Bioplastics from the Food Waste Liquid Fraction

(1)

IN

DEGREE PROJECT BIOTECHNOLOGY,

SECOND CYCLE, 30 CREDITS ,

STOCKHOLM SWEDEN 2017

Bioplastics from the Food

Waste Liquid Fraction

Characterization of production strain and product

expression

VIKTOR WESTERLUND

KTH ROYAL INSTITUTE OF TECHNOLOGY

Supervisor: Martin Gustafsson

Examiner: Antonius Van Maris

(2)

Abstract

Food waste is a rich source of nutrients and can be used in many microbial valorisation applications. Today, food waste in Sweden is treated mostly in anaerobic digesters for production of biogas. However, biogas-production is not an economically lucrative process and the economics are mainly balanced through tax regulations of the process power consumption [1]. Alternatively, other end-products of higher value could improve the economics of food waste treatment processes. One such end-product is polyhydroxybutyrate (PHB), a biodegradable bioplastic. In this study, we characterize

E. coli, W3110, !fadR harbouring the plasmid pCnCAB and investigate the nature of its PHB

production system. The strain produces PHB by expression of three genes from the Cupriavidus

necator PHB-operon. The natural function of PHB is carbon storage which accumulates during

(3)

1 Introduction

Today, we face a major challenge with anthropogenic climate change and loss of biodiversity [3, 4]. Increased large-scale fossil raw material exploitation [3] and non-degradable plastic accumulation in our terrestrial and aquatic environments are significant contributors to these adverse environmental effects [5]. Thus, a transition into renewable fuels but also renewable and biodegradable materials is required to stop this trend. With an increasing population, the consumption of the world’s resources is predicted to increase. As of mid-2014, a population of 7.2 billion is according to estimations predicted to grow by 1 billion within the next decade [6]. Thus, the production of waste is estimated to increase and possibly double by the year 2025 [7]. Thereby, it is important to have efficient waste treatment methods available, which can transform waste into valorised end-products.

As with other types of waste, food waste is a pronounced environmental and societal problem. Globally, 1.3 billion tonnes of food waste is produced annually, which corresponds to one third of the total food production [7, 8]. In Sweden, 1,210,000 tonnes of food waste were generated in 2012, of which 771,000 tonnes was produced from households. This corresponds to 81 kg/person annually [9]. The waste can be divided into two broad categories, preventable and non-preventable food waste [8]. The preventable fraction in this report, refers to wasted food that is suitable for human consumption and is thrown anyhow due to various reasons. The non-preventable fraction refers to food waste, which is produced as by-product from food industry or food waste not suitable for human-consumption. Directives aim at reducing the preventable fraction of food waste to reduce excessive resource consumption. Nevertheless, the non-preventable fraction is significant and requires efficient recycling and valorisation measures [10].

(5)

restaurant, industry) [14]. A biological process utilizing food waste as raw material must therefore be robust and might require adjustments to each batch to provide optimal results. Thus, the results obtained from this study might not be applicable in other cases, as food waste composition might vary. A recent review [14], summarized chemical characteristics of food waste for biohydrogen production. From three studies using food waste as raw material, the following was concluded and is summarized in table 1.

Table 1. Food waste characteristics from three individual studies [14].

Moisture content COD Carbohydrate C/N ratio

72-85.2% 19.3-346 g/L 25.5-143 g/L 9-21

Further, calculating an average pH from 14 studies on food waste resulted in an average pH of 5.4 [14]. The data from Table 1, will mainly provide rough estimations of food waste nutritional content and properties.

The food waste nutritional content is relatively high and thereby food waste has been identified as a resource productivity opportunity [15]. It presents a rich source of raw material for a wide range of biological valorisation applications, which enables new innovative products to be obtained from food waste [8]. As mentioned, food waste in Sweden is treated mostly in anaerobic digesters for production of biogas. However, biogas-production is not an economically lucrative process and the economics are mainly balanced through tax regulations of the process power consumption [1]. Alternatively, other end-products of higher value could improve the economics of food waste treatment processes. One such end-product is polyhydroxybutyrate (PHB), a biodegradable bioplastic, which is expressed intracellularly within certain microorganisms.

The advantage of producing the final end-product intracellularly is the simplified product separation from the complex medium components. In a complex medium such as food waste, which contains many different compounds, a selective separation procedure might be hard to achieve if the product would be secreted out into the medium. Instead, if the product is intracellular, simple centrifugation can separate the cells and thereby product from the cultivation medium, simplifying the purification process significantly. A simplified separation improves the overall economy of the process, however it is not the only advantage PHB production offers.

(6)

carbon excess and limitation/starvation of another key nutrient. As this key nutrient is present again, the PHB granules can be depolymerized and utilized as carbon source.

PHB is a biocompatible monomer, which makes it suitable to use in many medical applications [17] e.g. biocompatible stitches . PHB can also be utilized in the production of many single-use products, in which biodegradability would be beneficial in an environmental point of view. Examples of single-use products are plastic bags, disposable containers and plastic plates. As PHB derived from food waste might contain impurities, not suitable for medical application, single-use products are more suitable as end-use application.

More than 150 monomer constituents belong to the group of PHAs [18]. Poly-hydroxybutyrate, PHB is the most common of the various types of PHA polymers and was initially (in 1926) thought to be the only bacterial polyester naturally occurring. Subsequently, approximately 50 years later researchers found other constituents forming polyesters in microorganisms [19]. Consequently, this microbial polymeric material is referred to as PHA.

The general structure for PHA polyesters is shown in Figure 1. The structure of 3HB includes a methyl-group in R-position. The monomer backbone contains three consecutive carbon atoms and each monomer is separated by an ester-group [20].

Figure 1. The chemical structure of a general PHA monomer. Structure derived from [20].

(7)

In each production organism, a PHA synthase is responsible for polymerization of PHA to PHA granules [23]. Depending on the supply of substrates, PHA synthase can catalyse the polymerization of several PHA monomers as it has a relatively broad substrate spectrum. If the substrates are provided in the cultivation medium, they are often in the form of alternative intermediate metabolites, which provides PHA synthase with alternative substrates in addition to native substrates. Thus, resulting in various co-polymers. Another option is an in vivo approach, using metabolic engineering. The metabolic pathways are altered to provide PHA synthase with alternative substrates, and thus enable new homo-/copolymer combinations. Some microorganisms naturally express co-polymers while other microorganisms are genetically engineered to express them.

