Tailored molecular weight hyaluronic acid production by engineered Lactococcus lactis

Abstract

Background

Hyaluronic acid (HA) is a glycosaminoglycan with a wide range of biological functions that depend on its molecular weight (MW). Recently, there has been an increasing interest in producing HA at particular MWs for various cosmetic and biomedical applications. HA is traditionally produced by extraction or microbial fermentation, which is then subjected to chemical or enzymatic treatments to customize the MW. On the other hand, direct microbial synthesis at desired MWs has considerable advantages over conventional techniques. The present study introduces a combinatorial approach using four critical variables which influence the molecular weight of HA (MW_HA): (1) Expression of HA synthases from different Streptococcus species (S. parauberis, S. uberis, S. zooepidemicus, and S. pyogenes) which intrinsically produce different MW_HA; (2) Supply of HA precursors by varying heterlogous gene expression in the HA-precursor pathways (hasAB vs. hasABE); (3) Re-routing of metabolic fluxes by deletion of the lactate dehydrogenase (ldh) gene; and (4) Varying the initial glucose concentration in batch fermentation.

Results

Recombinant Lactococcus lactis strains expressing HA synthase genes taken from diverse Streptococcal sp. were found to produce varying MW_HA under otherwise identical genetic and bioreactor conditions. The HA synthases sourced from S. uberis and S. parauberis synthesized higher MW_HA, whereas those from S. pyogenes produced lower MW_HA. In silico analysis of the HA synthase sequences indicated that differences in the transmembrane regions among the various isoforms are the probable cause of variations in MW_HA. Compared to their wild-type counterparts, ldh-knockout L. lactis strains showed a noticeable increase in MW_HA due to a substantial increase in HA precursor levels. Further, the co-expression of hasE in addition to hasAB, considerably increased MW_HA due to a better balance of the intracellular HA-precursor ratios. This multiplexing approach, involving simultaneous manipulation of the above factors, allowed us to produce HA with tailored MW_HA over a broad range from 0.2 to 2.6 MDa.

Conclusion

Our technology eliminates the need for enzymatic desizing or post-processing of HA to achieve the desired MW_HA. In summary, this multiplexing approach enables one-pot synthesis of desired MW_HA, opening up new avenues for producing customized HA.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12934-026-02945-8.

Introduction

Hyaluronic acid (HA) is a glycosaminoglycan composed of alternating N-acetylglucosamine (UDP-GlcNAc) and glucuronic acid (UDP-GlcUA) units. It has a wide range of functions, such as the regulation of cell differentiation [1], signaling of immune receptors [2], promotion of angiogenesis [3], regulation of inflammation and cell proliferation in wound healing [4], and the lubrication of cartilage in joints [5]. These diverse biological functions arise from the HA’s ability to interact with specific receptors and trigger the activation of distinct signaling pathways [6]. A few research studies on the impact of different MW_HA in various biomedical applications are highlighted below. The presence of HA with different MWs in various body fluids, including blood serum, lymph, synovial fluid, and vitreous humor, underscores the importance of the molecular weight of HA (MW_HA) in facilitating these diverse functions [7–12]. For example, in the process of wound healing, Yang et al. (2020) distinguished the effects of two different MW_HA [4]: HA above 400 kDa contributes to hemostasis, whereas lower MW fragments (< 120 kDa) are associated with inflammation. Very high-MW_HA is predominantly present in synovial fluid, which functions as a lubricant and protects joint tissues from shear-induced damage [13].

In osteoarthritis treatment, several commercial formulations are available and administered for pain relief [6]; however, high-MW_HA (> 3000 kDa) is known to have greater efficacy with a lower reinjection frequency [14, 15]. In ophthalmic procedures, MW_HA is a significant factor to consider alongside concentration and ionic strength. However, few studies suggest that HA of ~ 500 kDa is preferred in eye drop formulations because of its persistence on the ocular surface and better hydration properties [16, 17]. Similarly, in cosmetics, HA of < 300 kDa is preferred for its ability to penetrate through the skin and act as a hydrating agent with antiaging effects [18]. In drug delivery, 200 kDa HA-linked conjugates are reported to have better binding ability with CD44 receptors, thereby enhancing their cytotoxic effects in cancer treatment [19]. While low-MW_HA (< 100 kDa) accelerates the initial phase of wound healing, combining it with high-MW_HA has also been shown to improve the overall repair process [20]. To our understanding, many applications can accept a range of MWs. The aforementioned examples are intended to highlight the importance of MW_HA and its distinctive role in different applications. MW_HA being a crucial quality parameter, it is also important to choose the right MW_HA for optimal performance in respective applications.

