anaerobic fermentation of glycerol in paenbacillus macerans

Published Ahead of Print 17 July 2009. 10.1128/AEM.01246-09.

2009, 75(18):5871. DOI:Appl. Environ. Microbiol. Ramon GonzalezAshutosh Gupta, Abhishek Murarka, Paul Campbell and Pathways and Environmental Determinants

: MetabolicPaenibacillus maceransAnaerobic Fermentation of Glycerol in

http://aem.asm.org/content/75/18/5871Updated information and services can be found at:

These include:

REFERENCEShttp://aem.asm.org/content/75/18/5871#ref-list-1at:

This article cites 40 articles, 15 of which can be accessed free

CONTENT ALERTS more»articles cite this article),

Receive: RSS Feeds, eTOCs, free email alerts (when new

http://journals.asm.org/site/misc/reprints.xhtmlInformation about commercial reprint orders: http://journals.asm.org/site/subscriptions/To subscribe to to another ASM Journal go to:

on July 29, 2012 by guesthttp://aem

.asm.org/

Dow

nloaded from

http://aem.asm.org/cgi/alerts

http://aem.asm.org/

APPLIED AND ENVIRONMENTAL MICROBIOLOGY, Sept. 2009, p. 5871–5883 Vol. 75, No. 180099-2240/09/$08.00�0 doi:10.1128/AEM.01246-09Copyright © 2009, American Society for Microbiology. All Rights Reserved.

Anaerobic Fermentation of Glycerol in Paenibacillus macerans:Metabolic Pathways and Environmental Determinants�

Ashutosh Gupta,1 Abhishek Murarka,1 Paul Campbell,2 and Ramon Gonzalez1,3*Department of Chemical and Biomolecular Engineering, Rice University, Houston, Texas1; Glycos Biotechnologies Inc.,

Houston, Texas2; and Department of Bioengineering, Rice University, Houston, Texas3

Received 30 May 2009/Accepted 13 July 2009

Paenibacillus macerans is one of the species with the broadest metabolic capabilities in the genus Paeniba-cillus, able to ferment hexoses, deoxyhexoses, pentoses, cellulose, and hemicellulose. However, little is knownabout glycerol metabolism in this organism, and some studies have reported that glycerol is not fermented.Despite these reports, we found that several P. macerans strains are capable of anaerobic fermentation ofglycerol. One of these strains, P. macerans N234A, grew fermentatively on glycerol at a maximum specificgrowth rate of 0.40 h�1 and was chosen for further characterization. The use of [U-13C]glycerol and furtheranalysis of extracellular metabolites and proteinogenic amino acids via nuclear magnetic resonance (NMR)spectroscopy allowed identification of ethanol, formate, acetate, succinate, and 1,2-propanediol (1,2-PDO) asfermentation products and demonstrated that glycerol is incorporated into cellular components. A mediumformulation with low concentrations of potassium and phosphate, cultivation at acidic pH, and the use of aCO2-enriched atmosphere stimulated glycerol fermentation and are proposed to be environmental deter-minants of this process. The pathways involved in glycerol utilization and synthesis of fermentationproducts were identified using NMR spectroscopy in combination with enzyme assays. Based on thesestudies, the synthesis of ethanol and 1,2-PDO is proposed to be a metabolic determinant of glycerolfermentation in P. macerans N234A. Conversion of glycerol to ethanol fulfills energy requirements bygenerating one molecule of ATP per molecule of ethanol synthesized. Conversion of glycerol to 1,2-PDOresults in the consumption of reducing equivalents, thus facilitating redox balance. Given the availability,low price, and high degree of reduction of glycerol, the high metabolic rates exhibited by P. maceransN234A are of paramount importance for the production of fuels and chemicals.

Although many microorganisms can metabolize glycerol inthe presence of external electron acceptors (respiratory me-tabolism), few are able to do so fermentatively (i.e., in theabsence of electron acceptors). Fermentative metabolism ofglycerol has been reported in species of the genera Klebsiella,Citrobacter, Enterobacter, Clostridium, Lactobacillus, Bacillus,Propionibacterium, and Anaerobiospirillum but has been stud-ied more extensively in a few species of the family Enterobac-teriaceae, namely, Citrobacter freundii and Klebsiella pneu-moniae (6, 9). Glycerol fermentation in these organisms ismediated by a two-branch pathway, which results in the syn-thesis of the glycolytic intermediate dihydroxyacetone (DHA)phosphate (DHAP) and the fermentation product 1,3-pro-panediol (1,3-PDO) (Fig. 1A) (6). In the oxidative branch,glycerol is dehydrogenated to DHA by a type I NAD-linkedglycerol dehydrogenase (glyDH). DHA is then phosphorylatedby ATP- or phosphoenolpyruvate (PEP)-dependent DHA ki-nases (DHAKs) to generate DHAP. In the parallel reductivebranch, glycerol is dehydrated by glycerol dehydratase, and3-hydroxypropionaldehyde (3-HPA) is formed. 3-HPA is thenreduced to the major fermentation product 1,3-PDO byan NADH-linked 1,3-PDO dehydrogenase (1,3-PDODH),thereby regenerating NAD� (Fig. 1A). Organisms that lack the

capacity to synthesize 1,3-PDO have been deemed unable toutilize glycerol in a fermentative manner (6, 9, 10). The me-tabolism of glycerol in these organisms is thought to require anelectron acceptor and takes place through a respiratory path-way that involves a glycerol kinase and two respiratory (aerobicand anaerobic) glycerol-3-phosphate dehydrogenases (G3PDHs)(6, 7, 24, 29, 35, 38) (Fig. 1B). A recent development in thisarea is the finding that Escherichia coli, an organism that isunable to produce 1,3-PDO, can indeed ferment glycerol in theabsence of external electron acceptors (15, 26). In this model,synthesis of the fermentation products 1,2-PDO and ethanolenables glycerol fermentation by facilitating redox balance andATP generation, respectively (Fig. 1C) (15). A type II glyDHand a PEP-dependent DHAK mediate the conversion of glyc-erol to glycolytic intermediates. glyDH also catalyzes the laststep in the synthesis of the key fermentation product 1,2-PDO(Fig. 1C).

Paenibacillus macerans, previously called Bacillus maceransand Bacillus acetoethylicum, is a gram-positive, spore-formingbacterium belonging to the genus Paenibacillus (17) that iscapable of fermentative metabolism of hexoses, deoxyhexoses,pentoses, cellulose, and hemicellulose (33, 39, 40, 41). Glyc-erol, however, is considered a nonfermentable carbon sourcefor P. macerans. The “nonfermentable status” of glycerol hasbeen used to determine whether certain electron acceptors,such as fumarate, trimethylamine N-oxide, nitrate, and nitrite,can mediate anaerobic respiration in this organism (34).

In this study we found that several P. macerans strains areable to ferment glycerol in the absence of external electron

* Corresponding author. Mailing address: Department of Chemicaland Biomolecular Engineering, Rice University, 6100 Main Street,MS-362, Houston, TX 77005. Phone: (713) 348-4893. Fax: (713) 348-5478. E-mail: [email protected].

� Published ahead of print on 17 July 2009.

5871


.asm.org/

Dow

nloaded from

http://aem.asm.org/

acceptors. The fermentation of glycerol by one of these strains,P. macerans N234A, occurred at high metabolic rates and inthe absence of an active 1,3-PDO pathway. Therefore, theenvironmental and metabolic determinants of glycerol fermen-tation in P. macerans N234A were investigated.

MATERIALS AND METHODS

Strains. Wild-type P. macerans Northrop 234A (� LMG 13285 � N234A) wasobtained from the Belgian Co-ordinated Collections of Microorganisms (BCCM/LMG, Gent, Belgium) and used throughout this study, unless otherwise speci-fied. P. macerans strains B-394 (NRRL collection, Peoria, IL), ATCC 7068(American Type Culture Collection, Manassas, VA), BKM B-51 (Bacillus Ge-netic Stock Center, Columbus, OH), B14029 (NRRL collection, Peoria, IL), andB-388 (NRRL collection, Peoria, IL) were also tested to determine their ability

to ferment glycerol. E. coli K-12 wild-type strain MG1655 was obtained from theUniversity of Wisconsin E. coli Genome Project (www.genome.wisc.edu) (19).The strains were kept in 32.5% glycerol stocks at �80°C. Plates were preparedusing Luria-Bertani (LB) medium containing 1.5% agar.

