In situ identification of keratin-hydrolyzing organisms in swine manure inoculated anaerobic digesters


Correspondence: Daniel I. Massé, Dairy and Swine Research and Development Centre, Agriculture and Agri-Food Canada, 2000 College Street, Sherbrooke, QC, Canada, J1M 0C8. Tel.: +1 819 565 9171; fax: +1 819 564 5507; e-mail: daniel.


Feathers, a poultry byproduct, are composed of > 90% keratin which is resistant to degradation during anaerobic digestion. In this study, four 42-L anaerobic digesters inoculated with adapted swine manure were used to investigate feather digestion. Ground feathers were added into two anaerobic digesters for biogas production, whereas another two without feathers were used as negative control. Feather degradation and enhanced methane production were recorded. Keratin-hydrolyzing organisms (KHOs) were visualized in the feather bag fluids after boron-dipyrromethene (BODIPY) fluorescence casein staining. Their abundances correlated (R2 = 0.96) to feather digestion rates. A 16S rRNA clone library was constructed for the bacterial populations attached to the feather particles. Ninety-three clones (> 1300 bp) were retrieved and 57 (61%) belonged to class Clostridia in the phylum Firmicutes, while 34 (37%) belonged to class Bacteroidia in the phylum Bacteroidetes. Four oligonucleotide FISH probes were designed for the major Clostridia clusters and used with other FISH probes to identify the KHOs. Probe FIMs1029 hybridized with most (> 80%) of the KHOs. Its targeted sequence perfectly matches that possessed by 10 Clostridia 16S rRNA gene clones belonging to a previously uncharacterized new genus closely related to Alkaliphilus in the subfamily Clostridiaceae 2 of family Clostridiaceae.


Feathers are a poultry byproduct composed of > 90% keratin (Astbury & Beighton, 1961), mostly as β-keratin (Sawyer et al., 2000). β-keratin is recalcitrant to degradation by many proteases because of the presence of extensive disulfide bonds and cross-linkages (Papadopoulos et al., 1986). Because feathers are generated in millions of tons annually, their disposal is an important part of solid waste management (Onifade et al., 1998). Traditionally, feathers are treated as an organic solid waste, being incinerated or disposed at waste disposal sites. These treatments are wasteful because feather keratin is rich in useful amino acids and its disposal by such methods generates greenhouse gases (Salminen & Rintala, 2002). Consequently, efforts have been made to develop biotechnologically more attractive alternatives. For example, feathers have been recycled after cooking under high pressure and temperature to make a feather meal suitable for use as an animal protein supplement (Odetallah et al., 2003) or as a slow-releasing nitrogen fertilizer (Choi & Nelson, 1996). However, hydrothermal treatment of feather keratin is expensive and destroys several valuable amino acids. The resulting product is poor in digestibility and has a variable nutrient quality (Wang & Parsons, 1997).

Feathers have been used in anaerobic digestion systems to produce methane biogas but are poorly degraded under anaerobic conditions (Salminen & Rintala, 2002). Pretreatments including thermal, chemical, and enzymatic methods have been used in attempts to increase their digestion rates (Onifade et al., 1998). When Williams & Shih (1989) enriched a mixture of bacteria using ball-milled feather agar, the resultant bacterial community showed keratinase activity capable of degrading autoclaved whole feathers. Feather degradation has been reported in anaerobic digesters (ADs) treating poultry waste including manure and/or mixed fractions of bone, trimmings and offal under thermophilic (Williams & Shih, 1989) or mesophilic (Salminen & Rintala, 2002) condition. However, the chemical transformation and the organisms responsible for this degradation have not been characterized in detail.

Keratin-hydrolyzing bacteria and keratinase enzyme mixtures have also been reported to digest prions responsible for transmissible spongiform encephalopathies or prion diseases including bovine spongiform encephalopathy (BSE), sheep scrapie, deer chronic waste disease, and human Creutzfeldt-Jakob disease (Gupta & Ramnani, 2006). Keratinases secreted by thermophilic bacterial strains VC13, VC15, S290 and Streptomyces sp. S6 (Tsiroulnikov et al., 2004), Bacillus licheniformis strain PWD-1 (Langeveld et al., 2003) and Bacillus MSK103 (Yoshioka et al., 2007) were shown to inactivate prions, because the structures of prions and feather keratin share many characteristics. Both are fibrous and insoluble proteins rich in β-sheet (Fraser et al., 1972). BSE is the most notorious prion disease, which has cost the Canadian cattle industry approximately $6.3 billion since 2003, as a result of closed markets and consumer concerns over food safety issues related to it ( Therefore, the control and inactivation of prions from cattle carcasses during processing have been major public and animal health concerns (Hill et al., 1997).