Commercial HA is predominantly produced through microbial fermentation of Streptococcus species, which are natural producers of high-MW_HA [21]. The HA produced is then subjected to chemical or enzymatic treatments to achieve targeted MW_HA [22]. While this approach mitigates zoonotic contamination risks associated with animal-derived sources (e.g., rooster combs, bovine vitreous humor), it introduces challenges such as the coproduction of endotoxins, which increases safety concerns [23]. Hence, metabolic engineering strategies have been explored with generally-recognized-as-safe (GRAS) organisms such as Bacillus subtilis, Corynebacterium glutamicum, and Lactococcus lactis. B. subtilis was the first recombinant organism engineered for HA synthesis, highlighting the critical role of heterologous genes in the microbial HA biosynthetic pathway [24]. Other engineered organisms, including Pichia pastoris, Escherichia coli, and L. lactis, typically achieve HA titers ranging from 1.2 to 1.7 g/L. Recent advancements, such as the co-expression of leech hyaluronidase, have substantially improved HA titers in fed-batch cultures, up to 19 g/L in B. subtilis [25] and up to 105 g/L in C. glutamicum [26]. HA was produced as low-MW_HA (< 10 kDa) in the latter two instances. While most of these reports have focused on enhancing HA titers, only a few emphasize MW_HA. Notably, L. lactis strains have been engineered to produce 4.6 g/L HA at ~ 1.2 MDa in batch cultures [27] and 3.2 g/L HA at ~ 3.4 MDa in fed-batch cultures [28]. Additionally, B. subtilis strains have yielded up to 3.7 g/L of ~ 3 MDa HA [29]. The increasing interest in MW_HA-specific applications has shifted focus toward producing HA with specific MWs. The regulation of leech hyaluronidase (HAase) expression in B. subtilis allowed the HA production to particular values of MW_HA ranging from 6.6 kDa to 1.4 MDa, where increased HAase activity reduced the MW_HA [25]. The drawbacks of such methods are the random and nonspecific cleavage of HA, which could lead to an increased polydispersity of the HA polymer. A more effective approach is synthesizing HA in vivo at the desired MW, avoiding the post-synthesis complications. For example, engineering Streptococcal HA synthase (HAS) transmembrane domains in recombinant C. glutamicum allowed in vivo HA synthesis with tunable MWs ranging from 300 kDa to 1.4 MDa [26]. This manuscript outlines a multi-factorial approach for the microbial production of HA with tailored MW.

Understanding the multifactorial determinants of HA molecular weight

To enable microbial production of HA at specific MWs, a detailed understanding of the factors that influence the MW_HA during synthesis is essential. These factors include the carbon flux distribution in the biosynthetic pathway of HA-precursors (UDP-GlcUA and UDP-GlcNAc) [30–32], the intracellular concentration, the ratio of the two HA-precursors [27], cofactor requirements [28, 33], the MW-specificity of the membrane-bound HA synthase [34], culture conditions [21, 27], etc. These factors can be categorized into two groups: one based on the polymerizing enzyme HAS and its impact on membrane integrity, and the second based on the metabolic regulation of HA-precursor pathways.

The most influential factor affecting the MW_HA is the HA synthase properties, such as polymerization kinetics, enzyme half-life, translocation efficiency, and the binding affinity of the precursors [35–37]. This synthase dependence was demonstrated by Schulte et al. [34], where the HA synthase (hasA) genes from diverse Streptococcus sp. expressed in L. lactis NZ9000 were found to produce widely different MW_HA, ranging from 0.4 to 2.1 MDa. Since all the other parameters were kept identical in these experiments, the differences in MW_HA could be attributed solely to the genetic variations in the polymerizing enzyme. There are also reports that enhancing the binding of HA synthase to the cell membrane by overexpressing cardiolipin improves membrane integrity, thus assisting in the synthesis of HA [38].

Recombinant hosts often carry HA-pathway genes (hasB, hasC, hasD, and hasE) as paralogs with distinct cellular functions [39]. Figure 1 illustrates the biochemical pathway of HA synthesis, highlighting the essential genes involved. The arrangement of these pathway genes in a single operon in the native producer Streptococcus sp. highlights the prominent involvement of these genes in HA synthesis.

Fig. 1.

Biosynthetic pathway for HA production in L. lactis. The highlighted pathway genes are crucial for HA biosynthesis and are present in a single operon in S. zooepidemicus

Due to the limited expression of homologous HA-pathway genes in a recombinant host [40], the overexpression of these genes is crucial for HA production. The precursor availability for HA biosynthesis is influenced by factors such as the overexpression of additional pathway genes [41, 42], downregulation or knockout of the genes involved in the competing metabolic pathways [33], and by altering the gene expression ratio of hasA to other pathway genes [43]. Cofactor availability (NAD + and acetyl-CoA) also influences the synthesis of HA precursors and their ratios, thereby influencing the MW_HA [28, 33]. In addition to genetic factors, culture conditions influence HA precursor fluxes and concentrations, thereby affecting the MW_HA. Parameters such as initial glucose concentration, pH, and C: N ratio will impact the specific growth rate, which is inversely correlated with MW_HA [30, 44, 45]. Supplementation with co-substrates such as acetate and glutamine has also been shown to enhance precursor availability [28, 46].

This study presents the production of HA with tailored MWs using recombinant L. lactis strains. Our approach involves multiplexing of various genetic factors that influence the MW_HA, such as HA synthase (hasA) genes from different Streptococcal species; co-expression of varying gene combinations (hasAB or hasABE) in the HA-precursor biosynthetic pathways; and re-routing of carbon flux (by using a Δldh strain), which also improves cofactor availability for HA-precursor synthesis. Finally, these genetic factors were also varied with a bioprocess parameter, viz. initial glucose concentration, which affects glucose uptake and distribution between the central carbon and HA biosynthetic pathways. We demonstrate the individual impact of each factor on the MW_HA produced during the fermentation of the recombinant L. lactis strains. By choosing various combinations of these four factors, we produce HA at specific MWs ranging from 0.2 to 2.6 MDa. This multiplexing of enzymatic, metabolic, and process factors for synthesizing customized MW_HA underscores the versatility of L. lactis as a robust platform for producing HA for diverse biomedical and industrial applications.

Materials and methods

Bacterial strains and vectors

Lactococcus lactis strains NZ9000 and NZ9020, obtained from NIZO Food Research (The Netherlands), were the hosts for recombinant HA production. Both strains carry chromosomally integrated nisRK genes induced by the nisin-controlled gene expression system (NICE®) [47]. The L. lactis NZ9020 strain is a lactate dehydrogenase (ldh) mutant in which two ldh genes have been knocked out [48]. Escherichia coli MC1061 was used as the cloning host, and appropriate antibiotics were added for selection. The expression vector pNZ8148, containing the PnisA promoter and a chloramphenicol resistance marker, was used for cloning [49].

