A process similar to autophagy is associated with cytocidal chloroquine resistance in Plasmodium falciparum

Abstract

Resistance to the cytostatic activity of the antimalarial drug chloroquine (CQ) is becoming well understood, however, resistance to cytocidal effects of CQ is largely unexplored. We find that PfCRT mutations that almost fully recapitulate P. falciparum cytostatic CQ resistance (CQR(CS)) as quantified by CQ IC50 shift, account for only 10-20% of cytocidal CQR (CQR(CC)) as quantified by CQ LD50 shift. Quantitative trait loci (QTL) analysis of the progeny of a chloroquine sensitive (CQS; strain HB3)×chloroquine resistant (CQR; strain Dd2) genetic cross identifies distinct genetic architectures for CQR(CS) vs CQR(CC) phenotypes, including identification of novel interacting chromosomal loci that influence CQ LD50. Candidate genes in these loci are consistent with a role for autophagy in CQR(CC), leading us to directly examine the autophagy pathway in intraerythrocytic CQR parasites. Indirect immunofluorescence of RBC infected with synchronized CQS vs CQR trophozoite stage parasites reveals differences in the distribution of the autophagy marker protein PfATG8 coinciding with CQR(CC). Taken together, the data show that an unusual autophagy-like process is either activated or inhibited for intraerythrocytic trophozoite parasites at LD50 doses (but not IC50 doses) of CQ, that the pathway is altered in CQR P. falciparum, and that it may contribute along with mutations in PfCRT to confer the CQR(CC) phenotype.

Figure 1. LD₅₀ vs IC₅₀ directed QTL analyses for CQR HB3 × Dd2 cross progeny.

A) IC₅₀ QTL scan for CQR progeny shows a peak on chr5 (asterix) that encompasses pfmdr1 as previously described . Notably, the chr6 locus that is pertinent for the LD₅₀ scan (see Fig. 1B) does not pass the suggestive threshold on this IC₅₀ scan. B) LD₅₀ QTL scan for CQR progeny shows a peak on chr6 (asterix; L.O.D. = 2.5, passing the suggestive threshold). The locus that includes pfmdr1 does not pass the suggestive threshold for this scan (see also Fig. 1C). C) To more clearly highlight the differences in genetic architecture for LD₅₀ vs IC₅₀ phenotypes, an overlay of the two QTL scans is shown. The CQR progeny LD₅₀ QTL scan is shown in blue, while the CQR progeny IC₅₀ QTL scan is in black. The overlay shows quite clearly that the pfmdr1 locus does not factor at all into the LD₅₀ phenotype, and that the chr6 locus does not factor at all into the IC₅₀ phenotype. Thus the IC₅₀ and LD₅₀ phenotypes are genetically distinct. D) Similarly, the interaction locus between chr6 & 8 (see text) does not appear on a IC₅₀ pair-wise scan. However, additive effects between chr5 and chr7 loci are seen, as previously reported . E) Pair-wise scan of the CQR progeny shows that chr6 and chr8 loci (circle) have additive effects on LD₅₀ (L.O.D. = 4.3).

Figure 2. PfATG8– positive puncta.

Shown are puncta for (A) control HB3 iRBC grown under normal culture conditions (B) HB3 iRBC grown for 6 hours under starvation conditions (see Methods) and (C) HB3 iRBC grown under starvation conditions plus the autophagy inhibitor 3 methyl adenine (3 MA). Shown are transmittance (left), immunofluorescence vs antiPfREX1 (Maurer’s cleft marker; red; second column), immunofluorescence vs antiTgATG8 (cross reacts with PfATG8; green, third column) and overlays (right). Bar = 5 µm. Also shown (D) are western blot data for iRBC harboring HB3 (CQS) and Dd2 (CQR) trophozoites grown under control culture conditions. Two separate gels for two independent sets of samples (two iRBC isolations for each culture) are shown. We note our data show a clear doublet at 15 and 17 kDa, similar to all other studies of eukaryotic ATG8 protein of which we are aware except one , which resolves only a single band instead of the usual doublet with a polyclonal antisera raised against a recombinant GST-PfATG8 fusion. We suggest three possible reasons for the discrepancy: 1) we use higher density [15% acrylate] gels relative to in order to resolve the low mass doublet, 2) we do not solubilize parasites with saponin as in which would release de - lipidated ATG8 into wash supernatant, 3) perhaps abundance of one PfATG8 species (presumably de lipidated) is higher in trophozoites relative to schizonts examined in .

