Appl Microbiol Biotechnol (1988) 27:549--552
Microbiology
The influence of lignin degradation products
on xylose fermentation by Klebsiella pneumoniae
Nora K. Nishikawa, Roger Sutcliffe, and John N. Saddler
Biotechnology & Chemistry Dept., Forintek Canada Corp., 800 Montreal Rd., Ottawa, Ontario, Canada, K1G 3Z5
Summary. The inhibitory effects of seven closely
related lignin degradation products on xylose fer-
mentation by Klebsiella pneumoniae were studied.
Compounds were added in varying concentra-
tions. Less heavily substituted phenolics (at con-
centrations of, 0.1--0.4 g/l) were more inhibitory
to growth and solvent production than vanillyl or
syringyl derivatives. All of the cultures recovered
from this inhibition after a prolonged incubation
period. When the mechanism of the organism's
recovery was investigated, GC and LC analysis
showed that 43.5% of the vanillin was metabol-
ized to vanillyl alcohol. Several unidentifiable
compounds were also detected in trace amounts.
K. pneumoniae also metabolized vanillyl alcohol
(54% of original supplement) and syringaldehyde;
however, unlike vanillin, there was no predomi-
nant metabolite derived from these compounds.
None of the metabolites derived from vanillyl al-
cohol could be identified while only the corre-
sponding alcohol and trimethoxybenzene were
identified among the syringaldehyde derived me-
tabolites.
Introduction
Fermentation of xylose to 2,3-butanediol by Kleb-
siella pneumoniae, when hydrolyzed water-sol-
ubles from steam-treated aspenwood were used as
the substrate, has been shown to be adversely af-
fected by the presence of inhibitory materials (Yu
et al. 1982; Saddler et al. 1983). The major inhibi-
tors produced during pretreatment of lignocellu-
losic substrates have been shown to be, water-sol-
Offprint requests to: N. K. Nishikawa
Applied
Biotechnology
© Springer-Verlag 1988
uble lignin degradation products, that are re-
leased during the pretreatment (Saddler et al.
1983; Clark and Mackie 1984; McCaskey and
Martin 1984). Some of the inhibitory effects have
been reported to be partially overcome by meth-
ods such as strain acclimatization, increased ino-
culum size or substrate treatment (Parekh et al.
1986; Beck 1986). Recently, other workers have
investigated the inhibitory effect of a small num-
ber of lignin compounds on yeasts (Tran and
Chambers 1985; Ando et al. 1986). However, little
attention has focused on the levels of individual
inhibitory compounds present, the accumulative
effect after fed-batch addition of these substrates,
or the mechanism by which their inhibitory ac-
tions can be overcome.
Analysis of the lignin degradation products
that accumulate in the water-soluble fraction of
pretreated aspenwood indicated that seven
closely related products predominate. In this
work, we have followed the growth and fermenta-
tion profile of K. pneumoniae as well as quantitat-
ing the disappearance of the supplemented lignin
derivatives and the products derived from these
compounds.
Materials and methods
Microoryanism. Klebsiella pneumoniae (formerly Aerobacter
aerogenes N R R L B-199; ATCC 8724) was obtained from the
National Research Council of Canada (NRCC 3006).
Supplementary compounds. Supplementary chemicals were all
commercially available and used as received.
Growth and solvent production experiments. K. pneumoniae was
grown in modified Kp medium, pH 6.5 (Yu and Saddler 1982).
Studies were carried out using 10 mL portions of medium in
60 mL Wheaton vials. Test compounds, with the exception of
, vanillyl alcohol, were prepared as stock solutions (pH 6.5).
