Effects of acetic acid, furfural and p-hydroxybenzoic acid on succinic acid fermentation by Escherichia coli BSS133

(1)

MASTER'S THESIS

Effects of Acetic Acid, Furfural and P- hydroxybenzoic Acid on Succinic Acid Fermentation by Escherichia coli BSS133

Mireille Ginésy 2014

Master of Science in Engineering Technology Chemical Engineering

Luleå University of Technology

Department of Civil, Environmental and Natural Resources Engineering

(2)

Effects of acetic acid, furfural and p-hydroxybenzoic acid on succinic acid fermentation by Escherichia coli BSS133

Mireille Gin´esy Thursday 20^thMarch, 2014

1

(3)

2

(4)

Abstract 3

Abstract

During hemicellulose hydrolysis, a mandatory step prior to fermentation, degradation products are formed from sugars and lignin In the case of hardwood hemicellulose, acetate is also released from xylan deacetylation. Those compounds might be inhibitory for both Escherichia coli growth and the succinic acid production. In the present work, E. coli growth was first tested on both glucose and xylose in the presence of different inhibitors: sodium acetate [0-12 g/L]; furfural [0-3 g/L]

and 4-hydroxybenzoic acid [0-2 g/L]. This investigation was performed in shake flasks, using small size inocula. Afterwards, eleven fermentations in batch mode were carried out in 1-L fermentors, following a full factorial design with three centred points. The ranges of inhibitors studied were:

acetic acid [0-10 g/L], furfural [0-2 g/L] and p-hydroxybenzoic acid [0-2 g/L]. Finally, hemicellulose was extracted from birch wood, hydrolysed with 2.7% sulfuric acid and used as a substrate for a fermentation.

The results showed that furfural was the most toxic compound. Besides, its association with acetic acid was synergistic since no growth occurred when furfural and acetic acid were present in the media at 2 and 10 g/L, respectively. Nevertheless, E. coli was able to hydrolyse these two compounds to some extent. Low POH concentrations did not have any significant effect on the growth and fermentation performance of E. coli. Acetic acid at high concentrations affected both the growth and the succinic acid production. Indeed, lower succinic acid concentrations were obtained when there was an initial concentration of 10 g/L of acetic acid in the media. Acetic acid, furfural and phenolswere found in the hydrolysates at concentration of 13.3 g/L, 2.2 g/L and 1.5 g/L respectively. As expected from the experimental model E. coli was thus not able to grow on undetoxified birtchwood hemicellulose hydrolysates when using a batch process.

(5)

4 Preface

Preface

This thesis work was conducted in spring 2011 at Lule˚a University of Technology (LTU).

It was financially supported by grants from the Swedish Energy Agency and Swedish Governmental Agency for Innovation System (VINNOVA).

First of all I would like to thanks Professor Kris Berglund for allowing me to perform my master thesis project within the biochemical technology group. I would also like to express my gratitude to Professor Ulrika Rova. Thank you for providing me a subject that I came to like very much, despite initial doubts. Finally I would like to thanks my supervisor Jonas Helmerius. I learned a lot under your supervision.

(6)

List of Figures

Figure 1. Basic principle of a biorefinery . . . 10

Figure 2. Structure of succinic acid . . . 12

Figure 3. Potential uses of biobased succinic acid . . . 13

Figure 4. Chemical synthesis of succinic acid . . . 14

Figure 5. Scanning electron micrograph of Escherichia coli . . . 17

Figure 6. Hexose and pentose metabolism in E. coli . . . 18

Figure 7. Tricarboxylic acid cycle . . . 19

Figure 8. Mixed acid fermentation pathway in E. coli . . . 19

Figure 9. Standing volume of different hardwood species in Sweden . . . 21

Figure 10. Plant cell wall . . . 22

Figure 11. Structure of glucuronoxylan from hardwood . . . 23

Figure 12. Effects of xylanotic enzymes on xylan . . . 24

Figure 13. Generation of inhibitors during hardwood hydrolysis . . . 25

Figure 14. Summary of the inhibition mechanisms of the toxic compounds in hemicellulosic hydrolysates . . . 28

Figure 15. Inflow of weak acid in the cytosol . . . 29

Figure 16. Schematic representation of the uncoupling theory . . . 29

Figure 17. Schematic representation of synthetic uncouplers action . . . 29

Figure 18. Modes of toxicity of furfural in E. coli . . . 32

Figure 19. Overcoming the inhibition problems . . . 36

Figure 20. Methods for the development of inhibitor-resistant microorganism . . . 41

Figure 21. Metabolic pathways in E. coli BSS133 . . . 43

Figure 22. Experimental matrix constructed with Modde 9.0 . . . 46

Figure 23. A fermentation run . . . 47

Figure 24. Evolution of the cell density for sodium acetate, furfural and POH for different inhibitors concentrations . . . 51

Figure 25. Evolution of the cell viability for sodium acetate and furfural for different inhibitors concentrations: . . . 53

Figure 26. A typical fermentation profile, initial concentrations: AA: 5 g/L; Fur: 1 g/L; POH: 1 g/L . . . 54

Figure 27. Cell density during the growth and production phases . . . 57

Figure 28. Evolution of acetic acid throughout the fermentations . . . 58 7

(9)

8 LIST OF FIGURES

Figure 29. Plot of normalized residuals for the three responses . . . 60

Figure 30. Plots of observed against predicted for the three responses . . . 60

Figure 31. Plots of distance to model for all the responses . . . 61

Figure 32. Effects plot . . . 61

Figure 33. Interaction plots for furfural and acetic acid . . . 62

Figure 34. Interaction plots for acetic acid and POH . . . 62

Figure 35. Interaction plots for furfural and POH . . . 63

Figure 36. Variable importance plot . . . 63

Figure 37. Response 4D contour plot . . . 63

Figure 40. Effect plot (production phase) . . . 65

Figure 41. Interaction plots for acetic acid and POH (production phase) . . . 65

Figure 42. VIP plot (production phase) . . . 66

(10)

List of Tables

Table 1. Comparison of the fermentation performances of different succinic acid

producers . . . 16

Table 2. Comparison of different hydrolysis methods . . . 24

Table 3. Phenols in hardwood . . . 26

Table 4. Composition of different hardwood hemicellulose hydrolysates . . . 27

Table 5. Inhibitor effect on E. coli LY01 . . . 33

Table 6. Effects of acetic acid, furfural, POH and the combination acetic acid and furfural on different parameters during ethanol fermentation by bakers’ yeast (as reported in [74]). . . 35

Table 7. CSL based medium recipe . . . 45

Table 8. PBS recipe (pH = 7) . . . 48

Table 9. Comparison of inhibition (using OD550) on glucose and on xylose with different inhibitors . . . 52

Table 10. Full factorial design: factors (acetic acid, furfural, and p-hydroxybenzoic acid) and responses (specific growth rate, lag phase, biomass yield, volumetric productivity, production yield and final succinic acid titre) . . . 56

