*2.2. Absence of Toxicity for Normal B- and T-Cells*

Potential use of native fucoidan or vLMW-F for biomedical application and specificity of the treatment require that they do not display toxicity for normal cells. We chose to focus on peripheral blood mononuclear cells and particularly on normal B-cells (for comparison with tumoral B-cells), T-cells and activated T-cells (frequently implicated in anti-tumor responses via the PD-L1/PD-1 axis). We performed an Annexin V-based flow cytometry test that enables assessing simultaneously apoptosis and viability. No cytotoxic effect was observed for either the native form or the two vLMW-F (Figure 3).

**Figure 3.** Native fucoidan or vLMW-F are not toxic for normal lymphocyte cells. Normal cells (mononuclear cells, B-cells, T-cells activated or not) were treated or not (control) with 100μg/mL of native fucoidan or vLMW-F (F1 and F2) for 48 h. Apoptosis analysis (Annexin V/PI staining) was realized by flow cytometry. Results were obtained from three independent experiments. An example of each population is shown (intact cells: green events–early apoptotic cells: blue events–late apoptotic cells: purple events) as well as the percentage of total Annexin V<sup>+</sup> cells for each condition. No apoptosis induction was observed. NS: not significant.

#### *2.3. Decrease in PD-L1 Transcriptional Expression*

*PD-L1* transcriptional expression involves signaling pathways that can be inhibited by fucoidan, such as PI3k/AKT, NF-κB or RAS/ERK1/2 [33–36]. Therefore, we studied the effect of native and vLMW-F (F1 and F2) on mRNA expression of *PD-L1* for the three LCLs and four DLBCLs cell lines. We showed that it was strongly decreased in LCLs regardless of treatment and to a lesser extent in DLBCLs (Figure 4). Our results emphasized that transcriptional expression of *PD-L1* can be strongly impaired by native fucoidan or vLMW-F.

**Figure 4.** Fucoidan downregulates transcriptional expression of *PD-L1* in LCLs and DLBCLs cells. LCLs (J1209, C0401, C1504) and DLBCLs (U2932, OCILy10, SUDHL4, SUDHL6) were treated with 100μg/mL of native fucoidan or vLMW-F (F1 and F2) for 48 h followed by RNA extraction and RT-qPCR. Results were obtained from three independent experiments. mRNA relative expression of *PD-L1* was decreased either for the native form or the fractions. NS: not significant; \* *p*< 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001; \*\*\*\* *p* < 0.0001.

#### *2.4. Decrease of membrane, but Not of total, PD-L1 protein expression*

Since mRNA expression can differ from protein expression, especially due to translational regulation, we studied the total protein expression of PD-L1 for the same cell lines and in the same treatment conditions. All the cell lines strongly expressed PD-L1 and, surprisingly, despite inhibition of transcriptional regulation, total PD-L1 expression remained unchanged or occasionally slightly modified whatever the cell line and treatment, as observed by Western blot (Figure 5) or flow cytometry (Figure S3). This could be explained by the fact that PD-L1 is stored in secretory lysosomes before their migration *via* the actin network and their fusion with the plasma membrane, which could mask transcriptional regulation.

**Figure 5.** Native fucoidan or vLMW-F do not modify PD-L1 total expression. PD-L1 total expression analysis for LCLs (J1209, C0401, C1504) and DLBCLs (U2932, OCILy10, SUDHL4, SUDHL6) by Western blot after 48 h of 100 μg/mL native fucoidan or vLMW-F (F1 and F2) treatment. No significant change was observed for PD-L1 total expression. Results were obtained from three independent experiments. NS: not significant.

