How does WzmWzt recognize and transport mycobacterial cell wall galactan?

Why LLG needs an ATP-driven transporter

LLG is the galactan polymer that becomes the backbone of arabinogalactan, a core component of the Mtb cell wall and a recognized virulence factor. In Mtb it is built in the cytosol as a linear chain of up to about 30 galactose units on a decaprenyl-pyrophosphate-GlcNAc-Rha linker, and it must then cross the membrane to reach the periplasm. Moving this lipid-linked polymer across a membrane is energetically unfavorable, so the cell assigns the job to WzmWzt, a polysaccharide ABC transporter.

WzmWzt is organized as a tetramer: one Wzm transmembrane domain plus one Wzt nucleotide-binding domain form a half transporter, and two halves make the functional unit. ATP hydrolysis at the Wzt subunits supplies the energy, but the paper's question is more specific than energetics. It asks how the transporter actually grips a natural mycobacterial lipid-linked sugar and how ATP binding converts that grip into forward movement.

Structures captured in three nucleotide and lipid states

The authors co-expressed the Mtb genes rv3781 (Wzt) and rv3783 (Wzm) in M. smegmatis, a host system the group had used to capture endogenous mycobacterial lipids bound to membrane proteins. Wild-type WzmWzt did not purify stably, so they introduced the Walker B mutation E183Q into Wzt, which allowed purification and cryo-EM analysis.

From one dataset they resolved an NBD-open state at 2.91 Å and an NBD-closed, ATP-bound state at 2.78 Å. A second dataset, purified in the detergent DDM, yielded a 2.61 Å map of a DDM/ATP-bound intermediate. The reported LLG3 density was modeled as a lipid-linked trigalactoside with a farnesyl (C15) tail, a pyrophosphate-linked GlcNAc-Rha unit, and three galactofuranose residues. The authors describe this assignment as tentative, based on the shape of the density and its position relative to a glycolipid seen in the related transporter KpsM, so the sugar length and exact chemistry of the bound species remain a model rather than a chemically confirmed structure.

LLG sits at the junction of both transporter halves

In the ATP-free, inward-facing state, the central cavity of the Wzm domains is split into two symmetric halves by paired TM5 residues (Phe194, Gln198, Phe201, Phe202) and closed at the periplasmic side by the Phe75 pair of the periplasmic gate and the Ile207 pair. Two continuous, helical-shaped extra densities lie inside this opening, each running alongside a TMD-NBD interface and the TM1-TM5 interface. A C1 reconstruction reproduced the same two symmetric densities, supporting the interpretation that two LLG molecules bind per tetramer.

Each density extends from a lateral opening in the TMD up to loops on the Wzt α-helical domain. That geometry is the basis for the paper's claim that the substrate is recognized through two Wzm and two Wzt subunits rather than by the transmembrane domain alone. The authors also note that the galactoside chain length appears to set where the substrate sits in the cavity, which is why the modeled trigalactoside is treated as a positional guide rather than a complete picture of a 30-unit polymer.

A fully upward gate helix primes the two Wzt subunits

The most distinctive feature of the LLG-bound structure is the Wzt gate helix. In this state it extends fully upward toward the membrane and contacts the TMDs, sealing the cytosolic end of the lateral opening. Comparisons with the related transporters WzmWzt from A. aeolicus, KpsMT, and TarGH show that this full upward extension is not seen in those structures; the corresponding region of TarH forms a loop rather than a helix.

The authors interpret this upward position as a priming step: by reaching up to the TMDs and the bound lipid, the gate helix holds the nucleotide-binding domains in a configuration ready to close. Residues from the gate helix and from the Wzm N-terminal interface helix (IFN) participate in binding the pyrophosphate group of LLG3, with hydrogen bonds indicated between the protein and the substrate. This is the structural reason the paper argues against a purely "lipid-first" recognition model: both the lipid tail and the sugar portion contact the transporter.