An E. coli, W3110, !fadR, pCnCAB was applied in this study. !fadR, implies a deletion of the fadR gene, which regulates genes in the β-oxidation. The plasmid, pCnCAB harbours genes from

Cupriavidus necator (C. necator) formerly known as Ralstonia eutropha, containing the PHA synthesis

operon (genes phaA, phaB and phaC) [24]. These express PHA synthase intracellularly in the cytosol and provide an alternative metabolic pathway for acetyl-CoA in order to generate 3-hydroxybutyric-CoA (3-HB-3-hydroxybutyric-CoA). The following metabolic pathway was the result: acetyl-3-hydroxybutyric-CoA (first step), two acetyl CoAs are condensed to acetoacetyl-CoA by β-ketoacyl-CoA thiolase (encoded by phaA). The product is subsequently reduced to 3-hydroxybuturyl-CoA by the NADPH-dependent enzyme, acetoacetyl-CoA dehydrogenase (encoded by phaB). 3HB-acetoacetyl-CoA is thereafter polymerized by the processive enzyme PHA synthase (encoded by phaC) [24], Figure 2.

Figure 2. Illustration of the PHB pathway from acetyl-CoA.

The main reason for incorporating the PHB pathway into E. coli is the increased growth rate. If efficient production of PHB is obtained from E. coli it can produce PHB at a faster rate than natural PHB producers. Further, E. coli does not have an system for utilizing PHB as carbon source. Thereby, it can not degrade PHB, which upon accumulation remains in the cells [2].

(8)

was in the 1970s. ICI, in Great Britain produced PHA in 200,000 l fermentation containers applying C.

necator as PHB-producing organism, by fermenting fructose. Monsanto continued the development

of PHA polymers by developing a co-polymer P(3HB-co-3HV) by feeding propionate to a C. necator mutant-strain utilizing glucose. This polymer exhibited better flexibility and impact resistance than PHB as mentioned before. Sucrose from molasses was also applied as feedstock for production of P(3HB) by Chemie Linz GmbH in Austria for a production rate of up to 1000 kg/week by using

Alcaligenes latus as production organism [25].

(9)

2 Material and methods

2.1 Bacterial strain and plasmid

The bacterial strain applied in this study was E. coli, W3110. For reference experiments, W3110 wild type was used and for PHB production, W3110, !fadR, pCnCAB was used (obtained by the courtesy of Prof. S.Y. Lee, KAIST, South Korea). The pCnCAB plasmid harbours the C. necator biosynthesis operon inserted into the ApR_{, cloning and expression vector, pBluescriptII KS(+) [26].}

2.2 Cultivation media

3 different cultivation media were applied in this study: Minimal resource medium (MR-medium):

The MR medium (pH 7.0) contained (per liter): 6.67 g KH2PO4, 4 g (NH4)2HPO4, 0.8 g MgSO4•7H2O,

0.8 g citric acid. Glucose (Glc) was added to a concentration of 20 g L-1_{and 5 ml of trace metal solution}

(both sterilized and added separately). Lastly, sterile antifoam was added to a concentration of 50 μL L−1_{. The trace metal solution contained (per liter of 0.12 M HCl): 10 g FeSO}

4•7H2O, 2 g CaCl2, 2.2 g

ZnSO4•7H2O, 0.34 g MnSO4•H2O, 1 g CuSO4•5H2O, 0.1 g (NH4)6Mo7O24•4H2O, and 0.02 g

Na2B4O7•10H2O.

Minimal salt medium (MS-medium):

The MS medium (pH ~7) contained (per liter): 5 g L−1_(NH

4)2SO4, 1.6 g L−1 KH2PO4, 0.7 g L−1 Na3C6H5O7•2H2O, 6.6 g L−1 Na2HPO4•2H2O. Glc was added to a concentration of 20 g L-1 and 1 ml of trace element solution and 1 ml of 1M MgSO4 (all sterilized and added separately). Lastly, sterile antifoam was added to a concentration of 50 μL L−1_{. The trace element solution contained (per liter):}

0.5 g L−1_{CaCl2 •2H2 O, 16.7 g L}−1_{FeCl3 •6H2 O, 0.18 g L}−1_{ZnSO4 •7H2 O, 0.16 g L}−1_{CuSO4 •5H2 O,} 0.15 g L−1_{MnSO4 •4H2 O, 0.18 g L}−1_{CoCl2 •6H2 O, 20.1 g L}−1_Na-EDTA.

Food waste liquid fraction (FWLF)

(10)

2.3 Cultivation procedures

2.3.1 pH-dependent growth on FWLF

The FWLF was prepared as described in 2.2. MOPS was added to FWLF corresponding to a concentration of 40 mM. E. coli cells from a glycerol stock at -80 °C were used to inoculate a sterile 1 L shake flask containing 100 ml of minimal salts medium (MS-medium) (pH of 7). As the seed culture reached OD600 of 1, 20 ml was extracted into a sterile sealed glass tube. The glass tube was centrifuged at 2000 g for 10 min (HERMLE, Z 206 A) and the supernatant was discarded. The cell pellet was then resuspended in 20 mL 0.9% w/v saline solution. Shake flasks containing FWLF and 40 mM MOPS, are defined as buffered FWLF. Shake flasks containing pure FWLF, are defined as unbuffered FWLF. Thus, shake flasks of 1 L containing 100 ml of buffered and unbuffered FWLF were inoculated with 5 mL of cell suspension. OD600 and pH was measured (SevenCompact, Mettler Toledo) continuously during the cultivation.

2.3.2 Nitrogen limitation trial on FWLF

The seed culture was prepared according to same procedure as described in 2.3.1. Additionally, the 250 mL shake flasks used in this experiment were prepared according to the 250 ml shake flasks containing 25 mL of cultivation medium were inoculated with 10 µl of cells suspension, which had been prepared by same procedure as 2.3.1. OD600 and pH was measured at different points throughout the cultivation.

Table 2. Additions made of each shake flask used to performed the nitrogen limitation trial on FWLF.

