Biomass-supported palladium catalysts on Desulfovibrio desulfuricans ...

6 downloads 0 Views 241KB Size Report
2005 25th-29th September 2005; Cape Town, S. Africa. Published by 16th Int. Biohydrometallurgy ... Platinum Review. Published by Johnson Matthey Plc.,.
ARTICLE Biomass-Supported Palladium Catalysts on Desulfovibrio desulfuricans and Rhodobacter sphaeroides Mark D. Redwood, Kevin Deplanche, Victoria S. Baxter-Plant,y Lynne E. Macaskie Unit of Functional Bionanomaterials, School of Biosciences, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK; telephone: 44-121-414-5889; fax: 44-121-414-5925; e-mail: [email protected] Received 11 July 2007; revision received 26 September 2007; accepted 8 October 2007 Published online 29 October 2007 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/bit.21689

ABSTRACT: A Rhodobacter sphaeroides-supported dried, ground palladium catalyst (‘‘Rs-Pd(0)’’) was compared with a Desulfovibrio desulfuricans-supported catalyst (‘‘Dd-Pd(0)’’) and with unsupported palladium metal particles made by reduction under H2 (‘‘Chem-Pd(0)’’). Cell surface-located clusters of Pd(0) nanoparticles were detected on both D. desulfuricans and R. sphaeroides but the size and location of deposits differed among comparably loaded preparations. These differences may underlie the observation of different activities of Dd-Pd(0) and Rs-Pd(0) when compared with respect to their ability to promote hydrogen release from hypophosphite and to catalyze chloride release from chlorinated aromatic compounds. Dd-Pd(0) was more effective in the reductive dehalogenation of polychlorinated biphenyls (PCBs), whereas Rs-Pd(0) was more effective in the initial dehalogenation of pentachlorophenol (PCP) although the rate of chloride release from PCP was comparable with both preparations after 2 h. Biotechnol. Bioeng. 2008;99: 1045–1054. ß 2007 Wiley Periodicals, Inc. KEYWORDS: catalysis; palladium; chlorophenols; polychlorinated biphenyls; PCB; reductive dehalogenation; Desulfovibrio desulfuricans; Rhodobacter sphaeroides

Introduction The application of Rhodobacter sphaeroides in the recovery of palladium from solution as supported metallic nanoparticles, and in catalysis using the resulting palladized biomass, was investigated in comparison with Pd(0) biomanufacy Deceased. Correspondence to: L.E. Macaskie Contract grant sponsor: BBSRC Contract grant number: BB/C516128/1

ß 2007 Wiley Periodicals, Inc.

tured using Desulfovibrio desulfuricans. The use of sulfatereducing bacteria (SRB) for palladium catalyst production results in a highly active ‘‘bionanocatalyst’’ containing a subpopulation of Pd(0) nanoparticles (approximately 5 nm), detected magnetically (Mikheenko, 2004; Mikheenko et al., 2005). However catalyst production is limited by the need for washing to remove H2S (the product of dissimilatory sulfate metabolism), which is a potent catalyst poison and hence an alternative route using another organism would cut the number of processing steps required from the Pd(II)reducing bacteria to dried, ground catalyst. Industrially, palladium is extracted from mixed ores with other platinum group metals (PGM), for use in automotive catalysts, jewellery, dental amalgams and as a catalyst in various industrial reactions (e.g., hydrogenations). Efficient use of PGM and their recovery are essential due to the increasing price of precious metals and finite ore resources (Kendal, 2006). Palladium can be reclaimed from wastes (e.g., spent automotive catalytic converters and other scrap) using aqua regia to oxidize Pd(0) to Pd(II) in the form of the [PdCl4]2 anion to create a Pd(II)-rich leachate (Mabbett et al., 2006; Yong et al., 2003). In order to regenerate Pd(0), a reducing step is required. Many bacteria can reduce metallic ions to lower valence species and, sometimes, to the metallic state by dissimilatory metal reduction, via which simple organic substrates are oxidized and metals can act as the primary or sole terminal electron acceptor. In the case of SRB, the ability to grow via dissimilatory metal reduction is described in only one strain to date (Tebo and Obraztsova, 1998) although many examples of metal reduction are documented (Lloyd, 2003; Lloyd et al., 2003, 2005). The rate of reduction of Pd(II) was significantly increased in the presence of cells of Desulfovibrio spp. (Baxter-Plant et al., 2003; Yong et al., 2002a, 2003). In this process the cells became ‘‘palladized’’ (coated with a layer of Pd(0) particles) and dry palladized biomass (Bio-Pd(0)) was shown to be an active catalyst using various test reactions (Mabbett et al.,

