NRP1 is overexpressed in human HCC tissue and is correlated with advanced stages and nodal metastasis status

To determine the potential interest of NRP1 in the development and progression of HCC, we analyzed the NRP1 expression levels in different HCC datasets from human databases (Fig. 1a–f). Representative images of NRP1 immunohistochemistry in normal liver and HCC tissue of the TCGA dataset were obtained from the HPA database with both HPA030278 and CAB004511 antibodies, observing a strong NRP1 staining in the liver HCC tissue (Fig. 1a). Through UALCAN database, a significantly increased NRP1 expression was identified in the primary tumor tissue of HCC samples compared to healthy normal tissue from the TCGA (Fig. 1b). It was also confirmed by an independent analysis performed with the GSE14520 dataset, in which differential-expressed genes were determined, identifying a significant overexpression of NRP1 in tumor tissues compared to paired non-tumor tissues of HCC patients (Fig. 1c).

Fig. 1: Characterization of NRP1 expression in human samples and HCC cell lines, and cell migration ability.

Representative images of NRP1 immunohistochemistry in normal liver and HCC tissues from the (a) HPA database and comparative analysis of NRP1 expression levels obtained from the (b) UALCAN database employing the TCGA gene datasets. Identification of significantly downregulated (green) or upregulated (red) expressed genes in HCC GSE14520 dataset from the (c) GEO database. Association of NRP1 levels with tumor stages from the (d) UALCAN and (e) GEPIA databases, and with different nodal metastasis status from the (f) UCSC Xena database. Significant differences when *P < 0.05. Comparison of the (g) mRNA and (h) protein levels of NRP1 determined in the three HCC cell lines HepG2, Huh-7 and Hep3B by qRT-PCR and Western blot, respectively. A representative immunoblot for each protein with the quantification of the corresponding triplicates is shown. Data are expressed as mean values of arbitrary units (a.u.) ± SD (n = 3). *P < 0.05, **P < 0.01, ***P < 0.001 vs HepG2; #P < 0.05 vs Hep3B. i Cell migration ability was evaluated by wound-healing assay, representing the % of the wound closure area after 4, 8, 12 and 24 h from the scratch, separately for each cell line and comparing the three HCC lines. Magnification 10×, scale bar 50 µm. For each cell line analysis *P < 0.05, **P < 0.01, ***P < 0.001 vs 4 h. For comparison analysis of the three cell lines *P < 0.05, **P < 0.01, ***P < 0.001 vs HepG2; #P < 0.05 vs Hep3B.

Association of NRP1 expression levels with tumor stages in human HCC was also evaluated, finding a significant increased expression of NRP1 in advanced HCC stages from UALCAN and GEPIA databases (Fig. 1d, e). Interestingly, when this analysis was conducted in the metastatic nodal status, NRP1 was found to be overexpressed in advanced stages of nodal metastasis in human samples from the UCSC Xena database (Fig. 1f).

The Hep3B and Huh-7 HCC cell lines showed an increased NRP1 expression and cell migration ability together with higher susceptibility to lenvatinib

To confirm the results obtained from the clinical databases and select a suitable in vitro model, we analyzed the NRP1 expression and the migration ability of different HCC cell lines (Fig. 1g, i), with different phenotypic and genotypic characteristics in accordance to the high heterogeneity of human HCC. Results showed markedly higher levels of both mRNA and protein NRP1 in Hep3B and Huh-7 in comparison to HepG2 (Fig. 1g, h). After thorough analysis of cell migration at 4, 8, 12 and 24 h, HepG2 cells exhibited the lowest wound closure ability, while both Hep3B and Huh-7 displayed greater cell migration, reaching a closure higher than 60% of the wound area at 24 h (Fig. 1i).

