Download pdf - Visualizing lignin coalescence and migration through maize cell walls following thermochemical pretreatment

ARTICLE

Visualizing Lignin Coalescence andMigration Through Maize Cell WallsFollowing Thermochemical Pretreatment

Bryon S. Donohoe,1 Stephen R. Decker,1 Melvin P. Tucker,2 Michael E. Himmel,1

Todd B. Vinzant1

1Chemical and Biosciences Center, National Renewable Energy Laboratory,

1617 Cole Boulevard, Golden, Colorado 80401; telephone: 303-384-7773; fax: 303-384-7752;

e-mail: [email protected] Bioenergy Center, National Renewable Energy Laboratory, 1617 Cole Boulevard,

Golden, Colorado 80401

Received 4 December 2007; revision received 3 March 2008; accepted 17 April 2008

Published online 2 May 2008 in Wiley InterScience (www.interscience.wiley.com). DO
I 10.1002/bit.21959
ABSTRACT: Plant cell walls are composed primarily ofcellulose, hemicelluloses, lignins, and pectins. Of thesecomponents, lignins exhibit unique chemistry and physio-logical functions. Although lignins can be used as a productfeedstock or as a fuel, lignins are also generally seen as abarrier to efficient enzymatic breakdown of biomass tosugars. Indeed, many pretreatment strategies focus onremoving a significant fraction of lignin from biomass tobetter enable saccharification. In order to better understandthe fate of biomass lignins that remain with the solidsfollowing dilute acid pretreatment, we undertook a struc-tural investigation to track lignins on and in biomass cellwalls. SEM and TEM imaging revealed a range of dropletmorphologies that appear on and within cell walls of pre-treated biomass; as well as the specific ultrastructural regionsthat accumulate the droplets. These droplets were shown tocontain lignin by FTIR, NMR, antibody labeling, and cyto-chemical staining. We provide evidence supporting the ideathat thermochemical pretreatments reaching temperaturesabove the range for lignin phase transition cause lignins tocoalesce into larger molten bodies that migrate within andout of the cell wall, and can redeposit on the surface of plantcell walls. This decompartmentalization and relocalizationof lignins is likely to be at least as important as ligninremoval in the quest to improve the digestibility of biomassfor sugars and fuels production.

Biotechnol. Bioeng. 2008;101: 913–925.

� 2008 Wiley Periodicals, Inc.

KEYWORDS: lignin; biomass; pretreatment; electronmicroscopy; tomography

Correspondence to: B.S. Donohoe

� 2008 Wiley Periodicals, Inc.

Introduction

Lignocellulosic biomass has long been recognized as apotential low-cost source of mixed sugars for fermentationto fuel ethanol, and now the objective is to make this processcost competitive in today’s market. Replacing 20% of U.S.2004 finished motor gasoline demand (or about 35 billiongallons) with ethanol by 2017 will require a significantincrease in ethanol production over today’s corn starch-based industry (President Bush State of the Union Address,2007). Current technology for the conversion of someagricultural resides and energy trees utilizes biochemicalprocessing that includes pretreatment, enzymatic hydrolysis,and fermentation (Himmel et al., 1999). However, thisprocess needs to be more efficient and less costly tocommercialize. In order to ensure a successful transitionfrom existing to 2030 technologies, pursuing knowledge-based solutions to critical barriers, such as plant cell walldecomposition, is critical.

Plant cell walls are composed primarily of cellulose,hemicelluloses, lignins, and pectins. Cellulose, a crystalline,insoluble polymer of cellobiose, comprises about 50% of theplant biomass. Although cellulose does not degrade easily, itwill hydrolyze to glucose by the synergistic action of threedistinct classes of enzymes: endoglucanases, exoglucanases,and cellobiases (Nidetzky et al., 1994). In contrast to theinsoluble linear cellulose homopolymer, hemicelluloses arewater- or base-soluble heteropolymers, comprised of a varietyof branched and substituted polysaccharides. Pectins are acidicpolysaccharides that retain large amounts of water and alongwith lignin, comprise much of the middle lamella. Thechemistry and physiological functions of lignins are quitedistinct from those of either the cell wall celluloses orhemicelluloses. They are formed as polymeric entities withinthe cell wall, where they help to reinforce the plant walls of the

Biotechnology and Bioengineering, Vol. 101, No. 5, December 1, 2008 913

vasculature (Sarkanen, 1998). Lignins enable vascular plants toform their water/nutrient conducting cells and to provide amechanism for withstanding compressive forces acting on theplant overall. While the lignin chemistry of primitive plants isnot well understood, it is well established that lignincomposition in higher plants is derived from the threemonolignols, p-coumaryl, p-coniferyl, and E-sinapyl alcohols;as well as p-hydroxycinnamyl alcohol-monolignol esters ingrasses (Sarkanen and Hergert, 1971).

Lignins are indeed a major component of the plant cellwall matrix and an important consideration for any biomassconversion regime. Although lignins have multiple potentialuses as a product feedstock or as a fuel in their own right,lignins are also generally seen as a formidable barrier toefficient enzymatic breakdown of biomass to sugars (Konget al., 1992). Many pretreatment strategies focus onremoving a significant fraction of lignin from the biomassto better enable saccharification (Himmel et al., 1997), andlignin content has also been shown to impact simultaneoussaccharification and fermentation (SSF) directly (Vinzantet al., 1997). However, some mild pretreatment strategies,such as dilute acid and hot water, significantly increase thedigestibility of biomass without removing much of theinsoluble lignin content. This paradox highlights the factthat details of the structural complexity of biomass cell wallsat the molecular level remain unclear.

In order to better understand the fate of biomass ligninsthat remain with the solids during dilute acid treatment, weundertook a structural investigation to track lignins withinbiomass cell walls. Lignins are notoriously complex andchallenging to study because they can be variable incomposition and degree of cross-linking (Grabber, 2005)and the formation and deposition of lignins in cell wallsvaries by cell type and tissue type (Donaldson, 2001;Morrison et al., 1998).

