Renewable and Sustainable Energy Reviews 166 (2022) 112687

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

journal homepage: www.elsevier.com/locate/rser

Unraveling PHA production from urban organic waste with purple

phototrophic bacteria via organic overload
Luis D. Allegue, Maria Ventura, Juan A. Melero, Daniel Puyol *
Group of Chemical and Environmental Engineering, Universidad Rey Juan Carlos, 28933 Mostoles, Madrid, Spain

A R T I C L E I N F O A B S T R A C T

Keywords: The production of polyhydroxyalkanoates (PHA) with purple phototrophic bacteria (PPB) has been limited due
Purple phototrophic bacteria to low yields and limited knowledge regarding the diverse routes used for carbon biosynthesis. The present study
Polyhydroxyalkanoates increases PHA accumulation yields using urban organic waste pretreated by steam explosion and acidogenic
Glycogen
fermentation as substrate. Throughout the PPB-based photoheterotrophic process in an anaerobic membrane
Hydrogen
The organic fraction of the municipal solid
photobioreactor, the organic loading rate (OLR) was modified to increase the amount of PHA and biomass in the
1
waste reactor. A maximum PHA accumulation of 42% (gPHA g−Biomass ) on a dry basis was achieved and maintained for
10 d for an OLR of 1 gCOD L− 1 d− 1, and hydraulic and sludge retention times of 2 and 6 d, respectively. This PHA
accumulation capacity is the maximum obtained using a mixed culture of PPB fed with waste. Also, a medium-
chain PHA (polyhydroxyhexanoate) has been quantified, enhancing the physicochemical properties and diver­
sifying their industrial applications. Furthermore, we show novel alternatives to PHA accumulation: carbon
storage as glycogen and extracellular polymers while deriving the excess electrons into hydrogen. Finally, a
statistical study of microbial communities has settled the environmental variables with the most significant
influence on these communities’ variability. This work demonstrates the importance of acquiring a thorough
understanding of carbon accumulation and electron allocation strategies of PPB under stressful conditions and
shows promising results for a larger scale implementation of a PPB-based photobiorefinery, which could valorize
urban organic waste to produce different high added-value products within the context of the circular
bioeconomy.

production.
1. Introduction Plastics derived from petroleum-based industries have revolution­
ized modern life since their commercialization due to their low cost,
The management of MSW challenges the EU as a critical element to durability, and good mechanical and thermal properties [5]. However,
transition to a circular economy. The EU 27 has produced 505 kg of they can remain in the environment for more than a century without
municipal waste per capita in 2020 [1]. However, only 30% of MSW is decomposing, and their recycling is a complicated process. Therefore,
recycled, far from the EU target of 55% by weight settled by 2025. In the more sustainable alternatives must be sought, compelling the transition
Waste Framework Directive and the Landfill Directive, high emphasis is to the bioeconomy and circular economy. Bioplastics have emerged as a
placed on the organic fraction of municipal solid waste (OFMSW), a key sustainable alternative in this context, as they can be produced from
element in planning the sustainable MSW management system [2]. renewable resources, are biodegradable, and have similar thermal and
OFMSW must be collected separately from 2023 onwards, a major step mechanical properties to petroleum-based plastics [6].
in establishing a circular bioeconomy in the EU [3]. The EU has defined Polyhydroxyalkanoates (PHA) are a family of environmentally sus­
a treatment hierarchy that prioritizes the valorization of organic wastes tainable polyesters produced by a wide variety of prokaryotes [7]. PHA
into biomaterials rather than compost or energy, thus defining the scope are synthesized in excess of carbon and limitation of a major nutrient, or
of the biorefineries for OFMSW [4]. Indeed, the fundamental contribu­ in situations of bacterial stress, in addition to acting as carbon and
tion of biorefineries to the concept of circular economy is the ‘pyramidal reducing power sink [8]. More than 150 monomers of PHA with a wide
approach,’ where the extraction and production of an extensive port­ range of structures are known, and PHA can be divided based on the size
folio of high added-value products have greater priority than bioenergy of their monomers into short-chain PHAs, consisting of monomers of 2–5

* Corresponding author. Universidad Rey Juan Carlos, Calle Tulipán, s/n, 28933 Móstoles, Madrid, Spain.
E-mail address: [email protected] (D. Puyol).

https://doi.org/10.1016/j.rser.2022.112687
Received 15 February 2022; Received in revised form 29 April 2022; Accepted 5 June 2022
Available online 10 June 2022
1364-0321/© 2022 Elsevier Ltd. All rights reserved.
L.D. Allegue et al. Renewable and Sustainable Energy Reviews 166 (2022) 112687

Several PPB species are strong PHA accumulators, most of them bearing
Abbreviations to the families Chromatiaceae, Rhodospirillaceae, Rhodobacteraceae,
and Bradyrhizobiaceae. However, the organic loading rate (OLR)
Biochemical methane potential BMP strongly conditions the community composition, where Rhodop­
Chemical Oxygen Demand COD seudomonas, Rhodobacter, Rhizobium, and some representatives from the
Extracellular polymers EPS Hyphomicrobiaceae family stand out at high OLR [26]. Still, a study that
Hydraulic retention time HTR examines the effect of operational factors over community transitions
Membrane photobioreactor MPBR and how they affect PHA production and elucidates the relationship
Organic fraction of municipal solid waste OFMSW between electron and carbon sinks in PPB metabolism is lacking.
Organic loading rate OLR The production of PHA by PPB enriched mixed cultures has been
Polyhydroxybutyrate PHB studied primarily on synthetic substrates and ideal illumination condi­
Polyhydroxyvalerate PHV tions (light intensities higher than 40 W m− 2) [21]. Our group has
Polyhydroxyalkanoate PHA studied PHA production using hydrolysates from the OFMSW and
Polyhydroxyheaxanoate PHH lignocellulosic wastes, reaching PHA accumulations of 5% [27] and 21%
Poly(hydroxybutyrate-co-hydroxyvalerate) PHBV [28], respectively. Several works have also studied the PHA accumula­
Purple phototrophic bacteria PPB tion using acidogenic fermentation effluents as substrate, achieving
Total Kjeldahl nitrogen TKN accumulations of 24% [24] and 31% [29]. A previous study analyzed the
Total phosphorus TP effect of transient OLR derived from a mixture of domestic wastewater
and fermented molasses on the performance of a sequencing batch
reactor in light sufficiency [28]. The OLR tested in that study varied
from 0.02 to 0.74 gCOD L− 1 d− 1, and the substrate concentration inside
carbon atoms, and medium-chain PHAs, containing 6 to 14 carbon the reactor was always below 0.2 gCOD L− 1 and was rapidly consumed
atoms [6]. The polyhydroxybutyrate (PHB) is the most studied polymer during the feeding sequence. Interestingly, the maximum PHA accu­
so far, being its thermomechanical properties very similar to those of mulation obtained in that study (around 700 mgCOD L− 1) coincided
polypropylene [9] but more stiff and brittle [10]. On the other hand, the with the period where the excess of the substrate was longer maintained
PHB and polyhydroxyvalerate (PHV) copolymer, called poly (hydrox­ (around 7 h). Thereby, maintaining an excess of substrate on the culture
ybutyrate-co-hydroxyvalerate), commonly known as PHBV, is less stiff seems to favor PHA accumulation. However, the analysis of how sub­
and brittle than the PHB monomer but retains most other mechanical strate overloading affects the overall performance of a photobioreactor
properties. In general, and depending on their composition, copolymers and how much substrate concentration a mixed culture can tolerate is
decrease the degree of crystallinity and melting temperature and in­ lacking, and they are critical questions that must be solved to allow
crease the extension to break [11]. Still, in particular, the combination scale-up.
of short-chain PHAs (PHB and PHV) with medium-chain PHAs such as PPB mainly obtains all the necessary energy through anoxygenic
polyhydroxyhexanoate (PHH) give polymers elastic properties that in­ photosynthesis by absorbing IR light through its photosystem [30]. This
crease their value in the industry as they can be used in different ap­ energy is most commonly used for photoheterotrophic growth on SCCA
plications [12]. However, the production cost of PHA can be 3–4 times and sugars and the accumulation of carbon in PHA and phosphorus in
more expensive than conventional fossil plastics [13,14]. Pure cultures polyphosphates, which can also serve as energy reserves or electron
typically produce commercial PHA using synthetic substrates, which are sinks for the regulation of intracellular redox reactions [31]. In addition,
responsible for 30–50% of production costs [15]. Mixed microbial cul­ they can also accumulate carbon to glycogen [32] and some extracel­
tures fed with organic and biodegradable wastes have been thoroughly lular polymeric substances (EPS), such as alginate [33], and can produce
investigated to reduce these costs [16]. These wastes are usually hydrogen via nitrogenase as well [34]. Furthermore, it is known that
accommodated by pretreatments depending on their complexity and competition between hydrogen production and PHA accumulation can
later treated by acidogenic fermentation, where short-chain carboxylic occur in some PPB species, as both functions constitute a way to dissi­
acids (SCCA), ideal precursors of PHA, are produced [17]. pate the excess reducing power [35]. Still, the distribution of carbon in
Among the potential feedstock, the organic fraction of municipal enriched mixed cultures between the different polymers that PPBs can
solid waste (OFMSW) is one of the most suitable options due to its high accumulate (PHA, glycogen, and alginate) and the energy distribution
availability and low cost. Previous studies have shown that thermal between the different electron sinks (Hydrogen and PHA) needs dedi­
pretreatments such as steam explosion solubilize a substantial amount of cated research.
carbon, reduce the final quantity of residues and increase the rates and Hence, this study aims to optimize the PHA accumulation in a PPB
productivities in fermentation processes [18]. The coupling of the enriched mixed microbial culture. The main research questions are how
acidogenic fermentation process and PHA production has already been the organic overloading affects the PHA production and the competence
tested at a pilot scale in several studies using simpler substrates such as for carbon and electron allocation, and how this overloading affects the
the solid phase of squeeze OFMSW plus biological sludge [19] or fruit microbial community transitions in a dynamic process. Furthermore,
waste [20], obtaining high fermentation yields (0.74 gCOD gCOD− 1). this study has been carried out in a novel membrane photobioreactor
However, these studies have been carried out with aerobic mixed mi­ (MPBR) using for the first time an OFMSW eluent pretreated by steam
crobial cultures, which, in addition to the aeration costs, require the explosion and fermentation as substrate. Finally, the holistic, organic
uncoupling of the famine and feast phases in two reactors. An exciting waste treatment was studied to understand the overall impact of this
strategy to reduce costs is using light-dependent microorganisms like multi-strategy approach for OFMSW valorization via a PPB-based
purple phototrophic bacteria (PPB), as they can grow under anaerobic biorefinery.
conditions, allowing easy enrichment without aeration. The PPBs also
1
allow PHA carbon accumulation ratios of up to 0.96 CmolPHB Cmol−Hac 2. Material and methods
[21], and unlike most chemoheterotrophs, PPB can grow and accumu­
late PHA simultaneously [22]. Thereby, PPB can perform PHA accu­ 2.1. OFMSW and inoculums
mulation in a continuous feeding regime, eliminating the costly and
non-productive periods associated with the famine phases used in aer­ The OFMSW was obtained and previously sorted by the solid waste
obic systems [23]. Another strategy studied has been the combined treatment plant Centro las Dehesas (Madrid). The organic fraction was
production of PHA and hydrogen in several publications [24,25]. separated at the collection point. The macro-characteristics of the