Component/Sample Shake flask 1 Shake flask 2 Shake flask 3

MOPS (400 mM) 5 mL 5 mL 5 mL FWLF 2.5 mL 2.5 mL 2.5 mL Saline 17.25 mL 17.5 mL 17.5 mL (NH4)SO4 (200 g L-1) 250 µl - - Trace - 10 µl - MgSO4 - 10 µl -

2.3.3 Reference, batch cultivation in bioreactor

E. coli cells from a glycerol stock at -80 °C were used to inoculate a sterile 5 L shake flask containing

(11)

2.3.4 Nitrogen limited and starved fed-batch cultivation on synthetic medium

From a glycerol stock stored at - 80 °C, E. coli cells harbouring plasmid pCnCAB were used to inoculate a pre-seed culture in a 100 ml shake flask containing 10 ml of MR-medium including 20 g L-1_{Glc and} 100 µg L-1 _{ampicillin. The cells were cultivated until they reached OD}₆₀₀_{of 1 at 30 °C and pH 7 on an} orbital shaker set to 180 rpm. The cultivated cells were then used to inoculate a seed culture in a 1000 ml baffled shake flask containing 100 ml of MR-medium, cultured in the same conditions until OD600 of 1. The seed culture was subsequently used to inoculate a 10L and 15L STR containing 4L and 7L MR-medium, respectively. The amount of Glc and ampicillin added corresponded to 20 g L-1_{and 100} µg mL-1_{, respectively}_{for each reactor. Inoculation of the STR was performed the day before to reach an} OD600 of 5 in the morning the next day, to be able to analyse batch, feed and starvation phase properly over one day. The DOT was maintained above 20 % during the cultivation by regulating the stirrer speed and gas flow into the reactor. Antifoam was added when required.

When nitrogen was depleted from the medium (indicated by a rapid increase of DOT) the feed was started. An exponentially increasing feed rate approach was applied, which is described by Equation 1.

Equation 1: "($) = "' )*+

The initial feed rate was calculated according to Equation 2: Equation 2: "'=_.* , -

//1 234

The initial feed rate, F0, was given by the specific growth rate, µ, the current cell concentration, x, total cultivation volume, V, yield of limiting substrate, Yx/s and limiting substrate concentration in feed, sin. The feed contained the following components (per liter): 245.6 g Glc, 66.2 g NH4Cl and 4.9 g MgSO4. To determine amounts of each component in the feed an elementary composition analysis of the E.

coli cell was used as starting point to estimate consumption at different cell concentrations. Regarding

(12)

these components. The pulse contained (per liter): 750 g Glc and 15 g MgSO4 and was applied in batch-, feed- and starvation phase when necessary. The Glc level was attempted to control between 10 and 20 g L-1_{. The Glc concentration was theoretically calculated at different time stages through the process} (assuming, Yx/s of 0.4 g cells/g Glc), and a correlation constant of 2.7 (between OD600 over DW) to approximate pulse additions online by measurements of OD600. Further, the Glc levels were estimated online with Glc measurement sticks (Uristix®, Siemens), to assure that Glc levels did not accumulate during the cultivation. As Glc levels decreased under 10 g L-1_{the pulse was manually triggered to} increase the Glc concentration with 10 g L -1_{. The exponential feed phase was applied for 3.5 h and the} starvation phase was applied for 5 h. 1M/3M NaOH was used for pH-titration in both cultivations. At initiated starvation 12 ml of 1M HCl was added to restore pH to 7. Samples for OD600, DW, Glc, ammonia, PHB and acetic acid were taken on regular intervals during the cultivation.

2.3.5 Fed batch cultivation on food waste liquid fraction (FWLF)

Same inoculation procedure as for the reference batch cultivation was applied to inoculate a 10L and 15L STR containing 5L and 8L of FWLF supplemented with 15 g/L Glc (same [Glc] used in shake flasks). Ampicillin corresponding to 100 µg mL-1_{was used in the shake flask and bioreactor. The pH,} temperature and dissolved oxygen tension (DOT) was maintained at 7, 30 °C and >5 % respectively, during the whole cultivation. The gassing and stirring was minimized during the cultivation and not increased before necessary to avoid cell lysis (a previously identified risk of FWLF cultivations). A sequential depletion of nitrogen sources from the food waste was indicated by small increases of the DOT. As cell growth almost ceased (indicated by a slow increase of DOT over time and confirmed with with OD600) the feeding was started. The feed consisted of FWLF supplemented with 15 g L-1 Glc and 5 g L-1_(NH₄₎₂_SO₄_{. An exponentially increasing feed rate approach was applied, which is described by} Equation 3. Equation 3 was applied instead of for the FWLF fedbatch as the limiting substrate concentration, sin, was unknown for FWLF.

Equation 3: 5($) = 5₀ )µ$

Equation 4: " $ = 5´ $ = µ 50 )µ$

(13)

2.4 Cell growth and PHB sampling

OD600 was used to monitor cell growth in all cultivations and was measured by diluting freshly taken cultivation samples in 0.9% w/w NaCl to an OD600 between 0.1-0.15 before measurement at 600 nm with a spectrometer (Genesys 20, Thermo scientific). Further, Dry weight (DW) samples of 5 ml each, were withdrawn in triplicate and added to dry and pre-weighed glass tubes and subsequently centrifuged at 2000 g for 10 min (HERMLE, Z 206 A). The pellet was washed once with 5 ml 0.9% w/w NaCl and centrifuged (2000 g for 10 min). The supernatant was discarded and the pellet was dried at 105 °C overnight. The dried pellets were allowed to cool to ambient temperature in a desiccator before being weighed. The same procedure was applied for PHB samples (withdrawn in duplicate) However, for this analysis glass tubes with sealed lids were used and subsequent to weighing stored in -20°C until analysis.

2.5 Sugar, ammonia and acetic acid analysis

Two methods were applied for sampling of Glc, ammonia and acetic acid. Either the sample was withdrawn directly from the STR or 2 ml were rapidly (>0.1 s) withdrawn from the reactor with an auto sampling device into a test tube. The test tube contained 2 ml cold (+4 °C), 0.13 M perchloric acid to stop metabolism [27]. The rapid sampling technique was applied for both the reference batch and the reference fed batch performed. The samples were then centrifuged at 2000 g for 10 minutes. 3.5 ml of the supernatant were collected and neutralized with 75 µl of saturated potassium carbonate (500 g/L). The samples were subsequently placed on ice to allow them to precipitate for 15 minutes and then centrifuged at 2000 g for 5 minutes. The supernatant was filtered (0.45 µm, VWR collection) and stored in -20 °C until analysis. The sugar and acetic acid concentration was determined using HPLC (Waters Alliance, UK) equipped with a Aminex HPX-87 H column (Bio-Rad, USA), using a mobile phase of 0.008 N H2SO4 and a flow rate of 0.5 ml/min at ambient temperature (20-22 °C). Detection was achieved using a refractive index detector (Waters Alliance, UK). Standards of Glc and HAc were applied to determine concentrations. For analysis of different sugars in FWLF, Aminex HPX-87 P column (Bio-Rad, USA) was applied, using a mobile phase of MilliQ H2O and a flow rate of 0.4 ml/min at 80 °C. The ammonia concentration was determined using the enzymatic kit, Megazyme Ammonia Kit Cat No. K-AMIAR.