Biotechnology and Bioengineering, Vol. 99, No. 5, April 1, 2008

1045

2001; Mikheenko, 2004; Yong et al., 2002a,b, 2003). The catalytic activity was seen also in ‘‘Bio-Pd(0)’’ sourced from waste leachate (Mabbett et al., 2006). The ability of biomass to remove Pd(II) from strong acids (Yong et al., 2003) and wastes (Creamer et al., 2006; Mabbett et al., 2006) is realized by first prepalladizing the cells lightly via hydrogenase activity under physiologically permissive conditions and then by using the nascent Pd(0) clusters to reduce further Pd(II) from the acidic solution autocatalytically (Mabbett et al., 2006). Therefore, the initial patterning of Pd(0) clusters in the ‘‘live’’ biological step determines the formatting of the nanoparticles, preventing their agglomeration, and hence this is vital to the activity of the finished product. In the process of Bio-Pd(0) catalyst preparation, [PdCl4]2 ions initially biosorb onto bacterial cells (de Vargas et al., 2004), coordinating to amine groups (de Vargas et al., 2005). Pd(II) reduction is initiated by cellular hydrogenase activity (de Vargas et al., 2004, 2005; Mikheenko, 2004; Mikheenko et al., 2005) and completed upon the provision of excess reductant (e.g., H2 or formate) via the ability of Pd(0) to catalyze reduction of Pd(II) abiotically (Mabbett et al., 2004; Yong et al., 2002a,b). Hence, Bio-Pd(0) of a known Pd:biomass loading can be produced by reducing a known mass of Pd(II) in the presence of a known mass of cells. Biomass-supported palladium catalysts could have potential environmental applications in ex-situ remediation. For example, Bio-Pd(0) prepared using D. desulfuricans catalyzed the reduction of toxic Cr(VI) to Cr(III) (Mabbett et al., 2001, 2006). Bio-Pd(0) was also effective in the reductive dehalogenation of chlorinated aromatic compounds (BaxterPlant et al., 2003; Harrad et al., 2007; Mabbett et al., 2001). This group of problematic environmental contaminants includes pentachlorophenol (PCP: used as a wood preservative and as a pesticide), and highly toxic polychlorinated biphenyls (PCBs), which were industrially prevalent due to their stability and thermal properties and now persist in the environment (Anon, 1999). The concept of using zerovalent metals is not new: zero-valent iron is well known to catalyze the reductive dehalogenation of PCBs (Chuang et al., 1995; Li et al., 2006) and palladized Fe was also highly effective (Kim et al., 2004), with Pd(0) being less susceptible to oxide formation than Fe(0). Bio-Pd(0)-mediated reductive dehalogenation was initially reported to liberate chloride ion from PCBs (Baxter-Plant et al., 2003, 2004) and subsequent studies confirmed the production of less heavily substituted intermediate species, and biphenyl (Harrad et al., 2007; Windt et al., 2005). Microbial (i.e., enzymatic) degradation of chlorinated aromatic compounds usually follows a prerequisite dehalogenation because the halide substitutions can cause steric hindrance of enzymatic attack on the aromatic rings (Lee and Carberry, 1992). Although microbially mediated breakdown occurs slowly in the environment, microorganisms can perform both dehalogenation and degradation of halogenated pollutants and microbial methods are receiving extensive attention (Bedard and Quensen, 1995; Bedard,

1046

Biotechnology and Bioengineering, Vol. 99, No. 5, April 1, 2008

2003; Wiegel and Wu, 2000). The present approach combines the utility of a biomass-immobilized catalyst with the robustness, high reactivity, and fast rate of a chemical catalyst. It is stressed that the bacteria are non-living in their application and although biobased in the initial Pd-cluster formation step the finished Bio-Pd(0) is essentially a chemical catalyst (Creamer et al., 2007). Use of Bio-Pd(0) represents a scalable technology for nanoparticle manufacture, particularly attractive since the latter can be made via biorecovery of Pd(II) from wastes (Mabbett et al., 2006). The recent emergence of environmental nanotechnology has prompted concern about the fates and effects of unsupported nanoparticles if released into the environment accidentally or by design (Li et al., 2006). Tethering of nanoparticles onto a bacterial carrier could alleviate some of these concerns while retaining full activity of the nanocatalyst. Microbial cell surface chemistry will influence the initial biosorption and subsequent enzymatically mediated reduction of Pd(II), thus affecting the patterning and hence, potentially, the catalytic properties of the resultant BioPd(0). Thus, within a single strain (D. fructosovorans) a mutant deficient in its periplasmic hydrogenases relocated its Pd(0) deposits to the cytoplasmic membrane—the location of the remaining hydrogenases (Mikheenko et al., 2005) with the effect of increasing the catalytic activity against Cr(VI) (Rousset et al., 2006). It was shown previously that Bio-Pd(0) was a more active catalyst than Pd(0) powder made from Pd(II) reduced chemically under H2 (Chem-Pd(0)) and this was attributed to the presence of nanoparticulate Pd(0) (Yong et al., 2002a). Hence, the first objective of this study was to biofabricate a new form of Bio-Pd(0), made using R. sphaeroides (a non-H2S-producing bacterium), in comparison with Chem-Pd(0). The second objective was to compare the catalytic efficacy of R. sphaeroides-supported Bio-Pd(0) (‘‘Rs-Pd(0)’’) and D. desulfuricans-supported Bio-Pd(0) (‘‘Dd-Pd(0)’’), with respect to the promotion of H2 release from hypophosphite (a test reaction involving hydrogen) and the reductive dehalogenation of chlorinated aromatic compounds (a test hydrogenolysis reaction (Nishimura, 2001) involving release of chloride ion into the aqueous solution). The third objective was to relate the locations of cellular Pd(0) deposits to measurements of catalytic activity in the context of potential factors known to be involved in metal bioreduction, the initiation of Pd(0) particle formation and in nanoparticle growth. Purple non-sulfur (PNS) bacteria, here focusing on R. sphaeroides, can exhibit intrinsically high resistance to various metallic species, for example, chromate, tellurite, selenite, and rhodium sesquioxide (Borsetti et al., 2003; Kessi et al., 1999; Moore and Kaplan, 1992; Nepple et al., 2000; Van Praag et al., 2002; Yamada et al., 1997). In PNS bacteria reductive metal precipitation is broadly considered to serve as a detoxification mechanism, although in the case of metalloid oxyanions, it may be considered dissimilatory,

