Repetitive Synthesis of High-Molecular-Weight Hyaluronic Acid with Immobilized Enzyme Cascades.

Industrial hyaluronic acid (HA) production comprises either fermentation with Streptococcus strains or extraction from rooster combs. The hard-to-control product quality is an obstacle to these processes. Enzymatic syntheses of HA were developed to produce high-molecular-weight HA with low dispersity. To facilitate enzyme recovery and biocatalyst re-use, here the immobilization of cascade enzymes onto magnetic beads was used for the synthesis of uridine-5'-diphosphate-α-d-N-acetyl-glucosamine (UDP-GlcNAc), UDP-glucuronic acid (UDP-GlcA), and HA. The combination of six enzymes in the UDP-sugar cascades with integrated adenosine-5'-triphosphate-regeneration reached yields between 60 and 100 % for 5 repetitive batches, proving the productivity. Immobilized HA synthase from Pasteurella multocida produced HA in repetitive batches for three days. Combining all seven immobilized enzymes in a one-pot synthesis, HA production was demonstrated for three days with a HA concentration of up to 0.37 g L-1 , an average MW of 2.7-3.6 MDa, and a dispersity of 1.02-1.03.

Introduction

Hyaluronic acid (HA) belongs to the important glycan group glycosaminoglycans (GAGs) and is composed of glucuronic acid (GlcA) and N‐acetylglucosamine (GlcNAc).[ , ] The linear molecule ([‐3)GlcNAc(β1‐4)GlcA(β1‐] ) can reach a molecular size of up to 107 Da. Due to its anionic character, HA binds water molecules resulting in a viscoelastic gel. HA occurs naturally in the human body and fulfills many biological tasks. As a passive structure, HA for example moisturizes the skin or is a space filler in joints absorbing the pressure. However, HA is also a signaling molecule and is involved in inflammation, tissue repair, and cell proliferation. The mode of action is highly dependent on the HA size. For example, high‐molecular‐weight (HMW) HA is anti‐inflammatory, while low‐molecular‐weight (LMW) HA has contrary effects. It is not surprising that HA as a biocompatible molecule is exploited for many cosmetic and medical applications like the treatment of arthritis or ophthalmology.[ , , ]

While the market of HA is rising, the request for defined HA especially for medical applications is not satisfied by common industrial processes.[ , ] HA is manufactured either by harsh extraction from rooster combs or fermentation with Streptococcus resulting in a highly disperse product and the risk of avian protein or endotoxin contamination.[ , , , ] Therefore, enzymatically oriented approaches to produce HA or other GAGs stay in focus of today‘s research.

Enzymatic catalysis has been intensively studied as a cleaner and green alternative and is increasingly important for the industry.[ , , , ] However, production costs, the need to improve stability, activity, and purity of enzymes are still obstacles for the industrial breakthrough. To address these obstacles immobilization of enzymes has become the focus of many studies (1500 papers each year).[ , , ] Activity and stability could be enhanced with immobilization strategies,[ , ] and there are even examples where selectivity and specificity could be changed.[ , , ] Different kinds of immobilization methods have been established like carrier‐bound covalent/noncovalent binding or encapsulation. All these techniques come with their advantages and disadvantages. A non‐covalent method is polyhistidine tag (His‐Tag) based immobilization, which is adapted from the purification of recombinant proteins with immobilized metal affinity chromatography (IMAC).[ , , ] This one‐step method enables selective and non‐destructive binding with high enzyme loading.[ , , , ] His‐Tag‐based immobilization is often combined with magnetic nanoparticles or beads.[ , , , ] Magnetic carriers offer a time‐saving inexpensive way to separate enzymes from the reaction just by applying a magnetic field. Therefore, easy reusability of enzymes is enabled.[ , ]

In our previous studies, we established one‐pot HA syntheses with soluble enzymes including enzyme cascades for uridine‐5'‐diphosphate (UDP)‐sugar synthesis and regeneration.[ , ] The enzyme cascades can be further extended by an adenosine‐5'‐triphosphate (ATP) regeneration system.

We here present the straightforward development of a repetitive one‐pot synthesis with immobilized enzymes on magnetic beads (MB) to produce HMW HA employing seven immobilized enzymes (Scheme 1).

Scheme 1

Immobilized enzyme cascades for hyaluronic acid synthesis. A schematic presentation about the methodical steps including individual and target enzyme immobilization from crude extract, assembling of the enzymes, and enzymatic reaction cycle. The one‐pot synthesis includes three enzyme modules (EMs): EM uridine‐5’‐diphosphate‐α‐d‐glucuronic acid (UDP‐GlcA), EM uridine‐5’‐diphosphate‐α‐d‐N‐acetyl‐glucosamine (UDP‐GlcNAc), and EM hyaluronan module. The EM UDP‐GlcA consists of the enzymes AtGlcAK (glucuronic acid kinase), AtUSP (UDP sugar pyrophosphorylase) (both from Arabidopsis thaliana), RpPPK2‐3 (polyphosphate kinase family 2 class 3 from Ruegeria pomeroyi), and PmPpA (pyrophosphatase from Pasteurella multocida). The EM UDP‐GlcNAc consists of the enzymes BlNahK (GlcNAc‐1‐phosphate kinase from Bifidobacterium longum), SzGlmU (UDP‐GlcNAc pyrophosphorylases from Streptococcus zooepidemicus), RpPPK2‐3, and PmPpA. In the hyaluronan module PmHAS1–703 (soluble class 2 hyaluronan synthase from P. multocida) polymerizes the HA chain using UDP‐GlcA and UDP‐GlcNAc as donor substrates. RpPPK2‐3 regenerates ATP and uridine‐5'‐triphosphate (UTP) from adenosine‐5'‐diphosphate (ADP) and UDP, respectively.

Results and Discussion

Immobilization and activity assays of immobilized enzymes

We here used His Mag SepharoseTM Ni from GE Healthcare (Sweden) as MBs, which were normally used for IMAC protein purification. For enzyme loading onto the MBs, we followed the standard protocol from the manufacturer (see Experimental Section; Scheme 1). Cells were disrupted with ultrasound and separated from the crude extract with centrifugation. The crude extract was mixed end‐over‐end with MBs. After binding, unbound enzymes were cast away with washing steps. Immobilized enzymes were mixed with substrates and cofactors and mixed end‐over‐end. After the reaction was completed, MBs and product solution was separated with a magnet. This His‐Tag‐based immobilization is a noncovalent method.[ , , ] Therefore, we had to make sure that the enzymes are not detaching during the reaction. Each step was analyzed by sodium dodecyl sulfate‐polyacrylamide gel electrophoresis (SDS‐PAGE) and Western‐blot (Figure S1) to ensure that only immobilized enzymes are used for substrate conversion. No detached enzymes were detected in the product solution. To confirm this result, we eluted the enzymes from the used MBs and here we could see a strong protein band in the SDS‐PAGE and Western‐blot. The amount of bound active enzyme per g MB was determined after testing the substrate conversion as follows (Table 1).

