Introduction

Recent advances in genomic sequencing and mass spectrometry (MS)-based proteomics have enabled detailed, cultivation-independent analyses of protein profiles in natural microbial communities. Low-diversity communities that constitute acid mine drainage (AMD) biofilms have served as a model system for the development of these methods and for ecological investigations (VerBerkmoes et al., 2009; Wilmes et al., 2009b). Most of the AMD biofilms are dominated by Leptospirillum Group II (LeptoII), a Fe(II)-oxidizing, chemoautotrophic bacterium (Schrenk et al., 1998; Tyson et al., 2004). The biofilms exhibit distinct developmental stages that vary in microbial community composition. Early developmental stages (DS1) are dominated by LeptoII; however, late developmental-stage biofilms (DS2) diversify, with increasing abundance of Leptospirillum Group III (LeptoIII), archaea and eukaryote populations (Wilmes et al., 2009a).

Proteomic measurements of an early developmental-stage biofilm identified two atypical cytochromes expressed at high levels. These were initially identified as LeptoII proteins of unknown function, and later were determined to be the membrane Cytochrome572 (Cyt572) and periplasmic Cytochrome579 (Cyt579) (Ram et al., 2005). Both cytochromes were purified from a mixed developmental-stage biofilm and characterized biochemically (Jeans et al., 2008; Singer et al., 2008). Cyt572, localized to the outer membrane, is a 57-kDa multimeric protein that oxidizes Fe(II) at low pH, and thus is likely the Fe(II) oxidase for LeptoII in the biofilm (Jeans et al., 2008). Inspection of metagenomic data sets showed six sequences corresponding to LeptoII variants of Cyt572 in the Ultra Back A (UBA) and 5way Community Genomics (CG) databases (5way CG and UBA databases refer to environmental genomic databases obtained from DNA isolated from biofilms collected at distinct sites in the Richmond Mine. 5way CG was sampled at the 5-way convergence of streams in June 2004, and UBA was sampled in the upper A drift in June 2005. Genomic sequences were assembled and these were deposited in the databases of these names (Tyson et al., 2004; Lo et al., 2007)), indicating that multiple variants of the cytochrome may be expressed in the same biofilm sample. Cyt579 was characterized as a ∼16-kDa monomeric protein localized to the periplasm of LeptoII (Singer et al., 2008). Cyt579 was isolated as a mixture of polypeptides with different N-terminal cleavage sites. Redox reactions with Fe(II) demonstrated pH-dependent Fe(II) oxidation that was inconsistent with its assignment as the Fe(II) oxidase for LeptoII and suggested an electron-transfer function. Our working model positions Cyt572 as the Fe(II) oxidase on the outer membrane of LeptoII cells, which oxidizes Fe(II) to Fe(III) and transfers electrons to Cyt579. This scheme is analogous to the proposed role of an outer membrane-bound c-type cytochrome, Cyc2, and rusticyanin, a periplasmic Cu protein, in Acidithiobacillus ferrooxidans, an acidophilic Fe(II)-oxidizing bacterium found in AMD environments (Castelle et al., 2008).

Recent analysis of multiple AMD biofilm proteomes from different developmental stages has shown that the community switches from rapid Fe(II)-dependent autotrophic growth in early developmental stages to partitioning of fixed carbon to heterotrophs in late developmental stages (Denef et al., 2010b). The effects of aging on biofilm morphology, microbial community composition and protein expression led us to speculate that electron-transfer proteins, critical for Fe(II) oxidation, could change physically over the biofilm life cycle, either from specific posttranslational modifications or from genetic variation as a result of mutation or recombination. Characterization of cytochromes from biofilms representing early and late developmental stages (DS1 and DS2, respectively) showed that both posttranslational modifications and expressed sequence variants are correlated with biofilm development.

Materials and methods

Biofilm collection

Biofilm samples were collected from the Richmond Mine (Iron Mountain, CA, USA) (40°40′38. 42″N, 122°31′19.90″W, elevation of <900 m); all samples were grown at the liquid–air interface. DS1 and DS2 samples were collected from ∼10 m into the C drift in November 2006. A DS1 sample was collected in August 2007 and a DS2 sample was collected in November 2006 from a dam at the confluence of the A and B drifts (AB-Muck dam site). DS1 samples were collected in June 2006 and November 2006 from a site ∼75 m into the C drift (C75 m location). Collections of the AB-End sample from November 2004 and the AB-Front sample from June 2004 have been described previously (Ram et al., 2005; Goltsman et al., 2009). Samples were frozen on dry ice immediately after sampling and stored at −80 °C.

