As microorganisms are able to produce de novo aromatic amino acids (AAAs)—i.e., L-Tyrosine (Tyr) and L-Phenylalanine (Phe)- through the shikimate pathway, they are a promising heterologous producer of derived molecules. Hence, a great deal of research has been carried out on this topic, as illustrated by the number of recent reviews from leading researchers in the field (Suástegui and Shao, 2016; Huccetogullari et al., 2019; Averesch and Kayser, 2020; Cao et al., 2020; Liu et al., 2020; Shen et al., 2020; Dickey et al., 2021).

p-Hydroxycinnamic acids are AAA derived molecules and, therefore, can be produced de novo through the heterologous expression of the relevant phenylpropanoid pathway enzymes presented Figure 1 (Vargas-Tah and Gosset, 2015). Chemical synthesis or extraction from plant biomass are two additional routes to access these p-hydroxycinnamic acids; however, they have drawbacks that biotechnological routes do not have. A recent review by Allais and co-workers discusses these three routes, with specific focus on the promising aspects of the biotechnology route (Flourat et al., 2021).

The latter heterologous biosynthesis using microorganisms starts with the deamination of Tyr and Phe by ammonia lyases to produce, respectively, p-coumaric acid (p-CA) and cinnamic acid. Cinnamic acid can then be hydroxylated into p-coumaric acid by a cinnamic acid 4-hydroxylase. p-CA is the first p-hydroxycinnamic acid of the pathway and it has been heterologously produced at industrially relevant titers. In fact, DuPont produced significant amount of p-CA from D-glucose in a two-step patented process using two engineered Escherichia coli strains with the final aim of producing 4-vinylphenol (Ben-Bassat et al., 2005; Sariaslani, 2007). Moreover, Liu et al. in 2019 reached a final production of 12.5 g.L⁻¹ of p-CA from D-glucose in a fed-batch fermentation using an engineered Saccharomyces cerevisiae strain. Although p-CA production has been reported in various other promising microorganisms including Pichia pastoris (Chen et al., 2021), Pseudomonas putida (Nijkamp et al., 2007; Calero et al., 2016), (Nijkamp et al., 2007; Calero et al., 2016), a S. cerevisiae that can use xylose as a sole carbon source (Borja et al., 2019) and Yarrowia lipolytica (Gu et al., 2020), the productions did not reach an industrially attractive titer yet and need further development.

The expression of a p-CA 3-hydroxylase allows the formation of caffeic acid from p-CA. The highest de novo bioproductions are, once again, with engineered E. coli and S. cerevisiae strains: 766.7 mg L⁻¹ and 569.0 mg L⁻¹ of caffeic acid, respectively (Huang et al., 2013; Zhou et al., 2021).

Following this, a caffeic acid O-methyltransferase allows the bioconversion of caffeic acid into ferulic acid. Kang et al., in 2012, (Kang et al., 2012), obtained the highest de novo ferulic acid biosynthesis published to-date: 196 mg L⁻¹ from D-glucose with an engineered E. coli. Since this finding, most of the work focuses on producing de novo ferulic acid-derived molecules.

To the best of our knowledge, no work has been published yet on de novo heterologous production of 5-hydroxyferulic acid and sinapic acid in any microorganism.

To summarise, most of the work concerning de novo bioproduction of p-hydroxycinnamic acids focuses on rewiring carbon flux towards Tyr and Phe synthesis through metabolism engineering, deletion of competing pathways, relief of feedback inhibitions, overexpression of rate-limiting enzymes and heterologous expressions of the relevant genes. However, some limitations are not straightforward to address with such strategies. Indeed, since hydroxycinnamic acids are hydrophobic compounds, their solubility in aqueous solutions such as fermentation media is low (Mota et al., 2008), they are also known to be antimicrobial agents (Alves et al., 2013; Pernin et al., 2019), hence, their accumulation in the broth leads to toxicity towards the producer microorganism (inhibition). In a previous study, a different strategy was proposed to simultaneously address all these concerns, by focusing on the intensification through process optimization and, more precisely, on the implementation of an in situ product recovery (ISPR) (Combes et al., 2021). The continuous recovery of heterologously synthesized p-CA through liquid-liquid extraction was implemented (Figure 2), enabling the detoxification of the medium and the enhancement of accumulation limits. This ISPR process coupled to fermentation is called biphasic fermentation, and in this particular case, oleyl alcohol was used as the extractant solvent.

Within this work, an endogenous decarboxylation of p-CA into 4-vinylphenol from the engineered S. cerevisiae strain was highlighted. The continuous removal of p-CA prevented this competitive pathway as well, intensifying further p-CA de novo production. This work demonstrates the importance and intensification capacities of process optimization for the heterologous microbial production of p-hydroxycinnamic acids. This approach will help engineer a viable biotechnological process, as it also eases the separation steps.

