Pnas 202302152
Pnas 202302152
Edited by L. Sibley, Washington University in St Louis, St. Louis, MO; received February 14, 2023; accepted March 20, 2023
The primary antigenic and virulence determinant of the human malaria parasite
Plasmodium falciparum is a variant surface protein called PfEMP1. Different forms Significance
of PfEMP1 are encoded by a multicopy gene family called var, and switching between
active genes enables the parasites to evade the antibody response of their human The parasites that cause malaria
hosts. var gene switching is key for the maintenance of chronic infections; however, avoid destruction by the human
what controls switching is unknown, although it has been suggested to occur at a immune response by
constant frequency with little or no environmental influence. var gene transcription is systematically changing the
controlled epigenetically through the activity of histone methyltransferases (HMTs). antigens they expose on the
Studies in model systems have shown that metabolism and epigenetic control of gene infected cell surface. This
expression are linked through the availability of intracellular S-adenosylmethionine process, called antigenic
(SAM), the principal methyl donor in biological methylation modifications, which
variation, is responsible for the
can fluctuate based on nutrient availability. To determine whether environmental
chronic nature and virulence of
conditions and changes in metabolism can influence var gene expression, P. falci-
parum was cultured in media with altered concentrations of nutrients involved in the disease. The mechanisms
SAM metabolism. We found that conditions that influence lipid metabolism induce enabling parasites to undergo
var gene switching, indicating that parasites can respond to changes in their envi- antigen switching remain
ronment by altering var gene expression patterns. Genetic modifications that directly unknown; however, it has been
modified expression of the enzymes that control SAM levels similarly led to profound assumed that this process occurs
changes in var gene expression, confirming that changes in SAM availability mod- stochastically over the course of
ulate var gene switching. These observations directly challenge the paradigm that an infection. Here we show that
antigenic variation in P. falciparum follows an intrinsic, programed switching rate, parasites can sense changes in
which operates independently of any external stimuli. their environment and undergo
methylation | metabolism | gene expression | var genes | immune evasion
antigen switching in response to
those changes. This implies a
Malaria parasites cause disease through invasion and replication within the circulating red much more sophisticated
blood cells of their vertebrate hosts. A defining feature contributing to the pathogenicity interaction between host and
of Plasmodium falciparum is its ability to express a series of cytoadherence proteins on the parasite than was previously
surface of the iRBC. P. falciparum erythrocyte membrane protein 1 (PfEMP1) is anchored appreciated and suggests that
in knob-like structures within the iRBC membrane, extending into the extracellular space parasites have evolved to
and facilitating cytoadherence to vascular endothelial receptors, thereby allowing the undergo antigenic switching
parasite to sequester within the microvasculature of various organs (1). Variant forms of
when needed, for example in
PfEMP1 are encoded by a large, multicopy gene family called var (2–4). Each P. falciparum
response to immune recognition.
genome possesses about 60 var genes, located primarily within subtelomeric regions of
the chromosomes (5, 6).
var genes are expressed in a mutually exclusive manner such that each parasite only
expresses a single gene over many cycles of asexual replication (7). Over the course of an
infection, the host’s immune system mounts an antibody response to the predominately
expressed variant of PfEMP1, drastically reducing the circulating parasite population, Author contributions: V.M.S., B.F.C.K., and K.W.D.
designed research; V.M.S., J.E.V., C.T.H., F.F., E.H., X.Z.,
often to undetectable levels. However, subpopulations of parasites expressing an alternative and M.R.G. performed research; K.Y.R., C.B.M., and
var gene can expand to high parasitemia, thereby leading to recurrent waves of parasites B.F.C.K. contributed new reagents/analytic tools; V.M.S.,
J.E.V., E.H., X.Z., B.F.C.K., and K.W.D. analyzed data; and
and enabling the infection to persist for a year or more (7, 8). The transcriptional activation V.M.S., J.E.V., C.T.H., F.F., C.B.M., B.F.C.K., and K.W.D.
and silencing of var genes are controlled epigenetically with transcriptionally silent var wrote the paper.
genes found in a state of condensed heterochromatin, whereas the single active var gene The authors declare no competing interest.
is maintained in a transcriptionally permissive euchromatic state (9–13). Three marks This article is a PNAS Direct Submission.
have been found to be of primary importance in var gene regulation. Nucleosomes span- Copyright © 2023 the Author(s). Published by PNAS.
This article is distributed under Creative Commons
ning the promoter region of the single active var gene are acetylated at the H3K9 position Attribution-NonCommercial-NoDerivatives License 4.0
(Histone 3, Lysine 9; H3K9ac), while those at silent var genes are tri-methylated (CC BY-NC-ND).
(H3K9me3). Conversely, the H3K36me3 mark is found at both active and silent var 1
To whom correspondence may be addressed. Email:
[email protected].
genes (14, 15). Di- and tri-methylation of H3K4 are enriched at actively transcribed genes
This article contains supporting information online at
in many organisms and similarly are found around the single var promoter marked for https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.
activation (13). Methyl groups at each of these positions are deposited by specific histone 2302152120/-/DCSupplemental.
methyltransferases (HMTs), which in P. falciparum are largely SET-domain-containing Published April 17, 2023.
2 of 12 https://doi.org/10.1073/pnas.2302152120 pnas.org
Fig. 1. Schematic drawing of the metabolic pathways resulting in the production of the lipid phosphatidylcholine (PtdCho, green) in P. falciparum infected red
blood cells. The inner, dark red circle represents the parasite, the larger circle surrounding the parasite represents the infected red blood cell (iRBC), while the
white background represents the extracellular environment. Uptake of three key nutrients is highlighted, choline and serine (blue) and methionine (Met, red).
Two alternative pathways for the production of CDP-choline are shown; the serine decarboxylase phosphoethanolamine methyltransferase (SDPM) and the
cytidine diphosphate (CDP)-choline pathway (also known as the Kennedy pathway). Importantly, the SDPM pathway requires the enzyme phosphoethanolamine
N-methyltransferase (PfPMT) and consumes three molecules of S-adenosylmethionine (SAM, red) for the production of phosphocholine. The role of SAM in
histone methylation by histone methyltransferases (HMTs) and demethylation by histone demethylases (HDMs) is highlighted in orange. The conversion of Met
to SAM by SAM synthase (PfSAMS) and the conversion of S-adenosylhomocysteine (SAH) to homocysteine by SAH hydrolase (PfSAHH) are also shown.