Culture medium and cultivation conditions. The minimal medium describedby Schepers et al. (33) was used, with minor modifications. This medium contains(per liter [final volume]) 6.8 g KH2PO4, 3 g NH4Cl, 1 g KCl, 0.5 g sodium citrate,0.2 g MgSO4 � 7H2O, 30 mg MnSO4 � H2O, 30 mg EDTA, 10 mg CaCl2 � 2H2O,5 mg Na2MoO4, 5 mg FeSO4 � 7H2O, 5 mg H3BO3, 3 mg CoCl2 � 6H2O, 1 mgCuSO4 � 5H2O, 1 mg ZnSO4 � 7H2O, 1 mg nicotinic acid, 2 mg biotin, 2 mgp-aminobenzoic acid, and 2 mg thiamine hydrochloride. Unless otherwise spec-ified, the medium was supplemented with 10 g/liter glycerol and 1 g/liter tryptone(final concentrations). “Low-phosphate and low-potassium” medium was pre-pared by replacing the KH2PO4 and KCl with 0.35 g NaH2PO4 per liter and8.37 g MOPS (morpholinopropanesulfonic acid) per liter. All chemicals wereobtained from Fisher Scientific (Pittsburg, PA) and Sigma Aldrich (St. Louis, MO).

Fermentations were conducted using a 750-ml fermentation system obtainedfrom Ward’s Natural Science (Rochester, NY) with a working volume of 500 mland independent control of the temperature (37°C), pH, and stirrer speed. ThepH was controlled at the desired values using a Jenco 3671 pH controller fittedwith a Jenco 600p pH probe (Jenco, San Diego, CA). A base (2 M NaOH) forpH control was added by gravity flow using a pinch valve (Bio-Chem Inc.,Boonton, NJ) connected to the pH controller. The stirrer speed was maintainedat 200 rpm by using a Fisher Scientific Isotemp stirring plate (Pittsburg, PA).Anaerobic conditions were maintained by sparging the medium with ultra-high-purity argon (Matheson Tri-Gas, Inc., Houston, TX). Sterile conditions wereachieved by using 0.2-�m and 0.45-�m HEPA filters (Millipore, Billerica, CA) inthe inlet and outlet lines, respectively. Experiments conducted in tubes (asspecified below) were carried out using 17-ml Hungate tubes (Bellco Glass, Inc.,Vineland, NJ) that were modified by piercing the septa with two luer lockneedles, one of which was used for oxygen-free argon sparging (20 gauge by 2 in.;Hamilton Company-USA, Reno, NV) and one of which was used for gas efflux(20 gauge by 8 in.; Hamilton Company-USA, Reno, NV). The contents weremixed by the rising gas bubbles. The working volume of these modified Hungatetubes was 10 ml.

Prior to use, the cultures (stored as glycerol stocks at �80°C) were streakedonto LB medium plates and incubated at 37°C. Single colonies were used toinoculate 17-ml modified Hungate tubes (see above) containing 10 ml of themedium described above supplemented with 10 g/liter tryptone, 5 g/liter yeastextract, and 5 g/liter glycerol. The tubes were incubated at 37°C until the opticaldensity at 550 nm was �0.4. An appropriate volume of each actively growingpreculture was centrifuged, and the pellet was washed with the medium de-scribed above (lacking glycerol) and used to inoculate 500 ml of medium in thefermentor, using a target starting optical density at 550 of 0.05 nm.

Analytical methods. The optical density at 550 nm was determined and used toestimate the cell mass (1 optical density unit � 0.43 g [dry weight]/liter). Aftercentrifugation, the supernatant was stored at �20°C before high-performanceliquid chromatography (HPLC) and nuclear magnetic resonance (NMR) analy-sis. Glycerol, organic acids, acetone, ethanol, and hydrogen were quantified aspreviously described (13, 14).

NMR experiments. An experiment with 100% [U-13C]glycerol was conductedto assess the incorporation of glycerol into proteinogenic biomass and to verifythat fermentation products originated from glycerol. This experiment was carriedout as previously described (26), but the culture was harvested after 30 h. Cellpellets were prepared and analyzed to determine 13C enrichment using a one-dimensional (1D) proton spin echo with and without a concurrent 90° pulse oncarbon (4, 26). 1D proton NMR spectroscopy was used to analyze the superna-tants from the 30-h culture mentioned above and from experiments performedwith unlabeled glycerol. The sample preparation and acquisition parametersused have been described previously (26). The resulting spectra were processedusing the software packages FELIX 2001 (Accelrys Software Inc., Burlington,MA) and MestRe Nova 5.0.3 (Mestrelab Research SL, Santiago de Compostela,Spain).

Enzyme activities. For enzyme assays, cells were grown under specified cultureconditions. Actively growing cells were harvested by centrifugation, washed twicewith a saline solution (9 g/liter NaCl), and stored as cell pellets at �80°C untilthey were used. The pellets were resuspended in the same saline solution toobtain �1 mg (dry weight) of cells/ml and sonicated with a Branson 500 Sonifier(NJ) for 15 min at 4°C. The microtip configuration with a power rating of 7 and50% pulse duration was used.

Glycerol kinase activity was assayed in a manner similar to that described byHayashi and Lin (16) by measuring the change in absorbance at 340 nm and 25°Cin a 1-ml reaction mixture containing 0.15 M glycine (pH 9), 11 mM MgCl2,

FIG. 1. Glycerol metabolism in bacteria. (A) 1,3-PDO model for thefermentative utilization of glycerol. (B) Respiratory metabolism of glyc-erol (i.e., metabolism in the presence of an electron acceptor). (C) 1,2-PDO–ethanol model for the fermentative utilization of glycerol. Dashedlines indicate multiple steps. glyD, glycerol dehydratase; glyDH-I, type IglyDH; GK, glycerol kinase; ae-G3PDH, aerobic G3PDH; an-G3PDH,anaerobic G3PDH; QH2, reduced quinone; glyDH-II, type II glyDH;FHL, formate hydrogen lyase; ADH, alcohol/acetaldehyde dehy-drogenase.

5872 GUPTA ET AL. APPL. ENVIRON. MICROBIOL.


.asm.org/

Dow

nloaded from

http://aem.asm.org/

0.27 M hydrazine, 1.2 mM NAD�, 5 mM ATP, 2 mM glycerol, 20 U of �-glyc-erolphosphate dehydrogenase, and 50 �l crude cell extract prepared as describedabove. Anaerobic G3PDH activity was measured with 50 �l crude cell extract asdescribed previously (21), except that 33 mM, instead of 10 mM, glycerol 3-phos-phate was used. The aerobic G3PDH activity was determined in the same way,except that flavines were omitted and sodium cyanide (10 �M) was included inthe assay mixture. The assay mixtures were monitored spectrophotometrically at570 nm. The extinction coefficient of reduced 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) was 17 mM�1 cm�1.

The activity of glyDH during the oxidation of glycerol was measured as de-scribed previously (15), with potassium carbonate (pH 9.5) as the buffer. Reduc-tive glyDH activity (i.e., activity with hydroxyacetone [HA]) was measured usinga similar mixture in which HA and NADH replaced glycerol and NAD�, re-spectively. Measurement of DHAK activity involved two different procedures forpreparation of cell extracts, depending on whether the ATP- or PEP-dependentenzyme was assayed. Cell extracts for the ATP-dependent assay were prepared asdescribed above, and a coupled assay for ATP-dependent conversion of DHA toDHAP and then NADH-dependent conversion of DHAP to glycerol 3-phos-phate was performed as previously described (3) with 30 �l crude cell extract.The PEP-dependent DHAK activity was assayed using the method reported byKornberg and Reeves (23), with minor modifications as previously described(43).

Alcohol dehydrogenase activity was determined by monitoring the NADH-dependent reduction of acetaldehyde at 340 nm as previously described (20) with30 �l crude cell extract. Coenzyme A (CoA)-linked acetaldehyde dehydrogenaseactivity was assayed as previously reported (36) with 30 �l crude cell extract,using 20 mM dithiothreitol instead of 2-mercaptoethanol.

The activity of methylglyoxal (MG) synthase was determined using a colori-metric assay as described by Berrios-Rivera et al. (5). MG-reducing activities(e.g., activities of MG reductase and aldo-keto reductases) were measured usinga 1-ml reaction mixture containing 10 mM MG, 0.1 mM NAD(P)H, and 100 mMpotassium phosphate buffer (pH 7) (22). 1D proton NMR spectroscopy was usedto identify the products of MG- and HA-reducing reactions. NMR measure-ments were obtained after cell debris was removed from a reaction mixturecontaining cell extract (30 �l), MG or HA (10 mM), coenzyme (1 mM NADH or1 mM NADPH), buffer (100 mM potassium phosphate, pH 7.0), and D2O. TheNMR data were collected after 4 h of incubation at 25°C.

Linearity of the reactions (protein concentration and time) was established forall preparations. All spectrophotometric measurements were obtained with aBioMate 5 spectrophotometer (Thermo Scientific, MA). The nonenzymatic rateswere subtracted from the observed initial reaction rates. Enzyme activities arereported below in micromoles of substrate per minute per milligram of cellprotein, and the values are averages for at least three cell preparations.