2.6 Depolymerisation of PHB and gas chromatography analysis

(14)

(15)

3 Theory

3.1 Calculation of yields

The DW, true cell dry weight (TCDW) and PHB accumulation yields per Glc was assumed to be constant during the exponential growth phase. TCDW was assumed to be proportional to consumption of NH3. Thus, the DW, TCDW and PHB data was therefore fitted with Equation 5, in exponential growth phase by least square regression.

Equation 5: 9 $ = : )* ;

Equation 5 was inserted into Equation 6 (including values for constants, k and µ), and substrate consumption data was fitted in Equation 6, by least square regression to obtain YX/S, see the example in Figure 3.

Equation 6: < $ = = − ( ? @_.A B

/₁ )

Figure 3. Illustration of least square regression method applied by use of Equation 5 and Equation 6 to estimate C_{, 2}_{from ammonia consumption.}

3.2 Calculation of rates

The data obtained from OD600 and DW measurements were fitted with (4), during exponential growth in batch phase by least square regression to obtain the maximum growth rate, µmax.

The TCDW and PHB accumulation data was inserted into (7) in a defined interval, using [PHB]>0.3 g L-1_{to obtain the specific productivity rate, q}_PHB_{for batch and exponential feed phase.}

(16)

The volumetric rate for PHB accumulation is defined according to (8). The [PHB]start, was chosen to estimate volumetric productivity at approximately the same [PHB]. [PHB]start was defined as ≈0.3 g L -1_.

DEFG$HIJ 8: ]_^ = Δ[PHB] Δt 3.3 Assumptions in determination of [PHB]

To estimate [PHB] a two assumptions were made:

The yield of methyl-3-hydroxybutyrate from PHB was estimated to be the same for all reactions, see Figure 4. Also, the assumption was made that all methyl-3-hydroxybutyrate formed during the acidic methanolysis, would dissolve in the organic phase.

(17)

4 Results

The results obtained from experiments performed are presented in this chapter. Firstly, initial pre-treatment experiments were performed on FWLF in attempt to increase its nutritional content. Secondly, the initial cultivations performed with the E. coli, W3110, wild-type on FWLF and medium are presented. This includes initial shake flask experiments on FWLF and a batch trial on MS-medium. The production of PHB was then tested on MR-medium in fedbatch mode with E. coli, W3110, !fadR, pCnCAB. Finally, the data obtained from previous trials was utilized to perform a FWLF fedbatch for production of PHB.

4.1 Pre-treatment aimed at increase of Glc content in FWLF

To increase the Glc content of FWLF, it was heated for 3 h at 70 °C. The result is illustrated in Figure 5, which demonstrates that [Glc] increases from 3.85 g L-1_{to 4.75 g L}-1_{, during this additional} pre-treatment step. Further [Co-elute sugars] and [HAc] was measured. From initial HPLC data, the Aminex HPX-87 H column displayed occurrence of xylose. Upon further investigation, an Aminex HPX-87 P column was applied (adapted for biomass hydrolysate) to analyse the FWLF. The data obtained from this analysis demonstrated that no xylose was present, instead sugars as maltose, mannose and fructose were identified as food waste components (data not shown). These sugars are derived from different starch-based food stuff and fruit, which might be expected to be found in FWLF. Thus, it is concluded that the Aminex HPX-87 H column lacked the capability to properly separate some sugars, which results in a Co-elution. Further, a comparison of Glc, Co-elute sugars and HAc concentrations were made with pre-treated FWLF, which did not undergo the heating step. The final concentration of Glc, Co-elute sugars and HAc were 3.47 g L-1_{, 7.60 g L}-1_{and 0.64 g L}-1_{, respectively.} The final concentration of the pre-heated sample of Glc, Co-elute sugars and HAc were 4.94 g L-1_{, 8.05} g L-1_{, 0.63 g L}-1_{, respectively, see Figure 5. From the results it can further be concluded that the HAc} concentration was not affected by the pre-treatment and that [Glc] increased during this pre-treatment step.

Figure 5. The diagram illustrates the concentrations of Glc, Co-elute sugars and HAc during the different pre-treatment steps. The final filtration step, illustrates a comparison with a pretreated reference FWLF, which did not undergo the 70 °C heating and mixing (HM).

Mixed FW 70 °C H& M Autocla ve 121 ° C Final fi ltration step 0 1 2 3 4 5 6 7 8 9 10 Gl c, Su ga r C o-elu te , H Ac [g L -1] Glc Sugar Co-elute HAc Glc (No PT)

(18)

4.2 pH-dependent growth on FWLF

To make initial shake flask trials on FWLF possible the buffering capacity of FWLF needed to be improved, as HAc production lowers pH and thereby hampers E. coli growth. To combat this problem 3-(N-morpholino) propanesulfonic acid (MOPS) was added to FWLF, which improves the buffering capacity at near neutral pH. The results from an experiment comparing E. coli, W3110, wild type growth on unbuffered FWLF and buffered FWLF is demonstrated in Figure 6. The results clearly indicate that the unbuffered FWLF, decreases faster in pH (as a result of HAc production) and thereby grows to a lower OD600, than the buffered sample. As can be seen, the growth is almost halted at 5 h cultivation time. At that time the OD600 is 1.58 vs 3.15 for the unbuffered sample and buffered sample, respectively. To investigate the pH sensitivity of E. coli, W3110 further an addition of NaOH was made at 5 h cultivation time to restore pH to neutral. This resulted in resumption of growth, which resulted in a final maximum OD600 of 2.98 and 5.27 for the unbuffered and buffered sample, respectively, see Figure 6.

Figure 6. A comparison of FWLF with addition of 40 mM MOPS (Buffered FWLF) and pure FWLF (Unbuffered FWLF) growth. The OD600 was and pH was measured. Additon of NaOH was made at 5 h to restore pH to neutral.