In this study, ground feathers were added in bags to ADs inoculated with adapted swine manure for methane production. The keratin-hydrolyzing organisms (KHOs) responsible for the feather digestion were detected microscopically by BODIPY fluorescent protease staining (BODIPY FL casein staining) and further identified using the full-cycle rRNA approach (Amann et al., 1995). A 16S rRNA clone library was constructed from the bacterial populations attached to feather particles inside feather bags. Oligonucleotide FISH probes were designed to target the major clusters of the 16S rRNA gene clones retrieved and used to identify the KHOs in situ.

Materials and methods

Source and preparation of chicken feathers

Freshly plucked white chicken feathers were collected from a slaughterhouse (Saint-Anselme, QC, Canada) and transferred to the laboratory within 4 h. The feathers were divided into 2 kg portions, placed in clean cotton bags and washed (delicate cycle) in a washing machine (Frigidaire, Martinez, GA) with tap water. Feather samples were then dried at 45 °C in a Unithern dryer (Construction CQLTD, England) until a constant weight was reached (after about 8 weeks). The samples were ground through a 4-mm screen (Thomas-Wiley Laboratory Mill) and divided into portions of ≈ 33 g each and placed in nitrogen-free polyester bags with a pore size of 50 μm (ANKOM Technology, Macedon, NY). Each bag was sealed with a plastic tie wrap and washed in the washing machine to remove any remaining residual particles with a diameter < 50 μm. Finally, the bags were dried at 45 °C until a constant weight (≈ 31 g each) was achieved. As required, 14 bags were then attached onto a steel stick and inserted into an AD.

Anaerobic digester set up

Four 42-L Plexiglas ADs and one 7-m3 semi-industrial AD, as described by Massé et al. (2001), were used in this study. The 7-m3 semi-industrial AD was used to adapt fresh swine manure collected from a commercial pig farm (Sherbrooke, Quebec) that processes about 4000 pigs per year, which was then used to inoculate (representing 100% of the volume) the 42-L ADs. The semi-industrial AD was fed with fresh swine manure with a retention time of 14 days. At the time of inoculation of the 42-L ADs, the semi-industrial 7 m3 AD had been operating at 25 °C for more than 2 years.

Feather bags were added to two 42-L ADs (feather ADs with duplicate) at different times. Fourteen bags (No.1 to No.14) were added to each AD on day 0 [initial load 0.88g g−1 VSS (volatile suspended solid) manure, representing 30% of the total Chemical oxygen demand (COD) in each AD]. Three bags (from bags No.15 to No. 26) were added at day 22, day 55, day 85, and day 113. The other two ADs (control ADs with duplicate) were operated without addition of feather bags serving as negative controls. The volume of the inoculum for these four ADs was 35 L. All ADs were operated in a closed room at 25 °C for 153 days.

Physiochemical characterization of anaerobic digesters

Total solids, volatile solids, total suspended solids (TSS), and volatile suspended solids were determined according to standard methods (APHA, 1998). Soluble COD was determined according to the closed reflux colorimetric method described in the standard methods. Methane (CH4), total Kjeldahl nitrogen (TKN), ammonia nitrogen (NH3-N), and liquid volatile fatty acids (VFAs) were analyzed following the procedures described by Massé et al. (2001). COD of raw feathers was determined following standard methods. At different time intervals (22–33 days), three feather bags (No.1 to No.12) were taken out, and washed with deionized water, dried for 48 h at 45 °C, and weighed to determine feather degradation. Bags (No.15 to No.26) were kept until the end of the experimental trial. Bags (No.13 and 14) were used to collect fresh samples for BODIPY FL casein staining; DNA extraction and FISH (see later sections for details). These two bags were put back immediately after sampling.

