PAS DOMAINS IN BACTERIAL SIGNAL TRANSDUCTION

Abstract

PAS domains are widespread, versatile domains found in proteins from all kingdoms of life. The PAS fold is composed of an antiparallel β-sheet with several flanking α-helices, and contains a conserved cleft for cofactor or ligand binding. The last few years have seen a prodigious increase in identified PAS domains and resolved PAS structures, including structures with effector and other domains. New bacterial PAS ligands have been discovered, and structure-function studies have improved our understanding of PAS signaling mechanisms. The list of bacterial PAS functions has now expanded to include roles in signal sensing, modulation, transduction, dimerization, protein interaction, and cellular localization.

Graphical abstract

PAS DOMAINS: WIDESPREAD, VERSATILE AND ADAPTABLE

Per-ARNT-Sim (PAS) domains are widespread sensing, signaling and dimerization domains found in proteins from all kingdoms of life [1]. The number of identified PAS domains has rapidly risen due to increased sequencing and algorithm refinements. As of May 2020, Pfam 33.1 includes 222,716 PAS sequences of which 84% derive from bacterial proteins, 12% are from eukaryotic proteins, and 4% are archaeal [2]. PAS domains are associated with 18,219 different protein architectures [2], although in bacteria, they most often reside at the N-terminus of histidine kinases, diguanylate cyclases/phosphodiesterases, and methyl-accepting chemotaxis proteins, where they regulate the functions of those proteins [3].

PAS domains are 100–120 amino acids in length and classically have divergent sequences (<20% average pairwise sequence identity), but conserved three-dimensional structures. The globular PAS core comprises an antiparallel β-sheet (with strands Aβ, Bβ, Gβ, Hβ, and Iβ) and several flanking α-helices (Cα, Dα, Eα, and Fα) (Figure 1a–d). The PAS core can also have adjoining N-terminal (Aα′) or C-terminal (Jα) helices that pack against the β-sheet. The β-sheet is the most conserved region of PAS domains, whereas the orientation, length, and number of α-helices, as well as loop lengths, vary between domains [4]. Commonly, PAS domains assemble into homodimers through a patch of hydrophobic residues on the outer surface of their β-sheet, often with contributions from the adjoining N-or C-terminal helices (Figure 1e, [4]). Less commonly, the β-sheet interfaces with another domain, e.g., PAS/HAMP interaction in the aerotaxis receptor Aer (Figure 1f, [5]).

Figure 1.