being linked to the disposal of excess reducing power from the photosynthetic apparatus, located on intracytoplasmic vesicles (Moore and Kaplan, 1994; Van Fleet-Stalder et al., 1997), along with the electron transport chain, uptake hydrogenases and respiratory nitrite, sulfite and glutathione reductases (Borsetti et al., 2003; Kessi, 2006; Kessi et al., 1999; Moore and Kaplan, 1992; Van Praag et al., 2002; Yamada et al., 1997). These factors and, indeed, any proteins having exposed reactive—SH groups could constitute candidate nucleation foci in R. sphaeroides. Pd(0) nanoparticle deposition effectively ‘‘tags’’ the site of Pd(II) reduction in vivo (Mikheenko et al., 2005) and the final objective was to gain insight into the localization of ‘‘Pd(II) reductase’’ activity in R. sphaeroides.

Materials and Methods Microorganisms and Culture Conditions D. desulfuricans (NCIMB 8307) was maintained and cultured as described previously (Yong et al., 2002a). R. sphaeroides O.U.001 (DSMZ 5864) was held in stock at 808C (in 15% glycerol v/v), revived on nutrient agar (308C) and cultured in filled sealed bottles under fluorescent illumination (39.5 mM photons m2 s1 measured using a PAR light meter SKP200, Skye Instruments, Llandrindod, Wales) at 308C using SyA medium (Hoekema et al., 2002).

palladium salt; substitution of the palladium amine salt gave a preparation of lower catalytic activity: I. Mikheenko and L.E. Macaskie, unpublished work), minimizing the background in Cl release assays (see below). For palladization, aliquots of cell concentrate and Pd(II) solution (Na2PdCl4; Sigma-Aldrich, Poole, Dorset, UK) were mixed to produce the desired mass ratio. For example, in order to produce Bio-Pd(0) loaded at 25% Pd(0) (w/w), 0.1 g Pd(II) and 0.3 g cell dry weight were mixed in 0.01 M HNO3, pH 2. Mixtures (50–100 mL) were sealed in 100 mL serum bottles with butyl rubber stoppers and aluminum tear seals, degassed under vacuum (5 min), purged with oxygen-free N2 (10 min) and incubated statically in the dark (308C, 60 min) to allow biosorption of Pd(II) before purging with H2 (15 min) after which complete removal of Pd(II) was confirmed by assay (below). The preparations were harvested by centrifugation, washed three times in sterile MOPS buffer (above) and once in acetone, dried at 608C and ground to give a dry black powder. Chemically reduced Pd(0) (Chem-Pd(0)) was prepared in parallel without bacterial cells and with complete Pd(II) reduction requiring 60 min under H2. Dried material was ground using a pestle and mortar before catalytic testing. For each type of catalyst, three independent batches were prepared and tested; data are means  SEM.

Assay of Pd(II) Determination of Dry Weight Biomass concentration (mg dry weight mL1) was calculated from the optical density (660 nm) with reference to a conversion factor, determined in triplicate by recording optical densities from dense cultures after various dilutions in deionized H2O. Cultures were washed twice in isotonic saline (8.5 g L1 of NaCl, pH 7; 2,400g, 20 min, 48C, 50 mL) before drying at 608C to constant mass.

Before harvesting palladized biomass the complete reduction of Pd(II) was confirmed by reading the A420 of sample supernatants in a variable wavelength spectrophotometer (Ultraspec III, Pharmacia Biotech, Little Chalfont, Buckinghamshire, UK). This assay method was validated previously using the SnCl2 method, and polarographically (Mikheenko, 2004).

