Bacterial Small Membrane Proteins: the Swiss Army Knife of Regulators at the Lipid Bilayer

ABSTRACT

Small membrane proteins represent a subset of recently discovered small proteins (≤100 amino acids), which are a ubiquitous class of emerging regulators underlying bacterial adaptation to environmental stressors. Until relatively recently, small open reading frames encoding these proteins were not designated genes in genome annotations. Therefore, our understanding of small protein biology was primarily limited to a few candidates associated with previously characterized larger partner proteins. Following the first systematic analyses of small proteins in Escherichia coli over a decade ago, numerous small proteins across different bacteria have been uncovered. An estimated one-third of these newly discovered proteins in E. coli are localized to the cell membrane, where they may interact with distinct groups of membrane proteins, such as signal receptors, transporters, and enzymes, and affect their activities. Recently, there has been considerable progress in functionally characterizing small membrane protein regulators aided by innovative tools adapted specifically to study small proteins. Our review covers prototypical proteins that modulate a broad range of cellular processes, such as transport, signal transduction, stress response, respiration, cell division, sporulation, and membrane stability. Thus, small membrane proteins represent a versatile group of physiology regulators at the membrane and the whole cell. Additionally, small membrane proteins have the potential for clinical applications, where some of the proteins may act as antibacterial agents themselves while others serve as alternative drug targets for the development of novel antimicrobials.

INTRODUCTION

Bacteria employ a wide array of regulatory mechanisms at various stages of gene expression to adapt to changing environments (1–3). While the roles of conventional transcription factors (4, 5) and regulatory RNAs (6–8) are well established, the significance of small proteins as regulators of gene expression and bacterial physiology is just beginning to be unraveled. Small proteins, typically ≤100 amino acids, are encoded by short, noncanonical, open reading frames. These genes were previously missed due to length cutoffs in initial genome annotations. In the past decade and a half, the development of methods specifically suited for small protein studies involving bioinformatics, transcriptomics, ribosome profiling, and proteomics have led to the identification of numerous small proteins in bacteria and eukaryotes, including humans (9, 10). Among prokaryotes, hundreds to thousands of small open reading frames (sORFs) have been reported in Escherichia coli (9), Salmonella enterica (11, 12), Mycoplasma pneumoniae (13), Mycobacterium tuberculosis (14), Listeria monocytogenes (15), Staphylococcus aureus (16) and human (17) microbiomes. Only a small subset of these proteins has been investigated so far, but it is evident that small proteins participate in a wide variety of cellular processes, such as cell division, signal transduction, regulation of transporters and drug efflux pumps, and stress responses (9, 18). A majority of the sORFs have yet to be validated for protein expression, and, for small proteins whose expression is confirmed, the functions have yet to be uncovered.

About 30% of all the small proteins identified in E. coli are predicted and/or shown to be localized to the membrane (9, 19, 20). A similar fraction (∼35%) of the total number of these newly found proteins in human microbiomes is predicted to contain a transmembrane helix (17). The number of small membrane proteins that have been identified may vary based on different factors, such as the organism of study, technique, and/or the prediction algorithm used to identify them. However, this subclass undoubtedly represents a significant proportion of small proteins with a great potential for regulatory functions at the membrane. Bacterial membranes contain various integral and peripheral membrane proteins that function as environmental receptors, transporters, pores, channels, chaperones, and enzymes (21). Consequently, small proteins at the membrane potentially have diverse roles in modulating transport, signal transduction, cell division, respiration, sporulation, and membrane integrity. Systematic functional analyses of the small membrane proteins are just beginning. The small size and hydrophobicity of these proteins place serious limitations on their identification and purification. In vitro characterization of membrane proteins also requires solubilization using detergents and other amphipathic reagents to maintain a membrane-like environment. Despite these hurdles, a combination of biochemical, genetic, proteomic, and structural approaches has deepened our understanding of several small membrane protein regulators. To this end, our review highlights the recent advances in our knowledge of the functions of selected bacterial small membrane proteins (Table 1). Readers may refer to other recent reviews for a more comprehensive overview of all bacterial small proteins (9, 22, 23). Here, we described and categorized some of the well-studied small membrane proteins and their biological functions in major cellular processes. We discussed structural details underlying the functions of these small proteins for those with available structures. Most of the small proteins listed below are integral membrane proteins predicted to contain at least one transmembrane α-helix.

TABLE 1.