Table 11. Differences between the two runs of the experiment number 6 (10 g/L acetic acid, 2 g/L POH). . . 56

Table 12. PLS total summary after exclusion . . . 59

Table 13. Summary of fit for individual response . . . 59

Table 14. Sugar and inhibitors concentrations during the hydrolysates fermentation . 66 Table 15. Inhibition of E. coli growth by acetic acid, furfural and POH. . . 69

9

(11)

10 Introduction

1 Introduction

1.1 Background

In the last decade, the rising environmental awareness combined with the decline of fossil fuel reserves have been a driving force behind the search for green alternatives to petroleum based industry. A potential solution is to use woody biomass, a readily available and renewable resource. To utilize it efficiently and sustainably, the biorefinery concept has to be developed. Biorefineries function according to the petroleum refineries model, i.e. highly integrated plants producing a wide range of valuable products, such as value added chemicals, fuels, polymers and other materials (for a review see [1]). Ideally, every components of the biomass feedstock should be used in the most beneficial way leading to minimal generation of wastes and relatively low production costs. Biorefineries could therefore economically compete with petroleum-based industry. Figure 1 illustrates the concept of biorefinery. The Sweden’s pulp

Figure 1.Basic principle of a biorefinery

and paper industry already has the necessary infrastructure to handle biomass [2], diversifying its production could be one way to face the rising competition of low cost producers using fast growing trees [3]. Another example of potential biorefinery are energy plants based on biomass combustion [4].

Hardwood trees (e.g. birch) are a common resource in Sweden. Hardwood consists mainly of three components: cellulose (39-53%), hemicellulose (19-36%) and lignin (17-24%) [5]. The hemicellulose fraction of wood is often underutilized in the forest based industry. However, due to its branched structure and its lower degree of polymerization, hemicelluloses are more soluble than cellulose and could therefore be pre-extracted from wood chips prior to the process, as reviewed by Walton S. [2].

Once extracted, the hemicellulose fraction could be used for the production of a great diver-

(12)

1.2 Aim 11

sity of fermentation products. Although ethanol is the most common product of hemicellulose hydrolysates fermentations, butanol can also be produced, as well as various chemicals such as xylitol, 2,3-butanediol and lactic, citric and butyric acids [6]. Succinic acid is also a high added value chemical that can be obtained from lignocellulosic hydrolysates fermentation.

However microorganisms cannot directly ferment the carbon fraction of the hydrolysate, the oligosaccharides of hemicellulose have to be broken down into simple sugars. This can be achieved either by dilute acid hydrolysis, concentrated acid hydrolysis or enzymatic hydrolysis [6]. The main problem of hydrolysis is that it releases a wide range of inhibitory compounds, which can be classified into three main groups: weak acids, furan derivatives and phenolic compounds [7]. Those inhibitors may affect the growth of the microorganisms, the production yield and the productivity, and thus need to be removed before the fermentation.

Many different detoxification methods exist [8]. However, most of them lead to additional process costs and, possibly, the loss of fermentable sugars. That is why it is important to determine to which extent the microorganisms can tolerate inhibitors, as this knowledge would permit to minimize the detoxification steps prior to the fermentation processes.

1.2 Aim

During this work the main and interaction effects of three inhibitors: acetic acid, furfural and p-hydroxybenzoic acid (POH) on the growth of E. coli BSS133 and its succinic acid production were investigated. The aim was to determine the detoxification requirements of this bacterium in order to minimize the pretreatment costs prior to succinic acid production by fermentation of hardwood hemicellulose hydrolysates.

In order to do that, this work was divided into three parts. First a growth investigation was performed in shake flasks to determine the limit of inhibitors that E. coli can tolerate. Low inocula (low cell density) were used in order to minimize the possible bioconversion of the toxic compounds. Afterwards, a certain number of fermentations (using high cell inocula) were run to study both the growth and the fermentation performance of E. coli. Finally, the ability of E. coli to growth on birch wood hemicellulose hydrolysates without prior detoxification was tested.

(13)

12 Succinic acid

2 Succinic acid

Succinic acid, also called butanedioic acid or amber acid, is a dicarboxylic acid having the molecular formula C4H6O4 (see Figure 2). Its name comes from Latin succinum which means amber, because succinic acid was first isolated by Georgius Agricola in 1546 via distillation of amber.

Figure 2.Structure of succinic acid

Nowadays succinic acid is mainly produced by petrochemical processes. However in the last decades much research has been focused on the development of fermentation processes as a green alternative to petrol-based succinic acid.

2.1 Succinic acid and industry 2.1.1 Markets and applications

The estimated 2010 worldwide use of succinic acid was around 20,000–30,000 tonnes per year with an annual growth rate of around 10% per year [9]. Nevertheless, this market could be even wider if the production costs were lower since succinic acid would then be an economically competitive intermediate for the production of many various commodity chemicals (Figure 3).

Indeed, succinic acid, together with other four carbon diacids (i.e. fumaric and malic acids), has been identified by U.S. Department of Energy as one of the twelve most promising building block chemicals that could serve as economics drivers for a biorefinery [10]. A building block chemical is a molecule with several functional groups that can be transformed into new families of useful molecules. Furthermore, the potential commodity chemical market for succinic acid was estimated at 270,000 tonnes per year [11].

According to an excellent review by Zeikus et Al. [12], the four current main markets for succinic acid in specialty chemicals are as:

• a surfactant/detergent extender/foaming agent;

• an ion chelator;

• an acidifiant/pH modifier, flavouring agent or antimicrobial agent in food industry;

• an intermediate for the production of health-related agent (e.g. pharmaceuticals, antibi- otics, amino acids, vitamins).

Succinic acid also have other growing applications such as feed additives (in the form of succinic salts), green solvents and plant growth stimulant [12].

(14)

2.1 Succinic acid and industry 13

Figure 3.Potential uses of biobased succinic acid [12]

The different possible applications for each succinic acid derivative are too numerous to be all cited but have been listed, with their targeted market price and potential 2020 market size for biobased products, in a report from the US Department of Energy [13]. Among the most important succinic-acid-based commodity chemicals are: 1,4 Butanediol (BDO), tetrahydrofuran (THF), γ-butyrolactone (GBL), the pyrrolidone family (methyl pyrrolidone and 2-pyrrolidone), succinate salts and adipic acid [13, 14]. Many of succinic acid derivatives could be used as green solvent, but another important application is the production of polymers and fibers, such as lycra [10], nylon [11], polyamides [11, 13] or polyesters [15, 16]. One interesting use of this latter is for instance the production of the biodegradable plastic Bionolle [17].

Succinic acid is therefore a chemical of great interest that should become more and more important in the coming years.

2.1.2 Chemical synthesis

Nowadays, succinic acid is mainly produced by hydrogenation of maleic anhydride to succinic anhydride and subsequent hydrations [18] or directly by hydrogenation of maleic acid (Fig- ure 4). This later can be reduced to succinic acid either chemically or electrochemically [19].