Since the active fraction of PD-L1 responsible for interaction with PD-1 is that at the cell surface, we have also studied its specific expression. Cytometry analysis was carried out on viable cells populations in order not to consider possible degradation of the protein on dead cells. As expected, cytometry analysis showed overexpression of surface PD-L1, which was much higher for LCLs [37] than for DLBCLs cells [8] (Figure 6A). However, unlike for total expression, fucoidan treatments significantly decreased PD-L1 membrane expression for the three LCLs and four DLBCLs, with generally better results for vLMW-F (Figure 6B). Otherwise, PD-L1 decrease was more important for LCLs than DLBCLs that have, however, lower baseline expression. Our results emphasized a specific decrease in PD-L1 membrane expression in the presence of fucoidan species, especially effective on LCLs when treated with vLMW-F. Since total expression remained unchanged, this could be due to regulation of membrane traffic. We verified that a decrease in membrane PD-L1 expression was due to the fractions of vLMW-F and not L-Fucose since it is the smallest subunit of fucoidan skeleton (Figure S4).

**Figure 6.** vLMW-F decrease PD-L1 surface expression in LCLs and DLBCLs cells. LCLs (J1209, C0401, C1504) and DLBCLs (U2932, OCILy10, SUDHL4, SUDHL6) cells were treated with 100μg/mL of native fucoidan or vLMW-F (F1 and F2) for 48 h followed by immunofluorescent staining for PD-L1 analyzed by flow cytometry. Results were obtained from three independent experiments. (**A**) PD-L1 is overexpressed by LCLs compared to DLBCLs. (**B**) Fold change (ratio of MFI test/MFI control, both normalized to isotypic control. MFI: mean fluorescence intensity) of PD-L1 surface expression for LCLs and DLBCLs. vLMW-F generally decreased more efficiently than the native form, PD-L1 surface expression for LCLs and DLBCLs. NS: not significant; \* *p*< 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001; \*\*\*\* *p* < 0.0001.

#### *2.5. Disruption of Actin Network and Decrease in Secretory Activity*

We have already shown for LCLs that PD-L1 membrane expression involved migration of LAMP2+/PD-L1+ secretory lysosome via the densified actin network and vesicular membrane traffic [37]. Therefore, we studied the impact of native and vLMW-F fractions F1 and F2 on actin network. F-actin was stained by fluorescent phalloidin. We emphasized a marked decrease in actin polymerization, especially for LCLs and to a lesser extent for DLBCLs cells, as visualized by confocal microscopy (Figure 7A) and quantified by flow cytometry (Figure 7B). Furthermore, we showed, by quantifying the F-actin, that, for normal B-cells, neither native fucoidan nor vLMW-F modify actin network (Figure 7C). To assess vesicular membrane traffic, we focused on LCLs, for which PD-L1 membrane expression and actin network were more affected. We studied native fucoidan and vLMW-F effects on two specific reporter molecules, which are constitutive of intracellular vesicles and expressed at the plasma membrane following their fusion: CD63 and LAMP2. Moreover, LAMP2 is otherwise a lysosomal-associated protein expressed by secretory lysosomes

(storage location of PD-L1). Analysis by flow cytometry showed a decrease in the two secretory vesicle markers at the plasma membrane for the native fucoidan and even more for vLMW-F F1 and F2 (Figure 7D). Our results suggest that fucoidan impedes secretory activity through the actin network and vesicle fusion with plasma membrane, which leads to decrease in PD-L1 membrane expression.

**Figure 7.** vLMW-F decrease actin polymerization and vesicle markers expression at plasma membrane. (**A**) LCLs (J1209, C0401, C1504) and DLBCLs (U2932, OCILy10, SUDHL4, SUDHL6) cells were treated or not (control) with 100μg/mL of native fucoidan or vLMW-F (F1 and F2) for 48 h, followed by phalloidin staining (F-actin, blue) and TOPRO-3 (nuclei, red) before confocal microscopy observations. (**B**, **C**) Quantification of F-actin in the same conditions analyzed by flow cytometry for (**B**) LCLs and DLBCLs or (**C**) normal B-cells. vLMW-F significantly decreased content of actin polymerization strongly in LCLs and to a lesser extent in DLBCLs but not in normal B-cells. Results were obtained for three independent experiments. (**D**) Flow cytometry fluorescence histograms of surface CD63 and LAMP2 (secretory vesicle markers) for J1209 (obtained for three independent experiments). The decrease is more efficient with vLMW-F than with the native form. NS: not significant; \* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001.

