The production of plastics, which are essential in all modern economies, is mostly achieved using fossil feedstocks that have high environmental implications. Moreover, worldwide production of plastics is drastically increasing (from 204 million tons in 2002 to 335 million tons in 2016) and is predicted to continue to increase (PlasticsEurope, 2015, 2017). Production of plastics has led to environmental concerns because plastics are not easily degradable and accumulate in the most varied environments (Barnes et al., 2009). Furthermore, despite remarkable efforts to increase recycling rates (in Europe, recycling increased by 79% from 2006 to 2016), only 31.1% of plastics were recycled in Europe in 2016. The remaining plastics are either disposed of in landfills (27.3%) or used for energy recovery (production of electricity or as fuel for industrial processes) (41.6%) (European Bioplastics)¹. Consequently, these “conventional” plastics urgently need to be substituted by biologically produced and/or biodegradable plastics, which are called bioplastics. However, even though it is thought that approximately 85% of the plastic products can be substituted by bioplastics, only approximately 1% of the plastics produced in the world are bioplastics (European Bioplastics)¹. The production of bioplastics is predicted to grow, reaching a production of 2.44 million tons in 2022, of which only 1.086 million tons will be biodegradable (European Bioplastics)². However, this predicted increase is still unsatisfactory compared to the ever-growing global plastic demand (PlasticsEurope, 2017). Polyhydroxyalkanoates (PHAs) are biologically produced polyesters that have been considered extremely valuable alternatives to commonly used petroleum-derived plastics (Madison and Huisman, 1999; Sudesh et al., 2000; Reddy et al., 2003; Verlinden et al., 2007; Akaraonye et al., 2010; Laycock et al., 2014; Li et al., 2016; Mozejko-Ciesielska and Kiewisz, 2016; Dietrich et al., 2017). These molecules accumulate in granules inside bacterial cells and, among other recently reported functions (Koller et al., 2017), provide carbon storage and reducing equivalents (Madison and Huisman, 1999). PHAs possess characteristics similar to common plastics and are biocompatible and biodegradable (Sudesh et al., 2000; Castilho et al., 2009; Laycock et al., 2014; Anjum et al., 2016). Therefore, PHAs are currently being industrially produced and have been used in a broad spectrum of end products, ranging from packaging to medical applications (Akaraonye et al., 2010; Carletto, 2014; Anjum et al., 2016; Mozejko-Ciesielska and Kiewisz, 2016; Dietrich et al., 2017).

Regardless of their excellent features and environmental advantages, more extensive use of PHAs is impaired by their high production costs (Byrom, 1987; Lee, 1996; Reddy et al., 2003; Koller et al., 2007a). One of the main factors contributing to these high costs is the use of expensive fermentation carbon sources, which have been shown to account for approximately 40% of the total PHA production costs (Choi and Lee, 1999; Kim, 2000). Thus, the use of waste materials as carbon sources for microbial-derived PHA production has been proposed in order to simultaneously reduce both PHA production and waste disposable costs (Choi and Lee, 1999; Kim, 2000; Koller et al., 2017; Nielsen et al., 2017). Several waste sources have been used to produce PHAs with relative success (Marshall et al., 2013; Nikodinovic-Runic et al., 2013; Anjum et al., 2016; Koller et al., 2017), including domestic wastewater (Carucci et al., 2001); food waste (Rhu et al., 2003); molasses (Albuquerque et al., 2007; Carvalho et al., 2014); olive oil mill effluents (Dionisi et al., 2005); palm oil mill effluents (Din et al., 2012); tomato cannery water (Liu et al., 2008); lignocellulosic biomass (Bhatia et al., 2019); coffee waste (Bhatia et al., 2018); starch (Bhatia et al., 2015); biodiesel industry waste (Kumar et al., 2014a; Sathiyanarayanan et al., 2017); used cooking oil (Ciesielski et al., 2015; Kourmentza et al., 2017); pea-shells (Patel et al., 2012; Kumar et al., 2014b); paper mill wastewater (Jiang et al., 2012); bio-oil from the fast-pyrolysis of chicken beds (Moita and Lemos, 2012); and cheese whey (Table 1).