After identifying the potential role of NRP1 in the progression of the HCC cell lines selected, we further assessed the molecular actions derived from the lenvatinib treatment. Initially, we observed a significant inhibition of cell viability from 1 µM in HepG2, and from the lowest dose (0.5 µM) in Hep3B and Huh-7 after 48 h of lenvatinib treatment (Fig. 2a). Interestingly, HepG2 cells were less susceptible to lenvatinib than Hep3B and Huh-7 cells, not reaching a 50% cell viability inhibition; while 1 µM and 2.5 µM of lenvatinib reduced to 50% the viability of Hep3B and Huh-7, respectively (Fig. 2a). For the following analysis, the two doses of lenvatinib closest to the IC50 were selected, being higher for the HepG2 line. As previously observed, lenvatinib was more effective in Hep3B and Huh-7 cells, showing a significant inhibition of colony formation ability and reduction of Ki67 proliferation index with both 2.5 and 5 µM lenvatinib (Fig. 2b, c, Supplementary Fig. S1a). However, HepG2 cell line was less sensitive, colony formation was only reduced after 48 h with 40 µM lenvatinib, while nuclear localization of Ki67 was significantly decreased with 20 µM lenvatinib (Fig. 2b, c, Supplementary Fig. S1a).

Fig. 2: Antitumor activity of lenvatinib and modulation of NRP1 expression in the in vitro models of HCC.

Different concentrations of lenvatinib ranging from 0.5 to 30 µM were used for the treatment of HCC cells during 48 h to determine the effects on (a) cell viability by CellTiter-Glo® assay, (b) colony formation assay, and (c) nuclear translocation of Ki67 by ICC and confocal microscopy. NRP1 expression was analyzed at (d) mRNA levels by qRT-PCR, and at protein levels by (e) Western blot and (f) ICC after 48 h of 2.5 and 5 µM lenvatinib treatment. Data from (a) are represented as % of cell viability relative to non-treated cells ± SD (n = 7). Data from (b–f) are represented as mean values of arbitrary units (a.u.) ± SD (n = 3). Bar graphs from (c) and (f) represent the nuclear CTCF ratio of Ki67 and NRP1 CTCF ratio, respectively. Magnification 63×, scale bar 10 µm. e A representative immunoblot is shown. *P < 0.05, **P < 0.01, ***P < 0.001 and ns, not significant vs non-treated cells.

Considering these findings and the close association of NRP1 and lenvatinib with cell migration and angiogenesis in cancer [9, 29], we performed the next experiments in the Hep3B and Huh-7 cell lines to precisely evaluate the role of NRP1 in lenvatinib efficacy in HCC.

Lenvatinib diminished NRP1 protein levels in HCC cells, being responsible for the antitumor effects of lenvatinib on cell proliferation and migration

In order to elucidate the association between NRP1 and lenvatinib efficacy in the human HCC cell lines, we analyzed the effects of lenvatinib treatment on NRP1 mRNA and protein levels (Fig. 2d–f). Remarkably, a significant alteration was not observed in NRP1 mRNA levels after 48 h treatment (Fig. 2d); but, protein expression by both Western blot and ICC experienced a strong reduction when 2.5 and 5 µM lenvatinib were administered (Fig. 2e, f, Supplementary Fig. S1b).

The NRP1 antagonist EG00229, hereinafter referred to as EG, acts through blockade of NRP1-VEGFA interaction, inhibiting NRP1 activity [30]. We employed EG to clarify the role of NRP1 in the antitumor actions of lenvatinib. Firstly, we selected an EG concentration by analyzing cell viability in Hep3B and Huh-7 after administration of different EG concentrations during 24 and 48 h (Supplementary Fig. S2). Although significant inhibition of cell viability was observed from the lowest dose (2.5 µM), the IC50 was only reached by 50 µM EG after 48 h in the Huh-7 cell line (Supplementary Fig. S2). Based on these results and published studies employing this NRP1 antagonist in cancer cells [31, 32], we selected 15 µM EG for the following experiments. In these analyses, we combined lenvatinib and EG, together with the specific gene silencing of NRP1 to assess its role in cell viability and migration in HCC (Fig. 3). Through determination of NRP1 expression by Western blot and ICC we proved that NRP1 gene silencing markedly reduced NRP1 protein levels, observing a slighter decrease in absence of gene silencing after lenvatinib administration alone and combined with EG (Fig. 3a, b, Supplementary Fig. S3a). We also observed that the individual treatment with the antagonist EG did not diminish the expression of NRP1 (Fig. 3a, b, Supplementary Fig. S3a).

