All procedures involving laboratory mice were in accordance with national (Portaria 1005/92) and European regulations (European Directive 86/609/CEE) on animal experimentation and were approved by the Instituto Gulbenkian de Ciência Ethics Committee, and the Direcção-Geral de Veterinária (the Official National Entity for regulation of laboratory animals usage). The mice were bred and maintained in conventional housing facilities at the Instituto Gulbenkian de Ciência. All experiments were conducted using male mice with 8–15 weeks of age from strains C57BL/6, BALB/c, and B6. lpr (B6.MRL-Fas^lpr/J).

Mouse primary hepatocytes were prepared as previously described (17). In short, mouse liver lobes were perfused with liver perfusion medium and liver digest medium (Gibco; Invitrogen) at 37°C. Dissociated cells from the tissue were separated using a 1.12, 1.08, and 1.06 g/ml Percoll (GE Healthcare) gradient. Hepatocytes were collected from the gradient solution and cultured in William’s E complete medium (Gibco; Invitrogen) in plates or glass coverslips coated with 0.2% gelatine. The viability of hepatocytes in each experiment preparation was assessed by vital staining with Trypan blue and by cell morphology in FACS analysis. All the experiments were performed with viability of hepatocytes preparations above 80% purity. The remaining dead cells were removed 12 h post-plating (12 h prior infection), as in freshly isolated preparations dead hepatocytes do not attach. HGF conditioning was performed by adding 100 ng/ml of recombinant mouse HGF (eBioscience) 1 h prior infection, with a subsequent wash before infection.

Green fluorescent protein (gfp)-expressing P. berghei ANKA (Pb) (18), P. yoelii 17XNL (Py), and Pb-spect(-)1 (19) sporozoites were obtained after dissection of infected salivary glands from Anopheles stephensi mosquitoes bred in the insectarium of the Instituto de Medicina Molecular, Lisbon. Briefly, mosquitoes were immobilized, and the thorax and abdomen were separated with fine forceps. The salivary glands were dissected from the thoraces into RPMI medium, collected to a 1.5 ml Eppendorf, and smashed to release the sporozoites. The sporozoites were collected from the supernatant upon a 2 min × 5 min 800 rpm centrifugation (20). The same procedure was applied to non-infected mosquitoes to obtain salivary glands suspensions. Mice were intravenously inoculated with 10⁴ sporozoites each. Hepatocytes were infected with 1.5 to 3.5 × 10⁴ sporozoites, and non-infected controls were mock infected with salivary glands suspensions of non-infected mosquitoes.

Non-parenchymal cells were isolated from mouse livers at 40 h post-infection. Cells were obtained by adapting a method previously described (21). Briefly, liver lobes were removed and perfused with liver perfusion medium (Gibco; Invitrogen) with 750 mg/l of Collagenase H (Roche) at 37°C. The resulting suspension was filtered through a 100 µm cell strainer (BD Falcon). The dissociated cells were suspended in liver perfusion medium and centrifuged for 10 min at 1,500 rpm, resuspended in RPMI complete medium (Gibco; Invitrogen), mixed in Percoll (GE Healthcare) solution to give a final concentration of 30% Percoll, and then centrifuged at 2,000 rpm for 10 min. The cell pellet was resuspended in RPMI and carefully laid on 30% Percoll solution and centrifuged at 2,000 rpm for 10 min. The cell pellet collected was washed and resuspended in ACK (NH₄Cl 0.15 M, KHCO₃ 10 mM, Na₂EDTA⋅2H₂O 0.1 mM and pH 7.2) for 3 min to lyse remaining erythrocytes. Cells were washed and centrifuged at 800 rpm for 20 s to discard the remaining hepatocytes; the supernatant was recovered, and NPCs were collected at 1,500 rpm for 5 min.