Whey constitutes the main by-product of the dairy industry and is obtained by precipitation and removal of milk casein during cheese-making processes. Approximately 120 million tons of whey are produced annually throughout the world, of which only 50% are used for products used in human and animal feed (Nikodinovic-Runic et al., 2013). The remaining whey needs to be disposed of, which raises environmental concerns due to its relatively high organic load, as whey is mainly composed of lactose (39–60 g/L), fats (0.99–10.58 g/L), proteins (27–60 g/L), and mineral salts (4.6–8 g/L) (Prazeres et al., 2012; Colombo et al., 2016). From these cheese whey components, lactose is responsible for most of the biochemical oxygen demand of whey; therefore, it is essential to find a biotechnological use for lactose (Siso, 1996; Prazeres et al., 2012; Carvalho et al., 2013; Nikodinovic-Runic et al., 2013). The value of using whey lactose for PHA production is clear as it would dramatically decrease the costs of PHA production, without competing with the production of food for humans and simultaneously solving an environmental problem.

In this review, current knowledge on the use of cheese whey as a substrate for PHA production is summarized (Figure 1). Moreover, the problems and prospects of using whey as a substrate for industrial scale production of PHAs are pinpointed and discussed.

Whey is a rich media that is clearly suited for microbial growth. However, finding microorganisms that efficiently produce PHAs when growing on whey has been shown to be extremely challenging. Unfortunately, some of the best-described PHA-producing microbial species have been shown to be unable to directly produce PHAs from whey. For instance, Cupriavidus necator [formerly known as Ralstonia eutropha, Wautersia eutropha, or Alcaligenes eutrophus (Vandamme and Coenye, 2004)] is capable of accumulating PHAs to up to 80% of its dry weight when growing on glucose (Ryu et al., 1997; Reinecke and Steinbüchel, 2008), but it is unable to efficiently grow and, consequently, produce PHAs on lactose, the predominant carbon source of whey (Povolo et al., 2010). Another example is the halophilic archaeal bacteria Haloferax mediterranei, which can accumulate PHAs to approximately 60% of its dry weight (Lillo and Rodriguez-Valera, 1990; Quillaguamán et al., 2010). Nevertheless, H. mediterranei, despite being capable of consuming lactose (Fernandez-Castillo et al., 1986), does not efficiently grow and produce PHA using this carbon source (Koller et al., 2007b). One exception is Alcaligenes latus, which is a known efficient PHA producer that can accumulate PHAs to up to 70% of its dry weight using diverse sugar sources (Yu et al., 1999; Gahlawat and Srivastava, 2017). Recently, one study showed that Alcaligenes latus is able to convert whey lactose into PHA (0.11 g L^-1h^-1) (Berwig et al., 2016) (Table 1). While the PHA yield obtained is not comparable to the yields obtained when Alcaligenus latus is grown in other carbon sources, more studies should be conducted to assess the full potential of using A. latus to produce PHAs from whey (Berwig et al., 2016).

