Small Proteins in Archaea, a Mainly Unexplored World

ABSTRACT

In recent years, increasing numbers of small proteins have moved into the focus of science. Small proteins have been identified and characterized in all three domains of life, but the majority remains functionally uncharacterized, lack secondary structure, and exhibit limited evolutionary conservation. While quite a few have already been described for bacteria and eukaryotic organisms, the amount of known and functionally analyzed archaeal small proteins is still very limited. In this review, we compile the current state of research, show strategies for systematic approaches for global identification of small archaeal proteins, and address selected functionally characterized examples. Besides, we document exemplarily for one archaeon the tool development and optimization to identify small proteins using genome-wide approaches.

INTRODUCTION

Short open reading frame (sORF)-encoded small proteins with a size of less than 70 amino acids (aa) have been reported in all three domains of life (e.g., references 1 to 3). Although the first small protein was reported in 1965 (4), small proteins have been overlooked in prokaryotes for a long time due to technical limitations in detection. However, modern genomics and transcriptomics technologies, combined with systematic genome-wide approaches, have uncovered an unexpected genome complexity in prokaryotes. As a result, over the past decade, global approaches have discovered a wealth of hidden small genes containing short open reading frames in many prokaryotic genomes in addition to genes encoding larger (standard) proteins and genes of noncoding RNAs (ncRNAs) (5, 6). In consequence, the number of experimentally identified sORFs and verified small proteins is constantly increasing. However, information regarding a specific cellular physiological role is lacking for the majority of confirmed small proteins, and only a small number identified in diverse prokaryotes have been functionally characterized in detail. They have been reported to play crucial roles in various cellular processes, including stress response, transcription and translation regulation, metabolism, and antibiotic resistance (reviewed in references 1, 3, and 6 to 18). In the respective reviews and original reports, the definitions of small proteins encoded by sORFs differ, particularly in their maximal size, which ranges between 50 to 100 aa. Here, we focused on proteins with a maximal length of 70 residues.

Due to the late discovery of archaea, as well as their frequent occurrence in extreme habitats (with high temperatures, high salt, extreme pH, or anoxic conditions), only a small proportion of the Archaea species has been described and successfully cultured in the lab so far. In the NCBI database are 1,985 genome assemblies of archaea, in contrast to 40,312 genome assemblies of bacteria (data retrieved 26 May 2021), clearly showing the underrepresentation of archaea. Consequently, in comparison to bacteria, significantly fewer small proteins have been identified in archaea. In this review, we report on the current research on small proteins in archaea and introduce the tools for genome-wide screening approaches generally used for prokaryotes. Finally, the development and optimization of tools to identify small proteins is exemplarily documented for the methanogenic archaeon Methanosarcina mazei.

STRATEGIES AND SYSTEMATIC APPROACHES FOR GLOBAL IDENTIFICATION OF SMALL PROTEINS IN ARCHAEA

In addition to the relatively small number of archaea described so far and the fact that they are often difficult to culture, several methodological reasons, not archaeon-specific but more general, have complicated the identification of small proteins in archaea; for example, automated gene prediction algorithms used are missing the small proteins (1, 3, 19–24). Another profound reason is the technical limitations of detecting small proteins (e.g., SDS-PAGE resolution or mass spectrometry) (25–29). Overall, a standard procedure for detecting small proteins in archaea is missing. Different strategies, including systematic approaches for the global identification of small proteins, are often combined. Since these methods have been used for both bacteria and archaea and are discussed in detail in a separate review within this issue, they are only briefly summarized here, as follows.