Appropriate aliquots of these stock solutions were added to
550 N. K. Nishikawa et al.: Effect of phenolics on xylose fermentation

duplicate vials of concentrated Kp medium and the dilution of Table 1. Growth and solvent production from xylose (2% w/v)
the supplements was adjusted with distilled water. Vanillyl al- for K. pneumoniae in the presence of inhibitory compounds
cohol was not readily soluble at high concentrations and was
added to vials as a dry supplement. The test media were then Compound Conc. Growth Solvent
sterilized by autoclaving (121 ° C, 15 min). Production
Xylose and acetic acid (for enhancement of solvent pro- (g/l) (% of Con- (% of Con-
duction; Yu and Saddler 1982) were added to the media from trol) trol)
pre-sterilized stock concentrations using a 1:10 dilution of
20% (w/v) and 5% (w/v) solutions, respectively. When 5% xy- 24h 48h 24h 48h
lose was required as a substrate, a 2 ml volume of 25% (w/v)
pre-sterilized xylose was added. I 0.1 92.5 98.8 72.1 93.1
A 5% (v/v) inoculum was added to experimental vials us- Ho._4/('~,~~_ co=H 0.2 84.7 94.0 62.3 78.4
ing Kp cultures grown aerobically with 1% glucose (Yu and 0.3 78.3 88.4 49.0 70.9
Saddler 1982). The test cultures were then incubated under fin- 0.4 74.2 83.0 37.5 65.6
ite air at 30°C (150 rpm shaking). Samples were analyzed at
24 h or 48 h. During studies of long term metabolism of test II ,----, 0.1 89.0 93.6 65.3 89.0
compounds, samples were taken at stated time intervals. HO~/(~_CHO 0.2 50.7 63.3 35.2 71.8
0.3 31.7 44.4 19.0 58.3
Analytical methods. Culture growth was monitored by reading 0.4 21.7 31.7 12.1 29.1
optical density at 540 nm after appropriate dilution of the III MeO 0.3 84.3 91.3 81.9 90.5
sample. Fermentation products were determined by GC as de- _ ~ 0.4 77.9 88.4 70.2 91.1
scribed previously (Yu and Saddler 1982). HO CO.=H 0.5 65.8 83.4 68.0 80.7
Fermented test media were analyzed for metabolism of in- 0.6 64.4 81.3 60.8 85.2
hibitor supplements using GC. Duplicate test media were
IV t~eo 0.3 94.9 92.5 85.4 102.
combined and centrifuged at 7500 rpm for 10 min. The super-
0.4 90.1 90.0 74.5 107.
natant was collected, saturated with sodium chloride, and then HO CHO 0.5 80.0 91.6 56.1 99.1
extracted with ether (2 x 25 ml). The combined ether extracts
0.6 87.0 91.5 27.1 110.
were washed with water, dried over magnesium sulphate, and
evaporated to remove the solvent. For GC analysis, the sam- V MeO 0.3 93.8 101. 86.8 97.1
ples were diluted to an estimated 1 m g / m l range based on ori- _ ~ 0.4 93.0 100. 79.8 96.3
ginal supplement concentration using methylene chloride with HO CH=OH 0.5 89.1 98.1 79.2 94.2
salicylaldehyde added as internal standard. GC analysis was 0.6 91.2 98.9 73.3 93.1
MeO
performed with a DB-I capillary column (10 m length; Chro-
matographic Specialties Ltd., Brockville, Ontario, Canada) on
a Varian model 3700 GC (Varian Assoc., Palo Alto, Ca). The
VI
HO~CO=H
~ 0.2
0.3
0.4
91.3
90.7
88.6
94.3
93.6
92.5
87.8
86.8
88.9
91.5
93.4
88.7
injector and FID detector temperatures were 280°C and UeO 0.5 80.1 94.0 74.6 92.2
300°C, respectively. Initial column temperature was 40°C
(1 rain) which was then raised at 10°C per min to 300 ° C. The VII MeOx__., 0.2 90.0 82.5 69.4 106.
linear carrier gas velocity was 20 ml/s. Reverse-phase LC was HO~/~'~'~"'.,,_.CHO 0.3 87.9 64.0 57.1 98.0
also used to verify the GC results. 0.4 39.4 62.8 38.0 106.