Construction of recombinant strains

The plasmid constructs used in this study are detailed in Table 1. The hasE gene from S. equi subsp. zooepidemicus ATCC35246 was amplified from the genomic DNA using the Phusion™ High Fidelity DNA Polymerase and cloned into pNZ8148-paB (Table 1) via the conventional restriction digestion and ligation method [42]. The resulting ligated products were then transformed into ultracompetent E. coli MC1061 and plated on LB (Lysogeny Broth) agar plates containing chloramphenicol (10 µg/mL) and streptomycin (10 µg/mL). A set of additional clones with the overexpression of hasE were constructed using Gibson assembly, with primers for inserts and backbone adapted from Schulte et al. [34]. After E. coli transformation, positive recombinants were screened using colony PCR and confirmed by DNA sequencing. Plasmids were transformed into competent L. lactis via electroporation and screened with appropriate antibiotic markers.

Table 1.

List of strains and plasmids used in this study. For clarity and ease of comparison, some strain labels have been modified from those reported by Schulte et al. [34], and the original labels are listed in supplementary table S3

Name	Description	Origin
pNZ8148	Nisin-inducible promoter, chloramphenicol resistance	Nizo Food Research (The Netherlands)
pNZ8148-paB	pNZ8148 vector containing hasA from Streptococcus parauberis and hasB from Streptococcus zooepidemicus	[34]
pNZ8148-uB	pNZ8148 vector containing hasA from Streptococcus uberis and hasB from Streptococcus zooepidemicus	[34]
pNZ8148-zB	pNZ8148 vector containing hasA and hasB from Streptococcus zooepidemicus	[34]
pNZ8148-pB	pNZ8148 vector containing hasA from Streptococcus pyogenes and hasB from Streptococcus zooepidemicus	[34]
pNZ8148-paBE	pNZ8148 vector containing hasA from Streptococcus parauberis, hasB and hasE from Streptococcus zooepidemicus	This study
pNZ8148-uBE	pNZ8148 vector containing hasA from Streptococcus uberis, hasB and hasE from Streptococcus zooepidemicus	This study
pNZ8148-zBE	pNZ8148 vector containing hasA, hasB, and hasE from Streptococcus zooepidemicus	[42]
pNZ8148-pBE	pNZ8148 vector containing hasA from Streptococcus pyogenes, hasB and hasE from Streptococcus zooepidemicus	This study
L. lactis NZ9000	pepN::nisRnisK genome integrated	Nizo food research
L. lactis NZ9020	pepN::nisRnisK genome integrated and Δldh, ΔldhB are knocked out of the genome	Nizo food research
L. lactis pauAB	L. lactis NZ9000, carrying pNZ8148-paB	[34]
L. lactis uAB	L. lactis NZ9000, carrying pNZ8148-uB	[34]
L. lactis zooAB	L. lactis NZ9000, carrying pNZ8148-zB	[34]
L. lactis pyAB	L. lactis NZ9000, carrying pNZ8148-pB	[34]
L. lactis pauABE	L. lactis NZ9000, carrying pNZ8148-paBE	This study
L. lactis pyABE	L. lactis NZ9000, carrying pNZ8148-pBE	This study
L. lactis pauABEΔldh	L. lactis NZ9020, carrying pNZ8148-paBE	This study
L. lactis uABEΔldh	L. lactis NZ9020, carrying pNZ8148-uBE	This study
L. lactis zooABEΔldh	L. lactis NZ9020, carrying pNZ8148-zBE	[27]
L. lactis pyABEΔldh	L. lactis NZ9020, carrying pNZ8148-pBE	This study

Bioreactor cultivation

HA production was performed in an unaerated batch bioreactor. Experiments were conducted in an in-situ sterilizable 2.4-L bioreactor (PasserA-2000, Biojenik Engineering, India) with a 1.2 L working volume at 30 °C, 200 rpm, and pH 7. Modified M17 media was used with the following composition: yeast extract: 5 g/L, brain heart infusion: 5 g/L, MgSO₄.7H₂O: 0.5 g/L, ascorbic acid: 0.5 g/L, KH₂PO₄: 0.5 g/L, and K₂HPO₄: 1.5 g/L. Unless otherwise stated, the initial glucose concentration was set to 30 g/L in all the experiments. The antibiotics chloramphenicol (10 µg/mL), tetracycline (2 µg/mL), and erythromycin (10 µg/mL) were added as per the selection marker requirements. The inoculum was prepared from glycerol stocks (-80 °C) by subculturing in a static flask. The inoculum was aseptically introduced into the bioreactor to reach a final biomass concentration of 0.02 to 0.04 OD₆₀₀. Once the culture reached the early log phase (~ 0.6 OD₆₀₀), the HA production was induced by adding nisin (2 ng/mL) into the bioreactor. Samples were collected at regular intervals (2–3 h) in all bioreactor experiments and stored at -80 °C for further analysis.

Analytical techniques

Estimation of biomass concentration

The fermentation broth was treated with an equal volume of 0.1% SDS for 10 min at room temperature with gentle mixing and subsequently centrifuged at 10,000×g for 10 min. The pellet was resuspended in 0.9% saline solution, and the absorbance was measured at 600 nm via a UV-visible spectrophotometer (Jasco, US). A correlation between dry cell concentrations and optical density (OD₆₀₀) was established, and the resulting data were used for process calculations.