The low capacity of highly PHA-producing microorganisms to produce PHA from whey lactose was, in some cases, bypassed by chemically or enzymatically converting whey lactose into glucose and galactose prior to fermentation. Using this hydrolyzed whey lactose, high PHA producers, such as C. necator and H. mediterranei, were shown to be able to produce PHAs (Marangoni et al., 2002; Pais et al., 2016) (Table 1). However, the steps required to convert whey lactose into glucose and galactose add to the final cost of PHAs and thus, from an industrial point of view, should be avoided. These conversion steps have, in fact, already been avoided by deploying genetic engineering techniques. For instance, Escherichia coli cells capable of consuming lactose were modified to express PHA biosynthesis genes from high PHA-producing microorganisms (Ahn et al., 2000, 2001). Alternatively, high producing strains were engineered to express lactose degradation genes (Povolo et al., 2010) (Table 1). While genetic engineering is a highly versatile and promising tool for enhancing PHA production from whey, the use of genetically engineered strains requires more controlled production plants. In fact, benefits and ethics aside, the wide application of genetic modified microorganisms (GMMs) in scientific research lab and biotechnology industry raises the concern of the accidental release of genetically modified species (or strains) into the environment and subsequent gene transfer to microorganisms/non-microorganisms host. In view of this, it has been considered necessary by biosafety regulations of individual countries to regulate their use avoiding any potential risks they may pose. The order requires notification, risk evaluation and approval of work with GMMs and approval of laboratories or plants where work with GMOs is to take place. Laboratories for working with GMMs are classified in four classes each with specific regulations with regard to containment, organization of the laboratories and the laboratory areas, plans for execution of the work, technical facilities, internal supervision, and control (Council Directive 98/81/EC of 26 October 1998 amending Directive 90/219/EEC on the contained use of genetically modified micro-organisms) (Devos et al., 2008; Guo et al., 2019). Therefore, biotechnology laboratories and industries require biosafety measures designed to protect their staff, the population, and the environment, which may be exposed to hazardous microorganisms and materials, as well as qualified staff, increasing the management costs of the plants and consequently the PHA production costs.

Considering all these observations, it is clear that an extremely valuable option for achieving economically competitive production of PHAs from whey would be to discover wild-type bacterial species that can directly produce high levels of PHA from whey lactose. However, currently, besides the abovementioned A. latus (Berwig et al., 2016), only a few wild-type microbes have been shown to be able to directly produce PHA from whey lactose and without exceptional PHA yields (Figure 1 and Table 1). Other microbes have been shown to produce PHAs from lactose-based media, including Bacillus sp. (Thirumala et al., 2010) and Enterobacter cloacae SU-1 (Samrot et al., 2011). Nevertheless, it remains unclear whether these microbes can use whey-derived lactose for PHA production.

Another option for whey-based PHA production that has recently received great attention is the use of mixed microbial cultures (MMCs) instead of single microbial strains (Colombo et al., 2016). While MMCs have been associated with lower yields of PHA production, they do not require sterile conditions and are able to adapt to changing industrial waste complex substrates (Reis et al., 2011; Colombo et al., 2016). MMCs have been used to directly produce PHAs from cheese whey lactose or after a first digestion of lactose by a different MMC (Carletto, 2014; Duque et al., 2014; Valentino et al., 2015; Colombo et al., 2016) (Table 1). MMCs are obtained from waste sources and are normally subsequently enriched for PHA production using feast and famine cycles designed to increase the presence of strains that are capable of creating PHA-based carbon reservoirs (Nielsen et al., 2017). Despite their great flexibility and cost advantages, using an unknown microbial community to produce PHAs can raise questions regarding the production efficiency, consistency of the polymer characteristics and biocompatibility when PHAs are to be used in medical applications. Nevertheless, the use of MMCs opens interesting prospects for identifying new PHA-producing microorganisms from these communities. Studies on MMCs showed that they consist of diverse bacterial genera, which change according to the tested fermentation carbon source (Janarthanan et al., 2016); however, a study in which single organisms were isolated for PHA production from whey has not, to our knowledge, been performed to date. Another possibility that would include the advantages of both MMCs and single strain fermentation would be to use MMCs to convert whey lactose into other carbon sources (such as organic acids) and then use a high PHA producer single strain to convert these organic acids into PHAs. This option has been proposed (Graziani and Fornasiero, 2007), but it has not yet been attempted for the production of PHAs from whey.