For this study, we have focused our microscopic analysis onthe xylem and sclerenchyma cells in the rind of corn stoverstalks. The logic behind choosing these cell types is three fold.First, we have followed a path of maximum mass fraction andmaximum recalcitrance from the anatomic scale down to thecellular scale and chosen the heaviest, most recalcitrant cellwall types to study. A related factor is that we looked at cellsthat contain significant amounts of lignins (Boerjan et al.,2003). Third, it is important to be consistent with the biomasstissue and cell types analyzed by microscopy. Because ofdramatic differences in their resistance to pretreatment,different tissue and cell types change in appearance followingpretreatment making it easy to mistakenly analyze differentbiomass fractions before and after pretreatment.

Methods

Corn Stem Rind

Corn stem rind was selected for this study because of its highmass and high lignin content. Field-senesced corn stems(Round-Up Ready Pioneer hybrid-36N18) were hand-cut

914 Biotechnology and Bioengineering, Vol. 101, No. 5, December 1, 2008

from a single field (Gustafson Farm, Weld County, CO) inJanuary 2005. Stalks were air dried to a moisture content of�10% prior to dissecting. The rind portion was peeled fromthe fourth and fifth internodes from the base of the cornstalk. The rind was then hand-cut into 3–5 cm long� 0.3 cmdiameter pieces or alternatively milled to 20-mesh in aThomas Wiley (Swedesboro, NJ) cutting mill.

Pretreatments

In order to generate anatomically relevant samples forimaging, corn stover rind chips were treated with dilutesulfuric acid or hot water in either batch or flow throughreactors. Dilute acid pretreatments were run at 1508C in0.8% H2SO4 (w/w) for 20 min in 15 mL (25 mm dia.� 30 mmlong) gold coated Swage-Lok (Cleveland, OH) pipe-reactorsheated to temperature (�8 min ramp time) by submersionin an air-fluidized sand bath. The reactors were cooled byquenching in an ice water bath (�1 min ramp time). Thesesamples were used for SEM, immuno-SEM, TEM, andTEM-tomography imaging of in situ lignin droplets on andwithin pretreated biomass samples. Compositional analysisand enzymatic hydrolysis on similar samples was reportedpreviously (Selig et al., 2007).

Isolated lignin droplets were collected by direct adsorp-tion of lignins onto filter paper. The lignin preparations weremade by pretreating corn stover rind chips in 0.8% H2SO4 at1708C in a flow-through reactor with a total heat-up andresidence time of 12 min or with hot water only at 2208C.The flow rate was 9.9 mL/min with a total of 10 reactorvolumes. After pretreatment, both preparations had formedtwo distinct fractions as seen previously with Organosolvlignin preparations, a solid dark-brown fraction, and a finecolloidal light-brown fraction. The light-brown fraction waswashed and suspended in Millipore water and then pipettedonto Whatman #1 filter paper for imaging. These isolatedlignin droplets were used for FTIR and NMR analysis.Organosolv pretreatments were also applied to ground cornstover samples to generate reference lignin samples.

Reference Lignin Samples

Reference lignin samples used in this study were isolatedfrom whole corn stover and Jack Pine sulfite pulp liquor.The methods used to isolate lignin from the corn stover werea combined ball milling and dioxane extraction process asdescribed by Sun and coworkers (Sun et al., 2004) and anorganosolv extraction process described by Black andcoworkers (Black et al., 1998). Lignin prepared from JackPine was isolated as described previously (Gidh et al., 2006).Organosolv, ball milled-dioxane, and Jack Pine ligninpreparations were pretreated at 1708C in 0.8% sulfuric acidto produce isolated droplets from known lignin sources.Pretreatment of these preparations created a fine suspensionin the pretreatment hydrolysate. The suspension wasdeposited onto Whatman #1 filter and prepared for imagingby SEM and analysis by FTIR and NMR.

Scanning Electron Microscopy (SEM)

Imaging by scanning electron microscopy (SEM) wasperformed using an FEI Quanta 400 FEG instrument underhigh vacuum operating with the Everhart Thorney Detectorand a solid-state backscatter detector. Samples were pre-pared for imaging by freeze-drying. Dry samples weremounted on aluminum stubs using carbon tape withconductive silver paint applied to the sides to reduce samplecharging. The samples were then sputter-coated with 10–12 nm Ir using a Cressington (Waterford, England) model203 sputter coater. Imaging was performed at beamaccelerating voltages from 12.5 to 25 kV.

FTIR and NMR analysis

FTIR spectra were obtained from lignin droplets from thefollowing pretreatments: (1) hot water flow-throughpretreated corn stover that was pretreated at 2208C at aflow rate of 9.9 mL/min with a total of 10 reactor volumes ofhot water, (2) dilute-acid flow-though pretreated cornstover at 1708C in 0.8% sulfuric acid at a 9.9 mL/min flowrate, and (3) Organosolv extracted and dioxane extractedcorn stover.

A Thermo-Nicolet Magna 550 FTIR spectrometer wasused to obtain mid-infrared spectra of samples placed in aSpectra-Tech Continum IR microscope. The microscopewas equipped with a variable aperture (up to 150 mmsquare) objective lens and a high sensitivity liquid nitrogencooled mercury/cadmium/telluride detector (MCT). Sam-ples of pretreated corn stover and lignin derived from thehot water and hot dilute acid flow through experiments weredeposited on separate KBr windows for microscopic andspectroscopic analysis (Tucker et al., 2001). The KBrwindows are transparent in the visible and mid-IR regionused in this study. The samples were washed with ethanolfollowed by deionized water, and air-dried. The driedsamples were then iridium-coated and imaged by SEM.