The majority of the so far identified and described archaeal small proteins (summarized in Table 1) have been found by serendipity. Small proteins were identified using various experimental approaches not explicitly aiming to identify small proteins, like the 7-kDa DNA-binding proteins from Crenarchaea (Sulfolobales) detected by isopycnic centrifugation of cell extracts and subsequent isolation of the DNA-binding proteins (30). Brz1 from Halobacterium salinarum was identified by inverse structural genomics. The proteome, which is smaller than 20 kDa, was first enriched by filter membrane centrifugation, then separated by Tricine–SDS-PAGE, followed by a short digestion, and then analyzed by One-dimensional liquid chromatography–mass spectrometry/tandem mass spectrometry (LTQ-FT system) and mapped to the genome (31). In this approach, 20 proteins containing a characteristic DNA/RNA-binding zinc finger motif were discovered, including 10 proteins smaller than 70 aa (32). The small protein rubredoxin was originally described for Clostridium pasteurianum (4). Rubredoxin of Pyrococcus furiosus was isolated under anaerobic conditions by several different chromatographic separations (including ion exchange, hydroxyapatite chromatography, and size exclusion chromatography), followed by amino acid sequencing (repetitive Edman degradation) (33). Small proteins similar to experimentally discovered ones were later identified in the databases based on sequence comparisons (similarity-based identification). Today, modern and sophisticated sequencing techniques allow global approaches to predict sORFs and verify small proteins. In a first genome-wide transcriptomics approach comparing M. mazei growing under various nitrogen availability conditions, 44 short ORFs encoding proteins 30 to 90 amino acids (aa) in length were discovered. The identification of sORFs was achieved by manual curation of the cDNA reads, mainly in the noncoding region of the genome (34). Three of those were later experimentally verified to be translated by LC-MS/MS analysis and were predicted to be involved in regulation of nitrogen metabolism (2). Other examples of archaeal sORFs identified by transcriptome analysis are the respective ORFs encoding four small heat shock proteins and four TRAM-domain containing proteins with low molecular weight in the psychrophilic methanoarchaeon Methanolobus psychrophilus R15, which are upregulated at 4°C compared to expression during growth at 18°C (35).

TABLE 1.

Identified and functionally characterized small proteins in archaea

Small protein	Length (aa)	Organism(s)	Function	Reference(s)
VhuU	69	Methanococcus voltae	Posttranslational processes, crucial for [NiFe] hydrogenase activity	88, 89
Histone-like	55–69	Euryarcheota	Stabilization of DNA	63
7-kDa binding family	51–78	Crenaechaeota (Sulfolobis sp.)	Stabilization of DNA, transcription factors	30, 61, 65–67
Cold shock proteins	60–70	Several groups	RNA-binding chaperons, TRAM domain	35, 58, 82, 83
sORF56	56	Sulfolobus islandicus	Transcription factor	69, 70, 72
Nop10	60	Methanocaldococcus jannaschii, Pyrococcus abyssi, Pyrococcus furiosus	Crucial for pseudouridine synthetase	85 – 87
TIP	64	Thermococcus kodakarensis	PCNA inhibitor	76, 77
Spt4	59–61	Methanocaldococcus jannaschii, Pyrococcus furiosus	Crucial part of transcription elongation factor	78–80, 133
SecE	61	Pyrococcus furiosus	Crucial for SecYEG protein transport channel	114 – 119
SecG	53	Methanocaldococcus jannaschii	Plug of SecYEG protein transport channel
CedA2	52	Sulfolobus acidocaldarius	Part of the Ced DNA import channel	120
RelB	51–82	Pyrococcus horikoshii, Methanococcus jannaschii, Archaeoglobus fulgidus	Antitoxin of a TA system II	97, 98, 134
SAMP1/SAMP2	66, 69	Haloferax volcanii, Pyrococcus furiosus, Methanosarcina acetivorans	Ubiquitin-like modifiers	105–107, 135, 136
AfDmpI	63	Archaeoglobus fulgidus	Keto-enol tautomerase activity	137
Brz-1	60	Halobacterium salinarum	Transcription factor	31, 32
Dodecin	68	Halobacterium salinarum	Quenching based on dealkylation of riboflavin	127 – 129
Rubredoxin	45–55	Pyrococcus furiosus	Detoxification of reactive oxygen species	33, 121
HVO_0416	58	Haloferax volcanii	Affects biofilm formation
HVO_0649	68	Haloferax volcanii	Affects swarming
HVO_0758	56	Haloferax volcanii	Affects biofilm formation and swarming	56, 57, 59
HVO_2142	50	Haloferax volcanii	Affects biofilm formation and swarming
HVO_2523	50	Haloferax volcanii	Affects biofilm formation and swarming
HVO_2753	59	Haloferax volcanii	Affects biofilm formation and swarming
HVO_2901	54	Haloferax volcanii	Affects biofilm formation and swarming
HVO_2922	60	Haloferax volcanii	Upregulated under iron limitation
HVO_A0556	47	Haloferax volcanii	Affects biofilm formation and swarming
sP36	61	Methanosarcina mazei	Related to nitrogen metabolism
sP41	53	Methanosarcina mazei	Putative transcription factor	2
sP44	23	Methanosarcina mazei	Putative transcription factor
sP26	23	Methanosarcina mazei	Regulator of glutamine synthetase	52

