Structure and Function of Cold Shock Proteins in Archaea

PCR amplification, cloning, and sequencing. M. frigidum csp was amplified from an M. frigidum genomic library (45). csp genes were identified from available genome sequence data on H. marismortui (4), H. lacusprofundi (20), and Halobacterium sp. strain NRC-1 (12) and amplified from genomic DNA. “C. symbiosum” csp and uncultured Crenarchaeota csp genes were amplified from fosmids C7E8 and 4B7 (3). The M. burtonii CSD gene was amplified from M. burtonii genomic DNA. PCRs were carried out in a total volume of 50 μl containing 2.5 mM MgCl₂, 0.2 mM deoxynucleoside triphosphate, 2.5 U Taq DNA polymerase (Sigma), 1× Taq DNA polymerase buffer, 20 pmol of forward and reverse primers (sequences are available upon request), and approximately 50 ng of DNA templates. Forward primers were designed to introduce an NdeI restriction site overlapping the ATG initiation codon. Reverse primers were designed to generate either a BamHI restriction site for cloning into green fluorescent protein (57) and pET-15b (Novagen) vectors or a SapI restriction site for cloning M. frigidum csp into pTYB1 (New England BioLabs). PCRs were carried out in a Hybaid Thermal Cycle PCR machine for 30 cycles (94°C for 30 s, 55°C for 60 s, and 72°C for 90 s) after initial denaturation at 96°C for 60 s. The amplified products were gel purified with a Prep-A-gene kit (Bio-Rad) and ligated into a pGEM-T cloning vector (Promega) according to the manufacturer's instructions. Recombinant plasmids harboring the csp genes and M. burtonii CSD were double digested with NdeI and BamHI and cloned between the same two sites of a green fluorescent protein vector. The new constructs were digested with XbaI and BamHI, and csp genes were subcloned into a modified pIN-III expression vector, pIN-III-lpp^P-5. M. frigidum csp was also cloned into pTYB1 and pET-15b expression vectors between the NdeI/SapI and NdeI/BamHI restriction sites, respectively. In the former, M. frigidum csp was fused to the N terminus of an intein-chitin binding domain tag, whereas in the latter M. frigidum csp was fused to the C terminus of a His tag. All recombinant plasmids were transformed into chemically competent E. coli strain DH5α cells. Constructs were verified by DNA sequencing (ABI 377 DNA sequencer, Automated DNA Analysis Facility, The University of New South Wales).

Complementation testing and expression analysis of archaeal genes in E. coli BX04.pIN-III-lpp ^P-5 plasmids harboring the archaeal csp genes and M. burtonii CSD were transformed into E. coli strain BX04 (63). Cells were grown in LB medium supplemented with ampicillin (100 μg ml⁻¹) at 37°C to an optical density at 600 nm of 0.6. Gene expression was induced by addition of 1, 2, 10, or 70 mM isopropyl-β-d-thiogalactopyranoside (IPTG), and cultures were incubated at 15°C for 16 h. Cells were harvested at 4°C and resuspended in 20 mM Tris HCl (pH 7.5)-200 mM NaCl-1 mM EDTA-0.5 to 1 mg of lysozyme. Cells were lysed on ice by sonication with a Branson Sonifier for four to six cycles of 30 s on a 35% duty cycle at a power setting of 3. Soluble and insoluble fractions were separated by centrifugation at 12,000 rpm at 4°C and analyzed by 16% (wt/vol) sodium dodecyl sulfate (SDS)-gel electrophoresis. To assess complementation ability, single colonies of recombinant BX04 cells were inoculated into LB medium and grown at 37°C for 4 to 5 h. Cultures (2 μl) were streaked onto LB agar plates supplemented with ampicillin (100 μg ml⁻¹) or with ampicillin plus IPTG (1, 2, 10, or 70 mM), and plates were incubated at 4, 10, 15, 23, and 30°C. Plates were incubated for various lengths of time (depending on the temperature), and growth of BX04 with a plasmid backbone only (control) was compared with that of strain BX04 harboring the archaeal csp genes and M. burtonii CSD.