Whey is a substrate that has numerous difficulties for direct use in PHA production. First, whey has been reported to have a low C-N rate, which greatly hinders PHA production (Bosco and Chiampo, 2010). Additionally, as previously discussed, lactose is not the preferred sugar of PHA-producing strains. Moreover, whey is a complex, unsterile and often variable by-product, and thus, its direct application in both lab-scale experiments and industrial production pipelines results in an abundance of other difficulties. Therefore, many studies that report PHA production from whey have actually been performed on whey derivatives that have undergone a series of pre-treatments (Figure 1 and Table 1) instead of whey itself. One of these derivatives is whey powder, which is diluted in water prior to fermentation to obtain a specific lactose concentration. Whey powder, normally produced by spray-drying, allows longer storage of whey as well as avoids seasonal variations of this by-product (Písecký, 2005; Chegini and Taheri, 2013). Nevertheless, a spray-drying step will increase costs in a PHA production pipeline. Thus, while acknowledging the possible consistency-based reasoning behind using whey powder for screening and optimizing PHA production, it remains unclear whether fermentations using whey are comparable to those performed with a whey powder.

The other extensively used pre-treatment of whey, regardless of whether it is a liquid or powder, is the removal of the majority of its proteins and other solids by ultrafiltration, creating a whey permeate that retains most of the whey lactose (Koller et al., 2007b). In addition to ultrafiltration, this protein removal is also commonly obtained by acidifying whey to a pH close to 4 followed by a heat treatment, centrifugation and filtration, generating a so-called whey supernatant (Pantazaki et al., 2009). These whey permeates or supernatants are easier to work with, as they are sterile, homogeneous and clear solutions; however, their use increases the production costs. Moreover, it was shown that whole whey and whey supernatant, used as carbon sources in a complex mineral media, led to different biomass growth and PHA accumulation in Methylobacterium sp. ZP24 (Yellore and Desai, 1998). Therefore, attention should be paid when extrapolating results obtained from whey permeates or supernatants to whole whey.

Other types of additives, which are routinely used when dealing with whey in PHA production studies, consist of minerals and nitrogen sources (Table 1). The use of these additives could probably be avoided by using whole whey instead of whey permeates or supernatants. However, for some studies, it remains to be seen whether the used microorganisms are capable of accumulating high amounts of PHAs using whey without additives, which would be ideal for industrial production. For Bacillus megaterium, the addition of salts ((NH₄)₂SO₄, 5 g L^-1; Na₂HPO₄, 2.5 g L^-1; KH₂PO₄, 2.5 g L^-1; MgSO₄, 0.2 g L^-1; and MnSO₄, 0.01 g L^-1) to whey supernatant increased its biomass and PHA production by almost 10 times (Obruca et al., 2011). Therefore, if any addition should be made to whey for enhanced PHA production, it would be interesting to explore the use of other inexpensive waste products as whey additives; this point will be discussed subsequently in this paper.

Whole whey should be more widely used to study PHA production. However, the use of whole whey raises important experimental challenges, including sterility issues. These issues are not as relevant when using MMCs; however, for single cultures, sterilization is crucial (Koller et al., 2011). Sterilization of whole whey with high temperatures cannot be performed as total whey protein will precipitate, and filtration is difficult due to the high quantity of suspended solids present in whey. Low-temperature pasteurization could be used, but it normally involves several cycles of heat and cold treatments that result in expensive and time-consuming processes (Ghaly and Kamal, 2004). Moreover, pasteurization does not assure full sterilization, which can lead to sterility problems during fermentation processes. Therefore, alternative sterilization techniques, such as the use of ultraviolet radiation or high temperature-short time pasteurization, have been proposed to be used prior to fermentation with pure strains in whole whey (Morr, 1987; Ghaly and Kamal, 2004). Furthermore, the use of antibiotics has also been attempted to solve these issues. Vancomycin was shown to reduce Bacillus cereus contamination in Hydrogenophaga pseudoflava PHA production from whey without significantly affecting the PHA production efficiency (Koller et al., 2011). Another problem that arises by using untreated whole whey is the high lactose concentration. It was shown that the whey lactose concentration was crucial for biomass and PHA production by B. megaterium CCM 2037. In fact, while growing on whey supernatant with a lactose concentration of 40 g/L, this strain grew to approximately 1.5 g/L biomass and produced less than 0.5 g/L PHAs; when grown in a diluted whey supernatant with a concentration of 20 g/L, the production values increased to 2.51 g/L biomass and produced 0.79 g/L PHAs (Obruca et al., 2011).

