = G2) and blue bars representing the significant decrease relative to (Control = G1), respectively. ns and dotted line = non-significant difference.

Discussion
Curcumin's effects on all stages of tumorigenesis are well established. However, although its action on multiple cell-signaling pathways has been the subject of numerous studies on cancer cells in vitro, several questions remain on its complex mechanisms of action in vivo. To shed light on this point, we used a rat model of aggressive peritoneal mesothelioma [7] and an established treatment procedure with curcumin injected intraperitoneally to determine the secondary lymphoid organs' implication in the immune response induced, and to further identify potential associated biomarkers [8]. Taverna and . p values of the comparison between groups are indicated at the top of the bars. Red (light red representing the partial decrease between "Tumor + curcumin" = G3 vs. "Tumor" = G2) and blue bars representing the significant decrease relative to (Control = G1), respectively. ns and dotted line = non-significant difference.

Discussion
Curcumin's effects on all stages of tumorigenesis are well established. However, although its action on multiple cell-signaling pathways has been the subject of numerous studies on cancer cells in vitro, several questions remain on its complex mechanisms of action in vivo. To shed light on this point, we used a rat model of aggressive peritoneal mesothelioma [7] and an established treatment procedure with curcumin injected intraperitoneally to determine the secondary lymphoid organs' implication in the immune response induced, and to further identify potential associated biomarkers [8]. Taverna and colleagues previously reported that innovative quantitative proteomic techniques such as SWATH-MS helped reveal how curcumin modulated the composition of exosomes released by cancer cells, making it possible to reverse their pro-angiogenic activity [42]. As reviewed by Bronte and Pittet, investigations on mouse spleen functions have unveiled a wider role than previously expected, especially in systemic regulation of immunity [11]. In line with its crucial implication in immune responses, which is deleterious to the host in metastatic cancers, this organ was shown to represent the site of accumulation of immature myeloid-derived suppressor cells (MDSCs) in tumor-bearing animals [11]. Tu and colleagues have revealed that one of curcumin's most impressive effects in vivo was the inhibition of these myeloid cells' expansion in the spleen and polarization toward a M1-like phenotype [43]. Therefore, our work first aimed at providing information at a molecular level on spleen proteomic biomarkers which were affected by both tumor progression and curcumin treatment. To date, this field of research has been poorly documented experimentally, and limited to the case of non-medicated and non-immunized rats [44].
Among the 55 proteins showing significant abundance changes common to G2 vs. G1 and G3 vs. G2, and excluded from G3 vs. G1, 13 were already reported in the literature as being associated with curcumin's effects. One especially has been the subject of numerous studies since 1998 [12], related to curcumin's suppression of the transformation of the aryl hydrocarbon receptor (encoded by the Ahr gene) in cancer cells [14,20]. A recent study by Mohammadi-Barbori and colleagues has shown that this effect's mechanistic aspects also include chromatin remodeling [19] in addition to a previously documented modification of the cellular redox status [15]. Interestingly, another important target is represented by miRNA expression [30], and particularly miRNA-21 (miR-21), which is regulated by curcumin [27,29,31,32]. Even more interesting, Pdcd4 represents a functional target of miR-21 [45], while the anticarcinogenic effects of curcumin were reported to be partly mediated through modulation of miR-21 expression [32]. These two findings suggest that the decrease in PDCD4 immunostaining we observed in spleen cells of curcumin-treated rats vs. untreated rats might reflect the reversal produced by curcumin treatment on the deleterious effects induced by the tumor development on immune cell functions. RhoA, one of the most extensively studied members of the Rho family of small GTPase, is also involved in tumor cell migration and invasion [46], as well as fibrosis induction [36,37], both of which are prevented by curcumin through reversion of the epithelial-mesenchymal transition [35,38]. Finally, in our study, among the twelve proteins involved in the most significant abundance changes specifically observed in the spleen, which were related to tumor progression and completely reversed upon curcumin treatment, two already reported in the literature concerned complement C9 and transgelin-2. In the first case, Jacob and colleagues have previously reported the reduction of complement activation and C9 deposits by curcumin [22]. In our data, the fact that no change was observed in parallel in lymph nodes could be explained by the finding that activation of the complement cascade resulted in rapid disappearance of C9 from the plasma and accumulation in the spleen [47]. However, although the reversion induced by curcumin treatment was not complete, a common effect in spleen and lymph nodes was observed for another member of the complement system, C3, consistent with its known role in both the classical and alternative pathways [48]. Regarding transgelin-2, Ma and colleagues have revealed that the gene encoding for this protein belonged to a group of fifteen genes involved in curcumin's therapeutic effect against human tongue cancer [41].