RESULTS

Anaerobic fermentation of glycerol by P. macerans. Despiteprevious reports of the inability of P. macerans to fermentglycerol (34), we found that several strains of this species canutilize glycerol in the absence of external electron acceptors.Commonly used strains, such as N234A, B-394, and ATCC7068, all showed efficient growth and glycerol utilization (Table1). Other strains that were tested but were unable to fermentglycerol included BKM B-51, B14029, and B-388. When thefermentation of glycerol by one of these strains, P. maceransN234A, is compared to that reported for E. coli MG1655 in a

similar medium (13), it is evident that this metabolic process ismuch faster in P. macerans N234A; the rates of glycerol utili-zation and ethanol production were 7- and 12-fold higher,respectively. Since the conversion of glycerol to ethanol andother products via anaerobic fermentation has been proposedas a means of achieving economic viability for the biodieselindustry (42), the high metabolic rates exhibited by P. maceransN234A are of paramount importance.

To further investigate the fermentative metabolism of glyc-erol by strain N234A, we conducted experiments using a low-nutrient medium supplemented with 1 g/liter of tryptone in-stead of the 10 g/liter of tryptone and 5 g/liter of yeast extractused in the experiments described above. While tryptone sup-plementation was required to observe significant utilization ofglycerol, rich supplements have been used in previous studiesof the fermentative and respiratory metabolism of P. macerans(33, 39, 40). Figure 2A shows a typical fermentation profile forstrain N234A in the low-nutrient medium. Similar results wereobtained with the other strains described in Table 1 (data not

FIG. 2. Fermentation of glycerol by P. macerans N234A in minimalmedium supplemented with 10 g/liter glycerol and 1 g/liter tryptoneand identification of fermentation products by NMR spectroscopy.(A) Concentrations of glycerol (f), cells (Œ), ethanol (F), 1,2-PDO(}), and formic (�) and acetic (�) acids. (Inset) Log-linear plot of cellconcentration. (B) 1D 1H-NMR spectrum of the fermentation brothfrom a 16-h sample of the culture shown in panel A. (Inset) Magnifi-cation of the area of the spectrum where two peaks corresponding tothe methyl protons of 1,2-PDO were found (doublet at 1.15 ppm).

TABLE 1. Glycerol fermentation by P. macerans strainsa

Strain Cell growth (opticaldensity at 550 nm)

Amt of glycerolutilized (g/liter)

Amt of ethanolproduced (g/liter)

B-394 1.12 � 0.10 8.56 � 0.53 4.18 � 0.34N234A 1.24 � 0.11 9.92 � 0.74 4.84 � 0.36ATCC 7068 0.98 � 0.08 6.05 � 0.46 2.97 � 0.18

a Cells were grown for 24 h in modified Hungate tubes using minimal mediumsupplemented with 10 g/liter glycerol, 10 g/liter tryptone, and 5 g/liter yeastextract, as described in Materials and Methods. The values are averages �standard deviations for three different cultures.

VOL. 75, 2009 FERMENTATION OF GLYCEROL IN P. MACERANS 5873


.asm.org/

Dow

nloaded from

http://aem.asm.org/

shown). Exponential growth was observed for a period of 4 hwith a maximum specific growth rate (�max) of 0.402 h�1 (Fig.2A, inset) (ln cell concentration � 0.402 t � 6.149, where t istime; R2 � 0.9997). The �max for three independent fermen-tations was 0.40 � 0.03 h�1. This �max is about 10 times thatreported for E. coli MG1655 in a similar medium and withsimilar culture conditions (26). The growth yield once the cellsreached the stationary phase was 78.4 mg cells/g glycerol,which is about 2.4 times the yield reported for E. coli (26).

The fermentation products identified via 1D 1H-NMR spec-troscopy include ethanol (two multiplets at 3.66 and 1.19 ppm)and acetic (1.93 ppm) and formic (8.46 ppm) acids (Fig. 2B).While succinic acid (2.444 ppm) was found in fermentationsconducted at a basic pH (e.g., pH 8) (see below), no lactic acidwas detected in the extracellular medium under any of theconditions evaluated. A doublet was observed in the spectra oflate fermentation samples at a position with the same chemicalshift as that of methyl protons of 1,2-PDO (doublet at1.15 ppm) (Fig. 2B, inset). The same doublet has been found inthe spectra of fermentation samples of E. coli (26), and uponfurther investigation via two-dimensional 1H-1H correlationspectroscopy NMR, the peaks were identified as originatingfrom 1,2-PDO methyl protons (15). Therefore, we concludethat 1,2-PDO is a product of glycerol fermentation by strainN234A. While 1,2-PDO is a product of the fermentation of the6-deoxyhexose sugars fucose and rhamnose (40), our findingsrepresent the first report of 1,2-PDO synthesis during the me-tabolism of glycerol in P. macerans.

The concentrations of glycerol and fermentation products,as determined by HPLC, are shown in Fig. 2A. Ethanol, for-mate, and 1,2-PDO continuously accumulated in the extracel-lular medium as glycerol was consumed, although the formateconcentration decreased after 16 h. Conversion of formate toCO2 and hydrogen, catalyzed by the enzyme formate-hydrogenlyase (33, 39), appears to be responsible for the small amountsof formate present in the medium and the decrease in theformate concentration after 16 h. Acetate, on the other hand,exhibited a different profile; it rapidly accumulated during thefirst 8 h of fermentation, when very little glycerol had beenconsumed (Fig. 2A). This is in contrast to the findings forethanol and 1,2-PDO, whose synthesis paralleled the consump-tion of glycerol. The decrease in the acetate concentrationafter 8 h was likely due to its conversion to acetone (withacetoacetate as an intermediate), a well-known pathway in P.macerans (33, 39). Based on the amount of acetate present at8 h (0.192 g/liter) and the amount remaining at the end of thefermentation (0.046 g/liter), we calculated that 0.070 g/liter ofacetone was produced (2 acetate 3 acetone � CO2). Giventhe volatile nature of acetone (leading to significant evapora-tion at 37°C) and the small amounts produced, it was notsurprising to find that the acetone concentration was below thelimit of detection of our HPLC method. It is noteworthy thatthe amount of acetate produced during glycerol fermentationis only 1 to 5% (on a molar basis) of the amount of ethanolproduced; both of these products originate from pyruvate andacetyl-CoA. Moreover, a significant fraction of the smallamount of acetate was generated from tryptone components,as discussed below. The data mentioned above were used toconduct a fermentation balance analysis, which included bothcarbon and redox balances (Table 2). The amount of carbon

recovered as fermentation products was close to 100% of theamount of carbon consumed as glycerol (3.061/3 � 1.02) (Ta-ble 2). Similarly, about 99% of the reducing equivalents gen-erated during the fermentation of glycerol were captured in thesynthesis of fermentation products (�1.985/�2 � 0.99) (Table2). These data represent excellent closure of both redox andcarbon balances.

No significant cell growth was observed when glycerol wasomitted from the medium formulation (the cell concentrationincreased by less than 0.01 g/liter). However, the metabolism oftryptone components led to accumulation of 0.096 g/liter ofacetate and 0.004 g/liter of ethanol in the extracellular me-dium. This amount of acetate is roughly 50% of the amountproduced by the culture described in Fig. 2A, suggesting thattryptone components significantly contribute to acetate synthe-sis during glycerol fermentation. The amount of ethanol gen-erated from tryptone metabolism, on the other hand, is only0.14% of the total amount of ethanol produced during glycerolfermentation (Fig. 2A). No products were detected when bothtryptone and glycerol were excluded from the medium.

Glycerol fermentation was also observed when the growthmedium was supplemented with a mixture of proteinogenicamino acids at levels similar to those provided by tryptonesupplementation (i.e., 0.175 g/liter of cells and 3.8 g/liter ofglycerol fermented). However, amino acid supplementationresulted in a lower cell density and slower fermentation kinet-ics (specific growth rate, 0.17 h�1) than those shown in Fig. 2A.Inclusion of citrate in the culture medium was beneficial butnot required for glycerol fermentation, and the presence ofcitrate did not lead to synthesis of any fermentation product inthe absence of tryptone and glycerol. It is noteworthy thatcitrate is a component of a standard minimal medium (com-

TABLE 2. Fermentation balances for the growth of P. maceransN234A on glycerol at pH 6 and 37°C

Substrate consumedor product formed

Mol of product/mol of glycerol

consumeda

Oxidationstateb

Redoxbalancec

Carbonrecovery(no. of

C atoms)d

SubstrateGlycerol �2 �2 3

ProductsAcetic acid 0.011 � 0.001 0.00 0.000 0.022Formic acid 0.210 � 0.011 2.00 0.420 0.210Carbon dioxide 0.801 � 0.036 4.00 3.205 0.801Hydrogene 0.801 � 0.036 �2.00 �1.6021,2 PDO 0.005 � 0.001 �4.00 �0.007 0.028Ethanol 1.000 � 0.023 �4.00 �4.000 2.000

Total �1.985 3.061

a The data are for triplicate samples taken at 16 h from the culture describedin Fig. 2A. The values (averages � standard deviations) are the net number ofmoles produced per mole of glycerol fermented. Since 63.73% of the acetic acidoriginated from glycerol (see text), the values shown are 0.6373 times theamounts detected in the fermentation broth.

b The oxidation state of carbon atoms was calculated by assuming that theoxidation states were �2 and �1 for oxygen and hydrogen, respectively (31).

c The values were obtained by multiplying the number of moles of product permole of glycerol by the oxidation state of carbon atoms.

d Carbon recovery was calculated by multiplying the number of moles ofproduct per mole of glycerol by the number of carbon atoms in the molecule.

e The numbers of moles of carbon dioxide and hydrogen were calculated aspreviously described (13, 26).