The electrolytic reduction of maleic acid in acidic media on nickel, lead, Ti/ceramic TiO2, platinised platinium, graphite, mercury and zinc cathodes and on cooper electrode have been reported [19, 20]. Chemically, the hydrogenation of maleic acid can be achieved in aqueous cobalt chloride and potassium chloride solutions in the presence of zinc/mercury or in aqueous

(15)

14 Succinic acid

sodium hydroxyde/ethanol in the presence of phosphoric acid [20]. However, both maleic acid and maleic anhydride are obtained from oxidation of either butane or benzene. Therefore, chemical synthesis of succinic acid is environmentally unfriendly and highly dependant on the petroleum cost and availability.

Figure 4.Chemical synthesis of succinic acid

2.1.3 Fermentation

An emerging technology driven by the current need to find alternatives to petrochemical processing is the succinic acid production by fermentation. Not only is this process less polluting than the chemical one, but it could be cost efficient and thus allow for the succinic acid market to expend and replace other petrochemicals, such as the competing maleic anhydride. Besides, unlike the ethanol fermentation during which the greenhouse gas CO2 is formed, the succinic acid production is a CO2-fixing fermentation. [12]

Since fermentative organisms are usually not tolerant to acidic conditions, succinic acid fermentations operate optimally at pH value near the neutral point, consequently producing succinate salt rather than the free acid [12, 13]. Therefore, after fermentation succinate salts have to be separate from the solid cell mass and the undesirable by-products and converted into free succinic acid. Conventional methods for downstream processing, i.e. product recovery, concentration and purification, are electrodialysis, filtration and acidification or extraction [21].

Succinic acid fermentation is already an industrial process. Indeed, Bioamber [22] has been producing bio succinic acid in Pomacle, France since January 2010 and also has a scaled up facility in Sarnia, Ontario, Canada expected to be operational in 2012 for succinic acid and BDO production. Besides, another plant, located in Thailand, should be operational in 2014 and exclusively supply PTT-MCC Biochem for its polybutetylene succinate (PBS) production.

Myriant [23] has also developed a technology for succinic acid fermentation and should start the production in 2013 in Lake Providence, Louisiana, US. This facility will be the world’s

(16)

2.2 Succinic acid fermentation 15

largest biobased succinic acid plant and will be using sorghum grain, sorghum grits and lignocellulosic sugar hydrolysates as feedstocks. Finally, DSM [24] and Roquette, which operate a demonstration plant in Lestrem, France, started the construction of a bio-based succinic acid plant expected to open in fall 2012 in Cassano Spinola, Italy. First Strach derivatives and later on cellulosic biomass will be used as feedstock.

Nowadays, biosuccinic acid is manly used in the food market. However, the applications are diversifying as both Bioamber and Myriant have partnerships with various companies producing THF, GBL, PBS, plasticizers, polymers or solvents.

2.2 Succinic acid fermentation 2.2.1 Different succinic acid producers

Succinic acid is formed by almost all plants, animals and microorganisms since it is an intermediate metabolite in the tricarboxylic acid (TCA) cycle and one of the fermentation end-products of anaerobic metabolism [21]. Succinic acid is produced by some obligate or facultative anaerobes, all forming mixed-acid fermentations yielding different proportions of ethanol, succinic, acetic, formic and lactic acid [12].

Some bacteria naturally produce succinate at high level. In particular, the Gram-negative Anaerobiospirillum succiniproducens, Actinobacillus succinogenes, Mannheimia succiniproducens have been thoroughly investigated. The last two are facultative anaerobes that have been isolated from rumen, like other diverse succinate-producing microorganisms, such as Ruminococcus flavefaciens and Fibrobacter succinogenes [25]. As early as 1951, Sijpesteijn and Elsden [26] demonstrated that in this organ, succinic acid was rapidly converted to propionic acid, which is absorbed through the rumen wall for subsequent oxidation to provide energy and biosynthetic precursors for the animal [12]. Therefore, the rumen contains a variety of mircoorganisms capable of converting starch and lignocellulosic materials into succinic acid. Succinate-producers are also found in the digestive system or other animals [12]; for example, A. succiniproducens, an obligate anaerobe, has been isolated from the mouth of a beagle dog.

Mutants of A. succinogenes sp. 130Z, a succinic acid-producer, were developed by Guettler et Al [27]. The variants produce increased amounts of succinic acid. For instance FZ 53 produced 105.8 g/L of succinic acid in 78h with a productivity of 1.34 g/L/h and a yield of 0.81 g.g⁻¹ (compared to 67.2 g/L, 0.79 g/L/h and 0.68 g.g⁻¹for the parent 130Z in the same conditions).

Besides, Van der Werf et al. [28] demonstrated that A. succinogenes sp. 130Z was able to ferment various carbohydrates, including pentoses, although it utilises those much less efficiently than hexoses or sugar alcohols. In particular, with D-arabitol only little formation of by-products (acetate, formate, ethanol) occurred and a succinate yield as high as 0.80 g.g⁻¹ was obtained.

In comparison, the yield from the corresponding sugar, arabinose, was only 0.33 g.g⁻¹.

A similar study by Lee et al. [29] using M. succiniproducens MBEL55E showed that this organism also ferment D-arabitol more efficiently than other carbohydrates with a succinic

(17)

16 Succinic acid

acid yield of 0.82 g.g⁻¹. However, this strain utilized xylose with the same efficacy as glucose , i.e. a yield of about 0.66 g.g⁻¹and a final succinate concentration of about 13–14 g/L.

Using A. succiniproducens in a batch fermentation Glassner et al. obtained 50.3 g/L of succinic acid in 24h, with a productivity of 2.1 g/L/h and a succinate yield of 0.9 g/g [30]. Meynial- Salles et al. reported the continous production of succinic acid at high yield, titer and productivity by A. succiniproducens [31]. The use of an integrated membrane-bioreactor electrodialysis process resulted in 83 g/L of succinic acid with a succinate yield of 0.88 g/g and a productivity of 10.4 g/L/h. Finally, Lee et al. investigated the use of wood hydrolysates (from oak) for the succinic acid production by A. succiniproducens. The organism could not grow on the hydrolysates with addition of a nitrogen source, however, when the media was supplemented with corn steep liquor (CSL) up to 23.8 g/L succinic acid was obtained with a productivity of 0.744 g/L/h and a yield of 0.88 g/g [32]

Table 1.Comparison of the fermentation performances of different succinic acid producers Microorganism Time Succinic acid Productivity Yield Ref.