Overexpression and purification of M. frigidum Csp. M. frigidum csp cloned into pTYB1 was transformed into E. coli strain BL21(DE3) [F⁻ ompT hsdS _B (r_B ⁻ m_B ⁻) gal dcm(DE3)]. Cells were grown at 37°C in LB medium supplemented with ampicillin (100 μg ml⁻¹) to an optical density at 600 nm of 0.5 to 0.6. Gene expression was induced by the addition of 1 mM IPTG, and the culture was incubated at 15°C for approximately 16 h. Cells were harvested by centrifugation at 4°C at 5,000 × g for 30 min and resuspended in 1/15 culture volume of lysis buffer (20 mM Tris HCl [pH 7.5], 1 mM EDTA). Cells were lysed by sonication as described above. Soluble and insoluble fractions were separated by centrifugation at 12,000 × g at 4°C and analyzed by SDS-gel electrophoresis. M. frigidum Csp was purified on a chitin column by a procedure that was described previously (55), with minor modifications. The supernatant was loaded onto a chitin bead column (1/50 culture volume) preequilibrated with 10 bed volumes of column buffer (20 mM Tris HCl, 1 mM EDTA, 200 mM NaCl). The column was washed with 20 bed volumes of column buffer with a linear gradient of NaCl (200 mM → 950 mM → 200 mM). The cleavage of the fusion protein was induced by flushing the column with 3 bed volumes of column buffer containing 70 mM dithiothreitol (DTT). The reaction mixture was incubated at 22°C for 64 h, and the protein was eluted in column buffer. M. frigidum Csp concentration was determined by Bradford assay (5), and amino acid analysis was performed at the Australian Proteome Analysis Facility, Macquarie University, Sydney, Australia. Protein purity was determined by visualization of protein bands on SDS-gels.

TUG-GE.Transverse urea gradient gel electrophoresis (TUG-GE) was performed as described previously (10, 53). Gels were prepared with a urea gradient of 0 to 7 M and an inverse acrylamide gradient of 15 to 11%. The bottom solution (35 ml) consisted of 0.375 M Tris-HCl (pH 8.8), 7 M urea, 11% acrylamide solution (40% acrylamide, 3.3% bisacrylamide), 0.026% (vol/vol) N,N,N′,N′-tetraethylenediamine (TEMED), and 0.015% (wt/vol) ammonium persulfate. Gradient solutions (57.5 ml each) consisted of 0.375 M Tris-HCl (pH 8.8), 7 M or 0 M urea (gradient solution with or without urea, respectively), 11% or 15% acrylamide solution (gradient solution with or without urea, respectively), 0.026% (vol/vol) TEMED, and 0.013% (wt/vol) ammonium persulfate. The top solution (36 ml) consisted of 0.375 M Tris-HCl (pH 8.8), 0 M urea, 15% acrylamide solution, 0.026% (vol/vol) TEMED, and 0.013% (wt/vol) ammonium persulfate. M. frigidum Csp samples (total volume, 75 μl) contained 30 μg of purified protein, 50 mM Tris HCl (pH 8.8), 0.005% bromophenol blue, and either 10% glycerol (N⇆U transition) or 7 M urea (U⇆N transition). The latter sample was preincubated for 2 h at 7°C before commencing electrophoresis. Gels were assembled in a Hoefer SE-250 electrophoresis unit (Bio-Rad) connected to a temperature-controlled water bath (MultiTemp III; Pharmacia Biotech) set at 1°C and prerun at 10 mA for 30 min. The internal temperature was 7 to 10°C. Electrophoresis was performed in 0.3% (wt/vol) Tris-1.9% (wt/vol) glycine buffer (pH 8.3) at 10 mA for approximately 3 h. Gels were washed in Milli-Q water and stained with silver, and gel images were collected on a LAS3000 (Fujifilm, Melbourne, Australia) with ImageGauge v4.0 software. The conformational stability of M. frigidum Csp (ΔG [stability of the native conformation relative to that of the unfolded state] and [urea]_1/2 [urea concentration at ΔG = 0]) was calculated as described previously (10, 18, 53).

Matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS).All samples were purified with C₁₈ ZipTips (Millipore) and mixed with a matrix solution prepared as follows. 3-Hydroxypicolinic acid (3-HPA) was dissolved in 50% CH₃CN (50 mg/ml) and mixed 9:1 with a solution of ammonium citrate (50 mg/ml in Milli-Q water). Mixtures were spotted onto a steel MALDI target and allowed to air dry. MALDI-TOF mass spectra were acquired in linear mode on a Voyager DE STR mass spectrometer (PE Biosystems) with manual acquisition control (200 laser shots/spectrum), a 25-kV accelerating voltage, a 94.2% grid voltage, and a 125-ns extraction delay time. By this approach, a mass error of about ±2 Da was observed, allowing easy discrimination of most of the ribonucleotide base residue masses, except cytidine and uridine, the residue masses of which are 305.18 Da and 306.17 Da, respectively.

Homology modeling.Models were generated with the protein homology modeling server SWISS-MODEL, and models were manipulated with the Swiss-Pdb Viewer.

Archaeal Csp models were predicted to form a β-barrel structure homologous to the bacterial Csp structures. Similar models for the M. burtonii CSD were generated previously (45).

Selection of archaeal Csp homologs.Five Csp sequences from H. marismortui with high identity (78 to 95%) to each other and four from H. lacusprofundi (81 to 89%) were identified (Table 1). The H. marismortui gene with the highest similarity to other H. marismortui genes (H. marismortui contig 380453) was chosen for complementation studies. The H. lacusprofundi genes with the greatest (H. lacusprofundi contig 846) and least (H. lacusprofundi contig 1547) similarity to other H. lacusprofundi csp genes were chosen for complementation. The two csp genes from Halobacterium sp. strain NRC-1 (genes 101 and 1836), one csp gene from uncultured Crenarchaeota, one gene from “C. symbiosum”, and M. frigidum csp were chosen, providing a total of eight archaeal csp genes. Additionally, an M. frigidum csp construct with a His tag fused to the N terminus (M. frigidum csp-His) of the protein was included in the complementation assays. One of the two M. burtonii hypothetical proteins with a predicted CSD fold (M. burtonii CSD) was also chosen for complementation.

The similarity between the eight archaeal Csps and E. coli CspA varies between 36 and 59% amino acid identity (Table 2). M. frigidum Csp has higher similarity to E. coli CspA (59%) than to any of the other archaeal proteins (36 to 44%). The five haloarchaeal proteins are most similar to each other (68 to 91%), and the two Csps from Crenarchaeota are most similar to each other (73%). The M. burtonii CSD has very low sequence identity (5 to 7%) to any of the Csps.

Complementation of cold sensitivity in E. coli by archaeal genes.The archaeal genes were cloned into IPTG-inducible, high-level expression plasmid pIN-III-lpp ^P-5 in E. coli BX04, and growth was assessed on plates incubated at 30, 23, and 15°C. In the presence or absence of IPTG, BX04 pIN-III-lpp ^P-5 only formed colonies at 30°C (Fig. 1). In contrast, the strain harboring M. frigidum csp formed colonies at 23 and 15°C (Fig. 1). Similar patterns of complementation were achieved with IPTG concentrations of 1, 2, 10, and 70 mM (data not shown), and complementation only occurred in the presence of inducer (Fig. 1). When plates were incubated at 10 and 4°C, colonies did not form (data not shown). The fact that the M. frigidum csp complements the growth defect of the quadruple-deletion strain and suppresses its cold sensitivity at 15 and 23°C demonstrates that this protein from an archaeal psychrophile is biologically active and able to fulfill a critical functional role in a mesophilic bacterial host.

• Open in new tab
• Download powerpoint

FIG. 1.

Complementation of cold sensitivity in E. coli BX04 cells expressing M. frigidum csp. The pIN-csp plasmid and the pIN vector (control) were transformed into cold-sensitive strain BX04 and streaked onto LB plates containing ampicillin with (A, C, E) and without (B, D) 2 mM IPTG, and the plates were incubated at 30°C (A), 23°C (B, C), or 15°C (D, E).