Examples of PAS domain structures and models with bound cofactors and/or ligands. In all examples, the PAS β-strands occur in the order B-A-I-H-G, and the β-sheet is surrounded by helices C-D-E-F. Adjoining N-terminal Aα′ helices are also shown. (a-d) PAS monomer structures: (a) The PAS structural prototype photoactive yellow protein (PYP) from Halorhodospira halophila with its cofactor p-coumaric acid (PDB: INWZ). The inset shows p-coumaric acid covalently bound to PAS via an Eα-Cys residue and stabilized by Bβ-Tyr and Cα-Glu residues [30]. (b) Vibrio cholerae VqmA PAS with its ligand 3,5-dimethyl-pyrazine-2-ol (DPO) (PDB: 6UGL). The inset shows DPO, which is stabilized by PAS Cα-Tyr, Fα-Phe, Hβ-Phe, and Hβ-Lys residues [7]. (c) Cronobacter turicensis RpfR PAS with its ligand (2Z)-dodec-2-eoic acid (DSF) (PDB: 6DGJ). The disordered HI loop in RpfR is represented by a dashed line. The inset shows DSF, which is stabilized by Fα-Ser, Fα-Asn, Gβ-Tyr, Gβ-Arg and Hβ-Asn residues [9]. (d) Staphylococcus aureus WalK PAS with Zn²⁺ (yellow sphere) bound to the surface (PDB: 4MN5). The inset shows the four metal-coordinating residues, Asp and His residues on a loop N-terminal to Aβ, an Iβ-His residue, and a Glu residue on a loop C-terminal to Iβ [11]. (e-f) Example of PAS dimerization (e) via the PAS β-sheet and N-terminal Aα′ helix [Azotobacter vinelandii NifL PAS1 dimer with FAD (PDB: 2GJ3)], compared with (f) the PAS β-sheet interfacing with a HAMP domain instead (Escherichia coli Aer). Aer PAS and HAMP models are based on the structures of NifL PAS1 (PDB: 2GJ3) and Af1503 HAMP (PDB: 2ASX), whereas the Aer PAS-HAMP interface is based on experimental data [5]. (g-h) PAS structures in different signaling states: (g) Overlay of two Aer2-PAS structures from Pseudomonas aeruginosa including an unliganded dimer bound to heme cofactor (Fe³⁺, PDB: 4HI4, [25]) and a liganded monomer with cyanomet heme (Fe³⁺-CN, PDB: 3VOL, [31]). For clarity, b-type heme is only shown bound to the liganded structure. Unlike canonical PAS structures, Aer2 PAS has a combined Cα/Dα helix and an Eη (instead of Eα) helix. Arrows depict ligand-induced movements of the Aα′ helices and the DxT motifs. The top inset shows how the heme-bound ligand displaces the Hβ-Leu residue from the ligand-binding site and the Iβ-Trp rotates ~90° to H-bond with the ligand [24,25]. Although this structure has CN as the ligand, O₂ is the natural ligand for Aer2. The bottom inset shows the conserved DxT motif that couples Aer2-PAS to the C-terminal AS-1 (amphipathic sequence-1) helix of the adjoining HAMP domain. The Iβ-Asp of the DxT motif forms a putative salt-bridge with an Arg residue that begins PAS-Hβ, whereas the Thr residue (at the beginning of HAMP AS-1) H-bonds to the backbone amide of the subsequent Hβ-Arg. For clarity, only one liganded PAS monomer is shown, as is a portion of HAMP AS-1 C-terminal to the PAS domain. (h) Overlay of two YtvA-LOV monomers from Bacillus subtilis with bound FMN cofactor in dark (PDB: 2PR5), and blue-light illuminated (PDB: 2PR6), states. Both structures are bound to FMN, but for clarity, FMN is only shown bound to the light-illuminated structure. The inset shows how light induces the formation of a covalent bond between the Eα-Cys residue and FMN, which causes the side chain of the Iβ-Gln to flip 180° [32]. In the dark state, the side chain of the Eα-Cys assumes two distinct conformations, which are labelled “a” and “b” [32].

PAS domains were originally defined as intracellular sensors and it was recently argued that extracytoplasmic PAS domains should instead be classified as Cache domains [6]. For this review, we have only included PAS domains as defined by Pfam and have focused on recent findings on sensing and signaling in bacterial PAS, rather than Cache, domains.

PAS SENSORS: DIVERSE COFACTORS, LIGANDS AND SIGNALS

The best studied PAS domains have sensory functions derived from their ability to bind small molecules. The first small molecule identified was p-coumaric acid bound to PYP from Halorhodospira halophila (Figure 1a and Table 1), while the newest ligand is the pyrazine autoinducer 3,5-dimethyl-pyrazine-2-ol (DPO) bound to VqmA-PAS from Vibrio cholerae (Figure 1b and Table 1). PYP is a blue light sensor implicated in phototaxis, whereas VqmA is a member of the quorum-sensing LuxR-family that regulates gene expression (Figure 2a, [7,8]). Ligands can serve as either a primary PAS signal (like DPO binding to VqmA) or as a secondary signal after binding to an embedded cofactor. Examples include flavin cofactors, which are activated by electron reduction or photon absorption, and hemes, which are activated by gas binding (Table 1). Despite the diversity of PAS cofactors and ligands, most are bound within a conserved cleft formed by the inner surface of the PAS β-sheet and the Eα and Fα helices (Figure 1, [4]). The region around Eα-Fα is the least structurally conserved part of the PAS core [4]. In the case of VqmA, DPO recognition and binding requires four key PAS residues, including an Fα-Phe whose benzene ring provides π-π stacking interactions with the pyrazine ring of DPO (Figure 1b, [7]).