Bacterial small membrane proteins and their regulatory functions. Small proteins, ≤100 aa, predicted to contain at least one transmembrane α-helix or shown to be associated with the membrane, and their known functions are described below

Function and small protein	Length (aa) (Species)	Function
Transport
MgtR	30 (S. enterica)	Promotes degradation of Mg2+ transporters MgtA, MgtB, and virulence protein MgtC associated with phosphate transport by membrane-bound protease FtsH (24, 25, 29, 32, 33)
MgtU	28 (S. enterica)	Prevents MgtB degradation by FtsH (25)
MgtS	31 (E. coli)	Prevents MgtA degradation by FtsH and regulates Mg2+ efflux by PitA phosphate symporter (25, 31, 34)
KdpF	29 (E. coli)	Stabilizes ATP-dependent potassium importer Kdp complex (26, 36, 37)
AcrZ	49 (E. coli)	Interacts with the AcrAB-TolC multidrug efflux pump and aids in efficient transport of antibiotics (27, 35, 38–40)
SgrT	43 (E. coli)	Inhibitor of PtsG glucose transporter (41–43)
Signal transduction
MgrB	47 (E. coli)	Inhibits PhoQ sensor kinase (49–53)
SafA	65 (E. coli)	Activates PhoQ sensor kinase (55–60)
Stress response
PmrR	29 (S. enterica)	Inhibits LpxT, a membrane-bound enzyme that phosphorylates lipopolysaccharides (61, 63)
YshB	36 (E. coli, S. enterica)	Promotes replication within the host cells and regulates virulence (19, 66)
Prli42	31 (L. monocytogenes)	Interacts with RsbRST stressosome complex and mediates activation of the general stress response (15, 69)
YohP	27 (E. coli)	Compresses the nucleoid upon overexpression (64, 70, 71)
Respiration
CydX	37 (E. coli)	A subunit of the cytochrome bd-I complex and required for cytochrome oxidase activity (72–77)
CydH/CydY	29 (E. coli)	Associates with cytochrome bd-I complex and blocks the oxygen entry channel (78–80)
TorE	56 (S. oneidensis)	Acts as a membrane tether for the trimethylamine oxide complex for efficient catalysis (81, 82)
Cell division
SidA	29 (C. crescentus)	Inhibits cell division by interacting with FtsW/FtsN (83)
DidA	71 (C. crescentus)	Inhibits cell division by interacting with FtsW/FtsI/FtsN (84)
SosA	77 (S. aureus)	Inhibits cell division postseptal ring formation (85, 86)
Blr	41 (E. coli)	Localize to the septal ring and may regulate cell division under hypoosmolar conditions (87–89)
YmgF	72 (E. coli)	Localize to the septal ring and may regulate cell division under hypoosmolar conditions (87–89)
DrpB	100 (E. coli)	Localize to the septal ring and may regulate cell division under hypoosmolar conditions (87–89)
Sporulation
SpoVM	26 (B. subtilis)	Localizes itself to the convex membrane, interacts with sporulation protein SpoIVA and required for proper spore formation (90–98)
CmpA	37 (B. subtilis)	Targets SpoIVA for degradation by ClpXP protease by binding both SpoIVA and ClpX (99)
Toxins
HokB	49 (E. coli)	Form pores in the cell membrane causing leakage (101, 107–112, 124)
PepA1, PepA2, PepG1, PepG2	30 to 44 (S. aureus)	Form pores in the cell membrane causing leakage (101, 107–112, 124)
TisB	29 (E. coli)	Form pores in the cell membrane causing leakage (101, 107–112, 124)
ZorO	29 (E. coli)	Form pores in the cell membrane causing leakage (101, 107–112, 124)
TimP	38 (S. enterica)	Causes membrane leakage either directly or indirectly (102)

SMALL MEMBRANE PROTEINS TARGETING TRANSPORT

Small proteins interact with various membrane-located transporters (e.g., cation transporters [24–26], multidrug efflux pump [27], and the glucose transporter [28]), regulating their stability and function (Fig. 1). In Salmonella, the small protein MgtR (30 amino acids [aa]) promotes the degradation of magnesium transporters—MgtA and MgtB—and a virulence factor MgtC that is associated with phosphate uptake within host cells (24, 25, 29, 30). On the other hand, the small proteins MgtS (31 aa) and MgtU (28 aa) protect MgtA and MgtB from degradation in E. coli and Salmonella, respectively (25, 31). It appears that the abundance of magnesium transporters is meticulously regulated by an intricate network of small proteins at the posttranslational level. Further details of these small proteins and their mechanism of action are discussed below.

FIG 1.