(h) (g/L) (g/L/h) (g.g⁻¹) A. succiniproducens

ATCC 53488 27 32.2 1.2 0.99 [40]

ATCC 53488 24 50.3 2.1 0.90 [30]

ATCC 53488^a — 83 10.4 0.88 [31]

A. succinogenes

FZ 53 78 105.8 1.34 0.81 [27]

FZ 21 47 81.9 1.65 0.74 [27]

130Z 84 67.2 0.79 0.68 [27]

M. succiniproducens

MBEL55E 7.5 14 1.87 0.66 [29]

MBEL55E^c 11 13.5 1.21 0.72 [32]

MBEL55E^c 11 13.4 1.18 0.71 [32]

C. glutamicum

∆ldhA–pCRA717 46 146 3.2 0.92 [33]

E. coli

NZN111 44 12.8 0.29 0.64 [41]

AFP111 LpTrc99A-pyc^b 77 99.2 1.3 1.14 [42]

AFP184 32 40 1.27 0.83 [43]

AFP184^d 32 25 0.78 0.5 [43]

KJ122 96 80.9 0.84 0.96 [43]

Batch fermentations using glucose as carbon source unless stated otherwise.

acontinuous process,^b fed batch,^cwhey,^dxylose

(18)

Efforts have been made not only to improve those natural succinic acid-producers, but also to develop more efficient strains from bacteria usually producing succinate as a minor product.

Metabolic engineering of E. coli to enhance succinic acid production is described in section 2.2.4.

Another bacterium that has been under investigation is Corynebacterium glutamicum [33–36], a Gram-positive facultative anaerobe. Recently, Okino et al. [33] obtained 146 g/L in 46h with a productivity of 3.2 g/L/h using a C. glutamicum strain (∆ldhA-pCRA717). Interestingly, the volumetric productivity was much higher in the beginning of the experiment (11.8 g/L/h) and a succinic acid concentration of 83 g/L was reached within 7h. Two other coryneform bacteria, Brevibacterium flavum and Brevibacterium lactofermentum which are very closely related to C.

glutamicum have also been studied for succinic acid production [34–36].

Yeasts, which are commonly used in industry, can tolerate lower pH than the succinic acid producing bacteria. This is interesting because fermenation at low pH would yield mainly the free succinic acid instead of succinate salt and thus reduce the cost of purification [37]. Beside, it is well known that during the fermentation of rice by yeasts for sake production, succinate, which contribute to the taste of this beverage, is formed. DSM and Roquette already developed a low pH yeast-based fermentation process [38] and BioAmber is currently conducting research on a yeast strain capable off using lignocellulosic material and agricultural residues, also at low pH [39].

2.2.2 E. coli

E. coli is a Gram-negative, rod-shaped bacterium (see Figure 5) commonly found in the gut of warm-blood animals including humans. E. coli presents several advantages compare to other microorganisms. Indeed, it grows quickly (even on minimal media), has simple nutriment

Figure 5.Scanning electron micrograph of Escherichia coli Credit: Rocky Mountain Laboratories, NIAID, NIH

requirements, is capable of utilize not only hexoses but pentoses as well (which is the main limitation of yeasts) and to adapt to either aerobic or anaerobic conditions, which makes it easy to handle and to grow. This plus its competency as a host for foreign DNAs and its genetic simplicity has made of E. coli a prime prokaryotic model in biochemical genetics, molecular

(19)

18 Succinic acid

biology and biotechnology. Consequently, extensive knowledge on E. coli has been gained and this bacteria is now very well characterized, especially as its complete genome sequence is available [44]. This implies that numerous tools for genetic manipulation are now available which makes E. coli even more convenient to use.

2.2.3 Mixed acid fermentation in E. coli

The first step in glucose metabolism in E. coli is glycolysis (EMP pathway). In glycolysis, one molecule of glucose is converted to two molecules of pyruvate with a yield of two ATP. Xylose metabolism is more complex as it first must be degraded to D-xylose-5-phosphate before enter- ing the pentose phosphate pathway (PPP). This pathway forms D-glyceraldehyde-3-phosphate and D-fructofuranose-6-phosphate which can join the EMP pathway (see Figure 6).

Figure 6.Hexose and pentose metabolism in E. coli

Under aerobic conditions, pyruvate enters the TCA cycle (see Figure 7) and is oxidized to CO2 with oxygen being a final electron acceptor. This releases large amount of energy in the form of one ATP, three NADH and one FADH2 per molecule of pyruvate.

Under anaerobic conditions, cells undergo mixed acid fermentation instead of the TCA cycle even though this pathway produce much less energy. This is because in the absence of oxygen as an oxidizing agent, regeneration of the NAD⁺ required for glycolysis can only be achieved throught fermentation using NADH to reduce metabolic intermediates (see Figure 8).

(20)

Figure 7.Tricarboxylic acid cycle

Figure 8.Mixed acid fermentation pathway in E. coli constructed from EcoCyc [46]. Metabolites shown in bold are excreted during fermentation.

(21)

20 Succinic acid

That mixed acid fermentation takes place in E. coli in the absence of oxygen has been discovered as early as 1901 [45]. This pathway results in the formation of formate, acetate, succinate, lactate and ethanol.

2.2.4 Production of succinic acid by E. coli

Although E. coli is a natural succinate producer it forms it only as a minor fermentation product.

In order to viably produce succinic acid with this organism, recombined strains with enhanced production abilities must be constructed.

Skorokhodova et al. reviewed different metabolic engineering strategies to improve the succinic acid production in E. coli and compared the performances of different recombinant succinic acid producing strains [47].

Few research groups have attempted to construct strains for aerobic production of succinic acid, for instance by breaking the TCA cycle before conversion or succinic acid into fumarate [47]. However, the theoretical yield of succinic acid is higher for mixed acid fermentation (2 mol/mol_glucosevs 1 mol/mol_glucose) [47]. Most studies have therefore been focused on anaerobic succinate production. Fermentation performances of some of the most promising E.

coli strains are listed in Table 1, along with those of other succinic acid producers.

(22)

21

3 Utilisation of hardwood hemicelluloses

Woods can be classified into two categories: softwood, i.e. wood from gymnosperm trees, and hardwood, i.e. wood from angiosperm trees. Hardwoods represent 17% of the total wood stock in Sweden [48]. As shown on the Figure 9, the most abundant species is by far birch, which account for 67% of the hardwood volume [48]. Hardwood species commonly used in other countries includes for example eucalyptus, poplar or mapple.

Figure 9.Standing volume of different hardwood species in Sweden by % of the total hardwood stock (created from [48])

Hardwood chips are notably used for pulp and paper production and as a fuel to produce heat or electrical power. However, as mentioned previously, to make the most of the hemicellulose fraction, which accounts for 19–36% of hardwood, it might be more interesting to extract it from the wood chips and use it as feedstock for fermentation processes instead.

Indeed, in the Kraft processes, the lignin fraction and most of the hemicellulose are sep- arated from the pulp during the cooking. The black liquor obtained is generally burned to recover energy and chemicals needed for the Kraft process. However, the heating value of hemicellulose (13.6 MJ/Kg) is twice as low as that of lignin [3]. Therefore, extracting the hemicellulose fraction from wood chips before the pulping process and converting it into valuable products might potentially improve the overall economics. Nevertheless, hemicellulose contributes to the paper strength and thus a satisfactory balance has to be found between pulp yield and property and amount of hemicellulose extracted [49–51].