Mass balancing, digestion rate calculations

The extent of feather digestion and rates of feather digestion were estimated using a mass balance over COD as described by O'Sullivan & Burrell (2007). The CH4, H2, VFAs, and biomass data were all converted to units of grams of COD equivalents for application to the mass balance. The extent of feather digestion was estimated according to the following equation:

display math

where COD initial is feather COD added to an AD, SCOD CH4, SCOD H2, SCODVFA,and SCODbiomass were the COD of CH4, H2, VFAs, and biomass correspondently due to feather digestion. The feather digestion rate was calculated from the slope of the extent of digestion curve. It was equal to the SCOD at time [t + Δt] minus the SCOD at time [t], divided by the time step.

Sample preparation for microbiological analyses

To collect any bacteria attached to the feather particles, fresh feather samples (5–10 g) collected from the feather bags (No. 13–14) were drained for 3–5 min on a sieve and resuspended in bacteria-free AD filtrate (50–100 mL). The filtrate was prepared from the mixed liquor of the same AD by centrifugation at 16 000 g for 30 min and filtration through a series of filters (Whatman Nuclepore track-etched membranes) with pore sizes of 3, 1, 0.45, and 0.2 μm, used sequentially. The mixture was transferred to a thick-walled polyethylene bag (180 mm × 300 mm) and subjected to vigorous mechanical pummelling for 2 min in a Colworth Stomacher 400 (A. J. Seward & Co. Ltd, London). This suspension was then filtered through a sterilized eight-layer cheesecloth. A portion of filtrated bag fluids was fixed in ice-cold ethanol (50% final concentration) or paraformaldehyde (PFA) (4% final concentration) for FISH probing gram-positive (Roller et al., 1994) and gram-negative bacteria (Amann, 1995), respectively. Another portion was transferred into 2 mL Eppendorf tubes that were immediately frozen in liquid nitrogen and stored at −80 °C for DNA extraction. The remainder was centrifuged at 800 g for 15 min to remove large particles before being used for BODIPY FL casein staining.

BODIPY FL casein staining combined with DAPI staining and enumeration

The BODIPY FL casein staining was performed following the procedure described previously (Xia et al., 2007). To stain microorganisms attached to feather particles, briefly, a 200 μL aliquot of the supernatant was mixed with the same volume of freshly prepared 2 × Tris-HCl buffer (10 mM, pH 7.8) before 200 μL BODIPY FL casein working solution was added. The mixture was placed in a 10-mL serum bottle wrapped in aluminum foil and incubated at 25 °C for 30 min on a rotating disk (220 r.p.m) before microscopic examination. Samples after incubation with BODIPY FL casein were spread evenly on three-well (10 μL in each well) gelatin-coated Teflon printed slides (Electron Microscopy Sciences) and dried in a dark room before being mounted with antifade reagent CITI fluor (Electron Microscopy Sciences) and examined microscopically.

To stain the bacteria in the mixed liquor, fresh mixed liquor samples were collected directly from each AD, a 400 μL aliquot then centrifuged at 4500 g for 10 min, and the supernatant discarded. The biomass pellet was resuspended in 1 × Tris-HCl, and BODIPY FL casein working solution was added to a final concentration of 0.5 mg g−1 TSS with a final reaction volume of 600 μL. The incubation and microscopic examination conditions were the same as those used for staining and examination of bacteria attached to the feather particles described above.

The BODIPY FL casein staining combined with DAPI (4′,6-diamidino-2-phenylindole) staining was also performed according to the procedure described previously (Xia et al., 2007, 2008). Briefly, cells staining positively with BODIPY FL casein were detected microscopically and their positions on the slides recorded. Then, the CITI fluor was washed away by gently rinsing slides (for 1 min) with 70% ethanol. After being air-dried DAPI (100 μL, 0.003 mg mL−1) was spread over the slides and kept at room temperature for 10 min. Then, the slides were dipped in distilled water three to five times before being air-dried and examined microscopically. The abundances of the KHO cells stained with BODIPY FL casein were estimated as a percentage of total cell numbers stained by DAPI in the same microscopic field. At least 60 sets of images taken with the 100 × objective lens from three different wells (20 images from each) were counted for each sample. All cell number estimations were performed in ImageJ (Abramoff et al., 2004) using digital images.