2
L.D. Allegue et al. Renewable and Sustainable Energy Reviews 166 (2022) 112687

organic fraction of municipal solid waste are 135 gVS Kg− 1 with a ratio reducing it with Na2S + 9H2O (1 mL per liter). No more than 0.5% O2
of VS/TS of 86%. The total COD (TCOD) was 118.5 gTCOD L− 1, while was observed in the reactor headspace during the experimental period.
the soluble COD (SCOD) was 2.3 gSCOD L− 1. In addition, it had low The fermentate was stored at 4 ◦ C for no more than one week. Before
concentrations of Total Kjeldahl nitrogen (TKN: 2.9 gN kgTS− 1) and use, it was centrifuged for 10 min at 4000 rpm to separate the solid
total phosphorus (TP: 0.4 gP kgTS− 1). The OFMSW was manually phase from the liquid phase. Then, the solid phase was derived to a
shredded, homogenized, and stored at 4 ◦ C until further use. biochemical methane potential (BMP) test to determine the methano­
The inoculum to initiate the acidogenic fermentation reactor was a genic potential of this secondary stream, and the liquid phase was fed to
highly adapted inoculum from a thermophilic acidogenic fermenter fed the MPBR, as shown in Fig. 1.
with a complex synthetic substrate composed of starch, sucrose,
peptone, and frying oil in a mass ratio of 1:1:1:0.1 g kg− 1, for more than 2.2.3. The phototrophic process with PPB
two years. The inoculum used for the MPBR was a refrigerator-stored The liquid fraction resulting from the fermenter effluent was treated
inoculum from an MBR described in Ref. [36]. The inoculum used in in an MPBR for study in a photoheterotrophic process with a mixed
the BMP tests came from a mesophilic anaerobic digester placed at a culture of PPB. The scheme of the MPBR is shown in the supplementary
WWTP located at Mostoles, Madrid, Spain. material (Fig. S1). It consisted of a 2.5 L reactor (Vworking = 2 L) with a
submerged LED lamp emitting at 805 nm (Idea Bioprocess technology,
Italy) that provided a volumetric irradiance of 2.2 W L− 1. A submerged
2.2. Experimental set-up hollow fiber membrane (Zena s.r.o, Czech Republic) was also used to
discharge the filtered output. A pH-meter with a control system was used
The main steps for the proposed OFMSW valorization process are to maintain the pH above 6.5 by dosing KOH (0.1 M). The reactor head
briefly summarized in Fig. 1, and each step is explained in detail in the had two gas outlets, one sealed with a rubber stopper dedicated to taking
following sections. biogas samples and the other connected to a flowmeter (Ritter, Ger­
many) to measure the volume of gas produced.
2.2.1. Steam explosion pretreatment The MPBR head was opened every other day to clean the submerged
The OFMSW was pretreated with Steam Explosion at 150 ◦ C and 38 LED lamp to avoid biofilm formation. Each time this happened, the
min [37]. We aimed at avoiding the formation of melanoidins through reactor headspace was purged with Ar to ensure anaerobic conditions. It
Maillard reactions, which usually occur rapidly at temperatures above was also ensured that the substrate did not degrade by keeping the feed
160 ◦ C by the interaction of free amino acids and carbohydrates [38]. bottle in a cooler at 4 ◦ C. During the first 5 days of operation, the MPBR
The pilot plant consisted of a steam boiler (Certuss E56, Krefeld, Ger­ was fed with a synthetic substrate consisting of HAc:HPr:HBu:EtOH on a
many) and a 20 L total volume hydrolysis reactor (Vworking = 10 L) 1:1:1:1 COD basis and nutrients from the Ormerod growth medium
connected to a flash tank (V = 100 L). The reactor load was 6 kg of waste described elsewhere [39] for acclimatization. Afterward, the liquid
per batch. Upon the reaction, a sudden opening of the reactor caused a fraction of the fermenter effluent was fed, and the operating conditions
steam explosion in the flash tank, and the thermally treated wastes were were varied in 7 stages, as depicted in Table 1, which can be classified as
extracted and cooled to ambient temperature. Start-up (S0), operation under stable biomass growth (S1 and S2), first
carbon overload (S3), biomass recovery (S4 and S5) and second carbon
2.2.2. Acidogenic fermentation of the hydrolyzed waste overload (S6). Due to extreme weather conditions (snowstorm), access
The hydrolysate obtained after the steam explosion directly fed the to the laboratory was not possible, and thus the reactor was not sampled
acidogenic fermenter. This process was carried out in a CSTR reactor from day 54–60. The biomass yield is calculated as the COD equivalent
with a total volume of 2.5 L (working volume, Vworking = 2 L). An organic of the biomass in the culture (a theoretical 1.75 gCOD gVSS− 1 is taken as
loading rate (OLR) of 4 gCOD L− 1 d− 1 was fixed and kept for 99 days. a reference) divided by the COD consumed.
Twenty-two days were dedicated to the start-up process. The rest was
considered steady-state, and the effluent was stored for its use as a 2.2.4. BMP test
substrate in the MPBR. The thermophilic temperature (55 ◦ C) was Anaerobic digestion performance of the solid fraction obtained after
controlled with an external water bath connected to an external jacket of centrifugation of the acidogenic fermenter output was carried out
the reactor. The HRT was 5 days, and the pH was set at 5.5 with a PLC. through standard biochemical methane potential (BMP) tests [40]. In
One outlet on the headspace was connected to a flowmeter (Ritter, brief, the tests were performed by triplicate, yielding mLCH4 gVS− 1, in
Germany), while the other was sealed with a rubber stopper to take 160 mL serum bottles at mesophilic conditions (37 ± 0.5 ◦ C). The
biogas samples. The anaerobic state of the reaction was ensured both in organic substrate used was the solid fraction after centrifugation of the
the reactor and in the feed and outlet bottle. Anaerobic conditions were fermentate. An inoculum to substrate (I/S) ratio of 2:1 (as VS) and an
established in the substrate by sparging the medium with Ar and initial concentration of 10 gVS L− 1 were set up. A triplicate control of the
inoculum was used to subtract the methane production coming from its
endogenous digestion.