Fig. 3: Effects of targeting NRP1 on lenvatinib actions on cell proliferation and cell migration ability.

All the assays were performed 48 h post-silencing with the last 24 h of treatment with 2.5 µM lenvatinib (Lvt) and/or 15 µM EG00229 (EG). Protein levels of NRP1 were analyzed by (a) Western blot and (b) ICC, and cell viability was assessed by (c) CellTiter-Glo®, (d) colony formation assays, and (e) nuclear translocation of Ki67 by ICC and confocal microscopy. Data from (c) are represented as % of mean values relative to control (Ctrl) ± SD (n = 7). Data from (a), (b), (d) and (e) are represented as mean values of arbitrary units (a.u.) ± SD (n = 3). a A representative immunoblot is shown. Bar graphs from (b) and (e) represent the NRP1 CTCF ratio and nuclear CTCF ratio of Ki67, respectively. Magnification 63×, scale bar 10 µm. f Cell migration was evaluated by wound-healing assay, representing the % of the wound closure area after 24 h from the scratch. *P < 0.05, **P < 0.01, ***P < 0.001 vs non-treated cells in each siR group; #P < 0.05, ##P < 0.01, ###P < 0.001 vs siR control; σσσP < 0.001 Lvt+EG vs Lvt treatment.

The derived effects on cell viability were also evaluated through cell viability and colony formation assays, as well as the determination of the Ki67-based proliferation index. Results showed that both lenvatinib and EG significantly decreased cell viability (Fig. 3c), colony formation ability (Fig. 3d) and nuclear localization of Ki67 (Fig. 3e, Supplementary Fig. S3b) alone and combined, not displaying a synergistic effect in combination, regardless of colony formation inhibition in Huh-7 cells (Fig. 3d). However, the NRP1 silencing increased the antitumor effects in all cases in terms of cell viability (Fig. 3c) and colony formation (Fig. 3d), but only increased lenvatinib-derived effects on Ki67 proliferation index reduction in Huh-7 (Fig. 3e, Supplementary Fig. S3b).

Regarding cell migration, similar findings were also observed (Fig. 3f, Supplementary Fig. S4). The individual treatment with lenvatinib and EG significantly diminished cell migration ability, exhibiting a synergy when combined only in the Huh-7 cell line. Likewise, NRP1 gene silencing raised this migration inhibition of lenvatinib in both cell lines, but of EG only in the Hep3B cells. Interestingly, NRP1 silencing only increased the effects of lenvatinib and EG combination in Hep3B, whereas in control silenced cells a significative difference was not observed when the antagonist EG was co-administered with lenvatinib. This was not obtained in the Huh-7 cell line, in which EG did increase lenvatinib-derived inhibition of cell migration when NRP1 was silenced (Fig. 3f, Supplementary Fig. S4). Therefore, NRP1 silencing only augmented lenvatinib effects when the administration of the NRP1 antagonist EG did not achieve for increasing cell migration inhibition derived from lenvatinib.

Altogether, these results suggest that NRP1 might be mechanistically important for the antitumor effects of lenvatinib on cell proliferation and migration, but in a previous step from NRP1 activity, releasing the interest on determining the exact mechanism underlying lenvatinib actions associated with NRP1 in HCC cells.