Small proteins targeting transporters. The inner membrane surrounding the cytoplasm of a Gram-negative bacterium is shown. Small proteins MgtR, MgtU, MgtS are important for magnesium homeostasis through interactions with their targets (MgtA and/or MgtB and PitA as described in the text). SgrT binds and blocks the function of PtsG, a major glucose transporter. KdpF associates and stabilizes the KdpABC potassium transporter, and AcrZ binds to AcrB and regulates drug export through the AcrAB-TolC efflux pump. The KdpFABC and AcrBZ complexes were generated using PDB ID 7BH2 and 4CDI and the figure was created with www.BioRender.com.

MgtR consists of a single putative transmembrane helix (TMH) and is located in the inner membrane of Salmonella with an N-in-C-out topology (29). MgtR reduces the cellular copy numbers of MgtA, MgtB, and MgtC by presenting them to the membrane-bound protease, FtsH, which leads to a decrease in intramacrophage survival and virulence of Salmonella (24, 25, 29). Initial mutational studies indicated that MgtR interacts with MgtC mainly through its transmembrane helical region via Ala/Ser motifs (29). A later in vitro structural study revealed that MgtR TMH forms a complex with MgtC TMH4 (although from a different organism) on a native-like lipid bilayer (32). The complex is maintained via multiple Van der Waals interactions between the side chains throughout the helical interface. Compared to its affinity to MgtC, MgtR binds slightly weaker to MgtA and MgtB (25). Interestingly, MgtR and a MgtR variant with a point mutation (S17I), which loses its ability to interact with MgtC endogenously, appear more promiscuous when added exogenously (29, 33). In addition, these small proteins reduced the amount of several inner membrane proteins, such as PhoQ, EnvZ, and MalG. This promiscuity in targeting does not appear to depend on the presence of the Ala/Ser motif, a cysteine residue, or the C terminus of MgtR. Furthermore, the decrease of targeted inner membrane proteins does not depend on FtsH or other proteases tested, indicating the existence of an entirely different mechanism of action (33). Based on the current gene database in the National Center for Biotechnology Information, mgtR was found in pathogenic species in Salmonella, Klebsiella, Pluralibacter, Pantoea, and Citrobacter.

In Salmonella, the mgtR gene is in the same operon as another small protein gene, mgtU. MgtU (28 aa) protects MgtB, but not MgtA or MgtC, from FtsH-dependent degradation (25). It promotes the survival of Salmonella under low magnesium concentration as well as oxidative stress. MgtU has a putative TM helix, and mutational analysis showed that multiple clustered bulky hydrophobic residues near the C-terminal end of its TMH are essential for its activity. The exact binding mechanism and how MgtU prevents MgtB degradation are yet to be investigated. MgtR and MgtU regulate the abundance of the magnesium transporter MgtB in a temporal manner (25). The former dominantly promotes MgtB degradation at early growth times, and the latter exerts a major effect when magnesium availability is extremely low at late growth stages. Notably, the transcription of the entire operon is regulated by the PhoQ/PhoP two-component system. The amount of mgtR and mgtU transcripts are presumably at a comparable level. However, mgtU has its own ribosomal binding site, and the relative amount of these two small proteins vary in time (25). It will be interesting to discover and compare the half-life, abundance, binding site, and affinity of these two small proteins to their common target MgtB at different growth stages to have a better mechanistic understanding of the regulation dynamics.

In E. coli, the small protein MgtS (31 aa) regulates the stability of the magnesium transporter MgtA by protecting it from FtsH-mediated degradation (Fig. 1) (31). MgtS is a bitopic membrane protein with a C-in-N-out topology. The expression of MgtS is also regulated by the PhoQ/PhoP two-component system and is induced during extreme magnesium deficiency (31). Mutational studies showed that individually substituting the two C-terminal aspartates with alanine almost entirely abolishes the function of MgtS (31). Unexpectedly, the S. enterica MgtS, which has 80% sequence identity to the E. coli homolog, did not appear to regulate the amount of MgtA, at least in the tested conditions (25). It is unclear, however, whether Salmonella MgtS was expressed in this case, as E. coli MgtS was detected only when magnesium concentration was extremely low (31). Overexpressing MgtS in E. coli led to the discovery of its second protein target, PitA, a cation-phosphate symporter (34). High levels of MgtS have been shown to impact PitA function, increasing its phosphate import and preventing the leakage of Mg²⁺ ions. Changes in MgtS levels in the cell appear to have a mild effect on PitA abundance (35). These observations are consistent with a model where MgtS regulates both PitA and MgtA. The physiological role of MgtS converges to increase the cytosolic magnesium concentration by stabilizing the magnesium transporter MgtA and preventing magnesium efflux through PitA during magnesium starvation. MgtS is relatively more widespread in bacteria and does not necessarily cooccur with MgtR.