Electron Microscopy Preparation of Bio-Pd(0) The procedure was based on that described previously (Mabbett et al., 2001; Yong et al., 2002a) except that all cell washing was done in buffer containing no added chloride. Bacterial cells were palladized immediately after harvesting from the mid-logarithmic phase of growth by centrifugation (11,900g, 10 min). High concentrations of Cl were shown to interfere with the initial biosorption of Pd(II) during biomass palladization (de Vargas et al., 2004), hence the biomass was washed during harvesting three times in 20– 50 mL buffer (20 mM sodium morpholinopropanesulfonic acid (MOPS)-NaOH (pH 7)). The MOPS buffer, made using deionized water and analytical grade reagents, contained less than 0.5 mg Cl L1 (the limit of assay sensitivity). After palladization, preparations were washed a further three times (to remove the Cl introduced in the

Samples of palladized biomass were washed as above, omitting the acetone wash, and prepared for examination of cell sections by transmission electron microscopy (TEM) as described previously (Baxter-Plant et al., 2003).

Evaluation of Catalytic Activity by the Hypophosphite Test Reaction The method was developed from that described previously (Yong et al., 2002a). For assay, each reaction contained 0.5 mg Pd(0) as a variable mass of total material dependent upon the relative Pd(0) and biomass fractions. The preparations were suspended in 10 mL of 100 g L1 NaH2PO2 buffered with MOPS-NaOH (0.5 M, pH 8), at 258C. After the onset of gas release the volume of H2 generated over 30 min was measured by displacement of

Redwood et al.: Biomass-Supported Palladium Catalysts Biotechnology and Bioengineering

1047

water using an inverted measuring cylinder. The pH of the reaction mixtures was unchanged after 30 min.

Assay for Catalytic Dehalogenation of Chlorinated Aromatic Compounds The reductive dehalogenation of chlorophenols and PCBs catalyzed by Bio-Pd(0) was demonstrated previously (Baxter-Plant et al., 2003; Harrad et al., 2007). As an electron donor for catalysis in test reactions, formate was preferred over H2 since its concentration in the aqueous phase can be more accurately controlled; metallic Pd(0) on the bacteria catalyzes the cleavage of HCOOH to CO2 and H2 and then effects the homolytic fission of H2, holding highly reactive H. within the Pd crystal (Rhodin, 1979). The substrates tested are shown in Table I. Chlorophenols were purchased from Aldrich (Fancy Rd, Poole, Dorset, UK) and PCBs from QMX Laboratories Ltd. (Bedford St, Thaxted, Essex, UK). For assay, each reaction contained 2 mg of test catalyst (i.e., total material: Pd(0) and biomass component, Pd: biomass loadings as specified), resuspended in 9 mL sterile MOPS-NaOH buffer (20 mM, pH 7) and 1 mL aromatic substrate in hexane carrier. This was done to facilitate the recovery of the catalyst from unreacted PCB/chlorophenol and reaction intermediates (see Discussion Section), separating the chloride (aq.) for measurement. After shaking and settling of the hexane-in-water suspension (5 min) a 1 mL sample was taken from the aqueous fraction to determine the initial Cl background, which was on average 0.47 mg Cl L1 (0.10 SEM). The reaction was initiated by the addition of 1 mL 1 M sodium formate (pH 7), the suspension was shaken and further samples were taken from the aqueous phase at suitable intervals, centrifuged (13,000g, 4 min), and supernatant was transferred into cuvettes. Reductive dehalogenation was monitored by the release of Cl, as determined by the mercury (II) thiocyanate Table I.

method (Mendam et al., 2000). A standard curve was prepared using NaCl in MOPS buffer. Assay interference by the organic components and spontaneous Cl release from biomass or aromatic substrate was excluded using catalystfree controls and aromatic substrate-free controls supplemented with hexane alone. In the case of PCB 138 negligible Cl release was observed (see later) and this served as a negative control to show that PCB-in-hexane did not promote spontaneous Cl release from the biomass and also confirmed the lack of significant background Cl in the assay system (see later).