Table 1

Binding capacity for each enzyme on His Mag SepharoseTM Ni from GE Healthcare (Sweden).[a]

Enzyme	Enzyme elution [mg]	Enzyme/MBs [mg g−1]	Substrate	K M [b] [mm]	K iS [b] [mm]	V max/protein[b] [U mg−1]	V max/MBs[b] [U g−1]
AtGlcAK	0.0205	0.82	ATP	14.3app [c]	0.007app [c]	26.5app [c]	21.7
AtUSP	0.0195	0.78	UTP	9.6	–	14.8	11.5
BlNahK	0.0666	2.67	ATP	7.3	–	5.8	15.5
SzGlmU	0.0599	2.40	UTP	0.3	–	0.7	1.7
RpPPK2‐3	0.4384	17.54	ADP	9.5	–	64.8	1136.3
			UDP	2.7	–	63.4	1111.8
PmPpA	0.2791	11.16	PPi	9.8	–	2435	27184
PmHAS1–703	0.0928	3.71	UDP‐GlcA	8.0app [c]	1.1app [c]	0.99app [c]	3.7
			UDP‐GlcNAc	5.9	–	0.09	0.3

[a] The maximum capacity of the manufacture is 20 mg protein per g MBs. Kinetic parameters for enzymes immobilized on MBs. Nucleotides and nucleotide sugars were measured with multiplexed capillary electrophoresis (MP‐CE). Pi was measured with a Phosphate kit. Kinetic parameters were determined with Michaelis‐Menten kinetics. [b] Kinetic parameters calculated with Equations (1) and (2); K M: Michaelis constant; V max: maximal rate of the system; K : dissociation constant of the enzyme‐substrate‐substrate complex. [c] Apparent kinetic constant.

–

–

–

–

–

–

–

The amount of MB‐loaded enzyme is dependent on the expression of the enzyme in the crude extract. We know from our previous studies that RpPPK2‐3 and PmPpA are the best‐expressed enzymes in Escherichia coli among the used enzymes.[ , , ] Therefore, it is not surprising that RpPPK2‐3 and PmPpA showed the highest loading concentrations, and the plant enzymes AtGlcAK and AtUSP the lowest concentration. The highest loading capacity according to the manufacture's information is 20 mg protein per g MBs, which we nearly reached with RpPPK2‐3. Next to the expression level also the binding to the MBs may be different for each enzyme since the His‐tag presentation on the N‐or C‐terminal end could be covered for each enzyme, which could also influence the loading efficiency. Nevertheless, the samples, which were eluted from the magnetic beads after the reaction, showed a high target enzyme concentration and the product solution showed no enzyme. Out of these reasons, the measured activities (Figure S2) are caused by immobilized enzymes.

Each immobilized enzyme was tested for substrate conversion (Figure S2). For AtGlcAK and BlNahK, the formation of ADP indicated conversion of GlcA and GlcNAc to GlcA‐1‐P and GlcNAc‐1‐P, respectively. After 4 h, a yield of 93 % (Figure S2A, AtGlcAK) and 74 % (Figure S2C, BlNahK) was obtained. Because GlcA‐1‐P was commercially not available, AtUSP was tested with Glc‐1‐P as substrate (Figure S2B). The reactions of AtUSP and SzGlmU were not supplemented with pyrophosphatase. Therefore, yields of the corresponding UDP‐sugars were low due to the equilibrium reaction. After 5 h, a yield of 26 % for UDP‐Glc (Figure S2B) and 40 % for UDP‐GlcNAc (Figure S2D) was obtained confirming our previous study with soluble enzymes. With immobilized PmPpA, we observed a relatively fast degradation of pyrophosphate (PPi) to phosphate (Pi) (Figure S2E). After 2.5 min, the reaction already reached its equilibrium. The conversion of UDP to UTP by RpPPK2‐3 was fast as well. After 15 min, the equilibrium between UTP and UDP was reached (yield 70 %) (Figure S2F). Immobilized PmHAS1–703 showed good activity (Figure S2G). After 24 h, approximately 10 mm UDP was obtained. The reactions of these enzymes still needed optimization, but we showed for the first time, that active enzymes are immobilized via a His‐Tag onto Ni2+/Nitrilotriacetic acid (NTA) MBs.

Kinetics

It has been demonstrated that the immobilization of enzymes alters the kinetic parameters. We further determined the kinetic parameters of the immobilized enzymes (Table 1, Figure S3). Metal ion concentrations were set to 25 mm MgCl2, 10 mm KCl, and 1.5 mm MnCl2, which we used for the synthesis of HMW HA with the soluble enzymes.[ , ] Immobilized AtGlcAK has a 1.7‐fold higher apparent Michaelis constant (K M) value and 1.4‐fold lower maximal velocity (V max) compared to soluble AtGlcAK. It shows also substrate inhibition for ATP. Immobilization of AtUSP had a greater effect. V max is only one‐tenth of the soluble version and the K M value is also 20 times higher. In the soluble EMs, AtGlcAK was the only bottleneck. Here, immobilized AtGlcAK and immobilized AtUSP determine the efficiency of the EM. Also, the K M value of immobilized BlNahK is 73 times higher than described in the literature, but showed a 14.5 timeshigher specific activity. Immobilized SzGlmU has a similar K M value as soluble SzGlmU but showed a 49 times lower V max. The maximal specific activity of immobilized RpPPK2‐3 is for both substrates, ADP and UDP, increased (ADP: 20 times; UDP: 4 times). Surprisingly, the K M for UDP stays the same, while the K M value for ADP is ten times higher. Immobilization affected PmPpA so that no substrate inhibition could be seen compared to soluble PmPpA. This means that the already high specific activity of PmPpA is even further increased. In contrast, immobilization did not influence the substrate inhibition of PmHAS1–703 by UDP‐GlcA, but the V max was doubled in comparison to the soluble enzyme. For UDP‐GlcNAc the same V max value could be observed, but the K M value was markedly reduced (4 times lower). We concluded that the higher affinity for UDP‐GlcNAc, as well as the overall higher V max with UDP‐GlcA, will have a tremendous effect on the polymerization activity of PmHAS1–703 in the one‐pot synthesis as shown later.

In summary, all immobilized enzymes are active, and the impact of the altered kinetic parameters on the one‐pot syntheses must be investigated.