Protein purification and enrichment

Both Cyt579 and Cyt572 were purified from the biofilms as described previously (Jeans et al., 2008; Singer et al., 2008) and stored at 4 °C. No change in the redox properties of samples of either protein was observed after 6 months at 4 °C, and minimal degradation was observed by SDS-PAGE. In all cases, visible spectra for both oxidized and reduced samples were identical to our previously published spectra for Cyt579 and Cyt572.

Enrichment of c-type cytochromes was achieved by fractionation of the extracellular fraction of the C75 m sample from November 2006. The 95% (NH4)2SO4 precipitate of the acid wash fraction was dialyzed for 16 h against 20 mM H2SO4/100 mM (NH4)2SO4 pH 2.2. The dialysate was loaded onto a SP-Sepharose (GE Healthcare, Piscataway, NJ, USA) fast flow column and washed with this pH 2.2 buffer. Cyt579 was eluted by a step increase to pH 5.0 in 100 mM NaOAc, and the remaining proteins were eluted with a 0–2 M NaCl gradient in the same buffer. Characteristic visible absorption spectra for c-type cytochromes were observed throughout the 1.2–2.0 M fractions, with 1.4 and 1.5 M fractions containing the highest concentrations of the c-type cytochrome as measured by visible absorbance of the α-band at 552 nm for samples reduced with sodium ascorbate. The individual heme bands were visualized by separation of proteins on 15% SDS-PAGE by the method of Francis and Becker (1984). The stained bands were excised from the gel and digested with trypsin (Haveman et al., 2006).

Redox experiments

Cyt579 purified from C-drift GS1 (11 μg ml−1) and GS2 (7 μg ml−1) were reduced in the presence of 30 mM FeSO4, and the extent of reduction was estimated by comparing the peak at 579 nm to a fully reduced sample, as described previously (Singer et al., 2008). Redox measurements and poising were performed in 10 mM NH4OAc pH 4.3 with 50 μM each of 1,4-benzoquionone, N,N,N′,N′-tetramethyl-p-phenylenediamine, ferrocenecarboxylic acid and 1,1′-ferrocenedicarboxylic acid used as redox mediators (Castelle et al., 2008). Concentrated solutions of sodium dithionite were used as the reductant and sodium hexachloroiridate was used as the oxidant. Concentrations used in the redox experiments were Cyt579, 10–20 μg ml−1; Cyt572, 25 μg ml−1 (C-drift DS1); and 30 μg ml−1 (C-drift DS2). The combination microelectrode (Microelectrodes Inc., New Bedford, MA, USA) was calibrated at pH 4 and pH 7 using quinhydrone.

PCR amplification and cloning of genes encoding Cyt579

Genomic DNA was recovered from the biofilms by a modified phenol:chloroform extraction method and used as a template for PCR (Ram et al., 2005). UBA_8062_372 (Cyt579)was amplified using High-Fidelity Taq polymerase (Invitrogen, Carlsbad, CA, USA) with the primers 5′-ATGAGGATGTGGACAGTGGCTGTC-3′ and 5′-TTACTTTGCTGCTGTGTTCAGAAA-3′. PCR conditions were initial denaturation at 95 °C for 5 min; 30 cycles of 95 °C for 30 s, an annealing temperature of 56 °C for 30 s and 72 °C for 35 s, and an extension at 72 °C for 4 min. PCR products were cloned into TOPO TA10 (Invitrogen) using the manufacturer's conditions, and clones were sequenced at the Joint Genome Institute (Walnut Creek, CA, USA).

Cyt579 intact mass measurement

Purified samples of Cyt579 were further prepared for characterization of the intact proteins by high-resolution top-down MS. Cyt579-enriched samples were desalted with Zip-Tip (C4, Millipore, Billerica, MA, USA) pipette tips and eluted with 100% ACN (0.1% acetic acid, v/v). The protein fraction was then diluted into 50/50/0.1 (v/v/v) H2O/ACN/acetic acid and infused into the Micromass Z-Spray source attached to a Varian (Lake Forest, CA, USA) 9.4-T (Cryomagnetics Inc., Oak Ridge, TN, USA) HiRes electrospray FT-ICR (Fourier transform ion cyclotron resonance mass spectrometer or an electrospray source coupled to the LTQ-Orbitrap-XL (Thermo Fisher Scientific, San Jose, CA, USA). MS fragmentation was achieved through collisionally activated dissociation, electron-transfer dissociation or infrared multiphoton dissociation. Parent charge states of Cyt579 were manually selected, isolated and fragmented (collisionally activated dissociation or electron-transfer dissociation) in the ion trap before high-resolution mass measurement in the Orbitrap. For infrared multiphoton dissociation on FT-ICR, parent charge states of Cyt579 were manually selected and isolated in the selection quadrapole before mass analysis in the FTMS analyzer cell. A 350-nM spike of ubiquitin was introduced into the C-drift DS1 and AB-Muck DS1 samples for internal mass calibration. M/z values were manually extracted from spectra, deconvoluted and plotted with Origin 8 (OriginLab, Northhampton, MA, USA).