each culture after 2 and 4 wk and analysis of var expression was serine, which in turn can affect SAM pools via SAM consumption
performed using a standardized, qRT-PCR that determines the by PfPMT. In contrast, methionine, taken up by P. falciparum
expression level of each individual var gene (18). In addition, either from human serum or the digestion of hemoglobin, can
intracellular metabolites were extracted from parasites and meas- be used directly by the parasite to synthesize SAM (Fig. 1). If
ured by LC-MS as previously described (56). This enabled us to the changes in var gene expression induced by depletion of
directly determine if the changes in the concentrations of serine serine and/or excess choline are in fact due to an increase in
and choline in the media were reflected by changes of these nutri- SAM pools (Fig. 2A), we hypothesized that conditions of excess
ents within the parasites and if the changes affected SAM levels. methionine would have a similar effect. To test this hypothesis,
As hypothesized, alterations in the levels of choline and serine in we grew parasites in culture media in which the availability of
the culture medium greatly accelerated var switching in compar- methionine was manipulated (SI Appendix, Table S1). Similar
ison to parasites cultured under standard nutrient conditions to serine, methionine is present in hemoglobin; therefore, it
(Fig. 2A and Supplementary data). Specifically, depletion of serine can only be depleted in the growth medium and cannot be
led to an increase in intracellular SAM and accelerated activation completely eliminated from the culture. var expression patterns
of var2csa expression. This effect was greatly enhanced when com- were analyzed by Q-RT-PCR after 2 and 4 wk of growth in the
bined with addition of excess choline. Depletion of both choline modified media and intracellular metabolites were measured
and serine, or excess serine did not result in activation of var2csa using LC-MS. Under conditions of depleted methionine,
and instead mimicked the expression pattern observed in parasites var expression closely resembled that observed in parasites
grown in standard media. No significant growth phenotype was grown in standard media (Fig. 3A). In contrast, growth in the
observed when culturing parasites under these modified condi- presence of excess methionine led to a measurable increase in
tions (SI Appendix, Fig. S2). SAM Levels and a coordinated switch to var2csa, similar to the
The conditions that induced changes in var gene expression, conditions of excess choline (Fig. 3 and replicate experiment
excess choline and/or depleted serine, are both predicted to shift shown in Supplemental Data). No significant growth phenotype
lipid synthesis away from the SDPM pathway and toward the was observed when culturing parasites under these modified
CDP pathway (Fig. 1), thus reducing SAM consumption and conditions (SI Appendix, Fig. S2). These experiments, coupled
potentially increasing the availability of SAM for HMT activity. with those in which the levels of choline and serine were altered,
This in turn could destabilize epigenetic regulation of var expres- further suggest that conditions that elevate SAM pools accelerate
sion and trigger a transcriptional switch over time. These results var gene switching, specifically activation of var2csa.
phenocopy previous results in which the activity of specific HMTs
was directly manipulated (15, 27), which similarly let to a coor- Increasing Expression of PfSAMS Induces var Switching. The
dinated switch to var2csa. While the shifts in var gene expression changes in var gene expression observed upon altering the
to var2csa were dramatic, the metabolomic analysis indicates that availability of choline, serine or methionine correlated with
the changes in SAM and SAH levels were subtle although repro- changes in intracellular SAM levels, although other possible effects
ducible (Fig. 2B), consistent with the hypothesis that even small of nutrient availability were also possible. We therefore sought to
fluctuations in SAM/SAH levels can have significant effects on more directly test how changes in intracellular SAM can affect var
HMT activity and gene expression (32–34). gene expression. In eukaryotes, SAM synthetase (SAMS) catalyzes
the formation of SAM from methionine and ATP, while SAH
Excess Methionine Induces var Transcriptional Switching. As hydrolase catalyzes the reversible hydrolysis of SAH to adenosine
shown in Fig. 1, the balance between the CDP and SDPM and homocysteine (Fig. 1). Homocysteine is then secreted from
pathways can be influenced by the availability of either choline or the infected cell. The activity of SAMS can directly raise the
1.0
6.0
1.0
2 weeks 4 weeks 4.0 1.0
depleted choline 0.5
0.5
20X serine 2.0 0.5
0 0 0 0
1.0 1.0
1.0
2 weeks 4 weeks 0.5
1.0
0.5
1X choline 0.5
0.5
depleted serine
0 0 0 0
1.0
1.5 1.0 1.5
2 weeks 4 weeks
20X choline 1.0 1.0 0.5
0.5
depleted serine 0.5 0.5
0 0 0 0
Fig. 2. Effect of changes in serine and choline availability on var gene expression. (A) Pie charts representing var gene expression profiles in parasite populations.
The expression level of each individual var gene was determined using qRT-PCR using a standardized protocol originally described by Salanti et al. (18). Each slice
of the pie represents the expression level of a specific var gene. A single initial population of parasites primarily expressing the var gene Pf3D7_0421100 (orange)
was divided into parallel cultures and exposed to different concentrations of choline and serine as described in SI Appendix, Table S1. var expression profiles were
determined after 2 and 4 wk of culture. Expression of var2csa is shown in blue. (B) Intracellular abundance of choline (Cho), serine (Ser), S-adenosylmethionine
(SAM) and S-adenosylhomocysteine (SAH) are shown at the right of each condition.
methylation index (SAM/SAH ratio) by catalyzing the production To determine whether an increase in intracellular SAM leads
of SAM, and this enzyme is conserved in Plasmodium and encoded to changes in var gene expression, the PfSAMS-HA overexpres-
by a single copy gene (PfSAMS: PF3D7_0922200). To more sion construct was transfected into the same 3D7 parent clone
directly test the hypothesis that var switching is affected by the described above and the parasites assayed for var expression
levels of intracellular SAM, we altered the expression of PfSAMS both “on” and “off” Atc. A line transfected with an empty vector
through the use of an overexpression construct, applying the RNA (EV) served as a control. Remarkably, the overexpression line
aptamer system adapted for Plasmodium by Ganesan et al. (57). displayed profound changes in var expression compared to the
This system enables regulated expression of an episomally encoded parent population or the empty vector control (Fig. 4E).
protein through the placement of Tet repressor protein (TetR) Specifically, the culture had switched to expression of var2csa
aptamers in the 3′ untranslated region (UTR) of the transcript. at the initial timepoint, as soon as the culture grew to a sufficient
In the presence of anhydrotetracycline (Atc), the transcript is parasitemia to recover RNA. Switching to var2csa was near
stabilized, enabling overexpression of the protein of interest (in absolute, with very little expression of alternative var genes
addition to normal expression from the chromosomal copy) detectable in the population. This switch occurred even in the
(57) (Fig. 4A). The specific vector used here (pMG68 supplied absence of induction by Atc, suggesting that even leaky expres-
by Dr Jacquin Niles) expresses an epitope-tagged (HA) protein. sion from the episome was enough to robustly alter the epige-
The PfSAMS-HA overexpression construct (OE) was transfected netic control of var genes. The EV control line showed no
into NF54 parasites and overexpression verified by Western difference in var expression before or after application of Atc
blot probed with anti-SAMS antibodies. This enabled detection (Fig. 4E). An additional replicate is shown in Supplementary
of both the wild-type and HA-tagged proteins (Fig. 4B). This Data. It is noteworthy that the change in intracellular SAM
analysis showed leaky expression of SAMS-HA coming from the observed here was much greater than what resulted from altering
plasmid prior to induction, but significantly greater expression choline, serine or methionine levels in the media and that the
in the presence of Atc. Quantification of total SAMS expression shift to var2csa was also much more profound, providing addi-
(WT + SAMS-HA) showed a nearly 2.5-fold increase in SAMS tional evidence that it is the changes in SAM that are responsible
expression in the presence of Atc (Fig. 4C). To determine whether for alterations in var gene expression.