G C / M S analysis was performed at the Division of Chem- MeO 0.5 38.4 60.1 39.6 103.
istry NRCC, Ottawa using similar GC conditions to above but VIII MeO~-~-'~ 0.3 75.7 84.5 71.5 94.7
using an HP5890 chromatograph equipped with a Hewlett 0.4 76.6 85.8 76.0 96.3
Packard HP5970B mass selective detector. MeO 0.5 70.2 82.8 75.0 93.4

I, p-hydroxybenzoic acid; II, p-hydroxybenzaldehyde; III,

Results and discussion vanillic acid; IV, vanillin; V, vanillyl alcohol; VI, syringic
acid; VII, syringaldehyde; VIII, veratrole

Initially Klebsiella pneumoniae was grown in me-
dia containing xylose (2% w/v) supplemented
with varying initial concentrations (0.1--0.6 g/l)
of phenolic compounds I to VII (Table 1), to es-
tablish the concentrations required to inhibit
growth or butanediol fermentation.
All cultures showed a decrease in growth and
solvent production after 24 h (Table 1). Higher
concentrations of vanillyl (III--V) and syringyl
(VI, VII) compounds were required to produce
the same reduction as caused by the less heavily
ring-substituted phenolics (I, II). On prolonged
incubation (48 h) most of the cultures made at
least a partial recovery in both growth and solvent
production. The cultures supplemented at the
highest levels with vanillin (IV) and syringalde-
hyde (VII) showed the most dramatic recovery.
Previously we had shown that xylose utiliza-
tion by K. pneurnoniae was inhibited when the hy-
drolyzed water-soluble fraction from steam-
treated aspenwood was used at a 5% (w/v) sugar
substrate concentration (Saddler et al. 1983). Su-
gar streams from hydrolysed hardwoods have
been demonstrated to contain as much as 6 g/l of
phenolic compounds (McCaskey and Martin
1984). With pretreated aspenwood, only a few of
the individual components have been identified
and quantitated. Stanek (1958) demonstrated the
presence of p-hydroxybenzoic acid (I), vanillin
(IV), vanillic acid (III), syringaldehyde (VII), and
N. K. Nishikawa et al.: Effect of phenolics on xylose fermentation
syringic acid (VI) as well as several minor compo-
nents. We had previously determined that the wa-
ter-soluble stream from steam-treated aspenwood
when used a the substrate for growth of K. pneu-
moniae contained, for example, vanillin and sy-
ringaldehyde at concentrations of 0.03 and
0.09 g/l, respectively. The concentration of these
various components present in pretreated aspen-
wood are known to be influenced by the severity
of the steam-treatment, the catalyst used, and the
fractionation procedures used to separate the var-
ious lignin, cellulose and hemicellulose compo-
nents. The relatively high concentrations of indi-
vidual degradation products required before
growth and solvent production are severely inhi-
bited demonstrates that, even using fed-batch
conditions in which these materials can be ex-
pected to accumulate, no single compound could
be solely responsible for inhibition. If the numer-
ous components present in the water-soluble
stream have a potency similar to the chemicals re-
ported in Table 1 and their effects are cumulative,
as shown recently for mannose fermentation by K.
pneumoniae (Tran and Chambers 1986), then the
combined effective concentration required for sig-
nificant inhibition to occur should be easily at-
tainable.
Recently Japanese workers have shown that
the aromatic monomers from steam pretreated as-
penwood have a significant effect on ethanol pro-
duction by Saccharomyces cerevisiae (Ando et al.
1986). They found the functionality substituted on
the aromatic ring to be related to toxicity, where
acids were less inhibitory than aldehydes. How-
ever they limited their investigation to a single
concentration of each material. Our results indi-
cate that for the adoption of inhibitor removal
methods such as solvent-extraction or ion-ex-
change, extremely efficient methods must be de-
veloped to overcome the combined effect result-
ing from even low residual concentrations of indi-
vidual components. From the differing initial ef-
fects on growth and solvent production (the ex-
tent and timescale of recovery), it was difficult to
infer an order of toxicity of chemicals to the or-
ganism or the removal of which compounds (e. g.
acids, aldehydes, etc.) would be most important in
the production of an easily fermentable hydroly-
sate stream.