Another global approach for detecting small proteins is based on a proteome-wide approach. Small proteins are generally underrepresented in most proteomics studies, since the identification by classical methods of bottom-up proteomics is challenging and shows an inherent bias against small proteins (36, 37). These biases arise due to a number of factors, such as the sequence length, which following enzymatic digestion can result in only a limited number of peptides, or in extreme cases no or only a single peptide(s). Additionally, in many cases, small proteins are assumed to be in low abundance, which further hampers their detection against a background of high-abundance proteins capable of producing a large number of potentially detectable peptides. Thus, in recent years, tools have been developed and adapted to optimize for the detection of small proteins (often termed peptidomics approaches). One major issue is to reduce the complexity of the samples and to enrich for the small proteins or deplete for larger proteins. Examples for enrichment of small proteins are gel-based methods (38, 39), organic solvent depletion methods (40), solid-phase extraction (37, 41, 42), molecular weight cutoff filters (43), size exclusion chromatography (44), and adapted protein extraction methods (45). To reduce or circumvent the loss of small proteins due to binding to larger protein complexes during sample processing, chaotropic reagents, such as guanidinium hydrochloride, can be used (37, 42). Recently, a combined bottom-up and top-down proteomics workflow was utilized for the analysis of small proteins (37). The combined analysis provides the benefits of both methods, allows characterization of proteoforms and a limited number of posttranslational modifications, and results in a more confident identification of small proteins than has previously been reported (see “Small proteins identified in M. mazei”). Finally, a multiprotease approach can be used for the improved identification and characterization of small proteins, as recently shown for M. mazei (46).

The global tool of ribosome profiling (also called RiboSeq or ribosome foot-printing), which was originally established in eukaryotes (47), has been recently successfully adapted for prokaryotes to identify new small proteins (e.g., references 48–50). In general, this method can identify the translation of predicted ORFs (including sORFs). Initially, extracted RNA is depleted for rRNA and nucleolytically digested, except for those fragments that are protected by ribosomes. By sequencing the remaining mRNA fragments that are protected from nuclease digestion by actively translating ribosomes (covering ∼30 to 40 nucleotides), the method provides a global picture of transcripts that are actively translated. The method further enables determination of the ORF boundaries, or refining of sORF annotations (27, 51); see also the respective review in this special issue. Very recently, the first RiboSeq analysis for an archaeon, Haloferax volcanii, was reported (50). Evaluating the data sets showed that 75% of the footprints map to the first nucleotide of the start codon, observing 3-nucleotide periodicity, as reported for eukaryotes. Overall, 160 unknown translation start sites were detected in H. volcanii, of which 68 represent small proteins.

CATEGORIES AND STRUCTURE OF SMALL PROTEINS IN ARCHAEA

Depending on the localization, the experimentally verified sORFs can be differentiated into different categories (Fig. 1). The first category is translated sORFs located in intergenic regions, which have been identified in, for example, M. mazei (e.g., sORF36, sORF41, sORF44, and sORF26) (2, 52), as well as in Halobacterium salinarum (Brz1) (32) (Fig. 1A). The second category is sORFs encoded in the 5′ untranslated region (UTR) of an operon or gene, so-called upstream ORFs (uORFs) (Fig. 1B), which might regulate the translation of the downstream gene(s) as known for bacteria (e.g., attenuation [6, 53, 54]). Translation of uORFs has recently been shown in H. volcanii (50); however, little is currently known about the effects of the respective small proteins and whether they regulate the translation of the downstream ORF. Besides, internal out-of-frame sORFs (alternative ORFs) and antisense translation ORFs (asORFs) have been identified in H. volcanii by the recent ribosome profiling approach (Fig. 1C and D) (50). A recently published study reports on the structural analysis of 27 small proteins, identified to be expressed in nine different bacteria and archaea, by circular dichroism (CD) and nuclear magnetic resonance (NMR) spectroscopy (55). Overall, the systematic screen demonstrated that most of these small proteins appeared to be unstructured or only partially folded, especially if they contain fewer than 30 aa. Most interestingly, bioinformatics tools such as the FuzPred algorithm predict that some of these unstructured small proteins might fold into a structure through disorder-to-order transitions upon interacting with their target protein (55). Besides, the structures of two archaeal small proteins of H. volcanii have been recently determined and characterized by NMR analysis, namely, those of the HVO_2753 protein (59 aa), which is potentially involved in biofilm formation (56), and the stress-regulated HVO_2922 protein (60 aa) (57).

FIG 1.

Localization of different types of short ORFs. (A) Intergenic localization, has its own regulatory elements; (B) localization within 5′-UTR (uORFs); (C) localization within a large ORF; and (D) localization in antisense. Panels B, C, and D created based on the data from Gelsinger et al. (50).