PfSAMS overexpression leads to an increase in intracellular SAM, Across three replicates, the PfSAMS OE did not grow at a
metabolites were once again extracted from parasites and measured visibly different rate than untransfected WT parasites or the EV
by LC-MS (56). This analysis showed that SAMS overexpression transfected line (SI Appendix, Fig. S2). An addition of 0.5 mM of
induces a significant increase in intracellular SAM, as expected, Atc also did not detectably alter growth. These data further sup-
while levels of SAH did not change significantly (Fig. 4D). port the hypothesis that conditions that increase the ratio of SAM/
4 of 12 https://doi.org/10.1073/pnas.2302152120 pnas.org
A
2 weeks Pf3D7_0421100
complete
media var2csa Fig. 3. Effect of changes in methionine
B Met SAM SAH
availability on var gene expression. (A) Pie
charts representing var gene expression
1.0
1.0
profiles in parasite populations with each
1.0 slice of the pie representing the expression
2 weeks 4 weeks
0.5 level of a specific var gene. A single
depleted 0.5
0.5
initial population of parasites primarily
methionine
expressing the var gene Pf3D7_0421100
0 0 0
(orange) was divided into parallel cultures
1.5
and exposed to different concentrations
30 of methionine as described in SI Appendix
2 weeks 4 weeks 1.5 1.0
Table S1. var expression profiles were
20
excess 1.0
determined after 2 and 4 wk of culture.
methionine 10
0.5
0.5 Expression of var2csa is shown in blue.
0 0 0 (B) Intracellular abundance of methionine
(Met), S-adenosylmethionine (SAM) and
complete media
S-adenosylhomocysteine (SAH) are shown
modified media at the right of both conditions.
SAH alter var expression, and that this switch is uniquely to Growth over a 4-wk period either on or off GlcN did not induce
var2csa. a switch to var2csa across any of the three clones. To ensure that
the addition of GlcN alone does not alter var expression, an
PfSAMS Overexpression Lines Revert to Wild-Type upon Loss untransfected (WT) 3D7 line was grown in parallel on and off the
of Transgene Expression. The PfSAMS overexpression plasmid same concentration of GlcN (Fig. 5F). No significant change in
carries the blasticidin S deaminase gene (58), allowing for selection var expression was observed. Intriguingly, all three PfSAMS-glms
and maintenance of the episome through the inclusion of the clones displayed virtually identical, heterogenous var expression
antibiotic blasticidin S in the culture media. Transfected episomes patterns (high expression levels of six var genes: PF3D7_1000100,
in P. falciparum require continuous selection for maintenance, and PF3D7_0412400, PF3D7_0412700, PF3D7_0632500, PF3D7_
therefore, the growth in the absence of selection pressure leads 0809100, and PF3D7_0600200) at all time points (Fig. 5 D and
to rapid shedding of the plasmid by the parasites. The PfSAMS E and Supplemental Data) suggesting that var switching rates
overexpression line displayed a profound switch to expression of might be reduced in these lines, and the overall level of var
var2csa immediately after transfection (Fig. 4E), likely due to leaky expression was much greater when compared to wild-type parasites
expression of the transgene from the episome, even in the absence (SI Appendix, Fig. S3). Together with the nutrient manipulation
of Atc (Fig. 4B). To ensure that switching to var2csa was a result experiments and the PfSAMS overexpression studies, these results
of PfSAMS overexpression and not due to some other aspect of indicate that conditions that increased the intracellular level of
the transfection process, this line was grown in the absence of SAM/SAH are uniquely capable of inducing switching to var2csa.
blasticidin S selection pressure for 2 mo and the var expression Growth assay results for two PfSAMS-glms clones are shown in
profile determined. Representative results are shown in Fig. 4E SI Appendix, Fig. S2. Interestingly, the PfSAMS-glms lines appear to
(and an additional replicate is shown in the Supplemental Data). have a positive growth phenotype, growing at a slightly faster rate
The line displayed a shift in var expression away from var2csa than an untransfected WT line. An addition of 2.5 mM GlcN caused
when drug selection was removed and Q-RT-PCR analysis showed a slight reduction in growth rate for both the WT and PfSAMS-glms
that PfSAMS mRNA had returned to the levels found in the lines.
empty vector control (Fig. 4F), indicating that the stable var2csa
expression phenotype observed in the transfected line was indeed Changes in SAM Metabolism Result in Alterations in Histone
a result of the increased PfSAMS expression. Methylation. Our manipulation of nutrient availability or PfSAMS
expression altered levels of SAM, the methyl donor for the histone
Reducing PfSAMS Expression Does Not Induce var Switching. To methyltransferases involved in epigenetic regulation of gene
assess whether a decrease in the levels of SAM/SAH could affect expression. To determine whether manipulating SAM levels resulted
var gene expression, we assayed var expression profiles in three in changes in histone methylation, we performed whole genome
clones of a PfSAMS-glmS line previously described by Harris “cut and run” analysis (60) to examine levels of H3K9me3 and
et al (bioRxiv 2022.01.18.476397) in which the level of PfSAMS H3K4me3, histone methylation marks directly implicated in var
expression can be knocked down, leading to a reduction in SAM gene regulation (9, 61). Specifically, we compared parasites grown
levels by ~60%. This line was generated through integration under conditions that increase SAM availability (excess choline) to
of the glmS ribozyme into the 3′ UTR of the PfSAMS locus, parasites grown under conditions of depleted choline and methionine.
allowing for post-transcriptional regulation of gene expression as These experiments were performed 26 h after the parasites were
previously described by Prommana et al. (59). When glucosamine exposed to the altered media conditions, prior to any detectable
(GlcN) is added, the ribozyme is cleaved leading to degradation changes in var gene expression, thus enabling us to observe changes
of the mRNA (Fig. 5A) and a corresponding reduction in protein in histone modifications that are a direct response to altered nutrient
expression. In three clones, expression of PfSAMS was found to availability rather than any adaptive or long-term changes that might
be reduced by about half after application of GlcN (2.5 mM) occur over extended growth under these conditions. Whole genome
relative to those in the absence of GlcN (0 mM) (Fig. 5 B and C). analysis identified distinct increases of histone methylation at regions
var expression profiles were determined for each line, in both throughout the genome known to be enriched in heterochromatin.
