INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The drying of cells leads to crowding of cytoplasmic components, condensation of the nucleoid, and increases in the melting temperature (T_m) of membrane phase transitions, and it imposes stresses upon cell walls (31). Prolonged desiccation may cause damage to proteins, nucleic acids, and membrane components through Maillard reactions and metal-dependent Fenton reactions (17, 31). High photon flux densities and UV irradiation exacerbate the effects of reactive oxygen species in these processes. Mechanisms that protect cells from the effects of desiccation include those which circumvent oxidative damage (5, 7, 16). Of these, the synthesis of superoxide dismutase (SOD) is an important response. SOD mediates the disproportionation of superoxide radicals to hydrogen peroxide and dioxygen; the hydrogen peroxide is then scavenged either by ascorbate peroxidase or catalase (27). In many bacteria, SODs are present in the periplasmic space and their synthesis is repressed under anaerobic conditions (3, 15). Regulation of the expression of genes that encode these periplasmic SODs, as well as genes involved in the synthesis of other antioxidants, may be subject to control by the alternate sigma factor RpoS (13, 15). Other mechanisms that cells use to protect against oxidative damage from desiccation include the synthesis of carotenoids and tocopherols (12, 38, 45) and anthocyanins (in plants [43]).

Cyanobacteria must be especially prone to desiccation damage because they evolve intracellular oxygen through photosynthesis and persist in habitats subject to intense solar irradiation (32). In fact, many terrestrial cyanobacteria withstand the most acute long-term water deficit; how they do so is of considerable practical importance. Cyanobacteria are known to use both Fe- and Mn-containing SODs (Fe- and Mn-SODs) to scavenge superoxide radicals (11), but there is a paucity of information on the roles of these enzymes in response to water deficit. The sod genes of a number of cyanobacteria have been characterized. The genome of Synechocystis sp. strain PCC 6803 contains a single SOD gene (sodB; slr1516 [22]); in contrast, Plectonema boryanum UTEX 485 contains sodB as well as three additional Mn-SOD genes (10). Results from studies with Synechococcus sp. strain PCC 7942, in which sodB was inactivated, suggested that in this cyanobacterium Fe-SOD did not protect photosystem II during oxidative stress but that it did protect photosystem I. The enzyme was also able to protect cells from the effects of chilling in the light (S. K. Herbert and D. J. Thomas, Abstr. VIth Cyanobacterial Workshop, abstr. S15, 1998).

The cyanobacterium Nostoc commune exhibits a marked tolerance of desiccation and was the subject of a study that sought to identify structural, physiological, and molecular mechanisms which contribute to this tolerance (31). It is becoming clear that these mechanisms are both numerous and very diverse. Cells of N. commune CHEN/1986 contain sucrose and trehalose (18), which contribute to a marked depression in the T_m of membranes when these are desiccated and rehydrated (20). In each case, the T_m value is at least 20°C below the ambient temperature of the environment from which the materials were collected. Field colonies of N. commune CHEN/1986 also elaborate a complex extracellular glycan which is secreted in copious amounts by liquid cultures of a derivative strain, N. commune DRH1, and which lends a spherical appearance to colonies grown on solid media (18). The extracellular glycan has unusual rheological properties and can prevent fusion of membrane vesicles at very low relative humidities (400 MPa) in the presence of a nonreducing sugar (20).

In addition to trehalose, sucrose, and the glycan, significant amounts of at least two major classes of UV-absorbing compounds provide protection to colonies of N. commune (8, 18). The mycosporines (mycosporine-like amino acid derivatives or MAAs) are a series of colorless, low-molecular-weight, water-soluble compounds with a single absorption maximum within the solar UV-R range (40). These MAAs constitute as much as 10% of the total dry weight of desiccated colonies, and they are released and diffuse from colonies upon rehydration (18). The MAAs in N. commune CHEN/1986 form strong interactions with a collection of proteins, the water stress proteins (Wsp [41]), which are also present in substantial amounts in desiccated colonies (approximately 60% of the total soluble protein) and are released from colonies upon rehydration (18).

