The archaeal flagellum is a unique motility apparatus distinct in composition and likely in assembly from the bacterial flagellum. Gene families comprised of multiple flagellin genes co-transcribed with a number of conserved, archaeal-specific accessory genes have been identified in several archaea. However, no homologues of any bacterial genes involved in flagella structure have yet been identified in any archaeon, including those archaea in which the complete genome sequence has been published. Archaeal flagellins possess a highly conserved hydrophobic N-terminal sequence that is similar to that of type IV pilins and clearly unlike that of bacterial flagellins. Also unlike bacterial flagellins but similar to type IV pilins, archaeal flagellins are initially synthesized with a short leader peptide that is cleaved by a membrane-located peptidase. With recent advances in genetic transfer systems in archaea, knockouts have been reported in several genes involved in flagellation in different archaea. In addition, techniques to isolate flagella with attached hook and anchoring structures have been developed. Analysis of these preparations is under way to identify minor structural components of archaeal flagella. This and the continued isolation and characterization of flagella mutants should lead to significant advances in our knowledge of the composition and assembly of archaeal flagella.
Type IV pili
The concept of a third major branch of the tree of life was first proposed by Carl Woese and colleagues in the 1970s , based initially on sequence comparisons of small subunit rRNA and subsequently supported by a variety of biochemical and genetic data . Woese’s analyses revealed the presence of two distinct prokaryotic groups eventually designated the Domains Bacteria and Archaea , with the latter group initially composed of a limited number of microorganisms that tended to inhabit extreme environments. These charter members of the Archaea included methanogens, extreme halophiles and certain sulfur-dependent thermoacidophiles and hyperthermophiles. However, in recent years the ubiquity and diversity of archaea has been demonstrated using modern molecular biology techniques and it now appears certain that the contribution and importance of archaea to the earth’s ecosystems have been grossly underestimated [4, 5].
Recently, as the genome sequences of several archaeal species were completed and annotated, it became clear that a significant percentage of the predicted open reading frames (ORFs) had no discernible homologues in either of the other two Domains of Bacteria and Eucarya. While Archaea have features in common with both Bacteria (lack of a nuclear membrane, polycistronic mRNAs) and Eucarya (many features of the transcription system such as similar promoters and TATA binding proteins), they also possess a number of defining archaeal-specific features . These include ether-linked lipids and unique cell walls. No doubt, many of the novel genes found in archaeal genome sequences are ones that are involved in these archaeal-specific traits. One may also conclude that a large number of unique genes must be involved in flagellation in archaea. Flagellation is a widespread characteristic of archaea and several complete genome sequences from flagellated archaea have been published. Yet, no homologues of the bacterial flagella system have been identified in any archaea  indicating that the archaeal system, while superficially resembling the bacterial one, is composed of unique components which may be assembled in a completely novel fashion. Since the structure and assembly of a flagellum are complex and require more than 40 genes in bacteria, it is likely that many archaeal genes of unknown function are ones involved in archaeal flagellation and motility.
The study of bacterial flagella has greatly advanced our knowledge of prokaryotic biology, especially in areas of genetic regulation and mechanisms of export. Archaea as a group are much less studied than bacteria. It is expected that continued study of archaeal flagella will lead to an elucidation of the composition and assembly of this unique motility structure and that using archaeal flagella as a model system will contribute to our understanding of genetic regulation and protein export in these unusual and understudied microorganisms.
In this contribution we review the current knowledge of archaeal flagella and motility, pointing out the unique features of this system and comparing it to both the bacterial flagellum and type IV pilus systems.
2 Distribution of flagellation in archaea
Flagellation occurs in all the main groupings of the archaea (Fig. 1), including halophiles, methanogens, sulfur-dependent thermophiles and hyperthermophiles . Even Thermoplasma species which lack a cell wall have been shown to be flagellated [9–11]. The fact that flagellation occurs in archaea living in such a wide range of environments (extremes of temperature, salinity, and pH) is a testament to the stability of the flagella of this Domain. Studies on flagellar genes and proteins, their interactions and functions currently focus mainly on the methanogens and halophiles.
Distribution of flagellation throughout the Domain Archaea. +, at least some members of the genus are flagellated; −, no members of the genus are flagellated. Reprinted with updated material from .
There currently exist a variety of techniques for the isolation of flagella, both in bacteria and in archaea. Standard procedures to isolate ‘intact’ flagella (filament, hook, basal body complexes) from bacteria involve lysis of protoplasts or spheroplasts with the non-ionic detergent Triton X-100 with eventual banding in a CsCl gradient . However, use of this detergent has been shown to dissociate some archaeal flagella filaments, as described later. Techniques that have been successfully used for the isolation of archaeal flagella include shearing, phase separation using Triton X-114 and recovery of filaments from spent culture media .
Initial investigations into the flagella of Methanococcus voltae involved the isolation of flagellar filaments by shearing them from the cell surface with a Waring blender . The preparation was further purified by differential centrifugation and finally by banding in a KBr gradient. SDS–PAGE analysis of this preparation revealed two major protein bands of apparent molecular masses of 31 and 33 kDa, which are most likely flagellins FlaB1 and FlaB2. Negative staining electron microscopy of the crude flagellar preparation illustrated filaments with a diameter of 13 nm and a wide variety of lengths. This method has been successfully employed for the isolation of filaments from many methanogens .
The phase separation technique is traditionally used for the separation of hydrophobic from hydrophilic proteins . M. voltae envelopes may contain few hydrophilic proteins other than the cell wall protein (S layer protein) and those present in the flagella, and so this procedure was utilized for isolation of ‘intact’ flagella. After phase separation, the flagella were further purified by differential centrifugation before finally either banding in a KBr gradient or passing through a Sepharose column . SDS–PAGE examination of the resulting sample demonstrated predominantly the 33- and 31-kDa flagellins, along with a limited number of other proteins of mostly small molecular mass (<20 kDa), which were significantly less abundant than the flagellins. Electron microscopy showed a reasonable proportion of flagella with hooks and knoblike anchoring structures at their ends.
Centrifugation of spent culture media can be used to isolate flagella filaments from some archaea, particularly when dealing with superflagella-producing mutants of Halobacterium halobium. Superflagella are aggregates of loose flagella that form thick spiral bundles, which are 10–20 times longer than the cell length. They contain more than 200 individual filaments when examined under electron microscopy . In the case of certain methanogens, notably Methanospirillum hungatei and Methanoculleus marisnigri, flagella preparations of high purity could be readily obtained from the culture supernatant [17, 18]. In the case of M. hungatei this could arise by shearing of the flagella from the cells while shaking during growth but in the case of M. marisnigri, cells grown statically had large numbers of free flagella in the media.
Flagella filaments have been characterized from many archaea, including species from the genera Halobacterium, Methanococcus, Methanoculleus, Methanospirillum, Methanothermus, Natrialba (formerly Natronobacterium), Pyrococcus, Sulfolobus, Thermococcus and Thermoplasma (Table 1). For many other archaeal species, flagella and/or motility have been observed but no work has been done to characterize the flagellin proteins or genes.
bAs determine by thymol-sulfuric acid or periodic acid-Schiff staining.
Archaeal flagella filaments show many differences from bacterial flagella, with one difference being the diameter of the filament. Typically, the archaeal filaments are thinner than their bacterial counterparts (10–14 nm in archaea [18–20] compared to about 20 nm in bacteria ). Filaments thinner than usual have been reported under certain conditions in flagella preparations of Natrialba (formerly Natronobacterium) magadii and Thermoplasma volcanium. These so-called protofilaments appear to be intermediate structures in the process of dissociation . Generally, archaeal flagella filaments are featureless when observed under the electron microscope after 2% uranyl acetate staining . An exception is Sulfolobus shibatae flagella , which demonstrate well-defined subunits within the filament. To date there are no reports of sheathed flagella in archaea although such flagella are found in certain bacteria such as Bdellovibrio bacteriovorus, Helicobacter pylori and spirochetes .
Although there are exceptions [25, 26], typical bacterial flagellar filaments generally possess a left-handed helix, with swimming motility resulting from counter-clockwise rotation. This rotation results in the flagella filaments forming a bundle in peritrichously flagellated bacteria, such as Escherichia coli. It is during the ‘tumble’ or clockwise rotation that the bundles fly apart. In comparison to this, Halobacterium salinarum possesses unusual flagella, in that the filament forms a right-handed helix. The halophile flagella, and presumably all archaeal flagella, are rotating structures as observed by both direct light microscopy  and observation of tethered cells . In H. salinarum, a clockwise rotation of the flagellar bundle pushes the cell, while a counter-clockwise rotation pulls the cell, so the cell appears to swim with the flagella in front, and the flagellar bundle is never seen to fly apart . It was found that the switching behavior of the flagellar bundle and rotation are synchronous in H. salinarum, which is not the case in the peritrichously flagellated E. coli. Whether this ‘push-pull’ motility feature is unique to H. salinarum or commonly found in archaea is not known.
There are very few reports that measure archaeal swimming speed. A rare exception is the case of H. salinarum where the speed was measured at 2–3 μm s−1, about 10% of the speed reached by E. coli under similar conditions of observation [16, 28]. Cells were shown to swim almost twice as fast when the flagella are rotating clockwise as opposed to counter-clockwise. Whether this slow speed is a general trait of archaeal motility or a specific characteristic of this species is unknown. It may reflect the habitat of the extreme halophile where swimming occurs in near-saturated solutions of NaCl or it may be that with little competition in its extreme environment there is little need to improve an inefficient system .
4.2 Hook and anchoring structure
To date, the best methods for isolating archaeal flagella with some attached basal structure has been with phase separation using Triton X-114 detergent  or by solubilization of whole cells  or flagellated cell envelopes with Triton X-100 . However, even with these techniques, convincing repeatable evidence of basal body rings has not been obtained. The basal structure has been reported as knoblike in several Methanococcus species  and Halobacterium while a Gram-positive-like basal body with two rings has been reported for Methanococcus thermolithotrophicus and M. hungatei. In the latter case, caution is urged as the observation was made on a few Triton X-100-resistant flagella when the filaments of M. hungatei have been shown to be Triton X-100-sensitive [18, 20]. Other researchers have published a M. hungatei basal structure that lacks a well-defined hook and rings [24, 30]. Whether the poorly defined basal structure often observed is a technical problem indicative of a delicate nature of the basal structure in archaea or a true structural difference between bacterial and archaeal flagella remains an open question. It is possible that a typical basal body is present in archaeal flagella but it is sensitive to the treatments used thus far to try to isolate ‘intact’ archaeal flagella. However, when the phase separation technique was used to isolate flagella from three different Gram-negative bacterial species, all four rings and rod of the basal bodies were clearly seen . If the archaea do possess a bacterial-like basal body, the proteins comprising such a structure must be distinct from their bacterial equivalents as no homologues to basal body proteins are seen in the analysis of complete archaeal genomes . It may be important to remember that archaeal walls are quite distinct from those of bacteria: archaeal walls always lack murein for instance. In addition, the best-studied archaea with regard to flagellation are Methanococcus and Halobacterium, which have a wall profile not seen in bacteria . This is an S layer that overlayers the cytoplasmic membrane with no other intervening wall layer. Perhaps the insertion of archaeal flagella into these structurally unique walls has resulted in the development of a different anchoring mechanism. Due to this relative simplicity, additional anchoring may be needed, and it has been speculated that some archaea may stabilize the flagella with a sub-cytoplasmic membrane layer, such as the polar cap of H. salinarum. Researchers were unable to determine if this cap was part of the cytoplasmic membrane or a sub-membrane structure. However, it is reminiscent of polar membrane-like structures often associated with flagellar insertion and near the site of septum formation in bacteria . Similar polar membrane-like structures have been observed in many flagellated archaea including M. voltae [19, 32].