the presence and the absence of GlcN. Representative results Four representative examples that display a variety of gene families
for one PfSAMS-glms clone are shown in Fig. 5 D and E with are shown in Fig. 6, including a subtelomeric region of chromosome
data from additional clones provided in the Supplemental Data. 2, an internal var tandem array from chromosome 4, the clag3.1/
2 2
st
st
to
la
la
0
ec
B
B
- blasticidin
SA ty V
ff
SA On
ATc - + - + - + - +
O
2 months
S
Em
M
after transfection -ATc +ATc
Pf
Pf
Pf3D7_0421100 var2csa
Fig. 4. Effect of overexpression of PfSAMS on var gene expression. (A) Schematic representation of the TetR-DOZI system for protein overexpression. This construct
was transfected into cultured parasites and maintained as an episome, allowing inducible expression of an HA-tagged version of the protein through the addition of
anhydrotetracycline (ATc) to the culture media. (B) Western showing expression of the HA-tagged version of PfSAMS in transfected, late-stage parasites. Expression of
the protein is detectable even in the absence of ATc, although expression increases dramatically 36 h after addition of ATc. Antibodies against Hsp70, HA and SAMS were
used for detection. (C) Quantitation of total PfSAMS expression in the presence or absence of ATc as detected by Western blot. (D) Intracellular levels of SAM (Left) and
SAH (Right) in wild-type and PfSAMS overexpressing parasites with and without the addition of ATc to the media. Intracellular SAM and SAH levels were determined by
LC-MS and are displayed as the calculated area under the signal curve per 1 × 103 parasites for each metabolite. (E) Pie charts representing var gene expression profiles
in parasite populations with each slice of the pie representing the expression level of a specific var gene. Top: An initial population of parasites primarily expressing the
var gene Pf3D7_0421100 (orange) was transfected with an empty vector, then var gene expression profiles determined after growth in the presence or absence of ATc
for 2 wk. Bottom: var gene expression profiles are shown for a population of parasites immediately after transfection with the PfSAMS overexpression construct (Left)
as well as after 2 wk of growth in the presence or absence of ATc (middle pies). Growth in the absence of blastidin for 2 mo (Right) results in the loss of the expression
vector and a shift away from var2csa expression (blue). (F). Q-RT-PCR assays showing total transcript levels of PfSAMS in parasites transfected with an empty vector,
with the PfSAMS overexpression construct and grown in the presence of ATc, or after removal of blasticidin from the media for 2 mo. Removal of blasticidin selection
allows PfSAMS expression levels to return to those of the empty vector control. *P < 0.05, **P < 0.01, ***P < 0.005, ns P > 0.05, paired student’s t test.
clag3.2 locus from chromosome 3 and the subtelomeric region of were subjected to whole transcriptome analysis by RNA-
chromosome 12 where var2csa resides. The increases in histone seq. Differential gene expression analysis was performed by
methylation extended into regions containing members of other comparing the transcriptomes derived from each of these
clonally variant gene families, including, rif, stevor, phist and acs. In conditions to that obtained from wild-type parasites grown
all cases, we detected a consistent and distinct increase in H3K9me3 in standard media. Using relatively stringent criteria [at least
at the variant antigen encoding genes while H3K4me3 was much 10,000 basemean (mean of normalized counts of all samples)
less responsive (Fig. 6). The parasite line used was expressing var2csa, and a log2FoldChange of at least 1.95], we identified relatively
and this locus displayed little change in H3K9me3 or H3K4me3, few genes that were responsive to our manipulations (changes in
consistent with the effect being more pronounced at silent regions gene expression for all conditions are summarized in Dataset S1).
of the genome that are enriched in H3K9me3. The differential effect This was not unexpected given that these conditions had little
on levels of the two histone modifications likely reflects different or no effect on parasite growth. We did observe an expected
Km values of the specific methyltransferases that deposit the marks, decrease in PfPMT expression in trophozoites grown in excess
although this will require experiments with the purified enzymes to choline, as was previously described (53). We also observed a
verify. These results are consistent with a model in which increases decrease in expression of a large cohort of genes expressed very
in SAM availability leads to corresponding increases in histone late in the replicative cycle in the PfSAMS knockdown line. This
methyltransferase activity and histone methylation. is consistent with a small delay in cell cycle progression that was
previously observed when parasites were grown in the presence
Overall Gene Expression Changes in Response to Alterations of low doses of a histone methyltransferase inhibitor (62). It is
in SAM Metabolism. Culture conditions that increased SAM logical that reduced SAM availability would similarly reduce
levels (excess choline/depleted serine, excess methionine, and methyltransferase activity and produce a similar phenotype.
overexpression of PfSAMS) resulted in a switch in var gene Interestingly, in neither case is the overall parasite growth rate
expression to var2csa while knockdown of PfSAMS did not reduced, indicating that the cell cycle delay is transient and
(Figs. 2–5). To determine whether these changes had additional that the parasites complete the overall replicative cycle on time.
effects on expression of genes other than var genes, parasites As a more sensitive way to identify any genes that are responsive
reared under conditions of increased choline or methionine to changes in SAM availability, we performed differential gene
supplementation or after PfSAMS overexpression or knockdown expression analysis comparing the two manipulations that led to
6 of 12 https://doi.org/10.1073/pnas.2302152120 pnas.org
A B C
PfSAMS HA AAAn
-GlcN +GlcN
1
relative mRNA
0.75
+GlcN
expression
-GlcN SAMS 0.5
H3 0.25
TetR-DOZI 0
HA -GlcN +GlcN
PfSAMS HA AAAn
AAA
n
stable mRNA mRNA degraded
Fig. 5. Effect of knockdown of PfSAMS expression on var gene expression. (A) Schematic representation of the GlmS system for mRNA knockdown. The DNA
sequence encoding a HA-tag and the glms ribozyme was integrated into the 3′ end of the PfSAMS locus, thereby incorporating a ribozyme into the 3′ UTR of the
mRNA. Addition of glucosamine (GlcN) results in degradation of the mRNA and reduced protein expression. (B) Western blot of proteins extracted from late-
stage parasites showing reduced expression of the HA-tagged version of PfSAMS after 36 h of growth in the presence of 2.5 mM glucosamine. (C) Quantification
of PfSAMS showing an ~50% reduction in PfSAMS expression upon the addition of glucosamine. (D and E) Pie charts representing var gene expression profiles
in the PfSAMS-glms parasite populations with each slice of the pie representing the expression level of a specific var gene. The profile for the initial transfected
population is shown on the left with additional profiles from 2 wk and 4 wk with and without glucosamine added to the culture media. The annotation numbers
of the dominantly expressed var genes are shown in E. (F) var expression profiles for wild-type parasites grown with or without glucosamine for 4 wk.