The ability of other archaeal genes to complement the cold sensitivity of BX04 was assessed at 15°C (data not shown). The uncultured Crenarchaeota csp complemented BX04 at 15°C in both the presence and the absence of IPTG. Similarly, when M. frigidum csp-His was cloned into a pIN vector and expressed in BX04, growth occurred in both the presence and the absence of IPTG. The ability to complement in the absence of inducer may reflect the ability of pIN vectors to allow basal expression of cloned genes even in the absence of inducer (63). It is not clear whether the different complementation patterns for uninduced BX04 with M. frigidum csp-His compared to M. frigidum csp relates to differential gene expression levels, stability of the proteins, or functionality mediated by the His tag. However, the fact that M. frigidum csp-His is functional in vivo encourages its use for future in vitro studies.

The M. burtonii CSD construct enabled colonies to form at 15°C only when plates did not contain IPTG (data not shown). This may indicate that the M. burtonii CSD functions effectively in E. coli when present at a low concentration in the cell, whereas at higher concentrations it prevents cell growth.

None of the haloarchaeal csp genes or the “C. symbiosum” csp gene suppressed the cold sensitivity of BX04. In this regard, it was noteworthy that at 30°C all of the strains were able to form colonies (data not shown), indicating that none of the recombinant plasmids were inherently toxic to the cell. In order to determine if all of the csp genes were expressed in BX04, cells were grown in liquid culture at 37°C and induced with 2 mM IPTG, the temperature was decreased to 15°C, and soluble and insoluble fractions were analyzed by SDS-gel electrophoresis. Bands in the soluble fraction consistent with the predicted molecular weights of the Csps were observed for all of the archaeal genes except the two Halobacterium sp. strain NRC-1 genes and one of the H. lacusprofundi genes (data not shown). A band of the appropriate molecular weight was also apparent for the M. burtonii CSD and M. frigidum Csp-His (data not shown). All of the proteins visualized were in the soluble fraction, and none appeared to accumulate in the insoluble fraction (data not shown).

The inability of the haloarchaeal csp genes to complement in BX04 may indicate that they perform a function(s) in the cell that is not compatible with restoring the E. coli growth defect. Alternatively, the haloarchaeal Csps may have a requirement for a higher intracellular salt concentration than is present in E. coli. For example, a DNA ligase from H. volcanii, LigN, that was purified as a recombinant protein from E. coli was inactive in the absence of KCl and displayed increasing activity at KCl concentrations of up to 3.2 M (41).

Structural features predicted to be important for archaeal Csp function in E. coli.In order to identify structural features of the proteins that may explain the complementation data, analyses were performed on the amino acid sequences, predicted secondary structures, and tertiary structures generated by homology modeling. In particular, the aromatic and basic residues were examined as they are known to be involved in binding to nucleic acids (25, 51).

The aromatic residues are generally conserved among all of the Csps (Fig. 2), with three of the seven aromatic residues on the nucleic acid binding surface, F12, F20, and F31 (amino acid numbering based on E. coli CspA) (16), being identical in all nine sequences. In some positions, the type of aromatic residue varies; e.g., F18 is either phenylalanine or tyrosine. On the nucleic acid binding surface, E. coli CspA has five basic and four acidic residues, M. frigidum Csp and uncultured Crenarchaeota Csp have six basic and five acidic residues, “C. symbiosum” Csp has seven basic and three acidic residues, and the halophilic proteins have two basic and seven or eight acidic residues. The ratio of basic to acidic residues is higher in the proteins that complement (M. frigidum Csp and uncultured Crenarchaeota Csp) compared to those that do not complement (“C. symbiosum” Csp and the haloarchaeal proteins), with the ratio being especially low in the halophilic proteins. These data indicate that the main differences in the Csp sequences relate to the content and distribution of positively charged amino acids. It is noteworthy that the high content of acidic residues in the Csps from halophiles is consistent with the general overrepresentation of acidic residues as an adaptation to growth in highly saline environments (28).

• Open in new tab
• Download powerpoint

FIG. 2.

Primary-structure analysis of Csp homologs. Amino acid sequence alignments of Csp homologs from E. coli (CspA), M. frigidum, uncultured Crenarchaeota, “C. symbiosum,” H. marismortui, Halobacterium sp. strain NRC-1 (genes 101 and 1836), and H. lacusprofundi (genes 846 and 1547). Aromatic residues, negatively charged residues, and positively charged residues are highlighted in yellow, blue, and red, respectively. β-Strands in E. coli CspA are indicated by bars, and the residues composing the RNA binding motifs, RNP1 and RNP2, are boxed. The lengths of the proteins are shown on the right in amino acids.