Southam et al.  and Cruden et al.  have shown that flagella are inserted through the end plug of M. hungatei and not through the sheath. The end plug is a complex structure composed of several proteinaceous matrices possessing hexagonal symmetry and pores (15 nm) large enough to accommodate the diameter of the flagella (10 nm). These layers in the end plug may act as a bushing for the flagella.
The hook region of archaeal flagella has been considered in the past to be poorly defined , and was identified as a slight thickening of the filament located next to the knoblike end when staining of flagella was done with uranyl acetate. However, as first observed by Cruden et al. , staining with phosphotungstic acid routinely gives better resolution of archaeal flagella specimens than does uranyl acetate. We have also noted a clearly defined hook region on M. voltae flagella isolated by phase separation when these samples are stained with phosphotungstic acid at neutral pH (Fig. 2). In the case of M. voltae flagella, the hook length is almost twice as long as that of Salmonella flagella (S.-I. Aizawa, personal communication), which is 55 nm . Hooks of varying length (72–265 nm) located near the anchoring structures were seen in preparations of flagella isolated from M. thermolithotrophicus. This is a much larger variability in length than usually reported for bacterial hooks .
Negatively stained preparation of flagella from M. voltae showing hook regions. Samples were prepared by S.L. Bardy using the phase separation technique with Triton X-114  and stained with 2% phosphotungstic acid by S.-I. Aizawa. Note the curved hook structure present at the ends of many flagellar filaments. Bar=100 nm.
5 Protein composition
Throughout the archaea, several trends with regard to flagellation can be noted. These include the presence of more than one flagellin composing the flagellar filament and the fact that archaeal flagellins possess a highly conserved and hydrophobic N-terminal sequence and a short leader peptide. From amino acid sequences deduced from cloned flagellin genes, archaeal flagellins fall in the size range of 193–580 amino acids corresponding to 20.5–61.1 kDa. However, archaeal flagellins often migrate in SDS–PAGE as much larger proteins than predicted, typically 5000–10 000 larger in molecular mass (compare M. voltae flagellins of 31 000 and 33 000 molecular mass as determined by SDS–PAGE with the predicted molecular masses of 22 400–25 500 deduced from the genes ). In some cases, much larger flagellins have been reported by SDS–PAGE: e.g. Natrialba magadii with one flagellin at 105 000 . Whether there are flagellin genes in this organism that are correspondingly larger is not yet known but N. magadii flaB2 is only slightly longer than normal flagellin genes, encoding a protein of 259 amino acids (GenBank accession number AJ225175).
The amino acid compositions of flagellins from bacteria and archaea are generally similar with a preponderance of acidic and neutral amino acids and few basic amino acids . One unusual finding is the presence of cysteine in several archaeal flagellins including ones from M. voltae, M. hungatei, Pyrococcus kodakaraensis and Archaeoglobus fulgidus [30, 34, 36, 37]. Cysteine is an amino acid almost never found in bacterial flagellins  (an exception is Roseburia cecicola).
In M. voltae, comparison of the DNA sequence with N-terminal sequences from the mature flagellin proteins (FlaB1 and FlaB2) demonstrated for the first time the presence of a 12-amino acid signal sequence that is not seen in bacterial flagellins . A similar finding was recently published for Methanococcus vannielii and leader peptides appear to be found in all archaeal flagellins. Examination of the amino acid sequence within the N-terminus of archaeal flagellins illustrates a strong conservation and hydrophobicity in approximately the first 50 amino acids of the mature proteins. Due to this conservation, a consensus sequence for the mature (demonstrated or presumed) flagellin proteins of mesophiles, thermophiles and hyperthermophiles has been determined for the first 34 amino acids as A*GIGTLIVFIAMVLVAAVAA(S/G/A)VLINT(A/S)G(Y/F)LQQK (position 2 is variable, Table 2). All positions shown as part of the consensus are ones in which the indicated amino acid is found in at least 70% of the flagellins.
N-terminal amino acid alignment of all available archaeal flagellin sequences
* The amino acids in the consensus sequence are found in at least 70% of all archaeal flagellins. Invariant amino acids are shown in bold. The vertical arrow indicates the site at which cleavage of the leader peptide from the preflagellin occurs in methanogens.
The conservation of the N-terminal region throughout all archaeal flagellins implies that it has an important role in either the assembly and/or function of the flagella. In bacterial flagella, the N- and C-terminal ends of the flagellins demonstrate conservation of sequence [24, 41] and it is known that these regions of the molecule are important in export and polymerization .
The presence of multiple flagellins in archaeal flagella differs from the single flagellin found in most bacterial flagella, and raises many questions. One such question concerns the need for multiple flagellins, and suggests that there may be differing roles for each in the assembly of the organelle. Transcriptional data tend to support this theory, with different flagellin genes being transcribed at different levels [34, 37], and the differing levels of expression are confirmed through detection via SDS–PAGE. In addition, recent mutant work in both M. voltae and H. salinarum suggests that each flagellin plays a different role in the assembly of the filament and that the flagellins are not interchangeable (see Section 10).
Another question raised by the presence of multiple flagellins is that of spatial organization, which has not been directly studied within the archaea. Among the bacteria that do demonstrate more than one flagellin, some studies have been done to specifically determine the organization of these flagellins into the filament. H. pylori flagella are composed of two flagellins, with the minor 57-kDa protein located proximal to the hook and the major 56-kDa flagellin comprising the rest of the filament . The flagellins of Caulobacter crescentus and Rhizobium meliloti are also known to assemble in a spatially distinct manner. However, in the case of Campylobacter coli, the two flagellins appear to be intertwined along the length of the filament . Mutant studies recently published on H. salinarum suggest that there are spatially distinct regions of the flagella in this organism. It had been reported previously that all five flagellin genes (flgA1, flgA2, flgB1, flgB2, flgB3) in H. salinarum were transcribed  and that all five gene products were found in purified flagella filaments . The recent study suggests that the FlgB flagellins are located mainly in the cell proximal region of the flagellum while the FlgA flagellins comprise the rest of the filament .
Relatedness has been demonstrated among the flagellins of several Methanococcus species, including M. maripaludis, M. deltae, M. vannielii and M. voltae. A high degree of cross-reactivity was observed in immunoblotting experiments with these flagellins and polyclonal antiserum that was raised against the denatured flagellins of M. voltae. Specifically, the antisera raised against FlaB2 of M. voltae reacted strongly with multiple bands in samples from M. maripaludis, M. deltae, and M. vannielii, suggesting that these could be either different flagellins or different forms of the same flagellin, perhaps with differing amounts of posttranslational modification. Surprisingly, only one of the putative flagellins in the flagellar preparations of M. thermolithotrophicus cross-reacts with the M. voltae FlaB2 antiserum (N.A. Thomas and K.F. Jarrell, unpublished data), while the flagellins from M. jannaschii show only weak cross-reactivity against this antiserum . However, the bands that do react in immunoblots correspond to the molecular masses of the flagellins seen in purified filaments as determined by SDS–PAGE.
These results indicate the presence of conserved epitopes defined by the amino acid sequence, which is highly conserved among the flagellins from these methanococci, especially over the N-terminal 50 amino acids. The conserved nature of the 5′ end of the flagellin genes in Methanococcus species is also readily demonstrated through the use of oligonucleotide probes corresponding to the 5′ end of M. voltae flagellin genes. These probes have been shown to hybridize specifically to the flagellin genes of other Methanococcus species in Southern blotting experiments (N.A. Thomas and K.F. Jarrell, unpublished results; ) and they have been used as tools to clone the flagellin genes from M. thermolithotrophicus and M. maripaludis (N.A. Thomas and K.F. Jarrell, unpublished data). An atypical feature of the flagellins was observed in the case of M. thermolithotrophicus. Here, two of the flagellins are much larger (329 and 422 amino acids) than those found in the other methanogens (typically 215–240 amino acids in length). These larger flagellins have an extended internal region while maintaining the conserved N- and C-terminal regions found in methanococcal flagellins.
Polyclonal sera raised against the intact filaments of M. hungatei strain JF1 cross-reacted strongly with the flagellins of H. salinarum, as well as a single flagellin from M. marisnigri and two flagellins from Methanothermus fervidus. Since the flagellins from M. hungatei, H. salinarum and M. fervidus all appear to be glycosylated, this cross-reactivity may be due at least in part to shared glycosyl units.
5.1.1 Posttranslational modifications
Unlike all but a very few bacterial flagellins [50, 51] the flagellins of the archaea often appear to be glycosylated. However, in most cases there is only presumptive evidence based on positive staining reaction of flagellins with glycoprotein-specific stains, such as thymol-sulfuric acid and periodic acid-Schiff [9, 17]. It is only in H. salinarum that structural information and linkage data are available. Here the flagellins are sulfated glycoproteins. Analysis of sulfated glycopeptides of the halophile flagellins indicated that they were indistinguishable from the sulfated oligosaccharides found in the cell surface S layer protein (cell surface glycoprotein) . These saccharides are found attached to the flagellins in the same unusual asparaginylglucose linkage unit as in the S layer as well. The glycoconjugates are Glc 1→4 HexUA 1→4 HexUA 1→4 Glc-Asn and HexUA 1→4 HexUA 1→4 HexUA 1→4 Glc-Asn where HexUA is either glucuronic acid or iduronic acid .
The sequence of amino acids around the N-glycosidic linkages matches that common to all eukaryotic N-glycosidic linkages, namely Asn-X-Thr(Ser). This sequence is found three times in the flagellin genes of H. salinarum. There are also multiple copies (2–6) of this sequon in the flagellins of M. voltae even though M. voltae flagellins do not stain positively with glycoprotein-specific stains like periodic acid-Schiff and thymol-sulfuric acid .
On SDS–PAGE gels, the flagellins of H. salinarum display a ladder-like pattern of bands with three centers of intensity. This heterogeneity, also noted in the flagellins of M. hungatei, likely represents flagellins containing different numbers of sulfated glycoconjugates . When flagella from an overproducing mutant of H. salinarum are examined by SDS–PAGE, the entire set of flagellins are seen to run with a lower than normal molecular mass, which may be due to a decrease in carbohydrate content . Moreover, superflagella (flagella in the supernatant) are present in this mutant, which may indicate that glycosylation is required for proper incorporation into the Halobacterium cell envelope.
Bacitracin is an antibiotic that inhibits certain types of glycosylation by binding to the lipid carrier dolichol diphosphate and removing it from the dolichol diphosphate cycle . In H. salinarum, addition of bacitracin has been shown to prevent the transfer to the S layer protein of the repeating unit saccharide, one of three types of glycoconjugate found attached to this protein . Since bacitracin does not enter the cell, these results indicate that in this case, glycosylation occurs on the outer surface of the cytoplasmic membrane. A similar conclusion was reached when an artificial peptide, which could not enter the cell, was shown to be glycosylated when added externally . Since the halobacterial flagellins lack a repeating unit glycoconjugate, they are not affected when cells are exposed to bacitracin. However, when cells are treated with EDTA, the H. salinarum flagellins migrate as much smaller proteins in SDS–PAGE, equivalent to the molecular masses observed when the flagellins are deglycosylated. It is thought that the EDTA specifically inhibits an externally located oligosaccharyltransferase .