DNA extraction

The feather bag fluid samples stored at −80 °C were used for DNA extraction. Samples collected from day 60-113 when a high abundance of KHOs was detected, were used. DNA was extracted with a FastPrep®-24 System & Kit following the protocol provided by the provider (MP Biomedicals, 29525 Fountain Pkwy, Solon, OH). DNA extraction was performed as described by Kong et al. (2010). Briefly, 500-μL aliquots of feather bag fluids were centrifuged for 10 min at 10 000 g. The pellet obtained was re-suspended in 465 μL potassium phosphate buffer (0.4 M) which was supplemented with 33.2 μL of lysozyme (100 mg mL−1) and 3.32 μL of mutanolysin (2.5 U μL−1) to a final volume of about 500 μL. The suspension was incubated at 37 °C for 30 min before 6.65 μL proteinase-K (20 mg mL−1) was added and further incubated at 37 °C for 1 h. The DNA extracted individually from each sample was mixed in equal amounts and used as PCR templates.

Clone library construction

The bacterial universal primer pair 63F (5′-AGGCCTAACACATGCAAGTC-3′) and 1392R (5′-GGGCGGWGTGTACAAGGC-3′) (Marchesi et al., 1998) were used to amplify bacterial 16S rRNA genes. The PCR cycle was as follows: 25 cycles of denaturation (1 min at 95 °C), annealing (1 min at 55 °C), and extension (1.5 min at 72 °C) before a final extension at 72 °C for 8 min. To ensure products were of the expected size, the PCR amplicons were run on 1% agarose gels. Then, they were purified using a QIAquick PCR purification kit (Invitrogen, Burlington, Ontario, Canada) according to manufacturer's instructions, before being ligated into the pCRII-TOPO vector provided in the TOPO TA cloning kit (Invitrogen). Clones with the correct inserts were sequenced at Laval University ( with an ABI 3130xl Genetic Analyzer (Applied Biosystems-Hitachi, Foster City, CA).

Phylogenetic analysis

Partial 16S rRNA gene sequences obtained from sequencing were assembled using sequencher 4.5 (Gene Code Corperation) and evaluated using the Chimera check program provided in RDP (Cole et al., 2005) and Bellerophon (Huber et al., 2004). After removing any putative chimera sequences, they were aligned using “Aligner” in Silva (Schloss, 2009) against their most closely related 16S rRNA gene sequences. The database generated was retrieved into ARB (Ludwig et al., 2004) and used to build phylogenetic (parsimony) trees. Bootstrap values were calculated with Mega4 (Tamura et al., 2007) using the model Kumura 2-Parameter. Using “CLASSIFIER” and “SEQMATCH”, 16S rRNA gene based identifications and similarity comparisons were performed, respectively, provided in RDP. Operational taxonomic units (OTUs, determined using the furthest neighbor algorithm) and Chao 1 estimations were calculated using Mothur (version 1.10.2,

Design and specification of oligonucleotide probes

Oligonucleotide probes targeting 16S rRNA were designed using ARB software. Their names, specificities, and formamide concentrations for FISH are listed in Fig. 4. The optimal formamide concentrations of these probes were first determined using fixed cells in feather bag fluids. The highest formamide concentration (tested in 5% step-wise increases) at which a clear fluorescent signal was observed was selected as the putative optimal FA concentration for each gene probe. The formamide concentration of probe FIMs1029 hybridizing with the rod-shaped KHOs was finally determined using Clone-FISH technique before being used to identify the KHOs. Clone-FISH was performed according to Schramm et al. (2002). The plasmid and host cell pair used was pGEM-T and JM103 (DE3), respectively (Promega, Fisher Scientific). Briefly, the swine manure digester clones (Fig. 5) with zero (clone SPAD28) or one mismatch (clone SPAD22) with probe FIMs1029 were inserted into the plasmid pGEMT equipped with a T7 RNA polymerase, and cloned into JM109 (DE3) competent cells. These cells were incubated at 37 °C with IPTG (1 mM) and chloramphenicol (170 mg L−1) before being fixed in paraformaldehyde (4%) for FISH probing. The formamide concentration chosen for these probes was determined as that used previously with the fixed cells in feather bag fluids.