Secondary-structure predictions identified five β-strands in all of the Csps except that of Halobacterium sp. strain NRC-1 (encoded by gene 101), which had an additional β-strand predicted at the C-terminal end of the protein (data not shown). No three-dimensional structures are available for archaeal Csps. However, the three-dimensional structure of E. coli CspA has been resolved by X-ray crystallography (46) and nuclear magnetic resonance spectroscopy (16, 38). The nucleic acid binding surface is composed of residues 9 to 13 (on β1), 14 to 17 (on L1), 18 to 21 (on β2), 30 to 33 (on β3), 34 to 49 (on L3), and 58 to 61 (on L4) (16). The homology models of the eight archaeal Csps displayed the characteristic β-barrel structure, with β1, β2, and β3 strands opposing the β4 and β5 strands (Fig. 3). E. coli CspA contains eight basic residues, seven of which are accessible, and of these seven, four are located on the nucleic acid binding side of the barrel (Fig. 3A, strands β1 to β3 and loops L1, L3, and L4). M. frigidum Csp also contains eight basic amino acids, four of which are exposed on the binding side of the barrel (Fig. 3B). The two Csps from the Crenarchaeota have even higher numbers of positive residues (12 and 11), but only 4 of them appear accessibly located on the binding surface of the barrel (Fig. 3C and D). The Csp homologs from the haloarchaea possess between four and six basic residues, only two or three of which are accessible, and only one of these is located on the β1-to-β3 surface (Fig. 3E to I).

• Open in new tab
• Download powerpoint

FIG. 3.

Three-dimensional structure models of Csp homologs. The nuclear magnetic resonance structure of E. coli CspA (A) and homology models of M. frigidum Csp (B), uncultured Crenarchaeota Csp (C), “C. symbiosum” Csp (D), and Csps from H. marismortui (E), Halobacterium sp. strain NRC-1 (genes 101 and 1836) (F and G), H. lacusprofundi (genes 846 and 1547) (H and I), and M. burtonii CSD (J) are shown. Positively charged residues exposed on the surface are in red writing. E. coli Csp β-sheets 1 to 5 are in black writing. E. coli Csp loops 1 to 4 are in white writing. M. burtonii CSD cysteine residues typical of Zn ribbons are in green writing. The models are presented alphabetically from left to right, with A in the top left corner and J at the bottom.

The homology model of the M. burtonii CSD has three domains (Fig. 3K). The central domain consists of five strands organized in a β-barrel structure very similar to that of the Csps. This domain contains 11 basic amino acids, 5 of which are exposed on the surface that corresponds to the binding surface of Csps.

The four proteins that were able to suppress the cold sensitivity of E. coli BX04 exhibited the highest number of accessible positive charges on the binding surface of their barrel structures (Table 3). Residues K10, K16, and K60 are located on β1, L1, and L4, respectively, and are conserved in all three Csp homologs. The remaining solvent-exposed residues of E. coli CspA, M. frigidum Csp, and uncultured Crenarchaeota Csp belong to loop L3. These common structural features may be involved in the binding of E. coli nucleic acid.

Biochemical analysis of M. frigidum Csp. M. frigidum Csp was purified as a recombinant protein free of an affinity tag with the cleavable intein-chitin binding system (see Materials and Methods). Following expression in E. coli BL21(DE3), 5 mg liter⁻¹ pure M. frigidum Csp was yielded (determined by Bradford assay and amino acid analysis) (Fig. 4). The molecular mass of the purified protein determined by MALDI-TOF MS was 7,814.6 Da (data not shown), consistent with the theoretical mass of 7,815.6 Da.

• Open in new tab
• Download powerpoint

FIG. 4.

Expression and purification of M. frigidum Csp. SDS-gel electrophoresis analysis of the purification of M. frigidum Csp by affinity chromatography on chitin beads. M. frigidum Csp was purified from crude lysate of BL21(DE3) cells expressing the M. frigidum Csp-intein-CBD fusion. Shown are the insoluble fraction (lane 1), the soluble fraction in sample buffer without DTT (lane 2), the soluble fraction in sample buffer with 70 mM DTT (lane 3), the purified M. frigidum Csp (lane 4), and a broad-range protein molecular size standard (lane MW; Bio-Rad) corresponding to 224, 122, 90, 51.5, 35.3, 28.7, 21, and 7.2 kDa.

