The Apicomplexa-specific glucosamine-6-phosphate N-acetyltransferase gene family encodes a key enzyme for glycoconjugate synthesis with potential as therapeutic target

Apicomplexa form a phylum of obligate parasitic protozoa of great clinical and veterinary importance. These parasites synthesize glycoconjugates for their survival and infectivity, but the enzymatic steps required to generate the glycosylation precursors are not completely characterized. In particular, glucosamine-phosphate N-acetyltransferase (GNA1) activity, needed to produce the essential UDP-N-acetylglucosamine (UDP-GlcNAc) donor, has not been identified in any Apicomplexa. We scanned the genomes of Plasmodium falciparum and representatives from six additional main lineages of the phylum for proteins containing the Gcn5-related N-acetyltransferase (GNAT) domain. One family of GNAT-domain containing proteins, composed by a P. falciparum sequence and its six apicomplexan orthologs, rescued the growth of a yeast temperature-sensitive GNA1 mutant. Heterologous expression and in vitro assays confirmed the GNA1 enzymatic activity in all lineages. Sequence, phylogenetic and synteny analyses suggest an independent origin of the Apicomplexa-specific GNA1 family, parallel to the evolution of a different GNA1 family in other eukaryotes. The inability to disrupt an otherwise modifiable gene target suggests that the enzyme is essential for P. falciparum growth. The relevance of UDP-GlcNAc for parasite viability, together with the independent evolution and unique sequence features of Apicomplexa GNA1, highlights the potential of this enzyme as a selective therapeutic target against apicomplexans.


Classification of the superfamily of GNAT-domain containing proteins in Apicomplexa.
To identify apicomplexan proteins presenting a GNAT domain, we searched the genomes of P. falciparum and 6 additional species, representing the main lineages of the phylum, for sequences containing the INTERPRO protein functional domain IPR000182 (GNAT domain). 71 protein sequences were detected, belonging to 18 orthogroups plus one singleton in the EupathDB database (http://eupathdb.org/eupathdb/, Supplementary Tables S1 and S2). In order to reconstruct the evolutionary relationships between these sequences, we performed phylogenetic analyses using three alternative methods (see Methods). 9 out of the 18 orthogroups, encompassing 48 out of the 71 sequences, were retrieved by all three methods as well-supported clades in the tree (Fig. 2). Despite the overall high degree of divergence in terms of sequence composition and length, the protein domain architecture within each of these 9 clades appeared relatively well-conserved. Many of the non-GNAT domains in these clades matched INTERPRO protein functional domains generically associated to the GNAT domain, such as the Acyl-CoA N-acyltransferase domain, or associated to specific families of GNAT proteins, such as N-myristoyltransferases (NMT) or histone acetyltransferases (HAT) (Supplementary Table S3). The remaining 9 orthogroups, clustering 23 out of 71 sequences, were not properly resolved in our phylogenetic analyses. A possible reason for this may be related to the unusually long branches recovered for some T. gondii sequences, which may lead to unexpected clustering in the tree 37 . T. gondii showed 15 GNAT-domain containing proteins compared  Table S1).

Identification of the GNA1 family in Apicomplexa.
In yeast and other eukaryotic organisms, the glucosamine-6-phosphate acetyltransferase activity is known to be essential and encoded by GNA1, a specific family of GNAT proteins 26,27,33 . The evolutionarily conserved GNA1 gene family typically encodes proteins of 150-200 aa length and only displays the GNAT and/or Acyl-CoA N-acyltransferase domains 32 . Out of the ten P. falciparum GNAT sequences, five were discarded as candidate GNA1s as they also contained protein domains corresponding to specific families of GNAT proteins other than GNA1. The remaining five (Fig. 2) were selected as candidate sequences and assayed for their putative GNA1 activity.