Through strain engineering using the many tools and strategies of molecular biology and using process engineering and intensification, the heterologous bioproduction of p-hydroxycinnamic acids appears to be a promising, efficient and sustainable route (Krömer et al., 2020). p-Hydroxycinnamic acids are platform molecules and notably, allow easy access to the corresponding p-hydroxybenzaldehydes as detailed in the next section.

Vanillin (4-hydroxy-3-methoxybenzaldehyde), the most important flavoring agent in the industry, is a p-hydroxybenzaldehyde derived from ferulic acid. The significance and market size of vanillin, and especially natural vanillin, has led to very active research towards its biotechnological production, as extraction from plant biomass could not meet the market demand (Priefert et al., 2001). One popular strategy to biotechnologically produce natural vanillin is to use ferulic acid as substrate of the biotransformation. Some microorganisms are able to metabolize ferulic acid, and vanilla is one of the degradation intermediates. Indeed, since 2000, Rhodia (now Solvay) uses a wild Streptomyces setonii strain to bioconvert Rhovanil Natural (natural vanillin) from ferulic acid. The latter was obtained through a technology developed by Givaudan comprising the metabolism of ferulic acid by the strain, and the separation of the different products including vanilla, up to 16 g L⁻¹ (Muheim et al., 1998). Although the main strategy consists in using strains able to metabolize ferulic acid into vanillin, other strategies exist, including: 1) Genetically engineering those strains to increase their capacities; 2) heterologously expressing the enzymes of interest in other strains; and 3) producing de novo vanillin, combining previous detailed pathway (Section 2.1) and the enzymes enabling the biotransformation of ferulic acid into vanillin (Martău et al., 2021). Many reviews cover published strategies of successful ferulic acid biotransformations into vanillin and we refer the reader to these for further information (Priefert et al., 2001; Walton et al., 2003; Gallage and Møller, 2015; Martău et al., 2021).

For other corresponding p-hydroxybenzaldehydes, there is less published work, due to smaller industrial interest. Yet, there is evidence that the aforementioned strategy works for p-CA and sinapic acid biotransformation into their corresponding p-hydroxybenzaldehyde form, p-hydroxybenzaldehyde and syringaldehyde, respectively (Estrada Alvarado et al., 2001; Nimura et al., 2010).

Unfortunately, there are two major bottlenecks for this biotechnological approach: The formation of undesired by-products (e.g., oxidation of p-hydroxybenzaldehydes), often overcome by deletion of the enzymes of interest, and the toxicity of the p-hydroxybenzaldehyde. The latter has been overcome first by choosing tolerant strains, but also by the implementation of an in-stream product recovery process to the biotransformation (Hua et al., 2007; Sciubba et al., 2009). Such implementation could also resolve the issue related to competitive metabolic pathway as previously described for the case of p-CA decarboxylation and the biphasic fermentation study (Section 2.1).

To summarize, in one-step or two-step biotechnological approaches, p-hydroxybenzaldehydes can be produced from glucose, and it is well established for the ferulic acid-vanillin route. For the other p-hydroxycinnamic acids, there is still much work to be done to create viable routes, as they remain at the proof-of-concept stage. In other words, these biotechnological processes open the way for the production of novel UV-filters from simple sugar using engineered microorganisms.

Now that the bioproduction of p-hydroxybenzaldehydes has been addressed, we move to discuss the synthesis of UV-filters from these biotechnologically-generated phenolics building blocks.

Antioxidants are essential to prevent damage from UV radiation exposure by neutralizing the Reactive Oxygen Species (ROS) that may form (Krutmann, 2003; Liebel et al., 2012). ROS, in the form of free radicals, are highly reactive species capable of degrading materials and inducing cellular damage in living organisms (i.e., inflammation, oxidative stress or even carcinogenesis) (Pillai et al., 2005; Austin et al., 2018). Phenols, especially the aforementioned p-hydroxycinnamic acids and their derivatives, are known to exhibit interesting antioxidant activity through their high conjugation, facilitating their ability to neutralize free radicals (Peyrot et al., 2020a; Peyrot et al., 2020c; Mention et al., 2020; Mouterde et al., 2020; Rioux et al., 2020; Abiola et al., 2021a; Abiola et al., 2021b). Such molecules have proven to be competitive (Figure 4, notably those derived from caffeic, ferulic and sinapic acid), against the antioxidants conventionally used and that are strongly criticized for their suspected endocrine toxicity and carcinogenic effect like BHA (butylated hydroxyanisole), BHT (butylated hydroxytoluene) or Trolox (6-hydroxy-2,5,7,8-tetramethylchroman-2-carboxylic acid). With their ability to scavenge ROS, paired with a capacity to absorb UV radiation, these p-hydroxycinnamic acids derivatives are compounds of choice to protect from UV radiation exposure and its consequences.