Results Examination of the Palladized Biomasses Using Electron Microscopy R. sphaeroides and D. desulfuricans were successfully palladized without modification to the procedure, as shown by the appearance of black Pd(0) deposits under TEM (Fig. 1). Both species formed Pd(0) deposits in the periplasm and showed occasional outgrowth structures, indicating a probable periplasmic initiation of Pd(II) reduction followed by Pd(0) crystal growth and eruption beyond the cell surface. Several differences between Dd-Pd(0) and Rs-Pd(0) were noted (Fig. 1). At a Pd(0) loading of 1% (w/w) surfacelocated clusters of Pd(0) on R. sphaeroides were infrequent but visible, whereas no clusters were visible on D. desulfuricans, suggesting that the Pd(0) deposits were below the limit of detection. At a Pd(0) loading of 5% Pd(0) clusters were visible in the periplasm and at the cell surface of D. desulfuricans, while Rs-Pd(0) clusters were retained periplasmically (Fig. 1). At a Pd(0) loading of 25%, clusters of Pd(0) nanoparticles appeared smaller and fewer overall on R. sphaeroides than on D. desulfuricans and, while the latter showed a relatively even

Chlorinated aromatic compounds used in catalytic dehalogenation testing. % Chloride release

PCB

Substitution

28 52 101 118 138 153 180 2-chlorophenol Pentachlorophenol

0

2,4,4 2,20 ,5,50 2,20 ,4,5,50 2,30 ,4,40 ,5 2,20 ,3,4,40 ,50 2,20 ,4,40 ,5,50 2,20 ,3,4,40 ,5,50

log Kow

Concn.* (mM)

Concn.* of Cl (mg/L)

Dd-Pd(0)

Rs-Pd(0)

Chem-Pd(0)

5.71 5.79 6.39 6.57 6.73 6.80 7.21

0.308 0.274 0.123 0.123 0.111 0.111 0.051 5.00 0.50

32.76 38.85 21.80 21.80 23.61 23.61 12.66 177.5 88.75

39.9  8.1 2.6  0.1 18.0  4.5 19.7  6.1 5.0  0.5 10.8  2.4 18.9  6.0 20.4  1.6 5.3  1.8

4.3  2.2 2.7  1.1 4.3  1.2 NS NS NS 3.6 1.8 10.9  0.5 19.6  2.2

17.0  3.1 0.9  0.1 NS NS NS 2.3  0.1 2.1  0.2 6.7  2.7 0.5  3.4

Data were calculated from the increase in Cl concentration after 24 h (PCBs) or 60 min (chlorophenols). NS: no significant chloride detected. Data are means  SEM from at least three independent experiments. Note that data represent the proportion of the total potential Cl release, rather than the measured quantities of Cl release. Hexane/water partition coefficients were not known. Kow: octanol/water partition coefficient; log Kow values taken from (Makino, 1998) are shown for conciseness; aqueous solubility is indicated by 1/Kow. Note that since the Pd:biomass ratio was 1:3, the data for Bio-Pd(0) can be multiplied by four for direct comparison with Chem-Pd(0) on a mass of Pd(0) basis. *The substrates were used as hexane in water suspensions. The concentration shown is that in the 10 mL of test mixture; the actual concentration in the aqueous phase was not determined.

1048

Biotechnology and Bioengineering, Vol. 99, No. 5, April 1, 2008

Figure 1. TEM sections of D. desulfuricans (D) and R. sphaeroides (R) palladized to 0%, 1%, 5%, and 25% Pd(0) w/w. Scale-bar in D 0% applies to all eight main frames. Inset: D. desulfuricans: formation of outer membrane vesicles during palladization; OMVs are indicated with open arrows. Filled arrow: Pd(0)-particle in association with outer membrane materials. Inset: R. sphaeroides: detail of occasional heavy cell surface deposit visible in approximately 1 in 20 cells.

dispersion, the R. sphaeroides cellular deposits were more heterogeneous, with occasional very large deposits (Fig. 1, inset) visible on approximately 1 in 20 cells. While the overall Pd:biomass ratio was controlled, this skewed distribution of particle sizes resulted in an overall lighter deposition of Pd(0) for the majority of the R. sphaeroides cells. Evaluation of Catalytic Activity Using the Hypophosphite Test H2 evolution from hypophosphite was used as an initial simple indicator of catalytic activity for various preparations (Fig. 2). Bio-Pd(0) loaded at 1%, 5%, and 25% Pd(0) w/w was tested alongside Chem-Pd(0) (100% Pd(0) w/w) and non-palladized biomass (0% Pd(0) w/w). No catalytic activity was seen at 0% and 1% Pd(0) loading (Fig. 2). BioPd(0) preparations loaded at 5% or 25% on R. sphaeroides or D. desulfuricans were significantly more catalytically active

than Chem-Pd(0) ( P < 0.01 in all four comparisons). The highest rate of H2 release was seen using Dd-Pd(0) loaded at 25% Pd(0) w/w, showing more than four times the rate using Chem-Pd(0) and approximately double the rate for the corresponding Rs-Pd(0). Whether loaded at 5% or 25% Pd(0) w/w, Dd-Pd(0) was significantly more active than Rs-Pd(0) ( P ¼ 0.011, P ¼ 0.023, respectively). The Pd(0) loading (5% or 25%) did not significantly affect catalytic activity of Rs-Pd(0) or Dd-Pd(0), which suggests that a similar proportion of total Pd(0) was catalytically active in this range of loadings onto either type of cell. A Pd(0) loading of 25% was chosen for subsequent investigation. Evaluation of Catalytic Activity by Reductive Dehalogenation of Chlorinated Aromatic Compounds Rs-Pd(0), Dd-Pd(0), and Chem-Pd(0) were tested for their ability to catalyze reductive dechlorination of PCBs and