ATP binding moves the gate helix about 40 Å downward

In the ATP-bound, NBD-closed structure, Mg2+-ATP sits at each nucleotide-binding site and the α-helical subdomain presses closer to the opposing RecA subdomain. The trigalactoside-binding pocket collapses. The gate helix undergoes a dramatic rearrangement: its C-terminal region becomes an ordered β strand running antiparallel to the ABCβ subdomain, while the upward-extending N-terminal portion is no longer resolved. Measuring between the Cα of Ala36 in the two conformations gives a distance of roughly 40 Å.

The consequence is that the gate helix leaves the TMDs and joins ATP binding and NBD dimerization. Aligning one TMD from the open and closed states shows the TMDs move as rigid bodies, and the interface residues on TM5 that were closed in the substrate-bound state open up and take part in substrate binding. The authors propose that dissociation of IFN Arg48 from the positively charged LLG3-binding region weakens the interaction and helps push the substrate through the TMD cavity. The DDM/ATP-bound intermediate, with two DDM molecules in the Wzm central cavity, captures a state between these endpoints.

Conditional mutants link specific residues to LLG transport

To test whether the structural contacts matter in cells, the authors built two conditional mycobacterial strains in which the relevant gene is expressed only when anhydrotetracycline is present: Msm c-NBD-tetON, which interrogates the Wzt NBD in M. smegmatis, and Mtb c-TMD-tetON, which interrogates the Wzm TMD in Mtb. Complementation with wild-type or mutant copies of the genes then shows which residues are required.

For the Msm strain, three assays were used: growth by CFU enumeration, accumulation of [14C]-LLG by whole-cell labeling and extraction, and LAM accumulation by immunoblotting, based on the idea that impaired arabinogalactan biosynthesis can cause LAM to build up. Wild-type wzt restored all three phenotypes, whereas the E183Q mutant and empty vector did not. Alanine substitutions at Wzm IFN residues Lys29 and Arg48, and at Wzt helical-domain residue Arg163, impaired LLG recognition and translocation in vivo. These functional results are what connect the cryo-EM states to a working transport cycle rather than to a static picture.

The periplasmic release step is still unresolved

The two gate residue pairs, Phe75 and Ile207, stay closed in all three structures, and no lateral opening appears on the periplasmic side. How the galactan actually exits into the periplasm is therefore not captured here. The authors note that an ATP-bound WzmWzt structure from M. abscessus in LMNG showed such an opening, but they caution that the detergent, rather than the LLG substrate, may modulate it.

The LLG3 assignment itself is tentative, and the authors state that delineating the full transport mechanism will require chemically synthesized intact LLG substrates and orthogonal validation of substrate assignment. The LLGR linker density at the cytosolic TMD-NBD interface is interpreted as a possible lateral entry site for substrate before transport, which is a proposal rather than a demonstrated step. These limits matter because the paper's therapeutic framing rests on the LLG3 and DDM binding sites being genuine conformational-specific pockets.

Galactan transport as an unexploited drug target

The paper's practical argument is a comparison within Mtb cell wall biosynthesis. The arabinan pathway already has established drugs, including ethambutol and the nitroimidazoles pretomanid and delamanid, but no inhibitors targeting galactan biosynthesis have been reported despite the pathway's importance and the vulnerability of its genes. The structures here supply LLG3- and DDM-bound pockets in defined conformational states, which the authors present as a starting point for conformational-specific inhibitors.

The most decision-relevant distinction in this paper is therefore not simply that WzmWzt has a structure, but which state a compound would need to trap. The fully upward gate helix defines the substrate-primed state, the downward helix defines the ATP-driven state, and the DDM/ATP intermediate defines a third. Because the periplasmic release step and the exact identity of the natural substrate remain open, the galactan transporter is a promising but still structurally incomplete target: the value of these maps lies in the state-specific pockets they define, not in a finished transport model.

Sources used in this answer