Repetitive batch synthesis of UDP‐GlcA with the immobilized enzyme cascade

We first tested UDP‐GlcA one‐pot synthesis with ATP regeneration to demonstrate the reusability of immobilized enzymes for two repetitive cycles (Figure 1, Figure S4). The reaction parameters were adapted to the conditions for HA synthesis.[ , ] Metal ions were set to 25 mm MgCl2, 10 mm KCl, and 1.5 mm MnCl2, and the reaction time for one cycle was set to 24 h (Figure S2G). ATP regeneration was started with 1, 2.5, and 5 mm ATP, respectively. After 24 h, the immobilized enzymes were recovered by magnetic separation and then re‐used in a second reaction cycle. In the first cycle, UDP‐GlcA synthesis reached a maximum yield between 91–96 % from 10 mm GlcA already after 2 h (1 mm ATP: 94 %; 2.5 mm ATP: 96 %; 5 mm ATP: 91 %) (Table 2).

Figure 1

Repetitive synthesis of UDP‐GlcA with enzymes immobilized on MBs. UDP‐GlcA was analyzed with MP‐CE. See also Figure S4. Each reaction contained 10 mm GlcA, 10 mm UTP, 20 mm polyphosphate (PolyP), 25 mm MgCl2, 10 mm KCl, 1.5 mm MnCl2, 100 mm TRIS‐HCL [TRIS: tris(hydroxymethyl)aminomethane; pH 8.5], 4.3 U AtGlcAK, 1.0 U AtUSP, 109 U PmPpA, and 8.3 U RpPPK2‐3. Different ATP concentrations (1, 2.5, and 5 mm) were used, respectively. After 24 h, the reaction mixture was directly changed.

Table 2

Yields and efficiency of the immobilized EM for UDP‐GlcA synthesis combined with immobilized RpPPK2‐3 as an ATP regeneration system.[a]

Cycle	ATP [mm]	UDP‐GlcA [mm]		Yield [%]	Space‐time yield [mm h−1]	ATP regeneration cycle[b]		Efficiency [%]
		maximal	actual			maximal	actual
1	1	10	9.35[c]	94	4.67	9	8.35	93
	2.5	10	9.55[c]	96	4.77	3	2.82	94
	5	10	9.05[c]	91	4.52	1	0.81	81
2	1	10	8.57[d]	86	2.86	9	7.57	84
	2.5	10	8.44[d]	84	2.81	3	2.38	79
	5	10	8.48[e]	85	1.41	1	0.70	70

[a] Each reaction contained 10 mm GlcA, 10 mm UTP, 20 mm PolyP, 25 mm MgCl2, 10 mm KCl, 1.5 mm MnCl2, 100 mm TRIS‐HCL (pH 8.5), 4.3 U AtGlcAK, 1.0 U AtUSP, 109 U PmPpA, and 8.3 U RpPPK2‐3. Different ATP concentrations (1, 2.5, and 5 mm) were used, respectively. After 24 h, the reaction mixture was directly changed. The enzyme loading can be seen in Table 1. [b] The regeneration cycle was calculated with Equation (3). [c] UDP‐GlcA concentration and yield after 2 h on that cycle. [d] UDP‐GlcA concentration after 3 h on that cycle. [e] UDP‐GlcA concentration after 6 h on that cycle.

The observed decrease of UDP‐GlcA is caused by chemical decomposition in the presence of MnCl2 at pH 8.5 to 1,2‐cyclic sugar‐phosphate derivate and uridine‐5'‐monophosphate (UMP) (Figure S4).[ , ] Interestingly, the decomposition of UDP‐GlcA was in the first cycle higher than in the second cycle. However, in the second cycle, we could still reach with 1 or 2.5 mm ATP a high product yield of around 85 % after 3 h. With 5 mm ATP, a yield of 85 % was reached after 6 h due to high ATP concentrations, which hamper AtGlcAK (Table 1 and Figure S3). This confirms the need for an ATP regeneration system for the UDP‐GlcA synthesis. With an ATP regeneration efficiency of around 90 % in the first cycle and 80 % on the second, 1 and 2.5 mm ATP are the best starting conditions to also obtain high space‐time yields (Table 2). Another interesting effect is that we observed the formation of adenosine‐5'‐tetraphosphate (AP4), which was increased with higher ATP concentration (Figure S4). Later in the reaction, AP4 was again degraded by RpPPK2‐3. It is known that PPK2 s, especially in excess, can elongate the phosphate chain of ATP and even AP4.

In summary, we could produce UDP‐GlcA for 48 h in sufficient concentrations in presence of PolyP for ATP regeneration. Possible scavenging of Mg2+ by PolyP has no effect on the enzyme cascade as previously seen for the soluble enzymes. We also conclude that the fast conversion shall be exploited for more and shorter cycles over several days.

Repetitive batch synthesis of UDP‐GlcNAc with the immobilized enzyme cascade

We also investigated the immobilized enzyme cascade for UDP‐GlcNAc synthesis for its reusability (Figure 2, Figure S5). The reactions were performed as described for UDP‐GlcA synthesis. In the first cycle, all reactions showed a similar reaction performance independent of the initial ATP concentration (Figure 2). The ATP regeneration with 1 mm ATP starting concentration showed the best regeneration efficiency (97 %) in both cycles (Table 3). Moreover, high yields (97 and 98 % with 1 mm ATP) for the UDP‐GlcNAc synthesis were reached after 2 h or 3 h in the first and second cycle, respectively. In general, each cycle reached high yields of 96–98 % and a high ATP recycling efficiency (>91 %) (Table 3) Surprisingly, the immobilized EM UDP‐GlcNAc module was still efficient with 20 mm PolyP concentrations which is a significant improvement in comparison to the performance of EM UDP‐GlcNAc in the soluble one‐pot synthesis, where 20 mm PolyP led to reduced UDP‐GlcNAc synthesis. In summary, the syntheses of UDP‐GlcA and UDP‐GlcNAc with immobilized enzyme cascades including ATP regeneration was highly efficient. With high process stability of all immobilized enzymes for at least two days and high product yields after 2–3 h, the immobilized enzyme cascades shall be utilized in multiple cycles within two days.

Figure 2

Repetitive synthesis of UDP‐GlcNAc with enzymes immobilized on MBs. UDP‐GlcNAc was analyzed with MP‐CE. See also Figure S5. Each reaction contained 10 mm GlcNAc, 10 mm UTP, 20 mm PolyP, 25 mm MgCl2, 10 mm KCl, 1.5 mm MnCl2, 100 mm TRIS‐HCL (pH 8.5), 0.3 U BlNahK, 0.4 U SzGlmU, 109 U PmPpA, and 8.3 U RpPPK2‐3. Different ATP concentrations were used at 1, 2.5, and 5 mm, respectively. After 24 h, the reaction mixture was changed.