Solid-state 13C NMR spectra were collected at 4.7 T withcross-polarization (CP) and magic angle spinning (MAS) ina Bruker Avance 200 MHz spectrometer. Spectra wereobtained from a series of solid residues collected from a hotwater flow through reactor, water extractives, ethanolextractives, xylose, or dilute acid pretreated rind residuescollected on filter paper. Solid-state 13C CP/MAS spectro-scopy was not quantitative, but differences in relativepeak intensities were used to identify differences amongsamples.

Immuno-SEM

Anti-lignin antibody was obtained as a gift from Katia Ruelfrom the Centre de Recherches sur les MacromoleculesVegetales, France. For immuno-EM, samples were placed on�15 mL drops of 2.5% non-fat dry milk in 1X PBS-0.1%Tween (PBST) for 20 min, blotted, then directly placed on

�15 mL drops of a-Lignin DHP C/S antibodies 1:10 in 1%milk PBST. Following 3� 1 min rinses, grids were thenplaced on �15 mL drops of a-rabbit (IgG) antibodyconjugated to a 15 nm gold particle (British BioCell, TedPella) diluted 1:100 in PBST. To enhance the labeling for abetter backscatter detection by SEM, grids were placed on adrop of silver enhancement solution (Nanoprobes,Yaphank, NY) for 5 min to achieve silver depositionincreasing the gold/silver particle to the 10–40 nm sizerange. Samples were then rinsed 3� 1 min with PBST andagain in H2O. For every sample secondary-only controlsamples were prepared.

Transmission Electron Microscopy (TEM)

Corn stover tissue was processed using microwave EMprocessing. Samples were fixed 2� 6 min in 2.5%gluteraldehyde buffered in 0.1 M sodium cacodylate buffer(EMS, Hatfield, PS) under vacuum. Dehydration wascarried out in graded ethanol series for 1 min each (15%,30%, 60%, 90%, 2� 100% ethanol). Samples were infiltratedwith LR White resin in the microwave under vacuum andovernight incubation at room temperature (RT) inincreasing concentrations of resin (15%, 30%, 60%, 90%,3� 100% resin, diluted in ethanol). The samples weretransferred to gelatin capsules and the resin polymerized byheating to 608C overnight. LR White embedded sampleswere sectioned to �60 nm with a Diatome diamond knife ona Leica EM UTC ultramicrotome (Leica, Wetzlar, Ger-many). Sections were collected on Formvar coated copper ornickel slot grids (SPI Supplies, West Chester, PA). Gridswere post-stained for 4 min with 2% aqueous uranyl acetateand 2 min with Reynolds lead citrate. Alternatively, gridswere stained with 1% KMnO4 for 10 min to selectivelystain for lignins. Images were taken with a Gatan UltraScan1000 camera (Gatan, Pleasanton, CA) on a FEI Tecnai G220 Twin 200 kV LaB6 TEM (FEI, Hilsboro, OR).

Alternatively, tomograms were created by first capturingdual-axis þ/� 60–658 tilt series using Serial EM (http://bio3d.colorado.edu/). Tomograms were constructed usingan R-weighted back projection algorithm within the IMODsoftware package (http://bio3d.colorado.edu/). Single-axistomograms were then combined to yield dual-axistomograms using a warping algorithm within IMOD(Mastronarde, 1997). Tomograms were displayed andanalyzed using the IMOD software package (Kremeret al., 1996).

Results

A Range of Discrete Droplet Morphologies Appear onthe Cell Wall Surface of Pretreated Biomass

Direct SEM observation of thermal chemically pretreatedcorn stover biomass samples revealed discrete droplets on

Donohoe et al.: Lignin Coalescence and Migration 915

Biotechnology and Bioengineering

the cell wall surface (Fig. 1). Droplets formed on biomasssamples from a variety of pretreatment conditions as long asthe pretreatment temperature rose above�1208C for as littleas 1 min. Droplets were observed on both dilute acid(0.8% H2SO4) and hot water pretreated samples from bothbatch and flow-through type reactors. The pretreatmentchemistry and reactor type has an impact on the abundance,size, and distribution of the droplets (data not shown), buton the whole, the end results appear similar. Figure 1 givesan indication of the range of droplet sizes and coveragedensity that can be seen by SEM analysis of pretreated xylemand sclerenchyma cell wall surfaces.

After observing thousands of droplets by electronmicroscopy, it has become clear that there are multipleclasses of droplets that can be distinguished by morpho-logical criteria, such as size, shape, surface texture, anddensity (Fig. 2). This variability is probably not surprisingconsidering the complexity of lignin and the plant cell wallmatrix. In terms of size, we have observed droplets coveringthree orders of magnitude from 5 nm to 10 mm in diameter.Droplets tend to be spherical with a flattened surfacecontacting the cell wall. However, we have also seen othershapes that may indicate how some droplets are formed.Some droplet shapes appear to be the result to two smallerdroplets fusing together into a larger droplet (Fig. 2A). Thismay partly explain the large range of sizes. While mostdroplets have a smooth exterior, some display a roughsurface coating (Fig. 2B). We hypothesize that this coat mayrepresent a carbohydrate shell structure surrounding alignin core, in which case these droplets constitute a lignin-carbohydrate complex (LCC). Further studies are ongoingto determine the composition of these droplet coats. Whilewe have observed a wide range of droplet classes, the mostabundant surface droplets from these pretreatments are 20–100 nm diameter, spherical, exhibit a smooth surfacetexture, and can be isolated from the corn stover sample byadsorbing onto filter paper placed in a pretreatment reactionvessel with the biomass sample (Fig. 2C). To the extent thatdistinguishing droplet classes provides clues to theircomposition and the mechanism of their formation, weare continuing to investigate these distinct morphologicalfeatures.