Excessive exposure to the sun can also induce hyperpigmentation in the form of age spots on the exposed skin (Vashi and Kundu, 2013). This negative effect caused by cell degeneration is due to an abnormal production of melanin, a pigment usually produced to tan the skin and protect it against UV radiation exposure. One way to reduce the effect of this disorder is to inhibit tyrosinase activity, responsible for the production of melanin during melanogenesis. Tyrosinase inhibition, widely described in the literature, can be carried out using fungal tyrosinase to mimic human tyrosinase, allowing the in vitro identification of potential inhibitors (Chakraborty et al., 1998; Chawla et al., 2008; Minsat et al., 2021). By using Kojic acid as a reference (Neeley et al., 2009; Minsat et al., 2021), some compounds proved to have great potential as tyrosinase inhibitors, in particular p-hydroxycinnamyl Meldrum’s (6) (Figure 5) (Peyrot et al., 2020b). This secondary activity, paired with their UV-filter activity and antioxidant properties, allows p-hydroxycinnamic acid derivatives to offer further protection against an over exposition to the sun.

As mentioned in the introduction, petroleum-based molecules currently used in formulations are criticized for their potential toxicity, both on humans and the environment. One of the main criteria required for the novel bio-based alternatives to efficiently replace them is to ensure their innocuous nature. Recently, some studies have highlighted the ability of organic UV-filters to be endocrine disruptors (Maipas and Nicolopoulou-Stamati, 2015), which can cause serious negative effects on the central nervous system and reproductive organs (Crews and McLachlan, 2006). As a preliminary step to determine potential toxicity, the interactions between several aforementioned p-hydroxycinnamic acid derivatives and endocrine receptors were evaluated in vitro (Horbury et al., 2019; Peyrot et al., 2020b; Peyrot et al., 2020c; Abiola et al., 2021b). Regardless of the modification introduced on the molecules, no agonist or antagonist interactions toward the different receptors were found (Figure 6), as they kept their normal activity even while in the presence of high concentration of the studied molecules. With the innocuousness determined by this preliminary analysis, combined with the versatile properties of p-hydroxycinnamic acid derivatives toward the protection against UV radiation exposure, those compounds can be considered as serious bio-based alternatives of current organic UV-filters.

After having addressed the bioproduction of p-hydroxybenzaldehydes, their use for the synthesis of new nature-inspired UV-filters, the assessment of their UV-filtering and biological properties as well as endocrine toxicity, we conclude this review by considering how these nature-inspired UV-filters deal with radiation exposure at the molecular level.

As just discussed, geometric isomerization is one of the key non-radiative pathways by which many plant and plant-based UV-filters dissipate potentially damaging UV radiation energy into potentially harmless heat. The first case study we briefly introduce here is one of the earliest studies which provided strong evidence, through ultrafast laser spectroscopy, of E-to-Z isomerization in the model plant UV-filter ethyl sinapate, the molecular structure of which is presented in Figure 7A. In this work, Horbury et al. (2018) studied the dynamical processes in operation following absorption of UV radiation of the two isomeric forms of ethyl sinapate at their absorption peak maxima.

The isomer-specific synthesis developed by Allais and co-workers enabled Horbury et al. to study photoprotection pathways starting from either pure E- or Z-ethyl sinapate (Horbury et al., 2018). The transient absorption spectroscopy (TAS) data obtained from their fs to nanosecond (10^–9 s, ns) TAS setup were virtually identical for both isomers. In order to elucidate the dynamical processes in operation, a sequential global fit was applied resulting in three remarkably similar time-constants (between the two isomers), which were assigned to: 1) a very fast intramolecular vibrational rearrangement, taking place within approximately 300 fs; 2) a ∼5 ps geometric photoisomerisation (i.e. E-to-Z and vice versa); and 3) a long-lived photoproduct whose presence is observed beyond the experimental time window of 2 ns. In essence, these findings suggest that UV irradiation of these sinapate species induces first a fast molecular rearrangement within solute molecules and possibly the solvent surrounding them (300 fs), followed by the photoisomerisation mechanism that allows for dissipation of most excess energy within approximately 5 ps. The photoproduct identified by Horbury et al. relates to the generated isomer, i.e., E- or Z-ethyl sinapate, depending on the starting isomer.

Importantly, Horbury et al., were able to infer that photoisomerization was a crucial dynamical process within photoexcited ethyl sinapate by comparison of the TAS data with steady-state measurements, namely the difference spectrum resulting from subtracting the UV absorption spectrum of the pure E (or Z) ethyl sinapate from the UV absorption spectrum of the same species after 2 hours of irradiation at the molecule’s UV absorption maximum.