The five selected P. falciparum candidate sequences and Trypanosoma brucei GNA1 (TbGNA1) 26 were cloned into the pRS421 pt plasmid and used to transform the S. cerevisiae GNA1 thermosensitive (ScGNA1-ts) mutant 38 . Only PF3D7_0629000 and TbGNA1 could support yeast growth in the absence of endogenous GNA1 at 37 °C (Fig. 3A), and we consequently renamed PF3D7_0629000 as PfGNA1. Our phylogenetic analyses (Fig. 2) clustered PfGNA1 together with six additional sequences within a robustly supported clade corresponding to Figure 2. Phylogeny, protein domain architecture and taxonomic distribution of 71 GNAT proteins from seven apicomplexan species. The maximum-likelihood unrooted phylogenetic tree is drawn to scale, with branch lengths proportional to evolutionary distances between nodes. Branches leading to extant nodes in the tree are colored according to the species. Indicated in bold are sequences examined for GNA1 activity. Values next to the orthogroup-defining nodes indicate statistical support from maximum-likelihood, Bayesian and neighborjoining phylogenetic analysis, respectively. For EupathDB orthogroups retrieved as well-supported clades in the tree, colored strips indicate their taxonomic distribution (top right legend: Api, Apicomplexa; Euk, Eukaryota; Eug, Euglenozoa; Ubi, ubiquitous; Bab, Babesia; The, Theileria; Fun, Fungi; Tox, Toxoplasma; Eim, Eimeria). The apicomplexan-specific GNA1 family is enclosed within a continuous line box, while the HAT and the NMT families are enclosed within a dashed line box. A schematic protein architecture delineating the occurrence of conserved motifs detected using MEME is shown next to each protein (see Table S2). The distance between vertical lines on the right is 200 amino acids.
Independent evolutionary origin of the Apicomplexa-specific GNA1 family. According to the OrthoMCLDB database, of all orthogroups grouping Apicomplexa GNAT proteins, only OG5_144531 and OG5_147324 (Apicomplexa GNA1 family) showed a taxonomic distribution restricted to apicomplexan organisms (Supplementary Table S2), suggesting their independent evolutionary origin within the phylum. In order to examine whether this represents an unusual feature of the Apicomplexa GNA1 family, we used the information provided by OrthoMCLDB to look at the taxonomic distribution of 30 P. falciparum 29,34 enzymes involved in sugar nucleotide biosynthesis and related pathways across 150 species representing 12 eukaryote and prokaryote lineages (Fig. 4A, Supplementary Table S4). For each associated orthogroup, we plotted as a heat map the fraction of species within each taxonomic lineage represented by at least one orthologous sequence. The orthogroup containing Apicomplexa GNA1s clusters together with the presumed subunit H of phosphatidylinositol N-acetylglucosaminyltransferase (PIG-A subunit H, PF3D7_1141400), involved in GPI-anchor biosynthesis 39 . These two were the only orthogroups with no potential orthologs outside of Apicomplexa, including Tetrahymena termophila, belonging to Ciliophora, the sister phylum to Apicomplexa.
Moreover, using simple reciprocal best-hit BLAST searches between P. falciparum and human (E < 10 −5 ), no potential human orthologs could be unambiguously defined for only two of the 30 P. falciparum enzymes, including PfGNA1 and PF3D7_0517500 (Supplementary Table S4). All together these analyses supported a rather exceptional evolutionary history of the Apicomplexa GNA1 family, compared to that of other enzyme families involved in glycoconjugate synthesis.
To further substantiate the orthologous relationships of genes belonging to the Apicomplexa GNA1 family, we examined their syntenic arrangements. A pairwise synteny analysis was performed by comparing the genomic region containing PfGNA1 in P. falciparum and 10 additional Plasmodium species, revealing series of collinear genes between the two regions ( Supplementary Fig. S1). Similarly, synteny was examined both within and between species belonging to Aconoidasida (P. falciparum and B. bovis) and Conoidasida (C. parvum, T. gondii and T. annulata), the two main classes in the phylum ( Supplementary Fig. S2). Despite the deep evolutionary divergence between the two classes within the phylum (~817 MYA) 40 , synteny could also be observed in some pairwise comparisons along hundreds of Mb of genomic regions, further supporting the single and independent origin of the GNA1 family in Apicomplexa.