Figure 1. SEM micrographs of corn stover rind xylem cell wall surfaces before

(A) and after 0.8% H2SO4 dilute acid pretreatment at 1508C for 20 min (B and C). Image C

is a higher magnification of the region boxed in B. Numerous round droplets are

distinct on the pretreated surface. These initial observations led to an extensive study

of the composition and origin of these dramatic structural features. Scale bars A,

B¼ 2 mm, C¼ 0.5 mm.

Droplets were Demonstrated to Contain Lignin byMultiple Analytical Tools

We have referred to these pretreated biomass surfacefeatures simply as ‘‘droplets’’ and reported their localizationand morphology at the resolution of electron microscopy.Obviously, a compelling question regards their chemicalcomposition. Previous studies have noted the appearance ofdroplets on pretreated biomass surfaces and described themas ‘‘lignin droplets’’ (Micic et al., 2001; Radotic et al., 2005).However, lignins are notoriously complex and difficult toanalyze consistently. Because of this, we have taken a multi-technique approach to establish the following conclusions:


(1) fractionated lignins from corn stover also form dropletswhen exposed to dilute acid and hot water pretreatmentconditions, (2) during pretreatment, droplets from cornstover redeposit on cellulose substrates, (3) the droplets thatform in pretreated corn stover indeed contain lignin asconfirmed by FTIR spectroscopy, NMR analysis, antibodylabeling, and cytochemical staining, and (4) the positiveidentification of droplets on the cell wall surface as lignin

Figure 2. SEM micrographs display three distinct morphological classes of

droplets. (A) Three pairs of droplets that appear to be fusing (arrowheads) to form

larger droplets. (B) Spherical droplets with a rough, coated surface (arrowheads). The

droplets in A and B are naturally redeposited on the cell wall surface following dilute

acid pretreatment. (C) Isolated spherical droplets with smooth surfaces captured on

filter paper. Scale bars¼ 0.5 mm.

containing allows us to localize the more abundantcoalesced lignin containing material within the cell wall.

In order to establish a positive control for analyzingdroplets from corn stover, we looked at lignin fractions todetermine if they also formed droplets during pretreatment.

Figure 3 shows SEM images of lignin droplets fromreference lignin fractions collected on filter paper. Theselignins were derived from corn stover by ball milling anddioxane extraction (Fig. 3A) (Sun et al., 2004), Jack Pinelignin from pulping process (Fig. 3B) (Gidh et al., 2006), andcorn stover by Organosolv treatment (Fig. 3C) (Black et al.,1998). With each sample, droplets appeared nearly identicalto what we had seen on the surface of pretreated corn stoversamples. The lignin fractions were compared with dropletsisolated from 2208C hot water flow-through hydrolyzate byFTIR spectroscopy (Fig. 3D). The FTIR spectra show a peakpattern that is diagnostic for lignin. Note the intense C–Hout-of-plane deformation band for 1,3,5-tri substitutedbenzene rings at 835 cm�1, broad C–O stretch bands for thering alcohols in the carbohydrate moieties near 1,050 cm�1,the intense Ar–O stretch near 1,200–1,240 cm�1 for alkylaromatic ethers, and the intense ring C–C stretching bandsbetween 1,500 and 1,600 cm�1. This spectrum is distinctlydifferent than that obtained from a sample of pure alpha-cellulose (data not shown). The spectra lack the intensealcohol C–O stretching bands near 1,050 cm�1 associatedwith carbohydrates suggesting that the samples do notcontain carbohydrate moieties. Thus, it appears from FTIRspectroscopy that the majority of lignin droplets from cornstover samples treated at 2208C with hot water arecomposed primarily of lignin and not rich in lignin-carbohydrate complexes, which is consistent with a higherseverity pretreatment (Chum et al., 1990; Schell et al., 2003).

To further probe the composition of the lignin droplets,we analyzed them by SEM and NMR. Figure 4 shows insetSEM images of the same samples prepared by hot waterflow-through reactor that were analyzed by FTIR, confirm-ing that the droplets can be isolated on filter paper foranalysis by solid-state NMR. A set of negative controls wasused to investigate the possibility of false-positive peaksassociated with soluble phenolics in corn stover extractives.These controls included pretreating Whatman #1 filterpaper in the presence of acetone extracted corn stover rindsections, corn stover water extractives, and corn stoverethanol extractives. A xylose-only control was also run todetermine if xylose degradation products formed duringpretreatment are producing peaks that overlap with lignin.Figure 4 shows the NMR spectra from these samples. Whilethe cellulose signal from the filter paper dominates each ofthe spectra, a distinct phenolic peak region consistent withlignin can clearly be seen from the flow-through reactorsolids.

Although both FTIR and NMR analysis required that thelignin droplets be isolated from the biomass for analysis, wewere also interested in positively identifying lignin dropletsin situ. In order to label droplets on the cell wall surface, weacquired an anti-lignin antibody as a gift from Katia Ruel,CERMAV/CNRS, France. These antibodies were polyclonalantibodies directed against the syringyl and guaiacylpropane epitopes and had been used successfully to localizelignin in maize cell walls previously (Joseleau and Ruel 1997;Ruel et al., 1999; Ruel et al., 2001). We labeled pretreated



Figure 3. SEM micrographs (A–D) and FTIR spectra (E) from reference lignin samples of organosolv and dioxane extracted lignin (purple), organosolv lignin (red), Jack Pine

lignin (blue), and hot water flow-through pretreated corn stover residue pretreated at 2208C (green) droplets. The lignin droplets were deposited on KBr window for FTIR

microscopic imaging and spectral data collection. The spectra all feature intense ring C–C stretching bands between 1,500 and 1,600 cm�1, intense Ar–O stretch near 1,200–