Table 1:

Cofactors and Ligands of Bacterial PAS Domains

Cofactor	Ligands/Signal	Example PAS Proteins
p-Coumaric acid	Blue light	Halorhodospira halophila PYP [30]
Flavin adenine dinucleotide (FAD)	Redox	Escherichia coli Aer [33,34] Azotobacter vinelandii NifL [35] Methylococcus capsulatus MmoS [36]
Flavin mononucleotide (FMN)	Blue light	Light-O2-Voltage (LOV) domains: Bacillus subtilis YtvA [37] Brucella abortus LOV-HK [38]
Riboflavin	Blue light	Erythrobacter litoralis EL346 [39]
Heme	O2, CO, NO	b-type heme: Bradyrhizobium japonicum FixL [40] E. coli DOS [41] Pseudomonas aeruginosa Aer2 [42] and BdlA [43] c-type heme: Geobacter sulfurreducens GSU0935 and GSU0582 [44]
FeS cluster	O2	Staphylococcus carnosus NreB [45]
-	Carbohydrates	Cellvibrio japonicus AbfS [46]
-	Fatty acids	Chloroflexus aurantiacus MltR [47] Mycobacterium tuberculosis Rv1364 Cronobacter turicensis RpfR [9]
-	Pyrazine	Vibrio cholerae VqmA [7]
-	Aromatic hydrocarbons	Pseudomonas putida TodS [48]
-	Metal2+	E. coli PhoQ [10] Staphylococcus aureus WalK [11]

Figure 2.

The diversity of PAS domain roles in bacterial proteins. All cartoons show the bacterial cytoplasmic membrane and the dimeric form of each protein. (a) PAS as a signal sensor as illustrated by VqmA from Vibrio cholerae. VqmA-PAS binds to the autoinducer 3,5-dimethyl-pyrazine-2-ol (DPO) and signals to a helix-turn-helix (HTH) DNA-binding domain to regulate transcription [7,8]. (b) PAS as a signal modulator as illustrated by Aer2 from V. cholerae. PAS1 modulates the extent of O2-mediated signaling from PAS2, which sends a conformational signal through two HAMP domains and a kinase control module (KCM) [15]. The KCM regulates a series of downstream chemosensory proteins. (c) PAS as a signal transducer as illustrated by BvgS from Bordetella bronchiseptica. BvgS transduces signals from a periplasmic Venus fly-trap (VFT) domain via a cytoplasmic PAS domain to C-terminal histidine kinase (HK; comprised of DHp and CA subdomains), response regulator (Rec), and histidine phosphotransfer (Hpt) domains, thus regulating kinase and phosphatase functions [16]. (d) PAS as a dimerization motif as illustrated by DgcE from Escherichia coli. The three DgcE PAS domains are essential for protein dimerization, which activates DgcE’s GGDEF (diguanylate cyclase) domain to produce c-di-GMP, and overcomes the anti-oligomerizing effect of the EAL (phosphodiesterase) domain [18]. PAS dimerization occurs after a protein complex binds to the membrane-associated sensor (MASE1) domain of DgcE. The EAL domain is degenerate and does not degrade c-di-GMP to guanosine monophosphate (GMP). (e) PAS as a protein interaction motif as illustrated by CetA and CetB from Camplyobacter jejuni. CetA and CetB constitute a bipartite energy-taxis system with the domains of the aerotaxis receptor, Aer, divided between two proteins. CetB is a PAS domain that is predicted to bind FAD, sense redox, and transmit signals through interaction with the HAMP domain of CetA [19]. (f) PAS as a cellular localization determinant as illustrated by CckA from Caulobacter crescentus. PAS-A directs CckA to either cell pole, whereas PAS-B targets CckA to new cell poles [20]. PAS-A and PAS-B also have sensory roles that regulate CckA’s kinase (HK; comprised of DHp and CA subdomains) and phosphatase (Rec) functions.

The most recently discovered bacterial PAS ligands are autoinducers and fatty acids, including fatty acid autoinducers (Table 1). The fatty acid autoinducer, diffusible signal factor (DSF) [(2Z)-dodec-2-eoic acid; an α,β-unsaturated carboxylic acid], was recently observed in the structure of the Cronobacter turicensis RpfR PAS domain (Figure 1c, [9]). The specificity of ligand binding was facilitated by the amide nitrogen of a PAS-Hβ Asn that interacts with the electron deficient C_β of DSF (Figure 1c). DSF binding to RpfR activates its phosphodiesterase domain, resulting in c-di-GMP degradation and bacterial biofilm dispersal [9].

In addition to PAS domains that bind ligands within a conserved cleft, some bind metal ligands to the PAS surface (Table 1). This was previously demonstrated for PhoQ from E. coli [10], and more recently for WalK from Staphylococcus aureus. WalK tetrahedrally coordinates Zn²⁺ to a negatively-charged region on the PAS surface where it negatively regulates WalK’s autokinase activity (Figure 1d, [11]).