BODIPY FL casein staining combined with FISH and enumeration

BODIPY FL casein staining combined with FISH was performed following the procedure described previously (Xia et al., 2007). FISH was performed on the slides after BODIPY FL casein staining. FISH was performed according to Amann (1995). All oligonucleotide probes were purchased from Opron (Huntsville, AL). Probe BAC303 (Manz et al., 1996) targeting most Bacteroidaceae and Prevotellaceae was used to target all the Bacteroidetes clones (Fig. 4) retrieved in this study. Probe Clost I (Küsel et al., 1999) targeting subgroup of Clostridia clusters I and II was used to target the five clones belonging to subfamily Clostridiaceae 1 (Fig. 3). Probe NONEUB (Wallner et al., 1993) was used as a negative control for detecting any general autofluorescence. If necessary the pretreatment used in CARD-FISH (Ferrari et al., 2006) to improve the penetration ability of gene probes was also adopted in this study. The percentages of the KHO cells hybridizing with probe FIMs1029 were estimated as a percentage of total cell numbers stained by BODIPY FL casein staining in the same microscopic field. At least 60 sets of images taken with the 100 × objective lens from three different slide wells (20 images from each) were counted for each sample. All countings were performed in imagej using digital images.


Images of bacterial samples stained with BODIPY FL casein, DAPI, and FISH were captured randomly using an epifluorescence microscope (Image-Pro Plus, Olympus. 1x70) equipped with a CCD digital camera SC500J (CoolSNAP-Procf, Media Cybernetic).

Nucleotide sequence accession numbers

The 16S rRNA gene sequences obtained in this study have been deposited in the GenBank database under accession numbers HQ698156-HQ698248.


AD performance

Feather mass in the bags was gradually degraded over the 153-day period of the experiment. VFAs (measured weekly) including acetic acid, propionic acid, and isobutyric acid were detected in the feather ADs and control ADs but all at a low level. In the control ADs without feathers, NH3-N and TKN fluctuated between 5.5 and 6.1 g L−1 and between 6.2 and 6.9 g L−1, respectively, throughout the experiment. In contrast, NH3-N and TKN in the feather ADs fluctuated between 5.5 and 6.1 g L−1 and between 6.7 and 6.9 g L−1, respectively, in the first 60 days and then increased up to 7.5 and 8.4 g L−1, respectively, at the end of experiment. In both the feather ADs and control ADs, the pH remained relatively constant at between seven and eight throughout the digestion. At the end of the experiment the mean cumulative methane (measured weekly) produced in the feather ADs (219.39 L) was 1.84 times higher than that produced in the control ADs (119.15 L). Mass balance calculations show that a total of 94% of the feathers was degraded after 89 days at an average removal rate of 4.8 mg COD g−1 feather per day. Statistical analysis of repeated batch digestions of feather has revealed a variability coefficient of ± 1.10% (calculated as the 95% confidence limit of the mean).

Correlation of abundance of proteolytic organisms in feather bags to feather digestion rates

The proteolytic organisms responsible for the feather digestion inside the feather bags were detected using BODIPY FL staining. Most proteolytic cells were rod shaped, although a few cocci of different sizes were also seen (Fig. 1a). BODIPY FL casein staining combined with DAPI staining showed that their relative abundance inside the feather bags (against the total bacteria stained with DAPI) increased from 1.7% at day 22 to 8.2% at day 55, and then to 14.3% at day 85 before decreasing to 11.6% at day 113, which correlated (R2 = 0.96) to the feather digestion rates detected, indicating that these were the KHOs responsible for the hydrolysis of the feather keratin (Fig. 2a).

Figure 1.