the largest changes in SAM levels, the PfSAMS overexpression line observed that var genes appear to be more responsive to changes
to the PfSAMS knockdown line (SI Appendix, Table. S4). Using the in SAM levels that the other gene families. Interestingly, from
same stringency criteria, in ring-stages, we observed an upregulation this large group of genes, only four defied the general trend in
of eight genes in the PfSAMS overexpression line, including the response to both genetic manipulations, becoming more highly
gene encoding ornithine aminotransferase (OAT), which was pre- expressed in the PfSAMS overexpressing line and repressed in
viously shown to be responsive to changes in SAM metabolism (63), the PfSAMS knockdown line. These included one member each
as well as the gene encoding Plasmodium transcription enhancer of the PHISTa, PHISTb, and PHISTc families (Pf3D7_0832200,
factor (PTEF), a protein previously shown to be involved in trans- Pf3D7_1201000 and Pf3D7_0424000, respectively). The
lation of the var2csa mRNA (64). We also observed upregulation PHIST (Plasmodium helical interspersed subtelomeric) proteins
of transcripts encoding two AP2 transcription factors (AP2-FG and are exported into the host RBC where some members have been
AP2-I), the significance of which is not clear. Comparison of the proposed to play a role in PfEMP1 trafficking and display on
trophozoite transcriptomes only detected changes in expression of the RBC surface (66–69). Pf3D7_0424000 was previously
the very late-stage specific genes mentioned above that are down- shown to be transcriptionally upregulated with var2csa in clinical
regulated in the PfSAMS knockdown line. These are primarily genes isolates (70, 71), providing additional evidence that these genes
encoding proteins involved in erozoitee formation and their reduced might be coregulated with var2csa.
expression is consistent with slowed entry into the last phase of the To ensure that any changes in gene expression that we observed
replicative cycle of the PfSAMS knockdown line. were in fact the result of changes in SAM metabolism rather than
We were particularly interested in any changes in expression unexpected mutations in the parasite’s genome, whole genome
of genes known to be subject to clonally variant expression and sequencing was performed on the PfSAMS overexpressing and knock-
therefore likely to be responsive to changes in histone methyla- down transgenic parasite lines as well as parasites grown for 4 wk
tion. Specifically, we compare the PfSAMS overexpression and under culture conditions of excess choline/depleted serine or excess
knockdown lines separately to wild-type parasites and examined methionine. We identified no mutations shared by parasites display-
changes in mRNA abundance of the genes and gene families ing a shift toward var2csa expression or in genes thought to be
known to be linked to the histone modification H3K9me3, involved in lipid or SAM/SAH metabolism, chromatin modification
including var, rifin, stevor, Pfmc-2TM, clag, phist, hyp, acs and or transcription. Therefore, we conclude that the gene expression
others (65). To increase the likelihood of detecting meaningful changes we observed are unlikely to be due to genetic mutations.
changes in genes that are often expressed in only a subset of
parasites within a population, we reduced the threshold level of Subcloning of Transfected Parasite Lines Gives Additional
expression to a basemean of 500 for this analysis. As shown in Insight into their var Switching Patterns. The propensity of
Fig. 7 A and B, the vast majority of these genes displayed a parasites to undergo var gene switching can also be readily
reduction in transcript levels when PfSAMS was overexpressed detected by generating subcloned parasite lines by limiting
and a corresponding increase in expression when PfSAMS was dilution. For example, subcloned populations are typically
knocked down. This likely reflects the changes in the abundance recovered that either express a single dominant var gene or
of the silencing mark H3K9me3 shown in Fig. 6. We also alternatively express a very heterogenous mix of var genes (28),
0
change in
H3K4me3
-1
-2
2
0 change in
-2
H3K9me3
10 KB 20 KB 30 KB 40 KB 50 KB 60 KB 70 KB
(stevor) (stevor)
Fig. 6. Conditions that increase
SAM availability lead to increases in
B central region of chromosome 4 histone methylation. Whole genome
1
cut-and-run analysis was used to
change in
0
-1
H3K4me3 assess differences in the levels of
-2 H3K9me3 and H3K4me3 in parasites
grown under media conditions that
6
4 increase SAM levels (20× excess
2
change in choline) compared to conditions that
reduced SAM levels (depleted choline
0
-2
H3K9me3
and methionine). The plots display
530 KB 540 KB 550 KB 560 KB 570 KB 580 KB 590 KB 600 KB 610 KB 620 KB 630 KB
the mean change in H3K4me3 (blue)
PF3D7_0411900 PF3D7_0412400 PF3D7_0412800 PF3D7_0413000 PF3D7_0413200 PF3D7_0413500 PF3D7_0414000 and H3K9me3 (red) coverage across
(var) (ruf 6) (ruf 6) (rif) (pgm-ase 2) (smcp 3)
(pol )
PF3D7_0412000 PF3D7_0412500 PF3D7_0412900 PF3D7_0413300 PF3D7_0413700 PF3D7_0414100
three regions of heterochromatin
(zinc finger) (ruf 6) (var) (rif) (lysine DC) (unknown) found within the P. falciparum
PF3D7_0412100 PF3D7_0412600 PF3D7_0413100 PF3D7_04134000 PF3D7_0413900
(S12) (pseudo) (var) (pseudo) (uc-th 13) genome. Positive values indicate an
PF3D7_0412200 PF3D7_0412700 PF3D7_0413600
(unknown) (var) (26S subunit 6B) increase in histone methylation under
conditions of increased SAM. (A) The
PF3D7_0412300 PF3D7_0413800
(ppc ligase) (50S L10)
subtelomeric region at the left end of
chromosome 2. (B) A central region of
C central region of chromosome 3 chromosome 4 containing multicopy
1
change in
gene family members. (C) A central
0
H3K4me3 region of chromosome 3 containing the
-1
-2
clag3.1/3.2 genes. (D) The subtelomeric
region at the left end of chromosome
4
12. Green and light blue arrows beneath
each plot indicate open reading frames,
2
0 change in
-2 H3K9me3 with gene annotation numbers and gene
names displayed next to each arrow.