HA concentration and molecular weight

HA concentration was determined using the CTAB turbidimetric method [50]. The fermentation broth was treated with an equal volume of 0.1% SDS for 10 min at room temperature and centrifuged at 10,000×g for 10 min. The resulting supernatant was used for the CTAB assay [34]. For the MW_HA measurements, the same supernatant was mixed with four volumes of isopropanol and incubated overnight. The mixture was centrifuged at 8000×g for 15 min, and the HA pellet was redissolved in the mobile phase for analysis by size-exclusion chromatography (SEC). A Phenomenex Poly Sep-GFC-P 6000 size-exclusion column (300 × 7.8 mm) was attached to the Shimadzu HPLC system with a refractive index detector, and a 0.2 N NaNO₃ mobile phase was used at 0.6 mL/min. A calibration curve for MW_HA estimation via SEC was carried out with varying MW_HA (0.11, 0.73, 1.59, 2.67 MDa, Lifecore Biomedicals, USA). A regression equation [51] was applied to correct for hydrodynamic volume effects and improve the accuracy of MW_HA estimation.

HA precursor quantification

UDP-N-acetylglucosamine and UDP-glucuronic acid were quantified according to the method described by Franke et al. [52]. The Samples were collected at mid-log phase (~ 5 OD₆₀₀) and immediately quenched by freezing at -80 °C. After treatment with 0.05% SDS, cell pellets were obtained by centrifugation. The pellets were then resuspended in 80% methanol with 0.1 M HEPES buffer, pH 7, and subjected to sonication. The resulting supernatant was lyophilized, reconstituted in the mobile phase, and analyzed on a strong anion exchange column (Shodex QA-825) using a Shimadzu HPLC system equipped with a photodiode array detector [52]. Gradient elution was applied to the mobile phase (1 mL/min), consisting of eluent A (10 mM NaCl) and eluent B (250 mM NaCl); eluent B was increased from 2% to 100% over 50 min, reduced to 2% over 2 min, and maintained at 2% for 8 min. UDP-sugar concentrations were determined by measuring the area under each chromatographic peak and correlating it with a standard calibration curve.

Statistical analysis

All experiments were performed at least in triplicate, and results were expressed as a mean ± standard deviation. Based on the data structure, the statistical analysis was performed using Student’s t-test (unpaired, two-tailed), one-way ANOVA, or two-way ANOVA with Bonferroni as a post-test in GraphPad Prism (version 8.0) software. Statistical significance was denoted as follows by p values: * p < 0.05, ** p < 0.01, *** p < 0.001, and ns – non-significant.

Results

Diverse HA synthases produce HA with different molecular weights

It was previously shown [34] that, to some extent the MW_HA is an intrinsic property of the HA synthase. HA synthases, sourced from different Streptococcal sp., when cloned in L. lactis, produced diverse MW_HA (ranging from 0.4 to 2.1 MDa) under identical shake-flask culture conditions [34]. We validated the effect of this parameter in the bioreactor cultivation of recombinant L. lactis strains, each harboring a different hasA gene. The HA synthases (encoded by hasA) sourced from four different Streptococcal strains, viz. S. parauberis (spauhasA), S. uberis (suhasA), S. zooepidemicus (szhasA), and S. pyogenes (spyhasA), were co-expressed with the szhasB gene sourced from S. zooepidemicus to create four recombinant L. lactis strains (Table 1). The hasB gene was co-expressed to achieve reasonable HA titers [41]. The use of a bioreactor in HA production is well-known to support higher production. In our study, all the experiments were performed in 1.2 L bioreactors to enable stable control over pH and temperature, thereby eliminating the confounding effects of pH variation on the MW_HA produced. As a result, HA titers were similar in our studies for these strains, while the MW_HA differed significantly (Table 2). The strain L. lactis uAB produced high-MW_HA (1.35 MDa), whereas L. lactis pyAB produced much lower MW_HA (0.3 MDa). The HA produced by L. lactis zooAB and L. lactis pauAB had an intermediate MW_HA (0.6–0.7 MDa).

Table 2.

HA synthases from different Streptococcus species produce varying MW

Strain	MWHA (MDa)	Specific growth rate (h− 1)	Titer (g/l)
L. lactis uAB	1.35 ± 0.1 ****	0.15	0.69
L. lactis pauAB	0.72 ± 0.04 ns	0.14	0.74
L. lactis zooAB	0.62 ± 0.06	0.18	0.77
L. lactis pyAB	0.3 ± 0.05 **	0.17	0.71

MW_HA values are presented as mean ± SD. Statistical significance was determined by one-way ANOVA, with each strain compared against L. lactis zooAB

We conducted bioinformatics analyses focusing on sequence variations among HA synthases to investigate the basis of MW_HA variation. HA synthase diversity was previously characterized based on sequence and sub-structural elements (SSEs) [34, 53]. Here, we investigated topological variations among the synthases and sequence-level changes in predicted structures. Furthermore, we performed multiple sequence alignment (MSA) for various HAS isoforms and annotated the variations with SSEs. MSA revealed notable differences in transmembrane (TM) regions (Additional file 1, Fig. S1B), while most SSEs in catalytic domains were highly conserved. The analysis is detailed and discussed in Supplementary Section S1.

A recent study [26] utilizing recombinant C. glutamicum to express wild-type szHAS, identified TM helix 1 (TMH-1) as crucial for HA export due to its hydrophobic interactions with TMH-2. Similarly, cryo-EM studies [54] of viral HA synthases demonstrated that the glycosyltransferase domain interacts tightly with the channel-forming TM region, coupling HA synthesis with its translocation. These findings support our understanding that variations in TM domains among HA synthases likely modulate the MW_HA. Our findings establish baseline values for MW_HA achievable by combining the synthase properties with other genetic and process parameters.