2.3. High-throughput 16S rRNA gene sequencing

For the DNA extraction and microbial community analysis by PacBio

sequencing, representative samples of the MPBR were taken twice a
week to analyze the development of the bacterial communities, ensuring
that each phase studied had at least three samples each, except for the S0
acclimation phase, where only one sample was collected. Samples were
stored at − 4 ◦ C until use. Total DNA was extracted using a Microbial
DNA isolation kit (Soil DNA Isolation Kit, CANVAX, Cordoba, Spain) to
perform the molecular identification of prokaryotes and frozen to
− 20 ◦ C for 4 days until use. The quantification, amplification (PCR), 16S
rRNA gene measurements, and the amplicon taxonomic annotation and
comparative analysis were outsourced to Instituto ISABIAL-FISABIO,
Fig. 1. Process configuration adopted in the present study. Hospital General Universitario de Alicante, Alicante, Spain. SMRT

3
L.D. Allegue et al. Renewable and Sustainable Energy Reviews 166 (2022) 112687

Table 1
Description of the operational stages of the membrane photobioreactor.
Parameters Stage 0 Stage 1 Stage 2 Stage 3 Stage 4 Stage 5 Stage 6

Substrate Synthetic Liquid fraction of the fermentate

Time (days) 5 18 15 15 15 10 11
HRT (days) 2 2 2 2 2 2 2
1 − 1
OLR (gCOD L− d ) 1 1 1 3 1 1 3
SRT (days) 4 4 6 4 4 6 6

sequencing was performed on a PacBio RS II instrument. A more detailed 3. Results

description can be found in the Supplementary Material.
3.1. Steam explosion and acidogenic fermentation pretreatments
2.3.1. Statistical analysis
The evolution of the bacterial composition determined by PacBIO The steam explosion disruption apparatus and experimental condi­
was used to perform a principal component analysis (PCA) to explore the tions provided reproducible conditions between the different batches
variation of the communities over time. In addition, distance-based performed. Fig. 2 shows the release of soluble organic matter with a final
redundancy (RDA) was conducted to reveal potential associations and result of 40% soluble COD on average. Considering that the soluble COD
multivariate relationships between environmental factors (OLR, sludge before the steam explosion was 12%, there is an increase in soluble COD
retention time (SRT), PHA, Glycogen, volatile suspended solids (VSS), of 43%. An average of 7% SCCA is measured, mainly Hac, HBu, and HPr.
and hydrogen production) and the relative abundance of phylotypes at The majority of soluble COD is unidentified organic matter (33%), while
the genus level. Both PCA and RDA were carried out using the “Vegan” the remainder is particulate COD. This pretreatment reproducibly sol­
Package in RStudio (R Development Core Team, 2022.02.0 version). ubilizes a substantial percentage of the COD present in the initial waste.
Relationships between samples, bacterial communities, and environ­ This solubilization is essential to reduce hydrolysis, a limiting factor in
mental parameters were examined using multiple linear regression most biological processes, including acidogenic fermentation.
analysis to rule out linear variables (>95%). A thermophilic acidogenic fermentation was carried out for nearly
100 days as a complementary pretreatment. This process significantly
2.4. Analytical procedures increases the soluble COD, especially the percentage of SCCA produced,
as shown in Fig. 2. Soluble COD increased by 82.5%, up to 74% of the
TS, VS, total suspended solids (TSS), volatile suspended solids (VSS), total COD. 65.6% of all COD was transformed into SCCA, representing
chemical oxygen demand (COD), ammonium, phosphate, total Kjeldahl acidification of 90%. HAc and HBu are the two major acids produced,
3−
nitrogen (TKN), NH+ 4 , PO4 were determined according to Standard accounting for 63% of the COD equivalent of all SCCAs. Also, Fig. 3
Methods [41]. Liquid samples were filtered through a cellulose-ester shows the operation profile of the fermenter. Fig. 3a shows a stabilized
1
filter of 0.45 μm of pore size (Advantech, Japan). The SCOD is consid­ yield of about 0.66 gCODSCCA gCOD−feed . The combination of both pre­
ered to be the COD measured in the liquid fraction after filtration. TCOD treatments on the OFMSW for the production of SCCA achieves high
is the COD measured before any filtration. SCCA and monosaccharides yields and high stability over time. In addition, hydrogen is co-produced
are analyzed using an ion-exclusion RazexTM ROA-Organic Acid H+ at high yields during the acidogenic fermentation process.
HPLC column (Phenomenex, USA), coupled to a refractive index de­ Fig. 3b shows the hydrogen production and VSS amount present in
tector (Agilent, USA) and operated at 65 ◦ C and 1 mL min− 1, with 0.005
M H2SO4 as the eluent. Gas samples were quantified using the 7820A gas
chromatography (GC-TCD) system (Agilent Technologies, Santa Clara,
CA, USA). The mobile phase is Argon at a 5 mL min− 1 flow rate. The
temperature of the oven and the detectors are 45 and 220 ◦ C,
respectively.
Samples were taken directly from the reactor using a purge valve to
quantify PHA, glycogen, and EPS. Immediately after extraction, the
biomass was fixed with 0.2% (by volume) formaldehyde and freeze-
dried overnight. Commercial standards for PHB, PHV, and PHH
(Sigma-Aldrich, USA) were used to perform the calibration curves. The
methodology used for the digestion, extraction, identification (through
GC-MS), and quantification (through GC-FID) of PHA can be found in
the Supplementary Material. For the measurement of glycogen, a
method was adapted from Ref. [42]. In brief, after overnight lyophili­
zation, 1–2 mg of dry biomass are weighed, digested, supplemented with
2 mL of 0.9 M HCl, and dissolved in a thermoblock for 3 h at 100 ◦ C.
Then, they are quickly cooled on ice and filtered with a 0.45
cellulose-ester filter. Finally, glycogen was quantified by HPLC mea­
surement using the same method described for SCCA quantification (see
Supporting Information). The calibration curve was performed with
standard glycogen from oysters (Sigma-Aldrich, USA). EPS quantifica­
tion strictly followed the method explained in Ref. [43] (see Supple­
mentary Material).

Fig. 2. Performance of the steam explosion (n = 4) and subsequent acidogenic

fermentation (average of 70 days) in terms of COD mass balance. Average
values with 95% confidence intervals.

4
L.D. Allegue et al. Renewable and Sustainable Energy Reviews 166 (2022) 112687

Fig. 3. Acidogenic fermentation reactor operation profile for 98 d. a) Shows

TCOD ( ), SCOD ( ), and COD equivalent of all SCCA ( ). b) shows H2 pro­
ductivity ( ) and biomass ( ) within the reactor. Error bars are 95% confi­
dence intervals.

the fermenter. The stabilization of both parameters is similar to the COD

and hydrogen, which could serve as an online indicator of fermentation
performance. Average hydrogen productivity was 345 mLH2 L− 1 d− 1,
with a hydrogen percentage of 57% (43% CO2), and biomass stabilized
near 1.8 gVSS L− 1. Based on the total COD balance, hydrogen production
never exceeded 3% of the total COD equivalent, confirming that it is a
high-value co-product but represents a small percentage of the total
organic matter in the OFMSW valorization.

3.2. The photoheterotrophic treatment of the liquid fraction with PPB

Fig. 4. Membrane photobioreactor operation profile for 88 d. Panel a) shows
3.2.1. Start-up (S0)
biomass ( ) and COD removal ( ). Panel b) shows PHB + PHV percentage ( ),
The capacity of a mixed culture enriched in PPBs to accumulate PHA total percentage of PHA (PHB + PHV + PHH) ( ), glycogen percentage ( ) and
has been evaluated at different stages. Fig. 4 shows the main results. EPS percentage ( ). Panel c) shows hydrogen productivity ( ). Error bars are
During the first 5 days (Stage 0), the culture underwent an adaptation 95% confidence intervals.
period to promote PPB’s growth using a synthetic substrate. The biomass
went through an exponential growth phase from the second to the third a substrate.
day and increased to 1.5 gVSS L− 1, with COD consumption reaching
85%. Correct acclimatization is confirmed by a bacterial community 3.2.2. Operation under stable biomass growth (S1, S2)
study with a sample from day 4 (D4). Fig. 5 shows a culture highly Stable biomass growth and high PHA (up to 34% dry weight) accu­
enriched in PPBs (>80%). The most prominent genus was Rhodop­ mulations were obtained in stages 1 and 2. The low organic load and a
seudomonas sp. with more than 60% relative abundance, followed by substrate composed mainly of SCCA but with limited nutrients first
Rhodobacter sp. with 15%. Rhodovibrio sp., Rhodopila sp., and Rubrivivax favored the growth of biomass and PHA accumulation, as shown in
sp. were also identified with abundances lower than 1.2%. Glycogen was Fig. 4. In Stage 1, COD consumption decreased compared to Stage 0, to
not measured in the first 4 days of acclimatization, and the average an average of 50%. However, biomass continues to grow to an average
percentage of PHA accumulation was 5% in dry mass, mainly composed of 1.7 gVSS L− 1. This COD consumption was limited by the availability
of PHB monomer (98% PHB and 2% PHV). From the beginning of the of nutrients (as shown in Supplementary Material, Fig. S4). As a result,
operation, NH+ 4 consumption is complete (Fig. S4), but no hydrogen PHA accumulation increased significantly, and then it stabilized at
production is detected. The acclimatization stage was considered around 31% dry weight during the following 12 days. This PHA
finished on day 5 when COD removal efficiency stabilized and hydrogen comprised PHB, PHV (20% and 5% by dry weight), and PHH that was
production began. Then, the liquid fraction of the fermentate was fed as

5
L.D. Allegue et al. Renewable and Sustainable Energy Reviews 166 (2022) 112687

drastic reduction was PHA accumulation, reduced to 5%. This minimal

accumulation remained stable and did not increase during the 14-day
phase. As PHA accumulation dropped, glycogen increased to 12%, and
hydrogen production doubled. In addition, a decrease in PPBs to less
than 20% was observed, with Rhodovibrio sp. unusually increasing to
15% (see Fig. S4). The next step would be to recover the biomass and a
more PPB-enriched culture, thereby studying whether PHA accumula­
tion is regained.