Several studies such as the one just presented have established geometric photoisomerization to be an important mechanism for energy dissipation following UV excitation of plant-based filters. However, the different isomers (e.g., E and Z isomers, see Figure 7A) often present different behaviors within a biological environment (Sharma et al., 2017), as well as having different absorption coefficients which, in practice, alters the efficacy of the UV-filter before and after radiation (Horbury et al., 2018). The key aim therefore is to synthesize a compound which prevents the formation of the alternate isomer while also maintaining the effective relaxation mechanism of E-Z isomerization. To this end, Horbury et al. synthesized a geometrically symmetric compound based on the sinapoyl structure, termed diethyl sinapate (Horbury et al., 2019) and shown in Figure 7B. Through the addition of two ester moieties across the acrylic bond, the formation of a geometric isomer is negated leaving only one optimized ground state structure.

Horbury et al. studied the effects of these changes on the excited state dynamics of these molecules using TAS and complementary steady-state methods. The steady-state techniques revealed diethyl sinapate to be highly photostable against solar irradiation, with only 3% of decrease in absorbance observed over a two-hour period of irradiation. The TAS which accompany these results showed that, following photoexcitation, the majority of diethyl sinapate returns to its initial, low energy form within 3 ps. Interestingly, an additional time-constant of 330 fs was required to fully capture the relaxation mechanism. This time constant was not observed in singularly substituted sinapates discussed in the previous case study of ethyl sinapate (Horbury et al., 2018). The authors attribute this time-constant to movement towards the molecular arrangement that allows energy dissipation. Furthermore, the long-lived component (>2 ns) observed in these studies is not assigned to a geometric isomer, as for the sinapates previously discussed (Horbury et al., 2018), but instead to an excited state that remains populated beyond the time-window of the experiment. Horbury et al. also performed TAS of diethyl sinapate on a synthetic skin solvated in alkyl benzoate, which showed the same ultrafast relaxation and therefore provide insight into the behavior of these UV-filters in an environment more comparable to a cosmetic formulation.

The findings of Horbury et al. have shown that the symmetric substitution removes concerns over potentially harmful photoproducts while also maintaining the efficient energy dissipation mechanism that is common among sinapates.

As described, previous work identified geometric photoisomerisation as a key mechanism for energy dissipation in plant-based UV-filters, which then allowed for optimization of this mechanism to avoid significant changes in UV absorption and other properties upon photoisomerization. Building on this work, a novel class of nature-based phenolic compounds was developed with not only further optimized characteristics and photodynamics, but also absorption maxima into the UV-A range of the solar spectrum, a feature that is highly desirable in novel UV-filters (Abiola et al., 2021b). Two of these phenolic compounds, consisting of coumaryl and sinapoyl derivatives with a symmetric substitution around their acrylic bond by means of a Meldrum functional group, are shown in Figure 7C, were studied by Abiola et al. (2021b) employing ultrafast laser spectroscopy and steady-state techniques similar to those used in the studies presented above.

The studies carried out by Abiola et al. revealed that UV irradiation of these phenolic compounds leads to an initial geometry rearrangement along a charge transfer coordinate within approximately 200 fs, followed by isomerization which dissipates most excess energy within ∼450 fs (Abiola et al., 2021b). The new isomer is generated with some residual excess energy, which is dissipated as heat within 4–10 ps. The excess energy is thus almost completely dissipated within 10 ps in these phenolic compounds; there is evidence of some excess energy being trapped beyond 2 ns in the sinapoyl derivatives, but no evidence that this would lead to the generation of any potentially harmful photoproducts. These favorable ultrafast photodynamics results were also confirmed by steady-state measurements, which demonstrated that both the coumaryl and sinapoyl derivatives are highly photostable, losing less than 10% of their UV absorbance after 2 h of irradiation with a solar simulator. Furthermore, these compounds show promising antioxidant properties and no evidence of endocrine disruption effects, which makes these UV-filters suitable candidates for applications in cosmetics, for example (Peyrot et al., 2020b; Abiola et al., 2021b).

The symmetrically substituted phenolic compounds studied by Abiola et al. thus present as ideal UV-filters with strong UV-A absorption, ultrafast energy dissipation and high photostability. These results were obtained both in an industry-standard emollient and on a synthetic skin mimic, which emphasizes the suitability of the phenolic compounds under study for applications in complex environments without lessening their ideal characteristics.

The three case studies presented here demonstrate how gathering a deep understanding of UV-filter photodynamics has guided the molecular design of plant-based compounds that present ideal physico-chemical properties. This type of bottom-up approach to molecular design has proven valuable in real-life applications from photoprotective cosmetics to light-to-heat conversion materials (Abiola et al., 2021a).