Lastly, PAS domains can act as adjuvants for cofactor binding to other domains. Bacteriophytochromes, for example, harness a PAS-Aα′ extension to assist with GAF domain-biliverdin interactions (Table 2). The PAS-Aα′ extension threads through an elongated loop of the GAF domain, and a PAS-Cys residue covalently binds to the tetrapyrrole chromophore via a thioether linkage (see [12] and [13] for the highest resolution crystal and NMR structures of a photosensory module, respectively).

Table 2:

Bacterial Multi-Domain Structures Containing PAS Domains

Domain1	Structure Arrangement2	Protein (with PDB Identifier and Year Deposited)3
PAS	PAS-linker4-PAS PAS-linker-PAS-PAS GAF-PAS-PAS PAS-linker-PAS	M. capsulatus MmoS (PDB: 3EWK; 2008) Rhodobacter sphaeroides PpsR (PDB: 4HH2; 2012) M. tuberculosis Rv3651 (PDB: 4Q6U; 2014) B. abortus LOV-HK (PDB: 6PPS; 2019)
ACT	ACT-linker-PAS	E. coli TyrR (PDB: 2JHE: 2007)
ANTAR	PAS-ANTAR-linker-PAS	Nakamurella multipartita PAL (PDB: 6HMJ; 2018)
DHp	PAS-DHp-CA HAMP-linker-PAS-DHp-CA PAS-DHpL-CA	Thermatoga maritima ThkA (PDB: 3A0R; 2009) Streptococcus mutans VicK (PDB: 4I5S; 2012) E. litoralis EL346 (PDB: 4R3A; 2014)
GAF	N5-PAS-GAF N-PAS-GAF N-PAS-GAF-PHY N-PAS-GAF-PHY N-PAS-GAF N-PAS-GAF-PHY-linker-PAS N-PAS-GAF N-PAS-GAF-PHY-linker-PAS N-PAS-GAF-PHY N-PAS-GAF-PHY-linker-GGDEF N-PAS-GAF-PHY GAF-linker-PAS GAF-PAS-PAS	Bacteriophytochromes (Bph): Deinococcus radiodurans BphP (PDB: 1ZTU; 2005) Rhodopseudomonas palustris BphP3 (PDB: 2OOL; 2007) Synechocystis sp. Cph1 (PDB: 2VEA; 2007) P. aeruginosa BphP (PDB: 3C2W; 2008) R. palustris BphP2 (PDB: 4E04; 2012) R. palustris BphP1 (PDB: 4GW9; 2012) Stigmatella aurantiaca BphP (PDB: 4RQ9; 2014) Xanthomonas campestris BphP (PDB: 5AKP; 2015) Agrobacterium fabrum Agp1 (PDB: 5I5L; 2016) Idiomarina sp. PadC (PDB: 5LLX; 2016) A. fabrum Agp2 (PDB: 6G1Y; 2018) Others: E. coli DhaR (PDB: 4LRZ; 2013) M. tuberculosis Rv3651 (PDB: 4Q6U; 2014)
GGDEF	PAS-linker-GGDEF-linker-EAL	P. aeruginosa RbdA (PDB: 5XGB; 2017)
HAMP	HAMP-linker-PAS-DHp-CA	S. mutans VicK (PDB: 4I5S; 2012)
Hemerythrin	Hemerythrin-PAS	Pyrococcus furiosus PF0695 (PDB: 3CAX; 2008)
HTH	HTH-linker-PAS PAS-linker-HTH PAS-linker-HTH	C. aurantiacus MltR (PDB: 3PXP; 2010) E. litoralis EL222 (PDB: 3P7N; 2010) V. cholerae VqmA (PDB: 6IDE; 2019)
STAS	PAS-linker-STAS	B. subtilis YtvA (PDB: 2MWG; 2014)

¹PAS (Per-ARNT-Sim); ACT (Aspartate kinase, Chorismate mutase, and TyrA); ANTAR (AmiR and NasR Transcription Antitermination Regulators); DHp (Dimerization and phospho-accepting Histidine); GAF (cGMP-specific phosphodiesterases, Adenylate cyclases and FhlA); HAMP (Histidine kinases, Adenylate cyclases, Methyl-accepting chemotaxis proteins, and Phosphatases); HTH (Helix-Turn-Helix); STAS (Sulfate Transporter and AntiSigma factor antagonist).

²CA (Catalytic); DHpL (DHp-Like); PHY (Phytochrome).