3.2.4. Recovery of the biomass (S4, S5)

The organic load and other parameters were reduced to mimic those
studied in Stage 1. Subsequently, the culture gradually recovered its
performance, demonstrating the resilience of this technology. In Stage 4,
biomass stabilized at 1.8 gVSS L− 1, and COD consumption rose to 55%.
The PHA accumulation was recovered, achieving 28% dry weight (5%
PHH), and glycogen again lowered to 5%. The hydrogen production
Fig. 5. Evolution of the microbial community structure of the membrane trend was also negative, decreasing its productivity. In Stage 5, the SRT
photobioreactor at the genus level during the different operative stages. was increased and followed the same trend as Stage 2, increasing the
biomass concentration considerably up to a maximum of 3.6 gVSS L− 1
also detected, accounting for 6% by dry weight. Average production of and slightly the PHA concentration. In both Stages, the percentage of
6% glycogen was also observed in a very stable pattern. Interestingly, PHH is slightly reduced compared to Stage 3, with an average of 25% of
hydrogen evolved with a similar pattern to that of PHA. the total PHA dry mass. Glycogen accumulation and hydrogen produc­
In Stage 2, the SRT increased the sludge age, thus keeping the tion follow decreasing trends, as in Stage 4. Since Stage 3 showed slight
biomass longer in the reactor. In this Stage, the biomass concentration granulation in the culture (see Fig. S6), EPS were measured to see if it
slightly increased to 2.2 gVSS L− 1 while the COD consumption rose to was producing them. At Stage 4, the culture did not show as much
60%. PHA accumulation took 4 days to stabilize at an average of 44%, granulation, and no EPS was detected. However, at Stage 5, EPS reached
with PHH being about 30% of total PHA by weight. Glycogen accumu­ 5% dry mass. As for the communities, up to 55% of PPB were recovered
lation slightly dropped to 5% by weight. Hydrogen, unlike Stage 1, in Stage 4 (Fig. 5) but stabilized at around 40% in Stage 5. The only
seems to have an inverse relationship with PHA accumulation. As ex­ difference was a higher abundance of Rubrivivax sp. In short, biomass
pected, upon adding the liquid fraction of the fermentate, the pressure of could be restored, significantly increasing its concentration and PHA
cross-contamination on the culture increased, and the relative abun­ accumulation, while glycogen accumulation and hydrogen production
dance of PPB dropped to just over 40%, with Rhodopseudomonas sp. decreased again. We proceeded again to perform an organic overload to
being still the predominant genus (Fig. 5). In both Stages, a stable PPB- see if the tendencies observed were reproducible.
enriched culture was achieved, with high PHA accumulations and
hydrogen yields, with inverse trends. Given these promising results, the 3.2.5. Second carbon overload (S6)
organic load was increased to examine if the productivity of PHA can be Similar to Stage 3, the organic overload of Stage 6 destabilized the
further improved. culture, reducing biomass concentration, COD consumption, and PHA
accumulation. On the other hand, glycogen accumulation and hydrogen
3.2.3. First carbon overload (S3) production had the opposite tendency to achieve maximum hydrogen
In Stage 3, the OLR was increased from 1 to 3 gCOD L− 1 d− 1 to productivity at 310 mL L− 1 d− 1. The culture showed evident granula­
analyze the culture behavior under an overload episode. The culture tion, and the EPS accumulation increased to more than 10% dry weight.
immediately fails to accumulate PHA. However, glycogen accumulation PPB abundance was again reduced to less than 20%, confirming that an
and hydrogen production increased. As shown in Fig. 4, biomass organic overload destabilizes the PPB-enriched culture. The trends
collapsed, reducing to 1.3 gVSS L− 1, although after 5 days, it stabilized observed in Stage 3 are repeated in Stage 6, confirming the reproduc­
at 1.6 gVSS L− 1. The COD consumption fell to 36%, though the limita­ ibility of the results and the culture’s behavior under organic overload.
tion of COD consumption was not driven by nutrient availability, as both The trends observed at each Stage depicted in Fig. 4 are also
NH+ 4 and PO4 were not 100% consumed, as shown in Table 2. The most
+
corroborated when we observe the performance parameters in Table 2.
The highest PHA productivities occurred at Stages 2 and 5, while higher
glycogen and hydrogen productivities occurred at Stages 3 and 6,
Table 2 matching the organic overload. Biomass yields in the best performing
MBPR performance parameters on each phase. Average ± standard deviation. Stages (2 and 5) of 0.97 and 0.94 gCOD gCOD− 1 were obtained,
Parameters S1 S2 S3 S4 S5 S6 respectively. It is assumed that when the biomass yield drops, the non-
Yx/s (gCOD 0.83 ± 0.97 ± 0.62 ± 0.86 ± 0.94 ± 0.62 ±
consumed COD is being diverted to hydrogen production. This fact
gCOD− 1) 0.07 0.05 0.08 0.25 0.02 0.07 agrees with the experimental data where we see the highest hydrogen
-qS (mgCOD 541 ± 638 ± 677 ± 688 ± 1055 665 ± yields and the lowest biomass yields at the same Stages. A maximum qs
L− 1 d− 1) 45 32 62 28 ± 66 49 of 1 gCOD L− 1 d− 1 was achieved on Stage 5, but, interestingly, this
qPHA (gPHA 0.22 ± 0.45 ± 0.05 ± 0.28 ± 0.61 ± 0.06 ±
parameter was not significantly reduced in the organic-overloaded
L− 1 d− 1) 0.05 0.05 0.02 0.02 0.06 0.04
qH2 (mlH2 L− 1 88 ± 71 ± 140 ± 83 ± 51 ± 5 256 ± Stages. As discussed in previous sections, NH+ 4 is 100% consumed in
d− 1) 12 15 25 26 34 Stages 1, 2, 4, and 5. Although it was not wholly consumed in Stages 3
qGLY (gGLY L− 1 0.10 ± 0.09 ± 0.19 ± 0.09 ± 0.13 ± 0.17 ± and 6, its assimilation was still very high, up to 94 and 93%, respec­
d− 1) 0.03 0.01 0.04 0.00 0.01 0.02 tively. However, PO+ 4 was not fully consumed in any Stage, with Stages 5
NH+ 100 ± 100 ± 94 ± 2 100 ± 100 ± 93 ± 1
4
comsuption 1 1 2 1
and 6 having the lowest consumption, down to 60%. It was again
(%) confirmed that PHA is inversely related to glycogen accumulation and
PO+4 81 ± 9 85 ± 7 71 ± 3 84 ± 2 68 ± 7 60 ± 4 hydrogen production. The relationship of the environmental variables to
comsuption the transitions in the microbial communities was examined in more
(%)
depth.