The regulation of gene expression during desiccation and subsequent rehydration is of considerable interest. One of the most obvious features of gene expression in rehydrating N. commune is an ordered, apparently stringent, recovery of metabolic functions, beginning with respiration and followed by photosynthesis and finally nitrogen fixation (14, 33). The control of these processes is likely to be complex, given what is known about the sensing of other environmental signals in less complex (morphologically) cyanobacteria (42). Studies with field materials of N. commune HUN and N. commune UTEX 584 provided evidence that some, but not all, proteins remain stable despite extended periods of desiccation (18, 41). Although the drying of N. commune UTEX 584 cells led to a rapid cessation of nitrogenase activity (33, 34), no evidence was obtained for hydrolysis of at least one structural component of nitrogenase, Fe protein, which was present in cells of N. commune HUN following 10 years of desiccation (29). The intracellular ATP pool and the protein biosynthetic machinery of desiccated cells remained unperturbed for 30 min and 2 h, respectively, after rapid drying at 99.5 MPa (1, 33, 34). In contrast, short-term drying led to structural changes in the pigment antenna complexes of cyanobacteria, the phycobilisomes. In the light, the phycobiliproteins were degraded, and even in the dark, short-term drying led to subtle changes in the polydispersity of the complexes when they were analyzed in sucrose gradients (30, 41). De novo transcription in rehydrated cells of N. commune UTEX 584 is directed by extant RNA polymerase holoenzyme, which maintains its stability during desiccation, at least long enough for some transcripts to accumulate to control (predrying) levels (46).

During our studies a marked accumulation of active Fe-SOD was found in desiccated field materials of N. commune CHEN/1986 following rehydration. We describe the biochemical characterization of the enzyme and the isolation and structural characterization of sodF from N. commune DRH1 and propose an unusual but otherwise significant role for SodF in the response to desiccation.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Cyanobacterial strains and growth conditions. Desiccated colonies of N. commune (Table 1) were collected from field localities (19). All of these belong to one form species as determined through group I intron analysis (data not shown). Colonies were kept desiccated in the dark, for periods between 3 and 130 years, prior to the analyses described here. The relative amounts of each sample were determined, and the samples were used for particular experiments and replicate trials. When necessary, desiccated cells were rehydrated with sterile distilled water at room temperature in the light (see below); rehydrated colonies were dried either passively over a period of 24 h or more rapidly under a forced stream of sterile air.

2

1

2

1

Purification and identification of SodF. Desiccated colonies of N. commune CHEN/1986 were used to isolate preparative amounts of SodF. To achieve efficient cell lysis, colonies were frozen under liquid nitrogen and ground to a powder with a pestle and mortar. The powder was mixed with sterile distilled water, and cells were allowed to rehydrate for 1 h at room temperature to allow the release of soluble proteins. Phycobiliproteins were removed by batch adsorption with Q-Sepharose (Pharmacia) resin equilibrated in 50 mM Tris · HCl (pH 7.5) buffer. After dialysis, the sample was concentrated with a 10-kDa-cutoff filter and the retentate, containing approximately 14 µg of total protein, was subjected to preparative sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (10%, wt/vol, gels) as described previously (18). Two major classes of soluble protein were resolved: the 27- to 39-kDa Wsp isoforms (41) and a protein with a prominent band at approximately 20 kDa. Proteins were transferred to Immobilon P membrane (Millipore) via Western blotting in CAPS (3-[cyclohexylamino]-1-propoanesulfonic acid) buffer at pH 10.4 in a Hoeffer transfer chamber and visualized with Coomassie brilliant blue as described previously (18). Protein, immobilized on an excised fragment of membrane, was subjected to automated Edman degradation with an Applied Biosystems model 477A protein sequencer.

Enzyme characterization. Aggregates of desiccated colonies of N. commune ENG/1996 were dispersed gently with a mortar and pestle at room temperature and suspended in sterile distilled water (0.6 g of desiccated cells in 30 ml of sterile distilled water). The suspended cells were incubated quiescently for 5 h in the dark at room temperature and then collected through centrifugation. The supernatant fraction (exudate) was dialyzed at 4°C in 50 mM potassium phosphate-0.1 mM EDTA (pH 7.8) buffer. The exudate at this point was colorless, and only a trace of color (blue; attributed to phycobiliprotein) was observed after subsequent concentration (see below). The cell pellet was resuspended in 6 ml of phosphate buffer, mixed with 0.6 g of alumina (Sigma type A-5), and ground in a mortar for 12 min. The initial slurry, with an additional 6 ml of phosphate buffer, was centrifuged, and the supernatant fraction (extract) was dialyzed in the phosphate buffer. The dialyzed extract and exudate were concentrated approximately fourfold under N₂ (Amicon ultra filter, 10-kDa cutoff).