(POTENTIAL) FUNCTIONS OF SMALL PROTEINS IDENTIFIED IN ARCHAEA

In archaea overall, 335 small proteins (<70 aa) have been reported based on UniProt data (Fig. 2). Well-known examples of archaeal small proteins are able to interact with DNA or RNA or with other proteins. Some of these represent subunits of complexes. The distribution of these archaeal small proteins based on protein category and archaeal genus is summarized and depicted in Fig. 3. The color coding additionally indicates the respective characterization status, namely, structural evidence, evidence at protein or transcriptional levels, inferred from similarity, or bioinformatics prediction. The majority are regulated in response to a specific stress and, as mentioned, most have been found by serendipity or, more recently, by global screens. However, for most of these proteins, experimentally verified functional analysis is still scarce or missing (e.g., references 2, 38, 40, 50, and 58 to 60). Only for a few has a physiological function been predicted or experimentally identified. The current knowledge on small archaeal proteins and their respective predicted or experimentally verified functions is summarized in Table 1; selected examples will be introduced in more detail in the following sections.

FIG 2.

Distribution of small proteins within prokaryotes. Recorded reviewed small proteins (≤70 aa) within prokaryotes in the database UniProt (retrieved 5 October 2020), excluding ribosomal proteins and subunits of DNA polymerase.

FIG 3.

Distribution of small proteins within archaea. UpSet plot representing the number of small proteins (≤70 aa) in archaeal genera. Vertical bars represent the number of proteins per category shared between the specific archaeal genera. Protein categories are color coded by their characterization status based on the UniProt data (retrieved 5 October 2020). Horizontal bars in the lower panel indicate the total number of proteins contained in each genus. The different dots denote the presence of the protein category by genus. UpSet plot was generated using the library ggplot2 in R (132).

Small proteins interacting with DNA or RNA.

The majority of archaeal small proteins described to date are DNA/RNA-binding proteins. Archaeal histone-like proteins are generally smaller than eukaryotic histones (ranging from 55 to 69 aa), contain a simple histone fold (HF), and establish chromatin structures similar to those of eukaryotes; however, they form tetramers or hexamers (61–64). Some Crenarchaeota lack histone-like proteins but contain members of the so-called “7-kDa DNA-binding” protein family (ranging from 51 to 78 aa), which show similar chromatin structuring function. Within this protein family are small basic proteins like the homologous Sul7d (Sulfolobus islandicus), Sac7d (Sulfolobus acidocaldarius), and Sso7d (Sulfolobus solfataricus). Those proteins are thermostable up to 100°C and stable under an extreme pH range from 0 to 12 (65). The proteins bind double-stranded DNA (dsDNA) in the minor groove without sequence specificity, resulting in an increase of stability of the DNA, especially at high temperatures (61, 65–67). Moreover, it has been reported that some of the 7-kDa DNA-binding proteins can be covalently modified. Sul7d can be methylated on the solvent-exposed surface (61), whereas Sso7d can be ADP-ribosylated, which modulates its thermoprotective effect on bound dsDNA (66).

Another DNA-binding protein is the extremely stable ORF56 gene product (56 aa) encoded on a conserved plasmid (pRN1) from S. islandicus (68–70). This small protein is a dimer in solution, but binds as a tetramer to its own promoter region at inverted repeats (autoregulation). ORF56 is organized in an operon together with ORF904, which encodes a multifunctional replication protein (mediating primase, DNA polymerase, and helicase activities), and overlaps the start site of ORF904. The ORF56-ORF904 operon has been shown to be essential for plasmid replication (71). Modulation of the expression of the operon on a transcriptional level results in regulation of the plasmid copy number by ORF56 (68, 72).

Two small proteins have been reported to be involved in replication and transcription. Proliferating cell nuclear antigen (PCNA), a trimeric protein that acts as a DNA clamp on the lagging strand (73), plays an essential role in replication and repair (74, 75). The small protein TIP (Thermococcales inhibitor of PCNA) in Thermococcus kodakarensis, which consists of 64 aa, was reported to inhibit PCNA-dependent activities (76). TIP disrupts the intact PCNA complex by binding to one monomer and blocking the interaction side for the trimer formation; consequently, the PCNA complex dissociates. This interaction with a small protein represents a novel mechanism to regulate PCNA activity (77). The transcription elongation factor complex Spt4/5 in P. furiosus (78) and Methanocaldococcus jannaschii (79), where Spt4 contains 61 and 59 aa, respectively, binds within the RNA polymerase claw encircling the DNA-binding channel. Due to this interaction, the small proteins prevent an early disengagement of the RNA polymerase and, furthermore, inhibit the formation of DNA secondary structures inside the transcription bubble. Thereby, the Spt4/5 complex improves the processivity of the RNA polymerase and transcription (78–80).