Co-expression of additional HA pathway genes enhances MW_HA

Co-expressing a third heterologous gene from S. zooepidemicus operon significantly increased the HA titer and MW_HA [41, 55]. In particular, inclusion of the hasE gene alongside hasA and hasB has been reported to elevate MW_HA more effectively than the overexpression of other pathway genes [27]. We investigated the effects of pathway gene combinations, specifically hasAB and hasABE, on the MW_HA, as well as their impacts when combined with different HA synthases. The strain L. lactis pauABE was created, expressing hasBE from S. zooepidemicus, along with spauhasA (Table 1). As shown in Fig. 2A, L. lactis pauAB produced 0.72 MDa MW_HA, whereas the co-expression of hasE in the L. lactis pauABE strain increased the MW_HA to 1.65 MDa. To understand variations in MW_HA, we analyzed the intracellular concentrations of the two HA precursors, UDP-GlcUA and UDP-GlcNAc (Fig. 2B). The data suggest that the increase in MW_HA observed in L. lactis pauABE is probably due to the balance in the two HA precursors. This finding is consistent with earlier reports [27, 28] and emphasizes the importance of precursor balance [31] in influencing the MW_HA. The same gene combination hasAB and hasABE, were tested with a different HA synthase (spyhasA) from S. pyogenes, which typically produces low-MW_HA. L. lactis pyABE was constructed and compared with L. lactis pyAB, and the results corroborated our earlier findings. The strain L. lactis pyAB produced MW_HA of 0.3 MDa. In contrast, its hasABE counterpart L. lactis pyABE produced HA with a significantly higher MW_HA of 0.72 MDa (Fig. 2C). The corresponding intracellular precursor concentrations are shown in Fig. 2D and mirror the trends observed in Fig. 2B.

Fig. 2.

Effects of gene combinations (hasAB vs. hasABE) on MW_HA. A MW_HA from L. lactis pauAB and L. lactis pauABE expressing hasA from S. parauberis;B corresponding intracellular HA precursor concentrations. C MW_HA from L. lactis pyAB and L. lactis pyABE expressing hasA from S. pyogenes; D corresponding intracellular HA precursor concentrations. Statistical comparisons between the hasAB and hasABE groups were performed using appropriate tests, with significant differences (p-value) denoted by asterisks in the graph

Thus, independent of the specific HA synthase, incorporating the hasE altered the precursor ratio, thereby enhancing the MW_HA. While various studies have explored the impact of gene combinations on HA production [41, 42, 56], this study distinctly highlights the role of hasE in tuning intracellular precursor ratios to favor higher MW_HA synthesis. The findings further expand the MW_HA range by combining the effect of HA synthases with the additional co-expression of hasE.

MW_HA is enhanced in lactate dehydrogenase mutants of recombinant L. lactis strains

The role of glycolytic flux redistribution in enhancing HA production has been documented previously [27]. In particular, ldh-knockout strains of L. lactis (L. lactis NZ9020) have shown improved HA titers and molecular weights due to the increased precursor availability [33]. Disruption of ldh genes reduces lactate formation and redirects carbon flux toward heterolactic fermentation, increasing the NAD⁺/NADH ratio due to increased ethanol production [33]. This enhances the production of the HA-precursor, UDP-glucuronic acid, which requires NAD⁺ for its synthesis. While many studies have employed the ldh-knockout hosts to enhance HA production titers, their impact on MW_HA has been reported in only a few instances [33, 57]. In this study, we also used the L. lactis NZ9020 strain, which can potentially increase the precursor concentrations and MW_HA, in conjunction with various HA synthases and heterologous gene combinations. The strain L. lactis pauABEΔldh was created by introducing plasmid pNZ8148-paBE into L. lactis NZ9020, in which two ldh encoding genes were already deleted and replaced with selection markers. Batch cultures of L. lactis pauABEΔldh produced MW_HA of 2.55 MDa, a 52% increase compared with L. lactis pauABE, which is based on L. lactis NZ9000 (Fig. 3A). Precursor levels of L. lactis pauABEΔldh increased more than three-fold compared with those of L. lactis pauABE (Fig. 3B). Although the precursor balance was similar in the both the cases, the higher availability of both HA precursors in L. lactis pauABEΔldh culture increased not only the MW_HA but also the HA titer by 90%, reaching 1.4 g/L (Fig. 3A).

Fig. 3.

Effect of ldh deletion on HA production. A HA production in L. lactis pauABE (NZ9000) and L. lactis pauABEΔldh (NZ9020); B precursor concentrations in L. lactis pauABE and L. lactis pauABEΔldh. Statistical analysis of HA production by L. lactis 9020 (ldh-deficient strain) was performed using appropriate tests, with asterisks indicating significant differences (p-values) in the graph

Many reports in the literature have described a trade-off between titer and MW_HA [29, 58]. This is based on the assumption that high precursor availability reduces the chance of terminating the polymerization. Thus, high precursor availability and low HA synthase activity are an ideal combination for high MW_HA production. The use of ldh mutants improved both titer and MW_HA, highlighting its potential as a chassis to widen the MW_HA range and increase the HA titer.

Modulation of MW_HA by combining synthase diversity and precursor availability

Considering the potential of the ldh mutant to enhance precursor levels, we used L. lactis NZ9020 as the host strain for heterologous expression of the hasABE genes, with each strain harboring a different Streptococcal HA synthase gene. Each of these strains resulted in higher MW_HA (Fig. 4A), as exemplified by comparisons with L. lactis pauABE and L. lactis pyABE carrying similar constructs in the L. lactis NZ9000 host strain (Fig. 2A and C). The recombinant L. lactis NZ9020 cultures showed similarly enhanced levels of precursor concentrations (Fig. 4B), which are higher than those of the L. lactis NZ9000 cultures (Fig. 2B, D).