General methods

SDS-PAGE was performed according to the method of Laemmli (1970). Protein concentration was estimated according to the method of Bradford (1976). Trypsin digestion and N-terminal sequencing of proteins were performed as described previously (Ram et al., 2005). Trypsin- and pepsin-digested samples were analyzed by 6 h 2D-liquid chromatography-MS/MS as described previously (Jeans et al., 2008; Singer et al., 2008). Pepsin digestions of Cyt572 preparations were performed as described previously (Jeans et al., 2008).

Results

Selection of biofilms for protein extraction

Biofilm samples representing a single developmental stage and with sufficient biomass for biochemical analyses were selected for protein extraction (Table 1) (for locations, see Figure 6 in Denef et al. (2009)). A number of metrics were used to determine biofilm developmental stage. Visual inspection was initially used to distinguish the developmental stages, as early developmental stages are thinner (∼50 μm) and deep pink in color, whereas late developmental stages were thicker (150–200 μm) and light pink in color (Wilmes et al., 2009a). Microbial community analysis was consistent with the observation that DS1 samples were dominated by LeptoII, and DS2 samples had an increased ratio of archaea to bacteria. These observations were complemented by detailed analysis of LeptoII strain populations, which indicated that the strains related to the reconstructed UBA genome dominated the morphologically identified DS1 samples and a mixture of UBA and CG types was observed in DS2 samples (Denef et al., 2010a). As described in the ‘Introduction’ section, comparative proteomic analysis also distinguished biofilm developmental stage, as proteomes measured from DS1 samples were characteristic of rapid autotrophic growth, whereas DS2 proteomes were characteristic of a partitioning of fixed carbon between LeptoII and archaeal populations (Denef et al., 2010b). The initial DS1 samples from the C-drift 10 m site were collected from the center of a flowing AMD stream. DS2 samples were collected from the stream margins in the same location, where biofilms attach to pyrite sediment (Wilmes et al., 2009a). DS1 and DS2 samples were also collected from the AB-Muck site at different times; the morphology of these biofilms was very similar to the DS1 and DS2 samples collected at C-drift 10 m. Biofilms were observed on the surface of a standing pool of AMD at the C75 m. The thickness of the C75 m biofilm was consistently ∼50 μm, but the pH and community composition varied over time for samples collected from the C75 m site (Table 1).

Table 1 Geochemical and microbial community compositions for biofilm samples
Full size table

Cyt579 N terminus is dependent on biofilm developmental stage

In our previous studies, multiple isoforms of Cyt579 with different N-terminal truncations were purified from a mixed development-stage biofilm collected at the C-drift ∼10 m site in November 2005 (Singer et al., 2008). N-terminal sequencing of Cyt579 (AKAMKPPFPV) isolated from a relatively thick, mature biofilm collected at the AB-Front site in July 2004 was identical to the N terminus of the smallest polypeptide isolated from the C-drift biofilm (Goltsman et al., 2009). These preliminary results suggested that the observed N terminus of Cyt579 correlates with the maturity of the biofilm. To test this hypothesis, Cyt579 was isolated from the samples obtained from the DS1 and DS2 C-drift ∼10 m samples, as described above. Approximately six times more Cyt579 was isolated from the DS1 than the DS2 biofilm (Table 2). SDS-PAGE of the Cyt579 from these samples indicated that the DS1 protein had a higher molecular weight than the DS2 cytochrome, consistent with a shorter N terminus in the late developmental stage (Figure 1). Furthermore, N-terminal sequencing by Edman degradation determined two different N termini for the DS1 (AELDILKPRV:ILKPRVPADQ; 5:1 ratio) and DS2 (AKAMKPPFPV) proteins; these corresponded to N termini determined for the Cyt579 preparation from the mixed developmental-stage sample from the C drift (November 2005) (Singer et al., 2008).