.asm.org/

Dow

nloaded from

http://aem.asm.org/

monly referred to as “Spizizen salts medium”) used to cultivateBacillus species (2).

Identification of the origin of carbon in fermentation prod-ucts and cellular components. Since glycerol metabolism re-quires supplementation of the medium with small amounts oftryptone, we conducted experiments to investigate whetherfermentation products and cellular components are synthe-sized from glycerol or tryptone. To this end, cells were grown

on 100% [U-13C]labeled glycerol, as described in Materialsand Methods. Selected areas of the NMR spectrum of thesupernatant of a 30-h sample from this culture are shown inFig. 3. 1D 1H-NMR spectroscopy was used in these experi-ments to distinguish between 13C and 12C atoms (see Materialsand Methods for details). Since 13C is magnetic, protons at-tached to 13C atoms have two different chemical shifts due tothe positive and negative energy levels of these atoms. Protons

FIG. 3. 1D 1H-NMR spectra of the fermentation broth from a 30-h culture grown on 100% [U-13C]glycerol. (A) Region of the spectrumcorresponding to the 13C and 12C signals of ethanol and acetic and formic acids. Central 12C peaks are flanked by two satellite peak structuresresulting from protons attached to the 13C atoms. The ratio of the area of each peak to the total area (i.e., sum of all peaks) (expressed as apercentage) is indicated. (B) Region of the spectrum where 13C satellites and 12C signals of 1,2-PDO methyl protons are located. The peaksindicated by arrows represent 13C satellite signals. (Inset) Magnification of the area where the 12C methyl proton signals should be located.



.asm.org/

Dow

nloaded from

http://aem.asm.org/

attached to 12C atoms have chemical shifts between those of13C because all 12C atoms are in a neutral state. Thus, protonsattached to 12C carbon atoms lead to a central peak structureflanked by two satellite peak structures that result from otherprotons that are attached to 13C atoms. The ratio of the area of13C satellite peaks to the total area (expressed as a percentage)then reflects the 13C enrichment of the carbon atom.

As Fig. 3A shows, the level of 13C enrichment in ethanol andformic acid approached 100%, demonstrating that most car-bon atoms in these products originated from glycerol. 13Cenrichment was also observed in the case of acetic acid, al-though only about 64% of the carbon in this product originatedfrom glycerol (Fig. 3A). In the case of acetate, the calculationsassume that the area of the satellite peak structures located tothe left of the 12C central peak (which overlaps with otherpeaks) is equal to the area of the satellite peak structuresvisible to the right of the 12C peak. The large fraction ofunlabeled carbon in acetate results from the contribution oftryptone components to acetate synthesis, as reported above.Acetate accumulates only during the first 8 h of fermentation,when very little glycerol is consumed and therefore a largefraction of the carbon in this product originates from tryptonecomponents (Fig. 2A) (note as well that experiments per-formed in the absence of glycerol showed that synthesis ofsignificant amounts of acetate occurred). As fermentation ofglycerol occurs after 8 h, most of the carbon is channeled toethanol (Fig. 2A). It then follows that the fraction of carbonoriginating from tryptone (i.e., unlabeled carbon) should bemuch larger in acetate than in ethanol; i.e., most of the carbonin ethanol originates from glycerol, which is 13C labeled, whilealmost 50% of the carbon in acetate originates from tryptone,which is not labeled.

The area of the spectrum where the doublet correspondingto 1,2-PDO methyl protons was observed is shown separatelyin Fig. 3B. While the 13C signals were identified, no signal wasobserved at the 12C positions. The inset in Fig. 3B shows amagnification of the area of the spectrum where a doublet dueto the protons attached to 12C carbons should appear. Whenthis inset is compared to the inset in Fig. 2B, which shows the12C doublet observed when unlabeled (naturally labeled) glyc-erol was used, it is evident that 1,2-PDO is synthesized exclu-sively from glycerol.

Also in the experiment described above, the 13C enrichmentof proteinogenic biomass was assessed to determine whetherglycerol is used in the synthesis of cell mass. For this purpose,the 13C enrichment of proteinogenic amino acids in cells grownon [U-13C]glycerol was compared to that in a reference culturein which cells were grown on unlabeled glycerol. The cells fromthe two cultures were hydrolyzed to obtain a cocktail of theirproteinogenic amino acids, which was subsequently analyzedusing NMR. 13C enrichment was determined using a 1D pro-ton spin echo with and without a concurrent 90° pulse oncarbon as described in Materials and Methods. The 90° pulseon carbon refocused the 13C carbon atoms, thereby suppress-ing the 13C satellites arising due to proton-carbon spin cou-pling. The nucleus of 12C is nonmagnetic; thus, protons at-tached to 12C do not experience any differences in the twosituations. Therefore, 13C satellite peaks could be easily iden-tified upon comparison of the spectra obtained using these twomethods. The 1D NMR spectra obtained for the two samples

(with and without labeled carbon) contained small 13C satellitepeaks, which in several cases were hidden below bigger peaks.However, many of the small peaks were well resolved, withvisible 13C satellites. Parts of the NMR spectra depicting two ofthese resolved areas for a sample in which 13C-labeled glycerolwas used are shown in Fig. 4. The amino acid carbon atomsshown are threonine-� (left panels) and alanine- (right pan-els) atoms. In the case of unlabeled glycerol, the 13C satellitepeaks accounted for about 1% of the total signal (data notshown), which is approximately the natural abundance of thisisotope. However, when 100% [U-13C]glycerol was used (Fig.4, top panels), the 13C satellites accounted for about 22% ofthe total signal, indicating that about 20% of these amino acidsin the biomass originated from glycerol. Although not shownhere due to space limitations, similar spectra were obtained formany carbons corresponding to other amino acids in protei-nogenic biomass.

The use of tryptone supplementation also raised the ques-tion of whether glycerol utilization is truly fermentative orwhether compounds present in tryptone, or generated from it,serve as electron acceptors. The almost exclusive synthesis ofthe reduced product ethanol (Fig. 2A and Table 2) is a strongindication that glycerol is in fact metabolized in a fermentativemanner; otherwise, the presence of an electron acceptor wouldconsume the reducing equivalents generated from glycerol,and significant amounts of oxidized products (such as acetate)would be produced instead of ethanol. The excellent closure ofthe redox balance (�99%) (Table 2) is another indication thatglycerol metabolism is mediated by fermentative pathways.More direct evidence of the fermentative nature of this meta-bolic process was provided by the results of experiments inwhich cell growth and glycerol utilization were observed de-spite the use of the inhibitors cyanide and azide, which blockgeneral respiratory processes (data not shown).

Metabolic routes involved in the conversion of glycerol toglycolytic intermediates and the synthesis of fermentationproducts by P. macerans N234A. Two pathways have beenreported to mediate the microbial conversion of glycerol to theglycolytic intermediate DHAP (6, 24, 42) (Fig. 1). In an at-tempt to identify whether these pathways mediate glycerolutilization in P. macerans N234A, we measured the activities ofrelevant enzymes in crude cell extracts obtained from a 16-hsample of the culture shown in Fig. 2A. The results indicatethat the glycerol kinase-G3PDH route was inactive, as no ac-tivity was detected for these enzymes (Table 3). Both anaero-bic G3PDH activity and aerobic G3PDH activity were assayed.On the other hand, significant activity was observed for glyDHand DHAK, indicating that the glyDH-DHAK route mediatesthe utilization of glycerol in P. macerans N234A (Table 3).These activities were not present in cultures grown on LBmedium but were detected when glycerol was included in thecultures (Table 4), suggesting that they could be induced bythis carbon source. The glyDH activity was found to be NAD�

specific, as it was not detectable when NADP� was used as acofactor. The four types of bacterial glyDHs reported to datecan be differentiated on the basis of their inducibility by glyc-erol, DHA, and HA (9). The glyDH activity detected in P.macerans N234A extracts was induced by both glycerol andHA, but not by DHA (Table 4). Type II glyDHs from entericbacteria are induced by glycerol and HA (9); therefore, the



.asm.org/

Dow

nloaded from

http://aem.asm.org/

glyDH of P. macerans N234A appears to be a type II enzyme.The DHA generated by the action of glyDH was converted toDHAP by the action of both ATP- and PEP-dependentDHAKs, as inferred from the enzyme assays (Table 3). Bothenzymes were induced by glycerol, but DHA was able to in-duce only the ATP-dependent activity (Table 4). Taken to-gether, these results agree with current models for the micro-bial metabolism of glycerol in which the glyDH-DHAKpathway is associated with the fermentative utilization of thiscarbon source (Fig. 1) (6, 15, 24, 42). These activities were notfound in strains unable to ferment glycerol (data not shown).