4.3 Nitrogen limitation trial on FWLF

PHB accumulation to a very high cell content inside cells is often linked to limitation or starvation of certain macro nutrients for natural PHB producers. To investigate, which nutrient was limiting on FWLF, a limitation trial was performed. The FWLF was suspected to be limiting on nitrogen, which concluded the following approach with shake flasks containing MOPS and 10x diluted FWLF: 1. (NH4)2SO4 supplementation

2. MgSO4 + Trace elements supplementation

(19)

The result can be seen in Figure 7, which clearly demonstrates that the FWLF shake flask supplemented with (NH4)2SO4 grew to the highest OD600 of 5.25. The shake flask without supplement and the MgSO4 + Trace supplemented shake flaks reached both an OD600 of 2.25. Continuous measurements of pH concluded that the pH did not decrease below a pH of 6.5 (data not shown). Hence, this experiment concludes that FWLF supplemented with (NH4)2SO4 supported the highest cell concentrations, which supports the hypothesis of FWLF being nitrogen limited.

Figure 7. Cultivation growth curves for shake flasks containing 10x FWLF with 40 mM of MOPS with different nutrient supplementations. 1. (NH4)2SO4 supplementation 2. MgSO4 + Trace elements supplementation 3. No additions. As can be seen that (NH4)2SO4 supplementation supported the highest cell concentration.

4.4 W3110 reference batch

The reference batch on W3110, wild type was performed to obtain reference data for the E. coli to be used in production of PHB in upcoming experiments. The reference batch resulted in a µmax of 0.67 h-1 and YDW/Glc of 0.475 g g-1, see Figure 8.

Figure 8. Diagram of reference batch performed on MS-medium with E. coli, W3110, wild type. Diagram shows OD600, DW, [Glc] and [HAc] over time.

(20)

4.5 Early shake flask experiments on W3110, pCnCAB in synthetic medium

From shake flask cultures (results not shown), the maximum growth rate (µmax) was measured to 0.30 h-1_{by OD measurements in MR-medium. The µ}

max was measured at 30 °C and starting pH of 7.0. Initially, trials were performed in 37 °C. It was noticed that cultivations in 37 °C severely affected the plasmid stability and resulted in a rapid loss of plasmid from cells. This was realized as inoculation from shake flask to shake flask containing fresh ampicillin always failed on inoculations <5 % of total cultivation volume and hence resulted in no growth. A hypothesis is the simultaneous inoculation of β-lactamase upon larger inoculation volumes (<10 % of total cultivation volume), which thereby degrades the ampicillin allowing E. coli cells without plasmid to grow.

4.6 Reference fed-batch culture limited and starved on nitrogen source (on MR-medium)

To investigate the PHB production ability of E. coli, W3110, pCnCAB, in 30 °C a nitrogen limited and starved fed-batch was performed. From experiment performed in 4.3, it is known that the food waste liquid fraction is limited on nitrogen sources and therefore it was of our interest to investigate similar conditions on a defined minimal salts medium to see if it had any effect on the PHB accumulation rate and cell content of PHB. The purpose of this investigation is that some natural PHB producers can accumulate PHB to a very high cell content during nitrogen limitation or starvation. However, it must be emphasized that previous data on recombinant E. coli expressing PHB did not suggest that the cell content of PHB would increase significantly during nitrogen limited or starved conditions [2]. Still, this was not confirmed for the plasmid and strain applied in this study and therefore the main approach we chose for further investigation.

The results can be seen in Figure 9A & B. A µmax of 0.29 h-1 was obtained by measuring OD600 during exponential growth in batch phase in both cultivations. The cultivations resulted in a final DW and [PHB] of 18.7 g L-1_{vs 14.9 g L}-1_{and 5.7 g L}-1_{vs 3.3 g L}-1_{for cultivation A and B, respectively. During} the feed phase the growth was limited to 0.2 h-1_{by an exponential feed to investigate the effect on PHB} production during nitrogen limitation. As can be seen in Figure 9, the approach was successful in cultivation B, but not for A. Cultivation A, failed due to overfeeding, which resulted in accumulation of both Glc and ammonia to very high concentrations. However, the results were included in this report as comparison between a batch cultivation with a short nitrogen starvation phase (cultivation A) and a nitrogen limited and starved fedbatch (cultivation B). Pulse additions were made at 3.1 h, 5.4 h, see Figure 9A for cultivation A and at 16.7 h, 18.7 h for cultivation B, see Figure 9B.

(21)

resulting in an increased PHB cell content over time during limitation or starvation. This is not the case for the strain applied in this study. Interestingly enough, both Glc and HAc concentrations are decreasing during starvation indicating consumption of both carbon sources. The decrease of HAc is coupled to a sudden increase of pH from 7 to 7.25 (data not shown) a short time after nitrogen starvation is initiated (note, Glc is still in excess). Further, the CO2 levels of outgas air were stable during this phase with a very slow decrease over time. It seems therefore, that most carbon is consumed for maintenance of cells (thus the CO2 production), instead of PHB production.

A yield from DW per Glc obtained from HPLC data and DW measurements resulted in YDW/Glc of 0.45 g g-1_{. From PHB data, values for both Y}

TCDW/Glc and YPHB/Glc were obtained, which were calculated to 0.36 g g-1 _{and 0.09 g g}-1_{, respectively. Y}_PHB/Glc_{of 0.09 g g}-1_{corresponds to 28% of the maximum theoretical} yield of PHB from Glc, which is 0.32 g g-1_{based on biochemical stoichiometry analysis [29]. The lower} yield, YTCDW/Glc of 0.36 g g-1 vs YDW/Glc of 0.45 g g-1, can be explained by the partial carbon flux to PHB from the total DW yield, YDW/Glc of 0.45 g g-1. For cultivation A there was not enough data on Glc consumption to make an estimation of yields. The ammonia (NH3) consumption pattern for both cultivation A and B resulted in an average yield, YTCDW/NH3 of 6 g g-1 (5.8 g g-1 vs 6.2 g g-1, for A and B respectively). This corresponds to an average nitrogen content of 13.8 wt%, which correlates well with values found in literature.