Table 2 Yields of Cyt579 from different biofilms
Full size table
Figure 1
figure 1

Cyt579 isolated from C-drift 10 m (11/06) DS1 and DS2. Cyt579 samples were analyzed on 4–12% Bis-Tris polyacrylamide gels (Invitrogen): lane 1, DS1 Cyt579 (12 μg); lane 2, DS2 Cyt579 (14 μg). The color reproduction of this figure is available on the html full text version of the manuscript.

Full size image

Intact protein characterization of Cyt579

High-resolution MS measurement of the purified protein from the C-drift DS1 sample revealed two molecular species by distinct isotopic distribution packets. Identification of the most abundant isotopic masses from each distribution corresponded to molecular species of 16 131.541 and 16 119.562 kDa (Figure 2a; external calibration). The measured mass of 16 131.541 kDa corresponds to the gene UBA_8062_372_S98A (Cyt579), with a predicted signal peptide cleaved at the N terminus (N-AELDILKP). This observation was confirmed by PCR amplification of the gene encoding for Cyt579 from DNA recovered from the C-drift DS1 sample. A total of 29 clones were obtained after transformation of the PCR amplicon, all of which had an identical sequence to UBA_8062_372_S98A. The mature isoform of Cyt579 corresponds to the same sequence but lacks the seven C-terminal amino acids (…FLNTAAK) ending in …GNLKPE; this was the dominant variant expressed in the previously characterized Cyt579 preparations from C-drift biofilm samples (Singer et al., 2008). The second most abundant distribution, 16 119.562 kDa, was inferred to be a modified form of Cyt579. To obtain the most accurate mass measurement possible, an internal calibration using ubiquitin as a standard was used and resulted in a measured mass of 16 119.549 kDa (Figure 2b). Collisional dissociation of this molecular ion resulted in 16 fragment ions, all corresponding to the sequence of Cyt579, including abundant fragment ions corresponding to a sequence tag of MFWVVA, which is unique to Cyt579. This exact sequence tag was also confirmed by electron-transfer dissociation measurements. In addition, infrared multiphoton dissociation fragmentation resulted in an expanded sequence tag of FWVVANGS, confirming the collisionally activated dissociation and electron-transfer dissociation results. The mass errors of the predicted versus measured fragment ions corresponding to the sequence tag were each <10 p.p.m., providing significant confidence in the identification of the 16 119 kDa species as a modified form of Cyt579. Previous bottom-up peptide measurements of this modified protein verified the sequence of the expected Cyt579 protein, plus the presence of an oxidation of residue Met-21. Although the sum total of the fragment ions and peptide mass spectra uniquely support the assignment of the 16 119 kDa species as a Cyt579, neither set provided complete sequence coverage, and thus it was impossible to unambiguously determine the modified form of this truncated version. However, on the basis of all the information, the most likely assignment of the 16 119 kDa species is modification of Cyt579 by oxidation (likely at Met-21), accompanied by the loss of CO from the intact protein (CO loss from intact proteins is not unusual). We searched extensively for the location of the CO loss, but were unable to pin it down in the fragmentation or peptide data. However, the calculated mass for a Cyt579 protein with these specific modifications is 16 119.530 kDa, in excellent agreement with the measured value (<2 p.p.m. mass error). For the C-drift DS2 sample, measured masses correspond to additional N-terminal truncations, resulting in masses of 14 574 and 14 319 kDa (N-LAAAKAMKPP and N-AKAMKPPFPV, respectively; Figure 2c). These masses were previously observed in mixed developmental-stage C-drift samples and are also derived from UBA_8062_372_S98A, with an identical cleavage at the C terminus as the Cyt579 isoform characterized in C-drift DS1.

Figure 2
figure 2

MS spectra of C-drift Cyt579 N-terminal residues are highlighted in red. (a) The full-length sequence of Cyt579 corresponding to mass 16 131 kDa. Intact mass measurement of C-drift DS1 resulted in the identification of two abundant mass distributions, including the truncated product at mass 16 119 kDa as described in text. Residues highlighted in blue indicate the unique sequence tag. (b) Partial sequence displaying the truncated N terminus and accurate mass measurement of C-drift DS1 (mass 16 119 kDa) following mass calibration using an internally spiked ubiquitin standard (inset). (c) Intact mass measurement of C-drift DS2 exhibiting two states of additional N-terminal truncation. The doublets may indicate (−H2O) loss and were confirmed to be Cyt579 through MS fragmentation. Strikethrough highlights the cleaved sequence. The color reproduction of this figure is available on the html full text version of the manuscript.