The pathways involved in the synthesis of the fermentationproduct 1,2-PDO were investigated using enzyme activity mea-surements, characterization of enzymatic reactions via NMR,and supplementation of intermediates in the postulated path-ways. Based on previous studies of bacteria, the pathways thatcould mediate the synthesis of 1,2-PDO from DHAP are sum-

marized in Fig. 5A (1, 8, 12, 15, 22, 25, 30, 40). DHAP is aglycolytic intermediate generated during the utilization of glyc-erol via the glyDH-DHAK route discussed above. Since thesynthesis of MG from DHAP is a common step in the 1,2-PDOpathway, regardless of the branch used for the conversion ofMG to 1,2-PDO (Fig. 5A), MG synthase was assayed, andsignificant activity of this enzyme was found (Table 3). MGsynthase was not detected in the absence of glycerol (Table 4),indicating that it could be induced by this carbon source. Whencell extracts were assayed for MG-reducing enzymes, signifi-cant activity in the presence of either NADH or NADPH wasfound (Table 3). Since these activities could be involved in theconversion of MG to either HA or lactaldehyde, 1D 1H-NMRspectroscopy was used to characterize the reaction(s). Figure5B shows that the product of the MG-reducing activities wasindeed HA (no lactaldehyde was detected). Additionally, cellsfermenting glycerol exhibited significant HA-reducing activity

FIG. 4. NMR spectra of proteinogenic amino acids in cell biomass obtained from an experiment performed with 100% [U-13C]glycerol. Theidentity of 13C satellites as peaks arising due to labeled carbon was confirmed by performing a 13C decoupled experiment in which the 13C signalswere suppressed (bottom panels). The peaks indicated by arrows represent the incorporation of labeled carbon into threonine-� (left panels) andalanine- (right panels). (Insets) Magnifications of areas of interest in the spectra.



.asm.org/

Dow

nloaded from

http://aem.asm.org/

(Table 3). The enzyme catalyzing this conversion was able touse NADH and NADPH as cofactor. Further characterizationof the reaction(s) using 1D 1H-NMR spectroscopy showed thatthe product of HA reduction was 1,2-PDO (Fig. 5C). BothMG-reducing and HA-reducing activities were induced byglycerol (Table 4). It is noteworthy that these activities werenot found in strains unable to ferment glycerol (data notshown).

The results described above suggest that the synthesis of1,2-PDO during glycerol fermentation occurs through the con-version of DHAP to MG to HA to 1,2-PDO (Fig. 5A). Anexperiment in which amplification of the 1,2-PDO pathwayallowed glycerol fermentation in the absence of rich supple-ments provided evidence of the important role of this pathway:by supplementing the growth medium with 18.3 mM HA (1.36g/liter), glycerol fermentation was observed in the absence oftryptone. In 48 h the cells grew to a concentration of 0.23

g/liter, fermented 3.88 g/liter of glycerol, and produced 1.27g/liter 1,2-PDO, 1.50 g/liter ethanol, 0.038 g/liter acetate, and0.20 g/liter acetone. A carbon balance analysis in this experi-ment yielded 99% recovery of carbon in cell mass and fermen-tation products. These results suggest that the synthesis of1,2-PDO facilitates fermentation of glycerol in P. maceransN234A. The conversion of glycerol to 1,2-PDO results in theconsumption of reducing equivalents and has been proposed tofacilitate glycerol fermentation in E. coli by enabling redoxbalance in the absence of external electron acceptors (15, 26).

1,3-PDO, a structural isomer of 1,2-PDO, is a known meta-bolic product that enables fermentative utilization of glycerolin enteric bacteria (6, 24, 42). 1,3-PDO is synthesized in thesemicroorganisms via a pathway that involves the reduction ofthe intermediate 3-HPA to 1,3-PDO by an NADH-linked 1,3-PDODH (Fig. 1A). We did not find 1,3-PDO in the superna-tant of P. macerans N234A cultures, and 1,3-PDODH activitywas not detected in cell extracts (Table 3).

The almost homoethanologenic nature of glycerol fermen-tation (Fig. 2 and 3) reflects the highly reduced state of carbonin glycerol and suggests a central role for this pathway. Ethanolis synthesized in microorganisms through the reduction of ac-etaldehyde, a reaction catalyzed by alcohol dehydrogenase (27,31). Acetaldehyde, in turn, originates from either the oxidationof pyruvate (catalyzed by pyruvate decarboxylase [PDC]) orthe reduction of acetyl-CoA (catalyzed by acetaldehyde dehy-drogenase) (11, 27, 31). In previous studies, the dissimilation ofpyruvate in P. macerans strains during fermentative metabo-lism of sugars was mediated by pyruvate formate-lyase (PFL),an enzyme that converts pyruvate to acetyl-CoA and formate(35, 39, 40). Therefore, we postulated that the route to ethanolsynthesis should involve the consecutive reduction of acetyl-CoA to acetaldehyde to ethanol. To verify this hypothesis, weassayed cell extracts of P. macerans N234A grown on glycerolfor acetaldehyde and alcohol dehydrogenase activities. Bothactivities were present at significant levels (Table 3).

Effect of culture conditions and medium composition on thefermentative metabolism of glycerol. The fermentative metab-olism of glycerol by P. macerans N234A is negatively affectedby neutral and alkaline pH (Fig. 6). Both cell growth andglycerol utilization at pH 7 were less than one-half those at pH6. Even more surprising was the observation that no significant

TABLE 3. Activities of enzymes involved in the dissimilation ofglycerol and the synthesis of fermentation products

Enzyme Activity (�mol/min/mgof cell protein)a

Glycerol dissimilationglyDH (oxidative, toward glycerol) ..........................0.014 � 0.001ATP-dependent DHAK.............................................0.035 � 0.001PEP-dependent DHAK .............................................0.026 � 0.001Glycerol kinase ........................................................... NDG3PDH........................................................................ ND

1,2-PDO synthesisMG synthase ...............................................................0.067 � 0.004NADPH-dependent aldo-keto reductase/MG

reductase..................................................................0.013 � 0.001NADH-dependent aldo-keto reductase/MG

reductase..................................................................0.022 � 0.002glyDH (reductive, toward HA).................................0.198 � 0.002

1,3-PDO synthesis1,3-PDODH ................................................................ ND

Ethanol synthesisAlcohol dehydrogenase..............................................0.035 � 0.001CoA-linked aldehyde dehydrogenase ......................0.015 � 0.001

a Cell extracts were prepared from a 16-h sample from the culture described inFig. 2A. ND, no activity detected.

TABLE 4. Effect of glycerol, DHA, and HA on the activities of selected enzymes involved in glycerol fermentation

EnzymeActivity under different growth conditionsa

LB LB � Gly LB � DHA LB � HA

Glycerol dissimilationOxidative glyDH ND 0.014 � 0.002 ND 0.002 � 0.001ATP-dependent DHAK ND 0.033 � 0.001 0.008 � 0.001 NMPEP-dependent DHAK ND 0.007 � 0.001 ND NM

1,2-PDO synthesisMG synthase ND 0.096 � 0.005 NM NMMG reductase ND 0.022 � 0.002 NM NMReductive glyDH ND 0.070 � 0.003 0.066 � 0.005 0.043 � 0.002

a Oxidative and reductive glyDH and MG reductase activities were measured using NAD(H) as a cofactor, and the enzyme activities are expressed in �mol/min/mgof cell protein. Cells were grown in closed Hungate tubes using the minimal medium described in Materials and Methods supplemented with 10 g/liter tryptone and5 g/liter yeast extract (LB medium) and with 5 mM glycerol (Gly), DHA, or HA, as indicated. Crude cell extracts were prepared as described in Materials and Methodsusing cell pellets from actively growing cultures. ND, no activity detected; NM, not measured.



.asm.org/

Dow

nloaded from

http://aem.asm.org/

cell growth or glycerol consumption was observed at pH 8. Arelationship between pH and the ability to ferment glycerol hasrecently been reported for E. coli and attributed to severalfactors (13, 15, 26), which are examined in detail below.