Extraction of hemicellulose prior to combustion and gasification processes could be interesting for combined heat and power plants since it has been reported that hardwood chips have a relatively high heating value, a lower ash content and significantly lower concentrations of alkali metal after partial removal of the hemicellulose by hot water extraction [4].

(23)

22 Utilisation of hardwood hemicelluloses

3.1 Hemicellulose structure

In hardwood, or any other lignocellulosic material, cellulose, hemicellulose and lignin interact with each other and are tightly packed together (Figure 10). Cellulose chains are held together by hydrogen bonds, thereby forming microfibrils which are crosslinked with hemicellulose chains to form a stable network. This network is further reinforced by a matrix of lignin, an aromatic polymer, which is covalently linked to the hemicellulose and fills some of the gaps in the structure.

Figure 10.Plant cell wall

Hemicelluloses are heterogeneous polymers of pentoses (D-xylose and L-arabinose), hexoses (D-glucose, D-mannose and D-galactose) and sugar acids. They have a linear backbone consisting of either a single sugar (homopolymer) or a mixture of different sugars (heteropoly- mers), which is often substituted with shorter sugar chains or acetic, glucuronic and ferulic acids. According to their composition, hemicelluloses can be classified into xylans, mannans, arabians, galactans or glucans. While cellulose has a linear, crystaline structure, hemicelluloses are branched and amorphous which make them more soluble and easier to hydrolyse. [6, 52, 53]

The type of hemicellulose depends on the source of plant material. Although hemicelluloses also differ both quantitatively and qualitatively from one hardwood species to another, the major component is an O-acetyl-4-O-methylglucurono-β-D-xylan which represents 15-30% of the dry wood. It is often simply called glucuronoxylan. Its structure is shown Figure 11. It has a backbone of β-D-xylopyranose units linked by 1,4-bonds and about 60 to 70% of those xylose residues are O-acetylated at position 2 or 3 (acetyl groups are shown in purple Figure 11) [54].

Besides, about one out of ten xylose residue carries an 4-O-methyl-α-D-glucoronic acid group (shown in green on the Figure 11) linked to xylose by α-1,2-glycosidic linkages. [52, 53, 55]

(24)

3.2 Hydrolysis 23

Hardwoods contain 2-5% of a glucomannan as well. This polysaccharide consists of β-D- glucopyranose and β-D-mannopyranose units linked by 1,4-bonds; the glucose:mannose ratio depends on the wood species and varies between 1:1 and 1:2. Other types of hemicelluloses may be present in hardwoods, but in even lower amounts. [53, 55]

Figure 11.Structure of glucuronoxylan from hardwood

3.2 Hydrolysis

As mentioned before, hemicellulose could be pre-extracted from wood chips prior to the process in Kraft pulping mill or combined heat and power plant. Most commonly investigated methods are alkaline extraction [50, 51], hot water extraction [4, 49, 56], auto-hydrolysis [57]

and acid-hydrolysis [57].

However, except in the case of the latter method, the extracted hemicellulose needs to be further hydrolysed in order to break down the polysaccharides into fermentable sugars. Two approaches can be considered: either chemical or enzymatic hydrolysis.

The enzymes used for hemicellulose hydrolysis are called hemicellulases and although they are produced by different bacteria, yeasts and fungi only filamentous fungi (such as Trichoderma reesei, Aspergilus niger or Humicola insulens) produce enzymes titres high enough to be commer- cially used. Depending on the kind of hemicellulose, different hemicellulases are needed. The hemicellulase necessary to degrade hardwood xylan are shown in green in Figure 12. Xylanases are backbone degrading enzymes, whereas α-D-Glucuronidases and acetyl xylan esterases are cleaving off the side chains and β-xylosidases are cutting trisaccharides and disaccharides into xyloses monomers. Additional hemicellulases are required to hydrolyse the glucomannan as well.

Chemical hydrolysis is generally performed with an acid, either concentrated or dilute, as catalyst. Sulfuric acid (H2SO4) is the most common catalyst in acid-based hydrolysis, although other mineral acids, such as hydrochloric acid (HCl), might be used as well.

Concentrated acid hydrolysis is performed at low temperature (∼ 40^◦C) and with an acid concentration in a range of about 30–70%. Dilute acid hydrolysis however is carried out at a much higher temperature (120-200^◦C) and an acid concentration within the range of 0.5-1.5%.

(25)

Figure 12.Effects of xylanotic enzymes on xylan

Both enzymatic and chemical hydrolysis have been reviewed by Taherzadeh and Karimi [58, 59]; a summary of the different advantages and drawbacks of the three methods is presented in Table 2.

Table 2.Comparison of different hydrolysis methods (adapted from [58, 59]) Hydrolysis method

Characteristics Enzymatic Dilute acid Concentrated acid

Advantages

mild hydrolysis conditions X × X

high yield of hydrolysis X × X

low cost of catalyst × X ×

short time of hydrolysis × X ×

Disadvanages

equipment corrosion × X X

high formation of inhibitors × X ×

product inhibition during hydrolysis X × ×

high energy consumption for acid recovery × X

(26)

3.3 Inhibitors in hydrolysates 25 3.3 Inhibitors in hydrolysates

3.3.1 Generation of inhibitors

During hemicellulose hydrolysis, different compounds that are toxic for microorganisms are generated. They are classified in three groups: the furans derivatives (or furaldehydes), the weak (or aliphatic) acid and the phenols. The different pathways for inhibitors formation are shown Figure 13.

Figure 13.Generation of inhibitors during hardwood hydrolysis.

Sugars: pentoses (dark grey), hexoses (black) ; inhibitors: furan derivatives (green), weak acids (purple), phenolic compounds (blue).

In harsh conditions sugar monomers are further degraded into furans; furfural comes from pentoses degradation whereas HMF is formed from hexoses (as reviewed by Palmquist and Hahn-H¨agerdal [8]). Little furaldehydes generation occurs during enzymatic hydrolysis because they are performed at mild temperature and also because hemicellulotic enzymes do not catalyse those degradation reactions.

Formic and levulinic acid are formed from furfural and HMF degradation, respectively [8].

Acetic acid is typically the most important weak acid [60] as it comes directly from the deacetylation of hemicellulose xylan.

Finally, phenols are alcohols, acids, aldehydes or ketones generated from the breakdown of lignin. They can be classified into hydroxyl, guaiacyl or syringyl compounds depending on the lignin monomer they are coming from. The different phenols that have been identified in hemicellulose have been reviewed by Klinke et Al. [5].

(27)

26 Utilisation of hardwood hemicelluloses 3.3.2 Inhibitors in hardwood hemicellulose hydrolysates

The inhibitors generated during the hydrolysis of hemicellulose depend on both the type of wood and the type of hydrolysis. As described in 3.2, the main hemicellulose in hardwood is a glucuronoxylan, which means that xylose is the most present sugar monomers in the hydrolysates. Therefore, a certain amount of furfural, a degradation product of pentoses, is generated upon hydrolysis. Furthermore since hardwood hemicelluloses also contain a few hexoses polymers, such as glucomannans, very low concentrations of HMF might be found in the hydrolysates.