Unlike in H. salinarum, the addition of bacitracin but not EDTA affects the migration of the flagellins of M. deltae in SDS–PAGE . With increasing bacitracin levels, the flagellins appear to migrate as smaller proteins detected by immunoblotting, likely reflecting less attached carbohydrate. At bacitracin concentrations that lead to mostly hypoglycosylated flagellins, intact flagella filaments were not observed indicating that a minimal amount of glycosylation may be necessary for filament formation. Based on observations made in H. salinarum, the results obtained with M. deltae suggest that a bacitracin-sensitive dolichol diphosphate carrier is responsible for a large proportion of the carbohydrate attached to the flagellins. It may be possible that in this case as well, bacitracin does not enter the cell and that glycosylation occurs on the cellular surface, however, this has not yet been demonstrated.
The data obtained in H. salinarum and M. deltae suggest that some flagellin glycosylation may be necessary for proper assembly and/or anchoring of the flagella in these organisms. However, the suggestion that glycosylation is required does not explain why some archaea possess glycosylated flagellins while others, even in closely related species, apparently do not.
Not all archaeal flagella demonstrate glycosylation through traditional staining methods, indicating either that glycosylation is not universally present throughout the archaea or that the method of carbohydrate attachment may resist detection by conventional methods. In the case of M. vannielii, the predicted molecular masses of the FlaB1 and FlaB2 flagellins were 22 524 and 23 088 Da, respectively, but examination of flagella filaments by SDS–PAGE demonstrated that FlaB1 has an apparent molecular mass of 30.8 kDa, almost 2 kDa larger than FlaB2 . This may reflect different amounts of posttranslational modification for each flagellin. However, these flagellins do not stain positively with glycoprotein stains. In addition, neither flagellation nor the flagellin apparent molecular mass is affected by the presence of bacitracin.
In addition to glycosylation, other posttranslational modifications to archaeal flagellins may be found. It has been suggested from staining results, for example, that the flagellins of N. magadii are phosphorylated .
At this point, the majority of archaeal flagella characterized possess multiple flagellins, with the lone possible exception being T. volcanium. SDS–PAGE analysis of the flagella from this thermoacidophile shows only one flagellin running at 41 kDa, but there may be more than one flagellin present with differing amounts of posttranslational modifications resulting in similar molecular masses, especially since this flagellin stains as a glycoprotein. The impending completion of the genome sequence from T. volcanium should clarify this apparent anomaly.
5.2 Minor structural proteins
The bacterial flagellum requires a wide variety of proteins, both in its biosynthesis and in the final structure. While there is generally only one flagellin comprising the filament in bacteria, a large genetic system is involved in flagellation and motility. In Salmonella, there are currently 44 known flagellar genes, encoding structural components (flagellin, hook, basal body, cap and others) and fulfilling regulatory roles . Yet within the archaea, only genes encoding flagellins and a limited number of co-transcribed genes have currently been identified. One would expect that a significant number of genes involved in the regulation, structure and assembly of archaeal flagella remain to be found.
One way to identify potential novel genes involved in archaeal flagellation is to sequence the N-termini of minor structural proteins detected in SDS–PAGE analysis of flagella isolated by techniques that result in attached hooks and anchoring structures. SDS–PAGE of such flagella preparations from M. voltae reveals the presence of flagellins as the dominant components, but also the presence of minor proteins that are likely minor structural components of the flagella. These minor bands do not appear to be contaminating proteins from either the cytoplasm or membrane fractions (S.L. Bardy and K.F. Jarrell, unpublished data). These minor structural proteins have a molecular mass smaller than that of the two flagellins (33 and 31 kDa), with the most prominent bands appearing at ∼20 kDa or smaller. The predominance of the lower molecular mass bands is surprising. Examining the ‘intact’ flagellar preparations under electron microscopy demonstrates a reasonable proportion of intact hooks suggesting that a protein band corresponding to the hook protein would be observed in SDS–PAGE gels. If the hook protein is represented on these gels, it is presumably one of the low molecular mass bands and thus much smaller than the hook protein in enterics (42 kDa in E. coli and Salmonella typhimurium) or it is masked by the presence of the massive amounts of flagellins. Evidence for the latter has recently been obtained. M. voltae cells were sheared to remove the majority of the filament mass. The sheared cells were lysed and then the membranes were subjected to phase separation to isolate the flagella stubs with attached hooks. These flagella should have an increase in the level of the hook protein relative to the flagellins since the majority of the flagellin had been previously removed by shearing. SDS–PAGE examination of the sheared flagella and the flagella stubs isolated by phase separation revealed the presence of an extra band in the flagella stubs (S.L. Bardy and K.F. Jarrell, unpublished data). This extra band migrated only slightly slower than the flagellins and could easily be masked in examination of phase-separated flagella of regular length in which the amount of flagellin would be greatly increased relative to the hook protein. N-terminal sequencing of the extra band revealed that it was a minor flagellin, FlaB3, raising the intriguing possibility that a protein very similar to the major filament proteins forms the hook region in archaea. Interestingly, SDS–PAGE analysis of H. salinarum flagella with attached polar caps (and thus presumably containing hooks) revealed the flagellins as the dominant protein species but another prominent band is seen migrating at about 50 kDa . Perhaps this might represent the hook protein of H. salinarum. It is not known if hook–filament junction proteins (equivalent to hook-associated proteins HAP1 and HAP3 in bacteria) are required in archaeal flagella but if so they are presumably among the minor protein bands observed in SDS–PAGE of the phase separation isolated flagella.
It has recently been reported that specific cytoplasmic chaperones may be involved in flagella assembly in bacteria . In Salmonella, FliS has been postulated to be a flagellin-specific chaperone  while recently, FliJ was found to be necessary for the export of the hook protein and the hook capping protein. The suggestion was made that FliJ functions as a cytoplasmic chaperone for the hook type proteins . In both methanogens and haloalkaliphiles, ATP-dependent flagellin binding proteins have been isolated which may be specific cytoplasmic chaperones. Polosina et al.  recently identified nucleotide diphosphate (NDP) kinase as a flagellin binding protein of N. magadii. NDP kinases have been identified as regulatory proteins for the eukaryotic Hsc70 molecular chaperone . In N. magadii, this protein may be responsible for the phosphorylation believed to be present on the flagellins. Chaperones may be needed in archaea to prevent the newly synthesized flagellins from non-specifically aggregating in the cytoplasm through interactions of their hydrophobic N-termini, before they are secreted out of the cell. This would be similar to the Syc chaperones of Yersinia which are proposed to act as anti-aggregation factors in the secretion of Yop proteins . Additionally, flagellin-specific chaperones may be involved in the delivery of the flagellins to the export apparatus analogous to the role of SecB in the general secretory pathway .
Due to the extremes of environment from which the archaea are often isolated, it is necessary for the flagella to be extremely stable. Work done with the methanogens and halophiles indicates that the flagella filaments, in their native form, are resistant to protease treatment and are more stable at high temperatures than are their typical bacterial counterparts. Also, proteolytic cleavage has shown that the more hydrophilic C-terminus of the flagellins does not appear to partake in interactions and filament formation as cleavage of this section does not lead to destruction of the polymer structure .
Stability of filaments has been studied in different extreme halophiles. It has been shown that under salt-free conditions, the flagella of H. salinarum and N. magadii dissociate into flagellins. However, H. salinarum superflagella reform when dialyzed against 4 M NaCl medium , but haloalkaliphilic flagella do not reassemble after transfer from low to high salt. As the flagella of haloalkaliphilic archaea N. magadii and Natronomonas pharaonis were exposed to decreasing salt concentrations, thinner filaments, termed protofilaments, believed to be due to intermediate dissociation of the polymer, were observed by electron microscopy. With N. magadii, it was found that the filaments were stable between 10 and 25% NaCl, but below 10% the filaments began to dissociate. The initial stages resulted in filaments with alternating parts with variable thickness and structure. In the absence of NaCl, complete dissociation of the filaments occurred and only ball-like aggregates were visible. Differential centrifugation was attempted in order to isolate various components of the dissociated filaments, but was unsuccessful . Similar degradation of the flagellar filaments was seen in flagella preparations of T. volcanium stored in mineral medium at low pH with the resulting protofilaments showing a diameter of 3–5 nm .
Detailed biochemical studies on the stability of H. salinarum flagella with regard to heat absorption have demonstrated that the structural integrity of the intersubunit bonds is so strong that the flagellins will denature before the flagella dissociates. These studies also demonstrated that the melting temperature of H. salinarum flagella showed only one peak, indicating all five subunits have identical thermodynamic properties .
Microcalorimetry and circular dichroism showed that cooperative melting of the H. salinarum flagella corresponds to changes in tertiary and secondary structure. This indicates that only a small part of the flagellin molecule is responsible for polymerization. As the conserved N-terminus forms α-helices, one of which is completely hydrophobic, it is speculated that this region may be responsible for polymerization and that the cooperative melting domain may participate in protecting the hydrophobic α-helices from contact with the environment .
Other heat stability studies have been done with the purified flagellar filaments of several methanogens. In these studies, the filaments were incubated at temperatures ranging from 40 to 90°C for 30 min and then examined by electron microscopy for the presence of intact filament structures as a crude measure of their stability (N.A. Thomas and K.F. Jarrell, unpublished data). Flagella filaments from the mesophile M. maripaludis showed stability to 70°C, while the filaments of the thermophile M. thermolithotrophicus and hyperthermophile M. jannaschii both showed stability to 90°C, and potentially beyond.
Cruden et al.  reported that the integrity of flagella filaments from M. thermolithotrophicus was lost in the absence of high NaCl concentrations as well as when incubated at pH 11 for 1 h. However, Kostyukova et al.  reported that filaments of M. thermolithotrophicus were still observed by electron microscopy after either of these treatments.
An extensive study of the stability of M. hungatei flagella indicated resistance of the structure to several proteases . In addition, filament structure remained intact over a pH range of 4–10 and up to a temperature of 80°C. However, treatment of the filaments with a variety of detergents (Triton X-100, Tweens, Brij 58) resulted in dissociation of the filament as determined by electron microscopy. The sensitivity to detergents of the filaments, composed of glycosylated flagellins , is consistent with a previous report that noted a correlation between glycosylated flagellins in a filament and the sensitivity of the filament to dissociation by Triton X-100 . In that study treatment with Triton X-100 resulted in the dissociation of the flagella of M. deltae and M. fervidus at concentrations as low as 0.5%. The flagella of M. voltae, M. vannielii and M. marisnigri are composed of non-glycosylated flagellins (as determined by traditional staining techniques) and are resistant to dissociation by Triton X-100. This correlation does not extend to the flagella of non-methanogenic archaea: T. volcanium, Pyrococcus furiosus and H. salinarum all possess flagella composed of glycosylated flagellins but are resistant to dissociation by Triton X-100.
7 Regulation of flagellation
M. hungatei and M. jannaschii are the only archaea currently known to show differential regulation of flagellation with regard to growth conditions: specifically temperature and calcium concentrations in the case of M. hungatei and pH2 in the case of M. jannaschii. There are many cases of environmental factors, such as the presence of glucose or temperature, affecting the production of flagella in bacteria [70, 71]. Usually in bacteria, flagellation is inhibited at higher growth temperature but in M. hungatei flagellation is inhibited in the lower temperature range of growth . Two strains of M. hungatei were examined, and it was found that strain GP1 produced flagella at 30 and 37°C, but not at 25 or 40°C. This differs from strain JF1, which produces flagella at 37 and 40°C, but not at any lower temperature examined. In all cases, the cells grew as short filaments. In general, it appears that M. hungatei produces significant amounts of flagella only when experiencing near-optimal growth temperatures. Also unusual is the level at which regulation appears to occur. In most bacterial systems, inhibition occurs at the transcriptional level with little, if any, flagellin detected by immunoblotting and flagellin mRNA is generally not detected. However, in M. hungatei, Western blotting of non-flagellated cells demonstrated levels of flagellin comparable to that found in flagellated cells, indicating that inhibition occurs at the level of assembly .