Redwood et al.: Biomass-Supported Palladium Catalysts Biotechnology and Bioengineering

1049

Figure 2. Rates of H2 release via hypophosphite by various catalytic preparations. All reactions contained 0.5 mg Pd(0). Data are means and standard error of the mean from four experiments. No H2 was generated using unpalladized biomass or Bio-Pd(0) loaded at 1% Pd(0) w/w.

chlorophenols (2-chlorophenol and PCP) using formate as the electron donor. No Cl release was promoted by nonpalladized biomass or by any Pd(0) in the absence of electron donor. Several tests gave no significant chloride release, which serves as a negative control (Table I), ruling out the spontaneous appearance of chloride from substrate or hexane-induced leaching of internal chloride from the cells.

Chloride Release From Polychlorinated Biphenyls (PCBs) The hypophosphite test indicated that at 25% Pd(0) loading Dd-Pd(0) was 2.3-fold more effective than Rs-Pd(0) (Fig. 2). However, this relationship was not maintained in the case of PCBs (Table I) and there was no consistent comparator between Dd-Pd(0) and Rs-Pd(0). In no case did the catalytic activity of Rs-Pd(0) exceed that of Dd-Pd(0) and only Dd-Pd(0) was able to liberate Cl from PCBs 118 and 153. Chem-Pd(0) was unable to dehalogenate markedly any PCB except PCB 28 (2,4,40 -trichlorobiphenyl) which was the most water-soluble of the PCBs used (Makino, 1998) and was also the congener most susceptible to attack by both types of Bio-Pd(0).

and 33% of this, respectively at equal Pd(0) loadings (25% Pd(0) w/w). The proportion of potential chloride release (Table I) was calculated as the measured increase in the concentration of soluble Cl compared to the concentration of bound Cl initially present in the form of chlorophenols or PCBs. On a mass of Pd(0) basis the differences between BioPd(0) preparations and Chem-Pd(0) can be corrected by a factor of 4, that is, the Cl release for Dd-Pd(0) and Rs-Pd(0) were 12.2 and 6.5 times higher than for Chem-Pd(0), respectively. Figure 3 shows the Cl released from PCP after 1 h, at which point the only extensive dehalogenation was observed using Rs-Pd(0), which was comparable to that seen using 2-CP. For Dd-Pd(0) and Chem-Pd(0), the onset of Cl release occurred only after 1–2 h, while Rs-Pd(0) catalyzed a similar overall extent of dechlorination, but with the first Cl release detected within 40 min. After 2 h the release of Cl by all three preparations was comparable. On the basis of Cl release per mass of Pd(0) after 2 h, both Bio-Pd(0) preparations were over four times as active as Chem-Pd(0). The release of Cl via Chem-Pd(0) ceased after 4 h whereas the Bio-Pd(0) preparations continued to liberate Cl at slower, broadly comparable rates. One important observation was the biphasic behavior of Cl release by the BioPd(0)s over the first 2 h. Rs-Pd(0) showed an initially rapid Cl release followed by a constant slower rate, while with Dd-Pd(0) the converse applied. These results were highly significant but no significant differences were apparent between the Bio-Pd(0)s after 2 h.

Discussion The cell-surface localization of Pd(0) deposits was similar on cells of R. sphaeroides and D. desulfuricans at 5% and 25%

Chloride Release From 2-Chlorophenol (2-CP) and Pentachlorophenol (PCP) Chloride release from 2-CP and PCP was more rapid than with PCBs. Constant rates of Cl release from 2-CP were observed over 60 min, with no delay before onset, using all three catalysts (not shown). The highest Cl release (36.14 mg L1 h1) was observed using Dd-Pd(0). The corresponding rates using Rs-Pd(0) and Chem-Pd(0) were 53%

1050

Biotechnology and Bioengineering, Vol. 99, No. 5, April 1, 2008

Figure 3. Reductive dehalogenation of pentachlorophenol (PCP). Pd(0) catalysts (solid lines): (^) 25% Dd-Pd(0) (*) 25% Rs-Pd(0) (D) Chem-Pd(0). Pd(0)-free controls (dashed lines): (^) D. desulfuricans biomass alone. (*) R. sphaeroides biomass alone. Data are means and SEM from at least three independent experiments. Initial Cl backgrounds (on average 0.47 mg Cl L1  0.102 SEM) were subtracted. Note that since the Pd:biomass ratio was 1:3, the data for Bio-Pd(0) can be multiplied by four for direct comparison with Chem-Pd(0) on a mass of Pd(0) basis.