·

Generation of O₂^· from N. commune DRH1 glycan. The complex extracellular polysaccharide (glycan) of N. commune DRH1 was subjected to exhaustive purification as described previously (20). The glycan was sterilized through autoclaving and used either directly in assays (preparation A) or after solvent extraction to remove monosaccharides released upon autoclaving (preparation B). Glycan (0.9 ml, 10 mg ml¹) was added to a sterile petri dish and mixed with 0.1 ml of 1 mM ferricytochrome c (final concentration, 100 µM). The absorbance at 550 nm (A₅₅₀) was set to zero, and then the solution was irradiated with a 0.6-A Fisher FBUVLS-150 UV-A and -B radiation source with a spectral intensity maximum at 365 nm. The flux densities at the surfaces of the glycan solutions were measured with digital radiometers (Spectronics Corporation, Westbury, N.Y.) and were 200 µW cm² (UV-A) and 20 µW cm² (UV-B). The A₅₅₀ of the solution was measured at time intervals over the course of 70 min. Either 70 or 140 U of bovine erythrocyte Cu-Zn-SOD (1,400 U ml¹) or 1,000 U of bovine liver catalase (14,000 U ml¹) was added to separate aliquots of the glycan-ferricytochrome c solution immediately prior to irradiation. Catalase was assayed by its rate of disproportionation of H₂O₂, monitored at 245 nm (28). The enzymes were stored in 50 mM potassium phosphate-0.1 mM EDTA (pH 7.8) prior to use.

DNA purification. Cells of N. commune DRH1 were harvested, lyophilized, and ground to a powder under liquid nitrogen. The powder was rehydrated (approximately 1:1, powder-buffer) in extraction buffer (100 mM Tris · HCl, 1.4 M NaCl, 20 mM EDTA, 1.5% [wt/vol] cetyltrimethylammonium bromide, 1% [vol/vol] -mercaptoethanol [2]) at 65°C, frozen under liquid nitrogen, and thawed again at 65°C. The cycle was repeated six times. The mixture was extracted with chloroform-isoamyl alcohol (24:1) and centrifuged, and the aqueous phase was mixed with an equal volume of isopropanol. DNA was spooled, dried in air, rehydrated with a minimal volume of Tris-EDTA buffer, and purified further through cesium chloride buoyant-density ultracentrifugation.

Construction of gene libraries. Genomic DNA of N. commune DRH1 was digested partially with Sau3AI and ligated to Lambda Fix II arms prepared by a partial fill-in reaction with dTTP and dCTP after digestion with XhoI (Stratagene). Ligation reaction mixtures were packaged with Gigapack III XL (Stratagene) packaging extracts to select for inserts approximately 18 to 23 kb in size in recombinant phage grown on Escherichia coli XLI-Blue MRA (P2).

Purification of phage DNA. E. coli XLI-Blue MRA (P2) was infected with recombinant phage by standard protocols (44), and phage was recovered from plate lysates in SM buffer (50 mM Tris · HCl [pH 7.5], 100 mM NaCl, 8 mM MgSO₄, 0.01% [wt/vol] gelatin). Phage DNA was recovered with the Wizards Lambda Preps DNA purification system (Promega Corporation).

Isolation and cloning of sodF. A 475-bp fragment of sodF was amplified from N. commune DRH1 genomic DNA by a PCR assay. Two 19-mer oligonucleotides were designed based upon a portion of the sequence (PYGMK) at the N-terminal region of N. commune DRH1 SodF (see Results). The two N-terminal oligonucleotides each had a redundancy of 1:64 and differed only in their 3'-terminal purine; the sequence of SODN-2A is 5'GGAA/GCCNTAT/CGGNATGAAA3' and the sequence of SODN-2G is 5'GGAA/GCCNTAT/CGGNATGAAG3'. Two C-terminal 20-mer oligonucleotides were synthesized based upon the reverse complement of the DNA that encodes the conserved sequence DVWEHAYY found in the carboxy-terminal region of Sod proteins. These two primers had a redundancy of 1:128 and differed only in their 3'-terminal purine; the sequence of SODI-2A is 5'AA/GTANGCA/GTGT/CTCCCANACA3' and that of SODI-2G is 5'AA/GTANGCA/GTGT/CTCCCANACG3'. All four possible permutations of these primers were tested. Assays were performed with 50-µl reaction mixtures in pH 9.0 buffer (supplied by the manufacturer, Promega Biotec) containing 12.5 nmol (250 µM) of each deoxynucleoside triphosphate, 1 mM MgCl, and 2.5 U of Taq polymerase (Promega Biotec). The annealing (90 s) temperature was at 60°C in the first cycle, dropped 1.5°C per cycle for the first 10 cycles (to 45°C), and was kept constant at 45°C for the remaining 30 cycles. In each cycle, denaturation occurred at 95°C for 1 min and elongation was at 72°C for 90 s. The assay began with a denaturation temperature of 95°C for 2 min and ended with an elongation time of 10 min.