Lenvatinib promoted autophagy as a mechanism responsible for the NRP1 downregulation in HCC cells

To fully elucidate the exact mechanism through which lenvatinib is downregulating protein levels of NRP1, we used specific inhibitors of protein synthesis (cycloheximide, CHX, 300 µM) and protein degradation through proteasome (MG132, 30 µM) or autophagy (bafilomycin A1, 100 nM) (Fig. 4). Results exhibited that synthesis blockade reduced NRP1 protein expression; nonetheless, lenvatinib led to a higher downregulation of NRP1. Moreover, administration of the proteasome inhibitor MG132 did not alter NRP1 levels or lenvatinib effects (Fig. 4a), suggesting that inhibition of protein synthesis or induction of proteasome degradation are not the mechanisms responsible for the lenvatinib-associated reduction in NRP1 levels. Then, we employed bafilomycin A1 as a specific autophagy inhibitor of autophagosome-lysosome fusion alone and combined with lenvatinib for a time-course of 3, 6, 12 and 24 h to evaluate the dynamic process of autophagy (Fig. 4b–d). Interestingly, autophagy blockade restored the NRP1 protein levels when co-administered with lenvatinib from 12 h in Hep3B, and from 6 h in Huh-7, and this led to NRP1 protein accumulation in both HCC cell lines (Fig. 4b). In addition, the autophagy process was evaluated by acridine orange staining, showing an autophagy induction derived from lenvatinib treatment, which was significantly decreased to basal levels by bafilomycin A1 (Fig. 4c). This was also observed by protein expression analysis of the autophagic markers p62/SQSTM1 and LC3-II (Fig. 4d). Results showed an efficient blockade of lenvatinib-induced autophagy after bafilomycin A1 treatment, represented by LC3-II and p62/SQSTM1 protein accumulation, and by a significantly reduced autophagic flux index, which was also observed after bafilomycin A1 treatment alone (Fig. 4d). These results confirm the usefulness of this drug as an effective autophagy inhibitor.

Fig. 4: Determination of the mechanism underlying the lenvatinib-derived downregulation of NRP1.

a Protein expression of NRP1 was analyzed by Western blot after 24 h treatment with 2.5 µM lenvatinib (Lvt) and/or 300 µM cycloheximide (CHX) or 30 µM MG132 (MG). b NRP1 protein levels were also determined by Western blot after 3, 6, 12 and 24 h treatment with 2.5 µM lenvatinib (Lvt) alone and combined with 100 nM bafilomycin A1 (Baf). *P < 0.05, **P < 0.01, ***P < 0.001 vs control; #P < 0.05, ##P < 0.01, ###P < 0.001 combined treatment vs Lvt treatment; σP < 0.05, σσP < 0.01, σσσP < 0.001 combined treatment vs inhibitor treatment. c Analysis of autolysosome cell content by acridine orange staining and fluorescence microscopy. Magnification 40×, scale bar 25 µm. Bar graphs represent the quantification of red/green CTCF ratio (n = 5). d Protein levels of p62/SQSTM1 and LC3 (LC3-I and LC3-II) were analyzed by Western blot. LC3 turnover assay was performed to determine the autophagic flux index. Data from (a), (b) and (d) are represented as mean values of arbitrary units (a.u.) ± SD (n = 3), showing a representative immunoblot. *P < 0.05, **P < 0.01, ***P < 0.001 vs control for each time point; ##P < 0.01, ###P < 0.001 combined treatment vs Lvt treatment.

Therefore, these findings indicated that autophagy may be the mechanism that underlies the NRP1 downregulation exerted by lenvatinib in the HCC cells.

Autophagy-dependent degradation of NRP1 was the mechanism responsible for the antitumor effects of lenvatinib

Autophagy has shown to be a key process in the modulating actions of lenvatinib on NRP1 expression. Considering the double-edged role of autophagy in cancer [16] and the interesting role of NRP1 in the lenvatinib-derived inhibition of HCC cell proliferation and migration, we decided to assess the relationship between them (Table 1, Fig. 5). Firstly, we evaluated the potential correlation between the receptor NRP1 and autophagy-related genes in human HCC samples (Table 1). We obtained that transcriptional expression of NRP1 is positively correlated with up to 59 genes related to autophagy, ranging from the lowest Pearson-CC of +0.31 to the strongest Pearson-CC of +0.60 (Table 1). Moreover, statistical significance (P < 0.0001) was reached for all these correlations observed.