1,240 cm�1 for alkyl aromatic ethers, broad C–O stretch bands for the ring alcohols in the carbohydrate moieties near 1,050 cm�1, and C–H out-of-plane deformation band for 1,3,5-tri

substituted benzene rings at 835 cm�1.

corn stover internode samples with this antibody accordingto the procedures described in Methods Section, and locateddistinct droplets on carbon coated samples by SEMbackscatter electron detection (BSE). With this detector,heavy atoms appear light grey against the darker backgroundof the lighter atoms (H, C, O) that make up the carbohydratebiomass (Fig. 5). In this case, the heavy atoms were silveratoms precipitated onto 15 nm gold nanoparticles thatthemselves were conjugated to the antibodies labeling thelignin droplets. To confirm that the bright BSE signal wasdue to the silver enhancement, we analyzed the samples byelectron dispersive spectroscopy (EDS) on regions ofinterest (indicated in red on SEM micrograph). We foundsilver (Ag) EDS peaks on the droplets indicating thepresence of silver enhanced lignin antibody that was absentfrom the EDS signal from the surrounding biomass(compare Fig. 5A and B).

Another technique often used to localize lignin within thecell wall is cytochemical staining. A range of cytochemical


stains has been reported to be useful to detect and localizelignin by electron microscopy, including mercury, bromine,and potassium permanganate (Fromm et al., 2003; Hepleret al., 1970; Maurer and Fengel, 1990). Of these, the methodthat has been shown to provide the best ultra-structuralpreservation, while revealing the location of lignin, ispotassium permanganate (KMnO4) (Stein et al., 1992). Totest the utility of staining lignin droplets in pretreatedbiomass with KMnO4, we compared the density of stainingand the EDS spectrum of KMnO4 stained Organosolvderived lignin, Organosolv solids, Organosolv and acidchlorite treated solids, and dilute acid pretreated corn stover(Fig. 6). The KMnO4 treated lignin showed a very densestaining by TEM and also a strong Mn peak by EDS(Fig. 6C). A much lower, but still detectable Mn signal wascollected from Organosolv solids (Fig. 6B). This correlateswell to the small amount of lignin remaining in Organosolvsolids samples. The Organosolv solids that were thenbleached with acid chlorite showed no staining by KMnO4

Figure 4. SEM micrographs (inset) and NMR analysis spectra of droplets on

filter paper. In spectra (a) thru (e), the signal from the cellulose filter paper substrate

dominates the intensity range, but a distinct phenolic profile is seen from the rind with

filter paper, ethanol extractives with filter paper, and most importantly the flow through

reactor solids samples indicative of the presence of lignin in the droplets. Scale

bars¼ 5 mm.

igure 5. (A and B) SEM micrographs imaged with the backscatter electron detector (BSE) and electron dispersive spectroscopy (EDS). The samples were labeled with anti-

, S lignol rabbit polyclonal antibody and detected with a 15 nm gold conjugated secondary a-rabbit antibody. The 15 nm gold was then enhanced with a silver enhancement

eposition protocol to provide a larger (40–100 nm) heavy metal target for BSE and EDS detection. The regions of interest scanned by EDS are indicated by a red box (A) and red

ross (B). The droplets (arrowheads) are bright by BSE detection and positive for Ag by EDS indicating positive lignin antibody detection.

FG

d

c

and no detectable Mn signal by EDS (Fig. 6A). Finally, thestained corn stover sample showed very dark stainingdroplets and a strong Mn signal by EDS taken from thedroplet region (Fig. 7D). Combined with the history ofsuccessful use of KMnO4, these results demonstrate theutility of KMnO4 to detect and localize lignin withinpretreated biomass samples.

Taken together, the results from FTIR, NMR, antibodylabeling, and cytochemical staining strongly indicate thatthe droplets we have observed contain lignin and that wecan confidently continue to investigate the formation oflignin droplets and the re-localization of lignin withinbiomass, in situ, using antibody and cytochemical labeling aswell as morphological analysis.

Lignin Droplets Originate as Coalesced Lignin Withinthe Cell Wall and Migrate Out of the Wall to Redepositon the Surface

We found that re-deposited lignin droplets can be evenlydispersed across nearly any cell wall surface (Fig. 1B and C),but are most often found clustered around and withinspecific ultra-structural features. Three cell wall ultrastruc-tural regions that accumulated a high concentration ofdroplets were pits, cell corners, and delamination layers(Fig. 7). In living plants, pits and cell corners are known tobe important for transport among adjacent cells and tissues.We have previously suggested that these same structuresplay an important role in the transport of pretreatmentsolutions within biomass. By both SEM and TEM analysis,the pits show a tendency to accumulate droplets relative tothe adjacent cell wall region (Fig. 7A and B). Figure 7A showsa pit partly occluded by a cluster of droplets, and 7B shows



Figure 6. TEM and EDS analysis of reference lignin samples and droplets on corn stover stained with. (A) organosolv and acid chlorite treated solids with little or no lignin

content shows no KMnO4 staining and no detectable Mn EDS peak, (B) organosolv solids with minimal lignin content show light staining by KMnO4 and minimal Mn signin by EDS,

(C) organosolv lignin shows dense KMnO4 staining and a strong Mn peak, and (D) pretreated corn stover cell corner with dense staining droplets (arrow) and a moderate Mn peak

by EDS.

an example of single droplets nearly large enough to blockthe pit channel. In addition to pits, cell corners showed adramatic accumulation of droplets. It was common to seethe entire cell wall surface within the cell corner coated bydroplets (Fig. 7C) following pretreatment. In Figure 7D thevolume of the cell corner is nearly filled with dropletsmaking it difficult to even delineate the margin of the cellwall. The last notable region of droplet accumulation is theseparation zone created by delamination of the cell wallduring pretreatment. Again, both SEM and TEM haverevealed spherical and flattened droplets within thedelamination zone (Fig. 7E and F). The flattened dropletsseen in Figure 7E are exposed after the completedelamination of the S3 secondary cell wall layer. Lignindroplets in delamination zones appear to fill the spaceavailable to them and are usually restricted from formingspheres. It is noteworthy that droplets accumulate in thesame regions (pits, corners, delamination zones) that areimportant for transport of pretreatment solutions and


enzymes through the biomass structure and could thereforecreate a barrier inhibiting access through these regions.

