**4.** *C. albicans* **PDR/Cdr1 Holds All Conserved Motifs Critical for ABCG Function**

The atomic structures from X-ray crystallography and cryo-EM of mammalian ABCGs suggest a rather unique fold resembling an importer rather than an exporter [38,256–259]. Indeed, mammalian ABCGs and fungal PDR exporters share conserved and superimposable topologies with all functional motifs in equivalent places (Figure 1A, Figure 2B, Supplementary Figure S1 and Table 4). Hence, we generated homology models of Cdr1 from *C. albicans*, using human ABCG2 (PDB ID 6VXF) as a template (Figure 3A). Remarkably, the Cdr1 structural model perfectly mirrored the human ABCG2 conformation, since each pivotal motif important for function is also present in Cdr1 (Figure 3).

**Figure 3.** Type II mammalian and fungal ABCG/PDR exporters share functional domains. (**A**) A homology model of the PDR exporter Cdr1 from *C. albicans* was generated using the SWISS-MODEL tool (color ribbon, NBD: green, TMD: yellow, elbow helix: pink, ICL: orange, valve: red, re-entry helix: blue and ECL: purple). The Cdr1 model was superimposed with the cryo-EM structure of human ABCG2 (PDB ID: 6VXF) (gray ribbon). (**B**) Zoom-in top view of the NBD dimer from panel A at the NBD–NBD interface shows essential conserved motifs (Walker A: red, Q-loop: violet, signature: pink, Pro-loop: yellow, Walker B: orange, D-loop: cyan and H-loop: blue). The transmission interfaces of human ABCG2 (**C**) and *C. albicans* Cdr1 (**D**) show a network cluster of triple helical bundles (THB) with conserved tyrosine residues as part of a salt bridge between elbow helix and ICL. Two negative residues are shown as the intracellular gates. Conserved residues are shown as sticks with color-coding as in the topology model. The conserved phenylalanine F clamp F439 in human ABCG2 (**E**) and the putative F clamp in *C. albicans* Cdr1 (**F**) are indicated as sticks. At the extracellular gates, the valve-like structures at the top of central cavity (red) are shown at the corresponding position of human ABCG2 (**G**) and *C. albicans* Cdr1 (**H**). The lid-structure formed by ECLs is shown as a violet ribbon with a surface. The homo-dimeric human ABCG2 transporter has a symmetric lid (**G**), while the full-size *C. albicans* Cdr1 transporter has a larger ECL forming the outer lid and part of the roof architecture (**H**).

The NBDs in Cdr1 are in closer contact and hold all conserved regions required for ABCG2 function (ATPase activity, A-loop, Walker A and B, Q-loop, mutational hot spot helix, signature loop, Pro-loop, D-loop and H-loop). The N-terminal and C-terminal NBDs form a head-to-tail dimer upon ATP-binding [336,337], with a RecA-like and an α-helical subdomain [327] (Figure 3B). Nonetheless, the fungal NBD1 has three minor differences, in that (i) a glutamine residue in the Q-loop is replaced with glutamate, (ii) the histidine residue in the H-loop is substituted with tyrosine and (iii) the Pro-loop is missing but contains a glycine instead [337] (Table 4). The non-identical deviant NBDs in fungal PDRs [41,216,338,339] may support an asymmetric catalytic cycle as proposed for ABCG5/G8 [38,339–341], whereas "symmetric" cycles require the presence of fully conserved "canonical" ATP-binding sites in both NBDs.

Stabilizing salt bridges maintain proper folding and dynamics. At least two salt bridges appear conserved in Cdr1 (Supplementary Figure S1), connecting the elbow helix with the coupling helix (R503 to E576, and R1185 to E1261) (Figure 3D), respectively. However, the salt bridge within ICL1 seems absent in fungal PDR, and a salt bridge at the upper membrane leaflet is found only in the first half of Cdr1 (R456 in ECL1 to E704 in the re-entry helix) (Table 4).

The triple helical bundle (THB) is part of the transmission interface and extremely conserved in Cdr1. Remarkably, the THB is present in diverse ABC transporters, including ABCB1 and LPS extractor, as well as in the antibiotic exporter MacB [332]. This may reflect a universal function in mediating NBD–TMD cross-talk, thus constituting a key element for controlling and driving the conformational switch to drive substrates through the translocation pathway. Moreover, the most highly conserved Y464 residue is essential for ATP consumption, and engages in a salt bridge to stabilize the entire transmission interface in the center of NBD–elbow–ICL1 cluster (Table 4 and Figure 3C,D). Thus, the THB constitutes a cluster of limited conformational flexibility, taking advantage of Y464 and its salt bridge with E458 in ICL1 and/or between elbow helix R383 and E458 [332] (Table 4 and Figure 3C). As for Cdr1, the proposed THB cluster is present in both N- and C-terminal domains. At the N-terminus, Y584 connects to R503 (elbow helix) and E576 (in the coupling helix), whereas at the C-terminal domain, Y1257 bridges to R1185 (second elbow helix) and E1261 in the second coupling helix (Table 4 and Figure 3D). This finding strongly supports the notion for THB as an essential structure acting at the transmission interface to control the entire transport cycle [332].