The folding kinetics of M. frigidum Csp were assessed by monitoring its transition curve in the presence of a transverse urea gradient as previously described (10, 53). Unfolding (N⇆U) and refolding (U⇆N) transitions of M. frigidum Csp were both examined by TUG-GE (Fig. 5). Multiple bands indicative of protein variants were revealed. The low-intensity upper bands arose from a translation product that was initiated from an internal methionine (residue 6) that produced an unfolded protein (data not shown). This was determined by overexpressing and purifying an M. frigidum Csp hybrid that initiated translation at Met6, visualizing the product by TUG-GE, and analyzing the molecular masses of all of the bands produced by the full-length and hybrid M. frigidum Csps from TUG gels (L. Giaquinto and R. Cavicchioli, unpublished results).

• Open in new tab
• Download powerpoint

FIG. 5.

TUG-GE of M. frigidum Csp. Unfolding (N⇆U) (A) and refolding (U⇆N) (B) transition curves of M. frigidum Csp in a TUG (0 to 7 M) perpendicular to the direction of electrophoresis is shown. The urea gradient is counterbalanced by an inverse acrylamide gradient (15 to 11%). The free energy of unfolding at 0 M urea was calculated by graphically extrapolating from the transition region, and the concentration of urea at ΔG = 0 is [urea]_1/2. F_u, fraction of unfolded molecules; ΔG, free-energy difference between folded and unfolded states; [urea]_1/2, concentration of urea at equilibrium; R, universal gas constant (8.314 kJ mol⁻¹); T, absolute temperature.

For the high-intensity lower bands that correspond to the full-length M. frigidum Csp, a single transition was observed between the initial (0 M urea) and final (7 M urea) states. Identical transition profiles were observed for the N⇆U and U⇆N processes. These results demonstrate that M. frigidum Csp is a single-domain protein that folds by a reversible two-state mechanism. The presence of two bands at high urea concentrations (when the protein is fully unfolded) is indicative of the presence of two structural variants of the full-length M. frigidum Csp. These variants may arise as a result of posttranslational modification of the M. frigidum Csp in E. coli.

To assess the conformational stability of M. frigidum Csp, ΔG and [urea]_1/2 were calculated at 7°C for the bands corresponding to the full-length M. frigidum Csp (Fig. 5). The values calculated were as follows: ΔG, 11.4 kJ mol⁻¹; [urea]_1/2, 3.2 M. The folding state of M. frigidum Csp was further assessed by monitoring changes in its secondary structure in the far-UV region (180 to 250 nm) by CD spectroscopy, and after thermal denaturation at 60°C, M. frigidum Csp exhibited characteristics typical of nonnative molecules (data not shown) (47). These properties are consistent with the protein being from a psychrophile and being thermolabile.

The UV absorption spectrum of M. frigidum Csp showed a maximum at 260 nm and an A ₂₈₀/A ₂₆₀ ratio of 0.77 (data not shown). The strong signal at 260 nm is likely to indicate the presence of nucleic acid in the protein sample. To determine if E. coli RNA was bound by M. frigidum Csp, the protein was analyzed by MALDI-TOF MS with a 3-HPA matrix. Initial analysis in the absence of subtilisin (data not shown) revealed a few peaks of very weak signal intensity in the 900- to 5,500-Da mass range. However, when an analysis was performed with M. frigidum Csp samples to which either subtilisin (Fig. 6A) or both subtilisin and an RNase inhibitor (data not shown) had been added, the spectra showed a systematic progression of peaks differing by the mass of ribonucleotide residue mass units. In particular, a mass difference of 347 Da, close to 345 Da was observed, consistent with the residue mass of guanosine. This is 16 Da higher in mass than the residue mass of deoxyguanosine and indicates that the series corresponds to RNA rather than DNA. The MALDI-TOF spectrum of M. frigidum Csp hydrolyzed with subtilisin in the presence of an RNase inhibitor (data not shown) produced a similar mass distribution, indicating that it was unlikely that RNases were active during sample handling for MALDI. The combined addition of subtilisin and RNase A to M. frigidum Csp produced MALDI signals with a higher intensity and in a lower mass range (Fig. 6B), indicating that RNase A was able to more extensively hydrolyze the target RNA when the M. frigidum Csp was removed by subtilisin hydrolysis. This suggests that an in-solution association between M. frigidum Csp and RNA not only survives the extraction-and-purification procedure but also offers protection from RNase degradation in the vicinity of the binding site. The series of peaks observed in the MALDI-TOF spectra most probably reflect variable endogenous degradation of the exposed RNA end. Since RNase A is specific for single-stranded RNA, our data indicate that single-stranded RNA was specifically associated with M. frigidum Csp sample preparation.