Because most of the original lignin content of the cell walldoes not migrate out of the wall and redeposit on the cellwall surface, and because redeposited surface droplets musthave spent some time traversing the thickness of the cell wallbefore emerging at the surface, we analyzed the laminatestructure within the cell wall for other coalesced ligninforms. By staining ultra-thin sections with KMnO4, we wereable to emphasize the location of lignin within the cell wall.In addition to the droplets forming on surfaces such as thecell corners, a layered pattern of droplet-like densities wasrevealed throughout the cell wall (Fig. 8B). This pattern isstriking when compared to the more even staining patternseen in non-pretreated samples (Fig. 8A). These data revealthat the issue of droplets occluding pore structure andpotentially blocking enzyme accessibility is not restricted tothe outermost surfaces of the cell wall. On the other hand,the more important phenomenon is likely that a relatively

Figure 7. SEM (A, C, E) and TEM (B, D, F) micrographs of corn stover rind cell walls. Samples were pretreated in 0.8% H2SO4 for 20 min. Lignin droplets were particularly

abundant in three cell wall ultrastructure regions: pits (A and B), cell corners (C and D), and delamination zones (E and F). Pits can become nearly occluded by multiple small (A) or

single large (B) lignin droplets. The walls of the cell corners are often seen densely coated with droplets (C) and can become nearly filled with spherical and flattened droplets (D)

obscuring the margin of the secondary cell wall. Panels E and F illustrate the formation of flattened droplets (arrowheads) within delaminated regions of secondary cell walls. SCW-

secondary cell wall; P-pit. Scale bars C¼ 1 mm; A, B, D, E¼ 0.5 mm; F¼ 0.2 mm.

homogenous distribution of lignin coalesces and migrates toa more localized, concentrated distribution of lignin. Thisprocess will likely increase the accessibility of individualcellulose microfibrils deep within the cell wall.

To more thoroughly investigate how lignins coalescewithin the cell wall, we used the technique of electron

tomography. We have collected and analyzed eighttomograms of coalesced lignins, representing over 6,000individual images of data. This technique allowed us toanalyze the 3D structure of the cell wall at �6 nm resolutionin conjunction with using KMnO4 to localize lignin(Donohoe et al., 2006; McIntosh et al., 2005). In addition to



Figure 8. TEM micrographs of unpretreated (A) and dilute acid pretreated (B) corn stover samples. The tissue has been stained with KMnO4 to emphasize the location of

lignin. In the native sample (A), lignin appears relatively evenly dispersed with typical lamellar concentrations. Following pretreatment at 1508C for 20 min, lignin can be seen as the

small dark droplet-like pattern within the primary and secondary cell walls (B). Scale bars¼ 1 mm.

providing a view into the 3D ultrastructure of the cell wall,the data in an electron tomogram provides �3 nm opticalslices within the biomass specimen that can be re-sampled inany dimension (Mastronarde, 1997).

Figure 9A shows a single 3 nm slice of a tomogramrevealing surface droplets, droplets within a pit, accumula-tion of lignin material in delamination zones and the middlelamella, and a detailed pattern of lignin coalescence and re-localization within the bulk of the cell wall. The improved3D resolution of the tomogram slice over conventionalTEM micrographs and KMnO4 emphasis of the lignincreates the contrast and clarity in this image. Thetomograms confirm that the droplets found on the surfaceof the cell wall, when viewed in cross section display a nearlyspherical shape with a flattened surface in contact with thecell wall surface (Fig. 9B). The �3 nm tomogram slices alsoreveal that the droplets are nearly solid. Also, disk-shapedaccumulations of lignins were confirmed within delamina-tion zones (Fig. 9C). The convex shape of the disk wallsmight suggest that the lignins moved into the delaminationzone either during or after the delamination event and didnot create the delamination. Alternatively, the shape maysimply be a consequence of cooling. Confirming this analysiswill require further investigation. One of the more strikinglignin morphologies is the coalesced lignin that appears to beextruding from the cell wall into the lumen of the pit(Fig. 9D). Directly adjacent to these extruding lignin bodiesare large droplets accumulated within the pit. This directevidence of lignin extrusion may indicate a mechanism forlignin removal from the cell wall by temperature and theformation of lignin droplets. It is important to note;however, that the majority of the lignin remains within thecell wall coalesced and re-localized. It also appears that theregion of the cell wall directly adjacent to the coalescedlignin is opened up and would likely provide increased


accessibility to pretreatment chemistries and cellulolyticenzymes.

Discussion

Thermochemical Pretreatment Above Lignin MeltingTemperature Causes Lignin to Coalescence, Migrate,and Redeposit on Biomass Cell Walls

Figure 10 illustrates a possible mechanism for lignin dropletcoalescence, extrusion, and redeposition. It seems clear thata thermochemical pretreatment that exceeds the meltingtemperature of lignins allows them to expand and becomemobile within the cell wall matrix. In an aqueousenvironment, hydrophophobic lignins minimize their sur-face area contact with water causing them to coalesce andform spheres. Because most lignins are still trapped withinthe natural pore structure of the cell wall, they can onlycoalesce to form small elliptic cylinders. However, if thelignins migrate to a larger void, such as a delamination layer,cell corner, or pit, they continue to be shaped by theiraqueous environment and form flattened disks or sphericaldroplets. It also seems clear that lignin droplets that leave thecell wall matrix can move around the biomass sample, intothe hydrolysate solution and redeposit on the surface duringcooling. What is less clear is the mechanism for ligninextrusion from the cell wall. Whereas thermal expansion,diffusion and aqueous interactions surely play a principalrole, our observations have led us to hypothesize that ligninis eventually being wrung out of the cell wall as adjacentcellulose microfibrils become unsheathed and exposedcrystalline surfaces adhere to each other by hydrogenbonding. We envision this localized cell wall collapsephenomenon as a driving force for lignin migration andextrusion.