#### **3. Discussion**

Our results show that vLMW-F decrease cell proliferation and induce apoptosis of LCLs and DLBCLs tumoral B-cells without being toxic for normal B- and T-cells. Furthermore, these formulas decreased PD-L1 expression at transcriptional and cell surface levels. These results may be explained by actin network alteration that could be associated with deregulation of cells secretory activity, reducing thereby PD-L1 externalization.

Numerous studies report *in vitro* and *in vivo* anti-cancer effects (such as anti-proliferative, pro-apoptotic, anti-angiogenic and anti-metastatic) of fucoidan extracts or their low-molecularweight derivatives against multiple types of solid tumors, such as colon, breast, hepatocarcinoma, lung or bladder [16,32,38]. On the contrary, scarce data are reported regarding their effects on liquid tumors, especially concerning LMW derivatives of fucoidan. Few studies have shown that native fucoidan extracted from *Fucus vesiculosus* can inhibit proliferation of myeloid and monocytic leukemia cell lines by inducing their apoptosis [24,30,39] or arrest ABCor GCB-DLBCL cell cycle [40]. Therefore, we first confirmed these effects previously reported on lymphocytic B lineage and demonstrated for the first time that these anti-proliferative and pro-apoptotic activities are also extended to LCLs. We also highlighted that extremely depolymerized formulas (< 600 Da) maintain specific anti-tumor activity with more efficiency than native form while being non-toxic for normal B- and T-cells. This is consistent with the fact that depolymerized forms are generally more potent [21,23] and in agreement with non-toxicity of fucoidans [16,32,41]. Use of depolymerized form enables being more specific with fewer unwanted effects.

We wanted to confirm the role of fucoidans since they, or derivatives, cover a wide range of immunomodulatory effects and can participate in immune response against several cancers and infectious diseases [42,43], sulfate and acetyl groups mainly contributing to the activity [44]. For instance, they have been shown to enhance dendritic cell maturation, cytotoxic T-cell activation, antibody production or memory T-cells production [45,46]. They can also increase production of TNFα by neutrophils *in vitro* and *in vivo*, delaying their apoptosis [45,46], or induce NO synthesis [47] and activate *in vitro* and *in vivo* NK cells [45,48]. However, scarce data exist either for solid or liquid tumors in regard to their effects on the inhibitory immune checkpoints responsible for tumor cells escape from immune surveillance, especially regarding the PD1/PD-L1 axis. Yet, other data have shown that fucoidan impairs pathways responsible for PD-L1 expression, such as PI3k/AKT, NF-κB or RAS/ERK1/2 [33–36], making them very promising candidates for immune checkpoint modulation. To date, only two articles have reported a native fucoidan that can decrease PD-L1 expression in tumor tissues of experimental-induced mammary cancer [22] and a LMW one that can decrease transcriptional expression of *PD-L1* and *PD-L2* and PD-L1 protein level in fibrosarcoma cells [23]. Thus, it is particularly interesting to better understand effects on PD-L1 since it is particularly critical in immune escape and now of great interest in clinical care/immunotherapy approaches. The decrease in transcriptional expression that we found is in agreement with other studies that have demonstrated that signaling pathways involved in PD-L1 expression (PI3k/AKT, NF-κB or RAS/ERK1/2) can be inhibited by fucoidan [33–36]. Especially, our results confirm such effect for the first time in B-lymphoma cells, whether with the native form or vLMW-F.

In tumor cells, transcriptional expression of *PD-L1* is often not directly related to protein expression due to stabilizing post-translational modifications, such as N-glycosylation and phosphorylation [49,50]. Moreover, it can be accumulated in vesicles, such as exosomes [5] or secretory lysosomes, as demonstrated for LCLs [37], where we observed cytoplasmic accumulation. The relevant fraction of PD-L1 is that expressed at cell surface and able to interact with inhibitor receptor PD-1 expressed on anti-tumor immune cells. As expected, we observed overexpression for LCLs [37] compared with DLBCLs cells [8]. Interestingly, we next showed a PD-L1 decrease on lymphoma B-cells after treatment by native and our two fucoidan fractions, with little more efficiency for vLMW-F than the native form. It is of importance to note that PD-L1 decrease occurs on viable cells since it reduces their inhibition towards anti-tumor immune cells, while some of them die by apoptosis. The same tests realized with L-fucose monomer showed no effect, clearly demonstrating the real added value for these original vLMW-F.