Fluorescent images of bacteria present in the feather bag fluid and mixed liquor of the anaerobic digesters inoculated with adapted swine manure. (a) A BODIPY fluorescence-labeled casein staining image showing that most of the keratin-hydrolyzing organisms with cell-surface associated keratinase are rods. (b) and (c) are color-combined FISH images from feather bag fluid samples and the mixed liquor samples, respectively using probe FIMs1029 (Cy3 labeled, set as red) and EUBmix (Fluorescein labeled, set as green) showing the yellow-colored rods (as a combination of red and green) hybridized with probe FIMs1029. (d) A color-combined fluorescent image using FISH image with probe FIMs1029 (green-colored), a BODIPY fluorescence-labeled casein staining image (red-colored) and a DAPI staining image (blue-colored) showing that the rod-shaped keratin-hydrolyzing organisms (light-blue colored cells, e.g. those labeled by arrows) stained positively with BODIPY fluorescence-labeled casein hybridized with probe FIMs1029 and stained by DAPI staining. The bar in each image represents 10 μm.

Figure 2.

(a) Correlation of the abundance of KHOs (square with black line) inside the feather bags to the feather digestion rate (circle with broken line) in the anaerobic digesters with feather addition. (b) The abundance of the KHOs in the mixed liquor of anaerobic digesters with (square with black line) and without feather addition. (circle with broken line). Standard deviations of the KHO percentage values were calculated based on at least six countings.

BODIPY FL staining was also applied to the mixed liquor samples from the feather and control ADs. Proteolytic organisms were found in all samples, and their cell morphologies were the same as those attached to the feather particles. Their relative abundances in communities from the feather ADs remained quite constant (between 3.1% and 3.5% of total cell number) for the first 60 days, and then gradually increased to 4.3% at day 85 before decreasing to 3.0% at day 113 (Fig. 2b). In contrast, their relative abundances in the control ADs gradually decreased from 3.4% of the total cell number at the beginning of the experiment to 1.4% at day 113 (Fig. 2b).

Identification of the KHOs using clone library construction and FISH probing

A 16S rRNA gene clone library was constructed from bacterial populations attached to the feather particles inside the feather bags. One hundred 16S rRNA clones of near complete length (c. 1300 bp) were sequenced. Seven were identified putatively as chimeras and removed. Of the 93 remaining sequences, 72 belonged to previously undescribed bacterial species sharing < 97% similarity with all other available 16S rRNA gene sequences held in the RDP database (date checked 28 July 2010). Most (57 clones) of the 93 clones belonged to members of the low G+C Firmicutes and the remainder (36 clones), with the exception of two clones, one of which was Gammaproteobacteria and the other a member of Synergistetes (Synergistia), were all members of the Bacteroidetes (Table 1).

Table 1. Classification (> 80 confidence level) of 16S rRNA gene clones in the feather bags fed into the anaerobic digester inoculated adapted swine manure using classifier
Phylogenetic classificationNumber of clones
  1. a

    U, represents unclassified.

  2. b

    N/A, indicates not applicable.

   Ua- PorphyromonadaceaeN/Ab1
   Incertae Sedis XIVAnaerovirgula1
   Incertae Sedis XITissierella5
   U-Incertae Sedis XIN/A11
   Clostridiaceae 1Clostridium5
   Clostridiaceae 2Alkaliphilus5

All the Firmicutes identified (Table 1) were members of the Clostridia, consisting of members from the genera Syntrophomonas in the Syntrophomonadaceae, Anaerovirgula from the family Incertae Sedis XIV, Tissierella, Anaerosphaera and Peptoniphilus all from the family Incertae Sedis XI, Cryptanaerobacter from the Peptococcaceae, Clostridium and Alkaliphilus from the Clostridaceae, Lutispora from the Gradilibacteraceae. One clone could be identified only as an unclassified Clostridia (class) and one as an unclassified Clostridiales (order). Many could be identified only at the family level, and these included members of the Porphyromonadaceae, Incertae Sedis XI, Ruminococcaceae, Peptostreptococcaceae, and Clostridiaceae. Similarly, all the Bacteroidetes clones identified belong to the Bacteroidia consisting of members from the genera Porphyromonas and Petrimonas in the Porphyromonadaceae. One Bacteroidetes was identified as an unclassified member of the Porphyromonadaceae.