Table 3

Yields and efficiency of the immobilized EM for UDP‐GlcNAc synthesis combined with immobilized RpPPK2‐3 as an ATP regeneration system.[a]

Cycle	ATP [mm]	UDP‐GlcNAc [mm]		Yield [%]	Space‐time yield [mm h−1]	ATP regeneration cycle[b]		Efficiency [%]
		maximal	actual			maximal	actual
1	1	10	9.71[c]	97	3.24	9	8.71	97
	2.5	10	9.72[c]	97	3.24	3	2.89	96
	5	10	9.72[c]	97	3.24	1	0.94	94
2	1	10	9.76[d]	98	2.44	9	8.76	97
	2.5	10	9.60[c]	96	3.18	3	2.84	95
	5	10	9.56[d]	96	2.39	1	0.91	91

[a] Each reaction contained 10 mm GlcNAc, 10 mm UTP, 20 mm PolyP, 25 mm MgCl2, 10 mm KCl, 1.5 mm MnCl2, 100 mm TRIS‐HCL (pH 8.5), 0.3 U BlNahK, 0.4 U SzGlmU, 109 U PmPpA, and 8.3 U RpPPK2‐3. Different ATP concentrations were used at 1, 2.5, and 5 mm, respectively. After 24 h, the reaction mixture was changed. The enzyme loading can be seen in Table 1. [b] The regeneration cycle was calculated with Equation (3). [c] UDP‐GlcNAc concentration after 3 h on that cycle. [d] UDP‐GlcNAc concentration after 4 h on that cycle.

Repetitive batch synthesis of UDP‐GlcA and UDP‐GlcNAc in one‐pot with immobilized EMs

The next step was to combine both immobilized EMs for the synthesis of UDP‐sugars in one pot. Both enzyme cascades require ATP. Therefore, it was interesting how the regeneration system can provide enough ATP for both systems. We had to make sure that the ATP concentration is sufficient, and the total ATP/ADP concentration is not too high to reach high kinase activities and efficient ATP regeneration. As a summary of the results for the single immobilized cascades (Tables 2 and 3), we decided to use 2.5 mm ATP as a compromise. Since the previous results suggest that more cycles are possible, we prolonged the experiment for 5 days (5 cycles; Figure 3, Figure S6).

Figure 3

Repetitive one‐pot synthesis of UDP‐GlcA and UDP‐GlcNAc with EMs immobilized on MBs. All nucleotides and nucleotide sugars were analyzed with MP‐CE. See also Figure S6. (A) UDP‐GlcA development. (B) UDP‐GlcNAc development. Each reaction contained 10 mm GlcA, 10 mm GlcNAc, 2.5 mm ATP, 20 mm UTP, 20 mm PolyP, 25 mm MgCl2, 10 mm KCl, 1.5 mm MnCl2, 100 mm TRIS‐HCl (pH 8.5), 4.3 U AtGlcAK, 0.6 U AtUSP, 0.6 U BlNahK, 1.2 U SzGlmU, 109 U PmPpA, and 8.3 U RpPPK2‐3. Every 24 h, the reaction mixture was changed, and the reaction was observed for 6 h.

For the UDP‐GlcA synthesis, maximal product yields of 76–86 % were obtained in the first three cycles (Table 3). On the following cycles, only 70 and 59 % yield were obtained, respectively. This is also reflected in the space‐time yield, which is 2.4–2.7‐fold higher in the earlier cycles compared to the last cycle (Table 3). The decomposition of UDP‐GlcA by MnCl2 is here also noticeable, which can also be seen with the rising UMP concentration (Figure S6).

Over five cycles we reached 100 % conversion of GlcNAc to UDP‐GlcNAc (Table 4). However, we want to mention that we measured values higher than 10 mm UDP‐GlcNAc, which results because of the interaction of the remaining UDP‐GlcNAc after one cycle and the MP‐CE analysis. Nevertheless, the space‐time yield in the first fourth cycle (≥10.1 mm h−1) is higher compared to the last cycle (8.3 mm h−1). In conclusion, the enzymes of the EM UDP‐GlcNAc seem to be very stable for five days under reaction conditions. Although, we observed a reduction of the space‐time yield for UDP‐GlcNAc on the last cycle. The EM UDP‐GlcA is stronger affected by the repetition of the synthesis. Already on the fourth cycle, lower maximal yields were obtained with decreasing space‐time yield. The reason for this could be that under reaction conditions and end‐over‐end mixing for serval days the enzymes are colliding onto the wall of the reaction vessel and other MBs, which would destroy the enzymes over time. Besides, the enzymes are exposed to the reaction temperature and pH value, which could also accelerate the deactivation.

Table 4

Yields and efficiency of the immobilized EMs for the combined one‐pot synthesis of UDP‐GlcA and UDP‐GlcNAc with immobilized RpPPK2‐3 as an ATP regeneration system.[a]

Cycle	ATP [mm]	UDP‐GlcA		UDP‐GlcNAc		ATP regeneration cycle[b]		Efficiency [%]
		Space‐time yield [mm h−1]	Max. yield [%]	Space‐time yield [mm h−1]	Max. yield [%]	maximal	actual
1	2.5	7.8	86.3[d]	10.6	100[d]	7	6.7	92.1
2	2.5	8.3	84.1[c]	10.1	100[c]	7	6.4	91.7
3	2.5	7.5	75.6[c]	11.2	100[c]	7	6.5	93.2
4	2.5	6.4	69.8[d]	11.4	100[d]	7	6.4	91.4
5	2.5	3.1	58.7[e]	8.3	100[e]	7	5.8	82.8

[a] Each reaction contained 10 mm GlcA, 10 mm GlcNAc, 2.5 mm ATP, 20 mm UTP, 20 mm PolyP, 25 mm MgCl2, 10 mm KCl, 1.5 mm MnCl2, 100 mm TRIS‐HCl (pH 8.5), 4.3 U AtGlcAK, 0.6 U AtUSP, 0.6 U BlNahK, 1.2 U SzGlmU, 109 U PmPpA, and 8.3 U RpPPK2‐3. Every 24 h, the reaction mixture was changed, and the reaction was observed for 6 h. [b] The regeneration cycle was calculated with Equation (3) and relates to the maximal UDP‐sugar concentration. [c] After 1 h reaction time on that cycle. [d] After 2 h reaction time on that cycle. [e] After 4 h reaction time on that cycle.

The main result is that RpPPK2‐3 efficiently provides ATP for the synthesis of both UDP‐sugars with high yields over four days. The ATP recycling efficiency here was ≥91 %.

We concluded that HA one‐pot synthesis by the combination of the immobilized EMs for UDP‐sugar synthesis and PmHAS1–703 shall be productive over 3 days.