Distribution of sodF in N. commune. The distribution of sodF was investigated in desiccated colonies of form species N. commune strains obtained from diverse geographic locations as described in the figure legends (Table 1). Two primers were designed to amplify a 427-bp fragment from within the coding region of N. commune DRH1 sodF (5'GAGTATCACTATGGCAAGCA3' and 5'CTAAAGTCAATGTAGTAG3'). Conditions of the PCR were as described above.

Transcription analysis. Approximately 1 g of desiccated colonies of N. commune ENG/1996 or rehydrated colonies which were blotted dry quickly was frozen in liquid nitrogen and ground to a powder. The powder was mixed with 1.5 ml of RNA extraction buffer (0.25 M sucrose, 0.2 M NaCl, 10 mM magnesium acetate, 0.1 M Tris · HCl [pH 9.0]). After the addition of 1.5 ml of buffered phenol, the mixture was sonicated at a setting of 30 (Fisher sonic dismembrator model 300) three times for 15 s each time, with cooling on ice. Glass beads (4-mm diameter) were added (1:1 ratio), and the slurry was vortexed at top speed for 10 min. The aqueous phase was recovered through centrifugation and extracted with an equal volume of phenol-chloroform-isoamyl alcohol (25:24:1) multiple times until no deposit remained at the interface. The aqueous phase was then extracted with chloroform-isoamyl alcohol (24:1), mixed with 2 volumes of ice-cold 100% (vol/vol) ethanol, and left at 20°C for several hours. The precipitate was recovered, dissolved in 1.7 ml of diethyl pyrocarbonate-treated water, mixed with an equivalent volume of 8 M LiCl, and left overnight at 4°C. The pellet was washed with 70% (vol/vol) ice-cold ethanol and stored in 70% (vol/vol) ethanol at 70°C until needed.

Nucleotide sequence accession number. The DNA sequence data reported here were deposited in GenBank and assigned the accession no. AF177945.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Identification of SodF in desiccated N. commune CHEN/1986. Soluble proteins of N. commune CHEN/1986 were obtained from desiccated colonies after one cycle of freezing and thawing through the addition of water. Resolution of proteins with SDS-polyacrylamide gels identified Wsp isoforms as well as a single prominent band at approximately 21 kDa (Fig. 1a). The 21-kDa protein was recovered, and the unambiguous sequence of the first 20 amino acids of SodF (lacking N-formylmethionine) was determined to be AFVQDPLPFDINALEPYGMK (Fig. 1a). Based on the relative amounts of Wsp proteins, which are the most abundant proteins in desiccated N. commune CHEN/1986 (41), and phycobiliproteins, SodF was judged to be the third most abundant soluble protein in materials of N. commune CHEN/1986 desiccated for 13 years (Fig. 1a).

View larger version (50K): [in this window] [in a new window]   FIG. 1.   (a) SodF is abundant in desiccated N. commune CHEN/1986. An immunoblot stained with Coomassie blue and used for N-terminal sequence analysis of SodF is shown. Wsp isoforms migrate between 30 and 39 kDa. The N-terminal sequence determined for the 21-kDa peptide (SodF) is indicated. (b) Distribution of sodF in the form species N. commune. An agarose gel (stained with ethidium bromide) of sodF PCR amplification products from samples of N. commune (Table 1) is shown. Lane M, HindIII molecular weight markers (LTI Inc.); lane 2, N. commune ALD8122/1974; lane 3, N. commune MEL/1968; lane 4, N. commune WH2/1939; lane 5, N. commune WH4/1869; lane 6, N. commune WH14/1948; lane 7, N. commune DRH1.