The effect of small changes in the calcium concentration of the growth media was on cell morphology as well as flagellation . In low calcium media, M. hungatei grew as very long (up to 900 μm long) non-flagellated filaments. Calcium is known to be important to the morphology of M. hungatei, in that it is the major metal ion of the sheath . It is thought that the changes in mineral concentration may directly affect the cellular morphology, and only indirectly affect flagellation.
The effects of both growth temperatures and calcium concentrations are reversible, and allow for methods to control flagellation of the cells, in effect to synchronize the turn-on of flagella assembly. The potential laboratory application of this phenomenon includes the study of flagella polarity of growth in archaea.
Recently, M. jannaschii was grown at high and ultralow hydrogen partial pressures and the cells subjected to comparative proteomic analysis . Under conditions where hydrogen became limiting (at either low hydrogen partial pressure or high cell densities under high partial pressure), flagella synthesis occurred. Four flagella-related proteins (FlaB2, FlaB3, FlaD and FlaE) were shown to increase under hydrogen limitation. Electron microscopy confirmed the absence of flagella in hydrogen excess conditions and their presence under hydrogen limitation.
8 Archaeal flagella gene families
In the archaeal species where flagellin genes have been cloned and sequenced, multiple flagellin genes arranged in tandem are always found (Fig. 3). This includes sequences from various species with remarkably different physiologies (the methanogens, the extreme halophiles and the sulfur-dependent thermophiles). The presence of multiple flagellin genes in both the Euryarchaeota and Crenarchaeota suggests this is a common feature of the flagellar system in the Domain Archaea. Furthermore, in all archaea where sequence data exist, in close proximity to the flagellin genes, there are always a number of putative flagella accessory genes. A trend that is found for the flagella accessory genes is that at least some of them tend to be co-transcribed with the flagellin genes.
Flagella gene families of selected archaeal species. Similar colors indicate homologues shared among families. White unlabeled boxes are ORFs that have no sequence similarity to genes in universal databases. The genes are transcribed in the direction of the respective arrows. The 2-kb region separating the flagellin genes and the accessory genes in Aeropyrum pernix contains a number of overlapping ORFs that are not shown in this diagram. The flaF gene designated in H. salinarum is not a homologue of the flaF gene in M. voltae. The B flagellin genes of H. salinarum are adjacent to the accessory genes, whereas the A flagellins are located elsewhere on the chromosome.
8.1 Analysis of archaeal flagella gene families
The entire flagellin and flagella accessory gene region extends 7.5–14 kb, depending on the archaeon. In the genus Methanococcus, all the flagellated species that have had their flagellin gene families cloned and sequenced have their respective flagellin genes followed by at least eight genes. The genes are termed flaC, flaD, flaE, flaF, flaG, flaH, flaI, and flaJ ([74, 75]; N.A. Thomas and K.F. Jarrell, unpublished data). In addition, M. jannaschii appears to have at least another gene immediately downstream of flaJ that may be part of its flagella gene family. A. fulgidus has a homologue to this gene but it is not located near its flagella gene family. In H. salinarum, a gene designated flaK is located immediately downstream of its flaJ homologue, although it is not known with certainty if the former gene is part of the flagella gene family.
In every archaeon, short intergenic regions separate all of the flagella accessory genes, or in some cases (flaE, flaF) they overlap. flaF is absent from the flagella gene families outside of the methanogens, suggesting that it may have a specialized role. Furthermore, the gene designated flaF in H. salinarum is also unique (N. Patenge, A. Berendes, H. Engelhardt, S. Schuster and D. Oesterhelt, personal communication).
In the cases of A. fulgidus and A. pernix, only the flaH, flaI and flaJ homologues have been found in close proximity to the flagellin genes (GenBank accession numbers B69381, A69381, H69380 and B72577, H72576, G72576 respectively). Instead of having the other flagella accessory genes (flaC–flaG), a number of ORFs (with no homology to any genes in universal databases) separate the flagellin genes from the flaHIJ group of genes. Thus, it appears that flagellin gene families can differ with respect to their composition and number of genes, suggesting that some gene products may not be required in all archaea. Whether this is related to the different extreme environments inhabited by various archaea or their different cell wall profiles is unknown. Interestingly, all the flagellated archaeal species have flaH, flaI and flaJ, which suggest that these genes are essential for either flagellar assembly or motility.
None of the protein sequences encoded by these flagella accessory genes have features that can be confidently identified with a certain biological function. One possible exception is flaI, which has a Walker box and presumed ATPase activity. Interestingly, FlaI is a homologue of PilT and other proteins that are commonly found in type IV pilus systems. The finding of flaI always in close proximity to the flagellin genes suggests that flagellar biosynthesis may occur via a type II-like secretion system, where the flagellins for the filament are secreted across a membrane and then incorporated into the final structure.
FlaF and FlaG are proteins that appear to lack a leader peptide but they do possess a very hydrophobic N-terminus. In addition, FlaG has sequence similarity to archaeal flagellins at the N-terminus. Both are expected to be produced in small amounts judging from transcription data. These proteins may be equivalent to PilE and PilV in the type IV pilus system, i.e. proteins with hydrophobic N-termini with sequence similarity to the pilins and believed to be structural proteins at the cell proximal end of the pilus . In the case of PilE and PilV, however, the proteins have a leader peptide that is cleaved by the prepilin peptidase .
A summary of some of the properties of the flagella accessory proteins of M. voltae is provided in Table 3.
Summary of selected properties of the flagella accessory genes and proteins of M. voltae
Predicted molecular mass
Apparent molecular massa
nd: Not determined.
aAs determined by immunoblotting of SDS–PAGE-separated M. voltae proteins.
bAs determined by Tmpred.
cAs determined by immunoblotting of fractionated M. voltae cells with polyclonal antibodies against the respective flagella accessory protein.
8.2 Transcriptional studies of flagella gene families
8.2.1 M. voltae
In M. voltae, the flagellin genes are present in two transcriptional units clustered close together (Fig. 4) . One unit contains flaA, a single flagellin gene that is weakly transcribed. The other unit encodes three flagellin genes and the flagella accessory genes. Northern blotting has demonstrated that this latter unit encodes the major flagellins that are in the filament and at least some of the downstream flagella accessory genes. Several transcripts, all initiating from a promoter upstream of flaB1, have been identified. The most abundant transcript encodes flaB1 and flaB2 as an approximately 1.3-kb polycistronic message. Transcription then continues through a hairpin structure after flaB2 to further include another flagellin gene, flaB3, and at least some of the downstream genes. This transcriptional readthrough results in additional minor messages that are approximately 4.2 and 5.4 kb in length. The 5.4-kb transcript would end after flaG. These findings are in agreement with protein expression analysis, where the FlaB1 and FlaB2 flagellins are found in very high amounts while the flagella accessory proteins are less abundant. No messages have been detected for flaH, flaI or flaJ using Northern blotting and RT-PCR methods, suggesting that either these genes are weakly transcribed or their message is unstable or transcription is growth phase-dependent. There is the possibility that a separate transcript may be generated that encodes flaH, flaI and flaJ. The intergenic region between flaG and flaH is larger than the other intergenic regions within the gene family and methanogen consensus promoter sequences can be found just upstream of flaH. Since there are short intergenic regions between flaH/flaI, and flaI/flaJ, it is believed that these three genes (flaHIJ) are transcribed as a single unit.
Schematic diagram summarizing data obtained from transcriptional studies of archaeal flagella gene families. The genes are transcribed in the direction of the respective arrows. The thickness of bars above the genes is indicative of the relative abundance of the transcripts but is not meant to be quantitative. In the case of H. salinarum, the flagellin gene transcripts are likely more abundant than those for the accessory genes, however, transcriptional data for the flagellin genes and the accessory genes were obtained from two different studies, so the relative abundance of the transcripts is not known with certainty. See text for further details.
8.2.2 M. maripaludis
As with M. voltae, several different length transcripts likely initiating upstream of flaB1 are found in M. maripaludis (Fig. 4, N.A. Thomas and K.F. Jarrell, unpublished data). No evidence of a flaA gene has been found. An abundant 1.3-kb message encodes the first two flagellin genes of the gene family. Presumably, transcriptional readthrough occurs so as to include a third flagellin gene on a stable 2-kb transcript. Additional transcriptional readthrough then includes some of the downstream genes on a less abundant polycistronic transcript that is approximately 4 kb. A message of this size would extend from flaB1 to flaE. No obvious promoter sequences can be found immediately upstream of any of the genes in the family (flaB2 to flaJ), which suggests that all the genes are transcribed from the single strong promoter found upstream of flaB1.
8.2.3 M. thermolithotrophicus
One transcriptional unit has been identified for the flagellin gene family in M. thermolithotrophicus (Fig. 4, N.A. Thomas and K.F. Jarrell, unpublished data). Using a probe specific for the first flagellin gene in the family, flaB1, a single transcript of 3 kb has been identified that encodes the first three flagellin genes of the operon, flaB1, flaB2 and flaB3. However, using a probe specific for flaB3, two messages are detected, a 3-kb transcript and a short 0.6-kb transcript. Messenger RNA processing may generate the stable 0.6-kb message from the larger 3-kb message since the intergenic region between flaB2 and flaB3 is only 46 nucleotides. This short message does not encode the fourth flagellin gene flaB4 since when a probe specific for flaB4 is used in Northern blotting experiments, the 0.6-kb nucleotide transcript is not detected. Presumably, the fourth flagellin gene and the immediate downstream genes are transcribed, but at low levels that are not readily detectable. A nucleotide sequence with the potential of forming a strong hairpin structure is located in the intergenic region between flaB3 and flaB4. This stretch of DNA most likely limits the transcription of downstream genes by decreasing readthrough. This might be a rather simple way for the cell to regulate the number of accessory proteins that are required for flagellar biosynthesis and function.
8.2.4 M. jannaschii
In the hyperthermophilic methanogen M. jannaschii, only a single transcript of approximately 1.3 kb that corresponds to the flagellin genes has been detected (Fig. 4, N.A. Thomas and K.F. Jarrell, unpublished data). The message most likely corresponds to flaB1 and flaB2 since the probe was specific for flaB1 and a consensus methanogen promoter can be found upstream of the flaB1 gene. Unlike the mesophilic methanogens, longer messages that extend into flaB3 or any of the downstream genes could not be detected. The absence of a message encoding the third flagellin gene and the downstream genes could possibly be due to the message rapidly degrading or being unstable. Analysis of the nucleotide sequence of the flagellar gene family of M. jannaschii reveals that short intergenic regions separate all of the genes. Thus it is most likely that transcription is driven from the promoter upstream of flaB1 and continues to include the other flagellin genes and the downstream flagella accessory genes.
8.2.5 P. kodakaraensis
Four different transcripts have been detected that encode different combinations of the five flagellin genes in P. kodakaraensis (Fig. 4) . All of the messages are believed to be transcribed from a promoter found upstream of the first flagellin gene. The shortest message is 0.98 kb and encodes flaB1. A second 3.7-kb message encompasses the first three flagellin genes of the family. A third transcript that is 5.4 kb long encodes all five of the flagellin genes that are in the gene family. Interestingly, the longest transcript is 9.2 kb and encodes all the flagellin genes and presumably some unidentified downstream ORFs. The Northern blotting data reveal that the longer messages (i.e. 3.7-, 5.4- and 9.2-kb transcripts) are relatively more abundant than the short 0.98-kb message which is believed to correspond to flaB1 alone.