Immobilized PmHAS1–703

MBs with immobilized PmHAS1–703 were repetitively utilized over three days with an exchange of the substrate, cofactor, and buffer after 24 h reaction time (Figure 4). Each reaction started with 10 mm UDP‐GlcA and 10 mm UDP‐GlcNAc. The HA synthesis was detected by the formation of UDP. Within the first 2 h in the first cycle, the typical lag phase of PmHAS1–703 can be observed (Figure 4A). The lag phase is caused by the de novo synthesis of HA chains, which is significantly slower than the polymerization of HA chains. In the following cycles, this lag phase does not occur since probably HA chains are still entrapped in the porous MBs, and polymerization of HA chains is preferred over de novo synthesis by PmHAS1–703. After 24 h, a UDP yield of 61 % (12.2 mm) after one cycle was reached (Figure 4A). In the following cycles, the yield decreased (second: 11.1 mm, 55 %; third: 7.8 mm, 39 %) indicating a loss of activity of immobilized PmHAS1–703, which again could be caused by the end‐over‐end mixing. With a long incubation time and lower PmHAS1–703 activity, the chemical degradation of UDP‐GlcA becomes more obvious by the increased formation of UMP (Figure S7). The HA concentration reached after 24 in the third cycle with 461 mg L−1 (First cycle: 365 mg L−1; second cycle: 399 mg L−1) its highest concentration (Figure 4B), which is significantly lower than with soluble PmHAS1–703 (3 g L−1). In this measurement, it becomes also clear that in the second and third cycle HA is already present at the reactions start. This also influences the molecular weight (M W) of HA. Apart from agarose gel electrophoresis, the HA polymer size and dispersity were analyzed by right‐angle and low‐angle light scattering (RALS/LAS) (Figure 4C,D). Under the same conditions, with an average M W of 1.2 MDa after one cycle, we could reach a similar M W compared to soluble PmHAS1–703. However, the M W of HA is rising with each batch (Second cycle: 2.3 MDa; third cycle: 2.8 MDa). PmHAS1–703 is likely prolonging already existing HA chains rather than producing new chains since polymerization is preferred by PmHAS1–703 over de novo synthesis. In all cycles, we reached a dispersity between 1.07–1.18, which is similar to soluble approaches.[ , , ] In summary, we here demonstrate for the first time HA synthesis with immobilized PmHAS1–703 on MBs.

Figure 4

Repetitive HA synthesis with immobilized PmHAS1–703 on MBs. The reaction (5 mL) contained 10 mm UDP‐GlcA, 10 mm UDP‐GlcNAc, 25 mm MgCl2, 10 mm KCl, 1.5 mm MnCl2, 100 mm TRIS‐HCL (pH 8.5), and 0.18 U PmHAS1–703. Every 24 h, the reaction mixture was changed, and the reaction was observed for 24 h. Nucleotides and nucleotide sugars were analyzed with MP‐CE. See also Figure S7. (A) UDP development was measured with MP‐CE. (B) The HA concentration was measured with the CTAB turbidimetric method. (C) The samples taken after 6 and 24 h for each run were analyzed with agarose‐gel electrophoreses and HA was stained with stains all solution. S1: Select‐HATM HiLadder (Hyalose LLC) with a composition of defined HA preparations (495, 572, 966, 1090, and 1510 kDa). S2: high‐M W HA≥2 MDa. (D) The M W of HA was measured with RALS/LALS and presented as a box‐plot diagram. Ø M W: average molecular weight; Q: dispersity.

In a previous study, the single glycosyltransferase domains (GlcA−T and GlcNAc−T) of PmHAS were separately immobilized on agarose beads by covalent coupling and used in two batches of reactors, respectively. Starting from an HA4 tetrasaccharide GlcNAc was transferred in the GlcNAc−T reactor onto HA4. The product HA5 was prolonged with GlcA in the GlcA−T batch reactor. This stepwise process last 2 h each and was repeated until a HA20 oligosaccharide was synthesized. This approach is well suited for the production of HA oligosaccharides but not for HMW HA. We here present an approach where we use a non‐covalent but target method by which the synthesis of HMW HA is feasible.

One‐pot synthesis of HA with immobilized EMs

MBs carrying the enzymes for UDP‐sugar, HA synthesis, and ATP regeneration were combined for a repetitive one‐pot synthesis over three days (Figure 5). The reaction mixture was renewed after 24 and 48 h reaction time.

Figure 5

Repetitive one‐pot HA synthesis with immobilized enzymes. The reaction mixture (5 mL) contained 10 mm GlcA, 10 mm GlcNAc, 2.5 mm ATP, 20 mm UTP, 20 mm PolyP, 25 mm MgCl2, 10 mm KCl, 1.5 mm MnCl2, 100 mm TRIS‐HCl (pH 8.5), 3.4 U AtGlcAK, 0.5 U AtUSP, 0.5 U NahK, 1 U GlmU, 87 U PmPpA, 3.3 U RpPPK2‐3, and 0.29 U PmHAS1–703. After 24 and 48 h, the reaction mixture was exchanged. Nucleotides and nucleotide sugars were analyzed with MP‐CE. See also Figure S8. (A) UDP developments were analyzed with MP‐CE. (B) The HA concentration was measured with the CTAB turbidimetric method. (C) The samples taken after 6 and 24 h for each run were analyzed with agarose‐gel electrophoreses and HA was stained with stains all solution. S1: Select‐HATM HiLadder (Hyalose LLC) with a composition of defined HA preparations (495, 572, 966, 1090, and 1510 kDa). S2: high‐M W HA ≥2 MDa. (D) The M W, presented as a box‐plot diagram, and dispersity of HA was measured with RALS/LALS.

The decreasing yield for UDP (79 % after 1st cycle, 58 % after 2nd cycle, and 41 % after 3rd cycle) indicates a decrease in the activity of immobilized PmHAS1–703 over the cycles (Figure 5A). This is also seen in a decreasing HA concentration after each turn (1st cycle: 368 mg L−1; 2nd cycle: 104 mg L−1; 3rd cycle: 72 mg L−1) and by thinner bands in the agarose gel (Figure 5C). Compared to the soluble approach with RpPPK2‐3, we reached a similar HA concentration (368 vs. 410 mg L−1) after 24 h with 2.5 mm ATP and 25 mm MgCl2. Most importantly, the average M W of HA (3.6 MDa, Figure 5 D) is much larger compared to the soluble enzyme cascades (1.05 MDa). The low dispersity of 1.03 is in the same range as we have seen with soluble enzymes.[ , ] Unfortunately, we could not measure the M W and dispersity of HA on the third cycle since we detected either no or only a very weak refractive index (RI) signal with the RALS/LALS method. The reason for this is probably the low HA concentration after 24 h. Reaching HMW HA, we assume also here that HA chains are captured in the MBs.