Figure 3. Comparison of antisera vs monoclonal Ab staining.

Shown are results for TgATG8 antisera (bottom row) vs staining using a monoclonal antibody raised vs a highly conserved Apicomplexan ATG8 epitope (top row, see methods). Panels A,G,D,J are transmittance, B,H,E,K are ATG8 fluorescence, and C,I,F,L are overlay, respectively. Left side is control intraerythrocytic HB3 P. falciparum; right side is HB3 starved for 6 hrs as described in methods. Bar = 5 µm.

Figure 4. PfATG8 (green) re distribution, CQ vs time.

CQS (strain HB3; top panel A) and CQR (strain Dd2; bottom panel B) parasites were grown in control media (left column) or in media plus 2×LD₅₀ dose of CQ (250 nM for HB3 Fig. 4A, 30 µM for Dd2, Fig. 4B) for 2, 4 or 8 hrs (column 2,3,4 respectively). The top row in panels A and B is transmittance, the second (green) is staining for PfATG8, the third is DAPI to visualize the single trophozoite nucleus, and the bottom row for each panel presents the overlay. Fluorescence acquired at 35% power, 200 ms, 642 nm; emission 700/75 nm dichroic, 690 nm cut-off. Bar = 5 µm.

Figure 5. Semi automated computational method for quantifying the distribution of ATG8 puncta relative to hemozoin crystals within the DV.

(A) Hemozoin within the DV is detected by transmittance, the outline of <10% transmittance is defined (labeled “DV”, left panel) and the center of this outline then defined by voxel analysis using Imaris software. (B) The center of distinct ATG8 puncta (bright green dots) are labeled using the option ‘Add new measurement points’ in the Imaris “spots” subroutine. (C) Distances from the hemozoin center to the ATG8 puncta centers are then computed (white lines) and data collated in excel.

Figure 6. Quantified PfATG8 puncta distribution for synchronized trophozoite parasites.

Shown are CQS HB3 (top) CQR Dd2 (middle) and transfectant C4^Dd2 (bottom) under different conditions. Far left, “CM” = control culture conditions, far right, “SM” = iRBC in starvation media for 6 hr. In between are puncta quantified for iRBC treated for 6 hrs with the indicated [CQ]. Black bars in each panel denote 2×IC₅₀ [CQ] for the strain, hashed bars denote 2×LD₅₀ [CQ] for the strain. Data are the average of at least 20 iRBC, +/− s.d., and puncta that are ≥3.5 µm from DV hemozoin optical density are plotted. The three phenotypes are distinct, as evidenced by statistical comparison of results upon 250 nM and 3.6 µM treatments common to all three strains; for 3.6 µM data, HB3 vs Dd2, HB3 vs C4Dd2, and Dd2 vs C4Dd2, P value <0.05 in each case. For 250 nM data, HB3 vs Dd2 or HB3 vs C4Dd2, P value <0.05, but Dd2 vs C4Dd2>0.05. That is, as proposed in the text, C4Dd2 is intermediate relative to HB3 and Dd2: both Dd2 and C4Dd2 differ from HB3, but C4Dd2 shows behavior similar to Dd2 at lower dose CQ, but different behavior vs Dd2 at higher dose.

Figure 7. Quantified PfATG8 puncta at ≥3.5 µm from hemozoin for synchronized trophozoites.

Two additional CQR and two additional CQS strains are analyzed. Shown are puncta counts for at least 20+/− s.d. grown under control conditions (“CM”, left side each panel), upon starvation (SM; far right, each panel) and upon dosing for 6 hrs with either 2×IC₅₀ or 2×LD₅₀ concentrations of CQ (50 nM and 250 nM for CQS strains, and 250 nM and 32 µM for CQR strains, respectively).