We also used the alignment of the GNAT protein domain conserved regions (resulting after removing poorly aligned or highly diverged regions) from the seven Apicomplexa GNA1 family sequences in Fig. 2 as a seed to build a Hidden Markov Model (HMM) profile. The resulting HMM profile was in turn used as a query in iterative searches against the UNIPROT database, with the HMM profile being iteratively rebuilt on the basis of the retrieved hits, ultimately resulting in the identification of a set of 50 significant protein hits exclusive to 50 strains from 30 apicomplexan species. We selected a representative strain from each species, including the seven for which GNA1 activity was confirmed experimentally, to perform phylogenetic and sequence analyses (Supplementary Tables S5 and S6). Extensive diversification could be observed among lineages at the level of sequence length and architecture of protein motifs, reflecting the old age of the Apicomplexa clade 40 (Fig. 4B). However, the phylogenetic relationships among sequences in the tree generally correspond well to the accepted taxonomic relationships among the species represented 41 . This observed congruence between the gene and the species trees is also compatible with a single evolutionary origin of the gene family.
Comparative sequence analysis between apicomplexan and non-apicomplexan GNA1 families further supports their independent evolutionary origin. The large superfamily of GNAT-domain containing proteins groups different families of enzymes that use acetyl coenzyme A (AcCoA) to transfer an acetyl group to a substrate 30,31 . Subsequent evolutionary diversification of the superfamily would have been shaped by extensive sequence divergence and lineage-specific domain gain and loss. Although highly divergent at the sequence level, GNAT domains are well-conserved in structure and catalytic mechanism 26,32,33 , suggesting a common evolutionary origin. In order to search for signatures of remote amino acid sequence homology between GNAT domains of the Apicomplexa GNA1 and non-apicomplexan eukaryote GNA1 enzyme families, we compared the multiple sequence alignment of the conserved region of the GNAT domain in a dataset of 30 apicomplexan GNA1 sequences to the conserved regions of the GNAT domain from 30 GNA1 from non-apicomplexan eukaryote organisms, ranging from yeast to human (Supplementary Table S7). The conserved regions of the GNAT domain were extracted by removing poorly aligned or highly diverged regions from their respective alignments. The average percentage of pairwise sequence identity within the ca. 90 amino acid length alignment of apicomplexan GNA1s was 50.46% (similarity 67.71%), close to the ones observed within non-apicomplexan GNA1s (49.74%; similarity: 66.97%). In contrast, when apicomplexan and non-apicomplexan GNA1 sequences were aligned together ( Supplementary Fig. S3), the average percentage of pairwise sequence identity between the two sets at the level of the conserved region of the GNAT domain drops to 18.06% (similarity 26.49%). The alignment  Table S4. (B) Phylogeny and protein domain architecture of putative GNA1s from 30 apicomplexan species. The Bayesian circular phylogenetic tree is drawn to scale, with branch lengths proportional to evolutionary distances between nodes. The tree has been rooted using Cryptosporidiidae representatives, considered to be occupying a basal position among Apicomplexa 41 . Values next to relevant nodes/clades indicate statistical support from maximum-likelihood, Bayesian and neighbor-joining phylogenetic analysis, respectively. Leaves are colored according to their taxonomic family membership and two additional classification groups, including class (see Table S5). Sequences examined for GNA1 activity are indicated in bold. The architecture in terms of conserved protein motifs detected using MEME is shown next to each protein (see Table S6). Outer circular lines form a protein length scale, the distance between lines is 100 amino acids.
algorithm introduced a gap of 14 aa length in apicomplexan GNA1s, mostly corresponding to the B3 strand of the canonical motif D of the GNAT domain 31 .