Besides small heat shock proteins, proteins analogous to bacterial cold shock proteins (Csps) have been reported in archaea. In Methanolobus psychrophilus and Methanococcoides burtonii, several small proteins (60 to 70 aa) containing a single TRAM (TRM2 and MiaB) domain (81) were detected that are strongly induced under low temperatures (35, 82). The TRAM domains (60 to 75 aa) are supposed to bind RNA and belong to RNA-modifying proteins (81). Four M. psychrophilus TRAM-containing small proteins have been demonstrated to functionally complement an Escherichia coli Csp-deficient strain, proving its function as nonspecific RNA chaperone (58, 83). Recent work showed that the small TRAM0076 protein (69 aa) from Methanococcus maripaludis globally effects gene expression at the posttranscriptional level, in addition to the RNA chaperone activity (84).

The small protein Nop10 (60 aa), is essential for the function of pseudouridine synthase Cbf5 from Methanocaldococcus jannaschii, Pyrococcus abyssi, and P. furiosus. The conversion of uridine into pseudouridine is the most frequent modification of structured RNAs (85). Nop10 consists of a conserved N-terminal zinc ribbon and a C-terminal α-helix, which are connected by a linker. Both ends seem to interact with the pseudouridine synthase Cbf5, which potentially enables RNA binding. It has been shown that binding of Nop 10 to Cfb5 results in masking its negative charges and mediates an interaction site of Cfb5 for pseudouridylation to target RNA (85–87).

Small proteins interacting with proteins or representing subunits of complexes.

The small protein VhuU (25 aa) is one of the first functional characterized small proteins (88, 89). It represents a subunit of the [NiFeSe] hydrogenase (Vhu) from Methanococcus voltae, which is a key enzyme in methanogenesis under sufficient selenium conditions. The small protein VhuU contains selenocysteine and is a part of the primary reaction center of the Vhu hydrogenase. It has been shown to be posttranslationally processed. The preprotein consists of 69 aa and is shortened, leading to a 25-aa small protein (89). Interestingly, another [NiFeSe] hydrogenases from M. voltae (Fru), contains the corresponding peptide sequence of VhuU as C-terminal parts of the large subunit (90). A respective engineered fusion protein consisting of VhuA and VhuU showed no significant difference in hydrogen uptake activity of the Vhu hydrogenase, but a 10-fold higher K_m value for hydrogen compared to that of the wild-type enzyme. Thus, at low hydrogen concentrations, the wild-type enzyme Vhu, consisting of two subunits, might have an advantage in competing with other hydrogen-utilizing microorganisms in the environment (91, 92).

Several toxin-antitoxin systems are reported within the small proteins for Bacteria (reviewed in, e.g., references 18, 93, 94). In archaea, only one toxin-antitoxin system has been observed to date, the RelE-RelB toxin-antitoxin system II reported in M. jannaschii, Archaeoglobus fulgidus, and Pyrococcus horikoshii (95). The toxin-antitoxin systems consist of one small toxic protein (RelE, ranging from 85 to 92 aa) and the corresponding small antitoxin protein (RelB, ranging from 57 to 82 aa), which neutralizes the toxin on the protein level (95, 96). Both genes are organized in an operon induced by nutritional stress and represent homologs of the E. coli MazFE system. RelE encodes a ribosome-dependent RNase, which cleaves stop (UAG, UAA, and UGA) and sense codons (UCG and CAG), thus inhibiting protein synthesis and cell growth. However, as long as RelE and RelB are simultaneously present, the toxic effect of RelE is neutralized by RelB via protein-protein interaction that prevents binding of RelE to the ribosome (97, 98). Under certain conditions, RelB is degraded by the Lon protease, leading to rapid cell stasis or death (98).

Small proteins also play important roles in protein modification; one example is the process of “SAMPyling” other proteins. The ubiquitin-like proteins (Ubl), which do not have high sequence similarities to the eukaryotic ubiquitin (Ub) but do have a similar three-dimensional structure containing the β-grasp fold and a C-terminal glycine, are present in all three domains of life (99–104). In archaea, so called small archaeal modifier proteins (SAMPs) have been identified (105–110). Examples include the two small modifier proteins SAMP1 (87 aa) and SAMP2 (66 aa) from Haloferax volcanii, which have been shown to be involved in protein modification by covalent binding to lysine residues of the target proteins, subsequently leading to cleavage of the conjugates by HvJAMM1 (111–113). Besides a second function in sulfur transfer similar to that of the bacterial Ubl, known functions like those for MoaD or ThiS were implicated. Thereby, SAMP1 is essential for the activity of molybdenum cofactor-dependent enzymes, and SAMP2 is required for tRNA thiosylation (112, 113).