8.2.6 H. salinarum
In H. salinarum, two transcriptional units encoding a total of five flagellin genes have been identified (Fig. 4) . One of the units encodes two flagellin genes, flgA1 and flgA2, and is 1300 nucleotides in length, while the other message encodes the remaining three flagellin genes, flgB1, flgB2 and flgB3, and is 1900 nucleotides long. Messenger RNA encoding only the flagellins is different from the other archaea studied, where the accessory genes are co-transcribed. However, recently the explanation has been presented. In addition to the flagellin gene transcripts, there is a fla gene cluster that encodes at least some of the flagella accessory genes. This gene cluster is located immediately upstream of the flgB1–3 flagellin genes and is oriented in the opposite direction (N. Patenge, A. Berendes, H. Engelhardt, S. Schuster and D. Oesterhelt, personal communication). Interestingly, in H. salinarum, there is no flaC homologue, which is found in all the methanogen examples. Instead, in that area, a gene called htpIX encoding a putative halobacterial transducer protein is found. Northern blotting experiments identified a single message for both the htpIX and the flaD genes indicating that they are co-transcribed. Another polycistronic message encompassing flaE to flaK has been detected by RT-PCR analysis. While H. salinarum does have flaDEGHIJ homologues of the corresponding M. voltae genes, its flaF is not homologous to M. voltae flaF.
9 Overexpression of flagella accessory proteins of M. voltae
Secondary structure analysis and computational analysis of the proteins in the M. voltae flagella gene family predict that FlaCDEH do not contain any transmembrane helices and would be located in the cytoplasm. However, in the case of FlaJ, seven to nine transmembrane helices are likely present and it is predicted to be a membrane protein. FlaF and FlaG are each predicted to have a single N-terminal transmembrane helix that may associate these proteins with the membrane. Lastly, FlaI has one transmembrane helix in the central portion of the polypeptide. Our laboratory has detected most of the putative flagella accessory proteins in M. voltae, indicating that they are expressed in vivo. In these studies, the corresponding genes were cloned into a vector that allowed for the overexpression of His-tagged proteins in E. coli. The overexpressed proteins were subsequently purified using affinity chromatography and used to generate polyclonal antibodies for use in immunoblots of fractionated M. voltae cells. Immunoblots of fractionated M. voltae cells using anti-FlaC antibody demonstrated that FlaC is associated with the membrane. No FlaC protein was detected in the soluble fraction of the cells. In a similar fashion, it has been demonstrated that FlaD, FlaE, FlaH and FlaI are all associated with the membrane of M. voltae. The findings are the first data that characterize these putative flagella accessory proteins in M. voltae. Furthermore, the fact that all these proteins are found in the particulate fraction of cells suggests that they may have a structural role. It is unknown whether these proteins are integral or peripheral membrane proteins. Some of the proteins may be associated with the flagellum structure since they potentially may reside at the base of the flagellum (in or associated with the membrane) as part of a basal complex. However, when M. voltae flagella isolated by phase separation were examined in immunoblots using antisera raised against these accessory proteins, none of FlaC, FlaD, FlaE and FlaH could be detected. This indicated that these proteins do not appear to be the hook or anchoring components of the M. voltae flagellum. They may, however, still be located in the vicinity of the flagella in the cell membrane or may be part of the anchoring mechanism that is detergent-sensitive. As such they may be identified in the flagella with an associated polar cap in H. salinarum which seems to encompass a larger structure than the knob seen at the end of phase separation-purified flagella of methanococci . Indeed PilT is not found as a structural component of type IV pili but is presumed to be peripheral to the pilus. The archaeal homologue FlaI may be expected to be the same.
Immunoelectron microscopy might provide some answers as to where these proteins specifically reside and more importantly if they are likely to interact as part of the flagellar structure.
10 Mutant studies
Functional characterization of archaeal flagellin genes and the flagella-associated genes has been a difficult task due to a lack of efficient archaeal expression systems and the fact that genetic techniques are not well established in many archaeal species [78, 79]. Some progress has been made in the methanogens and halophiles where specific flagellin gene mutants have been isolated in M. voltae and H. salinarum respectively.
10.1 M. voltae
10.1.1 Flagellin gene mutants
In M. voltae, transformation of protoplasts with a vector harboring an internal flagellin gene fragment resulted in the isolation of mutant strains that had insertions in flaA and flaB2. Insertions into flaA resulted in cells having flagella that appeared the same as wild-type. However, these flaA mutants were less motile than the wild-type in semi-swarm plate experiments. Mutants in flaB2 were non-flagellated, non-motile and did not produce flagella. Due to the insertion of the homologous fragment and the vector, a polar effect was created such that none of the genes downstream of flaB2 are expressed, thus essentially knocking out the gene family. These findings were the first data that functionally determined that this gene family is required for the flagellated phenotype. Furthermore, these experiments demonstrated that the presence of multiple flagellins is required in M. voltae (and most likely other archaea) for an optimal functioning of the flagellum, i.e. that the multiple flagellins are not interchangeable and that the less transcribed flagellin genes such as flaA and flaB3 may have specialized roles that require their presence in much smaller amounts than the major filament structural proteins.
10.1.2 flaH mutant
FlaH homologues have been found in close proximity to the flagellin genes in all flagellated archaeal species where sequence data exist. FlaH has no significant homology to any proteins in universal databases; however, from localization studies in M. voltae, it is known that FlaH is associated with the membrane fraction of cells.
Our laboratory has isolated an insertional mutant in flaH (N.A. Thomas, C. Pawson and K.F. Jarrell, unpublished data). The isolated flaH mutant was non-motile and electron microscopy revealed that the cells were not flagellated. Northern blotting experiments revealed that the mutant transcribed all the B family flagellin genes (flaB1, flaB2 and flaB3), and at least the flagella-associated genes flaC to flaG, essentially at the same levels as wild-type. In the flaH mutant, FlaH expression could not be detected in immunoblotting experiments using anti-FlaH polyclonal antibodies. However, immunoblotting experiments using antibodies against M. voltae flagellin and some of the other putative flagella-associated proteins demonstrated the flaH mutant did contain wild-type levels of the flagellins, FlaC, FlaD and FlaE. Thus the mutant makes all the flagellins and at least some of the putative flagella accessory proteins, but is unable to assemble the components into a flagellum. It is not known whether the mutant is deficient in secreting essential components for a flagellum or whether FlaH is involved exclusively in assembly. It should be noted that the nature of the insertional inactivation most likely results in a polar effect, disrupting transcription of the downstream genes flaI and flaJ. The flaI gene is believed to encode an ATPase that could provide the energy for export, thus this may explain in part why the mutant cannot assemble a flagellum. Interestingly, the flaH mutant has peptidase activity and therefore preflagellin presumably reaches the cytoplasmic membrane and is processed, but not assembled into the flagellum.
10.2 H. salinarum
10.2.1 Flagellin gene mutants
H. salinarum is an extreme halophile that has polarly located flagella. There are two groups of genes (designated A and B) that together encode five structural flagellins that have been shown to comprise the flagellar filament . Mutants in the flagellin genes of H. salinarum have been generated using insertional inactivation . In this study, each flagellin gene cluster was disrupted to generate mutants in each locus. The flgA mutants were flagellated but less motile than wild-type cells. The flagella of these mutants were curved and spiral-shaped like wild-type cells, but they were significantly shorter and were located at both the poles and sides of cells. In addition, a mutant with an insertion exclusively in one of the flagellin genes, flgA2, only produced straight flagella, which were mainly located at the poles of cells. Mutants that disrupted the B flagellin genes were less motile, had spiral-shaped flagella only at the poles of cells, but also had outgrowths from the cell surface. Furthermore, some of these outgrowths were observed to have a flagellum at their ends.
Northern blotting experiments of these mutants with a flgA gene probe suggested that expression of the B flagellins is dependent on successful expression of the A flagellins. The flgA mutants do not express the 1300-nucleotide flgA transcript seen in the wild-type and the flgB transcript is reduced dramatically. The exact nature of this dependence is not known.
Taken together, the data derived from these mutant experiments suggest that there are specialized roles for the A and B flagellins in H. salinarum. Tarasov et al.  suggest that the A flagellins are the main components of the filament and are incorporated in the initial stages of assembly. Presumably, these flagellins are also sufficient to form a longitudinal and spiral filament. The B flagellins are less abundant, most likely located at the base of the filament in wild-type cells. The phenotype of the flgB− strains (spiral-shaped filaments with strange outgrowths from the cell surface) implicates the B flagellins as important factors for proper anchoring of the flagellum. Additional support for this interpretation is found in the mutants that do not express the A flagellins. These mutants had two major differences from the wild-type: they had flagella all over the cell surface (at the poles and sides) and the filaments were shorter than wild-type. Presumably the improper incorporation of the B flagellins at the initial stages of assembly and then as the major components of the filament (by default in the flgA mutants) resulted in truncated filaments that were not located at the poles.
10.2.2 Mutants in the fla gene cluster
The fla gene cluster of H. salinarum has recently been subjected to transcriptional and mutational analysis (N. Patenge, A. Berendes, H. Engelhardt, S. Schuster and D. Oesterhelt, personal communication). In-frame deletions in each of flaI, flaH and flaE were constructed by gene replacement. The three deletion strains lack motility and belong to the fla− mutant class. These experiments demonstrate that some of the genes in the fla gene cluster are involved in the biosynthesis, transport, and/or assembly of the flagellar apparatus in H. salinarum.
These experiments and other mutants in the flagellin and flagella accessory genes of the archaea are of paramount importance in elucidating the nature of the flagellum in archaeal species. The existing data suggest specialized roles for some archaeal flagellins, while others are the main components of the filament and provide the propeller structure for motility. From the examples presented in this report and from other studies, the disparity in the transcription of the multiple flagellin genes suggests that some are needed in abundance, and others are minor (but functionally important) components of the flagellum. The mutants that have been isolated in M. voltae flaH and in H. salinarum flaH, flaI and flaE are the first data that implicate genes other than those that encode the flagellins in flagellar biosynthesis or assembly. These findings are particularly exciting since the gene family in question is exclusively found in flagellated archaeal species. The gene family represents a novel locus that is involved in motility and although the function of each of the genes within the family is not known, the availability of structural and biochemical data will certainly help in interpreting the phenotypes of existing and soon to be isolated mutants.
11 Preflagellin peptidase activity
The finding that archaeal flagellins are initially made as preproteins with leader peptides indicated that a leader peptidase must be present in archaeal cells to process preflagellins. The leader peptide present on all archaeal flagellins is one of many hallmark features that differentiate archaeal and bacterial flagellins. Bacterial flagellins are not made as preproteins and are exported through a type III export pathway . In contrast, archaeal flagellins are made with leader sequences that are cleaved by a membrane-located peptidase . The scenario is believed to be similar to the cleavage of leader peptides from type IV pilins . In the case of type IV pilins, prepilins (precursor pilin molecules with a leader peptide) are substrates for a membrane-located prepilin peptidase. The best studied example is the prepilin peptidase of Pseudomonas aeruginosa, PilD, a bifunctional enzyme that removes a six-amino acid leader sequence from the prepilin and also N-methylates the amino acid at the newly formed N-terminus . This enzyme also processes a limited number of other homologous extracellular proteins termed pseudopilins in P. aeruginosa. Comparisons of all the substrates for the prepilin peptidase revealed that each of them has a conserved hydrophobic N-terminal region, with a positively charged leader sequence . Several mutant studies have elucidated the sequence requirements for proper processing of prepilins and pseudopilins by this dedicated peptidase [84–87].