18 aa positions in the joint alignment were found to be conserved between apicomplexan and non-apicomplexan GNA1s across more than 60% of the sequences (including seven and three positions conserved in more than 75% and 90% of the sequences, respectively; Fig. 5 and Supplementary Fig. S3). Most of the conserved positions were found within the AcCoA binding motif A, particularly around the consensus motif defined for GCN5-related N-acetyltransferases 31 . Notably, only two of the conserved positions correspond to key functional residues identified in the crystal structure of human GNA1 (Fig. 5) 33 .
Lastly, we examined the predicted secondary structure of PfGNA1 and other apicomplexan GNA1 proteins. The region corresponding to the GNAT domain of PfGNA1 showed the characteristic α/β fold 31 , a conformation that was conserved across the remaining six apicomplexan GNA1s examined here ( Supplementary Fig. S4). When compared to the secondary structure of the GNAT domain of human GNA1, obtained from its protein crystal structure, some degree of conservation at the structural level could be observed, in particular in motifs A and B at the C-terminal end of the domain. These motifs are involved in acetyl-CoA binding and the active site of the enzyme, respectively 30,31 . In vitro activity assays of Apicomplexa GNA1 purified proteins confirm GNA1 enzymatic activity.
To further confirm the GNA1 activity of orthologs from the Apicomplexa GNA1 family, we heterologously expressed and purified the corresponding proteins. The enzymes used GlcN-6P as substrate and AcCoA as donor to produce GlcNAc-6P, as observed by HPLC-MS/MS (Fig. 6A). Although PfGNA1 and TaGNA1 were deficiently expressed, they also showed residual activity as detected by HPLC-MS/MS (Supplementary Fig. S5). In addition, the remaining recombinant proteins demonstrated clear GNA1 activity in colorimetric assays ( Supplementary  Fig. S6).
Since CpGNA1 (cgd4_4000) has been previously described as a spermidine/spermine N 1 -acetyltransferase (SSAT), involved in the back-conversion of polyamine spermine to spermidine and putrescine in C. parvum 42 , we decided to further characterize its enzymatic activity. Kinetic analyses showed that CpGNA1 used GlcN-6P as substrate to generate GlcNAc-6P displaying classical Michaelis-Menten kinetics, with a K m for AcCoA of 241.7 ± 35.3 µM and 719.1 ± 119.0 µM for GlcN-6P (Fig. S7A). In addition, in our hands CpGNA1 was not able to acetylate spermine or spermidine and SSAT activity could not be detected either in colorimetric ( Supplementary  Fig. S7B) or HPLC-MS/MS based assays (Fig. 6B). Furthermore, on the basis of the relative levels of GlcNAc-6P generated, CpGNA1 activity was not inhibited by the presence of increasing concentrations of spermine or spermidine ( Supplementary Fig. S7C). Hence, our data clearly indicates that the primary biochemical function of CpGNA1 is the acetylation of GlcN-6P. Apparently, T. gondii also shows an active polyamine retroconversion metabolism, driven by a SSAT activity, although to our knowledge this enzyme has never been heterologously expressed and assayed in vitro 43 . Nevertheless, as in the case of CpGNA1, TgGNA1 distinctly acetylated GlcN-6P, and this activity was neither inhibited by spermine nor spermidine, suggesting that the function of TgGNA1 is indeed to acetylate GlcN-6P ( Supplementary Fig. S7C).