³Where multiple structures exist for the same protein, the earliest structure is cited

⁴α-Helical linkers (e.g., Jα- or S-helices)

⁵N-terminal α-helix or unstructured coil containing the covalent Cys anchor for biliverdin

PAS DOMAINS AS SIGNAL MODULATORS AND TRANSDUCERS

A majority of PAS domains have not been shown to bind cofactors or ligands. For some of these domains, the ligand-binding cleft is too small to bind ligands; instead these domains function as protein interaction or dimerization domains and/or have roles in signal modulation or transduction. These functions differ from those of pure sensors. For example, the two PAS domains of the Xanthomonas campestris histidine kinase RavS regulate RavS function by inhibiting autokinase activity [14]. In RavS, the sensor role appears to be in the catalytic domain, which binds c-di-GMP. Complexity increases with V. cholerae Aer2, which has tandem PAS domains that both coordinate b-type heme and bind O₂ [15]. However, PAS2 is an O₂ sensor whereas PAS1 modulates the extent of O₂-mediated signaling from PAS2 (Figure 2b, [15]). PAS domains are often located between input and output domains such that signal transduction is likely to be a major role. The PAS domain of Bordetella BvgS, for example, transduces signals from a periplasmic Venus fly-trap sensing domain to a cytoplasmic histidine kinase (Figure 2c). However, BvgS PAS might also sense cytoplasmic signals, and could potentially override Venus fly-trap signals, thus tuning BvgS responses to the bacterial environment [16]. Another example is the histidine kinase NifL from Azotobacter vinelandii, which has tandem PAS domains: dimeric PAS1 binds FAD and serves as a redox sensor (Figure 1e), whereas PAS2 has no cofactor and transduces PAS1-generated redox signals via changes in PAS oligomerization [17].

PAS DIMERIZATION OR PROTEIN INTERACTION AS A REQUIREMENT FOR FUNCTION

PAS domains commonly dimerize, but this can also be essential for protein function. The three consecutive PAS domains of the E. coli diguanylate cyclase DgcE were recently shown to be essential for dimerization and activation of DgcE’s GGDEF (diguanylate cyclase) domain (Figure 2d, [18]). It was suggested that PAS occurs proximal to GGDEF domains because GGDEF functions as a dimer and can’t dimerize on its own. In DgcE, the three PAS domains are sandwiched between the membrane-bound signal input domain (MASE1) and the cytoplasmic catalytic (GGDEF) domain, and are also required for signal transduction. Some bacterial PAS domains also promote protein-protein interaction. CetA and CetB from Campylobacter jejuni comprise a bipartite Aer-type receptor, whereby the PAS-CetB sensor interacts with the CetA-HAMP domain to control the activity of the kinase control module (Figure 2e, [19]).

PAS DOMAINS AS REGULATORS OF CELLULAR LOCALIZATION

In addition to the recognized roles of PAS in sensing, signaling and dimerization, a newly identified role for PAS domains is to regulate cellular localization for the spatiotemporal control of cell signaling. This has been elegantly demonstrated for the bifunctional histidine kinase/phosphatase CckA from the asymmetrically-dividing bacterium Caulobacter crescentus. CckA has two PAS domains, PAS-A and PAS-B, both of which direct CckA to cell poles: PAS-A directs CckA to either cell pole, whereas PAS-B targets CckA to new cell poles (Figure 2f, [20]). PAS-A is also required for density-dependent kinase activity through PAS-A mediated oligomerization of CckA [20,21]. In contrast, PAS-B inhibits kinase activity and stimulates CckA phosphatase activity through c-di-GMP-binding and interaction with a kinase called DivL [20,21]. CckA thus uses its two PAS domains to integrate multiple signals and modulate kinase versus phosphatase activity depending on its subcellular location and environment. Kinase activity is favored at new cell poles, with subsequent phosphotransfer to the transcription factor CtrA, which controls the expression of cell cycle-regulated promoters. In addition to CckA, the single domain PAS protein MopJ from C. crescentus also localizes to cell poles [22]. In Sinorhizobium meliloti, the PAS domain of RgsP similarly regulates RgsP localization [23]. RgsP is a phosphodiesterase involved in cell wall biogenesis, and its PAS domain helps target RgsP to new cell poles and to cell division sites where it connects c-di-GMP signaling to the control of peptidoglycan synthesis [23].