In the case of P. aeruginosa prepilin, a number of substitutions can be tolerated at the position +1, with cleavage only occurring at the peptide bond between the leader peptide and mature protein. However, changing the −1 glycine to any other amino acid, except alanine, resulted in the loss of cleavage . With a change to alanine, partial processing was observed. The −2 position in type IV pilins is almost always lysine [83, 87]. However, in the case of P. aeruginosa, changing the lysine to either threonine or asparagine had no effect on cleavage of pilin or assembly of pili . Lastly, the invariant glutamic acid at the +5 position was shown to be important for assembly of the pilin into the P. aeruginosa type IV pilus. When the +5 glutamic acid was changed to other amino acids, cleavage was observed but not N-methylation, and the processed substrate was not assembled into the pilus. These experiments and further in vitro studies on the purified type IV prepilin peptidase of P. aeruginosa have helped the understanding of extracellular protein secretion in bacteria.
Very little is known about protein secretion in the archaea. Independent studies and the complete genome sequences of a number of archaea have revealed a number of Sec homologues, namely SecD, SecE, SecF and SecY , which are known to form a multisubunit membrane bound complex termed translocase in E. coli. The presence of these homologues and other protein secretion machinery components in archaeal species suggests that a system similar to the bacterial general secretion pathway is likely present. In addition, there are type II secretion homologues in archaeal species (at least one in the flagella gene family, FlaI) that are presumably involved in exporting substrates across or into the cytoplasmic membrane. Specific archaeal proteins have been identified that have either putative or demonstrated leader peptides. Examples of demonstrated leader peptides from exported proteins are found in the S layer proteins and numerous flagellins of methanogens [34, 40, 89]. The glucose binding protein of Sulfolobus solfataricus also has a demonstrated leader peptide . In each case, a positively charged leader sequence that ends in glycine is found. Proteins with a putative leader sequence that are thought to be extracellular or membrane-located have been identified from complete genome sequences. These include cutinases, amylases and sugar binding proteins. However, there are no data describing how these proteins are processed and exported. Furthermore, these substrates have variable amino acids in the positions that are normally conserved in flagellins, specifically the −2 and −3 positions of the leader sequence . Furthermore, the S layer proteins from the flagellated hyperthermophilic methanogens Methanothermus sociabilis and M. fervidus have typical, bacterial-like leader peptides that are 22 amino acids long . The amino acids after the cleavage site in these S layer proteins are acidic and basic, but not hydrophobic. This variety in the nature of substrates suggests that there might be other peptidases in the archaeal cell membrane that are involved in processing these different classes of substrates. Specifically, a dedicated peptidase that processes preflagellin molecules may exist to support the high number of molecules that are required for flagellar filament biosynthesis.
Our laboratory has demonstrated preflagellin peptidase activity in methanogens . An in vitro assay based on that previously described for type IV prepilin peptidase was used to detect the cleavage of the leader sequence from the preflagellin of M. voltae. In the preflagellin peptidase assay, E. coli membranes containing the overexpressed preflagellin (as the substrate) are mixed with methanogen membranes (as the enzyme source). M. voltae preflagellin peptidase activity was found to be optimal at 37°C and pH 8.5, in the presence of 0.4 M KCl with 0.25% (v/v) Triton X-100. Cleavage of the leader peptide from the preflagellin substrate is monitored by immunoblotting using anti-flagellin antibody in which the processed preflagellin (i.e. without its 12-amino acid leader peptide) migrates faster in SDS–PAGE. The precise cleavage site has been confirmed by N-terminal sequencing of the processed preflagellin. Cleavage only occurs after an invariable glycine residue located at 12 amino acids into the preflagellin sequence. This finding confirmed previous data, which showed that numerous methanogen flagellins isolated from purified flagellar filaments have an N-terminal amino acid sequence starting 13 amino acids into the corresponding preflagellin sequence [34, 93].
Analysis of the leader sequence of all available archaeal preflagellins reveals that the amino acids in close proximity to the cleavage site are conserved (Table 2) . In each case, glycine is located at the −1 position relative to the cleavage site. The −2 position is always occupied by a basic amino acid, either lysine or arginine. Furthermore, the −3 position is almost always a charged amino acid. Presently, the only example where a charged residue is not in the −3 position is a preflagellin of N. magadii, where a glycine residue is found (GenBank accession number O93718). Based on this conservation of charged amino acids at the cleavage site, it was hypothesized that these specific amino acids were important for proper processing of archaeal preflagellins. We have used site-directed mutagenesis to change the amino acids at the −1, −2 and −3 positions of the leader sequence of M. voltae FlaB2 (N.A. Thomas, E. Chao and K.F. Jarrell, unpublished data). In addition, we have also examined the role of the length of the leader sequence in the cleavage reaction.
Changing the −1 glycine of the M. voltae flagellin to glutamic acid, phenylalanine or arginine resulted in a loss of processing by the preflagellin peptidase (Fig. 5). Substitution of the glycine residue with alanine at position −1 resulted in partial cleavage. These findings are similar to those previously reported for the prepilin of P. aeruginosa, where glycine at the −1 position is required for in vitro cleavage, and an alanine results in partial cleavage . It has been postulated that a small amino acid may be required at the cleavage site to allow access for the peptidase to mediate the cleavage event. Surprisingly, a wide variety of amino acids are tolerated in the +1 position of the prepilin as determined by substitution experiments. Although no substitutions have been performed at the +1 position of archaeal preflagellins, in nature a number of amino acids are found there in different flagellins from many archaeal species although alanine is by far the most common. This suggests that the −1 residue (which is always a glycine) most likely plays an integral role in positioning the peptidase for cleavage to occur.
Posttranslational processing of wild-type and mutant M. voltae flagellin FlaB2 by a membrane-located preflagellin peptidase. A: Schematic of mutant FlaB2 proteins with a single amino acid substitution or a deletion of amino acids. The wild-type sequence is shown at the top. B: Immunoblot of the cleavage reactions corresponding to the substitutions performed at the −1 position of FlaB2. After the peptidase assay was performed, processed flagellin was detected by immunoblotting. Time points correspond to 0 and 30 min after the addition of M. voltae membranes to start the reaction. Lane 1, −1 alanine 0 min; lane 2, −1 alanine 30 min; lane 3, −1 glutamic acid 0 min; lane 4, −1 glutamic acid 30 min; lane 5, −1 arginine 0 min; lane 6, −1 arginine 30 min; lane 7, −1 phenylalanine 0 min; lane 8, −1 phenylalanine 30 min; lane 9, −1 glycine (wild-type) 0 min; lane 10, −1 glycine 30 min. The arrow points to the cleavage product. Only partial processing was observed when alanine was substituted for glycine at position −1.
Changing the −2 lysine to glutamic acid in M. voltae FlaB2 resulted in a loss of cleavage in the in vitro peptidase assay. Furthermore, changing the −2 lysine to alanine (a neutral and small amino acid) also resulted in loss of cleavage. These findings implicate the positive charge of the lysine (or in some archaeal preflagellins arginine) as an important factor in either the recognition or cleavage event mediated by the preflagellin peptidase. This is in contrast to the prepilin of P. aeruginosa, where the conserved lysine at −2 can be changed to many different amino acids which still result in correct cleavage by the peptidase . Indeed, the conserved nature of the positive charge at the −2 position in all archaeal preflagellins suggests some important role. In addition to the positive charge, the large side groups of both arginine and lysine may be important in stabilizing the peptidase at the cleavage site. Alternatively, one could envision that the active site of the peptidase might require electrostatic interaction with the preflagellin substrate. An argument against size alone being the crucial factor is that glutamic acid at the −2 position (negatively charged and large) did not result in cleavage. This feature of a large positively charged amino acid immediately in the vicinity of the cleavage site is a characteristic of all known archaeal flagellin sequences.
Substituting the positively charged lysine at position −3 of M. voltae FlaB2 with glutamic acid resulted in significantly less cleavage. This change introduces a negative charge at the −3 position (opposite of the positive charge found in the wild-type M. voltae preflagellin). That this significant change still allows some cleavage by the peptidase was not totally unexpected since the −3 position in some archaeal preflagellins (H. salinarum and A. fulgidus) is glutamic acid. Similarly, introducing an alanine at the −3 position also resulted in significantly less cleavage than observed with the wild-type preflagellin. More substitutions incorporating other amino acids at the −3 position are needed to ascertain the nature of the conserved charged amino acid at that position in most archaeal flagellins. However, the fact that positively and negatively charged amino acids as well as neutral amino acids can occupy the −3 position and result in proper cleavage is puzzling. Perhaps the cleavage event might have a charge requirement that is provided (but not necessarily exclusively) by the amino acids located near the cleavage site. Alternatively, perhaps the −3 position is required for a different step in the preflagellin processing other than the actual cleavage.
In addition to the substitutions performed in the aforementioned experiments, a deletion resulting in a six-amino acid leader peptide (in lieu of the 12-amino acid leader of the wild-type) was not processed by the preflagellin peptidase. This deletion removed the first six amino acids of the preflagellin and kept the amino acids surrounding the cleavage site intact. Thus, in the case of M. voltae, it appears that not only the conserved cleavage site, but also the length of the leader sequence is important in proper processing by the preflagellin peptidase. Interestingly, shorter leader peptides deduced from flagellin gene sequences in Pyrococcus sp. have been reported [37, 94], although it is not known if these reported leader sequences are representative of what is actually expressed in vivo. Nonetheless, in these putative shorter leader sequences, the conserved cleavage site of charged amino acids and an invariable glycine at position −1 is maintained.
We have also investigated the possibility of whether a bacterial type IV pilus system is capable of recognizing and processing archaeal preflagellin. Numerous studies have been performed in type IV pilus systems which show that the type IV peptidase from one species is capable of processing a substrate from a different species [95, 96], although this is not always true. As a family of proteins, type IV pilins are not very similar, except for the N-terminus, where a hydrophobic region is conserved. This situation is also found for all archaeal flagellins. We hypothesized that the similarities between archaeal flagellins and type IV pilins may be strong enough for heterologous recognition and processing. Both groups of proteins have short, positively charged leader peptides, followed by a similar long stretch of hydrophobic amino acids. In addition, both groups of proteins are structural determinants that are often posttranslationally modified prior to being incorporated into an oligomeric structure external to the cell.
In studies attempting to address whether the type IV prepilin peptidase of P. aeruginosa can recognize and properly process M. voltae preflagellin, it was found that no cleavage of the leader sequence from the preflagellin occurred . These experiments were performed in vivo in P. aeruginosa, where M. voltae preflagellin was expressed from a plasmid and then evaluated for processing using immunoblotting. To further investigate whether the secretory machinery could recognize the M. voltae flagellin, a fusion between the P. aeruginosa leader peptide and the mature M. voltae flagellin was created. In this fusion protein, the +1 alanine of the M. voltae flagellin was changed to a phenylalanine, which is the amino acid found in the +1 position of P. aeruginosa type IV pilin. In P. aeruginosa, the +1 phenylalanine is methylated prior to assembly into the pilus . When this fusion was introduced into wild-type and a pilD mutant (lacking the type IV prepilin peptidase) of P. aeruginosa, immunoblotting could not detect the recombinant M. voltae flagellin in cell extracts. Presumably, the fusion protein was not stable and was degraded by host proteases. However, when the +5 glycine in the same fusion protein was changed to glutamic acid as found in type IV pilins (and some archaeal flagellins such as M. hungatei), the recombinant protein was detected in both wild-type and pilD P. aeruginosa. In the wild-type P. aeruginosa, the leader sequence was apparently cleaved from this fusion as the processed M. voltae flagellin migrated faster than in the pilD mutant in SDS–PAGE. These findings demonstrate that given the appropriate leader sequence, M. voltae preflagellin with minor changes can be recognized by the type IV prepilin peptidase of P. aeruginosa. Furthermore, the hydrophobic nature of the M. voltae flagellin was sufficient for localization to the membrane of P. aeruginosa. It has been demonstrated through alkaline phosphatase fusions that the conserved hydrophobic amino acids in the mature pilin and the short positively charged leader sequence serve as export signals . The fact that the archaeal flagellin/pilin leader fusion was processed indicates that the hydrophobic character and not the exact sequence of the N-terminal region (of the substrate, pilin or flagellin) is sufficient for cleavage to occur. Presumably other factors within the type IV secretion system did not recognize the cleaved substrate for export or assembly. In the reciprocal experiment, to determine if M. voltae preflagellin peptidase could recognize and cleave the P. aeruginosa prepilin, no cleavage of the prepilin leader from the prepilin was observed (N.A. Thomas and K.F. Jarrell, unpublished data). Since the preflagellin peptidase has not been isolated, this experiment was performed in vitro, using membranes of expressed P. aeruginosa prepilin mixed with membranes of M. voltae.