Figures 3 and 4 show the phylogenetic relationships of these clones to their closest relatives. Agreeing with data from analyses based on the RDP database, the Firmicutes clones (Fig. 3) identified are phylogenetically diverse, clustering mostly at different distances with sequences from uncultured bacteria. In contrast, most Bacteroidetes clones (Fig. 4) form two closely related clusters with one belonging to the Porphyromonadaceae and the other to members of unclassified Bacteroidales. mothur analysis showed that the 93 clones represented 39 OTUs at a distance of 0.03. Thirty-five OTUs belonged to new bacterial species, sharing < 97% 16S rRNA sequence similarity with all other available sequences. Twenty-eight OTUs are Firmicutes and five are Bacteroidetes. The Proteobacteria and Synergisteres are represented by only one OTU. Species richness estimated by Chao 1 estimation at a 0.03 distance for the bacterial populations inside the feather bags is 60.

Before designing probes, 16S rRNA sequences were matched to all available SSU rRNA targeted probes in probeBase (Loy et al., 2003) using arb. Probe BAC303 perfectly matches the target site of all the Bacteroidetes clones (Fig. 4). The target site of probe Clost I is possessed by the five clones belonging to the Clostridiaceae 1 (Fig. 3). No currently available probe target sites were detected in other Firmicutes clones. Consequently, four new probes (Fig. 5) were designed to target the sequences retrieved in this study, including those of members of Firmicutes clusters (Fig. 3) of Incertae Sedis XI (probe CLOSs146), Clostridiaceae 2 (probe FIMs1029), Grocilibacteraceae (probe CLOSs654), and Ruminococcaceae (probe RUMs278). After optimization of their stringencies (see Materials and methods for more details) these probes were used to FISH.

Figure 3.

A distance tree (neighbor-joining) built with the Firmicutes 16S rRNA gene sequences obtained in this study (those with names in bold). The oligonucleotide probes designed or chosen to perfectly match with these clones are also shown. The bootstrap values (only those > 50% are shown) were calculated in Mega 4 based on 1000 resamplings. Scale bar represents one substitution per 10 nucleotides.

Figure 4.

A distance tree (neighbor-joining) built with the Bacteroidetes 16S rRNA gene sequences retrieved in this study (those with names in bold, the two sequences belonging to Proteobacteria and Synergistetes, respectively, are also shown in this tree). The coverage of probe BAC303 is also shown. The bootstrap values (only those > 50% are shown) were calculated in Mega 4 based on 1000 resamplings. Scale bar represents 1 substitution per 10 nucleotides.

Figure 5.

Name, sequence, specificity, and formamide concentration of the new oligonucleotide probes designed in this study. The names in the parentheses are name based on the nomenclature of Alm et al. (1996). aTM represents melting temperature (ºC) of the probes. bFA represents formamide concentration (%) of each probe used in FISH. The target sites of Clones SPAD22 and SPAD28 by probe FIMs1029 are also shown.

In all samples examined, probe BAC303 hybridized with a few cocci (data not shown), whereas probes CLOSs146, CLOSs654, and RUMs278 each hybridized with a few rod-shaped cells (data not shown). In contrast, probe FIMs1029 hybridized with many rods inside the feather bags (Fig. 1b) and in the mixed liquor (Fig. 1c). No cells were detected hybridizing with the probe Clost I in any of the samples examined. Pretreatment using HCl and different enzymes did not change the FISH results. Probes FIMs1029, CLOSs146, CLOSs654, RUMs278, and BAC303 were used to identify the KHOs present in the feather bags using BODIPY FL casein staining combined with FISH. Only probe FIMs1029 hybridized KHO cells (Fig. 1d), and subjective counting showed that about 80–90% of KHO cells in feather bag fluids collected at different times hybridized with it. This probe targets clones SPAD 6, 23, 24, 28, 32, 37, 38, 75, 81, and 107 (Fig. 3). All were identified by classifier as members of genus Alkaliphilus in the subfamily Clostridiaceae 2. Clones SPAD 6, 23, 24, 28, 32 were identified at a higher (> 80%) confidence level than others (70–79%).


In this article chicken feather degradation is reported in AD microbial communities inoculated with adapted swine manure at 25 °C. The only pretreatment used was to grind the feathers into smaller particles (diameter < 4 mm). Mass balance data suggest that 94% of the feathers added were degraded after 89 days. No similar studies have been reported previously, but the data presented here suggest that digestion of feathers in swine manure inoculated ADs is feasible for methane biofuel production. In addition, as the chemical structure of the feather keratin is similar to that of prions (Fraser et al., 1972), the results from this study suggested that AD systems adapted to feather digestion could be used to study prion degradation.