Fig. 4.

Effect of various HA synthases and additional co-expression of hasE on the HA-precursor level and MW_HA produced by Δldh L. lactis NZ9020 strains A MW_HA in L. lactis 9020 (Δldh) carrying hasABE and B HA precursors (UDP-GlcUA and UDP-GlcNAc). Statistical differences in MW_HA and precursor levels were analyzed using ANOVA

L. lactis pyABEΔldh produced HA of 0.88 MDa, which was ~ 20% higher than the MW_HA produced by L. lactis pyABE, whereas L. lactis pauABEΔldh produced 2.55 MDa HA, which is ~ 50% higher MW_HA than that produced by L. lactis pauABE. This difference underscores that the increase in MW_HA does not scale proportionally with precursor availability. It was speculated that the L. lactis uABEΔldh strain would produce the highest MW_HA based on the previous results (Table 2). However, L. lactis uABEΔldh was found to produce MW_HA of 2.45 MDa, which is nearly the same as that of L. lactis pauABEΔldh (Fig. 4A). Some reports suggest that the MW_HA does not increase beyond a certain saturation threshold, even with an abundant precursor supply [31]. While increased precursor availability supports the synthesis of higher MW_HA, the intrinsic properties of the synthase ultimately determine the maximum achievable polymer size. The heatmap analysis of intracellular precursor ratios and precursor availability (as indicated by UDP-GlcNAc) in Fig. 5 provides a better understanding of this phenomenon.

Fig. 5.

Precursor ratio and UDP-GlcNAc levels for various genetic combinations. A The impact of precursor ratio for two different HA synthase genes (spauhasA and spyhasA) with various genetic modifications, and B the impact of precursor availability, are visualized using heatmaps, with the MW_HA specified for each condition

MW_HA is influenced by multiple factors rather than a single determinant. The expression of hasABE led to a more favorable precursor ratio and a corresponding increase in MW_HA for both the HA synthases (Fig. 5). Furthermore, the ∆ldh host elevates the precursor levels, resulting in increased MW_HA. However, the magnitude of this increase differed between the L. lactis cultures expressing the two synthases, spauhasA and spyhasA, underscoring the critical role of synthase-specific properties. These findings emphasize that, although enhanced precursor availability can improve MW_HA, the HA synthase remains a key factor in determining the upper limit of polymer size, even under optimized genetic and metabolic conditions.

Effect of initial glucose concentration on MW_HA

In addition to genetic factors, MW_HA is influenced by several process parameters such as temperature, pH, aeration, substrate concentration, and specific growth rate [21, 30, 59]. These parameters can affect MW_HA by affecting precursor production and utilization during HA synthesis. We studied the effect of one such process parameter, namely, initial glucose concentration, which affects glucose uptake and its distribution in the glycolytic and HA-biosynthesis pathways [42]. Recent studies have focused on the impact of different carbon sources and their concentrations on HA production [60]. To evaluate the impact of initial substrate concentration on the MW_HA, the strain L. lactis pauAB was cultivated in the bioreactor at three different glucose concentrations (10, 30, and 50 g/L). At an initial glucose concentration of 30 g/L, the MW_HA was twice that of 10 g/L glucose (Fig. 6A). However, only a slight increase in MW_HA was observed between initial glucose concentrations of 30 g/L and 50 g/L. Similar trends were observed for HA titers for experiments with varying initial glucose concentrations (Table 3). The differences in MW_HA can be attributed to the availability of the two HA precursors (Fig. 6B).

Fig. 6.

Substrate availability influences the MW_HA and titer. A Effect of initial glucose concentration on the MW_HA and titer produced by L. lactis pauAB; B intracellular HA-precursor concentrations (UDP-GlcUA and UDP-GlcNAc) in L. lactis pauAB cultured at different initial glucose concentrations. Statistical significance was analyzed using Student’s t-test and one-way ANOVA

Table 3.

Effects of initial glucose concentrations on the production of HA by L. lactis pauAB

Glucose (g/l)	Biomass (g/l)	HA (g/l)	Specific growth rate (h− 1)
10	1.46	0.35	0.39
30	2.68	0.72	0.17
50	3.68	0.90	0.17

Mean values of biomass and hyaluronic acid (HA) concentrations under each experimental condition are presented above

At 10 g/L glucose, the limited precursor availability resulted in a low MW_HA. Increasing the initial glucose concentration to 30 g/L enhanced the precursor availability, which reflected in the improved MW_HA. However, increasing the glucose concentration to 50 g/L did not significantly increase the MW_HA. Since the strain has heterologous hasB, only the UDP glucuronic acid content was increased. Moreover, the UDP-N-acetylglucosamine (UDP-GlcNAc) remained at the same level compared to the 30 g/L initial glucose. It is known that chain truncation in HA synthesis occurs when there is a limitation in precursor availability [30]. Some studies have also emphasized UDP-GlcNAc levels as the limiting factor [61]. Similar trends were observed for the effects of initial glucose concentration (10 and 30 g/L) in several other recombinant strains. The detailed data can be seen in Supplementary Table S2.

Representative fermentation profiles (biomass, substrate, product and other metabolites) are provided in Additional File 2 for selected strains. The metabolic distinction between L. lactis NZ9000 (wild-type) and L. lactis NZ9020 (∆ ldh) is well established. The NZ9000 strain exhibits typical homolactic fermentation, while the NZ9020 host exhibits heterolactic fermentation, characterized by the production of metabolites such as formate and acetate. Due to the large number of experiments involving multiple strains, we did not provide this data in the main manuscript and have also not shown the profiles for all the experiments in Additional File 2. Process factors do affect MW, but these factors play a lesser role in MW variation as compared to the effect of synthase and the genetic variables mentioned in the manuscript. Ultimately, the main way that process factors affect MW is through the precursor availability and their ratio, which is captured in our study.