loading (Fig. 1). A periplasmic origin for Pd(0) crystals in D. desulfuricans was shown previously to derive from an involvement of periplasmic hydrogenases in the reduction of sorbed Pd(II) (Mikheenko et al., 2005) and hydrogenase activity was confirmed throughout the incubation period by assay (Mikheenko, 2004). The TEM study (Fig. 1) provided evidence for a similar origin for Pd(0) crystals in the periplasm of R. sphaeroides but the specific involvement of hydrogenase or any specific reductase was not tested. No evidence was seen for deposition of Pd(0) within the cytoplasm, largely ruling out a contribution of intracytoplasmic photosynthetic components or other intracellular enzymes and implicating a periplasmic-based mechanism (see below). In D. fructosovorans (Mikheenko et al., 2005) and in E. coli (K. Deplanche, I. Mikheenko, and L.E. Macaskie, unpublished work) nucleation foci are primarily on or near the hydrogenase itself as shown by analysis of mutants of both organisms deficient in one of more of their hydrogenases (Mikheenko et al., 2005; Rousset et al., 2006; K. Deplanche et al., unpublished work). The higher frequency of periplasmic and surface-orientated Pd(0) particles on D. desulfuricans and the generally greater catalytic activity, in comparison with R. sphaeroides (small nanoparticles in addition to some larger ones erupting from the cell: Fig. 1) would suggest a potentially higher incidence of nucleation foci in D. desulfuricans than in R. sphaeroides, leading to more numerous deposits of smaller size, whereas fewer initiation foci in R. sphaeroides would promote the formation of fewer, larger clusters derived from the same initial mass of Pd(II). This was also observed in D. fructosovorans where removal of the periplasmic hydrogenases gave larger Pd(0) deposits at the location of the remaining hydrogenase enzymes (Mikheenko et al., 2005). Since in the initial biosorption stage coordination of Pd(II) to amine groups was observed (de Vargas et al., 2005) any redox enzyme might initiate the formation of Pd(0) clusters but it should be noted that in this study no apparent reduction occurred in either case before the addition of exogenous reductant (H2). Some endogenous reduction of Pt(IV) to Pt(II) by D. desulfuricans was suggested by X-ray photoelectron spectroscopy but results for Pd(II) were equivocal (de Vargas et al., 2005). Hence, a small degree of reduction may be instigated biologically prior to the addition of exogenous electron donor. It has been suggested that the intracytoplasmic membrane-associated photosynthetic apparatus of R. sphaeroides could provide an abundant source of reductant during photoheterotrophic metabolism and, in the case of metalloid oxyanions, intracellular deposits were visible (Borsetti et al., 2003; Kessi et al., 1999). However in this study cytoplasmic deposits of Pd(0) were not detected (Fig. 1); the hypothetical cytoplasmic reductase(s) may not have been active, since resting cells were challenged with Pd(II) in darkness, and cells were not pre-grown in the presence of Pd(II); any de novo protein synthesis as an adaptive response would probably not occur during the short incubation of resting

cells with Pd(II). The initiation of Pd(0) particle formation near the R. sphaeroides cell surface would probably preclude the direct involvement of intracellular or cytoplasmic mechanisms such as the photosystem, along with nitrogenase, which evolves H2 in the presence of light and in the absence of NHþ 4 ion (Vignais et al., 1985). Oxidoreductases such as NiFe hydrogenases, intrinsic to the cytoplasmic membrane, function anaerobically to oxidize H2 at the expense of electron acceptors such as nitrate (Vignais et al., 2001; Vignais and Colbeau, 2004). Therefore, these would be correctly located to bioreduce [PdCl4]2 ion, resulting in the observed patterning of Pd(0) deposits on R. sphaeroides (Fig. 1). In mutant strains of D. fructosovorans deficient in soluble periplasmic hydrogenases, the resultant Pd(0) particles were re-located from the periplasm to the cytoplasmic membrane, but the kinetics of Pd(II) reduction were unaffected (Mikheenko, 2004). Therefore the absence of known periplasm-orientated hydrogenase in R. sphaeroides would not necessarily exclude the involvement of membranebound, inwardly orientated hydrogenase. Use of biomass as a support in catalyst preparation augmented the catalytic activity of the resulting dead cell/Pd hybrid catalyst as compared to Pd(0) alone, which indicates a high availability of palladium catalytic surface due to the small crystal sizes (Macaskie et al., 2005; Yong et al., 2002b). A further contributing factor may be the increased dispersion of the material and stabilization of nanoparticles on the biomass, overcoming the natural tendency of nanoparticles to agglomerate. Inspection of Figure 1 (e.g., DdPd(0), 5%) shows that the Pd(0) deposits were composed of smaller clusters of sizes measured at approximately 3–10 nm which is in accordance with a subpopulation of nanoclusters of size approximately 5 nm as calculated from magnetic measurements (Mikheenko, 2004; Mikheenko et al., 2005). The hypophosphite test indicated that the Pd(0) loading affected the activity of the Pd(0) tested. Using a Pd(0) loading of 1% (w/w), the release of H2 was undetectable for both Dd-Pd(0) and Rs-Pd(0), whereas the effectiveness was not significantly different between Pd(0) loadings of 5% and 25% (w/w). As very low catalytic activity was observed previously when a loading of 1% was used (Creamer et al., 2007), it is possible that with a loading of 1%, the growth of Pd(0) particles was restricted by a limiting availability of Pd(II), preventing eruption onto the cell surface, as shown by the TEM study (Fig. 1), making the Pd(0) catalyst less easily accessible to the reactants. The lack of significant differences between preparations loaded at 5% and 25% Pd(0) (w/w) (i.e., no improvement using a heavier loading; see above) is consistent with previous findings as a previous study (Creamer et al., 2007) showed that Bio-Pd(0) prepared using Bacillus sphaericus was a more effective hydrogenation catalyst at a lighter loading onto the biomass. A survey of the optimal Pd(0) loadings on different biomasses is the subject of current investigations. For in situ environmental remediation the use of Bio-Pd(0) would be unattractive on economic grounds. However, a Pd-catalytic approach to the remediation of soil

