**5. Model for a Conserved Catalytic Transport Cycle of ABCG/PDR Transporters**

Mammalian ABCG and fungal PDR share all hallmark domains as well as numerous conserved residues essential for function (Figure 3). Based on these striking similarities, we wish to propose a unified mechanism for the transport cycle of type II ABCG/PDR multidrug transporters (Figure 4). Several studies suggest that more than half of all known ABC transporters including human ABCG5/G8 utilize non-equivalent or deviant NBDs [43,216,259,306,329,343,344]. Interestingly, ABCG2 appears as a perfect homo-dimer molecule, although it may have some asymmetries in the NBD dimers [325,341,345]. Therefore, the "primordial" alternating access model [30,36,324,346,347] forms a rational basis for our model (Figure 4). The ABCG/PDR transporters in the apo drug-free state are in an inward-facing configuration, with the bottom of the NBD dimer connected. We propose that ABCG/PDR subfamilies have asymmetric catalytic cycles, as proposed for human ABCG2 [341]. In this state, only one nucleotide-binding site is occupied by ATP, which could support an "intermediate" NBD dimerization state, with one free site remaining accessible for ATP. The intracellular gate formed by two negative residues at the membrane border of the ICL1 in the transmission interface is open, thus offering a path for substrate/inhibitor entry [40]. The central cavity provides free binding zones to accommodate compounds of variable chemical spaces [157,348]. The aromatic rings in the conserved F clamp establish accessible binding/trapping sites [333]. The hydrophobic valve at the top of the central cavity is almost completely closed and blocks water leakage through the translocation pathway [42]. The lid-forming roof architecture also remains closed in the inward-facing state, with a compact loop structure that limits the space in the upper cavity. Since ABCG/PDR exporters have an uncoupled ATP hydrolysis cycle, the catalytic cycle would still be active even without substrate(s) [33,40,42]. Drug substrates (2a) or

inhibitors (2b) can access the central cavity and the translocation pathway through the intracellular gate(s) [40] before getting trapped in the binding zones [157] by the F clamp located in TMH2 in each TMD [333]. This is a critical step which also prevents substrate escape from the central transport pathway. By contrast, the binding of an inhibitor (2b) at the region below the valve, would lock the conformation in the inward–open state and inhibit ATP hydrolysis activity as indicated by cryo-EM particle structures [257]. Whether a compound is an inhibitor or a transport substrate for ABCG/PDR transporters is solely determined by the affinity that sets the on versus off rates, and by the kinetics underlying the interactions in the binding zones [33]. Indeed, biochemistry data also suggest that certain inhibitors inhibit ATPase activity [349–351]. Binding of the second ATP molecule functions as a molecular glue that triggers complete NBD dimerization, thus inducing the conformational switch of the TMDs into a substrate-occluded state (3). The mechanical movement at the transmission interface requires the THB as a rigid structure of limited dynamics [332]. Subsequently, the central cavity is compressed, hence creating peristaltic pressure that drives substrates along the central translocation channel through the concomitantly opening valve. Further, the full NBD dimerization pushes the transporter into a compressed state, imposing a squeezing motion on the central cavity that generates pressure critical for opening the valve. The hydrophobic valve also serves as the first barrier for an extracellular gating to ensure unidirectional transport. A retrograde backflow of substrates is therefore prevented by the outward hydrostatic pressure and by the tight valve that would close when resetting the transporter [42]. Accordingly, the space of the upper cavity is then enlarged to accommodate substrates [42]. The limited dynamics and stability of the extracellular roof is supported by a conserved salt bridge between ECL1 and the re-entry helix [42]. The compact lid, which is mainly formed by ECL, then constitutes the second barrier at the extracellular interface. Once the lid has opened, it allows for substrate release into the extracellular space. Finally, ATP hydrolysis at one (or both sites) releases Pi and ADP, thus initiating the reset of the transporter with two NBDs in the original inward–open-facing state [43]. The resulting accessible ATP site also opens the intracellular gate enabling a new cycle of substrate recognition. This catalytic cycle reflects the current knowledge about the catalytic cycles of ABCG2/PDR transporters, whereby biochemical, structural, genetic and mutational data have been integrated.

**Figure 4.** Proposed catalytic cycles of mammalian/fungal type II exporters (ABCG/PDR). In the apo substrate-free state (**1**), the exporter adopts an inward-facing conformation. We propose that NBDs are open with ATP present in at least one NBD or both, which mediates partial NBD dimerization leaving only one accessible ATP-binding region. The intracellular gate(s) at the transmission interface provides access for substrate or inhibitor entry. The aromatic rings at the conserved F clamp form accessible binding sites at the closed transporter valve subtending the closed ECL. Drug substrates (**2a**) or inhibitors (**2b**) can enter through intracellular gate(s), preceding their trapping in distinct binding zones in the central cavity. Binding of ATP at the second binding site or both triggers full NBD dimerization and triggers a first conformational change, setting an occluded state (**3**). The communication between NBD and TMD is regulated via a rigid triple helical bundle (THB) as a key part of the transmission interface. The NBD dimerization compresses the central cavity space to drive substrate movement through the translocation, thus engaging a push and squeeze motion to open the valve. Substrates shift into the upper cavity and are released by the subsequent opening of the ECL lid (**4**). ATP hydrolysis at one NBD site may be enough to reset the catalytic cycle and to convert the transporter molecule into the inward-facing drug-recognizing state (**5**). The structures show NBDs (green), elbow helix (pink), TMDs (yellow), ICL (orange), ECL (purple), phenylalanine clamp (blue hexagon), valve (red), substrates (cyan) and inhibitors (red). For more details and references see the main text.