Acetic acid is usually in important amount since glucoronoxylans are highly acetylated. As for the other weak acids, since they are degradation products of furans (which are not present in very high concentrations in the hydrolysates), they are not likely to be formed, especially not levulinic acid which comes from HMF.

Hardwood lignin contains coniferyl and sinapyl alcohols as well as small amounts of p- coumaryl alcohols [5], therefore a wide range of phenolic compounds might be generated upon hydrolysis. Table 3 shows different phenolics that were identified in various hardwood, according to a review of Palmqvist and Hahn-H¨agerdal [8].

Table 3.Phenols in hardwood Phenols Type of wood Vanillic acid willow/poplar/red oak

Vanillin willow/poplar/red oak Syringic acid hardwoods Syringaldehyde hardwoods

catechol willow/birch

POH poplar/aspen/willow

The concentration of different inhibitors formed during hydrolysis depends on the wood species, the hydrolysis and even the extraction method as it was shown that changing the extraction temperature changed the composition of the extracted liquid and thus of the hydrolysates [4]. Table 4 shows the composition of hydrolysates obtained from different woods and with different methods.

In this work, one representative of each group of inhibitors was studied. Acetic acid and furfural were picked since they are, respectively, the most abundant weak acid and furan in hardwood hemicellulose hydrolysates and POH was chosen to represent the phenols. Besides, to compare the fermentability of the inhibitors mixture with that of hydrolysates, birchwood hemicellulose was used since birch is the most common hardwood in Sweden.

(28)

3.3 Inhibitors in hydrolysates 27

Table 4.Composition of different hardwood hemicellulose hydrolysates (concentrations in g/L) Wood Total sugar Xylose Acetic acid Furfural HMF Phenols Hydrolysis method Ref.

Eucalyptus 13.8 12.34 3.41 0.26 0.07 2.23

Dilute acid 0.65% H2SO4

157^◦C - 20 min

[61]

Eucalyptus 30 25.6 5.4 2.1 n.s. n.s.

Dilute acid 0.35% H2SO4

156^◦C - 27 min

[62]

Oak 63.5 43.5 10.9 0.9 0.3 n.s. Dilute acid [63, 64]

Birch n.s. 68 20.3 <2.3 0 n.s.

Dilute acid 4% H2SO4

121^◦C - 1h

[4]

Birch 52.1 46.2 9.0 n.s. n.s. n.s.

Enzymatic T. reesei^a 45^◦C - 24h

[65]

Birch 45.5 35.9 4.8 n.s. n.s. n.s.

Enzymatic Aspergillus awamori^a

45^◦C - 24h

[65]

Aspen 25.4 23.7 6.1 0.28 n.s. POH: 1.07

vanillin: 0.21

Dilute acid 3% H2SO4

121^◦C - 1h

[66]

Aspen 22.8 20.2 5.3 0.05 n.s. POH: 0.82

vanillin: 0.23

Enzymatic Trichoderma harzianum^a

45^◦C - pH 4.8 - 48h

[66]

n.s.: not specified

ahemicellulases mixture produced by

(29)

3.4 Toxicity

3.4.1 Inhibition mechanism

Modes of toxicity for the different groups of inhibitors are described bellow and a summary is presented in Figure 14.

Figure 14.Summary of the inhibition mechanisms of the toxic compounds in hemicellulosic hydrolysates [67]

Weak acids

The growth inhibition by weak acids is caused by the inflow of undissociated acid into the cytosol [2, 7, 8, 67]. Indeed, weak acids are present in this form in hydrolysates due to their relatively high pKa. The protonated form of weak acids is liposoluble and can thus permeate the cell membrane freely. Once inside it dissociates because of the higher intracellular pH (approx.

7.8 [68]), thereby releasing the anion and the proton (see Figure 15). The toxicity of weak acid is pH dependent since the proportion of undissociated acid increases with decreasing pH.

Primarily, the growth inhibition was believed to be due to the uncoupling effect. Protons are normally allowed in the cell through the plasma membrane ATPase via a process coupled to the phosphorylation of ADP into ATP, which is then used as energy for biomass formation.

However undissociated weak acids act as uncoupling agents in that they allow a proton across the membrane without the creation of ATP. Besides, the inflow of protons causes a decline of the intracellular pH. To counteract this pH drop, the plasma membrane ATPase pumps protons out of the cells, which requires ATP (see Figure 16). Consequently the maintenance of the cytosalic pH can only be achieved at the expense of biomass formation as less energy is available for growth. Furthermore, if too many protons are released in the cytosol, the proton pumping capacity of the cell will be exhausted, thereby depleting the proton motive force and causing the acidification of the cytoplasm [8], which will cause cell death.

Nevertheless this uncoupling theory has been questioned [68] since weak acids differ from

(30)

3.4 Toxicity 29

Figure 15.Inflow of weak acid in the cytosol:

the undissociated acid can permeate the membrane and then dissociate because of the high pH.

Figure 16.Schematic representation of the uncoupling theory:

to counter balance the inflow of protons and avoid pH drop the plasma membrane ATPase pumps protons out, thereby consuming energy.

Figure 17.Schematic representation of synthetic uncouplers action:

once the proton is released, the anion exits the cell, is protonated, brings another proton in the cytosol and so on. XCOOH: synthetic uncoupler.

(31)

synthetic uncouplers in that they are not hydrophilic. In other words, both dissociated and undissociated synthetic uncouplers can freely pass across the cell membrane, which allows them to transport protons inside the cell at a rapid rate (see Figure 17) [68]. On the contrary, the lipophobic weak acids anions are trapped inside the cell; each of them can therefore import only one proton, which seems insufficient to dissipate the proton motive force [68].

That is why another explanation for the toxicity of weak acids has been suggested: the anion accumulation [68]. As mentioned previously, once released the anion is captured inside the cell and the protonated acid permeate the membrane until equilibrium is reached. The resulting anion accumulation causes an increase in the internal osmotic pressure of the cell, thereby affecting the turgor pressure [69]. Other mechanisms are involved in the toxicity of this accumulation although they are not yet fully understood. For instance the accumulation of anions in E. coli cells has been shown to be partly compensated by a reduction of the intracellular pool of glutamate [69], an important precursor of many different amino acids [67].

Other unidentified anions are most likely to have their intracellular pool reduced as well [69].

Roe et al. [70] found that in presence of weak acid the intracellular methionine pool is depleted and homocysteine accumulates. Since homocysteine is a precursor of methionine, their findings suggested an inhibition of the methionine biosynthetic pathway, which they proposed to be caused by the reduction of the intracellular pool of important anions. They have also demonstrated that intracellular homocysteine is inhibiting E. coli growth and so is probably methionine depletion as this amino acid is essential for protein biosynthesis. The concentration of anions inside the cell being dependent on the pH gradient over the plasma membrane [8], the anion accumulation theory would, according to Russell [68], explain why bacteria that let their intracellular pH drop are more resistant.