CRISPR-Cas9-based gene disruption strongly suggests
PfGNA1 is essential for parasite growth. Finally, to gain further insight into the biological function of PfGNA1, and to evaluate its potential as a drug target, we assessed whether it was required for parasite growth using CRISPR-Cas9-based techniques. In a first set of experiments, we attempted to disrupt PfGNA1 by CRISPR-Cas9-assisted gene truncation. Whereas controls using analogous plasmids targeting other regions generated viable parasites, no parasites harbouring a truncated version of PfGNA1 could be obtained (Fig. 7A). This suggests that the PfGNA1 gene plays an essential role for the survival of the parasite. In a second set of experiments, we were also unable to introduce nonsense mutations in the N-terminus of the GNAT domain conserved region. In contrast, in three independent biological replicates carried out in parallel, we successfully managed to obtain viable parasites harbouring synonymous nucleotide substitutions in the same region of the GNAT conserved domain (Fig. 7C). Therefore, nucleotides in the catalytic GNAT domain of PfGNA1 are genetically modifiable as long as the open reading frame is not altered. These data strongly suggest that an unaltered version of the PfGNA1 protein is required for parasite viability, at least during P. falciparum asexual blood stages.

Discussion
Although GNA1 enzymes have been identified and characterized throughout the eukaryote kingdom, sequence similarity-based approaches failed to identify GNA1 genes in Apicomplexa 34 . This difficulty was likely due to the extensive divergence at the sequence level featuring the different GNAT domain-containing enzyme families with specific acetyl acceptor substrates and taxonomic coverages. Here, we performed a systematic classification of the large superfamily of GNAT domain-containing proteins in seven Apicomplexa species with fully sequenced genomes 30,31 . Our phylogenetic analysis revealed 9 well-supported gene families displaying conserved architecture of protein motifs. Among them, we identified protein families containing NMT or HAT domains, including representatives of the latter family from P. falciparum (PF3D7_0823300) and T. gondii (TGGT1_254555 and TGGT1_243440) that had been previously characterized 44,45 . We tested five P. falciparum sequences, showing no evident correspondence with other GNAT domain-containing gene families, by complementation assays using a yeast thermosensitive strain defective in GNA1. As a result, we identified PfGNA1 as able to support yeast growth at restrictive temperature. PfGNA1 belongs to a well-defined family containing putative orthologs in all six other Apicomplexa species, all of which rescued the mutant under non-permissive conditions. The GNA1 family in Apicomplexa appears to have had a single origin early during diversification of the phylum, an observation supported by synteny-based analysis. Furthermore, using a HMM profile based on Apicomplexa GNA1 sequences as a query to scan the UNIPROT database, only sequences belonging to the phylum resulted in significant hits. Consistently with a separate evolutionary origin of the family, the phylogenetic tree of putative GNA1 from 30 species of Apicomplexa essentially recapitulated the accepted evolutionary relationships among the species, based on both ultrastructural and developmental characterizations as well as on molecular phylogenies 41 . Altogether, our results support an independent origin of the GNA1 family in Apicomplexa, which can be traced back to between 817 MYA (estimated time of divergence between the two main classes of the phylum) and 1344 MYA (estimated time of divergence between Apicomplexa and its sister group of ciliates represented by T. termophila) 40 .
The comparison of the GNAT domains of apicomplexan and non-apicomplexan GNA1s showed that domain conservation was essentially restricted to a few specific amino acid positions around the GNAT six-amino acid consensus motif 31 . Furthermore, the conserved residues were, with very few exceptions, distinct from the ones reported as critical for the catalytic activity of human GNA1 32 . This observation suggests that, rather than evolving from an ancestral eukaryote GNA1 family, apicomplexan GNA1s might have evolved in parallel from a separate GNAT domain-containing gene lineage. Furthermore, considering the evolutionary time since divergence of the phylum from other eukaryotic lineages 40 , Apicomplexa-specific GNA1s may have evolved functionally divergent catalytic residues exploitable for the design of selective small-molecule inhibitors.