The decrease in carbonic anhydrase 1 (encoded by Ca1) and delta-aminolevulinic acid dehydratase (encoded by Alad) induced in the spleen by tumor progression concurred with our histological observations on the red pulp, as these enzymes are specific to erythroid cells. However, in addition to their balance, both components of the spleen appeared to be affected in G2 rats. The parallel decrease in the activity of nucleosome assembly protein 1-like 1 (encoded by Nap1l1) specific to this group could be explained by Tanaka and colleagues' findings that its downregulation renders the cell vulnerable to apoptotic cell death via attenuation of NF-kB transcriptional activity on Mcl-1 [49]. Indeed, the protein is quite abundant in immune cells. Moreover, this observation agrees with another report from Çevik et al., demonstrating that lack of NAP1L1 leads to inefficient TLR3 responses, which are involved in pathogen recognition and activation of innate immunity [50]. Among the nine other main spleen proteins concerned with a complete reversion of the increase produced by tumor progression, we found two proteins involved in cell transduction pathways, one located in the cytosol (encoded by Prkar1a), and the second in the nucleus (encoded by Mapk3). The interrelationship between cyclic AMP elevation and ERK1/2 activation has previously been documented in cancer cells [51], suggesting that a combined increase in untreated tumor-bearing rats could be associated with early tumor progression in this lymphoid organ. Another category of proteins exhibiting increased abundance in G2 rats includes calreticulin and leucine-rich repeat-containing protein 59, both located in the endoplasmic reticulum (ER). This observation could relate to the fact that enhanced ER activity is required to facilitate the folding, assembly, and transportation of membrane and secretory proteins, all these functions being carried out by chaperones [52]. However, calreticulin can also be co-located in the nucleus, elevated levels being associated with additional cellular processes, and particularly with poor outcomes in some cancers as reviewed by Fucikova et al., in line with enhanced angiogenesis and facilitation of the migration and proliferation of tumor cells [53]. As in our study, elevated levels were observed not in the tumor, but a secondary lymphoid organ, the spleen, perhaps due to efferocytosis, which corresponds to the uptake and removal of apoptotic cells by phagocytes leading to silenced immune responses [54]. The elevated levels of annexins A1 and A5 are also frequently associated with tumor invasiveness [55,56], while the ephrin receptor family plays a major role in modifications of the tumor microenvironment and tumor immune evasion [57]. In this context, the increase we observed in the peculiar case of ephrin-B1 in the spleen could be explained by Iwasaki et al.'s finding that activation of this protein's expression is related to chronic hypoxia [58], which is also consistent with our observations on the aforementioned decrease in erythrocytic cell biomarkers.
In our study, cross-comparing spleen and lymph node proteomic data led us to identify several common biomarkers of interest, among which C-reactive protein, which represents one member of the positive acute-phase response proteins commonly elevated in inflammatory diseases and cancers. Even more interestingly, eight biomarkers showed similar profiles of abundance changes in both secondary lymphoid organs. The protein exhibiting the most significant changes, encoded by Mybbp1a, was initially reported to belong to Aurora kinases, presenting an essential role in the normal progression of mitosis [59]. Subsequently, George and colleagues pointed to the critical role of MYBBP1A in the regulation of senescence in cancer cells under genotoxic stress [60], while Weng et al. emphasized its overexpression in the progression of hepatocellular carcinoma with poor prognosis [61]. Interestingly, Nahálkova recently reported that MYBBP1A is associated with p53, TPPII, SIRT6, and CD47 in a protein interaction network that controls the Warburg effect [62]. Two other proteins of interest that bind to single-stranded DNA in the nucleus are encoded by Purb and Fubp1. In the first case, purine-rich element-binding proteins A and B are implicated in the regulation of gene expression at both transcription and translation levels, the highest levels of PURB being observed in myeloid cells from patients with primary acute myelogenous leukemia displaying risk factors forecasting a poor prognosis [63]. At the end of the 2010s, FUBP1 represented one of the fifty types of specific factors already reported to regulate c-myc transcription [64], whose overexpression was associated with the regulation of proliferation and migration in liver cancer cells [65]. By 2019, Debaize and Troadec concluded that FUBP1 represented a potent pro-proliferative and anti-apoptotic factor which also appeared to be of clinical relevance in oncogenesis [66]. Recent findings by Ma and colleagues have confirmed this protein's value as a novel prognosis factor and therapeutic target for cervical carcinoma [67]. Another protein showing extremely significant abundance changes in both organs is WIPF1, which belongs to the WASP-interacting family of proteins [68] and is involved in the formation of actin-rich membrane protrusions degrading the extracellular matrix called invadopodia [69]. Aberrant expression of WIPF1 has been reported in several cancers, contributing to invasive and metastatic properties [70]. In the context of our study, an intriguing feature is the observation made by Ramesh et al. concerning the implication of WIPF1 in the regulation of TCR signaling, linked to cytoskeletal abnormalities in inherited immune deficiency [71]. Finally, two more important members are ROA2 and PSB10. The former is an RNA-binding protein of extracellular vesicles involved in intercellular communication [72,73], while the latter represents a key subunit of the immunoproteasome [74,75].