Small proteins are involved in the transport of other ions as well, such as potassium transport in E. coli (26). At low potassium concentrations, the small protein KdpF (29 aa) is induced by the KdpD/KdpE two-component system and forms a stable complex with the KdpABC, an ATP-dependent potassium translocator on the membrane (Fig. 1). The KdpFABC complex was purified, and its structure and different functional states were revealed by X-ray crystallography (36) and cryo-electron microscopy (cryo-EM) (37). In both studies, KdpF had an α-helical conformation located between KdpA and KdpB that interacted with both subunits. The conformation of KdpF stays the same in complexes with different functional states, indicating that KdpF is primarily involved in maintaining complex structural integrity. This is consistent with the earlier observation that KdpF stabilizes the detergent-solubilized-Kdp complex in vitro but is functionally dispensable in vivo under the growth conditions tested (26). Interestingly, in vitro, the KdpABC complex could also be stabilized by high concentrations of E. coli lipids, and it was suggested that KdpF acts as a lipid-like stabilizing element (26).

Another small protein interacting with a large membrane protein assembly is AcrZ. AcrZ (49 aa) complexes with the multidrug efflux pump, AcrAB-TolC, and influences its export of certain antibiotics (27). Structural studies revealed that AcrZ has a long helical structure with a C-in-N-out topology and is situated in the inner membrane at a 45° angle to the lipid plane. It interacts with AcrB exclusively, wrapping around the TMHs of AcrB from the outer surface of the TM helix bundle (Fig. 1) (35, 38–40). When reconstituted with AcrB on a lipid-disc, which closely resembles the in vivo situation, AcrZ showed a prominent kink due to a proline residue at position 16 (35). Replacement of this proline to alanine almost entirely abolishes the interaction between ArcZ and AcrB. Alanine replacement of individual interfacial residues, however, does not have a significant effect, indicating that the overall shape of AcrZ is essential for the small protein to fit into the binding groove on AcrB and that the binding surface likely contains a collection of multiple weak intermolecular forces. The interruption of interactions at one position by point mutation, thus, does not significantly affect the overall affinity. The binding of AcrZ exerts an allosteric effect on AcrB and is proposed to change the conformation of the drug entry channels and binding sites (35).

Small protein SgrT is translated from one of the few known dual-function sRNAs, SgrS, under glucose-phosphate stress (28). SgrT (43 aa) targets the major glucose transporter PtsG and inhibits its function without changing its abundance (Fig. 1) (41, 42). Additionally, the SgrS sRNA base-pairs with the ptsG mRNA destabilizes it in an Hfq-dependent manner, represses translation, and reduces the amount of newly synthesized glucose transporter (43, 44). With the two independent actions, SgrS efficiently inhibits glucose uptake, prevents inducer exclusion, and relieves glucose-phosphate stress in E. coli. Unlike other small proteins targeting membrane-located transporters, SgrT does not have a predicted transmembrane region. It is recruited to the membrane by interacting with the membrane-bound domain of PtsG, though the exact residues that SgrT interacts with are still under debate (41, 42), and the mechanism of how SgrT inhibits PtsG remains unknown.

SMALL MEMBRANE PROTEINS MODULATING SIGNAL TRANSDUCTION

The PhoQP two-component system is a well-studied signaling pathway present in E. coli, Salmonella, and related bacteria. The PhoQP signaling system regulates virulence and responds to low magnesium, acidic pH, and stress caused by cationic antimicrobial peptides and hyperosmotic conditions (45, 46). In the presence of a signal, the sensor PhoQ is activated by autophosphorylation, and the phosphate group is transferred to the response regulator PhoP, which then stimulates PhoQP-dependent gene expression (47, 48). MgrB (47 aa), a small protein localized to the inner membrane binds to PhoQ and inhibits its kinase activity and promoting the dephosphorylation of PhoP (49, 50). Transcription of the sORF for mgrB is itself activated by the PhoP response regulator, making MgrB a negative feedback regulator of the PhoQP system (Fig. 2). It is possible that a small protein regulator of a global network could serve as a point of control for many distinct input signals. Indeed, previous studies have shown that MgrB, which has two conserved Cys residues in its periplasmic region, can act as a redox sensor in the periplasm (51), which can modulate gene expression through the PhoQP pathway. More recently, the small size of MgrB and the ability to tag it with fluorescent proteins has led to detailed analyses of the molecular interactions between MgrB and its target PhoQ, identifying specific residues in the cytoplasmic, transmembrane (TM), and periplasmic regions critical to MgrB’s function (52). Hyperactivation of the PhoQP system in the absence of MgrB leads to disruption of cell division, leading to filamentous growth in E. coli (53). Furthermore, inactivation or downregulation of MgrB is a key mechanism by which Klebsiella pneumoniae acquires resistance to last-resort drugs like colistin in clinical settings (54).