Comparing the average volumetric productivity of PHB, rPHB, in batch phase both cultivations, resulted in a similar rate (270 vs 240 mg L-1 _h-1_,_{for A and B respectively). However, r}_PHB_{, was lower over the} whole cultivation for cultivation B due to nitrogen limitation and starvation phase, which restricted the growth and thereby the PHB accumulation, as earlier mentioned (note, cultivation A was only starved shortly on nitrogen-source after a longer batch phase), see Figure 9A & B. It resulted in a rPHB of 360 vs 220 mg L-1 _h-1_,_{for A and B respectively (in calculations only [PHB]>0.3 g L}-1_{was used, as the} relative error was determined to be significant for [PHB]<0.3 g L-1_{). The specific productivity rate, q}_PHB_, over batch phase was 91 mg g-1_h-1_{and 83 mg g}-1_h-1_{for cultivation A and B, respectively. It must be} emphasized that the specific productivity is calculated from TCDW and not DW, which is usually the case for other product expression systems. This approach was chosen to be more suitable as the product weight is a significant fraction of the DW. As TCDW is applied the productivity of each true cell is accounted for.

(22)

Figure 9. Reference fed-batch cultivations of W3110 , pCnCAB with limitation and starvation of ammonium. Measurements were conducted on OD600,DW , PHB, Glc, acetic acid and ammonium.

4.7 Fed- batch cultivation on food waste

To examine the PHB production from FWLF, a fed batch cultivation was performed applying W3110, pCnCAB. From previous experiments performed, the wild-type W3110 has successfully been cultivated to cell concentrations of 7.5 g L-1_{in batch mode on FWLF with (NH}₄₎₂_SO₄_{and Glc} supplementation. To investigate W3310, !fadR, pCnCAB, performance on FWLF, the experiment was developed further by addition of a FWLF feed supplemented with Glc and (NH4)2SO4.

The results from the fedbatch on FWLF can be seen in Figure 10. Due to severe cell lysis in the 10L STR, which resulted in termination of growth in mid batch phase, only the results obtained from the

(23)

of cell lysis seems to be coupled with bubble formation and foaming. At higher stirring speeds, more bubbles are formed and thereby an increased tendency of cell lysis has been observed. The reason behind the cell lysis is still unclear and further investigation is required.

However, the 15L STR cultivation evaded the sudden cell lysis and reached a highest DW of 7.11 g L-1 and 3.5 g L-1_{PHB, with a final PHB cell content of 50%. The average µ}_max_{obtained from exponential} growth in batch phase corresponded to 0.47 h-1_{(0.48 h}-1_{measured by OD}

600 vs 0.46 h-1 measured by DW), see Figure 10. The productivity during exponential growth in batch phase resulted in an average rPHB of 270 mg L-1 h-1. The whole cultivation, including the long nitrogen starvation phase performed overnight, resulted in a rPHB of 180 mg L-1 h-1. The specific productivity rate, qPHB, in batch phase was 318 mg g-1_h-1_{(calculated from [PHB]>0.3 g L}-1_).

(24)

Figure 10. FWLF fedbatch cultivations of W3110, !fadR, pCnCAB performed on MR-medium with limitation and starvation of nitrogen source. Measurements were conducted on OD600, DW and PHB.

B

5 10 15 20 25 30 0 2 4 6 8 CDW [g L -1] µmax=0.46 h-1 µ=0.09 h-1

A

5 10 15 20 25 30 0 10 20 OD 600 µmax=0.48 h-1

C

5 10 15 20 25 30 0 1 2 3 4 25 50 75 100 PH B , Tr ue C D W [ g L -1] PHB concentration PHB content Ce ll c ont ent P HB [% ]

True cell mass

Time [h]

Batch Exponential

(25)

5 Discussion

This chapter is aimed at discussing the results obtained mainly from the reference fedbatch performed on synthetic medium and the fedbtach performed on FWLF supplemented with Glc and NH4SO4. In Table 3, all values obtained from the experiments performed are summarized, for a more straightforward comparison.

5.1 Growth rate dependency

From the experiments performed, the growth rate of E. coli, demonstrated dependency on pH, temperature and cultivation medium. From early shake flask experiments on FWLF, the growth rate is clearly dependent on pH of the medium, see 4.2. pH<6 severely inhibits growth and pH<5 completely hampers the growth of cells. The pH-controll was mainly an issue during shake flask cultivation on FWLF due to poor buffering capacity of the medium. The issue was resolved in STR cultivations by pH titration or by addition of 80 mM of MOPS in shake flask cultivations.

Comparing the µmax obtained from the reference batch, the shake flask trials and reference fedbatch, the growth rate was 0.67 h-1_{(W3110, wild type, 37 °C), 0.30 h}-1_{(W3110, pCnCAB, 30 °C) and 0.29 h}-1 (W3110, pCnCAB, 30 °C), respectively. The significantly lower µmax obtained at 30 °C is probably a result of lower enzyme activity in the cell. Enzymatic reactions are temperature dependent and can commonly be described by Arrhenius equation. A similar expression to Arrhenius equation can similarly describe the dependence of specific growth rate, µ on temperature. In addition to lower temperature, the additional carbon flux to PHB might have limited the carbon available for production of new cells. These factors together result in a reduced growth rate. To streamline the process, the plasmid stability at 37 °C must be resolved. If the plasmid however were to be stable at 37 °C, the volumetric productivity might be significantly increased.

The growth rate demonstrated dependency on cultivation medium also. The reference fedbatch (synthetic minimal salts medium) and FWLF fedbatch obtained a µmax of 0.29 h-1 and 0.47 h-1, respectively, both at 30 °C. The same trend was recognized for W3110, wild type, when cultivating on both MS-medium and FWLF (data not shown). The increase in µmax was suspected to be an indication of amino acid consumption during initial growth on FWLF, which gives FWLF a more complex medium character, similar to yeast extract. The higher growth rate is a result of major intermediary precursors e.g. amino acids, cofactors and vitamins being readily available for the cells to incorporate into their metabolism. Thereby, the biosynthetic requirements are much lower in these media, which allows E. coli cells to grow faster.