Lastly, the pattern of abundance changes that we observed specifically in lymph nodes for purine nucleoside phosphorylase, which mimicked our previous finding in the liver of the same rats [8], tends to confirm its value as an immune deficiency biomarker. For groups G2 and G3, the M5-T1 original neoplastic cell line [7] was injected intraperitoneally (3 × 10 6 cells in 200 µL PBS buffer) at day 0 ( Figure 1A). At day 16, all rats were anesthetized in an isoflurane chamber (Forene ® , Abbott, Rungis cedex, France) and finally euthanized in their home cage with a rate of 30% volume displacement per minute of CO 2 . In untreated rats (G2), numerous metastatic nodules were collected in the peritoneal cavity, invading the liver, pancreas, diaphragm, gut, and parietal peritoneum, easily identified by their white color and dense texture. In curcumin-treated rats (G3), metastatic nodules were absent from the peritoneal cavity and diaphragm, and a few residual tumor masses were collected in the peritoneal cavity, attached to the omentum, liver, or gut (Supplementary Figure S1).

Proteomic Analyses
For SWATH-MS analyses, four 20 µm-thick sections of each spleen and lymph node sample were used. In a first step, HPS-stained sections were examined to select areas of interest (to remove any risk of tumor cell contamination), and then corresponding areas were removed with a scalpel and collected in a 1.5-mL Eppendorf ® microtube. The samples were deparaffinized with xylene washes, rehydrated in graded ethanol solutions, and vacuum-dried [8]. Dried tissues were resuspended in 200 µL of Rapigest SF (Waters, Milford, MA, USA) and dithiothreitol added to a final concentration of 5 mM (DTT, Ap-pliChem, Darmstadt, Germany). Tubes were incubated in a thermo shaker at 95 • C for one hour, and sonication was performed twice by Ultrasonic processor 75185 (130 W, 20 KHz, Bioblock Scientific, Illkirch, France). Subsequently, cysteine residues were alkylated by adding 200 mM S-methyl-methanethiosulfonate (MMTS) to a final concentration of 10 mM (incubated at 37 • C). Sequencing-grade trypsin was added in a ratio ≥ 2 µg mM −3 tissue (incubated at 37 • C overnight). The reaction was stopped with formic acid (9% final concentration) and incubated at 37 • C for one hour, and the acid-treated samples were centrifuged at 16,000× g for ten minutes. Salts were removed from the supernatant and collected into new reaction microtubes using self-packed C18 STAGE tips. Peptide concentrations were finally determined with the Micro BCA™ Protein Assay Kit (Thermo Fisher Scientific, Saint-Herblain, 44, France).
Five micrograms of each sample were analyzed by repeating the whole gradient cycle corresponding to the acquisition of 32 time-of-flight MS/MS scans of overlapping sequential precursor isolation windows (25 m/z isolation width, 1 m/z overlap, highsensitivity mode). Each MS/MS scan covered the 400 to 1200 m/z mass range, with a previous MS scan for each cycle. The accumulation time was 50 ms for each MS scan and 100 ms for the product ion scan (230 to 1500 m/z), leading to a total cycle time of 3.5 s. Peak extraction of the SWATH data and relative quantification were further performed as previously described [8].

Conclusions
Taken together, our observations suggest that multiple intraperitoneal administration of curcumin produced effects in secondary lymphoid organs of tumor-bearing rats at two complementary levels. At the spleen level, close to the initial location of tumor development, curcumin restored both red and white pulp functions and the delicate balance observed between these two compartments in normal animals, shown by the return to normal abundances of 53 proteomic biomarkers. In lymph nodes, the decrease in PNPH produced by tumor development and related to immune deficiency also returned to normal under curcumin treatment. Furthermore, curcumin contributed to reducing the elevation of twelve additional biomarkers, and to limiting the reduction of two others produced by tumor progression in untreated animals. At a second level, the fact that eight biomarkers showed similar profiles of abundance changes in the spleen and mesenteric lymph nodes, far from the place of tumor development, tends to demonstrate the existence of an out-offield systemic antitumor effect, probably related to massive induction of necrosis/apoptosis in tumor cells by curcumin treatment. The nature of biomarkers affected by this "abscopal effect" finally suggests, in addition to C-reactive protein secreted in the blood and related to inflammation, an action at multiple levels in immune cells within lymphoid follicles, including transcription regulation, reorganization of the cytoskeleton, TCR signaling, and regulation of proteasomal activity. These observations provide a good basis for future mechanistic studies.