FIG 2.

Small membrane proteins modulating signal transduction and stress responses in bacteria. Small proteins MgrB, SafA, PmrR, and Prli42 are all localized to the cell membrane (inner membrane in case of Gram-negative bacteria). In E. coli, MgrB and SafA interact with the target PhoQ sensor kinase. In S. enterica, PmrR binds to LpxT, an enzyme involved in lipopolysaccharide modification. In L. monocytogenes, Prli42 anchors RsbR, a component of the multiprotein stress complex (stressosome). The cartoon for the stressosome complex was generated using PDB ID 6QCM, and the figure was created with www.BioRender.com.

Interestingly, the E. coli PhoQ is also regulated by a second small protein called SafA (65 aa) (55, 56). Expression of safA is controlled by an acid-responsive two-component system, EvgSA, which makes SafA a connector between the two signaling pathways, EvgSA and PhoQ-PhoP. In contrast to MgrB’s role as an inhibitor, SafA is an activator of PhoQ. The C-terminal periplasmic domain (41 to 65 aa) of SafA interacts with the sensor domain in the periplasmic region of PhoQ, thereby increasing PhoQ autophosphorylation (57, 58). Recent crystal structures of the PhoQ sensor domain and its variant D179R revealed an internal cavity, which facilitates interactions between SafA and PhoQ (58). SafA’s TM sequence is not important for its PhoQ activation function, in contrast with MgrB where specific residues in the TM are critical for PhoQ inhibition. By connecting the EvgSA and PhoPQ signaling pathways, SafA was also shown to play an important role in acid resistance during the exponential growth phase in certain E. coli isolates (55, 59, 60).

SMALL MEMBRANE PROTEINS MEDIATING STRESS RESPONSES

In Salmonella and related pathogens, the PmrBA two-component system responds to elevated levels of Fe³⁺ in the environment and regulates gene expression required for the modification of the lipopolysaccharide (LPS) to protect cells from Fe³⁺ toxicity (61, 62). A 29 aa small protein called PmrR is transcriptionally regulated by the PmrBA two-component system (61). PmrR, which is localized to the inner membrane, plays a key role in reducing the net negative charge on the LPS by directly binding and inhibiting an inner membrane enzyme (LpxT) involved in the phosphorylation of the lipid A component of the LPS (Fig. 2). A decrease in the overall negative charge results in lower binding of Fe³⁺ ions to the LPS and, consequently, lower activation of the PmrBA, constituting a negative feedback mechanism for PmrA-regulated gene expression. By extension, PmrR-dependent LPS modifications may modulate resistance to different ionic compounds based on the net charge status. For instance, inhibition of LpxT by PmrR under PmrBA activation conditions increased the susceptibility of E. coli to deoxycholate, one of the bile acids, and affected intestinal colonization (63).

Some small membrane proteins were found to accumulate to high levels under specific stress conditions (64), which can play an important role as a “stress-induced regulator.” A 36 aa protein, YshB, whose expression was first validated using a SPA-tagged version of the protein (19) is predicted to be membrane-localized. YshB is expressed abundantly under stress or stationary phase conditions in E. coli (65), and homologs of YshB are also found in pathogenic Salmonella, Klebsiella, and related enterobacteria. In S. enterica, yshB expression is upregulated after entry into host phagocytes, and a yshB null mutant was only partially virulent in a mouse model of infection (66). Although the details of how YshB affects virulence are not fully understood, this study is an illustration of how small membrane proteins conserved across different bacteria may have specialized roles under specific stresses.

In Gram-positive bacteria, a large cytosolic complex (1.8 MDa) called the “stressosome” senses changes in the environment and mediates the general stress response (67, 68). The stressosome consists of three proteins, RsbR, RsbS, and RsbT, and initiates the sigma B-mediated signaling pathway to adapt and survive under stress. An N-terminal, proteomics-based approach was used to discover a 31 aa membrane protein, Prli42, which is required for activation of the stressosome in Listeria monocytogenes (15). Prli42 is conserved among Firmicutes and is anchored to the membrane via the C-terminal tail. The N-terminal region of Prli42 interacts with its cytoplasmic protein partner RsbR, a component of the stressosome, through the basic residues K5 and R8 and is responsible for tethering RsbR to the membrane (Fig. 2). A recent follow-up study using cryo-EM identified a set of acidic residues in RsbR (E109, E110, E130) potentially important for binding Prli42 through interactions with the complementary basic residues K4, K5, R8 (69). Indeed, a triple mutant RsbR-QQQ with three glutamate residues replaced showed a significant reduction in binding Prli42. Furthermore, RsbR-QQQ failed to complement the loss of viability of ΔrsbR strain under oxidative stress, suggesting that Prli42 might be a sensor that communicates the stress signal to the stressosome via its interaction with RsbR (69).