6
L.D. Allegue et al. Renewable and Sustainable Energy Reviews 166 (2022) 112687

3.3. Statistical analysis of the evolution of microbial community profiles

The development of the microbial communities was evaluated by

PCA ordination based on Hellinger transformed rarefied abundance and
shown in Fig. 6. This analysis showed an apparent clustering of the
samples according to their Stage and temporal progression. The major
statistical difference occurred in Stages 1, 3, and 6. It was due to a start-
up in which the feed was synthetic, and the culture was highly enriched
in PPB. Inversely, in Stages 3 and 6, an organic overload produced the
opposite effect. In terms of genera, we can observe a relationship be­
tween Rhodopseudomonas sp., Rhodobacter sp., and Dsyngomonas sp.,
which had more significant weight in the first three Stages. Rhodophila
sp. and Rhodovibrio sp. had a greater weight in Stages 2 and 5, respec­
tively. Paludibacter sp. and Dsulfovrio sp. were correlated and had the
highest weight in the first organic overload (Stage 3). Also, they had a
higher weight in the second one (Stage 6), which could mean cross-
contamination Sphingomonas sp. and Mesorhizobium sp. (see Fig. S3).
The clustering of the Stages and how most PPB genera were negatively
related to organic overloads were confirmed. To further analyze these
results, an RDA was carried out to assess the contributions of environ­ Fig. 7. Redundancy discriminate analysis (RDA) ordination diagram of bacteria
mental variables to variances in the microbial communities. related to environmental variables. The nodes show the 10 genera with the
RDA analysis, shown in Fig. 7, confirms the relationship of PPB most weight on both axis (PPB in purple, other genera in yellow), and the ar­
rows indicate the weight of each environmental variable.
genera (Rhodopseudomonas sp., Rhodobacter sp., Rhodopila sp., and
Rubrivivivax sp) with PHA accumulation and SRT and its inverse rela­
tionship with organic load (OLR). Also, Acetobacter sp. is highly related environmental factors.
to PHA accumulation. Furthermore, the inverse relationship between
PHA, glycogen, and hydrogen is again evident. A positive relationship is 4. Discussion
also observed between higher biomass concentration (VSS) and Pal­
udibacter sp. and Rhodovibrio sp. To find more evident trends, another The results obtained in the combination of steam explosion and
1
RDA analyzed only the statistical weight of time and OLR on the acidogenic fermentation pretreatments (0.66 gCODSCCA gCOD−feed ) can
composition of the bacterial communities. The PPB Rhodopseudomonas be considered a very efficient step in the PHA production pathway. This
sp., Rhodobacter sp., and Rhodopila sp., as well as Acetobacter sp. and result is slightly lower than that obtained recently in acidogenic
1
Dsyngomonas sp., are not affected by the time or OLR (see Fig. S5 in fermentation of fruit residue, where a yield of 0.74 gCODSCCA gCOD−feed
Supporting Information). However, we can see that Mesorhizobium sp. was achieved [20]. Another work showed fermentation of OFMSW to
and Sphingomonas sp. are positively related, indicating a possible obtain an overall yield of 0.65 ± 0.04 gCODVFA gVS−(0)1 and acidification
development of contamination from the acidogenic fermenter devel­ of 86% [44]. In our work, we obtained slightly higher acidification
oped over time. In a similar case, the PPBs that are gaining ground with (90%), which we know from previous work is essential when working
time are Rubrivivax sp. and Rhodovibrio sp., but also Paludibacter sp., with PPB since the high presence of sugars in the substrate can nega­
associated with time and higher organic loads. Finally, fermentative tively affect the accumulation of PHA [29]. The remainder of the un­
genera had a great weight on hydrogen production. Therefore, PHA and defined COD could be oligo- and monosaccharides derived from
SRT are inversely related to hydrogen, glycogen, and OLR production, hemicellulose, lipids, proteins, complex carbohydrates, and amino acids
with different fermentative genera having the most weight for the latter [45]. Within the SCCAs produced, acetic acid corresponded to 43.4% of
the COD-equivalent. Previous studies determined that it is the most
suitable substrate for PHA production with PPB [46]. The second SSCA
was butyric acid, corresponding to 25% of the COD-equivalent of SCCAs.
Propionic, valeric, and hexanoic acids represented 17, 7, and 7%,
respectively. It is well-known that the mixing of acids favors the pro­
duction of copolymers; for example, propionic and valeric acids are
oxidized to propionyl-CoA, then to β-hydroxyvalerate (PHV) to produce
PHV [21]. Moreover, although it has not yet been demonstrated in a PPB
mixed culture before, it has recently been discovered that the presence
of hexanoic acid promotes the accumulation of PHH in Rhodospirillum
rubrum [47]. If these results are maintained in the process scale-up, the
SCCA productivity and the fast reaching of a steady-state, reproduc­
ibility, and the process stability, highlight the practicality and high
suitability of these pretreatments for the production of PHA with PPB.
The PPB-enriched mixed culture accumulated high percentages of
PHA by assimilating fermented effluent from OFMSW feed. As shown in
Fig. 4, the maximum percentage of PHA accumulation achieved was
42%, which is the maximum achieved with mixed cultures of PPB using
residual substrates [48], the closest result being the treatment of fer­
mented domestic wastewater, where 30.8% was achieved [29]. Fig. 7
Fig. 6. PCA plot of the microbial communities. The samples are shown on day shows an unequivocal relationship between the abundance of PPB
code (D4: day 4) and colored according to the Stage. The names indicate the 10
genera and PHA production. Rhodopseudomonas sp. is the most abundant
genera that contribute the most weight to PC1 and PC2, colored according to
PPB genus in all Stages except S3, reaching up to 90% of the total, and
PPB (purple) or another genus (yellow). The arrows indicate the weight of each
genus in each PC. coincides with the most abundant genus in other studies performed in

7
L.D. Allegue et al. Renewable and Sustainable Energy Reviews 166 (2022) 112687

our laboratory [36]. In other studies, the bacterial communities fluc­ The organic load’s increase in the MPBR led to the destabilization of
tuate in their percentage of PPB, for example, from the dominance of the culture and the collapse of PHA accumulation (Stages 3 and 6,
Rhodopseudomonas to Rhodobacter [49]. Although Rhodobacter sp. is the Fig. 4). An increase in glycogen accumulation and hydrogen production
second most abundant PPB genus identified in our study, Rhodop­ was detected in both Stages. This evidences the competition between
seudomonas sp. always remained predominant among PPB. Rubrivivax carbon allocation to growth and accumulation of PHA and glycogen,
sp. is the PPB genus with the highest reported PHA accumulation, with a with electrons allocation between growth, PHA accumulation, and
percentage of 85% by weight in pure cultures [48], which is coincident hydrogen production, as outlined schematically in Fig. 8. Glycogen
with the high PHA accumulations in Stage 5. In addition, we can see a accumulation by PPBs has been known for a long time. Still, only one
recurrent closeness between these PPB and the genera of Dsyngomonas study has examined the relationship between PHB and glycogen pro­
sp. and Acetobacter sp., which may indicate symbiotic processes. Ace­ duction [32], showing that PHB is an electron sink that works as an
tobacter, for example, can oxidize short-chain acids and ethanol to intracellular reserve for reducing power. However, it rapidly decreases
acetic acid [49], and PPBs can assimilate this substrate to accumulate when biomass growth is reduced, thus diverting carbon storage to
PHA faster. A more comprehensive metaproteomics study could help to glycogen accumulation. A variety of metabolic principles are linked to
clarify these symbiotic processes in enriched mixed cultures since it is glycogen metabolism: the maintenance of (i) photosynthetic efficiency
essential to know which PPB species have the highest capacity to in light, (ii) viability in periods of starvation, such as in darkness or
accumulate PHA to improve this process. Among the different PHA macronutrient depletion, and (iii) the acclimation to macronutrients
monomers produced, PHB is the predominant one, but other less com­ deficiency [54]. It is noteworthy that after Stage 3 and the lowering of
mon ones, such as PHH, reach relevant percentages. OLR, there was a rapid recovery of the culture, demonstrating that the
Most PHA accumulation studies with PPB only considered PHB and PHA production is stable and resistant to severe organic stress, and
PHV accumulation. In our work, we found, in addition to PHB and PHV, although the community within the PPB mixed culture may change, its
the PHH in all stages except Stage 0. The rest of the Stages accumulated functionality remains consistent. However, once PHA is no longer
between 2% and 30% of the total PHA in dry mass. It was demonstrated available as an electron sink, hydrogen production appears as a possi­
that PHH accumulates when hexanoic acid is present in the medium, bility for the balance of reducing equivalents.
assimilates via β-oxidation, and incorporates into PHA chains [50]. In The competition between hydrogen production and PHA accumu­
addition, the presence of hexanoic acid increases the accumulation of lation in Rhodobacter sphaeroides and Rhodospirillum rubrum strains has
PHB by boosting the production of intermediates such as acetyl-CoA and been observed [55], and the conversion of PHB accumulated in Rhodo­
butyryl -CoA [47]. The acetyl-CoA and butyryl-CoA produced as in­ valum sulfidophilum to hydrogen in the absence of other substrate was
termediates and end products are then assimilated using anaplerotic demonstrated [56]. In our experiment, this inversely proportional
pathways, e.g., the ethylmalonyl-CoA and the methylbutanoyl-CoA relationship is evident in Fig. 7. However, the mechanisms of electron
synthesis and degradation pathways [51]. In Stage 0, PHH was not transfer and how both metabolic pathways compete for electrons allo­
detected because the synthetic substrate did not contain hexanoic acid. cation are poorly understood. While the decreasing trend in PHA and
In the rest of the Stages, PHH had similar behavior to the rest of the PHA. increase in glycogen accumulation and hydrogen production is observed
Only two papers reported a PHH production by phototrophic bacteria, at Stages 3 and 6, biomass granulation was also evident, whereby at
one where C6 and C7 monomers were detected in Rhodospirillum rubrum Stages 4, 5, and 6, the accumulation of EPS was measured. During
[52] and another where PHH was detected in a mixed PPB culture [52]. organic overload of these stages, PPBs lower their cellular efficiency and
In addition to the PHH, we detected another peak in the GC, and, divert excess electrons to EPS and hydrogen. Another hypothesis for the
considering its retention time, we assumed that it could be a 3-carbon increased hydrogen production is that EPS, which are exopolymers
monomer forming polyhydroxypropionate (PHP). This assumption was composed mainly of carbohydrates, might be optimal substrates for the
furtherly verified on GC-MS as explained in the supplementary material growth of fermentative bacteria. As explained in previous sections, these
(Fig. S2). Both PHH and PHP, which are generally not accounted for, bacteria can come from cross-contamination between bioreactors,
indicate that PHA accumulation is usually underestimated with mixed increasing their biomass over time and producing more hydrogen. In
PPB cultures, increasing the interest in the technology. addition, the biomass’s COD and the cell yield decreased considerably,
One of the significant limitations of advancing this technology is the which is linked to fermentative processes. Fermentative bacteria pro­
low productivity obtained so far. In this study, PHA productivity reached duce more oxidized compounds for PPBs, which can assimilate them for
very high values (qPHA up to 0.61 gPHA L− 1 d− 1) compared to previous growth, causing a symbiotic relationship already evidenced in Figs. 6
works. The highest data obtained to date in mixed cultures of PPB is 0.77 and 7. The strategy of achieving a granular culture of PPB first and then
gPHA L− 1 d− 1 [21], but being fed with acetic acid. In a recent study by deriving its metabolism to PHA production could lead to considerable
the same group using fermented wastewater, the highest productivity cost savings downstream of the process, as granular sludge can decrease
obtained with a permanent feeding regime was 0.23 gPHA L− 1 d− 1 [29]. the Operational Expenditure (OPEX) of the process by up to 50%, which
Although the results presented in this work are promising, they are still has been reported for other anaerobic bacteria [57].
far from the productivities obtained with aerobic cultures, as they report The granulation capacity of PPB has been studied recently, and EPS
average productivities of 5–10 g L− 1 d− 1 [53]. The primary constraint is accumulations of up to 35% in dry mass were observed [33]. At Stage 6,
biomass growth, in this work, limited by nutrients. However, the higher a 10% dry mass of EPS was reached. EPS are primarily composed of
the biomass concentration, the more this technology is limited by the carbohydrates, proteins, and lesser amounts of other components and
volumetric irradiance in the reactor. In this work, a submerged LED have multiple roles such as flotation and locomotion, feeding, protection
lamp with a very low volumetric irradiance (2.1 W L− 1 or 5 W m− 2) was against desiccation/UV/pollution, development of biofilms, and
used when most studies are usually above 30 W m− 2 [31]. Using the communication [58]. These compounds are widely employed in the
biomass productivity of Stage 5 (1.8 gVSS L− 1 d− 1), this would result in a cosmetic, pharmaceutical, and petroleum industries [59]. In the work
biomass energy yield of 61 gCOD kWh− 1, which is slightly higher than where PPB granulation is analyzed, the relative abundance obtained is
the highest recorded so far of 59 gCOD kWh− 1 [53], but still does not between 40 and 70% [33], slightly higher than obtained in this work,
allow to obtain a process energetically more favorable than an aerobic where the range moves from 22 to 80%. The relationship between PHA
process. New lighting systems and reactor configurations (especially and EPS in PPB has not been studied so far, but the negative correlation
systems to retain solids) should be investigated to produce more between EPS and PHA was evidenced in other microorganisms [60], as
concentrated biomass. Another critical element is increasing PHA shown in Fig. 4 of our study. However, it is known that the EPS confer
accumulation and understanding what other components, carbon, protective effects upon the cell, allowing microorganisms to grow in
electrons, or both, can be diverted when PHA accumulation ceases. attached mode and improving the settling. Although a study showed