The CpGNA1 sequence (cgd4_4000) had been previously characterized to encode a SSAT 42 , involved in the reverse polyamine biosynthetic pathway in C. parvum. Here, whereas a robust GNA1 activity could be detected with as low as 0.2 µg/mL of purified C. parvum protein, we did not detect a SSAT activity with up to 1000 µg/mL, both through colorimetric and HPLC-MS/MS based in vitro assays. In addition, neither CpGNA1 nor TgGNA1 43 activities were inhibited by the presence of spermine or spermidine. Considering that the rest of the apicomplexan proteins tested showed evident GNA1 activities, our results suggest that the primary function of this family of enzymes is the acetylation of GlcN6P. Thus, the SSAT activity previously described for CpGNA1 would be residual, secondary, and/or just restricted to specific taxonomic lineages, life cycle stages or physiological conditions.
Our multiple efforts to ablate PfGNA1 using CRISPR-Cas9-based techniques were unsuccessful, despite evidence that the locus was genetically modifiable. This strongly suggests that the gene is essential for the growth of the malaria parasite, in agreement with UDP-GlcNAc being a bottleneck metabolite 46 , likely due to its involvement in the biosynthesis of important glycoconjugates, such as GPI anchors. Nevertheless, the relevance of GNA1 in other apicomplexan organisms is at present unknown. Notably, in a recent work, a loss-of-function genome-wide screen was performed in T. gondii tachyzoites to assess the contribution of targeted genes to cell fitness 47 . With the exception of TgGNA1 (TGGT1_243600), all the genes putatively encoding for enzymatic activities involved in UDP-GlcNAc biosynthesis (i.e. glucosamine:fructose-6-P-amidotransferase, TGGT1_231350; and N-acetylglucosamine-phosphomutase, TGGT1_264650) or UDP-GlcNAc utilization (i.e. Phosphatidylinositol N-acetylglucosaminyltransferase, TGGT1_241860; and N-acetylglucosaminyl phosphate transferase, TGGT1_244520) contributed greatly to T. gondii survival. This underlines the importance of the amino sugar biosynthetic route, whose presence is confirmed by labelling the glycoproteins of the parasite with [ 3 H]GlcN 48,49 . The striking lack of contribution of TgGNA1 to T. gondii fitness 47 might be suggesting the existence of another gene in the parasite's genome encoding for a redundant GNA1 activity, although it might also be highlighting certain limitations of the aforementioned study, such as the effect on proteins with slow turnover or the time selected for the screening readout 47 . Indeed, considering the comprehensive analysis of GNAT-containing sequences in different Apicomplexa presented here, and the functional complementation assays performed with additional P. falciparum sequences, no other suitable candidate sequence encoding for a redundant GNA1 activity could be identified. The generation of conditional GNA1 mutants, still a challenging task in apicomplexan organisms, will contribute to comprehensively define whether GNA1 biochemical activity is essential in different species and to confirm its suitability as a potential therapeutic target for drug development.

Methods
Sequence and phylogenetic analysis. Phylogenetic analyses were performed on the basis of multiple alignments of amino acid sequences obtained using MUSCLE 50 . Bayesian and maximum-likelihood analyses were carried out using the Blosum62 + G (eight categories, shape parameter: 1.87) + F protein evolution model 51 , selected by ProtTest v3.2 as the best-fitting to the data 52 . Bayesian analysis was implemented in MrBayes 3.2.5 53 . Searches were run with four Markov chains for one million generations sampling every 100th tree. After the stationary phase was reached, determined by the average standard deviation of split sequences approaching 0 (<0.05), the first 2,500 trees were discarded as burn-in. A consensus tree was then constructed to evaluate posterior probabilities on clades. Maximum-likelihood trees were constructed using PhyML v3.1, with tree topology searching optimized using the subtree pruning and regrafting option 54 . The statistical support of the retrieved topology was assessed using the Shimodaira-Hasegawa-like approximate likelihood ratio test 55 . Neighbor-joining phylogenetic analyses were conducted in Seaview v4.5.4 56 , with statistical support on clades assessed using a bootstrap analysis with 1,000 replicates. Trees were represented and edited using iTOL v3.3.2 57 . MEME v4.11.2 was used to identify conserved motifs shared among proteins 58 . Profile hidden Markov models (HMMs) were generated and calibrated using HMMER v3.0 59 , on the basis of MUSCLE protein alignments 50 further edited with Gblocks 60 . Logo representations of multiple protein sequence alignments were obtained using Skylign 61 . Protein secondary structures for apicomplexan GNA1 sequences were predicted using SABLE 62 .