In E. coli, the 27 aa small membrane protein YohP showed differences in the level of expression in a condition-specific manner (64). YohP levels were shown to be significantly higher when cells are grown in minimal glucose medium versus Luria broth, in oxygen-limiting versus aerobic conditions, and under envelope stress or hydrogen peroxide stress versus no stress. How the modulation in YohP levels affects bacterial physiology is not understood yet. Overexpression of YohP in E. coli caused nucleoid condensation (70)—a strategy proposed to protect cells against stress-induced damage by decreasing transcription and translation (23). Furthermore, YohP has been a model protein to study the topology and membrane targeting pathway for small proteins, where YohP was found to interact posttranslationally with the signal recognition particle (SRP) (70, 71). Site-specific, cross-linking experiments performed both in vivo and in vitro also indicated that SRP recognizes YohP in a ribosome-independent mechanism, which is in stark contrast to the canonical cotranslational ribosome-SRP-mediated pathway used for membrane protein translocation. A detailed discussion of small membrane protein targeting by SRP can be found in recent publications (23, 70).

SMALL MEMBRANE PROTEINS ASSOCIATED WITH THE RESPIRATORY CYCLE

Two small membrane proteins were found to associate with cytochrome oxidase complexes. One of these proteins, CydX, is 37 aa and encoded by a sORF in the cydAB operon (72). CydX was first identified in E. coli as part of a biochemical study of cytochrome bd oxidase (73). Bioinformatic analyses found that cydX is widespread in eubacteria. Interestingly, its presence was correlated with the occurrence of a CydA variant that has a longer quinol-binding domain or the Q-loop (74). Recent cryo-EM structures of the E. coli cytochrome bd-I oxidase revealed that CydX is part of this complex and binds CydA. CydX was essential for cytochrome bd oxidase function (72) and may contribute to the folding and stability of the complex (75). Consistent with the vital role of cytochrome oxidase complex in bacterial survival within the host, CydX was shown to be critical for replication within macrophages and the stress response to nitric oxide in S. enterica (76). In Brucella, CydX plays a key role in combating hydrogen peroxide and acid stress (77). Another small membrane protein CydH (or CydY) is 29 aa, bound to this complex, and makes contact with the CydA Q-loop region at a site distinct from the CydX/CydA interface (78, 79). In the structure, CydH blocks the oxygen entry site of the cytochrome bd oxidase complex, and CydH may have more than just a structural role, perhaps, in the evolution of this complex with a secondary channel to transmit oxygen (80).

Another example of a small membrane protein involved in respiration is the 56 aa TorE from Shewanella oneidensis MR1, which is involved in the proper assembly and stabilization of the alternative trimethylamine oxide (TMAO) respiratory complex (81). Specifically, TorE interacts with cytochrome type c oxidase, TorC and anchors it to the membrane enhancing its catalytic efficiency. Deletion of torE leads to a decrease in TMAO reduction in vivo (82).

SMALL MEMBRANE PROTEINS AFFECTING CELL DIVISION

Two small proteins—SidA and DidA—act as cell division inhibitors in Caulobacter crescentus. SidA is an SOS-inducible, 29 aa, transmembrane protein encoded by a sORF containing a LexA-binding site in its promoter region (83). Overexpression of SidA is sufficient to cause a block in cell division. SidA directly interacts with a late cell division protein, FtsW, in a bacterial two-hybrid assay and the two proteins together can form a complex together with FtsI (83). It is proposed that SidA might be degraded in the membrane via a protease-dependent mechanism. The second membrane protein, DidA (71 aa), is an SOS-independent cell division inhibitor, which also interacts with FtsW/FtsI and is regulated by a transcription factor DriD (84). DriD itself is activated by DNA damage. However, the details of this activation mechanism are unknown. Together SidA and DidA represent redundant DNA damage checkpoints in C. crescentus cells under stress conditions. In Staphylococcus aureus, a 77 aa membrane protein called SosA serves as an SOS-induced, late-stage, cell division inhibitor. The sORF for sosA is transcribed divergently from the lexA promoter (85). Interestingly, SosA accumulates in cells lacking the ctpA gene, which encodes a membrane-bound protease, suggesting that SosA can be cleaved by a CtpA. This mechanism where a membrane-associated SOS-induced cell division inhibitor is cleaved by a membrane-bound protease is also conserved in B. subtilis (86).