Full size image

Divergent redox behavior of Cyt579 from distinct developmental stages

Previous experiments involving reaction of Fe(II) with Cyt579 isolated from the Richmond Mine biofilms exhibited pH-dependent redox behavior, with the oxidized form of Cyt579 favored at pH<3, even in the presence of a large excess of Fe(II). To test the redox behavior of the Cyt579 preparations from C-drift ∼10 m DS1 and DS2 samples, extracted proteins were exposed to 30 mM Fe(II) at pH 2. The DS1 preparation was reduced by >95% under these conditions, whereas the DS2 preparation was reduced by <5% (Figure 3a). Consistent with these results, redox titrations at pH 4.3 with redox mediators indicated that the DS1 Cyt579 had a midpoint potential of 605 mV, whereas the DS2 preparation had a midpoint potential of 430 mV (Figure 3b).

Figure 3
figure 3

Fe(II) oxidation of C-drift 10 m (11/06) DS1 and DS2. Cyt579 preparations were exchanged into 200 mM glycine/SO42− buffer pH 2.0 by dialysis and oxidized with a small amount of Fe2(SO4)3. The oxidized samples were diluted into pH 2.0 buffer containing 30 mM FeSO4 and spectra were acquired after 5 min: (a) C-drift DS1 Cyt579 (11 μg ml−1); (b) C-drift DS2 (7 μg ml−1).

Full size image

Cyt579 isolated from AB-Muck developmental stages

To establish the generality of the developmental stage-dependent alterations of Cyt579, the protein was purified from the AB-Muck DS1 samples and DS2 samples. As with C-drift samples, substantially more Cyt579 was extracted from the early development-stage sample (Table 2). MS analysis of intact protein from the AB-Muck DS1 samples revealed a molecular ion identical to that identified from the C-drift DS1 samples (16 119.540 kDa), indicating that the N terminus and Cyt579 variant were the same for both (Figure 4a). Internal mass calibration of the ABM DS1 sample resulted in a 0.022 kDa difference between the ABM DS1 and C-drift DS1 Cyt579. The AB-Muck DS2 sample had molecular ions corresponding to the LAAA N terminus previously observed for the C-drift DS2 sample, but lacked the ion distribution corresponding to the AKA N terminus (Figure 2b). Redox experiments in the presence of 30 mM Fe(II) were very similar to the results obtained for the C-drift samples, and redox titrations at pH 4.3 indicated that the DS1 Cyt579 had a midpoint potential of 600 mV, whereas the DS2 preparation had a midpoint potential of 450 mV.

Figure 4
figure 4

MS spectrum of AB-Muck. N-terminal residues are highlighted in red. (a) MS measurement of ABM DS1 exhibiting the same N-terminal truncation as C-drift DS1. (b) Intact MS measurement of ABM DS2 resulting in conformation of the LAA N-terminal truncation. The color reproduction of this figure is available on the html full text version of the manuscript.

Full size image

Cyt579 from C75 m site

Although the biofilms collected from the C75 m site remained fairly constant in thickness across sampling times, the pH was lower (0.70) for the June than for the November 2006 sample, at which time the pH was 1.0. In addition, archaea were significantly more abundant than in the biofilm growing at this site in June than in November. Cyt579 was extracted from both C75 m biofilms, which was more typical in morphology and microbial community composition to the early developmental-stage biofilms from C-drift and AB-Muck sites described above. The yield of Cyt579 from the November C75 m sample was similar to the yields obtained from the other early growth-stage biofilms; however, the yield from the June sample was dramatically lower (75 times less Cyt579 was extracted from the June biofilm). This result is consistent with an ultrastructural study of the same biofilm (Wilmes et al., 2009a), in which immunohistochemical detection of Cyt579 showed that Cyt579 expression was localized only at the biofilm–water interface, and that a majority of the LeptoII cells did not express Cyt579. Intact protein MS characterization of Cyt579 indicated that the proteins purified from the June and November samples had identical masses of 15 690 kDa, which is the mass of the 8062_372 S98A Cyt579 with an ILKP N terminus, identical to the isoform observed in the previous study on Cyt579 from the biofilms (Figure 5). Although insufficient Cyt579 was recovered from the June C75 m sample for redox analysis, a midpoint potential of 590 mV was determined for Cyt579 from the November C75 m sample.

Figure 5
figure 5

MS measurement of C75 m Cyt579 resulted in an accurate mass of 15 690 kDa corresponding to an ILKP N terminus. The N-terminal residue is highlighted in red. The color reproduction of this figure is available on the html full text version of the manuscript.