In the feather ADs, NH4-N levels increased substantially with feather digestion. Methanogens are known to have very low tolerances to elevated NH4-N levels (Kayhanian, 1994), and loss of their activity (~ 56.5%) has been reported at NH4-N concentrations ranging from 4.1 to 5.7 g L−1 (Chen et al., 2008). However, in the current study, CH4 generation could be detected throughout the operation of the feather ADs. Therefore, it appears that the NH4-N levels measured here did not affect markedly biogas production, which may result from the selection and adaption of methanogen populations able to survive under these conditions (Jarrell et al., 1987).

BODIPY FL casein staining was used to label the KHOs in this study. This is because enzymes in anaerobic digestion are mainly cell-associated (Angelidaki & Sanders, 2004). Its principle is that once quenched BODIPY dye-labeled casein is hydrolyzed by any proteases, fluorescent precipitates attached onto the surfaces of bacteria excreting these enzymes thus enabling single cell tagging of proteolytic bacteria. BODIPY FL casein has mainly been used to study protease activity of purified enzymes (e.g. Welder et al., 2002). Only quite recently has it been applied to study protease activity in situ in bacterial cells (Yoshioka et al., 2003) and in uncultured bacterial populations in communities of activated sludge wastewater treatment plants (Xia et al., 2007; Xia et al., 2008). BODIPY FL casein detects serine, metallo-, acid, and sulfhydryl proteases (Molecular Probes). Microbial keratinases are mainly metallo- or serine proteases (Lin et al., 1996; Riffel et al., 2003; Thys & Brandelli, 2006) and thus can be detected with this substrate at a single cell level of resolution.

Most (80–90%) of the KHOs in the feather bags hybridized with FISH probe FIMs1029 designed in this study. Their relative abundances correlated well with feather digestion rates. Consequently, they are considered to be the main KHOs responsible for the feather digestion in this study. The target site of probe FIMs1029 is possessed by 10 16S rRNA cloned sequences retrieved in this study (Fig. 3). These are only distantly related to the type strains of Alkaliphilus species, sharing < 75% similarity with A. oremlandii their closest relative, and other Alkaliphilus species. However, these clone sequences share > 94% similarity with each other (data not shown). Therefore, they almost certainly represent members of a new genus in the subfamily Clostridiaceae 2. These clones constitute three OTUs indicating that the KHOs may be diverse and consist of at least three different species. This view is supported by the observation that 10–20% of the KHO cells staining with BODIPY FL casein did not hybridize with the probe FIMs1029. Interestingly, some Alkaliphilus are known to be proteolytic bacteria in anaerobic ecosystems. A. transvaalensis isolated from a South African gold mine is a strictly anaerobic chemoorganotroph capable of utilizing proteinaceous substrates including yeast extract, peptone, tryptone, and casein (Takai et al., 2001). Similarly, A. crotonatoxidans isolated from a methanogenic environment is a strictly anaerobic bacterium capable of utilizing yeast extract, peptone, and tryptone (Cao et al., 2003).

Proteolytic bacteria positively stained with BODIPY FL casein were also detected in the inocula and the mixed liquor of the control ADs without added feathers. Most of them also hybridized with the FISH probe FIMs1029 (data not shown). This indicates that the KHOs responsible for feather degradation may also use proteins other than keratin, which may be present in the adapted swine manure used here, thus allowing these KHOs to grow there. The gradual decrease in their relative abundance in these control ADs suggests these other protein may be metabolized and rapidly disappear.

All that was known previously about KHOs had been derived from pure culture studies. The KHO isolates currently described are mainly members of the genus Bacillus, but other representatives are found among Fervidobacterium, Thermoanaerobacter, and Streptomyces (Gupta & Ramnani, 2006). Thus, our knowledge about their identity and ecophysiology in complex natural ecosystems like ADs remains limited. This study has shown for the first time that bacteria other than those cultured isolates may play important roles there in keratin degradation.


Financial support for Y.X was through a peer review project and the specified risk material disposal project of Agriculture and Agri-Food Canada. We thank Denis Deslauriers, Frederic Beaudoin, and Gilles Grondin for their excellent technical support.