Maintenance of high intracellular level of PD-L1 and decrease on cell surface emphasized that native fucoidan, as well as vLMW-F, can regulate membrane secretory traffic of this immune checkpoint. A prerequisite for vesicle secretion is their migration close to the plasma membrane *via* the actin network [51,52]. Remodeling and molecular modification of actin network occur in tumor cells [53,54] and constitute a specific and original therapeutic target [55,56]. We previously showed that actin network is increased in LCLs by EBV latency III program and responsible for fusion of secretory lysosome with plasma membrane, leading to PD-L1 overexpression at the cell surface [37]. Presently, we show that native fucoidan, as well as vLMW-F, decrease actin polymerization for LCLs and to a lesser extent for DLBCLs cells, which can be associated with reduction in PD-L1 membrane expression. This is also supported by decrease in LAMP2 membrane expression, particularly after treatment with vLMW-F, which has been correlated with vesicles secretory activity of cells. Data in the literature confirm some links between fucoidan and actin pathway, such as decrease in expression of genes implicated in polymerization, organization and stabilization [57] or disruption of F-actin stress fibers [58,59]. In this context and as a first hypothesis, the difference between native and vLMW forms might be explained by variation in the internalized amount of the different species, especially for their molecular weight differences. No effect on normal cells, associated with lack of toxicity, suggests very interesting specific sensitivity for tumoral actin network. Our results also suggest that expression of other inhibitory immune checkpoints could be decreased insofar as they pass through secretory vesicles [4], secretory lysosomes [37,60] or exosomes [61,62].

As native fucoidan is mainly composed of fucose and because the F1 and F2 fractions present very low DP, such formulas may comprise mono-, di- or tri-saccharides of fucose that are imputable of the effects observed on our cellular models. Results obtained with the single L-fucose monosaccharide did not show any effect. This suggests that F1 and F2 present other original fucoidan-derived compounds of vLMW responsible for these bioactivities, perhaps including glucuronic acid, galactose and xylose (the other principal sugar found in fucoidan) [63], or backbone modifications because of the depolymerization method (opening or creation of insaturation because of H2O2-based hydrolysis) [64].

Altogether, our results suggest that efficiency of vLMW-F is correlated to their low molecular weight (by comparison with the native form), which is consistent with the literature [23,65]. It will be interesting to study if this can be associated with different capacities of internalization. However, no noteworthy differences were observed between the two fractions, F1 and F2, that differ mainly by their sulfation degree. These slight differences in sulfation percentage could explain the different effects observed with the two vLMW-F in some experiments. Negatively charged sulfate groups are responsible for interaction with numerous molecules and involved in various biological process. They contribute to fucoidan activity. However, at a very low fraction size, this parameter does not seem to be a key factor in the bioactivities observed. Considering their very small size, high internalization of vLMW-F could explain that a lower degree of sulfation is sufficient for optimal activity. Establishing a complete picture of the structure (Mn and % S)/bioactivity relationship of such a complex fucoidan molecular structure is tedious as both parameters can differently influence the outcome of the effect according to the bioactivity sought and the experimental model used. However, in perspective, it could be interesting to compare the results obtained in this work with ones of desulfated native fucoidan and/or oversulfated vLMW-F.