Table 1 Significantly correlated genes from autophagy with NRP1 in human HCC samplesFig. 5: Effects derived from autophagy inhibition on NRP1 protein expression and NRP1-associated antitumor actions of lenvatinib.

All the assays were performed 48 h post-silencing with the last 24 h of treatment with 2.5 µM lenvatinib (Lvt) and/or 100 nM bafilomycin A1 (Baf). Protein levels of NRP1 were analyzed by (a) Western blot and (b) ICC. Cell proliferation was assessed by (c) CellTiter-Glo® assay, (d) colony formation assay, and (e) Ki67 proliferation index determination. Data from (a), (b), (d) and (e) are represented as mean values of arbitrary units (a.u.) ± SD (n = 3), showing for (a) a representative immunoblot. Data from (c) are represented as % of mean values relative to control ±SD (n = 7). Bar graphs from (b) and (d) represent the NRP1 CTCF ratio and the nuclear CTCF ratio of Ki67, respectively. Magnification 63×, scale bar 10 µm. f Cell migration ability was evaluated by wound-healing assay, representing the % of the wound closure area after 24 h from the scratch. Magnification 10× and scale bar 50 µm. *P < 0.05, **P < 0.01, ***P < 0.001 vs non-treated cells in each siR group; #P < 0.05, ##P < 0.01, ###P < 0.001 vs siR Control; σP < 0.05, σσP < 0.01, σσσP < 0.001 Lvt+Baf vs Lvt treatment.

Based on these findings, we further analyzed the relationship between NRP1, autophagy and lenvatinib efficacy in our in vitro models of human HCC. As previously observed, combination of bafilomycin A1 with lenvatinib partially restored NRP1 protein levels; however, when NRP1 was silenced autophagy blockade did not prevent NRP1 downregulation (Fig. 5a, b, Supplementary Fig. S5a). Cell proliferation and migration processes were also assessed in these conditions, exhibiting a synergistic inhibition effect on cell viability (Fig. 5c), colony formation ability (Fig. 5d) and Ki67 proliferation index (Fig. 5e, Supplementary Fig. S5b) when HCC cells were treated with lenvatinib after NRP1 silencing. Furthermore, we demonstrated that autophagy blockage partially prevented antitumor effects of lenvatinib; nevertheless, NRP1 silencing prevented this loss of in vitro effectiveness of lenvatinib even in presence of bafilomycin A1 (Fig. 5c–e).

Regarding cell migration, autophagy inhibition also reduced the lenvatinib inhibitory effects, increasing the wound closure ability of the HCC cells in the presence of the drug (Fig. 5f, Supplementary Fig. S6). Although in this analysis NRP1 silencing and lenvatinib treatment did not show a synergy, the silencing strategy impeded the efficacy inhibition exerted by bafilomycin A1 on lenvatinib, thus avoiding loss of lenvatinib activity on cell migration (Fig. 5f, Supplementary Fig. S6).

Altogether, autophagy-dependent degradation of NRP1 seems to be a crucial mechanism in the loss of lenvatinib efficacy, which could be modulated by HCC cells during drug resistance development in order to avoid antitumor actions of lenvatinib. Therefore, NRP1 could be supposed as an interesting molecular target in human HCC in order to prevent autophagy-related lenvatinib resistance.

NRP1 was downregulated under a hypoxic microenvironment through autophagy induction in HCC cells