The ability to encircle other proteins or to form clamps has been observed several times for small proteins. Examples include SecE (61 aa) in P. furiosus and SecG (53 aa) in M. jannaschii, both subunits of the SecYEG translocation channel, which is involved in protein transport. Whereas SecE is essential and acts as a clamp and opens or closes the transport channel, SecG functions as a plug to keep the channel closed, preventing small molecules passing the channel (114–119). Another small protein involved in transport is CedA2 (52 aa) in Sulfolobus acidocaldarius. This small protein is a component of the Ced system, which is crucial for DNA import. Under UV stress, S. acidocaldarius induces the pilus-producing ups operon, resulting in a mating-pair formation in which the UV-damaged cell imports DNA from a nondamaged cell via the Ced complex (120).

Rubredoxins (45 to 55 aa) are nonheme iron proteins that are involved in the reduction of superoxide in some anaerobic bacteria or act as electron carriers in many biochemical processes (84). Homologues of rubredoxin have been found in different anaerobic archaea, such as P. furiosus, Pyrococcus abyssi, Methanothermobacter thermoautotrophicus, and Thermotoga maritima (121, 122). Although in vitro studies using a reconstituted system have indicated that rubredoxin from P. furiosus is involved in detoxification of reactive oxygen species using a reconstituted system by donating electrons for superoxide reductase (SOR) and rubrerythrin (123), the in vivo function of rubredoxin in P. furiosus is still not clear. A generated rubredoxin deletion strain did not show higher O₂ sensitivity than that of the parental strain, implying that another redox protein is able to directly reduce SOR or flavodiiron protein A (FdpA) (124). P. furiosus rubredoxin is monomeric and might be the most thermostable protein, with a melting temperature of nearly 200°C (33, 121, 125, 126).

Just recently, the first biochemical and detailed functional characterizations of one M. mazei small protein were reported, elucidating the physiological role of the first archaeal small protein (52). The small protein (sP26) of 23 aa, including the noncanonical start (valine), was highly conserved within numerous Methanosarcina strains on the nucleotide and amino acid levels. The transcript levels of small peptide RNA 26 (spRNA26) were increased under nitrogen limitation conditions or under salt or oxygen stress. Under nitrogen limitation, when the 2-oxoglutarat level in the cell is high, glutamine synthetase GlnA₁ forms dodecameric structures, resulting in activation. GlnA₁ activity has been shown to be activated in the presence of sP26, most likely based on a stabilizing effect of sP26 on the dodecameric GlnA₁ or by forming tighter dodecameric GlnA₁ complexes when binding to sP26. A further stabilization occurred by binding of the PII protein GlnK₁ to GlnA₁, mediated by sp26. Due to the fact that homologues to sP26 have been identified in several Methanosarcina strains, it can be assumed that these are involved in nitrogen fixation in a similar way (52).

Small proteins interacting with DNA or proteins in Haloarchaea.

Quite a high number of small proteins have been identified in Haloarchaea. In an early study elucidating the low-molecular-weight proteome (<20 kDa) of Halobacterium salinarum, 45 small proteins were discovered. In this study, the small protein bacteriorhodopsin-regulating zinc finger protein (Brz-1; 60 aa) was identified (see supporting material for reference 31). The respective brz-1 gene, as well as another sORF (brp-1), is located in the intergenic region between the bop and brp operons, encoding the key component bacteriorhodopsin of the photosynthetic system of H. salinarum and a bacteriorhodopsin-related protein, respectively. Brz-1 represents a zinc finger-containing transcription factor and is crucial for expression of the photosynthetic system, including the bop cluster and the crtB1 gene, which encodes the first enzyme in the carotenoid biosynthesis pathway (31, 32).

Another small protein, dodecin (68 aa), has been identified in H. salinarum under stress conditions. Solving the structure by X-ray crystallography showed a riboflavin-binding site provided by three β-sheets and one α-helix (127, 128). Upon binding the riboflavin, dodecin forces a relaxation by dealkylating the riboflavin to 7,8-dimethylalloxazine (lumichrome; fast quenching), which results in reducing the risk for uncontrolled light-induced damage by riboflavin and simultaneously leads to the accumulation of the important precursor. Thus, dodecin buffers the riboflavin concentration inside the cell. When conditions are more favorable, dodecin is downregulated, freeing the riboflavin for flavin adenine dinucleotide (FAD) and flavinmononucleotid biosynthesis (129).