The observed ability of the organism to re-
cover rapidly from even relatively high concentra-
tions of added phenolics led us to investigate the
mechanism of recovery. The fermentation of xy-
lose (5% w/v) supplemented with vanillin or sy-
ringaldehyde was examined. These systems were
551
chosen because of the speed and extent of recov-
ery of solvent production. Higher substrate levels
were used because in normal fermentation we
have found that it takes 4 days for the xylose to be
fully utilized as opposed to 2 days at the lower
substrate level used in the earlier studies.
The fermentation profiles for control and van-
illin added cultures are shown in Fig. la. With the
control the rapid onset of growth was followed
approximately 12 h later by solvent production.
In the vanillin supplemented culture there was a
significant induction period before growth was es-
tablished. However once initiated the develop-
ment paralleled that of the control. Again solvent
production was first apparent about 12 h after ini-
tial growth and paralleled that of the control.
Similar results for syringaldehyde (0.3 g/l) sup-
plemented cultures, were observed, although only
80% of the growth and 96% of the normal solvent
production of the control were obtained (results
not shown).
Analysis of extracts of the culture media with
added vanillin over the course of the fermentation
showed a decrease in vanillin concentration (Fig.
lb), that coincided with the onset of growth. GC
and LC analysis of the extracts showed the pro-
duction of only one major product. This was iden-
tified by G C / M S and IR spectroscopy to be van-
illyl alcohol. The production profile of this alco-
hol is also shown in Figure lb. Vanillyl alcohol
accounted for 43.5% of the original vanillin ad-
ded. Several minor products were also present at
extremely low concentrations. Although veratrole
(VIII) was identified, the other compounds ap-
peared to be self-condensation products of vanil-
lin that had been further modified and their iden-
tities could not be conclusively established.
Cultures supplemented with syringaldehyde
yielded a multitude of minor products, however
no one product dominated. Using G C / M S we
were able to identify the corresponding alcohol
formed by reduction of syringaldehyde as well as
trimethoxybenzene. However, the majority of
products were higher molecular weight com-
pounds presumably formed by various cross-con-
densation reactions and modified by further com-
binations of demethoxylation, methylation, reduc-
tion etc. We could not positively identify these
compounds as it would require the production
and isolation of these materials as well as the syn-
thesis of authentic samples for comparison.
The results suggest that, at least in the case of
vanillin, inhibition is overcome by the metabolism
of this compound to a less toxic material such as
vanillyl alcohol (V) (see Table 1). As veratrole
552 N.K. Nishikawa et al.: Effect of phenolics on xylose fermentation

15- 6r- bilized cell reactor. They suggested that other al-
dehydes were also bioconverted by S. cerevisiae,
but no product identification or distribution were
given.
Our results, even from the limited analysis on
v

o vanillin, vanillyl alcohol and syringaldehyde, de-

monstrate that the metabolism of inhibitors by K.
8 pneumoniae is extremely sensitive to structure.
Furthermore, the expectation that accumulation
I II I o co~t,-o, of inhibitory compounds on fed-batch addition
o p // / evanillinadded
would occur is simplistic and misleading. It was
/ // / Solvent Production
I I ocon! :p, _ . apparent that some of the inhibitors could be me-
tabolised by the fermentative organism, which re-
suited in either the removal of toxic materials, or
the production of further inhibitors.
0
0.6,
Acknowledgement. This work was funded by Energy, Mines
and Resources as a part of the liquid fuels program.