Together with autophagy, the hypoxia-derived response has been also closely related to chemotherapeutic failure and drug resistance acquisition in HCC [4, 5]. For this reason, we analyzed the likely modulation derived from the induction of a hypoxic microenvironment on NRP1 expression and the role of the autophagy process through stabilization of HIFs with the employment of CoCl2 as hypoximimetic (Fig. 6). Results displayed a significant downregulation of NRP1 protein levels after 24 h and 48 h of hypoxia induction (Fig. 6a, b, Supplementary Fig. S7). Since autophagy showed to be the main process responsible for lenvatinib-derived downregulation of NRP1, we also tested if the same modulation was being conducted under hypoxia. Interestingly, after 12 h of hypoxia induction, a marked decrease in NRP1 expression was observed, but also a recovery by bafilomycin A1 administration not only in the selected cell lines Hep3B and Huh-7 (Fig. 6c), but this was even observed in the HepG2 cell line, which experienced a recovery of NRP1 protein levels after autophagy blockade (Supplementary Fig. S8). In the case of the Hep3B and Huh-7 lines, an efficient inhibition of hypoxia-derived autophagy was achieved by bafilomycin A1 (Fig. 6d, e). Results showed that the autophagolysosome content was significantly reduced after bafilomycin A1 administration (Fig. 6d), an accumulation of p62/SQSTM1 and LC3-II and a strong decrease of the autophagic flux index were also osberved when bafilomycin A1 was added (Fig. 6e). Therefore, NRP1 degradation by hypoxia-induced autophagy could be an interesting cell mechanism and, together with the extensive processes already known, constitute the complex response to hypoxia in HCC.

Fig. 6: Analysis of the modulation on NRP1 protein levels by an in vitro hypoxic microenvironment.

Hypoxia (Hx) was induced by incubating HCC cell lines with 100 µM CoCl2 for the corresponding period of times. Protein expression of NRP1 was analyzed by (a) Western blot and (b) ICC. Bar graphs from (b) represent the NRP1 CTCF ratio. Magnification 63× and scale bar 10 µm. *P < 0.05, **P < 0.01, ***P < 0.001 vs normoxia (Nx). c NRP1 expression was also determined by Western blot in normoxia (Nx) and after hypoxia induction and/or autophagy inhibition by treatment with 100 nM bafilomycin A1 (Baf). *P < 0.05, **P < 0.01, ***P < 0.001 vs Nx; #P < 0.05, ##P < 0.01, ###P < 0.001 Hx+Baf vs Hx for each time point. d Autophagy was evaluated through analysis of autolysosome cell content by acridine orange staining and fluorescence microscopy. Magnification 40×, scale bar 25 µm. Bar graphs represent the quantification of red/green CTCF ratio (n = 5). ***P < 0.001 vs Hx for each time point. e Protein levels of p62/SQSTM1 and LC3 (LC3-I and LC3-II) were analyzed by Western blot. LC3 turnover assay was performed to determine the autophagic flux index. *P < 0.05, **P < 0.01, ***P < 0.001 vs Nx; #P < 0.05, ##P < 0.01, ###P < 0.001 Hx+Baf vs Hx for each time point. Data from (a), (b), (c) and (e) are represented as mean values of arbitrary units (a.u.) ± SD (n = 3), showing a representative immunoblot.

HIF-1α modulated NRP1 expression and was involved in the loss of lenvatinib efficacy derived from the autophagy-dependent degradation of NRP1 as part of the hypoxia response

The hypoxic microenvironment has shown to be a relevant mechanism on cancer progression and is mainly modulated by the HIF-1α, which has been closely related to development of drug resistance in HCC [5, 6]. Based on the results obtained and considering the interesting role of HIF-1α-associated response to hypoxia on loss of therapeutic efficacy, we evaluated the correlation between NRP1 and HIF-1α, as well as the modulation on the lenvatinib effects on both HCC lines (Fig. 7). At first, we determined the potential gene correlation of NRP1 with HIF-1α in human HCC samples by a comprehensive analysis in three different databases. We obtained a strong positive correlation between both genes, finding the correlation coefficients of +0.51 (Fig. 7a), +0.39 (Fig. 7b) and +0.52 (Fig. 7c), with statistical significance in all cases (P < 0.0001). We further performed an in vitro analysis by specifically silencing HIF-1α. Interestingly, we observed that an effective HIF-1α silencing under hypoxia led to a significant downregulation of NRP1 protein levels (Fig. 7d). Additionally, when HIF-1α silencing was combined with bafilomycin A1, HIF-1α was slightly decreased only in Hep3B and successfully silenced in both cell lines (Fig. 7e). On the other hand, NRP1 expression decreased under hypoxia and, as expected, was recovered when autophagy was inhibited; while HIF-1α gene silencing prevented this increase on NRP1 expression derived from bafilomycin A1 (Fig. 7e). We also found that acute hypoxia induction decreased cell viability of both HCC lines, showing a synergy when HIF-1α was silenced. Autophagy blockade increased Hep3B and Huh-7 viability; however, HIF-1α silencing prevented this reversal effect of bafilomycin A1, thus improving the inhibition exerted by hypoxia and HIF-1α silencing (Fig. 7f).