Three other nonessential small membrane proteins—41 aa Blr, 72 aa YmgF, and 100 aa DrpB—have been shown to associate with cell division proteins and localize to the septal ring in E. coli (87–89). These proteins are thought to be important for modulation of division under low osmolarity and other stress conditions; however, details of their precise roles have yet to be investigated.

SMALL MEMBRANE PROTEINS REQUIRED FOR SPORULATION

In Bacillus subtills, the small protein SpoVM (26 aa) is involved in spore formation. Mutants with dysfunctional SpoVM fail to produce viable spores (90, 91). SpoVM serves as a landmark molecule and specifically localizes to the outer membrane of the forespore, anchoring the surrounding proteinaceous spore coat (92, 93). Remarkably, this small protein can detect the positive curvature of the forespore outer membrane by its slightly increased binding affinity and cooperativity toward convex membranes (94–96). SpoVM consists of a flexible N terminus and a short amphipathic α-helix near its C terminus, which sits relatively deep in the lipid bilayer parallel to the plane of the membrane (92, 94, 95, 97). The only charged residue in the helical region, R17, presumably snorkels up to the aqueous/lipid interface, maintaining the delicate balance of overall hydrophilic/hydrophobic interactions as well as its deep membrane location. Mutating R17 to aspartate reduces the ability of SpoVM to localize in the membrane and leads to its cytosolic distribution (94). The flexibility of the N terminus is ensured by a conserved proline residue at position 9. The precise position of this helix breaker is essential for SpoVM preferential localizing to the convex membrane and its interaction with other proteins during spore formation (95). Replacing Pro9 with Ala leads to an elongated α-helical region, a rigid N terminus, an off-balance of hydrophilic/hydrophobic forces along the two sides of the helix, and loss of membrane location specificity. Moving the proline away from the N terminus retains its ability to target the convex membrane but results in a deficiency in sporulation (95). SpoVM was found to interact in vivo with SpoIVA, a structural spore coat protein, which polymerizes and prevents the dissociation of SpoVM from the membrane (92, 98). Interestingly, in vitro analysis indicated that SpoVM is a substrate of the membrane-bound protease FtsH (90), which might also play a role in sporulation.

In the same pathway as SpoVM, another small protein CmpA (37 aa), is required for proper sporulation (99). CmpA is transiently localized on the surface of the forespore in a SpoVM-dependent manner and serves as a checkpoint for proper spore coat assembly. CmpA targets SpoIVA for degradation by the ClpXP protease by interacting directly with both SpoIVA and ClpX, thereby inhibiting sporulation. CmpA is degraded posttranslationally to allow sporulation to progress. Both SpoVM and CmpA are widespread and well conserved in spore-forming Bacillus species, indicating their essentiality in viable spore formation (97, 99).

SMALL MEMBRANE PROTEINS TARGETING THE LIPID BILAYER—TOXINS

Toxin-antitoxin systems have been well-studied and are prevalent in bacteria (reviewed in reference 100). One class of toxin molecules are the small proteins that target the lipid bilayer and compromise the bacterial membrane integrity, leading to cell death or dormancy (101–106). Here, we used the HokB toxin as an example to show one mechanism of action. It is important to note that functional characterization of these toxins is often performed using overexpression constructs and must be considered with caution because their functions, when expressed at endogenous levels in bacterial cells, are not well understood.

HokB (49 aa) has a putative transmembrane helix, a short N-terminal region, and an extended C terminus (107). It is located on the inner membrane with an N-in-C-out topology. HokB was shown to form pores of different sizes in vitro and in vivo (108). Overexpression of HokB led to the increased persister cells, although its absence did not significantly affect persistence, likely due to the redundancy of toxin-antitoxin modules (104). It was proposed that metabolically active cells with high membrane potential promote large mature HokB-pore formation, which allows the leakage of intracellular metabolites and induces persistence. With this mechanism, HokB-dependent persister cell formation specifically selects metabolically active cells, which can resume growth rapidly upon awakening (108).

Other remarkable toxic small membrane proteins include PepA1-A2/G1-G2 (30 to 44 aa) in S. aureus (101, 109), the ZorO (29 aa) (110), the SOS-inducible TisB (29 aa) in E. coli (111, 112), and the more recently discovered TimP (38 aa) in S. enterica (102). Similar to HokB, these toxins also cause membrane leakage and destroy membrane potential. Notably, effective antimicrobial agents are being developed from PepA1 (95) and ZorO (110), demonstrating the potential of small membrane proteins in clinical applications.