• Open in new tab
• Download powerpoint

FIG. 6.

Mass spectra of M. frigidum Csp. Shown are the positive-ion MALDI-TOF mass spectra of M. frigidum Csp incubated in subtilisin (A) and M. frigidum Csp incubated in subtilisin and RNase A (B) determined by C₁₈ ZipTip extraction and a 3-HPA matrix. The ribonucleotide bases corresponding to the mass difference between consecutive peaks are indicated. The average residue mass for each of the ribonucleotide bases is as follows: adenosine, 329.21; cytidine, 305.18; uridine, 306.17; guanosine, 345.21.

Collectively, these data may indicate that an M. frigidum Csp-RNA complex protects the nucleic acid from degradation, since (i) the RNA series is easily observed only when the protein is removed by subtilisin hydrolysis and (ii) more extensive hydrolysis of the RNA is observed when RNase A is added in addition to subtilisin.

Characteristics of archaeal csp and CSD complementation. M. frigidum csp complemented BX04 in the presence of IPTG (Fig. 1), whereas uncultured Crenarchaeota csp was able to complement the growth defect with and without IPTG (data not shown). This is consistent with a previous study with E. coli csp genes which found that cspA, cspE, and cspF were able to complement only in the presence of IPTG and cspG and cspI were able to do so with and without IPTG (63). The level of amino acid sequence similarity does not explain our complementation results, as the uncultured Crenarchaeota Csp has lower primary sequence identity to E. coli CspA (39%) than M. frigidum Csp has to E. coli CspA (59%) (Table 2). However, the three proteins able to complement the growth defect of BX04, M. frigidum Csp, uncultured Crenarchaeota Csp, and M. burtonii CSD, do have particular structural features identifiable in their modeled tertiary structures in common with E. coli CspA that are not features of the other archaeal Csps. In E. coli CspA (25) and Bacillus subtilis CspB (51), a cluster of aromatic or basic amino acids on the β-barrel surface plays a key role in the interaction of the protein with nucleic acids. Positive charges attract the negatively charged nucleic acid by electrostatic interaction, and the aromatic patches bind and melt nucleic acid secondary structure to facilitate transcription and translation at low temperatures (40). In the archaeal Csps, while the aromatic cluster of residues is largely conserved in all of the proteins, the number and location of the acidic and basic residues vary among the Csp homologs and the M. burtonii CSD. The M. frigidum Csp, uncultured Crenarchaeota Csp, and M. burtonii CSD have a higher content of solvent-exposed basic residues (Table 3), and these are located on the nucleic acid binding surface of their β-barrel structures (Fig. 3B, C, and J), similar to the number and location of solvent-exposed basic residues for E. coli CspA (Fig. 3A). These structural features are the main characteristics that can be identified that explain the complementation results.

The phylogenetic relationship of the archaeal Csp homologs with E. coli CspA reveals that E. coli CspA, the M. frigidum Csp, and the two Crenarchaeota Csps form a cluster distinct from the other Csps (Fig. 7). The “C. symbiosum” Csp clusters with the complementing genes. As the “C. symbiosum” Csp also has four accessible positive charges predicted on the binding surface of the protein (Fig. 3 and Table 3), it may be expected to complement BX04. However, the ratio of basic to acidic residues on the nucleic acid binding surface is considerably higher for the “C. symbiosum” Csp (7:3) compared to the M. frigidum Csp and the uncultured Crenarchaeota Csp (6:5) and the E. coli Csp (5:4), and this charge balance for the “C. symbiosum” Csp may adversely affect its interactions with nucleic acid targets involved in complementation. We also noted that although it is not apparent in the E. coli Csp three-dimensional structure, between strands β3 and β4, a short α-helix region was predicted in the secondary-structure plot of the E. coli Csp, M. frigidum Csp, and uncultured Crenarchaeota Csp that is not predicted in the “C. symbiosum” Csp (data not shown).