One reason for the absence of glycerol fermentation at pH 8could be the limited availability of CO2 under alkaline condi-tions, which in turn could be caused by the low activity at thispH of CO2-generating enzymes, such as formate-hydrogen

FIG. 5. Investigation of the pathways involved in the synthesis of 1,2-PDO from DHAP. (A) Enzymes reported to mediate the synthesis of1,2-PDO in bacteria. The thick lines indicate the route used by P. macerans N234A for 1,2-PDO synthesis during glycerol fermentation, as inferredfrom our studies (see text, Tables 3 and 4, and Fig. 5B and 5C). MGS, MG synthase; AKR, aldo-keto reductase; MGR, MG reductase; 1,2 PDOR,1,2-PDO reductase. (B) 1D 1H-NMR characterization of the reactions involved in the reduction of MG to HA or lactaldehyde (see panel A). Thelower and upper spectra are the spectra for the initial and final (4-h) samples, respectively. The arrows pointing down and the asterisk indicateMG and HA peaks, respectively. The arrow pointing up indicates the peak for acetate, an impurity in MG. No lactaldehyde was found as a productof this activity. (C) 1D 1H-NMR characterization of the reaction(s) converting HA to 1,2-PDO, catalyzed by glyDH (see panel A). The lower andupper spectra are spectra for the initial and final (4-h) samples, respectively. The arrow and the asterisk indicate HA and 1,2-PDO peaks,respectively.



.asm.org/

Dow

nloaded from

http://aem.asm.org/

lyase (13). While CO2 could be generated by conversion ofacetoacetate to acetone, this pathway appears to be inactiveduring glycerol fermentation, as we did not find acetone in theextracellular medium. This agrees with previous studies thatreported no acetone production by P. macerans under alkalineconditions (39). The CO2 limitation hypothesis was investi-gated by providing a gas atmosphere containing 20% CO2

(with the balance argon) while controlling the pH by additionof base (Fig. 6). At pH 8 the cells even transitioned from anon-glycerol-fermenting state in an argon atmosphere to aglycerol-fermenting state in a CO2-enriched atmosphere.

Another reason for the negative impact of alkaline condi-tions could be that a basic pH, in combination with high levelsof potassium and phosphate, has a negative impact on theactivity of two key pathways involved in the fermentative uti-lization of glycerol, namely the 1,2-PDO and glyDH-DHAKpathways (15). While evidence for this originates from studiesconducted with E. coli, these pathways are active in P. macer-ans N234A as well (Tables 3 and 4 and Fig. 5). The use of amedium with low levels of phosphate and potassium had abeneficial effect on glycerol fermentation at pH 7 and 8 (Fig.6). Low levels of both phosphate and potassium were requiredto observe this improvement. Interestingly, the use of a CO2-enriched atmosphere in combination with a low-phosphateand low-potassium medium did not lead to further improve-ments in cell growth or glycerol utilization (Fig. 6). Glycerolfermentation at pH 6 was not affected by addition of CO2 orby the concentrations of phosphate and potassium (data notshown).

Since the use of a CO2-enriched atmosphere and a low-phosphate and low-potassium medium allowed fermentationof glycerol under basic conditions, we examined the effect of

pH on the distribution of fermentation products (Fig. 7). Whileat pH 6 the fermentation broth contained almost exclusivelyethanol, the use of alkaline conditions led to a slight decreasein the ethanol yield, along with the production of significantamounts of formic, acetic, and succinic acids (Fig. 7). Differ-ences in organic acid yields at different pHs have also beenreported for other P. macerans strains during glucose fermen-tation (39). However, the shift from almost homoethanolo-

FIG. 7. Effect of pH on product and biomass yields. Experimentswere conducted using a low-potassium and low-phosphate medium(see the legend to Fig. 6 and the text for details). The data are for pH6 (black bars), pH 7 (gray bars), and pH 8 (open bars). The barsindicate the means and the error bars indicate the standard deviationsfor three samples taken once the cultures reached the stationary phase.Differences in product and biomass yields discussed in the text weresignificant at P values of �0.045 (Student’s t test).

FIG. 6. Effect of pH, carbon dioxide, and concentrations of potassium and phosphate on cell growth (filled bars) and glycerol fermentation(open bars) by P. macerans N234A. The basal medium used in this study contained 6.8 g/liter KH2PO4 and 1 g/liter KCl (see Materials andMethods) (“high potassium and phosphates”). A low-potassium and low-phosphate medium was prepared by replacing KH2PO4 and KCl in thebasal medium with 0.35 g/liter of Na2HPO4 and 8.36 g/liter of MOPS (“low potassium and phosphates”). Experiments were conducted at 37°C,using the media described above supplemented with 10 g/liter glycerol and 1 g/liter tryptone. The gas atmosphere and pH used are indicated. Thebars indicate the means and the error bars indicate the standard deviations for three samples taken once the cultures reached the stationary phase.Differences in cell growth and glycerol utilization discussed in the text were significant at P values of �0.037 (Student’s t test).



.asm.org/

Dow

nloaded from

http://aem.asm.org/

genic fermentation to the mixed-acid type of fermentation isunique to the fermentation of glycerol. To illustrate this, it ishelpful to compare the ratio of organic acids to ethanol (on aweight basis), which changed from 0.04 at pH 6 to 1.6 at pH 8,a 37-fold difference. The higher biomass yield under alkalineconditions was accompanied by a higher 1,2-PDO yield, whichfurther supports the enabling role of 1,2-PDO synthesis. An-other important finding was the identification of succinic acidas a product of glycerol fermentation under alkaline conditions(Fig. 7). Previous studies of the fermentation of glucose, xy-lose, and deoxy sugars by P. macerans strains did not reportsuccinic acid as a product (33, 39, 40).

DISCUSSION

Although glycerol metabolism in P. macerans strains wasthought to be restricted to respiratory conditions (34), theresults reported here demonstrate that these organisms canutilize glycerol in a fermentative manner (Table 1 and Fig. 2and 3). The use of 13C-labeled glycerol and analysis of thefermentation broth using different NMR techniques allowedidentification of fermentation products and provided evidencethat they were synthesized from glycerol (Fig. 3 and 4). An-other NMR technique, 1D proton spin echo with and withouta concurrent 90° pulse on carbon, was used to analyze the 13Cenrichment of proteinogenic biomass, which demonstratedthat glycerol was incorporated into macromolecules. Excellentclosure of redox and carbon balances was observed (Table 2).

The most established model for the fermentative metabo-lism of glycerol in bacteria entails an active 1,3-PDO pathway(Fig. 1A). However, no 1,3-PDO was found in the fermenta-tion broth, nor was 1,3-PDODH activity detected in cell ex-tracts of P. macerans N234A (Fig. 2 and 3 and Table 3).Instead, production of 1,2-PDO, a structural isomer of 1,3-PDO, was observed (Fig. 3B and 4B), and the pathway in-volved in its synthesis was identified (Fig. 5). Based on thisevidence, we conclude that glycerol fermentation in P.

macerans N234A follows the “1,2-PDO–ethanol model” previ-ously proposed for the fermentative utilization of glycerol in E.coli (15, 26) (Fig. 1C). In this model, the synthesis of 1,2-PDOand ethanol enables glycerol fermentation by facilitating redoxbalance and ATP generation, respectively.

The role of the 1,2-PDO pathway in glycerol fermentation isbetter illustrated by performing a generalized degree-of-reduc-tion balance analysis, as shown in Table 5. The synthesis of1,2-PDO allows the cells to attain redox balance by consump-tion of the reducing equivalents generated during the incorpo-ration of glycerol into cell mass and oxidized products. Inagreement with the importance of this pathway, we found thatstimulation of 1,2-PDO synthesis by addition of the pathwayintermediate HA allows glycerol fermentation in the absenceof tryptone. In this experiment, 42.1 mM glycerol was con-sumed, which resulted in the synthesis of 32.6 mM ethanol,16.7 mM 1,2-PDO, 3.5 mM acetone, and 0.6 mM acetate (Ta-ble 5). Table 5 shows the calculated reducing equivalents gen-erated or consumed in the synthesis of biomass and the prod-ucts described above. Overall, the analysis of the degree ofreduction shows that the reducing equivalents consumed in thesynthesis of 1,2-PDO (16.7 mM) account for 99% of the re-ducing equivalents generated in the synthesis of cell mass andthe fermentation products acetate and acetone (16.8 mM).