Zaldivar et al. [60] investigated the effect of organic acids on membrane integrity by study- ing the magnesium leakage in the presence of those toxins. They showed that while increased cell permeability may contribute to the weak acid inhibition, leakage was too small for it to be the primary mechanism of toxicity.

Furan derivatives

Like for most inhibitors, the inhibition mechanism of furans appears to be rather complex. It has been demonstrated that different bacteria [65], including E. coli [71, 72], and yeasts [73–78]

were able to transform furfural into its alcohol (furfuryl alcohol). Similarly the conversion of HMF into 5-(hydroxymethyl)-furfuryl alcohol by E. coli [72], Clostridium acetobutylicum [79] and yeast [75–78, 80] has been observed. Cell growth is inhibited during furfural conversion but resumes after complete reduction to furfuryl alcohol [81, 82]. Since furfural reductase activity in E. coli is NADPH-dependent [81, 82], it was proposed that furfural inhibits growth by diverting NADPH away from biosynthesis [82]. A number of findings support this hypothesis. Indeed Miller et al. demonstrated that furfural inhibition in E. coli could be relieved by silencing ydhD, a NADPH-dependant oxidoreductase [82]. YdhD have a higher affinity (lower Km) for NADPH than many of the biosynthetic enzymes [82]. Besides overexpression of the fuco gene

(32)

3.4 Toxicity 31

was showed to increase furfural tolerance in E. coli [83]. FucO exhibit a furfural reductase activity that is, unlike that of YdhD, NADH dependent. Overexpression of this gene thus allows for furfural reduction without NADPH depletion. Finally, overexpression of the pntAB gene also decreased furfural toxicity in E. coli [84]. PntAB is a cytoplasmic NADH/NADPH transhydrogenase that plays an important role in NADPH formation [85]. It can thus reason- ably be assumed that the higher tolerance to furfural is due to an increase of the NADPH pool, which give another evidence of the link between furfural inhibition and NADPH limitation.

The biosynthetic reaction that seems the most affected by furfural is the sulfate assimilation required to form cysteine and methianine (sulfur-containing amino acids) [84]. Miller et al.

observed that supplementation with reduced sulfur sources but not that of taurine increased E.

coli resistance to furfural [84]. It therefore appears that furfural prevents the reduction of sulfite by CysIJ [84] an enzyme with a lower Km for NADPH than YqhD [76].

Furfural is known to alter DNA structure and sequence [86, 87]. Recently, it was demonstrated that overexpression of thyA increases E. coli tolerance to furfural. ThyA is involved in the biosynthesis of dTMP, an essential nucleotide for DNA repair [88]. Therefore, it is most likely that furfural inhibition is also linked to DNA damages. Need for DNA repair might participate in NADPH starvation as this reducing agent is necessary for dTMP biosynthesis.

Furfural could also have other sites of actions. Global transcripts analysis was used to study the effect of furfural on E. coli [84]. However many of the genes significantly perturbed by furfural either have putative functions or were unclassified or unknown. Their role in the furfural response is therefore not yet understood. Other mechanisms of toxicity have been identified in Saccharomyces cerevisiae such as inhibition of glycotytic [89–91] and non-glycolytic enzymes [77, 92] and accumulation of reactive oxygen species [93]. The mechanisms of inhibition of furfural in E. coli and possible solutions to alleviate it are represented in Figure 18.

Furfural and HMF have a partially common mechanism of inhibition as a furfural resistant strain acquired tolerance to HMF [94]. It was concluded that HMF also inhibits growth by depleting the pool of NADPH necessary for biosynthesis. Nevertheless, it is likely that both furaldehydes have modes of toxicity of their own as well [94].

Phenols

As mentioned previously, the phenolic group of inhibitors in hydrolysates gather a wide range of compounds, i.e. aldehydes, acids, alcohols and ketones (see Section 3.3). This heterogeneity, added to the lack of accurate qualitative and quantitative analytic methods for phenolic derivatives, has rendered the identification of their mechanisms of inhibition rather complex [7].

Phenolic compounds have been reported to increase the permeability of cells membranes [95–

97]. For instance, Heipieper et al. observed an efflux of potassium, cellular metabolites (such as ATP) and nucleotides when cells were grown in the presence of phenol [95].

The high lipid solubility of phenol was suggested to be responsible of its fluidizing action by disturbing the structural order of lipids in the membrane [96].

(33)

Figure 18.Modes of toxicity of furfural in E. coli and possible solutions for increased tolerance.

Furfural acts by damaging the DNA and depleting the pool of sulfur-containing amino acids thereby stopping biosynthesis. Overexpression of fuco, pntAB and thyA (shown in purple) and silencing of

yqhD (green) are beneficial for E. coli resistance.

(34)

3.4 Toxicity 33

Zaldivar et al. used magnesium leakage as an indicator of membrane damage caused by different phenolic derivatives [60, 98, 99]. The extend of membrane damage was shown to depend on the type of phenol. Indeed, magnesium leakage for alcohols was more important than for the other phenolics compounds, in particularly aldehydes. This shown that other mechanisms must be involved in the toxicity of phenols, at least in that of acids and aldehydes.

One of those other mechanisms could be the disruption of the proton motive force. Indeed, weakly acidic phenolic compounds may act in a similar fashion than acetic, formic and lactic acid (described previously).

3.4.2 Inhibition effects

E. coli

Zaldivar et al. investigated the effect of a wide range of inhibitors (furans, weak acids and phenolic derivatives) on the growth and fermentation of E. coli LY01 [60, 98, 99]. A summary of their findings is presented in Table 5. IC25and MIC represent the concentrations inhibiting 25% and 100% of growth, respectively.

Table 5.Inhibitor effect on E. coli LY01 Inhibitor IC25 MIC Ref.

Furan derivatives

furfural 2.0 3.5 [98]

HMF 2.3 40 [98]

Weak acids

formic acid 1.15 17.5 [60]

levulinic acid 3.0 40 [60]

acetic acid 5.0 25 [60]

Phenolic aldehydes

4-hydroxybenzaldehyde 0.15 1.12 [98]

syringaldehyde 0.3 2.5 [98]

vanillin 0.4 1.5 [98]

Phenolic acids

4-hydroxybenzoic acid 0.4 15 [60]

syringic acid 0.7 17.5 [60]

vanillic acid 0.55 15 [60]

gallic acid 2.3 40 [60]

Phenolic alcohols

coniferyl alcohol 0.15 3 [99]

methylcatechol 0.20 1.5 [99]

catechol 0.35 3 [99]

vanillyl alcohol 1.0 9.0 [99]

(35)

As it can be seen, phenolic derivatives exhibited very different toxicities. A general trend however was that phenolic aldehydes were the most toxic, followed by phenolic alcohols and then phenolic acids. Although large concentrations of those latter were required for complete inhibition (e.g. MIC= 15 g/L for POH), they were already toxic at very low concentrations (e.g.