A combinatorial strategy for producing an array of MW_HA

In summary, a combinatorial strategy is developed to produce an array of MW_HA by multiplexing the following four factors: (i) different hyaluronan synthases (HASs) with intrinsic ability to produce different MW_HA; (ii) the role of the hasE gene in enhancing MW_HA by modulating the balance of HA precursors; (iii) the enhancement of HA-precursor concentrations by rerouting carbon flux; and (iv) the modulation of precursor availability by initial substrate concentrations. To produce an array of MW_HA, we employed four different HAS variants with two gene combinations (hasAB and hasABE) in both the wild-type (L. lactis NZ9000) and ldh mutant (L. lactis NZ9020) strains. We grew these strains at two glucose concentrations (10 and 30 g/L). This approach yielded a broad range of MW_HA, as illustrated in Fig. 7. For better clarity, some strains/conditions that produced very similar MW_HA are omitted. A comprehensive dataset of MW_HA and conditions is given in Additional file 1 (Section S2, Fig. S2, and Table S2). Also, to understand the differences in MWs produced by these various cultivation conditions, the Chromatogram overlay is available in the supplementary information to visually understand the diversity of MWs produced (Additional file 1; Fig. S3A and S3B).

Fig. 7.

MW_HA a la carte by multiplexing HA synthase with genetic and cultivation factors

The combinatorial strategy enabled the production of HA across a range of MWs, from 0.2 to 2.6 MDa. As an outcome, a map correlating MW_HA and titer was generated for all the recombinant L. lactis strains, specifically, strains with each of the four hasA variants, in the hasAB and hasABE gene combinations, expressed in L. lactis NZ9000 and L. lactis NZ9020, cultured at 10 and 30 g/L initial glucose concentration (Supplementary section S2). This provides a framework for selecting an optimal strain and production strategy based on the MW_HA requirements.

Discussion

The combinatorial strategy employed in this study, manipulating both synthase diversity and precursor supply, aligns conceptually with some of the approaches used for other microbially-produced glycosaminoglycans and biopolymers. For instance, in the case of γ-polyglutamic acid (PGA) produced by Bacillus subtilis [62], the exchange of native PGA synthetase with synthetases from other Bacillus species was found to be the primary determinant for controlling molecular weight, achieving a range from 40 kDa to over 8,500 kDa. This is similar to our study, where expressing the hasA gene from various Streptococcus species in recombinant L. lactis strains resulted in varying MW_HA. Our study strongly supports the finding that the intrinsic properties of the HAS isoform exerts a dominant influence on the resulting MW_HA, independent of the host’s metabolic state. Elucidating the precise structural elements of the synthase that govern polymer size remains a significant challenge, given the complex, multi-domain, membrane-bound nature of HAS. While the catalytic domains were highly conserved, our in-silico analysis identified significant sequence variations within the TM domains of the four HAS isoforms (Additional file 1; Fig. S1B and Table S1B). We hypothesize that these TM region variations are a likely contributor to the observed differences in MW_HA, a concept supported by recent studies [26, 63] that identified TM helices as crucial for HA export and translocation. We are further investigating the specific functions of each domain to elucidate their effects on HA molecular weight. A deeper understanding of these structure-function relationships, building on work that has identified key motifs for synthase processivity and size control, will ultimately enable the rational engineering of synthases for even more precise targeting of MW_HA.

Beyond the choice of synthase, our study demonstrates that the final MW_HA can be tuned by modulating the intracellular precursor availability (UDP-GlcNAc and UDP-GlcUA) and its balance. We successfully manipulated this precursor pool through three distinct but complementary approaches: (1) Modulation of the initial glucose concentration in the medium directly influenced the intracellular precursor pools (Fig. 6B). (2) Balancing the precursor ratio via co-expression of hasE, consistently yielded higher MW_HA for the hasABE constructs which had a better precursor balance compared to the hasAB constructs (Fig. 2A, C). (3) Enhancement of the flux in the HA-precursor pathways by effectively re-routing the central carbon flux away from lactate and towards the HA biosynthetic pathway, using a lactate dehydrogenase-deficient host strain (L. lactis 9020), led to an increase in the intracellular concentrations of both HA-precursors (Fig. 3B). The work of Chen et al. (2009) provided a direct mechanistic basis for our observations by demonstrating that in Streptococcus zooepidemicus, the MW_HA is directly controlled by the intracellular concentration of UDP-GlcNAc, which they identified as the rate-limiting precursor for chain elongation. While numerous studies have explored various heterologous gene combinations for HA production [41, 42, 64–66], our work clearly elucidates the quantitative impact of these genetic and process modifications on MW_HA by correlating them with measured changes in the precursor pool.

Similar to our approach, studies with other glycosaminoglycans (GAGs) [67]have shown two E. coli host systems (E. coli K5 and E. coli Nissle 1917) being able to produce different MWs of chondroitin due to their genetic and metabolic makeup., The overexpression of the transcriptional factor rfaH alters the precursor synthesis, thereby increasing the chondroitin molecular weight significantly. The studies further highlight the critical role of the intracellular UDP-sugar precursor pool [68], demonstrating that precursor availability is a key bottleneck influencing the final synthesis of polysaccharides. Furthermore, Hu et al. demonstrated that while enhanced precursor supply improved the titer of various GAGs [69], it was the balance of precursors that was crucial for increasing the molecular weight of hyaluronic acid, heparosan, and chondroitin. This aligns directly with our findings, where interventions leading to an increase in the MW_HA have either boosted overall precursor levels (Figs. 3B and 6B) or specifically balanced their ratios (Fig. 2B, D). Taken together, our work solidifies the conclusion that the metabolic engineering of precursor supply is also a critical component in determining the MWs and can be a universally applicable tool for tailoring the molecular weight of GAGs.