120 KB 130 KB 140 KB
Abbreviations: pseudo- pseudogene;
pol α- DNA polymerase alpha catalytic
PF3D7_0302300 PF3D7_0302500
subunit A; zinc finger- LITAF-like zinc
PF3D7_0302100
(protein kinase) (pseudo) (clag 3.1)
PF3D7_0302200 PF3D7_0302400 PF3D7_0302600 finger protein, putative; S12- ribosomal
(clag 3.2) (ncRNA) (ABC B4) protein S12, mitochondrial; ppc
ligase- phosphopantothenoylcysteine
synthetase, putative; ruf 6- RNA of
D subtelomeric region of chromosome 12 unknown function 6; pgm-ase 2-
1
phosphoglucomutase-2; 26S subunit
0 change in 6B- 26S protease regulatory subunit 6B,
-1
H3K4me3 putative; lysine DC- lysine decarboxylase-
-2 like protein, putative; 50S L10- 50S
6 ribosomal protein L10, putative; uc-
4
th 13- ubiquitin carboxyl-terminal
2
0 change in hydrolase 13, putative; smcp 3- structural
-2
H3K9me3 maintenance of chromosomes protein
10 KB 20 KB 30 KB 40 KB 50 KB 60 KB 70 KB
3; protein kinase- serine/threonine
PF3D7_1200610
(var2csa
protein kinase; ncRNA- noncoding
RNA; ABC B4- ABC transporter B family
PF3D7_1200100 PF3D7_1200300 PF3D7_1200500 PF3D7_1200800
(var) (rif) (rif) uORF) (fikk)
member 4, putative; acs7- acyl-CoA
PF3D7_1200200 PF3D7_1200400 PF3D7_1200600 PF3D7_1200700
synthetase 7; fikk- serine/threonine
(rif) (var) (var2csa) (acs7) protein kinase, FIKK family.
with the latter parasites presumably reflecting a population that For the empty vector (EV) control line, we analyzed eight indi-
arose from a single parasite that was actively undergoing var vidual subclones and the var expression profiles are pictured as pie
gene switching at or near the moment that it was isolated. We charts in Fig. 8A. Out of the eight screened EV clones, one had a
applied this method to examine var gene switching patterns for unique var expression profile from the parent clone, manifesting
parasites in which we had earlier altered PfSAMS expression. as a heterogenous expression profile and indicative of active
Specifically, we subcloned populations of the empty vector switching. It is notable that each of the seven remaining clones
control, PfSAMS overexpression and PfSAMS knockdown lines, had maintained high expression of the dominant “on” var, and
maintaining a selection pressure of 1 μg/mL blastidicin S to none had switched to var2csa.
facilitate maintenance of the episomal expression construct. For the PfSAMS overexpression line, we anticipated that all
We then examined the var expression pattern in each of the would be found to continue expressing var2csa at very high levels.
subcloned lines. We isolated a total of 14 individual subclones and found that 12
8 of 12 https://doi.org/10.1073/pnas.2302152120 pnas.org
A B
PfSAMS overexpression PfSAMS knockdown
GBPH2
rif
40 40
FIKK
Differential Expression
30
significantly down
30
significantly up
var genes hyp17
−log10(padj)
−log10(padj)
rif
PHISTa
20 20
PHISTa surfin
CLAG8
GBA140
CLAG9
PHISTa
CLAG2
10 10
hyp9 Pf3D7_0424000 (PHISTc) CLAG3.1
GBPH2
Pf3D7_0832200 (PHISTa)
PHISTa
Pf3D7_1201000 (PHISTb)
Pf3D7_1206000 (var2csa)
stevor
0 0
−8 −4 0 −5 0 5 10
log2FoldChange log2FoldChange
Fig. 7. Volcano plots displaying differential expression of genes known to undergo clonally variant expression linked to changes in the deposition of the
chromatin mark H3K9me3. Upregulated genes are shown in red while down regulated genes are in blue. var genes are displayed as open circles while non-
var genes are labeled with the name of the encoded protein. Full annotation numbers for all genes are included in Dataset S1. (A) Volcano plot derived from
comparison of ring-stage parasites overexpressing PfSAMS to wild-type ring-stage parasites grown under standard culture conditions. (B) Volcano plot derived
from comparison of ring-stage parasites in which PfSAMS has been knocked down to wild-type ring-stage parasites grown under standard culture conditions.
Differentially expressed genes were defined as having a log2foldchange of at least 1.95 and an adjusted P value is less than 0.05. All var genes are shown by
open circles, while all other genes are shown by solid circles and labeled appropriately.
continued to express var2csa as the dominant var gene, as expected Interestingly, all clones maintained generally the same heteroge-
(Fig. 8B). Two clones, however, had switched to expressing a het- nous var expression pattern as the parent population. If these
erogenous mix of var genes at low levels, indicative of active parasites were undergoing normal or rapid switching, we would
switching. Given that overexpression of PfSAMS resulted in vir- expect the heterogenous profiles to differ between each clone.
tually complete switching to var2csa (Fig. 4E), we expected these However, the patterns were very similar, with the most highly
lines to express almost exclusively var2csa as long as PfSAMS expressed genes shared between all eight clones. This is surprising
overexpression was maintained, therefore we were surprised to and indicates either that switching has become skewed toward this
observe clones that displayed active switching. Consequently, we subset of highly expressed genes, or alternatively, that mutually
more closely examined PfSAMS expression in these two lines using exclusive expression has been altered and that individual parasites
Q-RT-PCR. Surprisingly, PfSAMS expression was greatly reduced are now expressing all these genes. The overall increase in total var
(Fig. 8C, clones H9 and C7), indicating that this population was gene expression in the knockdown lines is consistent with the
no longer overexpressing PfSAMS and providing an explanation latter hypothesis (SI Appendix, Fig. S3). Interestingly, some of the
for why the parasites had stopped exclusively expressing var2csa. genes that are expressed at a low level differ between the clones,
Expression of PfSAMS in these two clones was found to be even suggesting that the original population maintains the ability to
lower than an untransfected WT line. The reason for this loss of switch var genes to some extent under conditions of lower PfSAMS
PfSAMS overexpression is not known, but likely is the result of expression. Together, results from manipulation of PfSAMS
an alteration in the activity of the episomal promoter driving expression further support a model in which higher levels of SAM/
expression of the PfSAMS overexpression construct. To determine SAH push parasite populations to express var2csa.
if the var2csa expression phenotype could be rescued by increasing
PfSAMS expression from the transfected episome, the concentra- Discussion
tion of blasticidin S in the culture media was increased slightly to
2 μg/mL, a condition that is known to increase transgene expres- How parasites avoid rapid exhaustion of their var repertoire has been
sion (58, 72). As anticipated, these cultures reverted to var2csa a topic of speculation ever since the var gene family was first
expression after 3 wk (Fig. 8D), further supporting that increased described. Several hypotheses have been proposed to address this
PfSAMS expression is responsible for var2csa transcriptional question, including that parasites have evolved an inherent switching
activation. frequency that matches the time required for the host to mount an
The clones obtained from the PfSAMS knockdown line grown effective antibody response against the expressed form of PfEMP1
in the presence of glucosamine are pictured as pie charts in Fig. 6E. (reviewed in ref. 73). Alternatively, it has been suggested that parasites
parent parent
population population
subcloning by subcloning by
limiting dilution limiting dilution
subcloned populations
Pf3D7_0421100 var2csa
H9 B7 B11 C7
subcloned populations
C D E PfSAMS knockdown
3 parent
Relative PfSAMS expression
2 subcloning by
limiting dilution
H9
1
C7
3 weeks
0
lo B o Tc
r c
7 .)