5.2 Yields, volumetric productivity rates and specific productivity rates

(26)

calculated yields. In addition, the YDW/Glc from the FWLF fedbatch, should in theory be higher than the previous yields measured on MS/MR-medium. That is, if our hypothesis is correct regarding occurrence of complex nitrogen sources such as amino acids in the food waste. The explanation can be found in the concurrent uptake of carbon within these complex nitrogen sources in addition to the cells normal carbon uptake. Thereby more DW is produced per gram Glc, which results in a higher yield. Unfortunately, due to technical problems with the HPLC, this hypothesis was left unconfirmed. Interestingly enough, no Glc was supplemented to the 10x diluted FWLF in 4.3, which nevertheless reached significant cell concentrations. Thus, other fermentable carbon sources must have been available as previous results demonstrate in 4.1. Still, for 10x diluted food waste, which probably contains ~0.5 g L-1_{Glc, an OD}₆₀₀_{of 1 would be difficult to reach assumuming an Y}_x/s_{of 0.475 g g}-1_{, see} 4.4 and a conversion factor between OD600 and DW of 2.7. The results suggest that other fermentable sugars are present in significant amount to enable growth to an OD600 of 5.25, which was reached for the (NH4)2SO4 supplemented shake flask.

During exponential growth in batch phase the volumetric productivities, rPHB, were similar to both fedbatch cultivations on synthetic and FWLF medium, between 240-270 mg L-1 _h-1_{(calculated from} approximately same [PHB]start). The total cultivation, rPHB, was highest (360 mg L-1 h-1) for cultivation A on synthetic medium and lowest (180 mg L-1 _h-1_{) for the fedbatch on FWLF. The explanation for the} low rPHB, can be found in the 10 h long nitrogen starvation phase for the FWLF fedbatch, which lowered the rPHB significantly. Also, the fact that cultivation A and B reached higher [PHB] and TCDW is considered. By excluding the starvation phase from the FWLF fedbatch the rPHB obtained was 232 mg L-1 _h-1_{similar to cultivation B’s total r}_PHB_{of 220 mg L}-1 _h-1_{but lower than cultivation A’s r}_PHB_{of 360 mg} L-1 _h-1_{. This indicates that the PHB accumulation and cell growth for the FWLF fedbatch was impeded} most probably by some kind of nutrient limitation/depletion. In either way it can be concluded that nitrogen starvation is not suitable for W3110, pCnCAB for such long period of time.

Comparing the specific productivities, qPHB in batch phase of the cultivation performed, it is obvious that the FWLF fedbatch produces more PHB per cell, which is also reflected by the high final PHB cell content obtained, see Table 3 and Figure 10C. The feed phase for both cultivation B and FWLF fedbatch displayed increased specific productivity rates 13o mg g-1_h-1_{vs 923 mg g}-1_h-1_{compared to 83 mg g}-1_h -1_{vs 318 mg g}-1_h-1_{, respectively. Note that q}

PHB did only increased slightly for the second batch phase of cultivation A. This is interesting, as a reduced growth rate during nitrogen limitation while keeping Glc in excess seems to result in an increased qPHB.

(27)

content occurred alongside the initial suspected nitrogen source shifts and thus it seems as if limitation of certain amino acids might be beneficial, if a high PHB cell content is desired.

However, the specific productivity, qPHB for the FWLF fedbatch is many times higher than for the fedbatch on synthetic medium. The underlying reason is yet unknown but the availability of acetyl-CoA might offer an explanation to why amino acid consumption is beneficial for PHB production. The biosynthesis of PHA is enabled through the expression of three consecutive enzymes, which are dependent on the flux of acetyl-CoA as substrate. Thereby, the PHB biosynthesis is coupled to availability of acetyl-CoA. As major intermediary precursors are readily available in a complex medium such as FWLF, the excess acetyl-CoA from the glycolysis can be utilized for PHB production. In a defined minimal salt medium, the [acetyl-CoA] is lower because it is used as a precursor in other biosynthetic pathways, restricting the flux of acetyl-CoA to PHB. Another alternative explanation might be poor regeneration of Nicotinamide adenine dinucleotide phosphate (NADPH) when cells are grown on defined media. NADPH is important because it functions as cofactor for acetoacetyl-CoA reductase (second enzyme in PHB biosynthesis pathway), and depletion of it will halt PHB synthesis. Further, in a defined minimal salt medium, the TCA-cycle is fully active during cell growth. The first enzyme of the TCA cycle, citrate synthase, uses acetyl-CoA but also produces free CoA. Free CoA is an inhibitor of _-ketothiolase (the first enzyme of PHB biosynthesis pathway), thereby inhibiting PHB production. The above-mentioned theories could explain why PHB production on a defined minimal salts medium results in a reduced qPHB compared to a complex medium cultivation. However, this has not been confirmed and further studies are necessary to acquire a better understanding of the PHB biosynthesis pathway.

Another option, not coupled to the PHB biosynthesis pathway is cell lysis. From Figure 10C, the last data point resulted in a PHB cell content of 67% from 50% in just 2 h. The 10 h overnight nitrogen starvation on the other hand resulted in only 3% increase of PHB cell content. Previous to measurement of the last data point, severe foaming was observed in the reactor, which was suspected to be caused by cell lysis. This most probably caused the major loss of true cell mass while PHB concentration remained constant. The reason that the lysed cells did not account for the DW is that the fragment were probably to light to segment during centrifugation while the PHB granules probably did. Thus, this resulted in the higher PHB content in the last data point. If cell lysis on the other hand occurred continuously during the cultivation, this would also result in a higher PHB content per cell. In addition the specific productivity of PHB would also be higher. As cell lysis was a pronounced risk with FWLF cultivations, and one of the duplicate cultivations was terminated due to cell lysis, this theory must be emphasized and reflected over properly. A batch cultivation on complex medium e.g. LB should have been made to investigate PHB cell content on complex medium. This would probably have demonstrated if the PHB cell content obtained in FWLF was feasible.

5.3 Valorisation options for food waste

(28)

(29)

Table 3. List of all data obtained from experiments. * A and B refers to cultivation A and B in Figure 9.