In translational medicine, fucoidan extracts are considered of great interest as adjuvants for cancer therapy [66,67]; benefits are also observed with anti-PD-L1 or anti-PD-1 immunotherapies in mice models. Fucoidan extracted from *Fucus vesiculosus* promoted activation of tumor-infiltrating CD8+ T-cells and strongly inhibited growth of melanoma cells when co-administrated with anti-PD-1, particularly when applied before immunotherapy [35]. Fucoidan extracted from three species of algae, among them *Fucus vesiculosus*, promoted proliferation and activity of *ex vivo* PBMCs and potentiated anti-PD-1 effects [68]. Intranasal administration of *Ecklonia-cava*-extracted fucoidan enhanced anti-PD-L1 mediated anti-cancer activities against melanoma and carcinoma tumor growth in lungs [69]. Fucoidan of *Luminaria japonica* enhanced anticancer efficacy of anti-PD-L1 antibodies against Lewis lung carcinoma [70].

Algal polysaccharides, such as fucoidan, could provide novel therapeutic alternatives and promising supplements, especially for cancer treatments. Clinical trials are actually in course for hepatocellular carcinoma (NCT04066660) and rectal cancer (NCT04342949). It could be useful combined with immune checkpoint blockade therapies to treat lymphoproliferative malignancies.

Our results suggest that vLMW-F (<600 Da) could be effective potential adjuvants of anti-PD-L1 or anti-PD-1 immunotherapy thanks to their favorable anti-proliferative and pro-apoptotic effects associated with their ability to decrease membrane PD-L1 *via* actin depolymerization. Since fucoidan possesses anti-inflammatory properties, which could interfere with antitumoral functions (reduction in recruitment of antitumoral killing cells and increase in cancer cell apoptosis and chemo-sensitivity), it will be important to further investigate this activity for vLMW-F. Otherwise, the PD-L1/PD-1 axis plays a crucial role in the tumor microenvironment and interactions between antitumor and cancerous cells. Therefore, functional *in vitro* studies on co-culture models with autologous T-cells or NK cells and *in vivo* studies on mice models are the next steps. Treatments with anti-PD-L1 or anti-PD-1 antibodies and fucoidan (native and vLMW-F) remain to be evaluated, independently and in combination.

#### **4. Materials and Methods**

#### *4.1. Fucoidan Samples*

Native fucoidan of *Fucus vesiculosus* was obtained from Sigma-Aldrich. Original formulas were depolymerized by our collaborators from LIENSs laboratory (UMR CNRS 7266, La Rochelle University, France) from the native fucoidan using a radical H2O2-based hydrolysis method previously published with other types of polysaccharides [71]. Briefly, native fucoidan was dissolved in Milli-Q water (25 mg/mL), and then the solution was purged with argon and heated until 60◦C. Addition of H2O2 30% (Sigma-Aldrich) at a weight/weight ratio of 0.5 and 1.5 resulted in production of two different vLMW-F fractions, named F1 and F2, respectively, after 96 h and 72 h of depolymerization. The chromatographic profile shows that F1 and F2 fractions are practically eluted at the same time and after the native fucoidan, which attests that the H2O2-based hydrolysis reaction worked well (Figure S1). Number-average molecular weights (Mn), degree of polymerization (DP) and polydispersity index (I) were estimated by SEC-HPLC according procedures already published [71] using calibrant curves made of pullulans standards (Polymer Standards Service GmbH, Mainz, Germany) for native fucoidan and heparin standards (Iduron, UK) for F1 and F2 fractions. Degree of sulfation (DS) was calculated by an Azure-A-basedcolorimetric assay (Sigma-Aldrich) according to a state-of-the-art technique. Results of these characterizations are summarized in Table 1.

**Table 1.** Characteristics of native fucoidan and vLMW-F (F1 and F2). Number-average molecular weights (Mn), degree of polymerization (DP), polydispersity index (I) and degree of sulfation (DS)**.** \*: calculated with pullulans standards. \*\*: calculated with heparin standards.


As shown in Table 1, the F1 and F2 fractions contain a mixture of different fucoidanderived compounds with similar very-low-mean DPs between 3.4 and 3.5 but are distinct by their mean sulfation degree, 6% against 2.1%, respectively. Further, LC–MS (data not shown) confirmed that these fractions contain predominantly di- and tri-oligofucoidan with, interestingly, several unsaturated species.