.)
T
ne r2 )
ne ( )
subcloned populations
lo a a
B (va aT
lo H9 sa
et
C et
C SA S o W
C 1 (v 2cs
C ne MS n a
(h
h
c
ne 7 ff
Pf M
SA
C
Pf
lo
Fig. 8. Detection of var expression switching through limiting dilution cloning. (A) Pie charts representing var gene expression profiles for wild-type parasites
transfected with an empty expression vector are shown. Expression profiles for populations isolated from a single “parent” population by limiting dilution are
shown. All but one clone display dominant expression of the var gene Pf3D7_0421100 (orange). (B) Using the same methodology as in A, 14 subclones of a
PfSAMS overexpression line were obtained and var expression profiles determined. Two subclones (H9 and C7) had switched away from expressing var2csa (blue).
(C) Expression levels of PfSAMS were determine using Q-RT-PCR. Two clones (H9 and C7) that displayed a heterogenous var expression profile also displayed reduced
levels of PfSAMS expression. (D) var expression profiles from the PfSAMS overexpression lines H9 and C7. Initial expression profiles from populations displaying
reduced PfSAMS expression (Left) and after 3 wk of growth in media containing increased blasticidin concentration (Right). (E) Using the same methodology as
in A, eight subclones of the PfSAMS knockdown line were obtained and var expression profiles determined.
might be able to sense increasing titers of antibodies that recognize as local concentrations of polyamines, numerous cytokines and reac-
the infected RBC surface and respond by undergoing var gene tive oxygen species (reviewed by refs. 75 and 76). Such changes can
switching. Experimental data supporting the existence of a sensing result in downstream effects on lipid metabolism and SAM availabil-
mechanism have been presented for the monkey parasite Plasmodium ity, which are known to cause profound changes in epigenetic marks
knowlesi (74); however, no similar data have been reported for P. associated with heterochromatin in model systems (39, 40, 77).
falciparum or any other malaria parasite species. The data we present Thus, the immune response stimulated by antibodies against the
here however indicate that P. falciparum can in fact sense environ- expressed form of PfEMP1 could result in drastic changes in local
mental changes and undergo var gene switching in response. nutrient availability, thereby stimulating var gene switching through
Specifically, conditions that alter SAM metabolism can lead to robust the pathways described here. A similar model was previously pro-
changes in var gene expression. These results parallel recent obser- posed for regulating sexual commitment based on the observation
vations that changes in the availability of certain environmental that levels of LysoPC can change dramatically in response to immune
metabolites can drastically alter rates of sexual differentiation (42) activation or infection by various pathogens (42, 78–82). var gene
and rates of multiplication (43). We have performed similar exper- switching could be similarly modulated, albeit separately from sexual
iments in a gametocyte-producing line of P. falciparum and found commitment.
that conditions that induce var gene switching (excess choline or In the experiments described here, manipulations that increased
methionine) suppress pfap2-g expression and sexual commitment SAM levels resulted in a corresponding increase in histone methyl-
(Harris et al, bioRxiv 2022.01.18.476397), indicating that these two ation throughout the heterochromatic regions of the genome
phenomena are responding to similar environmental stimuli, but in (Fig. 6). Transcriptomic analysis detected very little changes in over-
opposite directions. all gene expression, suggesting that the increased histone methyla-
If parasites can respond to environmental changes by undergoing tion reinforced heterochromatic gene silencing that was already in
var gene switching, what conditions might elicit this response? If place in these regions. With respect to var genes, promoter compe-
parasites evolved to be able to undergo changes in var gene expression tition has frequently been cited as a model for mutually exclusive
in response to immune recognition, rising antibody titers targeting gene expression (83–87), with a single gene outcompeting other
PfEMP1 and immune cell activation by immune complexes could members of a family to become the transcriptionally active copy.
be key to triggering var gene switching. For example, it is known We hypothesize that increased silencing efficiency at UpsA/B/C var
that activation of various antigen presenting cells (dendritic cells, loci shifts competition to var2csa, leading to its activation under
macrophages, B cells) can cause dramatic changes in local availability conditions of increased SAM. Interestingly, decreased SAM levels
of various amino acids, including arginine and tryptophan, as well appear to loosen competition, leading to activation of several var
10 of 12 https://doi.org/10.1073/pnas.2302152120 pnas.org
genes and an overall increase in total var transcript levels protocols (91) and are explained in detail in the SI Appendix, Supplementary
(Figs. 5 and 8 and SI Appendix, Fig. S3). These data suggest a model Methods section. Custom media preparations are described in SI Appendix,
in which the unique var2csa promoter becomes selectively activated Tables S1 and S2.
under conditions of increased heterochromatic silencing, enabling
Genomic, Transcriptomic, and Metabolomic Analysis. Primers for Q-RT-PCR
it to serve as a switching node (88). While the precise nature of the
and plasmid construction are listed in SI Appendix, Table S3. Measurements of
competition between var promoters is not understood, histone
metabolites were performed by LC-MS as previously described (56) and described
methylation levels appear to strongly influence the interactions. in detail in the Supplemental Data section. The genome sequencing data can
Interestingly, we did not detect activation of specific members of be accessed at this link: http://www.ncbi.nlm.nih.gov/bioproject/931344 (92).
the rifin, stevor or Pfmc-2TM gene families in response to our The RNAseq data can be accessed at this link: http://www.ncbi.nlm.nih.gov/bio-
manipulations, suggesting that expression of these gene families is project/931991 (93). Metabolomic datasets and Cut & Run ChIP-seq data are
regulated differently. The co-induction of PF3D7_0424000, a gene associated with Harris et al. (bioRxiv 2022.01.18.476397).
previously detected as being upregulated along with var2csa in clin-
ical isolates (70, 71), and the gene encoding PTEF, a protein impli- Assays for Histone Modifications. To assess changes in H3K9me3 and
cated in overcoming the repressive effect of the var2csa uORF (64), H3K4me3 levels, Cleavage Under Targets & Release Using Nuclease (CUT&RUN)
reinforces a possible regulatory link between these genes and var2csa. was performed as described in ref. 60 with details provided in the Supplementary
While we observed changes in histone modifications at hetero- data section. See https://github.com/KafsackLab/MetChoH3K9me3 for the full
chromatic regions throughout the genome, it is also possible that analysis pipeline (94).