Full size image

Identification of c-type cytochromes from LeptoII

Previously, we reported that during purification of Cyt579, a yellow protein fraction bound to the SP-Sepharose column and was eluted with a NaCl gradient (Singer et al., 2008). Preliminary characterization indicated that these high NaCl fractions contained c-type cytochromes. In concert with yields of Cyt579, early developmental-stage samples contained more than five-fold higher levels of c-type cytochromes; therefore, biochemical studies focused on these biofilms. To identify these cytochromes, NaCl gradient fractions from Cyt579 purification from the C75 m November 2006 sample were analyzed in detail. Visible spectroscopy indicated that the concentration of c-type cytochromes was highest in 1.4 and 1.5 M NaCl fractions (Figure 6). These spectra were indicative of c-type cytochromes, with an α-band for the reduced sample at 552 nm. Visualization of the proteins by heme staining following SDS-PAGE revealed two prominent bands at ∼8 and 12 kDa, with a minor band at ∼10 kDa, in the 1.4 M NaCl fraction. In the 1.5 M NaCl fraction, two prominent bands were observed at ∼24 and ∼29 kDa (Figure 7). All of these bands were excised from the gel, digested with trypsin and analyzed by liquid chromatography-MS. The 8 and 12 kDa bands were identified as products of genes UBA_LII_8524_245 and 8524_197, respectively (Supplementary Information, Supplementary Figures S1 and S2). The 24 and 29 kDa proteins were identified as products of genes UBA_LII_7931_111 and 7931_112, respectively, and therefore are adjacent on the LeptoII genome. These protein sequences are similar to diheme cytochromes, and may be coexpressed in a predicted operon. Peptides for all of these c-type cytochromes were previously identified in multiple proteomic data sets obtained from extracts of AMD biofilms (Goltsman et al., 2009). Peptides for an additional predicted 10 kDa monoheme c-type cytochrome, UBA_LII_7931_87, were detected in multiple proteomic data sets; although a corresponding band was not detected in the SP-Sepharose fractions, the faint heme-stained band at 10 kDa is a likely candidate.

Figure 6
figure 6

Visible spectra of c-type cytochromes. Pale yellow fractions obtained from pH 5.0 NaCl gradient elution (0–2 M) of SP-Sepharose column after elution of Cyt579 were concentrated and measured by visible spectroscopy. Using the Soret band of oxidized samples as an indicator, fractions eluting at 1.4 M (0.1 mg ml−1 protein) and 1.5 M NaCl (0.2 mg ml−1) contained the highest concentration of c-type cytochromes. Samples were reduced by addition of 1 μl of a concentrated (100 mM) solution of sodium dithionite: (a) 1.4 M NaCl; (b) 1.5 M NaCl. The color reproduction of this figure is available on the html full text version of the manuscript.

Full size image
Figure 7
figure 7

Heme stain of c-type cytochromes. 1.4 and 1.5 M NaCl fractions were treated with cold 10% TCA and the precipitated protein pelleted by centrifugation. The pellet was resuspended in 10 μl of SDS-PAGE sample buffer and protein separated by SDS-PAGE on a 15% Tris-HCl polyacrylamide gel (Bio-Rad, Hercules, CA, USA). Heme-containing proteins were then visualized by staining (Francis and Becker, 1984). The bands were excised and digested with tryspin according to a previously published procedure (Haveman et al., 2006). Gene products identified are from lane 1 (1.4 M NaCl): 1. UBA_8524_235; 2. UBA_8524_197; from lane 2 (1.5 M NaCl): 3. UBA_7931_111; 4. UBA_7931_112. The color reproduction of this figure is available on the html full text version of the manuscript.

Full size image

Diversification of monoheme cytochromes in late biofilm-developmental stage

Inspection of the reconstructed genomes for LeptoII from the 5way CG and UBA sites indicated that the sequences for the diheme c-type cytochromes were identical in the two genomes; however, the two monoheme c-type cytochromes displayed sequence variation when the two genomes were compared (Figure 8). To determine whether these variants were expressed in different developmental stages, their expression in the extracellular proteome of the AB-End, an early developmental-stage biofilm and AB-Front, a late-stage biofilm, were compared (Goltsman et al., 2009). In the AB-End proteome, only peptides corresponding to the UBA variants of these two monoheme cytochromes were detected, whereas peptides for both the UBA and 5way CG variants were detected in the AB-Front proteome (Table 3), consistent with the dominance of these samples by single versus multiple genotypes.