The results of the generalized degree-of-reduction analysisalso supported the assumption that pyruvate dissimilation dur-ing glycerol fermentation takes place via PFL. Pyruvate dis-similation via any other known pyruvate-dissimilating enzyme(e.g., PDC, pyruvate dehydrogenase, or pyruvate oxidase)would result in the generation of one reducing equivalent perpyruvate molecule dissimilated. The degree of reduction forethanol synthesis via PDC is shown in Table 5 (“glycerol 3ethanol � CO2” pathway). If the analysis described above wereconducted with the assumption that pyruvate dissimilation oc-curs via PDC, only 34% of the reducing equivalents generatedduring the metabolism of glycerol would be accounted for bythe 1,2-PDO pathway. These results indicate that the opera-

TABLE 5. Generalized degree-of-reduction balances for the conversion of glycerol into cell mass and selected fermentation products

Theoretical analysis Experimental data

Pathway Stoichiometry ()a �� (H)b Concn of firstproduct (mM)c HHA

d

Glycerol 3 biomass C3H8O3 (14/3) 3 3CH1.9O0.5N0.2 (4.3)e 1.1 (0.55H) 9.26 1.75Glycerol 3 ethanol � CO2

f C3H8O3 (14/3) 3 C2H6O (6) � CO2 (0) 2 (1H)Glycerol 3 ethanol � formateg C3H8O3 (14/3) 3 C2H6O (6) � CH2O2 (2) 0 (0H) 32.56 0.00Glycerol 3 acetate � formateg C3H8O3 (14/3) 3 C2H4O2 (4) � CH2O2 (2) 4 (2H) 0.63 1.27Glycerol 3 0.5 acetone � formate �

0.5 CO2

C3H8O3 (14/3) 3 0.5C3H6O (16/3) � CH2O2 (2) � 0.5CO2 (0) 4 (2H) 3.46 13.82

Glycerol 3 1,2-PDO C3H8O3 (14/3) 3 C3H8O2 (16/3) �2 (�1H)HA 3 1,2-PDO C3H6O2 (14/3) 3 C3H8O2 (16/3) �2 (�1H) 16.67 �16.67

a Pathway stoichiometry accounts only for the carbon balance between reactants and products. The degree of reduction per carbon () was estimated as describedelsewhere (28) and is indicated in parentheses.

b The degree of reduction balance (�K) is estimated as follows: over i reactants�icii � over j products�jcij, where � and c are the stoichiometric coefficient and thenumber of carbon atoms for each compound, respectively. The net number of redox units (H) is expressed per mole of glycerol (H � NAD(P)H � reduce flavin adeninedinucleotide � H2).

c An experiment was conducted using a medium without tryptone but supplemented with HA (see text for details). The values are the concentrations of the firstproduct in the corresponding pathway or reaction.

d The net number of redox units (HHA) was calculated for the experiment in which HA was included in the medium (see footnote c) as follows: net number of redoxunits (H) obtained in the theoretical analysis � product concentration/product stoichiometric coefficient.

e The cell mass formula is the average reported for a microbial cell (28).f Pyruvate dissimilation via pyruvate decarboxylase was assumed.g Pyruvate dissimilation via PFL was assumed. The same results were obtained if the conversion of formate to H2 and CO2 was considered.



.asm.org/

Dow

nloaded from

http://aem.asm.org/

tion of the PDC pathway results in a significant redox imbal-ance that prevents the fermentative metabolism of glycerol.The PFL route avoids this situation by “storing” the “excess”reducing equivalents as formate or “releasing” them as hydro-gen. Further evidence for the operation of PFL was obtained inthe analysis of the composition of fermentation products forexperiments conducted under alkaline conditions. Since a highpH inhibits the enzyme formate hydrogen lyase (32), whichconverts formic acid to CO2 and hydrogen, the operation ofthe PFL route implies that the molar amount of formic acidaccumulated in the extracellular medium should approach thesum of the amounts of ethanol and acetic acid. Using the dataobtained at pH 8 (Fig. 7), the number of moles of formic acidwas calculated to be equivalent to 92% of the sum of thenumbers of moles of ethanol and acetate.

The degree-of-reduction analysis presented above was moredifficult to determine in experiments in which tryptone wasincluded in the culture medium because of the lack of preciseinformation about the relative contributions of glycerol andtryptone to the synthesis of biomass and acetate. However, ifone assumes that the carbon from glycerol is incorporated intoother cellular macromolecules in the same proportion that it isincorporated into proteinogenic biomass (i.e., 20%; Fig. 4),0.56 mM reducing equivalents would be generated as a resultof the cell growth shown in Fig. 2A (0.385 g/liter). Since theconversion of glycerol to 1,2-PDO consumes 1 mol of reducingequivalents per mol of 1,2-PDO synthesized (Table 5), it fol-lows that the amount of 1,2-PDO found in the fermentationbroth (�0.33 mM) is close to the amount needed to provide asink for the reducing equivalents generated in the synthesis ofcell mass.

A key component of the proposed metabolic model is theoperation of an oxidative pathway that channels glycerol intothe glycolytic intermediate DHAP and is composed of threeenzymes, namely a glyDH and PEP- and ATP-dependentDHAKs. The reaction catalyzed by glyDH is the first step inthe fermentative metabolism of glycerol in all microorganismsin which this metabolic process has been characterized to date(6, 24, 42). Although PEP-dependent DHAK is the most com-mon DHAK found in glycerol-fermenting organisms (6, 15,24), P. macerans N234A also possesses a second DHAK, ATP-dependent DHAK.

The pathways described above provide a framework to ex-plain the observed changes in cell growth and glycerol fermen-tation as a function of the pH and the concentrations of po-tassium and phosphate (Fig. 6). High levels of phosphatepromote the decomposition of both DHA and HA, two keyintermediates in these pathways, and have also been shown tonegatively affect the activity and inducibility of glyDH in otherbacteria (37). glyDH is one of the most important enzymes inthe proposed pathways (Fig. 1C). Moreover, bacterial MGsynthases, which are key enzymes in the synthesis of 1,2-PDO(Fig. 5A), are inhibited by high levels of phosphate (12, 18, 44).High concentrations of potassium, on the other hand, increasethe toxicity of MG in E. coli (6), and MG is a key intermediatein the synthesis of 1,2-PDO (Fig. 5). The effect of pH onglycerol fermentation (Fig. 6) can also be related to its impacton the pathways mentioned above. DHA, an intermediate inthe glyDH-DHAK pathway, is very unstable in alkaline envi-ronments. The higher toxicity of MG under alkaline conditions

(6) also limits the synthesis of 1,2-PDO. Interestingly, the con-ditions described here as conditions that negatively affect glyc-erol fermentation (i.e., alkaline pH and high concentrations ofpotassium and phosphates) were used by other investigatorsduring studies of glycerol metabolism in P. macerans (34). Thismay be one of the reasons why the fermentative utilization ofglycerol by this organism was not observed previously.

Finally, the proposed model suggests that the small amountof tryptone (or amino acids) required for glycerol fermentationto proceed is a consequence of the low activity of the 1,2-PDOpathway. The synthesis of 1,2-PDO appears to be the onlypathway in P. macerans N234A able to “dispose of” the “ex-cess” reducing equivalents generated during the incorporationof glycerol into cell mass; i.e., it is the only pathway that resultsin the net consumption of reducing equivalents (Table 5). Theutilization of building blocks present in tryptone (e.g., aminoacids) then reduces the use of glycerol in the synthesis of cellmass and, therefore, the associated redox imbalance. In agree-ment with this hypothesis, stimulation of the 1,2-PDO pathway(via addition of the intermediate HA) led to cell growth andglycerol fermentation in the absence of tryptone supplemen-tation.

ACKNOWLEDGMENTS

We thank S. Moran and E. Nikonowicz for assistance with NMRexperiments.

We thank Glycos Biotechnologies Inc. for financial support.

REFERENCES

1. Altaras, N. E., and D. C. Cameron. 1999. Metabolic engineering of a 1,2-propanediol pathway in Escherichia coli. Appl. Environ. Microbiol. 65:1180–1185.

2. Anagnostopoulos, C., and J. Spizizen. 1961. Requirements for transforma-tion in Bacillus subtilis. J. Bacteriol. 81:741–746.

3. Barbirato, F., S. Astruc, P. Soucaille, C. Camarasa, J. M. Salmon, and A.Bories. 1997. Anaerobic pathways of glycerol dissimilation by Enterobacteragglomerans CNCM 1210: limitations and regulations. Microbiology 143:2423–2432.

4. Bax, A. 1983. A simple method for the calibration of the decoupler radio-frequency field-strength. J. Magn. Res. 52:76–80.

5. Berrios-Rivera, S. J., K. Y. San, and G. N. Bennett. 2003. The effect ofcarbon sources and lactate dehydrogenase deletion on 1,2-propanediol pro-duction in Escherichia coli. J. Ind. Microbiol. Biotechnol. 30:34–40.

6. Booth, I. R. March 2005, posting date. Chapter 3.4.3, Glycerol and methyl-glyoxal metabolism. In R. Curtis III et al. (ed.), EcoSal—Escherichia coli andSalmonella: cellular and molecular biology. ASM Press, Washington, DC.http://www.ecosal.org/.

7. Borgnia, M. J., and P. Agre. 2001. Reconstitution and functional comparisonof purified GlpF and AqpZ, the glycerol and water channels from Esche-richia coli. Proc. Natl. Acad. Sci. USA 98:2888–2893.

8. Boronat, A., and J. Aguilar. 1979. Rhamnose-induced propanediol oxi-doreductase in Escherichia coli: purification, properties, and comparison withthe fucose-induced enzyme. J. Bacteriol. 140:320–326.