SAM availability might not influence the entire genome equally. For Antibody Generation and Immunoblotting. Rabbit polyclonal antibodies
example, while our genetic and metabolic manipulations led to pro- were obtained via a 28-d protocol (Pocono Rabbit Farm and Lab). Total protein
found changes in var gene expression, we observed little to no change lysates were prepared using saponin-lysed parasites, separated in 12% pol-
in parasite growth and the overall transcriptome changed very little, yacrylamide gels and transferred to 0.2-μm membranes. Primary antibodies
indicating that major effects on gene expression are likely limited to included anti-Hsp70 (1:2,000 dilution, StressMarq SPC-186), anti-HA (1:2,000
a relatively narrow portion of the genome. In a recent paper (89), dilution, Abcam, ab9110) and anti-SAMS-706 (1:500 dilution). Secondary
Heinberg and colleagues provided evidence for how, through the antibodies were used at a 1:2,000 dilution (anti-rabbit IgG, Millipore
formation of R-loops, var antisense lncRNAs serve to recruit specific Sigma 12-348). Detailed procedures are described in the Supplemental Data
protein complexes to the active var locus, thereby facilitating the section.
localized assembly of transcriptionally active chromatin within the
subnuclear var expression site. Remarkably, among the recruited Data, Materials, and Software Availability. Genome sequence, transcrip-
proteins are metabolic enzymes, including thioredoxin peroxidase tomic data and analysis pipeline data have been deposited in NCBI Bioproject and
(PfTPx-1), which was previously shown to interact directly with both Github (PRJNA931991 (93); PRJNA931344 (92); https://github.com/KafsackLab/
PfSAMS and PfSAHH (90), thus linking SAM metabolism to chro- MetChoH3K9me3 (94)). All genetically modified parasite lines are available from
matin assembly at the active var gene. These authors showed that the authors upon request.
manipulation of expression of either PfTPx-1 or PfSAMS alters var
gene expression switching, specifically activating var2csa transcrip- ACKNOWLEDGMENTS. This project was funded by NIH grants AI52390 and
tion, further implicating these proteins in a transcriptional response AI99327 to K.W.D. and AI141965 to B.F.C.K. V.M.S. was supported by the David
pathway. Together with the observations described here, this work Rockefeller Graduate Program in Bioscience at the Rockefeller University.
defines a molecular mechanism that enables parasites to sense alter- F.F. received support from the Swiss NSF (Early Postdoc.Mobility grant
ations to their environment and respond through epigenetic changes P2BEP3_191777). C.T.H. and J.E.V. received support from F31 Predoctoral
Fellowships 5F31AI136405 and F31AI164897, respectively, from the NIH.
in gene expression. In addition, by employing lncRNAs to anchor
The Department of Microbiology and Immunology at Weill Medical College of
protein complexes to specific loci, parasites are able to target epige-
Cornell University acknowledges the support of the William Randolph Hearst
netic changes to individual genes, thus differentiating the assembly Foundation.
of active versus silent chromatin and contributing to mutually exclu-
sive var gene expression.
Author affiliations: aDepartment of Microbiology and Immunology, Weill Cornell Medicine,
Materials and Methods Cornell University, Ithaca, NY 14853; bLaboratory of Chemical Biology and Microbial
Pathogenesis, Rockefeller University, New York, NY 10065; cDivision of Infectious
Diseases, Department of Medicine, Weill Cornell Medicine, Cornell University, Ithaca, NY
Parasite Cultivation and Genetic Manipulation. Methods for parasite culture, 14853; and dSection of Infectious Disease, Department of Microbial Pathogenesis, Yale
transfection, analysis of gene expression and growth assays follow established School of Medicine, Yale University New Haven, CT 06510
1. J. A. Boddey, A. F. Cowman, Plasmodium nesting: Remaking the erythrocyte from the inside out. 9. T. Chookajorn et al., Epigenetic memory at malaria virulence genes. Proc. Natl. Acad. Sci. U.S.A. 104,
Annu. Rev. Microbiol. 67, 243–269 (2013). 899–902 (2007).
2. X. Su et al., A large and diverse gene family (var) encodes 200–350 kD proteins implicated in the 10. J. J. Lopez-Rubio et al., 5’ flanking region of var genes nucleate histone modification patterns linked
antigenic variation and cytoadherence of Plasmodium falciparum-infected erythrocytes. Cell 82, to phenotypic inheritance of virulence traits in malaria parasites. Mol. Microbiol. 66, 1296–1305
89–100 (1995). (2007).
3. D. I. Baruch et al., Cloning the P. falciparum gene encoding PfEMP1, a malarial variant antigen 11. C. Flueck et al., Plasmodium falciparum heterochromatin protein 1 marks genomic loci linked to
and adherence receptor on the surface of parasitized human erythrocytes. Cell 82, 77–87 phenotypic variation of exported virulence factors. PLoS. Pathog. 5, e1000569 (2009).
(1995). 12. K. Perez-Toledo et al., Plasmodium falciparum heterochromatin protein 1 binds to tri-methylated
4. J. D. Smith et al., Switches in expression of Plasmodium falciparum var genes correlate with changes histone 3 lysine 9 and is linked to mutually exclusive expression of var genes. Nucleic Acids Res. 37,
in antigenic and cytoadherent phenotypes of infected erythrocytes. Cell 82, 101–110 (1995). 2596–2606 (2009).
5. M. J. Gardner et al., Genome sequence of the human malaria parasite Plasmodium falciparum. 13. A. Cortes, K. W. Deitsch, Malaria epigenetics. Cold Spring Harb. Perspect. Med. 7, a025528 (2017),
Nature 419, 498–511 (2002). 10.1101/cshperspect.a025528.
6. T. Otto et al., Evolutionary analysis of the most polymorphic gene family in falciparum malaria. 14. L. Jiang et al., PfSETvs methylation of histone H3K36 represses virulence genes in Plasmodium
Wellcome Open Res. 4, 193 (2019). falciparum. Nature 499, 223–227 (2013).
7. K. W. Deitsch, R. Dzikowski, Variant gene expression and antigenic variation by malaria parasites. 15. U. E. Ukaegbu et al., Recruitment of PfSET2 by RNA polymerase II to variant antigen encoding loci
Annu. Rev. Microbiol 71, 625–641 (2017). contributes to antigenic variation in P. falciparum. PLoS. Pathog. 10, e1003854 (2014).
8. L. H. Miller, D. I. Baruch, K. Marsh, O. K. Doumbo, The pathogenic basis of malaria. Nature 415, 16. L. Cui, Q. Fan, L. Cui, J. Miao, Histone lysine methyltransferases and demethylases in Plasmodium
673–679 (2002). falciparum. Int. J. Parasitol. 38, 1083–1097 (2008).
12 of 12 https://doi.org/10.1073/pnas.2302152120 pnas.org