Fig. 7: Study of the HIF-1α-dependent modulation of NRP1 expression and its role on the autophagy-derived regulation of the hypoxia response and the loss of lenvatinib efficacy.

Plots of gene expression correlation between NRP1 and HIF-1α are shown, with the corresponding correlation coefficients and p-values obtained from the (a) UALCAN, (b) UCSC Xena and (c) GEPIA databases. All the following assays were performed 48 h post-silencing with the last 24 h of treatment with 100 µM CoCl2 to induce hypoxia (Hx), 100 nM bafilomycin A1 (Baf) and/or 2.5 µM lenvatinib (Lvt) in both HCC cell lines Hep3B and Huh-7. Protein expression was analyzed by Western blot and cell viability by MTT assay. d Protein levels of HIF-1α and NRP1 after HIF-1α silencing in hypoxia. *P < 0.05, **P < 0.01 vs siR Control. e Protein levels of HIF-1α and NRP1 and (f) cell viability after HIF-1α silencing under hypoxia and in combination with bafilomycin A1. **P < 0.01, ***P < 0.001 vs Nx; ##P < 0.01, ###P < 0.001 vs Hx; σσP < 0.01, σσσP < 0.001 vs siR Control. g Protein levels of HIF-1α and NRP1 and (h) cell viability after HIF-1α silencing under hypoxia and in combination with lenvatinib and/or bafilomycin A1. ***P < 0.001 vs Nx; #P < 0.05, ###P < 0.001 vs siR Control; σP < 0.05, σσP < 0.01, σσσP < 0.001 Hx+Lvt vs Hx; ττP < 0.01, τττP < 0.001 Hx+Lvt+Baf vs Hx+Lvt. Data from (d), (e) and (g) are represented as mean values of arbitrary units (a.u.) ± SD (n = 3), showing one representative immunoblot. Data from (f) and (h) are represented as % of mean values relative to normoxia (Nx) ± SD (n = 7).

Since autophagy inhibition could act as a relevant mechanism in the loss of lenvatinib efficacy through NRP1 modulation, as well as in the HIF-1α-associated response to hypoxia, crucial in the adaptive cellular response to chemotherapy, we decided to elucidate the interesting relationship among them in both HCC cell lines. Lenvatinib treatment under hypoxia increased HIF-1α protein expression, while augmented the NRP1 downregulation exerted by hypoxia (Fig. 7g). Remarkably, autophagy blockade by bafilomycin A1 restored NRP1 protein levels even in presence of lenvatinib and a hypoxic microenvironment, while no changes were observed in HIF-1α expression (Fig. 7g). When HIF-1α was specifically silenced, these alterations were not observed in both HIF-1α and NRP1 expression, preventing the upregulation of NRP1 caused by autophagy inhibition (Fig. 7g). Regarding cell viability, similar findings were obtained in which the synergistic effects of lenvatinib and hypoxia induction on decreasing cell viability were partially restored by bafilomycin A1. Likewise, HIF-1α silencing not only increased the inhibitory actions of combined lenvatinib and hypoxia, but also prevented the loss of effectiveness derived from autophagy blockade (Fig. 7h).

Altogether, these results suggest that NRP1 is directly modulated by HIF-1α under hypoxia and that autophagy plays a crucial role in lenvatinib efficacy and cell response to hypoxia through NRP1 modulation. Therefore, not only NRP1, but also HIF-1α, could act as potential targets in order to prevent therapeutic failure overpassing an adaptive cellular response through autophagy modulation.