Livers and NPCs were collected at 40 h post-infection, immediately homogenized in denaturing solution and total RNA was obtained using High Pure RNA Tissue Kit and High Pure RNA Kit (Roche). One microgram of total RNA was converted to cDNA (Transcriptor First Strand cDNA Synthesis Kit; Roche). Hepatocytes in culture were collected at 24 h post-infection, the lysis and reverse transcriptase reaction were performed using TaqMan Gene Expression Cell-to-CT kit (Ambion).

cDNA specific to P. berghei 18S rRNA was amplified by using the Taqman specific primers: IGC-Pb1 (forward) 5′-CCGATAACGAACGAGATCTTAACCT-3′, IGC-Pb2 (reverse) 5′-CGTCAAAACCAATCTCCCAATAAAGG-3′ and IGC-Pb3 (specific fluorogenic probe) 5′-FAM–ACTCGCCGCTAATTAG–MGB-3′ (7). cDNA specific to P. yoelii 18S rRNA was amplified by using the SYBR green system and the specific primers NYU-Py3 5′-GGGGATTGGTTTTGACGTTTTTGCG-3′ and NYU-Py5 5′-AAGCATTAAATAAA GCGAATACATCCTTAT-3′ (22). Hgf, Bid, Casp3, Hifα, Pik3r1, Cflar, Fadd, and Casp8 expression was quantified using TaqMan Gene Expression Assays from ABI (Mm01135185_m1, Mm00432073_m1, Mm01195084_m1, Mm00468872_m1, Mm01282780, Mm01255576_m1, Mm00438861_m1, and Mm00802247_m1, respectively). Endogenous control Gapdh (Mouse GAPD Endogenous Control; ABI) was used in multiplex PCR with target genes. PCR reactions were performed with ABI Prism 7900HT, and relative quantification was calculated by the ΔΔCt method, using the mean of the control group as calibrator.

Hepatocytes cultured on coverslips were fixed 24 or 40 h post-infection with 4% paraformaldehyde (PFA) and incubated with 50 nM NH₄Cl to extinguish PFA activity. Cells were permeabilized and blocked with blocking solution for 1 h [3% of Bovine Serum Albumin (Calbiochem), 100 mM of glycine, 0.1% of saponin, and 10% of Fetal Calf Serum (Gibco; Invitrogen) in PBS 1×] and stained for apoptosis by Terminal Deoxynucleotidyl Transferase dUTP Nick End Labeling (TUNEL) using “In Situ Cell Death Detection—TMR” (Roche); to stain EEFs coverslips were then incubated 45 min at room temperature with blocking solution containing anti-GFP IgG Alexa-488 conjugated antibody (Invitrogen); followed by DAPI staining (Invitrogen) and mounted with MOWIOL (Calbiochem); to stain FasL, cells were incubated 45 min at room temperature with blocking solution containing anti-FasL (CD178.1 Kay10; BD Pharmingen) followed by DAPI staining (Invitrogen) and mounted with MOWIOL (Calbiochem). Cell culture images were acquired using an automated Nikon Eclipse TE2000-S inverted fluorescence microscope, with a 20× objective, covering the total coverslip area. ImageJ (NIH) software was used to differentially identify TUNEL-positive or -negative (live) cells, EEFs and FasL in each coverslip, and to estimate single EEFs area. Results are expressed as the number of TUNEL-positive cells and live cells with EEFs per 100 microscopic fields, or FasL⁺ cells per 10³ cells.

Liver macrophages were stained, at 4°C, by incubating liver NPCs with F4/80 APC (A3-1; Serotec), CD11c PE (HL3; BD Pharmingen), and CD49b Biotin (DX5 and streptavidin-PerCP; BD Pharmigen) antibodies. Subsequently, a two-step HGF intracellular staining (H170; Santa Cruz and goat anti-rabbit IgG-FITC) was performed by using the BD Cytofix/Cytoperm kit. FACScalibur cytofluorometer (Becton Dickinson) and FlowJo software were used for FACS analysis. Purified Kupffer cells were obtained through high-speed cell sorting (FACSAria; Becton Dickinson), after staining with F4/80 APC (Serotec).

We used a 0.4-µm transwell culture system (Millipore) to allow purified Kupffer cells (upper chamber) to exchange soluble factors during 12 h with primary hepatocyte cultures (lower chamber), before infection. Hepatocytes were then infected with P. berghei ANKA sporozoites and fixed for parasite and apoptosis staining or collected for RNA analysis at 24 h post-infection. For HGF/MET signaling blocking, we added anti-Met antibodies (N-17; Santa Cruz Biotechnology) at 400 ng/ml when purified Kupffer cells were put in culture with primary hepatocytes followed by an additional 400 ng/ml added 1 h prior infection.

Hepatocyte growth factor secretion in the supernatants from transwell cultures was evaluated by sandwich ELISA. Briefly, maxisorp plates were coated with a capture anti-HGF antibody (H-170;, Santa Cruz) in 0.5 M Carbonate Bicarbonate Solution. Detection used anti-HGF-biotinylated antibody (BAF2207; R&D). Samples were quantified against a standard curve with mouse recombinant HGF (eBioscience).