8
L.D. Allegue et al. Renewable and Sustainable Energy Reviews 166 (2022) 112687

Fig. 8. Schematic description of the different biopolymers for carbon storage and consequent electron allocation between PHA, glycogen, EPS, and hydrogen, related
to increasing organic load stress.

that the settling cohesion was enhanced and stabilized by purple pho­ [53], the most logical solution being the scaling of the technology using
totrophic bacteria [61], little is known about the stabilization and the irradiance of the sun. Scaling up outdoors involves many un­
sedimentation in PPB cultures, even less with high EPS content. certainties, such as contamination by other microorganisms, response to
Considering that the downstream processing, including settling and non-fully anaerobic systems, and intermittent solar illumination. These
biomass extraction, is a bottleneck in PHA production [62], the granu­ problems could be solved by sunlight collection and filtration systems
lation of mixed PPB cultures and the simultaneous production of PHA [31]. Operational strategies adaptable to the different seasons of the
and EPS can reduce costs in the scale-up of the technology [63] and year and the feedstock arriving in the biorefinery is a crucial point for its
opens up exciting research opportunities for the future. development.
The possibility of producing different products seasonally (hydrogen,
glycogen, EPS) and PHA can provide the system with more economic
4.1. Implications for a future photo-biorefinery
resilience. This study demonstrates that if the reactor is operated at a
higher organic load, the carbon accumulation is redirected to glycogen,
Cascading biorefineries for waste treatment are indispensable for a
or even granulation can occur in the culture due to stress and derived to
sustainable future because they can turn a problem into an opportunity.
EPS production. This links to higher hydrogen productivities through
The innovative biorefinery platform herby proposed integrates a steam
electron allocation in PPB metabolism or synergies generated with
explosion pretreatment that can significantly increase the fraction of
fermentative bacteria. In addition, this work opens the way for the
fermentable organic carbon from the OFMSW with an acidogenic
production of longer chain biopolymers that could diversify their in­
fermentation that produces high yields of SCCA, which are the perfect
dustrial applications, considering that the PHB-PHV-PHH polymers
substrate for PHA accumulation in a PPB-enriched culture. This bio­
present rubber-like elastomeric properties and can therefore be used in a
refinery is completed with the anaerobic digestion of the solid fraction
different set of applications than a polymer composed solely of PHB and
obtained after fermentation, yielding 336 LCH4 kgVS− 1 (see Fig. S7 in
PHV [12]. Thus, it would still be essential for the scale-up of this tech­
Supporting Information). This additional energy source would close the
nology to optimize PHA production and to be able to tune the type of
carbon and energy cycle of the biorefinery. As for the COD removed,
polymer obtained. Finally, we evidenced which PPB genera are more
more than 40% of the COD is not assimilated, mainly due to nutrient
critical in the PHA accumulation process and how although the com­
limitations. This residual COD flow from the MPBR could be combined
munity evolves, its functionality remains the same. All this Information
with the final digestate from anaerobic digestion, a high nutrient side
gives us tools for optimizing and controlling this technology and paves
stream, in a possible second phototrophic reactor with PPB dedicated to
the way for future large-scale feasibility analyses.
protein production. This technology has already been tested several
times [64,65]. Nonetheless, the biggest challenge of this technology is
5. Conclusions
the industrial scalability of the MBPR.
The industrial production of PHA is currently carried out by pure
This study demonstrates the possibility of increasing PHA production
aerobic cultures based on sugars or other similar substrates. Compara­
by combining different operating strategies. Firstly, the association of
tively, the production of PHA with mixed PPB cultures has some ad­
steam explosion pretreatment and thermophilic acidogenic fermenta­
vantages: (i) no need for sterilization of equipment, (ii) they can use a 1
tion leads to stable production of up to 0.74 gCODSCCA gCOD−feed and 345
wide variety of waste as substrate, eliminating a large part of the pro­
mLH2 L− 1 d− 1 from OFMSW. Secondly, understanding the different
duction costs, (iii) they do not need aeration, (iv) higher yields of PHA
metabolic pathways of carbon assimilation and consequent electron
are possible, and (v) they can accumulate and produce PHA in the same
allocation during transient OLR is critical to enhancing PHA production.
reactor by IR illumination, eliminating the requirement for sorting and
This work achieved a maximum PHA accumulation of 42% on a dry
an accumulation reactor. An intermediate approach is aerobic mixed
basis by controlling the OLR. This production is stable and resistant to
cultures, which are already studied at a pilot plant scale [19,20]. This
severe stress caused by organic overload, and although the community
technology’s best performance is 0.45 gCODPHA gCODFeed [20]. A pre­
within the PPB mixed culture may change due to this stress, its func­
liminary mass balance of the overall novel photobiorefinery proposed
tionality remains consistent, and the stress is reversible after the over­
here shows a total yield of 0.15 gCODPHA gCODFeed. Moreover, although
load. Finally, this comprehensive analysis elucidates critical
this is the highest biomass energy yield seen so far, as discussed in
mechanisms that allow the design of a PPB-based pho-biorefinery with
previous sections, for the time being, the production of PHA with PPBs
seasonal PHA, EPS, and hydrogen production, depending on the culture
and artificial illumination is considered to be economically unfeasible