Redwood et al.: Biomass-Supported Palladium Catalysts Biotechnology and Bioengineering

1051

washings or contaminated water could be attractive as compared to purely microbial methods. For example, 2-CP was removed using acclimated sludge at a rate of approximately 0.046 mg Cl L1 h1 (Boyd and Shelton, 1984) while, in this study, Rs-Pd(0) and Dd-Pd(0) catalyzed the dehalogenation of 2-CP at rates of 420- and 786-fold faster rates, respectively. Recently, a pure culture of Pseudomonas stutzeri was reported to degrade 2,4-dichlorophenol at a rate equivalent to 0.087 g Cl g catalyst1 h1 (Sahinkaya and Dilek, 2006). This is approximately 100- to 1,000-fold slower than biomass-supported Pd(0) catalysts, shown to be capable of dehalogenating 2-CP and PCP at rates in the range 9.3–72.3 g Cl g Pd1 h1 (this study). Microbial dehalogenation (oxidatively or reductively) is usually prerequisite to further degradation; in these comparisons, the biological methods refer not only to dehalogenation but also to degradation and it is assumed that the former is the rate-limiting step. In contrast the BioPd(0) method appeared (by GCMS analysis of PCB 28) to give only lesser chlorinated intermediates and biphenyl (Harrad et al., 2007) and an additional microbial treatment would probably be required for complete degradation. Different chlorinated aromatic compounds have different industrial applications and environmental occurrences and a catalyst targeting a specific compound could be potentially beneficial. These studies show that in the case of PCBs the D. desulfuricans-supported catalyst was superior in all cases whereas, in the case of PCP, although the activity of both biomass-supported catalysts was comparable overall after an extended period, the Rs-Pd(0) could offer an advantage for the treatment of waste streams where a rapid flow rate might require a short flow residence time (Fig. 1). The reason for this result is not clear, nor is it apparent why the initial rapid rate slows after 2 h, but the difference between the two catalysts is significant and underlies that the specific mechanisms by which dehalogenation is effected (see below) would warrant further study. Other studies using atomic force microscopy have shown that 5 nm Pd(0) nanoparticles (made by selected size cluster pinning in vitro) tend to accumulate PCB substrate (I.P. Mikheenko, personal communication) and partial dehalogenation products, requiring a hexane wash to scavenge the latter from the catalyst (Harrad et al., 2007). As a general trend, increasingly substituted PCB congeners are increasingly stable and recalcitrant, which is attributed to a stabilizing effect of adjacent chloride substitutions and steric hindrance against enzymatic attack (Wiegel and Wu, 2000). Another factor may relate to the aqueous solubility of different PCB congeners. Aqueous solubility, routinely indicated using the octanol/water partition coefficient (Kow), is an important factor in the study of environmental fates of organic pollutants such as PCBs. Aqueous solubility (described by 1/Kow: Table I) follows a decreasing trend with increasing number of chloride substitutions. For example the aqueous solubility of the trichlorinated congener 28 is more than 30-fold greater than that of the heptachlorinated congener 180 (Makino,

1052

Biotechnology and Bioengineering, Vol. 99, No. 5, April 1, 2008

1998, 1999; Yeh and Hong, 2002). In hydrogenolysis tests the catalyst was located in the aqueous phase, hence the more heavily substituted species were limited in aqueous availability. This could provide a basis for the observed differences in Cl release from the most and least soluble PCBs (PCBs 28 and 180, respectively): 5.5-fold in the case of Dd-Pd(0) and 3.1-fold in the case of Rs-Pd(0) and 20.8-fold in the case of Chem-Pd(0) (differences are significant, P