9. Bouvet, O. M. M., P. Lenormand, E. Ageron, and P. A. D. Grimont. 1995.Taxonomic diversity of anaerobic glycerol dissimilation in the Enterobacte-riaceae. Res. Microbiol. 146:279–290.

10. Bouvet, O. M. M., P. Lenormand, J. P. Carlier, and P. A. D. Grimont. 1994.Phenotypic diversity of anaerobic glycerol dissimilation shown by 7 entero-bacterial species. Res. Microbiol. 145:129–139.

11. Candy, J. M., and R. G. Duggleby. 1998. Structure and properties of pyruvatedecarboxylase and site-directed mutagenesis of the Zymomonas mobilis en-zyme. Biochim. Biophys. Acta 1385:323–338.

12. Cooper, R. A. 1984. Metabolism of methylglyoxal in microorganisms. Annu.Rev. Microbiol. 38:49–68.

13. Dharmadi, Y., A. Murarka, and R. Gonzalez. 2006. Anaerobic fermentationof glycerol by Escherichia coli: a new platform for metabolic engineering.Biotechnol. Bioeng. 94:821–829.

14. Dharmadi, Y., and R. Gonzalez. 2005. A better global resolution function and anovel iterative stochastic search method for optimization of high-performanceliquid chromatographic separation. J. Chromatogr. A 1070:89–101.

15. Gonzalez, R., A. Murarka, Y. Dharmadi, and S. S. Yazdani. 2008. A newmodel for the anaerobic fermentation of glycerol in enteric bacteria: trunkand auxiliary pathways in Escherichia coli. Metab. Eng. 10:234–245.



.asm.org/

Dow

nloaded from

http://aem.asm.org/

16. Hayashi, S., and E. C. C. Lin. 1967. Purification and properties of glycerolkinase from Escherichia coli. J. Biol. Chem. 242:1030–1035.

17. Heyndrickx, M., K. Vandemeulebroecke, P. Scheldeman, K. Kersters, P.DeVos, N. A. Logan, A. M. Aziz, N. Ali, and R. C. W. Berkeley. 1996. Apolyphasic reassessment of the genus Paenibacillus, reclassification of Bacil-lus lautus (Nakamura 1984) as Paenibacillus lautus comb. nov. and of Bacilluspeoriae (Montefusco et al. 1993.) as Paenibacillus peoriae comb. nov., andemended descriptions of P. lautus and of P. peoriae. Int. J. Syst. Evol.Microbiol. 46:988–1003.

18. Hopper, D. J., and R. A. Cooper. 1971. Regulation of Escherichia coli meth-ylglyoxal synthase: new control site in glycolysis? FEBS Lett. 13:213–216.

19. Kang, Y. S., T. Durfee, J. D. Glasner, Y. Qiu, D. Frisch, K. M. Winterberg,and F. Blattner. 2004. Systematic mutagenesis of the Escherichia coli ge-nome. J. Bacteriol. 186:4921–4930.

20. Kessler, D., I. Leibrecht, and J. Knappe. 1991. Pyruvate-formate-lyase-de-activase and acetyl-CoA reductase activities of Escherichia coli reside on apolymeric protein particle encoded by adhE. FEBS Lett. 281:59–63.

21. Kistler, W. S., and E. C. C. Lin. 1972. Purification and properties of theflavine-stimulated anaerobic l-�-glycerophosphate dehydrogenase of Esche-richia coli. J. Bacteriol. 112:539–547.

22. Ko, J., I. Kim, S. Yoo, B. Min, K. Kim, and C. Park. 2005. Conversion ofmethylglyoxal to acetol by Escherichia coli aldo-keto reductases. J. Bacteriol.187:5782–5789.

23. Kornberg, H. L., and R. E. Reeves. 1972. Inducible phosphoenolpyruvate-dependent hexose phosphotransferase activities in Escherichia coli. Biochem.J. 128:1339–1344.

24. Lin, E. C. C. 1976. Glycerol dissimilation and its regulation in bacteria.Annu. Rev. Microbiol. 30:535–578.

25. Misra, K., A. B. Banerjee, S. Ray, and M. Ray. 1996. Reduction of methyl-glyoxal in Escherichia coli K12 by an aldehyde reductase and alcohol dehy-drogenase. Mol. Cell. Biochem. 156:117–124.

26. Murarka, A., Y. Dharmadi, S. S. Yazdani, and R. Gonzalez. 2008. Fermen-tative utilization of glycerol by Escherichia coli and its implications for theproduction of fuels and chemicals. Appl. Environ. Microbiol. 74:1124–1135.

27. Nakano, M. M., Y. P. Dailly, P. Zuber, and D. P. Clark. 1997. Character-ization of anaerobic fermentative growth of Bacillus subtilis: identification offermentation end products and genes required for growth. J. Bacteriol.179:6749–6755.

28. Nielsen, J., J. Villadsen, and G. Liden. 2003. Bioreaction engineering prin-ciples, p. 60–73. Kluwer Academic/Plenum Publishers, New York, NY.

29. Pettigrew, D. W., G. J. Yu, and Y. G. Liu. 1990. Nucleotide regulation ofEscherichia coli glycerol kinase: initial-velocity and substrate binding studies.Biochemistry 29:8620–8627.

30. Saikusa, T., H. Rhee, K. Watanabe, K. Murata, and A. Kimura. 1987.

Metabolism of 2-oxoaldehydes in bacteria: purification and characterizationof methylglyoxal reductase from Escherichia coli. Biosci. Biotechnol. Bio-chem. 51:1893–1899.

31. Sawers, R. G., and D. P. Clark. July 2004, posting date. Chapter 3.5.3,Fermentative pyruvate and acetyl-coenzyme A metabolism. In R. Curtiss IIIet al. (ed.), EcoSal—Escherichia coli and Salmonella: cellular and molecularbiology. ASM Press, Washington, DC. http://www.ecosal.org/.

32. Sawers, R. G., M. Blokesch, and A. Bock. September 2004, posting date.Chapter 3.5.4, Anaerobic formate and hydrogen metabolism. In R. CurtissIII et al. (ed.), EcoSal—Escherichia coli and Salmonella: cellular and molec-ular biology. ASM Press, Washington, DC. http://www.ecosal.org/.

33. Schepers, H. J., S. Bringermeyer, and H. Sahm. 1987. Fermentation ofD-xylose to ethanol by Bacillus macerans. Z. Naturforsch. 42c:401–407.

34. Schirawski, J., and G. Unden. 1995. Anaerobic respiration of Bacillus mac-erans with fumarate, TMAO, nitrate and nitrite and regulation of the path-ways by oxygen and nitrate. Arch. Microbiol. 163:148–154.

35. Schryvers, A., and J. H. Weiner. 1982. The anaerobic sn-glycerol-3-phos-phate dehydrogenase: cloning and expression of the glpA gene of Escherichiacoli and identification of the glpA products. Can. J. Biochem. 60:224–231.

36. Shone, C. C., and H. J. Fromm. 1981. Steady-state and pre-steady-statekinetics of coenzyme A-linked aldehyde dehydrogenase from Escherichiacoli. Biochemistry 20:7494–7501.

37. Truniger, V., and W. Boos. 1994. Mapping and cloning of gldA, the structuralgene of the Escherichia coli glycerol dehydrogenase. J. Bacteriol. 176:1796–1800.

38. Walz, A. C., R. A. Demel, B. de Kruijff, and R. Mutzel. 2002. Aerobicsn-glycerol-3-phosphate dehydrogenase from Escherichia coli binds to thecytoplasmic membrane through an amphipathic �-helix. Biochem. J. 365:471–479.

39. Weimer, P. J. 1984. Control of product formation during glucose fermenta-tion by Bacillus macerans. J. Gen. Microbiol. 130:103–111.

40. Weimer, P. J. 1984. Fermentation of 6-deoxyhexoses by Bacillus macerans.Appl. Environ. Microbiol. 47:263–267.

41. Williams, A. G., and S. E. Withers. 1985. The production of hemicellulose-degrading enzymes by Bacillus macerans in anaerobic culture. Appl. Micro-biol. Biotechnol. 22:318–324.

42. Yazdani, S. S., and R. Gonzalez. 2007. Anaerobic fermentation of glycerol: a path toeconomic viability for the biofuels industry. Curr. Opin. Biotechnol. 18:213–219.

43. Yazdani, S. S., and R. Gonzalez. 2008. Engineering Escherichia coli for theefficient conversion of glycerol to ethanol and co-products. Metab. Eng.10:340–351.

44. Zhu, M. M., F. A. Skraly, and D. C. Cameron. 2001. Accumulation ofmethylglyoxal in anaerobically grown Escherichia coli and its detoxificationby expression of the Pseudomonas putida glyoxalase I gene. Metab. Eng.3:218–225.



.asm.org/

Dow

nloaded from

http://aem.asm.org/

anaerobic fermentation of glycerol in paenbacillus macerans

Documents