The reduced HA synthesis with the re‐use of the immobilized enzymes could be rationalized by the analysis of the EMs for UDP‐sugar synthesis. In the first cycle, the syntheses of both UDP‐sugars were amazingly fast (Figure S8). After 1–2 h, each UDP‐sugar reached a yield of 74 % UDP‐GlcNAc and 100 % UDP‐GlcA. However, in the next cycle, we observed that UDP‐sugar synthesis is reduced (74 % UDP‐GlcA; 54 % UDP‐GlcA). In the last cycle, the highest yields are reached after 3 h (78 % UDP‐GlcA, 56 % UDP‐GlcNAc). However, this is still sufficient having in mind that PmHAS1–703 is simultaneously using both UDP‐sugars to produce HA. With 2.5 mm ATP and 25 mm Mg2+, the RpPPK2‐3 regeneration system also here was efficient enough to provide enough ATP for both syntheses (1st cycle: 85 % efficiency; 2nd cycle: 59 % efficiency; 3rd cycle: 62 % efficiency). We conclude that at 25 mm MgCl2 UDP‐sugar syntheses are also working in the HA one‐pot synthesis effectively.

RpPPK2‐3 can also use UDP to produce UTP, which would be a nice bonus in this one‐pot synthesis because too high UDP concentration can hamper PmHAS1–703 activity. RpPPK2‐3 is probably occupied with the ATP regeneration in the early beginning of the reaction, and later on, it could be that not enough PolyP is present for efficient UDP regeneration as we have seen earlier.

A comparison with our previous studies[ , , ] is necessary to rationalize the outcome of HA synthesis with immobilized enzyme modules. With soluble enzyme cascades a high UDP‐GlcNAc concentration gives high M W of HA which is due to a high K M for UDP‐GlcNAc of PmHAS1–703. Furthermore, the strong substrate inhibition by UDP‐GlcA limits the substrate concentration for high PmHAS1–703activity. Therefore, the UDP‐sugar ratio dictates the M W of HA for soluble PmHAS1–703 by the interplay between de novo synthesis and polymerization at the early beginning of the reaction.[ , ] In general, polymerization is faster and preferred over the de novo synthesis by soluble PmHAS1–703.[ , ] A high UDP‐GlcNAc concentration increases the activity of soluble PmHAS1–703, which we believe is the outcome of the preference of polymerization resulting in less but larger HA chains reaching 1.55 and >2MDa. The addition of PolyP in one‐pot synthesis limits UDP‐sugar synthesis and decreased the activity of soluble PmHAS1–703 which results in a lower HA M W (1.17 MDa). In the present study, immobilized PmHAS1–703, however, showed different kinetic characteristics The affinity towards UDP‐GlcNAc is greatly increased for immobilized PmHAS1–703 (K M 5.9 mm) (Table 1) in comparison with soluble PmHAS1–703 (K M 23.4 mm). Therefore, higher activity of immobilized PmHAS1–703 is reached at a lower UDP‐GlcNAc concentration compared to soluble PmHAS1–703. In addition, immobilized PmHAS1–703 showed generally higher activity with higher UDP‐GlcA concentrations than soluble PmHAS1–703 (Table 1). The consequence is that with higher activity also polymerization is preferred over de novo synthesis, which leads to 3 times longer HA chains (3.6 MDa, Figure 5D) in the presence of PolyP. We assume that the activity increase of PmHAS1–703 caused by immobilization is exceeding the activity decrease caused by PolyP. Another reason could be that the immobilization could affect the accessibility of UDP‐sugars and already produced HA chains. For example, the porous MBs capture the produced HA chains, and since polymerization is preferred the HA chain is elongated before a new HA chain is produced. All these reasons could rationalize, why we observe a lower HA concentration and a higher average M W with immobilized EMs in one‐pot synthesis compared to soluble EMs.[ , ]

With immobilized PmHAS1–703 alone we observe a lower M W after one cycle (1.2 MDa) when compared to the one‐pot synthesis (Figures 4 and 5). The reason for this could be that the UDP‐GlcA concentration is already high at the beginning (Figure S7), while in the one‐pot synthesis UDP‐GlcA must first be synthesized (Figure S8). At 10 mm UDP‐GlcA, the activity of immobilized PmHAS1–703 is already reduced due to substrate inhibition (Table 1) giving smaller HA chains at similar HA concentrations.

Most importantly, we demonstrate that the immobilized EMs can be re‐used at least three times over three days to produce UDP‐sugars and HA. Surprisingly, immobilized PmHAS1–703 produces HA over three days (Figure 4). Nevertheless, we observe that enzyme performance diminishes over time. Especially PmHAS1–703 reuse is extremely affected after the first cycle (Figure 5) ending in a low HA concentration. However, improving its stability could enhance the productivity of immobilized PmHAS1–703. This effort may be rewarding because enzyme production especially with seven enzymes like in this case is still cost‐ and time‐intensive. A prolongation and improvement of the reusability of such a reaction with so many enzymes could be promising.

Conclusions

As a straightforward development to our previous work, we intended to improve the enzymatic hyaluronic acid synthesis. We established a one‐pot synthesis for uridine‐5'‐diphosphate (UDP)‐sugar synthesis with an integrated adenosine‐5'‐triphosphate regeneration system that was capable to provide the donor substrates for hyaluronic acid (HA) synthesis. HA concentration of 368 mg L−1 with a high average molecular weight (3.6 MDa) and low dispersity (1.03) was reached. We can conclude that this is the first one‐pot synthesis of high‐molecular‐weight (HMW) HA with immobilized enzyme cascades. However, we believe that further improvement of the PmHAS1–703 stability could enhance our approach. This may be achieved by protein engineering. Also, continuous HA synthesis in a flow reactor can be considered to reduce mechanical stress and contact time with inhibitory reaction compounds (e. g., polyphosphate). This comes with new challenges like optimizing substrate residence time and adjusting UDP‐sugar ratios to balance de novo synthesis and polymerization. Also, the rising viscosity shall be considered.

The effort could be worth it since immobilization of PmHAS1–703 was rewarded with higher activity and with higher molecular weight of HA. HMW HA is preferred in medical applications like the treatment of arthritis since the half‐life in the human body is increased.

Experimental Section

Materials and methods were already used in our previous studies.[ , ]

Genes and enzymes

All genes in this work were cloned in our previous studies (Table 5).

Table 5

List of the cloned genes used in this study.