IC25= 0.4g/L for POH).

Both furans were inhibitory at fairly low concentrations. Furfural, which is, as mentioned before (see Section 3.3.2), the main furan in hardwood hemicellulose hydrolysates, was slightly more toxic than HMF.

E. coli could grow even with high concentrations of weak acids in the medium. Acetic acid, the most important inhibitor in hardwood hemicellulose hydrolysates (see Section 3.3.2) was globally the least toxic weak acid, except at high concentrations (more toxic than levulinic acid).

Furans and phenolic aldehydes (with the exception syringaldehyde) were the only inhibitors shown to have a positive effect on ethanol production, when present in low enough concentrations.

Effect of binary combinations of inhibitors on growth were also studied [60, 98, 99]. Most combinations tested were less than additive. Nevertheless, combinations with furfural were an exception: association with acetic acid, HMF and phenolic aldehydes were synergistic and greatly inhibited the growth. In constrast, combinations of furfural with some phenolic acids (ferulic acid and POH) and phenolic alcohols (methylcatechol and vanillyl alcohol) were less than additive.

Effects of different carbon sources on growth inhibition by acetic acid [100] and furfural [82]

have been investigated. Both studies showed that toxicities were increased when xylose was the sugar compare to glucose. Indeed xylose consumption by E. coli KO11 was sligthly more affected by acetate than that of glucose [100]. Besides, ethanol yield and volumetric productivity were stimulated by acetate (up to 12 g/L) when glucose was the carbon source but not when it was xylose. Miller et al. observed that furfural MIC for E. coli LY180 was 1.0 g/L in a mineral salts medium supplemented by xylose; replacing xylose by glucose resulted in a MIC of 1.5 g/L. The authors suggested this to be due to the increased availability of NADPH (see Section 14).

Yeast

This work used a study on the influence of acetic acid (0–10 g/L), furfural (0–3 g/L) and POH (0–2 g/L) on bakers’ yeast (S. cerevisiae) as a model [74]. In that study four parameters were investigated: the specific growth rate (µnet), the biomass yield (Y^X_/_S), the ethanol volumetric productivity (QP) and the ethanol yield (Y^P_/_S). The main and interaction effects they observed are summarized in Table 6.

As it can be seen, the interaction effects were the same then those observed on E. coli: acetic acid and furfural had a synergistic effects whereas other combination were not particularly harmful. The negative interaction of acetic acid and furfural was also reported for the yeast

(36)

3.5 Overcoming inhibition problems 35 Table 6. Effects of acetic acid, furfural, POH and the combination acetic acid and furfural on different parameters during ethanol fermentation by bakers’ yeast (as reported in [74]).

Inhibitor µnet YX/S QP YP/S

Main effects

Acetic acid × × % up to 9 g/L &

Furfural & % up to 1 g/L

& above 1 g/L & % up to 2 g/L

& above 2 g/L

POH × × × ×

Interaction effects

AA + Fur – – × –

AA + POH × × × ×

POH + Fur × × × ×

×: no significant effect was observed –: negative interaction

Kluyveromyces marxianus [73]. However, bakers’ yeast was not affected by POH (at least up to 2 g/L), whereas even though E. coli LY01 could tolerate fairly high concentrations (up to 15 g/L) of this phenol, 0.15 g/L was enough for a 25% inhibition growth. Similarly, in contrast to E. coli LY01 which was very sensitive to catechol (see Table 5), up to 2 g/L did not influence growth and ethanol fermentation by K. marxianus [73].

Interestingly, acetic acid and furfural could, in some cases, slightly stimulate the ethanol production by S. cerevisiae (see Table 6).

3.5 Overcoming inhibition problems

The production of succinic acid or other valuable products from hemicellulosic feed-stock is a real challenge due to the need for hydrolysis and the strongly inhibitory compounds thereby generated. Different strategies to overcome those inhibition problems are presented in Figure 19. The hydrolysate may be detoxified although this has a cost, more efficient process may be designed or improvements can be done on the organism itself.

3.5.1 Detoxification

Different methods to detoxify lignocellulosic hydrolysates have been extensively studied and reviewed [101–103]. These methods can be classified as physical, chemical or biological. With physical methods, inhibitors are removed from the hydrolysates. This can be achieved by phase equilibria-based separations based on either solubility or volatility, for instance, liquid- liquid extraction or roto-evaporation, respectively [104]. Extractions performed with different solvents such as diethyl ether or ethylacetate have been reported to remove furans, acetic acid and some phenols [101]. Separations based on physical absorption onto a solid substrate are

(37)

Figure 19.Overcoming the inhibition problems

another type of physical detoxification methods [104]. Examples are treatment with activated carbon, ions exchange resins or membrane separations [103, 104].

Chemicals methods consist in modifying the inhibitors to less toxic or non-toxic compounds [104]. Sulphite has been used as a reducing agent for several decades and has been shown to reduce furans concentration [101]. Alkali treatment is a common chemical method based on the instability of some inhibitors where the pH is increased to 9-10 with a strong base such as sodium, potassium or calcium hydroxyde. Alkali treatment using the latter is called overliming and is usually more efficient than treatment with other bases since it leads to the precipitation of some inhibitory compounds. Overliming has been reported to remove furans and ketones [101].

Biological detoxification methods are either enzymatic or microbial. The enzymes used for detoxification, i.e. laccases and peroxidases, are mainly provided by the white rot fungus Trametes versicolor, but alternative fungi such as Cyathusc stercoreus or Cythus bulleri can be used.

Treatment with those enzymes has been reported to remove the phenols compounds from the hydrolysates. Some microorganisms are able to degrade the inhibitory compounds and can therefore be grown directly on lignocellulosic hydrolysates to detoxify them, hence the name of in-situ detoxification. Althought the soft-rot fungus Trichoderma reesei is a widely used microorganism, various fungi (e.g. Coniochaeta ligniaria), yeasts (e.g. S. cerevisiae) and bacteria (e.g.

Ureibacillus themosphaericus) are used for detoxification. The nature of the inhibitors removed depends on the microorganism. [101–103]

Since the concentrations of inhibitors generated during hydrolysis depends on both the lignocellulosic material and the hydrolysis method and since different microorganisms have different inhibitor tolerance, the most suitable detoxification method depends highly on the process. For example, a study comparing activated carbon treatment and overliming for the detoxification of softwood hydrolysates showed that even though activated carbon treatment

Effects of acetic acid, furfural and p-hydroxybenzoic acid on succinic acid fermentation by Escherichia coli BSS133

MASTER'S THESIS

Effects of Acetic Acid, Furfural and P- hydroxybenzoic Acid on Succinic Acid Fermentation by Escherichia coli BSS133

Mireille Ginésy 2014

Effects of acetic acid, furfural and p-hydroxybenzoic acid on succinic acid fermentation by Escherichia coli BSS133

Abstract

Preface

Contents

List of Figures

List of Tables

1 Introduction

2 Succinic acid

3 Utilisation of hardwood hemicelluloses