SOD activity in vivo. After identification of SodF, aliquots of the retentate (see Materials and Methods) stored at 70°C for approximately 1 month, as well as rehydrating cells of N. commune CHEN/1986, were subjected to enzyme assay. In each case azide-inhibitable SOD activity was detected. In one well-controlled experiment, azide-inhibitable SOD activity was distributed almost equally between the rehydrating cells and the extracellular fluid (~170 U g [dry weight]¹ in each fraction). However, these experiments depleted the supply of desiccated N. commune CHEN/1986. Further experiments were therefore undertaken with other, more abundant, desiccated materials and the readily available clonal strain N. commune DRH1, which was derived from N. commune CHEN/1986.

View larger version (113K): [in this window] [in a new window]   FIG. 2.   A single Fe-SOD activity is identified in N. commune strains. Shown are SOD activities demonstrated in 7% (wt/vol) native tube gels with 0.7 U of N. commune ENG/1996 cell extract (lane 1), 1.3 U of N. commune ENG/1996 exudate (lane 2), 2.7 U of N. commune DRH1 cell extract (lane 3), and 2.4 U of N. commune DRH1 exudate (lane 4). The relative mobility (Rf value) was calculated by using the migration distance of the tracking dye as a reference.

Cloning of sodF. Based upon published alignments of cyanobacterial Sod proteins (data not shown), PCR amplification was expected to generate an approximately 483-bp DNA fragment of sodF. A single prominent 475-bp product was obtained with N. commune DRH1 genomic DNA with primers SODN2A and SODI2A; a much less prominent band of equivalent size was produced with primers SODN2G and SODI2A (data not shown). The identity of the former product as a fragment of sodF was confirmed through DNA sequence analysis (data not shown).

View larger version (53K): [in this window] [in a new window]   FIG. 3.   DNA sequence of sodF and flanking regions from N. commune DRH1. Shown is the complete coding region of sodF, with the putative amino acid sequence (underlined) and the putative ribosome-binding site (boldface underlining). Sequences of primers used to amplify the original 475-bp sodF fragment from genomic DNA correspond to positions 768 to 787 (5' to 3') and 1192 to 1210 (3' to 5') and are underlined; compare the N-terminal amino acid sequence with that of the native protein (Fig. 1a), which shows a single discrepancy (ED) at position 16 of the open reading frame.

View larger version (61K): [in this window] [in a new window]   FIG. 4.   SodF from N. commune DRH1 (SODF-NOSDRH1 [this study]) shows greatest sequence similarity to SodF from P. boryanum UTEX 485 (SODF-PLEBO [10]). The C-terminal regions of PanC proteins from Bacillus subtilis (KCY-BACSU), E. coli (KCY-ECOLI), and Haemophilus influenzae (KCY1-HAEIN) correspond to the partial sequence of PanC from N. commune DRH1 (KCY-NOSDRH1 [this study]). A partial alignment of Phr from Synechococcus sp. strain PCC 6301 (PHR-PCC6301) and N. commune DRH1 (PHR-NOSDRH1 [this study]) is shown. Sequences and alignments were manipulated with Lasergene software (DNASTAR, Madison, Wis.) and the software of the Biology Workbench, NCSA, University of Illinois (http://biology.ncsa.uiuc.edu/).

Distribution of sodF in N. commune. A single amplification product of approximately 427 bp was obtained with sodF-specific primers under stringent conditions of PCR from all strains tested (Fig. 1b), including the three strains (N. commune CHEN/1986, ENG/1996, and DRH1) which were used for the biochemical and enzymatic characterizations of SodF.