Figure 9. TEM tomogram images of pretreated cell walls. (A) A �3 nm tomogram slice showing a pit containing section of pretreated corn stover cell wall between two

adjacent cells. Surface droplets are seen on the far right (arrowheads), within the pit (arrow), and within the delamination zones of both secondary cell walls. Another region of

dramatic lignin accumulation in this cell wall is the middle lamella. (B) Surface droplets are solid and spherical with a flattened contact surface on the cell wall. (C) Disk-shaped

droplets in delamination zones display convex shaped sides. (D) An oblique section through the pit reveals a droplet extruding from the cell wall. P-pit; DZ-delamination zone;

SCW-secondary cell wall; ML-middle lamella. Scale bars A¼ 0.5 mm, B–D¼ 0.2 mm.

Lignin droplets can be found redeposited on nearly anysurface in pretreated biomass however; some sub-cellularstructures appear to accumulate especially high densities ofdroplets. These regions included pits, cell corners, delami-nation zones, and the middle lamella. We were particularlyinterested in the surfaces that accumulate a high density ofdroplets because there is the potential to coat surfaces,occlude pore structure and block enzymes from reaching thecell wall. Whereas droplets found in and around the cell pitscould have come from anywhere in the biomass andredeposited around the pits, the same is not true for thedroplets found in cell corners and in delamination zones.

These ultra-structural regions are deep within the biomassand the droplets accumulating in them can only have beenre-localized from adjacent cell walls.

This study has established that the droplets that extrudefrom the cell walls and redeposit on the surface arecomposed, perhaps predominantly, of lignin. Our FTIRresults indicate that in the case of droplets formed by hotwater pretreatment at 2208C the lignin droplets appear tonot contain detectable residual carbohydrates. However,further investigation will be required to determine underwhat conditions the lignin droplets remain complexed withcarbohydrates as LCCs. We suspect that at least in the case of



Figure 10. Model of a proposed mechanism for the driving force of lignin

migration based on the melting of lignin, coalescence caused by the aqueous

environment, and eventually the progressive collapse of microfibrils driving molten

lignin to the surface of the biomass cell wall.

coated lignin droplets, the coat represents a carbohydrateshell surrounding a lignin core to form a lignin-carbohy-drate complex. A semi-pure lignin droplet and a carbohy-drate-coated droplet would have very different implicationsfor their interaction with cellulolytic enzymes and thus hassignificant impact on biomass conversion technologies.

Relocalization of Lignin is Likely to be as Important asLignin Removal to Improve Digestibility

Our characterization of the lignin droplets emphasizes theneed to use multiple modes of lignin identification. By usingboth mophological criteria and rigorous analytical tools, wewere able to demonstrate the utility of KMnO4 to detect andlocalize lignin within the cell walls of pretreated biomasssamples. By combining KMnO4 labeling with electrontomography, we were able to visualize a striking pattern oflignin re-localization within the pretreated cell wall. Thesedata revealed the probability that droplets occluding porestructure and blocking enzyme accessibility is not restrictedto the outermost surfaces of biomass, but likely has animpact throughout the cell wall. More importantly, whatwas revealed was a pattern of lignin re-localization thatappears to dramatically open up the structure of the cell wallmatrix and improve the accessibility of the majority ofcellulose microfibrils. Indeed, this re-localization phenom-enon likely explains a critical mechanism for the enhanceddigestibility of dilute acid and hot water pretreated biomass.The important concept is that lignins can be moved awayfrom much of the cellulose microfibril surfaces withoutbeing removed from the biomass altogether. Our workand the recent work of others (Zeng et al., 2007) isdemonstrating the utility of structure analysis of pretreatedbiomass. Our ultimate goal is to gain understanding of themolecular mechanisms of pretreatment in order to optimizecurrent or invent new pretreatment strategies.


We would like to thank Mark Davis for NMR analysis, David Johnson

for preparing pretreated materials, and Katia Ruel for the generous gift

of antibodies. This work was funded by the DOE Office of the Biomass

Program.

References

Black SK, Hames BR, Myers MD. Method of separating lignocellulosic

material into lignin, cellulose, and dissolved sugars. U.S. patent 1998.

5,730,837.

Boerjan W, Ralph J, Baucher M. 2003. Lignin biosynthesis. Annu Rev Plant

Biol 54:519–546.

Chum HL, Johnson DK, Black SK, Overend RP. 1990. Pretreatment Catalyst

Effects and the Combined Severity Parameter. Appl Biochem Biotech-

nol 24-5:1–14.

Donaldson LA. 2001. Lignification and lignin topochemistry—an ultra-

structural view. Phytochemistry 57(6):859–873.

Donohoe BS, Mogelsvang S, Staehelin LA. 2006. Electron tomography of

ER, Golgi and related membrane systems. Methods 39(2):154–162.

Fromm J, Rockel B, Lautner S, Windeisen E, Wanner G. 2003. Lignin

distribution in wood cell walls determined by TEM and backscattered

SEM techniques. J Struct Biol 143(1):77–84.

Gidh AV, Decker SR, Vinzant TB, Himmel ME, Williford C. 2006. Deter-

mination of lignin by size exclusion chromatography using multi angle

laser light scattering. J Chromatogr A 1114(1):102–110.

Grabber JH. 2005. How do lignin composition, structure, and cross-linking

affect degradability? A review of cell wall model studies. Crop Sci

45(3):820–831.