In addition to these whey pre-treatments, for specific microorganisms, as previously mentioned, transformation of whey lactose into galactose and glucose needs to be performed prior to fermentation (Table 1). Some authors have avoided lactose hydrolysis by using media that mimics hydrolyzed whey containing galactose and glucose as carbon sources (Wallner et al., 2001; Koller et al., 2011). The majority of the remaining studies proceeded to hydrolysis using β-galactosidase enzymes and showed that microbes were able to use glucose and galactose originating from enzymatic hydrolysis of cheese whey lactose (Marangoni et al., 2002; Koller et al., 2005, 2007a,b, 2008; Koller, 2015). However, the use of enzymatic hydrolysis greatly increases the costs of a possible PHA production plan. Thus, it is necessary to highlight that it has also been shown that chemical lactose hydrolysis using HCl resulted in whey lactose hydrolysates that could be used for microbial production of PHAs (Pais et al., 2016; Kucera et al., 2018). Nevertheless, preferably, these treatments should be avoided by finding either wild-type microorganisms that accumulate high levels of PHAs using lactose as a carbon source or microorganisms that convert lactose into more widely used carbon sources for PHA production.

The use of MMCs avoids most of these problems, as, in theory, a community is able to thrive and evolve in diverse environments and still produce PHAs. However, this hypothesis has not been widely tested, as most studies used whey powder derivatives to test MMC production of PHAs (Table 1). In addition to whey powder, whey permeate has also been used (Carletto, 2014) as well as the supernatant of whey, which has been subjected to a first fermentation to transform whey lactose (Colombo et al., 2016). Interestingly, the use of these whey supernatants yielded more PHA production than those obtained by using whey powder and whey permeates (Colombo et al., 2016). These results led us to believe that the use of MMCs may be the correct approach for producing PHAs from whole whey. However, these results also suggest that there are considerable PHA productivity differences when using diverse whey derivatives. Therefore, MMCs should be tested for PHA production using whole whey.

Finally, most of the current knowledge on using whey for PHA production is based on studies that use whey derivatives to ensure experimental consistency and ease (Table 1). However, it is now time to use whole whey, as this is crucial for obtaining a comprehensive view of the actual feasibility of using whey for PHA production in an industrially viable manner.

In this review, the focus is on two important factors that need to be taken into consideration when aiming at achieving an economically viable production of PHA from cheese whey: microorganism choice and whey pre-treatment requirements. While acknowledging that not all factors can be reviewed here, it is necessary to mention, nonetheless, that some other important aspects need to be considered to achieve efficient production of PHAs from whey.

Economically and environmentally viable production of bioplastics is required to solve the environmental problems that arise from the production and accumulation of conventional plastics. PHAs could play an important role in this bioplastic solution because they have market-desired properties and can be biologically produced from inexpensive waste substrates, such as whey. However, there are still numerous challenges that need to be overcome to achieve efficient production of PHAs from whey. In this work, the studies that describe PHA production from whey focusing on whey pre-treatments and the choice of producing microorganisms have been summarized (Table 1). Our current knowledge on PHA production from whey provides clear indications that producing PHA from whey is possible and can be achieved using diverse strategies. Notwithstanding, this review also highlights that there is a shortage of studies that aim at using whey to produce PHAs at a large scale and consider the costs and environmental impacts of future industrial producing plants. Therefore, this study could motivate and guide future studies focusing on obtaining economically viable production of PHAs from whey.

TA was responsible for data collection, wrote the first draft of the manuscript, and revised the article. DR and GC revised the manuscript. LI was responsible for the conception and design of the manuscript and revised it critically.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.