Expression of sodF in N. commune ENG/1996. Probing of RNA extracts from desiccated and rehydrating materials of N. commune ENG/1996, as well as liquid cultures of N. commune DRH1 in the late logarithmic phase of growth, identified a major transcript of around 750 bases. Two minor signals between 400 and 600 bases were detected (Fig. 5a, lanes 2, 3, 4 and C), but their resolution was varied in repeat trials. sodF mRNA was abundant in cells after 3 years of desiccation (Fig. 5a, lane 2) and remained so up to 15 min after the onset of rehydration (Fig. 5a, lanes 3 and 4). After this time the intensity of the signals diminished and then gradually increased over the next 24 h (Fig. 5a, lanes 5 through 9). The decrease in sodF mRNA after 15 min of rehydration accompanied an increase in the intensity of signals attributed to 23S rRNA that appeared to reach steady state after approx 3 h of rehydration (Fig. 5b). Multiple bands detected when we used the 23S rDNA probe may represent precursor and processed forms of one or more copies of 23S rRNA (1). The fluctuation in the amounts of sodF mRNA in cells, apparently according to their degree of hydration, raised the question of the variability in mRNA content within different colonies. However, the pattern was consistent in repeat trials with different samples of cells of N. commune ENG/1996; most notably, the depletion in sodf mRNA after 15 min of rehydration and the subsequent increase in the pool of sodF mRNA to controls levels after approximately 24 h were reproducible (data not shown). The patterns in the de novo synthesis of 23S rRNA in response to rehydration (Fig. 5c) were consistent when different samples of the desiccated colonies of N. commune ENG/1996 were used (data not shown).

View larger version (76K): [in this window] [in a new window]   FIG. 5.   (a) sodF expression in N. commune ENG/1996 by Northern blotting. Lane M contains RNA molecular weight markers (in kilobases). Other lanes contain mRNAs from colonies desiccated 3 years (lane 2) or rehydrated for 5 min (lane 3), 15 min (lane 4), 1 h (lane 5), 3 h (lane 6), 6 h (lane 7), 12 h (lane 8), or 24 h (lane 9). A liquid culture of N. commune DRH1 in late logarithmic phase of growth is shown in lane C. (b) 23S rRNA in N. commune ENG/1996 by Northern blotting. The lanes contain identical amounts of the same samples applied in panel a.

Function of SodF in vivo. In view of the release of significant amounts of SodF upon cell rehydration, we questioned the physiological relevance of this extracellular activity and the possible role of the glycan, because there is some evidence that superoxide radicals can lead to depolymerization of polysaccharides (10). Ferricytochrome c was reduced in a time-dependent fashion when it was irradiated with UV in the presence of the glycan (Fig. 6). The rates of ferricytochrome c reduction in the presence of glycan preparation A (Fig. 6a) or glycan preparation B (Fig. 6b) were approximately 150 × 10³ min¹ and 80 × 10³ min¹, respectively, in comparison to the autoreduction rate in these experiments (20 × 10³ min¹). Reduction of ferricytochrome c was inhibited approximately 50% in the presence of SOD, and the degrees of inhibition were equivalent when two different concentrations of SOD were used (see Materials and Methods). Catalase did not prevent the reduction of ferricytochrome c in this system (Fig. 6).

View larger version (13K): [in this window] [in a new window]   FIG. 6.   Glycan of N. commune DRH1 generates O2· when it is exposed to UV irradiation. (a) Glycan preparation A; (b) glycan preparation B. Shown is the reduction of ferricytochrome c (measured as increases in A550) in the presence of glycan (), glycan plus SOD (), glycan plus catalase (), and the control (no glycan present in assay) (). Values are determinations from single experiments; experiments were repeated four times, and the data presented are representative of those obtained in each trial. Variations in the A550 at zero time are due to differences in the buffer solutions of the different components. The slopes of relevant curves were used to estimate the rates described in the text.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Genetic mechanisms that contribute to desiccation tolerance remain poorly understood. It was proposed, in view of the destructive effects of oxygen, that genes involved in oxygen-scavenging mechanisms are likely to be important in the responses of prokaryotes to air drying (31). The quantities of active SodF in N. commune strains CHEN/1986 and ENG/1996, despite prolonged desiccation of cells, suggest that SOD is an important component of the repair process of this cyanobacterium. The fact that SodF remains active after such long-term storage may suggest that the enzyme has structural features which permit it to remain in the native state despite removal of most of the water from the cells or that it may be sequestered in a particular cell microenvironment. The amount of water in such desiccated (anhydrobiotic) cells can be as small as 0.02 g g (dry weight) of cells¹ in which proteins lack a monolayer coverage of water (31, 32). However, not all proteins remain stable under these conditions (30, 41).

SOD activity is located in extracellular fluids, such as synovial fluid (24), or the periplasmic space of E. coli (6) and other bacteria (3, 13, 15), and Nocardia asteroides synthesizes and secretes a SOD which adheres to the outer cell membrane (4). In each of these instances, SOD intercepts extracellular superoxide radicals and diminishes damage to the cell. In none of these examples, however, does the quantity of SOD released from the cells approach the magnitude of that released by rehydrating N. commune. This finding is all the more significant considering the time the cells were stored in a dormant state.