Gene	Origin	Gene ID	Vector	His‐Tag	Ref.
atglcak	Arabidopsis thaliana	819902	pET‐22b(+)	3'‐end	[37]
atusp	Arabidopsis thaliana	835333	pET‐16b(+)	5'‐end	[37]
pmppa	Pasteurella multocida	29387852	pET‐22b(+)	3'‐end	[37]
blnahk	Bifidobacterium longum	AB303839.1	pET‐22b(+)	3'‐end	[38]
szglmu	Streptococcus zooepidemicus	AF347022	pET‐22b(+)	3'‐end	[38]
pmhas1‐703	Pasteurella multocida	AF036004.2	pET‐22b(+)	3'‐end	[38]
rpppk2‐3	Ruegeria pomeroyi	SPO1727	pET‐22b(+)	3'‐end	[39]

Enzyme production with Escherichia coli BL21 (DE3)

Plasmids were transformed into E. coli BL21 (DE3) via heat shock. Positive clones were selected on agar plates with lysogeny broth (LB) medium and ampicillin (100 μg mL−1). As preculture, 20 mL liquid LB medium with ampicillin in 100 mL baffled flasks were inoculated with positive clones. The preculture was cultivated at 37 °C for 18 h and shaken with 120 rpm. 10 mL of the preculture was used to inoculate 1 L terrific broth medium in a 5 L baffled shaking flask (80 rpm, 37 °C). Cells were grown to an optical density at 600 nm wavelength (OD600) of 0.6–0.8. Simultaneously, the expression was initiated with 0.1 mm isopropyl β‐d‐1‐thiogalactopyranoside (IPTG, AppliChem GmbH), and the temperature was lowered to 25 °C. After 20 h, the cultivation was centrifuged (7000 rpm, 30 min, 4 °C) and the supernatant was poured off. The remaining cells were stored at −20 °C.

Cell disruption

Work was all the time done on ice. 4 g of frozen cells was dispersed in 10 mL binding buffer (20 mm sodium phosphate, 500 mm sodium chloride, 30 mm imidazole, pH 7.4). Cell disruption was done with ultrasound (6 : 30 min, 15 s pulse, 60 s pause, 52 % amplitude). Subsequently, cell particles were separated with centrifugation (15000 rpm, 30 min, 4 °C) and the supernatant was aliquoted (1 mL) and stored at −20 °C.

Immobilization on MBs

The used MBs are His Mag SepharoseTM Ni from GE Healthcare (Sweden) (37–100 μm). Because of the pET‐vector system, all used enzymes have either an N‐terminal or C‐terminal His‐tag. Therefore, we could load the proteins on the MBs following the manufacturer‘s instructions: 200 μL homogenous slurry (≈25 mg MBs) was dispensed into a 1.5 mL reaction tube and the storage solution was removed by applying the tube on a magnetic rack. 500 μL of binding buffer (20 mm sodium phosphate, 500 mm sodium chloride, 30 mm imidazole, pH 7.4) was used to equilibrate the beads. After removing the binding buffer, 1 mL crude extract was added and the beads were resuspended for 30 min with slow end‐over‐end mixing (20 rpm, 25 °C). Thereafter, the supernatant was removed and the MBs were washed three times with 500 μL binding buffer. The immobilized enzymes were further used for different reaction experiments. If desired, the enzymes were eluted after the reaction with 100 μL elution buffer (20 mm sodium phosphate, 500 mm sodium chloride, 500 mm imidazole, pH 7.4). During this working process, samples were taken and analyzed with Bradford assay, SDS‐PAGE, and Western blot. The immunostaining was done with a His6‐tag monoclonal antibody, which is conjugated with horseradish peroxidase (Roche Diagnostics GmbH, Germany).

Activity assays

Activity assays with immobilized enzymes took place in a volume of 2 mL in a 2 mL reaction tube at 25 °C and end‐over‐end mixing (20 rpm). Each reaction was implemented with 100 mm HEPES‐NaOH pH 8 and 10 mm MgCl2. The following substrates were added for each enzyme assay. AtGlcAK: 5 mm ATP and 5 GlcA; AtUSP: 5 mm UTP and 5 mm Glc‐1‐P; BlNahK: 5 mm ATP and 5 mm GlcNAc; SzGlmU: 5 mm UTP and 5 mm GlcNAc‐1‐P; PmPpA: 5 mm PPi; RpPPK2‐3: 10 mm UDP and 10 mm PolyP; PmHAS1–703: 8 mm UDP‐GlcA, 15 mm UDP‐GlcNAc, 25 mm MgCl2 (final concentration), 10 mm KCl, and 1.5 mm MnCl2. Within 24 h hourly, a sample was taken and the reaction was stopped by adding the sample 1 : 1 to 14 mm SDS, 2 mm para‐aminobenzoic acid (PABA, Sigma Aldrich, Munich. Germany), and 2 mm para‐aminophthalic acid (PAPA, Sigma Aldrich, Munich, Germany). Analysis of the samples of AtGlcAK, AtUSP, BlNaHK, SzGlmU, RpPPK2‐3, and PmHAS1–703 reactions was done with MP‐CE and samples of PmPpA were analyzed with a phosphate assay kit (MAK308, Sigma Aldrich, Germany). Following products were observed for each enzyme: ADP for AtGlcAK and NahK; UDP‐Glc for AtUSP; UDP‐GlcNAc for GlmU; ATP or UTP for PPK3, UDP for PmHAS1–703; Pi for PmPpA. The volumetric activity (U mL−1) was calculated with the slope of each product trend and the specific activity (U mg−1) was calculated with the enzyme concentration (Bradford assay).

Kinetic assay

Kinetics with immobilized enzymes took place in a volume of 2 mL in a 2 mL reaction tube at 25 °C and end‐over‐end mixing (20 rpm). The following additives were applied to every reaction: 100 mm Tris‐HCl pH 8.5, 25 mm MgCl2, 10 mm KCl, 1.5 mm MnCl2. The following changes were done for each enzyme reaction. AtGlcAK: 0.25–20 mm ATP, 10 mm GlcA; AtUSP: 0.25–20 mm UTP, 10 mm Glc‐1‐P; BlNahK: 0.25–20 mm ATP, 10 mm GlcNAc; SzGlmU: 0.5–20 mm UTP, 10 mm GlcNAc‐1‐P; PmPpA: 1–40 mm PPi; RpPPK2‐3: 0.25–20 mm ADP, 20 mm PolyP or 0.25–15 mm UDP, 20 mm PolyP; PmHAS1–703: 0.5‐20 mm UDP‐GlcA, 20 mm UDP‐GlcNAc or 1–30 mm UDP‐GlcNAc, 8 mm UDP‐GlcA. For AtGlcAK, AtUSP, BlNahK, SzGlmU, PmPpA, and RpPPK2‐3, several samples were taken between 0–30 min. For PmHAS1–703. several samples were taken between 0–24 h. Similar to the activity test every sample was stopped with 14 mm SDS, 2 mm PABA, and 2 mm PAPA and analyzed with MP‐CE (all other enzymes) or phosphate assay kit (PmPpA), respectively, and the volumetric and specific activities were calculated. The computation was done with the program SigmaPlot 10.0 (SPSS Software GmbH, Erkrath, Germany). Using either the Michaelis–Menten equation [Eq. (1)] or substrate inhibition (uncompetitive) equation [Eq. (2)], the kinetic parameters were determined:

with [S]: concentration of the substrate [mm]; K M: Michaelis constant [mm]; v: rate of formation of product [μmol min−1 mg−1]; V max: maximal rate of the system [μmol min−1 mg−1]; K iS: Dissociation constant of the enzyme‐substrate‐substrate complex.