Hepler PK, Fosket DE, Newcomb EH. 1970. Lignification during secondary

wall formation in coleus. an Electron Microscopic Study. Am J Bot

57(1):85–96.

Himmel ME, Adney WS, Baker JO, Elander R, McMillan JD, Nieves RA,

Sheehan JJ, Thomas SR, Vinzant TB, Zhang M. 1997. Advanced

bioethanol production technologies: A perspective. Fuels Chem

Biomass p 2–45.

Himmel ME, Ruth M, Wyman CE. 1999. Cellulase for commodity products

from cellulosic biomass. Curr Opin Biotechnol 10:358–364.

Joseleau JP, Ruel K. 1997. Study of lignification by noninvasive techniques

in growing maize internodes - An investigation by Fourier transform

infrared cross-polarization magic angle spinning C-13-nuclear mag-

netic resonance spectroscopy and immunocytochemical transmission

electron microscopy. Plant Physiol 114(3):1123–1133.

Kong FR, Engler CR, Soltes EJ. 1992. Effects of cell-wall acetate, xylan

backbone, and lignin on enzymatic-hydrolysis of aspen wood. Appl

Biochem Biotechnol 34-5:23–35.

Kremer JR, Mastronarde DN, McIntosh JR. 1996. Computer visualization

of three-dimensional image data using IMOD. J Struct Biol 116(1):

71–76.

Mastronarde DN. 1997. Dual-axis tomography: an approach with align-

ment methods that preserve resolution. J Struct Biol 120(3):343–

352.

Maurer A, Fengel D. 1990. A new process for improving the quality and

lignin staining of ultrathin sections from wood tissue. Holzforschung

44(6):453–460.

McIntosh R, Nicastro D, Mastronarde D. 2005. New views of cells in 3D:

an introduction to electron tomography. Trends Cell Biol 15(1):

43–51.

Micic M, Benitez I, Ruano M, Mavers M, Jeremic M, Radotic K, Moy V,

Leblanc RM. 2001. Probing the lignin nanomechanical properties and

lignin-lignin interactions using the atomic force microscopy. Chem

Phys Lett 347(1–3):41–45.

Morrison TA, Jung HG, Buxton DR, Hatfield RD. 1998. Cell-wall composi-

tion of maize internodes of varying maturity. Crop Sci 38(2):455–

460.

Nidetzky B, Steiner W, Hayn M, Claeyssens M. 1994. Cellulose hydrolysis by

the cellulases from Trichoderma reesei: a new model for synergistic

interaction. Biochem J 298(Pt 3):705–710.

President Bush Delivers State of the Union Address. United States Capitol,

Washington, D.C., Office of the Press Secretary, January 23, 2007

http://www.whitehouse.gov/news/releases/2007/01/20070123-2.html.

Radotic K, Micic M, Jeremic M. 2005. New insights into the structural

organization of the plant polymer lignin. Biophysics from Molecules to

Brain: in Memory of Radoslav K. Andjus. New York: New York Acad

Sciences. p 215–229.

Ruel K, Burlat V, Joseleau JP. 1999. Relationship between ultrastructural

topochemistry of lignin and wood properties. Iawa J 20(2):203–211.

Ruel K, Chabannes M, Boudet AM, Legrand M, Joseleau JP. 2001. Reassess-

ment of qualitative changes in lignification of transgenic tobacco plants

and their impact on cell wall assembly. Phytochemistry 57(6):875–882.

Sarkanen S. 1998. Template polymerization in lignin biosynthesis. In: Lewis

NG, Sarkanen S, editors. Lignin and Lignan Biosynthesis, Vol. 697,

Washington, DC: ACS Symposium Series. p 194–208.

Sarkanen KV, Hergert HL. 1971. Classification and distribution. In:

Sarkanen KV, Ludwig CH, editors. Lignins. Occurrence, Formation,

Structure and Reactions. NY, USA: Wiley-Interscience, New York.

p 43–94.

Schell DJ, Farmer J, Newman M, McMillan JD. 2003. Dilute-sulfuric acid

pretreatment of corn stover in pilot-scale reactor - investigation of

yields, kinetics, and enzymatic digestibilities of solids. Appl Biochem

Biotechnol 105:69–85.

Selig MJ, Viamajala S, Decker SR, Tucker MP, Himmel ME, Vinzant TB.

2007. Deposition of lignin droplets produced during dilute acid pre-

treatment of maize stems retards enzymatic hydrolysis of cellulose.

Biotechnol Progr 23(6):1333–1339.

Stein BD, Klomparens KL, Hammerschmidt R. 1992. Comparison of

bromine and permanganate as ultrastructural stains for lignin in plants

infected by the fungus Colletotrichum-Lagenarium. Microsc Res Tech-

nique 23(3):201–206.

Sun XF, Sun RC, Fowler P, Baird MS. 2004. Physicochemical characteriza-

tion of liginin isolated with high yield purity from wheat straw. Int J

Polym Anal Characterization 9(5–6):317–337.

Tucker MP, Nguyen QA, Eddy FP, Kadam KL, Gedvilas LM, Webb JD.

2001. Fourier transform infrared quantitative analysis of sugars and

lignin in pretreated softwood solid residues. Appl Biochem Biotechnol

91-3:51–61.

Vinzant TB, Ehrman CI, Adney WS, Thomas SR, Himmel ME. 1997.

Simultaneous saccharification and fermentation of pretreated hard-

woods-effect of native lignin content. Appl Biochem Biotechnol 62(1):

99–104.

Zeng M, Mosier NS, Huang C-P, Sherman DM, Ladisch MR. 2007.

Microscopic examination of changes of plant cell structure in corn

stover due to hot water pretreatment and enzymatic hydrolysis.

Biotechnol Bioeng 97:265–278.