PAS SIGNALING MECHANISMS

Signals that originate within PAS cores most often propagate from the cofactor/ligand site to the PAS β-sheet as a conformational or dynamic signal. Aer2 receptors, for example, have a PAS-heme whose ligand-binding site is occupied by an Hβ-Leu. When O₂ binds, the Hβ-Leu moves out of the ligand-binding site and an Iβ-Trp rotates ~90° to bond with and stabilize heme-bound O₂ (Figure 1g, [15,23,24]). This initiates a conformational signal at the PAS β-sheet that propagates to C-terminal HAMP and kinase control domains. Perturbations in PAS β-sheets can also cause flanking N- or C-terminal helices to reposition or unfold, altering dimer interactions and inducing signaling. Ligand-binding to Aer2-PAS moves the Aα′ helices ~4 Å (Figure 1g), and a PAS signal-on mutant has an unfolded Aα′ helix that is dissociated from the PAS core [15]. Perhaps the best understood mechanism comes from a subset of PAS called LOV (Light-O₂-Voltage) domains (Table 1). In LOV domains, light-induced protonation of the flavin cofactor (e.g., via formation of a thioadduct, Figure 1h) results in a PAS Iβ-Gln flipping its sidechain 180°; its oxygen-atom then bonds with the proton while the amide bonds with a residue on PAS-Aβ. This destabilizes interactions between the PAS β-sheet and flanking N- or C-terminal helices, promoting their detachment or rearrangement (reviewed by [26]). In a chimeric protein of the YtvA LOV domain (Figure 1h) fused to the effector module of FixL (YF1), weakening PAS β-sheet/Aα′ interactions rotates and tilts the PAS domains and separates the Jα helices, providing a mechanistic basis for signaling to the effector domain [26]. The tilting and rotating of short α-helices (many of which are amphipathic and form coiled-coils) appears to be a common theme in PAS-to-effector-domain signaling. PAS-Iβ often connects to this helix via a conserved “DxT” motif. The Asp residue of “DxT” forms a salt-bridge to a positively-charged PAS-GH loop residue, whereas the Thr actually begins the adjoining helix and H-bonds to a backbone amide in PAS-Hβ, thus providing a structural basis for coupling PAS to other domains (Figure 1g, [4]). The critical role of the “DxT” motif is evident for the Treponema denticola histidine kinase Hpk2, where an Asp to Lys mutant reduces PAS cofactor (hemin) binding, protein dimerization, and autophosphorylation [27].

The last decade has seen substantial increases in PAS structures containing effector and other domains (Table 2), but few structures have been resolved for the same protein in different signaling states. Generally, PAS signaling is thought to alter effector domain dynamics and quaternary structure. For example, DPO-binding to VqmA likely stabilizes the PAS β-sheet and modulates the PAS-effector domain interface, promoting enhanced DNA binding (Figure 2a, [7]). Similarly, zinc-binding to WalK stabilizes PAS and potentially modulates the angle between the PAS and C-terminal catalytic domains, regulating kinase activity [11]. Bacteriophytochrome PAS domains show minimal crystallographic changes between dark-adapted and photoactivated states (e.g., [28]), but recent NMR data suggests substantial light-induced remodeling at the PAS-GAF interface that could be involved in signaling [13]. Finally, protein interactions can also induce conformational shifts in PAS domains. When the DhaL kinase binds to the E. coli transcription factor DhaR, the DhaR coiled-coil linker region reorganizes and the connected PAS domain pivots, displacing some Cα atoms up to 15 Å; this triggers DhaR’s ability to activate gene expression [29].

CONCLUSIONS

The last few years has seen a surge in identified PAS domains, resolved PAS structures, and multidomain structures. New PAS ligands have been discovered, and structure-function studies have improved our understanding of PAS sensing and signaling mechanisms. The list of PAS functions is growing, highlighting their incredible versatility and utility. In addition to classic roles as sensors and dimerization domains, PAS domains modulate protein function and interactions, transduce signals, and determine cellular localization. Future progress will be facilitated by developing tools to identify non-covalently bound PAS ligands, and by exploiting PAS-effector domains in different signaling states to uncover structure-function relationships.

HIGHLIGHTS.

• PAS domains are common sensing, signaling, dimerization, and localization domains

• Sensing versatility derives from a conserved cleft that binds cofactors and ligands

• PAS signaling involves repositioning of the PAS β-sheet and adjoining helices

• PAS connects to diverse effector domains to regulate distinct cellular functions

• PAS-effector domain structures have begun to clarify protein signaling mechanisms