The mechanism of intracellular gating in PDR/ABCG2 must be crucial for substrate/inhibitor entry into the transport pathway of the exporter, but as yet is little understood. In human ABCG2, two negative residues are conserved in all ABCG/PDR transporters, represented by E451 (between TMH2 and ICL1) and D477 (beginning of TMH3). Interestingly enough, they are pivotal for drug transport, but do not affect ATP hydrolysis [40] (Table 4 and Figure 3C). Since this region around the transmission interface undergoes dynamic movements during the catalytic cycle, E451 and D477 may not be part of a drug-binding zone but rather provide an entry route and gating functions [332]. In fungal PDRs, both negative residues are conserved (Supplementary Figure S1) and Cdr1 indeed contains the corresponding positions with E570 and E597 in the N-terminal part, and D1255 and E1280 in the C-terminal domain (Table 4 and Figure 3D).

The so-called phenylalanine clamp formed by two F residues is located in the substratebinding zone of ABCG/PDR transporters. Interestingly, the THM2 in the ABCG/PDR subfamily contains at least 4–5 conserved F residues (Supplementary Figure S1). TMH2 occupies space in the middle of the transmembrane core, where a putative binding zone around the central cavity is present. In human ABCG2, F439 (Figure 3E), located in the middle of TMH2, is implicated in binding both substrates and inhibitors [333]. Remarkably, a new cryo-EM structure of ABCG2 illustrates that the aromatic side chains of both phenylalanine residues could contribute to a binding site [259], and thus play a role as a clamp for both substrate and inhibitor recognition [259,333]. Remarkably, the Cdr1 homology model suggests that the conserved residues in both TMDs (Supplementary Figure S1) are F559 in

TMH2 and F1239 in TMH8 (Figure 3F) and equivalent to F439 in ABCG2 (Figure 3E). Hence, conserved phenylalanine in or nearby substrate/inhibitor-binding zones of ABCG/PDR provide a clamping mechanism to trap substrates and/or inhibitors.

Extracellular gating at the membrane interface and subsequent drug release from the outward-facing state is regulated by two conserved motifs, a hydrophobic valve and a flexible lid architecture in the roof [42]. The hydrophobic valve in ABCG/PDR transporters is contributed by two half-molecules, thus generating a physical gate for outward-directed substrate translocation from the central into the upper cavity. The atomic structures of mammalian ABCGs [38,256,259] indeed reveal a unique valve-like motif in the core of the transporter, separating the central cavity from the upper cavity. This conserved "glycine-Φ-Φ" motif plays a critical role as a hydrophobic valve that controls water flow through the transport pathway [42] (Supplementary Figure S1). The glycine adds a flexible kink, whereas two hydrophobic aliphatic leucines build a hydrophobic barrier to prevent water flow or substrate leakage. Human ABCG2 has the "G553-L554-L555" motif, whereas Cdr1 has "G672-F673-V674" and "G1362-V1363-L1364" in the first and second half, respectively (Table 4 and Figure 3G,H). Interestingly, a similar extracellular gating mechanism may also operate in CmABCB1, which possesses a gate at the outer membrane border to regulate substrate translocation [35,295].

A flexible but compact lid is part of the roof architecture formed by a rather large extracellular domain in ABCG/PDR. The roof is another unique motif in the ABCG/PDR subfamily. In human ABCG2, this roof is maintained by an intramolecular C592–C608 disulfide bond that strengthens the compact lid architecture. This lid may establish the second gating mechanism to regulate drug release from the outer cavity [334,335] (Figure 3G). Notably, the covalent C603–C603 inter-molecular link is key for homo-dimer formation in human ABCG2, but not essential for function [334,342]. The last extracellular loop of fungal PDR is slightly larger than the equivalent loop in mammalian ABCG2 (Supplementary Figure S1), where 2–3 conserved cysteine residues are present (Table 4). The overall similarity in the roof architecture also supports the notion of a conserved extracellular gating mechanism in ABCG/PDR transporters (Figure 3G,H).