Reference Batch, W3110, Wild Type

Growth rate Value Unit

µmax (37 °C) 0.67 h-1

Yields Value Unit

YDW/Glc 0.475 g g-1

Shake flask, W3110, pCnCAB

µmax (30 °C) 0.3 h-1

Reference Fedbatch on Synthetic Medium

µmax (30 °C) 0.29 h-1

Yields Value Unit

YDW/Glc YTCDW/Glc YPHB/Glc YTCDW/NH3 0.45 0.36 0.09 6 g g-1 g g-1 g g-1 g g-1

Volumetric Productivity Value Unit

rPHB, Batch (A)* rPHB, Batch (B)* rPHB, Total (A)* rPHB, Total (B)* 270 240 360 220 mg L-1 _h-1 mg L-1 _h-1 mg L-1 _h-1 mg L-1 _h-1

Specific Productivity Value Unit

qPHB, Batch (A) qPHB, Batch (B) qPHB, Exponential Feed (B) 91 83 130 mg g-1_h-1 mg g-1_h-1 mg g-1_h-1 Fedbatch on FWLF

Growth Rate Value Unit

µmax (30 °C) 0.47 h-1

Volumetric Productivity Value Unit

rPHB, Batch rPHB, Total 270 180 mg L-1_h-1 mg L-1_h-1

Specific Productivity Value Unit

(30)

6 Conclusion

(31)

Acknowledgements

(32)

References

1. Regeringskansliet. Förlängda statsstödsgodkännanden för skattebefrielse av biodrivmedel. 2015 [cited 2017 21st of June]; Available from:

http://www.regeringen.se/pressmeddelanden/2015/12/forlangda-statsstodsgodkannanden-for-skattebefrielse-av-biodrivmedel/.

2. Lee, S.Y., H.N. Chang, and Y.K. Chang, Production of Poly (β‐Hydroxybutyric Acid) by

Recombinant Escherichia colia. Annals of the New York Academy of Sciences, 1994. 721(1):

p. 43-52.

3. IPCC special report on renewable energy sources and climate change mitigation , 2011.

4. Bellard, C., et al., Impacts of climate change on the future of biodiversity. Ecology Letters, 2012. 15(4): p. 365-377.

5. Oehlmann, J., et al., A critical analysis of the biological impacts of plasticizers on wildlife. Philosophical Transactions of the Royal Society B: Biological Sciences, 2009. 364(1526): p. 2047-2062.

6. FAO, I., WFP (2015), The State of Food Insecurity in the World 2015. Meeting the 2015

international hunger targets: taking stock of uneven progress. Food and Agriculture

Organization Publications, Rome.

7. Turon, X., et al., Food Waste and Byproduct Valorization through Bio-processing:

Opportunities and Challenges. BioResources, 2014. 9(4): p. 5774-5777.

8. Pleissner, D. and C.S.K. Lin, Valorisation of food waste in biotechnological processes. Sustainable chemical processes, 2013. 1(1): p. 21.

9. Food waste volumes in Sweden. 2013, Naturvårdsverket, Swedish Environmental Protection

Agency. p. 20.

10. Tuck, C.O., et al., Valorization of biomass: deriving more value from waste. Science, 2012.

337(6095): p. 695-699.

11. Bernstad, A. and J. la Cour Jansen, A life cycle approach to the management of household food

waste – A Swedish full-scale case study. Waste Management, 2011. 31(8): p. 1879-1896.

12. Leal Filho, W. and M. Kovaleva, Food waste and sustainable food waste management in the

Baltic sea region. 2015: Springer.

13. Møller, J., A. Boldrin, and T.H. Christensen, Anaerobic digestion and digestate use:

accounting of greenhouse gases and global warming contribution. Waste management &

research, 2009. 27(8): p. 813-824.

14. Yasin, N.H.M., et al., Food waste and food processing waste for biohydrogen production: A

review. Journal of Environmental Management, 2013. 130: p. 375-385.

15. Lin, C.S.K., et al., Food waste as a valuable resource for the production of chemicals, materials

and fuels. Current situation and global perspective. Energy & Environmental Science, 2013. 6(2): p. 426-464.

16. Lee, S.Y., et al., Comparison of recombinant Escherichia coli strains for synthesis and

accumulation of poly‐(3‐hydroxybutyric acid) and morphological changes. Biotechnology and

(33)

17. Wei, Y.-H., et al., Biodegradable and Biocompatible Biomaterial, Polyhydroxybutyrate,

Produced by an Indigenous Vibrio sp. BM-1 Isolated from Marine Environment. Marine

Drugs, 2011. 9(4): p. 615-624.

18. Steinbüchel, A. and T. Lütke-Eversloh, Metabolic engineering and pathway construction for

biotechnological production of relevant polyhydroxyalkanoates in microorganisms.

Biochemical Engineering Journal, 2003. 16(2): p. 81-96.

19. Wallen, L.L. and W.K. Rohwedder, Poly-. beta.-hydroxyalkanoate from activated sludge. Environmental science & technology, 1974. 8(6): p. 576-579.

20. Lee, S.Y., Bacterial polyhydroxyalkanoates. Biotechnology and bioengineering, 1996. 49(1): p. 1.

21. Bengtsson, S., et al., Production of polyhydroxyalkanoates by activated sludge treating a paper

mill wastewater. Bioresource technology, 2008. 99(3): p. 509-516.

22. Bugnicourt, E., et al., Polyhydroxyalkanoate (PHA): Review of synthesis, characteristics,

processing and potential applications in packaging. 2014.

23. Steinbüchel, A. and H.E. Valentin, Diversity of bacterial polyhydroxyalkanoic acids. FEMS Microbiology Letters, 1995. 128(3): p. 219-228.

24. Madison, L.L. and G.W. Huisman, Metabolic engineering of poly (3-hydroxyalkanoates): from

DNA to plastic. Microbiology and molecular biology reviews, 1999. 63(1): p. 21-53.

25. Sudesh, K., Polyhydroxyalkanoates from Palm Oil Biodegradable Plastics. SpringerBriefs in Microbiology. 2012, Dordrecht: Dordrecht : Springer.

26. Yang, T.H., et al., Biosynthesis of polylactic acid and its copolymers using evolved propionate

CoA transferase and PHA synthase. Biotechnology and bioengineering, 2010. 105(1): p.

150-160.

27. Larsson, G. and M. Törnkvist, Rapid sampling, cell inactivation and evaluation of low

extracellular glucose concentrations during fed-batch cultivation. Journal of biotechnology,

1996. 49(1-3): p. 69-82.

28. Braunegg, G., B. Sonnleitner, and R. Lafferty, A rapid gas chromatographic method for the

determination of poly-β-hydroxybutyric acid in microbial biomass. Applied Microbiology and

Biotechnology, 1978. 6(1): p. 29-37.

29. Yamane, T., Yield of poly‐D (‐)‐3‐hydroxybutyrate from various carbon sources: A theoretical

study. Biotechnology and bioengineering, 1993. 41(1): p. 165-170.

30. Rosander, E., M. Svedendahl Humble, and A. Veide, Municipal Solid Waste as Carbon and