Fas-blocking experiments were performed by using either a blocking antibody or a pan-caspase inhibitor. We used 10 µg/ml of anti-FasL antibody (CD178.1 Kay10; BD Biosciences) and 20 µM of Z-VAD-FMK (pan-caspase inhibitor – 550377; BD Biosciences), added 1 h prior infection and using the appropriate controls. Fas signaling was induced with 1 µg/ml of anti-Fas antibody (CD95 Jo2; BD Biosciences) 18 h post-infection with P. berghei. Cflar knockdown was achieved through a mixture of three distinct siRNAs (s63907, s63908, and s63909; Ambion), at a concentration of 20 µM that was delivered to primary hepatocytes with Oligofectamine (Invitrogen). Control samples included untransfected cells and cells transfected with a negative control siRNA (Ambion) or with a positive control targeting mouse Gapdh (Ambion).

All experiments were performed with a minimum of triplicate samples, and data are representative of at least three independent experiments. Results are presented as average values and error bars represent SD. Statistical analysis used a Mann–Whitney test, and p < 0.05 was considered statistically significant.

Liver NPCs, specifically Kupffer cells, have a role in the natural control of Plasmodium liver stage expansion (7). We found that Hgf mRNA expression in isolated non-parenchymal liver cells was highly increased upon in vivo infection with P. berghei sporozoites (Figure 1A) and correlated with relative resistance to parasite liver stage expansion in BALB/c as compared to C57BL/6 mice (Figure 1B). It has been reported that HGF and its mRNA localize exclusively to NPCs in the liver (23–25), and we have confirmed that expression of Hgf mRNA in primary hepatocytes was not detectable, irrespective to infection with P. berghei (data not shown). Given that HGF has been shown to facilitate liver stage infection (13, 14, 16), we investigated the cellular source of HGF in infected livers. FACS analysis of NPCs isolated from sporozoite-infected mice revealed that F4/80⁺ cells, representing Kupffer cells, expressed intracellular HGF as opposed to other NPCs, e.g., CD11c⁺ (dendritic cells) or CD49b⁺ cells that include NKT and NK cells (Figure 1C). Furthermore, we found that depletion of liver macrophages abrogated the augmented expression of HGF in the liver induced by infection (Figure S1 in Supplementary Material) suggesting that liver macrophages are the main in vivo source of HGF in malaria liver stage infection. To untangle the effect of HGF in Plasmodium liver stage infection, we pre-treated hepatocyte primary cultures of C57BL/6 and BALB/c mice with HGF 1 h prior infection and observed that HGF conditioning similarly impairs parasite yield in both mouse strains (Figure 1D). Moreover, HGF pre-treatment of C57BL/6 hepatocytes infected with P. yoelii (Figure 1E) or P. berghei-spect(-)1 (Figure 1F) revealed that reduction of parasite yield induced by HGF transcends Plasmodium species and is independent of hepatocyte traversing (19). In addition, we found that reduction of parasite burden by HGF was attributable to a significant decrease in the number of hepatocytes carrying EEFs (Figure 1G), detectable from 24 h post-infection onward. In contrast, HGF conditioning was not associated with altered EEFs size, neither at earlier stages (24 h) (Figure 1H) nor at the end of the liver infection (40 h) (Figure 1I). These results show that HGF conditioning reduces the number of hepatocytes that carry EEFs, rather than affecting its development suggesting that Kupffer cell-derived HGF impacts on the survival of infected hepatocytes.

We investigated whether the reduced number of developed EEFs upon HGF pre-treatment was due to apoptosis of infected hepatocytes. Analysis of mRNA expression of key informative genes showed that HGF induces downregulation of PI3K a component of the HGF/c-Met pathway that is involved in apoptosis protection (26, 27) (Figure 2A). Moreover, HGF downregulated Hif1-alpha that has been associated with hepatocyte apoptosis and efficiency of P. berghei infection (28, 29) (Figure 2B). Furthermore, upregulation of Caspase 3 (Figure 2C), with a concomitant decrease of Bid expression (Figure 2D), suggests that HGF acts through a mitochondrial-independent apoptosis pathway in response to hepatocyte Plasmodium infection (30).