CONCLUSIONS AND OUTLOOK

It is evident that bacterial small membrane proteins have broad functional diversity in terms of their targets and their mechanisms of action. Extraordinarily, this was somewhat predicted 7 years ago based on viral transmembrane miniproteins, which regulate various biological activities with a versatile mechanism (113). Despite this diversity, several reoccurring themes among the well-studied small membrane proteins have emerged and might guide future studies of newly discovered small membrane proteins.

From a structural perspective, α-helical conformation dominantly appears in small membrane proteins as a structural element that anchors at the lipid bilayer, as a specificity determinant to target other membrane proteins, or as a sensor to detect membrane physiological properties. The overall shape of a small membrane protein is often more essential for its function, which is defined by intramolecular disulfide linkage or kink-inducing amino acid residues, such as proline and glycine. In contrast, many residues in a small protein have less functional importance, potentially providing the sequence space for engineering or evolving new molecules. Lastly, small protein-mediated target degradation appears to be a favored mechanism for regulating protein abundance. FtsH is an essential ATP-dependent zinc-metalloprotease on the membrane with a weak unfoldase activity (reviewed in references 114 and 115). It is conceivable that the binding of small protein might locally stabilize or denature its target protein and make it less or more susceptible to FtsH degradation. Alternatively, small proteins may also simply serve as a tag for the recognition of membrane-bound protease and guide the degradation by binding to its protein target.

From a technical standpoint, there has been slow but steady progress in the functional analysis of small membrane proteins using a variety of approaches. In vivo studies using genetic and biochemical approaches in combination with epitope tagging when feasible are powerful tools in elucidating function, association partners, and mechanisms of gene regulation, but they are fraught with challenges. While gene deletions are useful in some instances, a big challenge commonly encountered in functionally characterizing a small protein is that a deletion mutant has no detectable phenotype, at least under routine growth conditions employed in laboratories. Small protein modulators are likely important for fine-tuning functions of larger target proteins, resulting in deletion phenotypes that are rather subtle and difficult to detect. Phenotype identification is possible under very specialized growth conditions but may require tedious and prolonged investigation. Therefore, small protein overexpression studies are often utilized to gain functional insights. Results from such experiments must be carefully considered keeping in mind the potential for various pleiotropic effects and must be validated using alternate or complementary methods. Overexpression of small, especially hydrophobic, proteins may cause nonphysiological responses, such as membrane disruption, binding to nonspecific targets, or oligomerization because of high levels of the protein.

Application of proteomic and structural analysis tools, such as mass spectrometry, cryo-EM, and X-ray crystallography, help validate the expression and reveal the structures of these proteins in complex with their interacting partners. In vitro purification and characterization of small membrane proteins are often hindered due to the following factors: (i) low expression levels, (ii) protein size constraints, and (iii) a need for solubilization with amphipathic molecules. While more challenging, the short length of these proteins may be advantageous in the affordable chemical synthesis of a small membrane protein or its variants (116). Additionally, the use of synthetic membrane mimics like nanodiscs (117), styrene maleic acid copolymers (118), and dendrimersomes (119) can facilitate functional characterization of small membrane proteins in vitro. As the structure-function analyses of these proteins continue, the combined knowledge of all the characterized proteins may provide a method to predict and categorize newly discovered small proteins into distinct functional classes.

Accumulation of functional information on small membrane proteins also gives rise to more questions, especially for those that have common protein targets or in the case of one small protein regulating multiple targets. What are the mechanisms and dynamics of regulation of small proteins with respect to their gene transcription, translation, protein stability, active conformation, oligomerization, and target interaction? Does the promiscuity indicate an intermediate stage during evolution? How did the genes for small proteins initially evolve? Some small protein genes appear to have been acquired through lateral gene transfer based on their phylogenetic distribution and clustering with their target genes on the genome. More interestingly, recent studies showed de novo gene birth from nongenic sequences in the yeast genome and proposed a “TM-first” model (120, 121). The evolutionary history of a new, adaptive yeast gene was traced, supporting the hypothesis that a novel ORF encoding a small membrane protein emerges from a thymine-rich noncoding region probably via pervasive translation (121). Although bacterial genomes are more compact and have a wide range of GC content, it is still tempting to speculate that a similar mechanism of gene birth may exist, at least in a subset of bacteria. Additionally, in the laboratory, multiple transmembrane peptides genes that confer colistin resistance were successfully selected from artificial random sequence space (122). Furthermore, synthetic small membrane proteins have been engineered de novo to specifically target pathogens without damaging commensal bacteria (123). It is clear that small membrane proteins are highly useful, regulatory tools in nature and will likely prove to be equally useful for biotechnology as we learn more about this versatile class of proteins.