SodF does not have a recognizable N-terminal signal sequence, and microsequencing of the protein through Edman degradation confirmed that there was no processing at the N-terminal region in vivo. Three observations suggest that SodF is secreted in the exudate specifically, rather than via large-scale breakage of cells. First, the specific activity of extracellular Fe-SOD was significantly greater (fivefold in N. commune DRH-1) in the exudate than in the cell extract. Second, there was only a limited number of proteins observed upon electrophoretic examination of the exudate compared to the number present in the crude cell extract. Third, the exudate was colorless and carried only a faint tinge of color following concentration; i.e., there was an inverse correlation between the presence of phycobiliprotein (a sensitive indicator of cell lysis) and SodF activity distributed between the exudate and cell extract. The mechanism of release of SodF from N. commune is unknown, but it represents the third secretion event we have identified that may be of physiological relevance to rehydrating N. commune. Both the predominant acidic water stress proteins (Wsp) and the water-soluble UV-absorbing pigments (MAAs) are released in conspicuous amounts upon rehydration of desiccated cells in the absence of cell lysis (18, 19). A recent characterization of the wsp gene suggests that Wsp, like SodF, lacks an N-terminal signal sequence for secretion (data not shown).

It is reasonable to suspect that crowding and elevated salt concentrations within desiccated cells, as well as intense solar irradiation, contribute to the production of both intra- and extracellular superoxide radicals during rehydration as cells become metabolically active in an environment laced with organic compounds. Metabolic activity exacerbates the damage acquired during desiccation, because cells recover the capacity for photosynthesis and thus oxygen evolution (39).

Cells are metabolically inactive following dehydration and during subsequent storage in the dry state. Transcription analysis indicates that there is turnover of sodF mRNA immediately upon rewetting and that then sodF mRNA synthesis resumes. Although not directly measured, sodF mRNA synthesis must continue to be made and/or be stable as water is removed from cells (to account for the large amounts in all desiccated samples analyzed). The cells after 24 h of rehydration (Fig. 5a, lane 9) are in recovery phase, which may be equated with exponential growth. However, the system is complex. For example, in previous studies we demonstrated a sequential recovery of metabolic functions upon rehydration, beginning with respiration and followed by photosynthesis, etc. (for references, see reference 31). Nitrogen fixation is recovered after 24 h of rehydration. The fact that sodF mRNA is abundant after 24 h of rehydration emphasizes the possible importance of SodF during the recovery phase.

The extracellular glycan may be the primary potential source of damage for rehydrating N. commune in view of its abundance, its capacity to generate significant quantities of superoxide radicals upon UV irradiation, and the fact that it completely envelops the cells and makes intimate contact with the outer membrane (18, 19; in particular, see Fig. 1 in reference 20). Autoclaving the glycan leads to release of ribose from the structure (data not shown) and thus an increase in the concentration of free reducing ends. This finding is consistent with the higher rate of ferricytochrome c reduction obtained with glycan preparation A (Fig. 6a) and mediated through increased O₂^· generation. Again, this may be of physiological relevance in situ, because the ribose linkage is labile, and changes in the rheological properties of the glycan during growth of the cells may involve release of ribose through enzyme hydrolysis (data not shown). In situ, the UV-absorbing MAAs (especially in the presence of reducing sugars) may also be a source of superoxide; they were largely removed during purification of the glycan for the experiments described here. We therefore interpret the release of SodF to be a beneficial biological response to rehydration of the desiccated cells. During this upshift, cells must deal with damage accumulated during desiccation and in part in response to UV radiation. There is clear evidence that SODs can mediate such repair. For example, Deinococcus radiodurans R1 is extremely resistant to oxidative stress and sodA mutants of this organism are much more sensitive to ionizing radiation than the wild type (23). The flux densities encountered by N. commune ENG/1996 in situ on a clear sunny day were measured as 2,150 and 300 µW cm² for UV-A and UV-B, respectively. These values are an order of magnitude greater than those used in the experiments with the purified glycan. The contribution of the glycan to superoxide generation in situ may therefore be significant.

Although sodF is transcribed as a monocistronic transcript, its proximity to at least two genes involved in DNA repair and synthesis may suggest that this region of the genome of N. commune DRH1 functions as part of a global response to environmental stress.