Figure 8
figure 8

Sequence comparison for monoheme c-type cytochromes. Sequence comparison for LeptoII monoheme cytochromes identified by heme stain (Simmons et al., 2008; Goltsman et al., 2009). For CG_8_58/UBA_8524_235, the N terminus was predicted by SignalP and observed by Edman degradation (APDVK; see Supporting Information, Figure 2). For CG_8_20/UBA_8524_197, N termini were predicted using SignalP (http://www.cbs.dtu.dk/services/SignalP/).

Full size image
Table 3 Comparison of unique peptidesa from monoheme cytochromes from AB-End and AB-front samples
Full size table

Comparison of Cyt572 from biofilms

To compare the effects of biofilm development stage and environmental conditions on the properties of Cyt572, the membrane protein was purified from the November 2006 C-drift DS1 and DS2 samples and from the June and November 2006 C75 m samples. In contrast to Cyt579, approximately equal amounts of Cyt572 were isolated from all four biofilms; ∼0.5 mg of Cyt572 per 5 g of material remaining after extraction of each of the biofilms with dilute acid. N-terminal sequencing of the purified proteins revealed an identical sequence (YPGFAR), corresponding to the predicted N terminus based on the sequences for Cyt572 in the reconstructed LeptoII and LeptoIII genomes, and also identical to the previously purified preparations of Cyt572 (Jeans et al., 2008).

Analysis of the reconstructed LeptoII and LeptoIII genomes has indicated a substantial number of sequence variants for the gene encoding for Cyt572 (Jeans et al., 2008). A search of the peptides obtained from each of the preparations of Cyt572 against this genomic database showed that most of the biofilms contained multiple variants of Cyt572 proteins that copurified (Table 4, Supplementary Figure S3). Preliminary experiments indicate that additional LeptoII and LeptoIII sequence variants can be recovered from biofilm samples by shotgun genomic sequencing and PCR amplification, indicating the sequence diversity of genes encoding Cyt572 is considerably broader than currently represented in the 5way CG and UBA genomic data sets (data not shown).

Table 4 Detected Cyt572 variants
Full size table

Redox behavior of Cyt572 from biofilms

When incubated with 1 mM Fe(II) at pH 2.0, the four preparations of Cyt572 were fully reduced, as observed for the previously reported sample. As proteomic measurements showed that Cyt572 preparations were mixtures of many sequence variants, formal redox titrations were not performed. However, to compare early and late developmental-stage samples, Cyt572 preparations isolated from C-drift DS1 and DS2 samples were poised with a mixture of redox buffers at 570 mV and visible spectra recorded (Figure 9). In both cases, the preparation was reduced by 75–80%, indicating that despite the sequence variation, the overall redox behavior is similar between developmental stages.

Figure 9
figure 9

Poising of C-drift DS1 and DS2 Cyt572 preparations at −570 mV. C-drift Cyt572 preparations were diluted into 10 mM NH4OAc containing redox mediators as described in Experimental Procedures and poised with Na2IrCl6 at −570 mV: (a) C-drift DS1 (25 μg ml−1); (b) DS1 (30 μg ml−1).

Full size image

Discussion

We isolated redox-active proteins from a natural biofilm community and showed that posttranslational modification and expression of sequence variants are dependent on biofilm developmental stage. Isolation of Cyt579, hypothesized to be the dominant electron-transfer protein in the periplasm of LeptoII (Singer et al., 2008), established that protein abundance, N-terminal cleavage sites and midpoint potentials are correlated with biofilm growth stage. A C-terminal cleavage site was identified that is common to all isoforms of Cyt579 recovered from Richmond Mine biofilms and also inferred, based on MS analysis, for intact Cyt579 purified from the Leptospirllum ferriphilum P3A isolate (RC Blake II, unpublished data). Therefore, C-terminal cleavage seems to be a general posttranslational modification of Cyt579 by Leptospirillum strains. The N-terminal cleavage sites do not appear to be adventitious or represent nonspecific protein degradation, as they have been observed in Cyt579 samples prepared from multiple biofilms collected over 4 years. In addition, the N-terminal cleavage sites of late developmental-stage biofilms are unchanged when Cyt579 purification was performed in the presence of protease inhibitors (data not shown). Therefore, the N terminus of Cyt579 functions as a molecular ‘clock’ that correlates the protein isoform with the maturity of the biofilm. Geochemical conditions also affect the expression of Cyt579, as substantially less Cyt579 was isolated from a biofilm (C75 m in June 2006) collected under low pH conditions that had higher ratio of archaeal to bacterial community membership. Despite the significant changes in the properties of Cyt579 in different biofilms, a striking observation enabled by high-resolution MS is that one dominant sequence variant is expressed in all of the biofilms from which the protein has been isolated.