Although specific amino acids involved in processing have been identified in the leader sequence of archaeal flagellins, it is not known if they are critical for recognition of the substrate, or whether they are important in the actual peptidase mediated cleavage reaction. However, the apparent requirement of a positively charged amino acid at the −2 position of archaeal preflagellins is unique. Studies addressing the role of the hydrophobic N-terminal region of archaeal flagellins have yet to be performed, although the overwhelming conservation of amino acids at the N-terminal region of all archaeal flagellins suggests some functional importance. The conserved region may serve as a recognition sequence or may be important in assembly of the flagellin subunits into the flagellar filament.
12 Comparison of archaeal flagella to bacterial flagella
While superficially similar to bacterial flagella, it is obvious that the archaeal organelle for motility is a distinct entity. The evidence is summarized as follows. All archaeal flagella are likely composed of multiple flagellins , while most bacterial flagella are composed of a single flagellin species (although there are numerous exceptions, e.g. [44, 100, 101]). Archaeal flagellins usually migrate in SDS–PAGE as much larger proteins than predicted from their gene sequence. This apparent larger molecular mass has been attributed by either direct  or indirect means [9, 17, 18, 56, 69] to posttranslational modifications, often glycosylation. While modifications like glycosylation have been reported for bacterial flagellins , these are rare occurrences. Sequence comparisons of archaeal flagellins to bacterial flagellins fail to reveal any similarity even though flagellins from even distantly related bacteria share N- and C-terminal similarities [24, 41]. Of course, archaeal flagellins bear the archaeal-specific trait of having a short signal peptide that is cleaved from the preflagellin and not found in the monomers incorporated in the flagella filament. Bacterial flagellins are not made with leader peptides (although the sheath proteins of Spirochaeta aurantia periplasmic flagella have a leader peptide , these proteins appear from sequence analysis not to be true flagellins ). In addition, none of the archaeal flagella accessory genes (flaC to flaJ in M. voltae) have homology to any bacterial genes including ones involved in bacterial flagellation and motility. In fact, with the availability of several completely sequenced genomes from flagellated archaea, it became clear that these species (and most likely all archaea) do not contain homologues of any genes involved in bacterial flagellation . There are no homologues of flagellins, rod proteins, the hook protein, hook-associated proteins, ring proteins, switch proteins or mot proteins.
The lack of homologues in archaea to any of the bacterial flagella structural proteins is intriguing in light of the presence of readily identifiable bacterial-like chemotaxis genes in several flagellated archaea. For example, a very bacterial-like set of chemotaxis genes, including cheA, cheW, cheR, cheB and cheY, has been found in H. salinarum. Analysis of completely sequenced genomes has revealed bacterial-like chemotaxis homologues in P. furiosus and A. fulgidus but not in M. jannaschii. In the bacterial system CheY is a response regulator which, when phosphorylated by CheA in response to a change in the external level of attractant or repellent, binds to FliM, a component of the bacterial flagella switch . This binding results in the clockwise rotation of the flagella, which leads to a tumble, or reorientation of the cell. However, as mentioned, no archaeal equivalent to the switch protein that binds CheY has been reported in any archaea. It seems that somehow most archaea have adapted a bacterial chemotaxis system to a unique motility structure.
Bacterial flagella are assembled by a type III export apparatus with similarities to the export pathway of many virulence factors [58, 105, 106]. In bacteria, there is a flagellum-specific export pathway located presumably within the basal body MS ring structure. The rod proteins, hook, hook-associated proteins and flagellin all reach their final destination beyond the cytoplasmic membrane by passing through the hollow structure formed by the MS ring and later the rod, hook and filament itself. The new subunits of flagellin are added to the filament at the distal tip furthest away from the cell only after passing all the way through the hollow interior of the entire basal body–hook–filament structure, truly one of the most remarkable assembly systems ever discovered. Indeed “the thought of flagellin molecules rushing down through the core of the flagellar motor, rod, hook and filament while all the time the structure is rapidly rotating is an astonishing one” . Macnab and DeRosier describe the assembly as “somewhat bizarre”  while Jones and Aizawa call this mode of transport “somewhat implausible, [although] several lines of evidence support it” . None of the rod proteins, hook, hook-associated proteins or flagellin has a signal peptide as found in substrates of the type II secretion systems. This lack of signal peptide cleavage is a hallmark of the type III pathway .
Examination of several observations of the archaeal flagella system makes this bacterial type of assembly pathway very unlikely. First there is the much thinner size of the archaeal flagella which typically average 10–14 nm in diameter [19, 20], about intermediate in diameter between bacterial flagella  and pili . It is not known if archaeal flagella are hollow like bacterial flagella and, if they are, whether the hole would be large enough to accommodate flagellin subunits passing through it. The bacterial flagellum has an interior hole of 2 nm and this channel is thought to be so small that the subunit must pass through it in a partially unfolded form . Secondly, and perhaps most strikingly, is an examination of the archaeal flagellins themselves. In addition to archaeal flagellins not bearing any sequence similarity to bacterial flagellins, the fundamental difference of the flagellins having a signal peptide alone would suggest that the mechanism of assembly of the archaeal flagellum is distinct from that of the bacterial counterpart. It is possible that the signal peptide somehow acts as a signal to direct flagellin monomers to the base of the flagella for subsequent transport through the structure for assembly at the distal tip. However, a simpler explanation is that the signal peptide is present to direct the transport of the subunits across the cytoplasmic membrane in a process involving signal peptide cleavage by a preflagellin peptidase. Clearly this suggests that assembly at the tip of the growing filament is unlikely and subunits would most likely be added at the base. The other piece of evidence related to assembly is the observation that the glycosylation of flagellin subunits in H. salinarum occurs outside the cytoplasmic membrane [54, 111]. This is readily explained if the flagellins cross the cytoplasmic membrane in a signal peptide-dependent fashion and are then glycosylated by a glycosylase located outside the cell. It would appear to be more difficult to explain the glycosylation evidence if the flagellins enter the hollow flagella structure from the cytoplasm and emerge outside the cell only at distal tip of the filament where the glycosylase would be waiting to modify the flagellin. As early as 1988, Gerl and Sumper recognized the logical extension of the glycosylation observations: “the flagellin polypeptides must be translocated across the cytoplasmic membrane before glycosylation. If so, aggregation to a functional flagellum is likely to occur by a mechanism different from that proposed for the assembly of eubacterial flagella” .
13 Comparison of archaeal flagella to type IV pili
The lack of similarity of archaeal flagella to bacterial flagella in terms of electron microscopic appearance and structural components contrasts with similarities noted between archaeal flagella and another structure that can be involved in bacterial motility, namely type IV pili. Type IV pili are adhesive organelles found in a variety of Gram-negative bacteria, such as P. aeruginosa, enteropathogenic E. coli, Myxococcus xanthus, Neisseria gonorrhoeae and Dicholebacter nodosus, whose primary function appears to be twitching motility . Early comparisons identified similarity between the N-termini of archaeal flagellins and type IV pilins  and both types of proteins reveal strikingly similar hydropathy plots with a very hydrophobic N-terminus, distinct from bacterial flagellins . In addition, type IV pilins are made initially as prepilins with a short (often six) amino acid signal peptide, which is cleaved by a specific membrane-localized enzyme, called the prepilin peptidase. Archaeal flagellins are made with short leader peptides that are cleaved by an enzymatic activity located in the cytoplasmic membrane. Also indicating a similarity between type IV pili and archaeal flagella is the homology observed between one of the flagella-associated proteins of archaea (FlaI) and PilT , which is a putative ATP binding protein involved in the type IV pilus system and one of the most conserved features of the type IV pilus systems of P. aeruginosa, M. xanthus and N. gonorrhoeae. Interestingly, mutations in pilT result in an overproduction of assembled pili but ones that cannot mediate twitching , much like mutations in mot genes where an assembled but non-rotating flagellum results. It has been suggested that the PilT may be the motor for the type IV pilus  and thus the possibility is raised that the archaeal equivalent, FlaI, may be a motor component for the archaeal flagella.
Several proteins involved in pilus function in P. aeruginosa have been reported by Darzins and Russell  as having remarkable similarities to well characterized chemotaxis proteins. These include PilG and PilH which have significant homology to the response regulator CheY, as well as proteins that are homologues to CheW (PilI), CheA (PilL), CheR (PilK), CheB (ChpB) and methyl-accepting chemotaxis proteins (PilJ). Both PilG and PilH have the correctly located aspartic acid and lysine residues that correspond to the phosphorylation site in CheY. The pilGHIJK gene cluster along with pilL is thought to be part of a second signal transduction pathway involved in regulating twitching motility. A key unanswered question is whether PilG and/or PilH interact directly at the base of the type IV pilus to effect a response as CheY does in its binding to the flagella switch. If they do, it has been suggested that they might interact with PilT and/or PilU, the genes for which happen to be located immediately downstream of the che-like cluster. Interestingly, the PilT equivalent FlaI is the only type IV pilus protein homologue found in the archaeal flagella system. However, it should be noted that there is no evidence for che-like genes in other type IV pilus systems such as the enteropathogenic E. coli or N. gonorrhoeae.
It should be pointed out that the mechanism of twitching motility remains poorly understood. While the original suggestion by Bradley  of the so-called pilus retraction theory has been generally accepted, it has been suggested that pilus rotation is a theoretically conceivable motor , an intriguing possibility that would be more like flagella.
While not shown specifically for type IV pili, the growth of certain types of pili has been shown to be through incorporation of the pilus subunits at the base of the structure [116, 117]. This is the presumed mechanism for type IV pili as well since there is no channel in the interior of the fiber, which precludes the bacterial flagella method of growth via incorporation of the subunits at the distal tip . Since archaeal flagellins have signal peptides as possessed by the pilins, it has been suggested that the archaeal flagellum grows by addition of new subunits to the base of the structure after the preflagellins have been processed by the membrane-bound preflagellin peptidase activity .
The type II protein export system, also called the general secretory pathway, has been found in a large number of Gram-negative bacteria . Exoproteins processed by the type II secretion system are synthesized with a typical 20–30-amino acid leader peptide at the N-terminus. They cross the cytoplasmic membrane in a sec-dependent manner, processed by leader peptidase and first appear in the periplasm before crossing the outer membrane.