Interestingly, we observed that HGF pre-treatment specifically increased apoptosis susceptibility in parasitized hepatocytes but did not affect bulk apoptosis in the infected cultures (Figures 2E,F), as detected by TUNEL assays. Remarkably, in HGF pre-treated cultures, the size distribution of EEFs inside apoptotic hepatocytes significantly deviated to higher values (p-value <0.0001) (Figure 2G). Thus, in untreated infected cultures, hepatocytes carrying small EEFs were the major target of spontaneous apoptosis, but under HGF conditioning significant apoptosis was observed in infected hepatocytes with larger EEFs (Figures 2G,H). Collectively, these data indicate that parasitized primary hepatocytes respond to HGF treatment through a mitochondrial-independent apoptosis mechanism, which operates in hepatocytes carrying early liver schizont stages that are refractory to spontaneous apoptosis.

We ask whether HGF secreted by Kupffer cells could impair hepatocyte infection in vitro. We performed transwell co-culture experiments using FACS sorted F4/80⁺ cells from infected livers and naïve primary hepatocytes. Remarkably, we observed that parasite yield and the number of infected hepatocytes were reduced in the presence of Kupffer cells, to an extent comparable to the cultures pre-treated with HGF (Figures 3A,B), indicating that this effect is attributable to soluble factors derived from Kupffer cells. Moreover, HGF was detected in the supernatants of Kupffer cells/hepatocyte cultures (Figure 3C), which were able to reduce parasite yield when used as media in pure hepatocyte cultures (Figure 3D). Furthermore, adding anti-MET antibodies in the co-cultures to block the action of Kupffer cells-derived HGF led to increased parasite yield in hepatocytes (Figure 3E). Altogether, these results strongly suggest that HGF is secreted by Kupffer cells and acts in an endocrine fashion to reduce successful progression of infection in parasitized hepatocytes.

Additionally, it has been described that the failure of EEFs to attain full development in vitro is attributable to an hepatocyte autonomous mitochondrial-dependent apoptosis response (10). In line with these observations, we asked whether reduction of the number of EEFs in infected primary hepatocyte cultures occurs as well in the absence of HGF. To ascertain the success of P. berghei sporozoites infection in hepatocyte primary cultures, we performed time course counting of infected hepatocytes, which showed that less than 50% of the initial infected cells attained EEF full development (Figure 4A). To investigate whether the aborted EEFs were related to apoptosis induction in P. berghei-infected cultures, we analyzed the expression of apoptosis genes in hepatocyte primary cultures and found out that Fas pathway signaling-related genes were significantly altered at 24 h p.i. A strong upregulation of Bid coupled with a slight upregulation of other proapoptotic genes and downregulation of the anti-apoptotic Cflar, represent a signature that Fas pathway signaling was induced in infected primary cultures (Figure 4B), presumably executing a Type II apoptotic response that requires mitochondrial participation (31). To test whether Fas pathway signaling was detrimental to successful hepatocyte infection, we analyzed primary hepatocytes from lpr mice (Fas-null mutant), which demonstrated that at 24 h p.i. the parasite yield was significantly increased in the absence of Fas pathway signaling. An effect that was also observed by blocking Fas pathway signaling with anti-Fas Ligand antibodies or with a pan-Caspase inhibitor (Figure 4C). Furthermore, the absence of Fas pathway signaling improved the number of successfully infected hepatocytes (Figure 4D), possibly because apoptosis induced by infection was partially dependent on Fas pathway signaling (Figure 4E). Interestingly, immunofluorescence microscopy of infected cultures revealed prominent FasL expression in non-infected hepatocytes as compared to infected hepatocytes (Figure 4F), indicating that induction of FasL expression by P. berghei infection (Figure 4G) impacts the viability of both infected and non-infected hepatocytes. Conversely, enhancing Fas pathway signaling in hepatocytes either by adding anti-Fas antibodies or by knocking down Cflar, a potent Fas pathway inhibitor (32), increased apoptosis susceptibility in infected hepatocyte cultures (Figures 4H,K). In these conditions, we observed decreased number of hepatocytes containing EEFs (Figures 4I,L) and reduced parasite yield (Figures 4J,M). The effects of Cflar knockdown were exacerbated by anti-Fas stimulation (Figures 4K–M) and strongly suggest that Cflar inhibition sensitized hepatocytes to undergo apoptosis through the Fas pathway. These experiments provide evidence that apoptosis signals evoked by the parasite favor Fas-dependent hepatocyte apoptosis in infected cultures leading to a decrease in the number of infected primary hepatocytes that support full parasite development.