9
L.D. Allegue et al. Renewable and Sustainable Energy Reviews 166 (2022) 112687

feed, contributing to urban solid waste management, and taking a step [15] Pérez V, Mota CR, Muñoz R, Lebrero R. Polyhydroxyalkanoates (PHA) production
from biogas in waste treatment facilities: assessing the potential impacts on
forward in the direction of a circular bioeconomy.
economy, environment and society. Chemosphere 2020;255. https://doi.org/
10.1016/j.chemosphere.2020.126929.
Author Contributions [16] Guleria S, Singh H, Sharma V, Bhardwaj N, Arya SK, Puri S, et al.
Polyhydroxyalkanoates production from domestic waste feedstock: a sustainable
approach towards bio-economy. J Clean Prod 2022;340:130661. https://doi.org/
LDA: Data curation; Formal analysis; Investigation; Software; Visu­ 10.1016/j.jclepro.2022.130661.
alization; Writing – original draft; MV: Data curation; Investigation; [17] Pandey AK, Pilli S, Bhunia P, Tyagi RD, Surampalli RY, Zhang TC, et al. Dark
fermentation: production and utilization of volatile fatty acid from different
Writing – review & editing; JAM: Funding acquisition; Project admin­ wastes- A review. Chemosphere 2022;288:132444. https://doi.org/10.1016/j.
istration; Resources; Supervision; Validation; Writing – review & edit­ chemosphere.2021.132444.
ing; DP: Conceptualization; Data curation; Formal analysis; Funding [18] Zhao G, Kuang G, Wang Y, Yao Y, Zhang J, Pan ZH. Effect of steam explosion on
physicochemical properties and fermentation characteristics of sorghum (Sorghum
acquisition; Methodology; Project administration; Supervision; Valida­
bicolor (L.) Moench). LWT (Lebensm-Wiss & Technol) 2020;129:109579. https://
tion; Writing – review & editing doi.org/10.1016/j.lwt.2020.109579.
[19] Moretto G, Lorini L, Pavan P, Crognale S, Tonanzi B, Rossetti S, et al. Biopolymers
from urban organic waste: influence of the solid retention time to cycle length ratio
Declaration of competing interest in the enrichment of a Mixed Microbial Culture (MMC). ACS Sustainable Chem Eng
2020;8. https://doi.org/10.1021/acssuschemeng.0c04980.
[20] Matos M, Cruz RAP, Cardoso P, Silva F, Freitas EB, Carvalho G, et al. Combined
The authors declare the following financial interests/personal re­ strategies to boost polyhydroxyalkanoate production from fruit waste in a three-
lationships which may be considered as potential competing interests: stage pilot plant. ACS Sustainable Chem Eng 2021;9:8270–9. https://doi.org/
10.1021/acssuschemeng.1c02432.
Juan Antonio Melero, Luis D Allegue, Daniel Puyol reports financial [21] Fradinho JC, Oehmen A, Reis MAM. Improving polyhydroxyalkanoates production
support was provided by Regional Government of Madrid. Daniel Puyol in phototrophic mixed cultures by optimizing accumulator reactor operating
reports financial support was provided by Spanish Ministry of Economy. conditions. Int J Biol Macromol 2019;126:1085–92. https://doi.org/10.1016/j.
ijbiomac.2018.12.270.
[22] Higuchi-Takeuchi M, Numata K. Acetate-inducing metabolic states enhance
Acknowledgments polyhydroxyalkanoate production in marine purple non-sulfur bacteria under
aerobic conditions. Front Bioeng Biotechnol 2019;7:1–10. https://doi.org/
10.3389/fbioe.2019.00118.
Financial support from the Community of Madrid through project [23] Fradinho JC, Reis MAM, Oehmen A. Beyond feast and famine: selecting a PHA
S2018/EMT-4344 BIOTRES-CM is gratefully acknowledged. D. Puyol accumulating photosynthetic mixed culture in a permanent feast regime. Water Res
wishes to thank the Govern of Spain. Ministry of Economy and 2016;105:421–8. https://doi.org/10.1016/j.watres.2016.09.022.
[24] Ghimire A, Valentino S, Frunzo L, Pirozzi F, Lens PNL, Esposito G. Concomitant
Competitiveness for the Ramon y Cajal grant. biohydrogen and poly- b -hydroxybutyrate production from dark fermentation
effluents by adapted Rhodobacter sphaeroides and mixed photofermentative
cultures. Bioresour Technol 2016;217:157–64. https://doi.org/10.1016/j.
Appendix A. Supplementary data
biortech.2016.03.017.
[25] Policastro G, Luongo V, Fabbricino M. Biohydrogen and poly-β-hydroxybutyrate
Supplementary data to this article can be found online at https://doi. production by winery wastewater photofermentation: effect of substrate
org/10.1016/j.rser.2022.112687. concentration and nitrogen source. J Environ Manag 2020;271:111006. https://
doi.org/10.1016/j.jenvman.2020.111006.
[26] Almeida JR, Fradinho JC, Carvalho G, Oehmen A, Reis MAM. Dynamics of
References microbial communities in phototrophic polyhydroxyalkanoate accumulating
cultures. Microorganisms 2022;10. https://doi.org/10.3390/
microorganisms10020351.
[1] Eurostat. Energy, transport and environment statistics 2020 edition. Luxembourg,
[27] Allegue LD, Puyol D, Melero JA. Novel approach for the treatment of the organic
https://doi.org/10.2785/463410; 2020.
fraction of municipal solid waste: coupling thermal hydrolysis with anaerobic
[2] Council of the European Union. Council Directive 1999/31/EC on the landfill of
digestion and photo-fermentation. Sci Total Environ 2020:714. https://doi.org/
waste. https://doi.org/10.1039/ap9842100196; 1999.
10.1016/j.scitotenv.2020.136845.
[3] Sherwood J. The significance of biomass in a circular economy. Bioresour Technol
[28] Allegue LD, Ventura M, Melero JA, Puyol D. Integrated sustainable process for
2020;300. https://doi.org/10.1016/j.biortech.2020.122755.
polyhydroxyalkanoates production from lignocellulosic waste by purple
[4] European Commission. Circular economy action plan. https://doi.org/10.277
phototrophic bacteria. GCB Bioenergy 2021;13(5):862–75. https://doi.org/
5/855540; 2020.
10.1111/gcbb.12807.
[5] Kumar M, Rathour R, Singh R, Sun Y, Pandey A, Gnansounou E, et al. Bacterial
[29] Almeida JR, Serrano E, Fernandez M, Fradinho JC, Oehmen A, Reis MAM.
polyhydroxyalkanoates: opportunities, challenges, and prospects. J Clean Prod
Polyhydroxyalkanoates production from fermented domestic wastewater using
2020;263:121500. https://doi.org/10.1016/j.jclepro.2020.121500.
phototrophic mixed cultures. Water Res 2021;197:117101. https://doi.org/
[6] Tan D, Wang Y, Tong Y, Chen GQ. Grand challenges for industrializing
10.1016/j.watres.2021.117101.
polyhydroxyalkanoates (PHAs). Trends Biotechnol 2021;39:953–63. https://doi.
[30] Grattieri M, Rhodes Z, Hickey DP, Beaver K, Minteer SD. Understanding
org/10.1016/j.tibtech.2020.11.010.
biophotocurrent generation in photosynthetic purple bacteria. ACS Catal 2019;9:
[7] Mannina G, Presti D, Montiel-Jarillo G, Carrera J, Suárez-Ojeda ME. Recovery of
867–73. https://doi.org/10.1021/acscatal.8b04464.
polyhydroxyalkanoates (PHAs) from wastewater: a review. Bioresour Technol
[31] Fradinho J, Allegue LD, Ventura M, Melero JA, Reis MAM, Puyol D. Up-scale
2020;297:122478. https://doi.org/10.1016/j.biortech.2019.122478.
challenges on biopolymer production from waste streams by Purple Phototrophic
[8] Dawes EA, Senior PJ. The role and regulation of energy reserve polymers in micro-
Bacteria mixed cultures: a critical review. Bioresour Technol 2021;327. https://
organisms. Adv Microb Physiol 1973;10:135–266. https://doi.org/10.1016/
doi.org/10.1016/j.biortech.2021.124820.
S0065-2911(08)60088-0.
[32] Philippis R De, Ena A, Guastini M, Sili C, Vincenzini M. Factors affecting poly-/3-
[9] Jain R, Tiwari A. Biosynthesis of planet friendly bioplastics using renewable carbon
hydroxybutyrate accumulation in cyanobacteria and in purple non-sulfur bacteria,
source. J Environ Health Sci Eng 2015;13:1–5. https://doi.org/10.1186/s40201-
vol. 103; 1992. p. 187–94.
015-0165-3.
[33] Stegman S, Batstone DJ, Rozendal R, Jensen PD, Hülsen T. Purple phototrophic
[10] Verlinden RAJ, Hill DJ, Kenward MA, Williams CD, Radecka I. Bacterial synthesis
bacteria granules under high and low upflow velocities. Water Res 2021;190:
of biodegradable polyhydroxyalkanoates. J Appl Microbiol 2007;102:1437–49.
116760. https://doi.org/10.1016/j.watres.2020.116760.
https://doi.org/10.1111/j.1365-2672.2007.03335.x.
[34] Puyol D, Barry EM, Hülsen T, Batstone DJ. A mechanistic model for anaerobic
[11] Dhania S, Bernela M, Rani R, Parsad M, Grewal S, Kumari S, et al. Scaffolds the
phototrophs in domestic wastewater applications: photo-anaerobic model (PAnM).
backbone of tissue engineering: advancements in use of polyhydroxyalkanoates
Water Res 2017;116:241–53. https://doi.org/10.1016/j.watres.2017.03.022.
(PHA). Int J Biol Macromol 2022;208:243–59. https://doi.org/10.1016/j.
[35] Carlozzi P, Giovannelli A, Laura M, Touloupakis E, Di T. Poly-3-hydroxybutyrate
ijbiomac.2022.03.030.
and H 2 production by Rhodopseudomonas sp . S16-VOGS3 grown in a new
[12] Pereira JR, Araújo D, Marques AC, Neves LA, Grandfils C, Sevrin C, et al.
generation photobioreactor under single or combined nutrient de ficiency. Int J
Demonstration of the adhesive properties of the medium-chain-length
Biol Macromol 2019;135:821–8. https://doi.org/10.1016/j.ijbiomac.2019.05.220.
polyhydroxyalkanoate produced by Pseudomonas chlororaphis subsp. aurantiaca
[36] de las Heras I, Molina R, Segura Y, Hülsen T, Molina MC, Gonzalez-Benítez N, et al.
from glycerol. Int J Biol Macromol 2019;122:1144–51. https://doi.org/10.1016/j.
Contamination of N-poor wastewater with emerging pollutants does not affect the
ijbiomac.2018.09.064.
performance of purple phototrophic bacteria and the subsequent resource recovery
[13] Rosenboom J-G, Langer R, Traverso G. Bioplastics for a circular economy. Nat Rev
potential. J Hazard Mater 2020;385:121617. https://doi.org/10.1016/j.
Mater 2022;7:117–37. https://doi.org/10.1038/s41578-021-00407-8.
jhazmat.2019.121617.
[14] Koller M, Mukherjee A. A new wave of industrialization of PHA biopolyesters.
Bioengineering 2022;9:74. https://doi.org/10.3390/bioengineering9020074.