The type II secretion system shares about 10 homologues with the type IV pilus system including three to five homologues with prepilin-like signal peptides (dubbed pseudopilins) of the main structural pilin (PilA), as well as PilB, PilC and the outer membrane gated channel PilQ [113, 118]. Finally the prepilin peptidase (PilD) involved in cleaving the short atypical leader peptide from the prepilin of the type IV pilus system is the same enzyme that processes the PilA homologues (pseudopilins) of the type II secretion system [119, 120]. The pseudopilins have been proposed to form a pilus-like secretion tube for export of substrates outside the cell, although evidence for such a pseudopilus is lacking [121, 122]. Substrates of the type II secretion system cross the cytoplasmic membrane in a sec-dependent fashion. However, despite similarities to the type II system, the type IV pilus substrates may be secreted via a sec-independent route  and may insert into the cytoplasmic membrane spontaneously .
In addition, homologues of these components are also involved in DNA uptake in Bacillus subtilis. Here the shared components include four prepilin-like proteins, a peptidase and a protein with an ATP binding motif (PilB homologue). In the archaeal flagella system, we also find multiple prepilin like proteins (two to five flagellins with a short atypical leader peptide and homology at the N-terminus to type IV pilins), a protein with an ATP binding motif (FlaI) and a peptidase. Found in the type II systems of Gram-negative bacteria but missing in the B. subtilis case and in archaea is the PilQ homologue since neither B. subtilis nor the archaea have an outer membrane. It may be that this ancient system developed to form a pilus in certain bacteria, while being involved in secretion in others, DNA uptake in still other bacteria and the formation of flagella in archaea .
14 Model for assembly of archaeal flagella
Do archaeal flagella assemble like bacterial flagella or like type IV pili or does the assembly of archaeal flagella represent a completely novel mechanism not yet observed in a bacterial structure? The latter is certainly a possibility when one considers that only a fraction of the total genes involved in archaeal flagellation has likely been identified. Assembly of type IV pili from enteropathogenic E. coli could be accomplished with a cluster of 14 genes in addition to the sec machinery  but in P. aeruginosa it has been estimated that as many as 40 genes may be required for production and function of type IV pili . Likewise, currently 44 genes have been identified that are required for assembly and function for bacterial flagella . Furthermore, there are no definitive functions for any of the archaeal genes identified as important for flagellation, other than certain flagellins. In fact, all that is known is that FlaB1 and FlaB2 are the major components of the filament in M. vannielii (and likely M. voltae) and that all five flagellin proteins are present in the filaments of H. salinarum. The role of the least abundant flagellins (FlaA and FlaB3 in M. voltae) or any of FlaC, D, E, F, G, H, I or J is not known. No bacterial flagella structural protein homologues have been detected in archaea and only one type IV pilus homologue (FlaI and PilT) has been detected in addition to the similarities between type IV pilins and archaeal flagellins themselves. Nevertheless, we have used, as a working model, a mechanism of assembly of archaeal flagella that contains some basic features of the type IV pilus system. This includes transport of the subunits to the cytoplasmic membrane with cleavage of the signal peptide by a dedicated enzyme with subsequent incorporation of the flagellins at the base of the growing structure (Fig. 6). As Macnab stated in 1990, in reference to polarity of growth of bacterial flagella, “It is interesting to ask why a mechanism involving distal addition should have evolved rather than the apparently simpler one of progressive insertion of subunits at the base” . In bacteria “The evidence against use of the primary export pathway (aside from the horrendous losses it would entail) is simply that none of the rod proteins, hook protein, hook-associated proteins, or filament protein have a cleaved signal peptide” . This is clearly not the case for the archaeal flagellins, which have leader peptides that are cleaved, and perhaps the members of this Domain have evolved the ‘simpler’ mechanism of subunit addition at the base.
Speculative model for the structure and assembly of a M. voltae flagellum. Flagellin subunits are escorted to the cytoplasmic membrane by chaperones where they processed by the preflagellin peptidase and inserted at the base of the growing structure. S.L., S layer; C.M., cytoplasmic membrane; P.C., polar cap; P.P., preflagellin peptidase.
It is believed that flagellin-specific chaperones bind to the preflagellins in the cytoplasm, possibly to prevent their aggregation via their hydrophobic N-termini. Chaperone-escorted preflagellin monomers are delivered to the cytoplasmic membrane where their leader peptide is cleaved by a membrane-located preflagellin peptidase, an enzyme thought to be specific for the flagella system. A preflagellin peptidase activity has been demonstrated in M. voltae membranes and the sequence immediately surrounding the cleavage site has been shown to be important for processing. Whether this transport involves an archaeal Sec system is unknown. If the flagellins are to be glycosylated as many archaeal flagellins are, this modification would occur on the exterior of the cytoplasmic membrane, as shown for H. salinarum. This could occur with the flagellins embedded in the cytoplasmic membrane via their N-termini with much of the protein exposed on the outer leaflet. These data point to a novel mode of assembly of flagella in the archaea where new subunits would be added at the base and not at the distal tip after passing through the hollow structure as is the case for bacterial flagella.
There is evidence that not all flagellins are equal. Mutants in the minor flagellin FlaA of M. voltae assemble normal-looking flagella when viewed in the electron microscope but the cells are less motile by semi-swarm plate analysis. Transcription data would suggest that FlaB3 is present in much smaller amounts than FlaB1 and FlaB2 as well. Does it play a specific role in assembly or function? Recent analysis of mutant flagellins of H. salinarum suggests that in the case of this extreme halophile, the A flagellins make up the bulk of the filament (corresponding to the FlaB1 and FlaB2 flagellins of M. voltae and M. vannielii). The B flagellins may be located at the proximal end of the flagella adjacent to the anchoring mechanism and may be terminators for growth or play the role of hook-associated proteins.
Of the archaeal proteins so far identified as likely required for flagella assembly/function (FlaA–FlaJ), it does not appear that FlaC, D, E, or H are minor components of the ‘intact’ flagella isolated by Triton X-114 phase separation since they have not been detected using immunoblotting techniques with specific antisera (S.L. Bardy, N.A. Thomas and K.F. Jarrell, unpublished observation). They may still be associated peripherally with the flagella as all of these proteins (as well as FlaI) fractionate to the membrane. It may be that, like the type IV pilus system, many of the proteins identified in the archaeal flagella system thus far locate to the cytoplasmic membrane, possibly forming a complex basal structure or secretion apparatus. This may not be stable to Triton X-114 extraction in the phase separation technique. The only genes found near the flagellin genes in all archaea are flaH, flaI and flaJ. FlaI is a homologue of PilT and likely an ATP binding protein which may supply the energy for assembly or be a motor for the system. FlaJ is a membrane protein containing many transmembrane domains. FlaI and FlaH have been shown to fractionate to the membrane of M. voltae. The universal presence and the strong conservation of sequence of these proteins throughout the archaea argue for an essential function in flagellation. Perhaps they interact to form a secretion complex that in concert with the preflagellin peptidase is necessary for the processing and insertion of the flagellins into the filament. The presence of the hook raises questions about the likely presence of hook-associated proteins at the junction of the hook and filament. These may be represented in the low molecular bands seen on SDS–PAGE analysis of the M. voltae‘intact’ flagella. Alternatively the minor flagellins such as FlaB3 of M. voltae may fulfil this role if needed. The hook protein itself remains unidentified, although recent evidence suggests that it may be FlaB3.
The anchoring mechanism can be assembled first on the polar cap, which may exist submembranously. Of the proteins identified thus far we think the most likely candidate to be a component of the anchoring structure is FlaG, which has a hydrophobic N-terminus and N-terminal similarity to the flagellins. In this way it is similar to certain type IV pilus proteins (PilE in P. aeruginosa) which have sequence similarities to the main pilin and which are believed to be minor structural proteins of the pilus . FlaG and FlaF have not yet been overexpressed and antibodies to them are not available so their presence in the ‘intact’ flagella has not been tested. FlaA and FlaB3 are most certainly minor structural components of the flagella of M. voltae, likely performing a specific role. A number of such roles requiring minor amounts of proteins can be envisioned, such as hook–filament junction proteins or capping proteins. The major flagellins, with their signal peptide cleaved but likely still anchored in the cytoplasmic membrane by their hydrophobic N-termini, would be incorporated at the base of the structure. In this way it might be analogous to the model proposed by Fussenegger et al.  for type IV pili in N. gonorrhoeae. Perhaps a cap protein is incorporated first as an initiator of filament growth and the flagellin incorporated under the cap with new subunits added at the base.
Whether a special pore-forming protein is needed to allow the flagella to pass through the outer wall layer is unknown. Type IV pili do have a component (PilQ in P. aeruginosa) which forms a pore through the outer membrane allowing passage of the pilus . Archaea do not have an outer membrane but most are covered with an S layer which is often the only wall component located external to the cytoplasmic membrane. Bacterial flagella are not believed to need a pore-forming protein to allow passage of flagella through an S layer located on the exterior of a cell wall. Instead the S layer array is distorted by the flagellum [128, 129]. However, we are not aware of any studies in archaea specifically addressing this point.
15 Concluding remarks
The archaeal flagellum is a motility organelle widely distributed throughout the third Domain of life. While it performs the same function as its bacterial namesake, the compositions of the two structures bear no common homologues identified in searches of the sequenced genomes. There is evidence, especially the presence of a cleaved leader peptide, which suggests that the assembly mechanism for the two structures may also be fundamentally different. Polarity of growth experiments are needed to resolve the question of whether new subunits are added at the base or distal tip of the archaeal filament. The answer to that question is paramount to any model of assembly. Also needed is a determination of the function of the flagella-accessory gene products already identified and the identification of new genes involved in the biosynthesis and function of archaeal flagella. The former problem is being addressed by isolation of insertional mutations in these genes in both methanogens and extreme halophiles. The latter question is being currently approached by N-terminal sequencing of minor components (non-flagellins) of flagella isolated with some attached anchoring structure present.
The study of archaeal flagella has provided a model system for addressing both protein export and organelle assembly in archaea, two areas of research that have received little attention thus far from researchers.
Research in the authors’ laboratory was funded by operating and equipment grants from the Natural Sciences and Engineering Research Council of Canada (NSERC). The authors are grateful to N. Patenge and D. Oesterhelt for supplying data prior to publication and to S.-I. Aizawa for valuable advice and electron microscopy. K.F.J. wishes to thank former graduate students M. Kalmokoff, D. Bayley, D. Faguy, T. Karnauchow and J. Correia for their contributions to this work.
(1997) Diversity of alkaliphilic halobacteria: proposals for transfer of Natronobacterium vacuolatum, Natronobacterium magadii, and Natronobacterium pharaonis to Halorubrum, Natrialba, and Natronomonas gen. nov., respectively as Halorubrum vacuolatum comb. nov., Natrialba magadii comb. nov., and Natronomonas pharaonis comb. nov., respectively. Int. J. Syst. Bacteriol. 47, 853–857.
(1998) Flagellin genes of Methanococcus vannielii: amplification by the polymerase chain reaction, demonstration of signal peptides and identification of major components of the flagellar filament. Mol. Gen. Genet. 258, 639–645.
(1997) A 16-kDa protein functions as a new regulatory protein for Hsc70 molecular chaperone and is identified as a member of the Nm23/nucleoside diphosphate kinase family. J. Biol. Chem. 272, 2607–2614.
(1991) Analysis and nucleotide sequence of the genes encoding the surface-layer glycoproteins of the hyperthermophilic methanogens Methanothermus fervidus and Methanothermus sociabilis. Eur. J. Biochem. 199, 147–152.
Macnab, R.M. (1990) Genetics, structure, and assembly of the bacterial flagellum. In: Biology of the Chemotactic Response (Armitage, J.P. and Lackie, J.M., Eds.), pp. 77–106. Cambridge University Press, Cambridge.