0.5
i (b)
0.4
~0.3
o
0.2
0.1
02 3 4 1 5 6
INCUBATION TIME (days)
Fig. 1. (a) Fermentation profiles for K. pneumoniae grown on
xylose (5% w/v) with added vanillin (0.6 g/l) and (b) metabol-
ism of vanillin and production of vanillyl alcohol during the
fermentation
(VIII) was identified as one of the derived meta-
bolic products, the effects of differing concentra-
tions of this material were examined. The results
in Table 1 show that this compound is relatively
toxic, however, the maximum amount that was
detected in the fermentation broth was always less
than 0.05 g/1. As the final concentration of vanil-
lyl alcohol did not account for the complete meta-
bolism of vanillin, the possible metabolism of
vanillyl alcohol was also examined. It was found
that 54% of the original material added was meta-
bolized within 7 days. Again there was no forma-
tion of a major product although a range of uni-
dentified compounds were detected at low con-
centrations.
Recently, De Wulf et al. (1986) also demon-
strated the bioconversion of vanillin to vanillyl al-
cohol by a yeast, Saccharomyces cerevisiae. These
workers optimized fermentation conditions for
the production of vanillyl alcohol using an immo-
• vanillin
o vanillyl alcohol
References
Ando S, Arai I, Kiyoto K, Hanai S (1986) Identification of
aromatic monomers in steam-exploded poplar and their in-
fluences on ethanol fermentation by Saccharomyees cerevi-
siae. J Ferment Technol 64:567--570
Beck MJ (1986) Effect of intermittent feeding of cellulose hy-
drolyzate to hemicellulose hydrolyzate on ethanol yield by
Paehysolen tannophilus. Biotechnol Lett 8 : 513-- 516
Clark TA, Mackie KL (1984) Fermentation inhibitors in wood
hydrolysates derived from the softwood Pinus radiata. J
Chem Tech Biotechnol 34B: 101--110
De Wulf O, Thonart P, Gaignage Ph, Marlier M, Paris A, Pa-
7
quot M (1986) Bioconversion of vanillin to vanillyl alcohol
by Saccharomyces cerevisiae. Biotechnol Bioeng Syrup
17:605--616
McCaskey TA, Martin CP (1984) '°Components of hardwood
hydrolysate inhibitory to Pachysolen tannophilus" in Pro-
ceedings of the Technical Association of the Pulp and Pa-
per Industry. Research and Development Conference, Ap-
pleton, WS. Sept 30--Oct 3, 1984
Parekh SR, Yu S, Wayman M (1986) Adaptation of Candida
shehatae and Pichia stipitis to wood hydrolysates for in-
creased ethanol production. Appl Microbiol Biotechnol
25:300--304
Saddler JN, Mes-Hartree M, Yu EKC, Brownell HH (1983)
Enzymatic hydrolysis of various pretreated lignocellulosic
substrates and the fermentation of the liberated sugars to
ethanol and butanediol, Biotechnol Bioeng Symp 13 : 225--238
Stanek DA (1958) A study of low-molecular weight phenols
formed upon hydrolysis of aspenwood. TAPPI 41:601--609
Tran AV, Chambers RP (1985) Red oak wood derived inhibi-
tots in the ethanol fermentation of xylose by Pichia stipitis
CBS 5776. Biotechnol Lett 7:841--846
Tran AV, Chambers RP (1986) Lignin and extractives derived
inhibitors in the 2,3-butanediol fermentation of mannose-
rich prehydrolysates. Appl Microbiol Biotechno123 : 191-- 197
Yu EKC, Saddler JN (1982) Enhanced production of 2,3-buta-
nediol by Klebsiella pneumoniae grown on high sugar con-
centrations in the presence of ~icetic acid. Appl Environ
Microbiol 44:777--784
Yu EKC, Levitin N, Saddler JN (1982) Production of 2,3-buta-
nediol by Klebsiella pneumoniae grown on acid hydrolyzed
wood hemicellulose. Biotechnol Lett 4:741--746
Received June 11, 1987/Accepted September 15, 1987