Cloning of apicomplexan GNA1 candidate sequences. pRS421 vector (ATCC 87475), containing ampicillin and MET15 63 , was modified by cloning S. cerevisiae GPD promoter and CYC1 terminator sequences using SacII/BamHI and XhoI/KpnI restriction sites. The new vector generated, pRS421 pt was used to clone different GNA1 candidate sequences. Genomic DNAs of C. parvum, E. tenella, T. gondii, and T. brucei were used as templates for PCR amplification with Platinum DNA Polymerase High Fidelity (Thermo Scientific), using specific primers including BamHI and XhoI restriction sites (Supplementary Table S8). The remaining sequences cloned in pRS421 pt were codon-optimized synthetic genes (GenScript or Integrated DNA Technologies) based on the predicted amino acid sequences.
Yeast complementation assays. A S. cerevisiae GNA1 thermosensitive mutant, ScGNA1-ts 38 , was used for GNA1 complementation assays. The mutant contains a temperature-sensitive copy of GNA1 (YFL017C), which encodes a GNA1 enzymatic activity essential for yeast survival. The thermosensitive allele is marked by URA3 and the mutant was not able to grow above 34 °C 38 . ScGNA1-ts contains also a mutated MET15 gene that allows auxotrophic complementation with pRS421 pt vector in minimal media without uracil, methionine and cysteine (-ura, -met, -cys). After transformation with pRS421 pt based constructs, yeast cells were grown in serial dilution at permissive (23 °C) and restrictive (37 °C) temperatures.
To determine SSAT activity the same protocol was used but using as substrate 500 µM of spermine and 3 different CpGNA1 concentrations (250 µg/ml, 500 µg/ml and 1000 µg/ml). For SSAT colorimetric assays 500 µM of spermine or spermidine were used as substrate and 1, 20 or 500 µg/ml of CpGNA1. Reactions were allowed to proceed for 30 min at room temperature before being stopped by boiling for 10 min.

CRISPR-Cas9
PfGNA1-disruption and PfGNA1-editing constructs. Parasites were cultured and transfected either by electroporating ring-stage parasites or by nucleofection of schizont stages, as previously described 64 . All the methods were carried out in accordance with relevant guidelines and regulations and human erythrocytes and serum were purchased from the Banc de Sang i Teixits (Catalonia, Spain), after ethical approval from the Comitè Ètic Investigació Clínica Hospital Clínic de Barcelona. A single guide RNA (sgRNA) targeting the PfGNA1 consensus motif [(R/Q)-X-X-Q-X-G] was chosen using the Eukaryotic Pathogen CRISPR gRNA Design Tool 65 . For PfGNA1-disruption homology regions (HR) 1 and 2 were amplified from 3D7 P. falciparum genomic DNA using primers P1/P2 and P3/P4 and cloned in plasmid pL7 using SpeI/AflII and EcoRI/NcoI restriction sites, respectively 64 . sgRNA was integrated replacing pL7 BtgZI-adaptor (Fig. 7A). For PfGNA1 edition, sgRNA and Cas9-expressing construct (pDC2-Cas9-hDHFRyFCU) 66 and linearized pUC19 plasmids were used as backbone. The sgRNA sequence was cloned in the pDC2-Cas9-sgRNA using primers P5/P6. PfGNA1 coding sequences with a shield mutation in the protospacer-adjacent motif (PAM) 64 and silent or nonsense mutations were cloned in pUC19 to generate pUC19-silent PfGNA1 or pUC19-nonsense PfGNA1 (Fig. 7B). All primers used are described in Supplementary Table S8.