• Open in new tab
• Download powerpoint

FIG. 7.

Phylogenetic tree of Csp homologs. The tree was constructed with the PHYLIP server.

M. burtonii CSD.In addition to the cold sensitivity of the E. coli quadruple-deletion mutant BX04 (63), the deletion of two of the three csp genes in B. subtilis causes a severe reduction in growth ability (23). The growth defects can be complemented by overexpressing a specific set of proteins that includes homologous Csps, or the S1 domain of PNPase in E. coli, or initiation factor IF1 in B. subtilis. The primary sequences of both the S1 domain and IF1 have little similarity to those of Csps, but their three-dimensional structures are very similar and functional overlap between these proteins has been proposed (59, 63). The M. burtonii CSD is the first example of an archaeal protein that fulfills an functional role analogous to that of the E. coli S1 domain and the B. subtilis IF1 domain.

The M. burtonii CSD is one of two M. burtonii CSD proteins originally identified in a comparative genomic analysis as hypothetical proteins with structural features similar to those of Csps (45). Subsequent proteomic analysis found that both proteins were synthesized in cells growing at 4°C, thereby demonstrating that they are not only functional but important for low-temperature growth (21). Both proteins are encoded within a region of archaeal genomes that was previously described as a superoperon involved in RNA and protein processing, with their predicted RNA binding fold lending support for their role in RNA processing in M. burtonii (21, 30). The C-terminal domain of M. burtonii CSD has characteristics of Zn ribbon proteins, containing two pairs of closely spaced cysteine residues separated by beta strands (Fig. 3J). Archaeal proteins that contain Zn ribbons are likely to be involved in nucleic acid binding and particularly transcription (1), providing further support for a putative role in RNA processing. The present study extends our knowledge of the M. burtonii proteins from the superoperon by demonstrating that the M. burtonii CSD can complement csp deletions that cause cold sensitivity in E. coli.

The M. frigidum Csp.Biochemical analysis of the M. frigidum Csp demonstrate that it is a single-domain protein that folds by a reversible two-state mechanism (Fig. 5), consistent with the folding kinetics described for bacterial Csp homologs (33, 39, 43, 47, 48, 49). The ΔG of 11.36 kJ mol⁻¹ and the [urea]_1/2 of 3.2 M calculated at 7°C are lower than the ΔG and [urea]_1/2 values of Csp homologs from E. coli (43) and B. subtilis (50) that were calculated by fluorescence at pH 7 and 25°C. The lower conformational stability of the M. frigidum Csp is likely to reflect the increased flexibility and decreased stability of the protein, consistent with adaptations typical of cold-adapted proteins (52).

The ability of M. frigidum csp to complement function in vivo is consistent with the ability of the M. frigidum Csp to bind E. coli RNA. Evidence for the latter came from UV absorption spectroscopy and MALDI-TOF MS analyses (Fig. 6). Moreover, the specificity of RNase A for single-stranded RNA indicates that M. frigidum Csp binds single-stranded RNA. The M. frigidum Csp-RNA association appears to offer some protection from RNase degradation, since the RNA series is observed only when the protein is hydrolyzed by subtilisin, and removal of the protein by subtilisin hydrolysis allows more extensive degradation of the RNA by RNase A (Fig. 6).

M. frigidum is incapable of growth above ∼18°C and has proven difficult to culture in the laboratory, and few studies examining the organism's growth properties or probing mechanisms of adaptation have been reported (17, 36). The ability to perform heterologous protein studies provides opportunities for examining the molecular characteristics that distinguish cold-adapted archaea that possess a restricted growth temperature range (stenopsychrophiles, e.g., M. frigidum) from eurypsychrophilic archaea (e.g., M. burtonii) that can tolerate a wide range of temperatures extending into the mesophilic range (8), a question about thermal adaptation that is of broad relevance to all microorganisms that live in cold environments (15).