A more recent study focused on 16 genes encoding putative small zinc-finger proteins in H. volcanii (ranging from 50 to 68 aa). Since all respective chromosomal deletion mutants were successfully generated, it is assumed that none of those small proteins is essential. However, 12 mutants showed a clear phenotype related to motility under nonoptimal conditions. Six of those mutants (HVO_0649, HVO_0758, HVO_2142, HVO_2523, HVO_2901, and HVO_A0556) lost their swarming ability, whereas others (HVO_0416, HVO_0758, HVO_2142, HVO_2523, HVO_2901, and HVO_A0556) showed increased biofilm formation (3- to 9-fold increase) (59). Interestingly, the small protein HVO_2753, which is conserved within euryarchaeota, contains two zinc-binding motifs. The respective deletion strain lost its swarming ability and did not form proper biofilms (56).

The H. volcanii small protein HVO_2922 (60 aa) has been shown to be upregulated under iron limitation conditions and downregulated under high salt concentrations, suggesting that it is stress regulated. The NMR solution structure of the protein and mutant derivatives showed that the protein forms a symmetrical dimer that is probably essential for its functionality. HVO_2922 is conserved in Haloarchaea; based on the structure of the dimer this protein belongs to the UPF0339 family with unknown function (57).

Tool development and optimization of genome-wide identification of small proteins in Methanosarcina mazei.

In the methanogenic archeaon M. mazei, several sORFs have been discovered genome-wide using a differential transcriptome sequencing (RNA-seq) approach (34). In this study, 44 RNAs that each contain a putative sORF have been identified and designated small peptide RNAs (spRNAs), and their respective small proteins have been predicted. Several of these spRNAs have been demonstrated to be differentially transcribed in response to nitrogen availability. Today, based on several genome-wide RNA-seq studies of M. mazei grown under various growth and stress conditions (e.g., nitrogen starvation; oxygen, temperature, and salt stresses; and/or biofilm formation), a total of 1,442 small RNAs in M. mazei containing predicted sORFs (<70 aa) have been identified by manual analysis, as well as by automated bioinformatics tools such as ANNOgesic (130). In total, 1,330 of those comprised putative sORFs encoding small proteins ranging from 8 to 49 aa, the length distribution of which is depicted in Fig. 4A. Interestingly, the majority of the predicted small proteins start with a noncanonical, alternative start codon (70.4%) and fewer with the canonical methionine. These identified sORFs in general have to be verified to be translated by using, e.g., proteomics or ribosome profiling approaches. Since experimental verification of respective small proteins is entirely based on earlier identified sORFs, they are often also designated short open reading frame-encoded peptides (SEPs).

FIG 4.

Predicted very small proteins in M. mazei. (A) Bar plot representing the frequency of all predicted sORFs that are less than 50 aa in length (1,330 ORFs), color coded by green for canonical (394 ORFs) and blue for alternative start (936 ORFs). (B) Bar plot representing the frequency of all predicted very small proteins (gray) (1,330 sORFs) based on numerous transcriptome sequencing (RNA-seq) sets and all verified very small proteins (blue), depending on their length in amino acids. Plots were generated using the library ggplot2 in R (132).