10
L.D. Allegue et al. Renewable and Sustainable Energy Reviews 166 (2022) 112687

[37] Allegue LD, Puyol D, Melero JA. Synergistic thermophilic co-fermentation of food [52] Liebergesell M, Mayer F, Steinbüchel A. Anaylsis of polyhydroxyalkanoic acid-
and lignocellulosic urban waste with steam explosion pretreatment for efficient biosynthesis genes of anoxygenic phototrophic bacteria reveals synthesis of a
hydrogen and carboxylic acid production. Biofuels Bioprod Biorefin 2021:1–11. polyester exhibiting an unusal composition. Appl Microbiol Biotechnol 1993;40:
https://doi.org/10.1002/bbb.2318. 292–300. https://doi.org/10.1007/BF00170383.
[38] Gao S, Lu D, Qian T, Zhou Y. Thermal hydrolyzed food waste liquor as liquid [53] Capson-tojo G, Batstone DJ, Grassino M, Vlaeminck SE, Puyol D, Verstraete W,
organic fertilizer. Sci Total Environ 2021;775:145786. https://doi.org/10.1016/J. et al. Purple phototrophic bacteria for resource recovery: challenges and
SCITOTENV.2021.145786. opportunities. Biotechnol Adv 2020;43:107567. https://doi.org/10.1016/j.
[39] Allegue LD, Puyol D, Melero JA. Food waste valorization by purple phototrophic biotechadv.2020.107567.
bacteria and anaerobic digestion after thermal hydrolysis. Biomass Bioenergy [54] Mas J, Van Gemerden H. Storage products in purple and green sulfur bacteria. In:
2020;142. https://doi.org/10.1016/j.biombioe.2020.105803. Blankenship RE, Madigan MT, Bauer CE, editors. Anoxygenic photosynthetic
[40] Angelidaki I, Alves M, Bolzonella D, Borzacconi L, Campos JL, Guwy AJ, et al. bacteria. Dordrecht: Springer Netherlands; 1995. p. 973–90. https://doi.org/
Defining the biomethane potential (BMP) of solid organic wastes and energy crops: 10.1007/0-306-47954-0_45.
a proposed protocol for batch assays. Water Sci Technol 2009;59:927–34. https:// [55] Hustede E, Steinbüchel A, Schlegel HG. Relationship between the photoproduction
doi.org/10.2166/wst.2009.040. of hydrogen and the accumulation of PHB in non-sulphur purple bacteria. Appl
[41] APHA. Standard methods for the examination of water and wastewater. 21 ed., vol. Microbiol Biotechnol 1993;39:87–93. https://doi.org/10.1007/BF00166854.
51. Washington: American Public Health Association; 2005. [56] Maeda I, Idehara K, Okayama N, Miura Y, Yagi K, Mizoguchi T. Poly(3-
[42] Lanham AB, Ricardo AR, Coma M, Fradinho J, Carvalheira M, Oehmen A, et al. hydroxybutyrate) as an endogeneous substrate for H2 evolution in Rhodovulum
Optimisation of glycogen quantification in mixed microbial cultures. Bioresour sulfidophilum. Biotechnol Lett 1997;19:1209–12. https://doi.org/10.1023/A:
Technol 2012;118:518–25. https://doi.org/10.1016/j.biortech.2012.05.087. 1018437921006.
[43] Felz S, Al-Zuhairy S, Aarstad OA, van Loosdrecht MCM, Lin YM. Extraction of [57] Tavares Ferreira TJ, Luiz de Sousa Rollemberg S, Nascimento de Barros A,
structural extracellular polymeric substances from aerobic granular sludge. JoVE Machado de Lima JP, Bezerra dos Santos A. Integrated review of resource recovery
2016;2016:1–8. https://doi.org/10.3791/54534. on aerobic granular sludge systems: possibilities and challenges for the application
[44] Moretto G, Russo I, Bolzonella D, Pavan P, Majone M, Valentino F. An urban of the biorefinery concept. J Environ Manag 2021:291. https://doi.org/10.1016/j.
biorefinery for food waste and biological sludge conversion into jenvman.2021.112718.
polyhydroxyalkanoates and biogas. Water Res 2020;170:115371. https://doi.org/ [58] Decho AW, Gutierrez T. Microbial extracellular polymeric substances (EPSs) in
10.1016/j.watres.2019.115371. ocean systems. Front Microbiol 2017;8:1–28. https://doi.org/10.3389/
[45] Romero-Cedillo L, Poggi-Varaldo HM, Ponce-Noyola T, Ríos-Leal E, Ramos- fmicb.2017.00922.
Valdivia AC, Cerda-García Rojas CM, et al. A review of the potential of pretreated [59] Ghosh S, Lahiri D, Nag M, Dey A, Sarkar T, Pathak SK, et al. Bacterial biopolymer:
solids to improve gas biofuels production in the context of an OFMSW biorefinery. its role in pathogenesis to effective biomaterials. Polymers 2021;13:1–28. https://
J Chem Technol Biotechnol 2017;92:937–58. https://doi.org/10.1002/jctb.5116. doi.org/10.3390/polym13081242.
[46] Fradinho JC, Oehmen A, Reis MAM. Photosynthetic mixed culture [60] Zhao L, Bao M, Zhao D, Li F. Correlation between polyhydroxyalkanoates and
polyhydroxyalkanoate (PHA) production from individual and mixed volatile fatty extracellular polymeric substances in the activated sludge biosystems with
acids (VFAs): substrate preferences and co-substrate uptake. J Biotechnol 2014; different carbon to nitrogen ratio. Biochem Eng J 2021;176:108204. https://doi.
185:19–27. https://doi.org/10.1016/j.jbiotec.2014.05.035. org/10.1016/j.bej.2021.108204.
[47] Cabecas Segura P, Onderwater R, Deutschbauer A, Dewasme L, Wattiez R, Leroy B. [61] Larson F, Lubarsky H, Gerbersdorf SU, Paterson DM. Surface adhesion
Study of the production of poly(hydroxybutyrate- co -hydroxyhexanoate) and poly measurements in aquatic biofilms using magnetic particle induction. MagPI.
(hydroxybutyrate- co -hydroxyvalerate- co-hydroxy hexanoate) in Rhodospirillum Limnol Oceanogr: Methods 2009;7:490–7. https://doi.org/10.4319/
rubrum. Appl Environ Microbiol 2022;6. https://doi.org/10.1128/aem.01586-21. lom.2009.7.490.
[48] Sali S, Mackey HR. The application of purple non-sulfur bacteria for microbial [62] Fernández-Dacosta C, Posada JA, Kleerebezem R, Cuellar MC, Ramirez A.
mixed culture polyhydroxyalkanoates production. Rev Environ Sci Biotechnol Microbial community-based polyhydroxyalkanoates (PHAs) production from
2021;7:959–83. https://doi.org/10.1007/s11157-021-09597-7. wastewater: techno-economic analysis and ex-ante environmental assessment.
[49] Nakano S, Fukaya M. Analysis of proteins responsive to acetic acid in Acetobacter: Bioresour Technol 2015. https://doi.org/10.1016/j.biortech.2015.03.025.
molecular mechanisms conferring acetic acid resistance in acetic acid bacteria. Int [63] Kopperi H, Amulya K, Venkata Mohan S. Simultaneous biosynthesis of bacterial
J Food Microbiol 2008;125:54–9. https://doi.org/10.1016/j. polyhydroxybutyrate (PHB) and extracellular polymeric substances (EPS): process
ijfoodmicro.2007.05.015. optimization and Scale-up. Bioresour Technol 2021;341:125735. https://doi.org/
[50] Silva F, Matos M, Pereira B, Ralo C, Pequito D, Marques N, et al. An integrated 10.1016/j.biortech.2021.125735.
process for mixed culture production of 3-hydroxyhexanoate-rich [64] Delamare-Deboutteville J, Batstone DJ, Kawasaki M, Stegman S, Salini M,
polyhydroxyalkanoates from fruit waste. Chem Eng J 2022;427:131908. https:// Tabrett S, et al. Mixed culture purple phototrophic bacteria is an effective fishmeal
doi.org/10.1016/j.cej.2021.131908. replacement in aquaculture. Water Res X 2019;4. https://doi.org/10.1016/j.
[51] Cabecas Segura P, Onderwater R, Deutschbauer A, Dewasme L, Wattiez R, Leroy B. wroa.2019.100031.
Study of the production of poly(hydroxybutyrate- co -hydroxyhexanoate) and poly [65] Alloul A, Wuyts S, Lebeer S, Vlaeminck SE. Volatile fatty acids impacting
(hydroxybutyrate- co -hydroxyvalerate- co- hydroxyhexanoate) in Rhodospirillum phototrophic growth kinetics of purple bacteria: paving the way for protein
rubrum. Appl Environ Microbiol 2022;88. https://doi.org/10.1128/aem.01586- production on fermented wastewater. Water Res 2019;152:138–47. https://doi.
21. org/10.1016/j.watres.2018.12.025.

11