The distinct contribution of this study is the demonstration of multiplexing these metabolic and process tools and quantifying their synchronous impact, thereby creating a predictive matrix of conditions to achieve a desired MW_HA and titer. Based on the desired MW_HA requirements, one or more of these conditions can then be selected for large-scale production. In our study, an illustration of the various combinations for spauhasA and spyhasA is shown in Fig. 8.

Fig. 8.

MW_HA vs. HA-titer map for HA production under selected conditions by recombinant L. lactis strains expressing two different HA synthases (spauhasA and spyhasA)

From these data, it becomes possible to select a suitable combination according to the required MW and titer. A comprehensive dataset and representation of all the tested combinations are shown in the supplementary information (Additional file 1 ; Fig. S2 and Table S2).

We acknowledge that the HA titers obtained in this study, ranging from 0.4 to 1.5 g/L, are modest compared to the high yields achieved by other dedicated microbial platforms. However, the primary objective of this work was not to maximize titer, but to establish a versatile and tunable “à la carte” system. The central contribution of our technology is the demonstration that a single bacterial platform, L. lactis, can be systematically manipulated to produce a broad and predictable spectrum of MW_HA (0.2–2.6 MDa).

The resulting MW–titer map (Additional file 1; Fig. S2) provides a framework for selecting an optimal strain and production strategy. Depending on the stringency of the MW requirement for a specific application, a trade-off in titer may be acceptable. Furthermore, these titers do not limit the platform’s capabilities. Significant opportunities exist to improve yields by integrating established process strategies, which were beyond the scope of the current combinatorial study. Future work will focus on enhancing productivity by implementing Glucostat fed-batch operations [42] which can control MW while improving titer, enhancing Acetyl-CoA availability [28] to improve the limiting precursor (UDP-GlcNAc), or harnessing heme-induced respiration [27]. This multiplexing approach, combined with future process optimization, establishes a robust foundation for the one-pot synthesis of customized HA.

Conclusions

This study presents the production of HA with tailored molecular weights based on the multiplexing of various genetic factors in engineered Lactococcus lactis and cultivation conditions. By incorporating a diverse range of HA synthases, modifying key pathway genes, and fine-tuning cultivation conditions, we achieved the multiplexing concept. Unlike traditional methods that rely on harsh chemical or enzymatic treatments to achieve the desired molecular weight, our technique offers a streamlined, one-pot synthesis that produces HA with superior quality and reduced downstream processing requirements. The use of L. lactis, a well-characterized, safe organism, further enhances the efficiency and safety of the process. This study reports the broadest range of MW values (0.2–2.6 MDa) for producing MW-specific HA, making it a suitable HA production platform for diverse biomedical and cosmetic applications.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ANOVA

Analysis of variance

CTAB

Cetyltrimethylammonium bromide

GRAS

Generally Recognized as Safe

HA

Hyaluronic acid

hasA

Hyaluronan synthase gene

hasAB

Combination of heterologous genes-Hyaluronan synthase (hasA) and UDP-glucose dehydrogenase (hasB)

hasABE

Combination of heterologous genes-Hyaluronan synthase (hasA), UDP-glucose dehydrogenase (hasB), and phosphoglucoisomerase (hasE)

hasB

UDP-glucose dehydrogenase gene from Streptococcus zooepidemicus

hasE

Phosphoglucoisomerase gene from Streptococcus zooepidemicus

HEPES

4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid

HPLC

High-performance liquid chromatography

kDa

Kilo Dalton

LB

Lysogeny broth

ldh

Lactate dehydrogenase

MDa

Million Dalton

MW_HA

Hyaluronic acid Molecular weight

NICE

Nisin-controlled gene expression system

OD₆₀₀

Optical density at 600 nm

PCR

Polymerase chain reaction

SEC

Size exclusion chromatography

spahasA

hasA gene from Streptococcus parauberis

sphasA

hasA gene from Streptococcus pyogenes

suhasA

hasA gene from Streptococcus uberis

szhasA

hasA gene from Streptococcus zooepidemicus

UDP-GlcNAc

Uridine diphosphate N-acetylglucosamine

UDP-GlcUA

Uridine diphosphate glucuronic acid

WT

Wild type

Δldh

Lactate dehydrogenase knockout

Author contributions

- Sharath Soundiraraj : Conceptualization, Methodology, Experimental Investigation and Data Acquisition, Formal analysis, Data curation, Visualization, Writing – original draft, review, editing.- Nakul Ravishankar : Bioinformatic analysis of HA synthase genes- Pandeeswarri Jeeva : Molecular cloning- Lars M. Blank : Collaboration – provision of HA synthase genes from diverse Streptococcal strains for this study; review, and editing- Guhan Jayaraman : Conceptualization, Supervision, Project administration, Funding acquisition, Writing – review & editing.

Funding

Open Access funding enabled and organized by Projekt DEAL. This work was supported by the Biotechnology Industry Research Assistance Council (BIRAC), Government of India (Grant No. BT/SB/0012/01/18).

Data availability

All the data is provided within the manuscript or in a supplementary information file.

Declarations

Ethics approval and consent to participate

This article contains no studies performed by the authors with human participants or animals.

Competing interests

The authors declare no competing interests.