The first experimental verification of small proteins from M. mazei was performed in 2015 (2). The soluble proteome was analyzed by LC-MS/MS using the plasma-free separation method to detect seven small proteins under strict criteria (FDR = 1%), experimentally verifying three small proteins predicted by Jäger et al. (34), namely sP44 (23 aa), sP41 (53 aa), and sP36 (61 aa). Based on the significant upregulation of spRNA36 and the respective small protein (sP36) under nitrogen starvation, a function of sP36 in the context of the nitrogen metabolism was proposed (2). In a following semi-top-down approach comparing 2D LC-MS and a gel-free LC approach, eight small proteins were verified with high confidence (FDR = 1%). Not using the strict criteria, an additional six small proteins were verified and identified with medium confidence (FDR = 5%), including sP26 (38). Within the scope of method optimization, an acetonitrile-based depletion method for the high-molecular-weight proteome (<15 kDa) was established and evaluated. Using this approach, three additional so far unknown small proteins were verified. While the depletion method removed approximately 70% of the high-molecular-weight proteome, it did not result in a significant higher detection rate for small proteins. However, depletion of the high-molecular-weight proteome resulted in significantly improved identification metrics for those small proteins that were detected (40). To evaluate the impact of the gel staining methods in a GeLC-MS, in which an SDS-PAGE is performed prior to LC-MS/MS, the low-molecular-weight proteome of M. mazei grown under nitrogen starvation conditions was analyzed. Within this study, an additional 24 predicted small proteins, ranging from 8 to 69 aa, were verified under strict criteria (39). In another recent study, a multidimensional top-down analysis was performed that enabled the detection of several small proteins and provided evidence for a number of posttranslational modifications (37). The utilization of a solid-phase extraction methodology, followed by strong cation exchange and then either top-down or bottom-up LC-MS allowed detection of 36 proteoforms belonging to 12 small proteins. The major advantage of this approach was the ability to identified intact proteoforms, allowing for the identification of N and C termini, as well as a number of biological posttranslational modifications (e.g., disulfide bridges). Importantly, this workflow enabled the characterization of five small proteins, of which one was newly verified. The extensive analysis also uncovered several interesting characteristics of the verified small proteins, e.g., alternate translation initiation sites, and argued for posttranslational processing of several small proteins (N- or C-terminal truncations). Interestingly, this approach also revealed the existence of N-terminal formylation (of small proteins) in archaea, which was not previously known, and thus further illustrates the diversity of modifications of small proteins (37) Very recently, a systematic evaluation of a multiprotease approach to improve both the sequence coverage and number of identified peptides for small proteins was performed. This comprehensive approach resulted in the identification of 91 proteins with at least two unique peptides and clearly improved the identification and molecular characterization of small proteins (46). Overall, with the different optimization steps described above, 58 small proteins smaller than 51 aa have been identified and experimentally verified with high confidence (FDR = 1%). The length distribution of these is depicted in Fig. 4B. In agreement with the predictions, the majority of them (70.4%) show a noncanonical start. A recent study concerning the average number of sORFs per genome that analyzed 109 bacterial genomes showed that approximately 16% ± 9% of total coding ORFs might be coding for small proteins (21). If this tendency is applicable for Archaea, one can predict 263 to 713 sORFs for M. mazei (compare to 3,490 large ORFs identified in M. mazei [131]).

CONCLUSION AND OUTLOOK

Most small proteins are single discoveries and only recently has the hidden world of small proteins present in archaeal genomes been recognized using genome-wide approaches. First biochemical characterizations indicate that the majority of identified small proteins have specific functions. As has been shown in bacteria, small archaeal proteins are often involved in survival under stressful conditions, under which it is important to adapt to various degrees. The establishment of effective identification and verification tools for small proteins in archaea (as well as in bacteria) has advanced in recent years, often through the combination of genome-wide approaches (RNA-seq, RiboSeq, and peptidomics). Besides, combining bottom-up and top-down proteomics enables identification of posttranslational modifications. Moreover, a better understanding of the molecular mechanisms involved in sORF evolution, their evolutionary rates, and selection pressure on codon usage will help to identify sORFs using computational tools. Consequently, future studies of the small proteome in archaea and the elucidation of their biological functions will provide novel insights into the full repertoire and function of the archaeal small proteome and substantially complement our understanding of small proteins in archaea. Accordingly, unveiling this new layer of biological information in archaea as well as in bacteria is challenging, but it has an enormous potential to discover surprising and unexpected results and thus might trigger a major change in our current thinking.

Biographies

Katrin Weidenbach performed her Ph.D. at Göttingen University (Germany). She is a senior postdoctoral researcher in the group of Ruth Schmitz, Institute for General Microbiology (IFAM) at Kiel University (Germany), focusing on small proteins in archaea.

Miriam Gutt is currently performing her Ph.D. on small archaeal proteins in the group of Ruth Schmitz.

Liam Cassidy acquired a Ph.D. from the Australian National University and is now a postdoctoral researcher in the Systematic Proteomics Workgroup of Andreas Tholey at the University of Kiel.

Cynthia Chibani performed her Ph.D. at Göttingen University. She is currently a postdoctoral researcher in the group of Ruth Schmitz, IFAM at Kiel University, mainly focusing on bioinformatic analysis of metagenomes and viromes.

Ruth A. Schmitz performed her Ph.D. in biochemistry and microbiology at Marburg University, worked as a postdoctoral researcher at the University of California, Berkeley (USA), and habilitated in microbiology and genetics at Göttingen University. Since 2004, she heads the IFAM at Kiel University.