For c-type cytochromes expressed by LeptoII, the diheme cytochromes identified by proteomic and biochemical techniques were invariant, whereas two monoheme cytochromes had distinct sequence variants in the reconstructed 5way CG and UBA genomes of LeptoII. The sequences identified in the LeptoII UBA genome are expressed in both early and late developmental-stage biofilms; however, the LeptoII 5way CG variants are detected by proteomics only in late developmental-stage samples. This observation is consistent with appearance of strains of LeptoII related to the reconstructed 5way CG strain only in late developmental-stage biofilms, as assayed by proteomic measurements and fluorescent in situ hybridization (Denef et al., 2009). The monoheme variants from the 5way CG LeptoII strains expressed in the late developmental-stage biofilms may have altered redox potentials in comparison with the UBA LeptoII variants.

For Cyt572, the N-terminal cleavage site and redox behavior were not altered by biofilm age and changing environmental conditions. In most of biofilm samples from which Cyt572 was isolated, peptides that mapped to multiple genomic variants were detected from pepsin-digested samples. The significant amount of variation observed for genes coding for Cyt572 may be a result of fine-scale environmental interactions, as distinct sub-populations of LeptoII colonize micro-niches in the biofilm. Localization of Cyt572 to the outer membrane would cause the protein to interact directly with the environment and therefore be sensitive to small perturbations.

Previously, we proposed that Cyt572 may function as the Fe(II) oxidase for Leptospirillum Group II, oxidizing Fe(II) on the surface of cells and transferring electrons to Cyt579 (Figure 10a) (Jeans et al., 2008; Singer et al., 2008). This study suggests that c-type cytochromes are present at significantly lower levels and may bifurcate electrons from the dominant proteins to the cytochrome c oxidase and cytochrome bc1 complexes. Mono-and diheme cytochromes have been shown to transfer electrons from rusticyanin to cytochrome c oxidase and cytochrome bc1 complex in A. ferrooxidans (Castelle et al., 2008) (Taha et al., 2008). However, the observation of developmental stage-dependent changes in the cytochromes complicates this model of Fe(II) oxidation. Our initial model was based on the observation that Cyt572 is capable of oxidizing Fe(II) at pH 1–3, whereas Cyt579 was only active at pH above 3. Our current study shows that Cyt579 isolated from carefully collected samples of early developmental-stage biofilms is readily reduced by Fe(II) at low pH and that the Cyt579 samples with lower redox potentials are from late developmental stages. These results suggest that Cyt572, the redox potential of which is invariant as a function of developmental stage, and Cyt579 have similar redox potentials in early developmental-stage biofilms. Alternatively, Cyt579 may function as an alternative Fe(II) oxidase in early developmental-stage biofilms, as has been previously proposed (Figure 10b) (Ram et al., 2005). In late developmental-stage biofilms, the lowering of the redox potential of Cyt579 may represent an alteration of the pathway to enable the use of different electron acceptors (Figure 10c). Elucidation of these alternative pathways for electron transfer will require cytoplasmic membrane preparations from the biofilm that contain active preparations of cytochrome c oxidase and the cytochrome bc1 complex from Leptospirillum Group II.

Figure 10
figure 10

Working models for Fe(II) oxidation by LeptoII in Richmond Mine biofilms. OM represents the outer membrane of LeptoII cells, IM represents the cytoplasmic membrane of LeptoII cell: (a) mechanism of Fe(II) oxidation based on previously published experiments on Cyt579 and Cyt572(Jeans et al., 2008; Singer et al., 2008); (b) alternative mechanism for Fe(II) oxidation in DS1 biofilms with both Cyt572 and Cyt579 as Fe(II) oxidases; (c) alternative mechanism for Fe(II) oxidation in DS2 biofilms with c-type cytochromes as initial electron acceptors from Cyt572. The color reproduction of this figure is available on the html full text version of the manuscript.

Full size image

Previous studies with single species biofilms have described changes in protein type and abundance as these biofilms mature (Southey-Pillig et al., 2005, Teal et al., 2006). Our work is unique in that it shows both posttranslational modification and sequence variation in individual proteins in the life cycle of a natural, multispecies biofilm. This study shows that combining both high-throughput proteomics measurements and targeted biochemical studies can identify highly expressed proteins in natural microbial communities that may be sensitive to changes in the environment or species composition. These observations are critical to link biochemical pathways to the functioning of natural microbial communities.