One‐pot synthesis with immobilized enzyme

Enzymes were immobilized on MBs, as described above. The one‐pot synthesis took place in 1 mL in a 1.5 mL reaction tube or 5 mL in a 5.5 mL reaction tube. Tubes were mixed end‐over‐end at 25 °C. For the EM UDP‐GlcA (AtGlcAK, AtUSP, PmPpA, and RpPPK2‐3) following conditions were applied: 1–5 mm ATP, 10 mm UTP, 10 mm GlcA, 20 mm PolyP, 25 mm MgCl2, 10 mm KCl, 1.5 mm MnCl2, and 100 mm TRIS‐HCl (pH 8.5). For the EM UDP‐GlcNAc (BlNahK, SzGlmU, PmPpA, and RpPPK2‐3) following conditions were applied: 1–5 mm ATP, 10 mm UTP, 10 mm GlcA, 20 mm PolyP, 25 mm MgCl2, 10 mm KCl, 1.5 mm MnCl2, and 100 mm TRIS‐HCl (pH 8.5).

For the combination of EM UDP‐GlcA and EM UDP‐GlcNAc, 10 mm GlcA, 10 mm GlcNAc, 2.5 mm ATP, 20 mm UTP, 20 mm PolyP, 25 mm MgCl2, 10 mm KCl, 1.5 mm MnCl2, and 100 mm TRIS‐HCl (pH 8.5) were applied. For EM HA 10 mm UDP‐GlcA, 10 mm UDP‐GlcNAc, 25 mm MgCl2, 10 mm KCl, 1.5 mm MnCl2, 100 mm TRIS‐HCL (pH 8.5) were applied.

For the whole one‐pot synthesis 10 mm GlcA, 10 mm GlcNAc, 2.5 mm ATP, 20 mm UTP, 20 mm PolyP, 25 mm MgCl2, 10 mm KCl, 1.5 mm MnCl2, and 100 mm TRIS‐HCl (pH 8.5) were used. Samples were taken during a period of 24 h and stopped with SDS, PAPA, and PABA. Samples were analyzed with MP‐CE. After each cycle, a magnetic field was applied. MBs with immobilized enzymes were stuck on the wall of the reaction vessel. The reaction mixture was removed with a pipette and a fresh substrate/cofactor solution with the same components as the starting solution was added. If needed, the regeneration cycle was calculated [Eq. 3]:

MP‐CE analysis

We analyzed samples of enzymatic reactions (except PmPpA) with a multiplexed cePRO 9600™ system (Advanced Analytical Technologies, Ames, IA, USA). which is operated by the pK a‐analyzer software (Advanced Analytical Technologies. Ames. IA. USA). It is the same method we described earlier. The system has a 96‐capillary array, where one capillary had an inner diameter of 50 μm, an effective length of 55 cm, a total length of 80 cm. Before separation was started, the system was washed with 0.1 m NaOH followed by distilled H2O. As electrophoresis buffer, we used 70 mm ammonium acetate, 1 mm EDTA, and pH 9.2. Samples were sucked in by a vacuum of −0.7 psi for 10 s. The separation of nucleotides and nucleotide sugars was achieved by applying a voltage of 10 kV and measured by UV (254 nm). The resulting peaks were determined with the pK a‐analyzer software. The concentration of each analyte was calculated with corresponding calibration curves.

Phosphate assay kit

We used a phosphate assay kit (Sigma‐Aldrich. MAK308) to determine Pi concentration following the manufacturer's instructions. The concentration of the samples was determined with a standard curve ranging from 4–40 μm Pi. Therefore, we diluted the samples to adapt them to the standard range. The test took place in a 96 well plate, where 50 μL diluted sample and 100 μL malachite green reagent were mixed and incubated at room temperature for 30 min. The malachite green absorbance could be measured at 620 nm with the Multi‐Mode Microplate Reader SynergyTM 2 (BioTek®. Vermont. USA).

Agarose gel electrophoresis

Agarose gel electrophoresis was performed as described by Lee and Cowman. We applied as markers Select‐HA™ HiLadder (Hyalose LLC; 495, 572, 966, 1090, and 1510 kDa) and high‐M w HA≥2 MDa (Contipro Biotech. Dolní Dobrouč. Czech Republic). 4 μL loading buffer were merged with 20 μL of samples or markers, respectively. The separation was done with a 0.5 % agarose gel in TAE buffer [40 mm TRIS acetate, 1 mm ethylenediaminetetraacetic acid (EDTA), pH 8] by applying a voltage of 105 V. The separation took 55 min. Afterward, the gel was washed for 30 min in 30 % ethanol and colored with Stains‐All solution (Sigma‐Aldrich, 30 % ethanol, 6.25 μg mL−1 Stains‐All) for 24 h. The background was discolored with distilled H2O for 24 h. The gel was documented with the GelDoc (BioRad, Düsseldorf, Germany).

Size‐exclusion chromatography

The molecular weight distribution (M w and M n) and the relative molecular mass dispersity (dispersity index, DI) within the HA samples were analyzed with size‐exclusion chromatography (SEC). For separation, two columns were connected (A6000 M, A7000, Malvern Instruments) and samples were analyzed via RALS/LALS in combination with a RI‐detector to determine the HA concentration. Samples were diluted in buffer (10 mm PBS, pH 7.4) to a total volume of 180 μL. A volume of 80 μL sample was injected and eluted with buffer (10 mm PBS, pH 7.4, 0,35 mL min−1, 35 °C). Calibration was carried out using a PEO standard (0.5 mg mL−1 in PBS, Malvern Instruments). The samples were measured in duplicate and analyzed by using OmniSEC software and the dispersity was calculated with a quotient of M w and M n [Eqs. (4)–6]:

with M i: molar mass of chains with i units; n i: Number of chains with i units; m i: total mass of chains with i units.

CTAB turbidimetric method

As described in the literature,[ , , ] the concentration of HA can be measured through the formation of turbidity between HA and cetyltrimethylammonium bromide (CTAB). Therefore, we mixed 25 μL acetate buffer (200 mm sodium acetate, 150 mm sodium chloride, pH 6) with a 25 μL sample (1 : 20 diluted) in a 96 well plate and mixed it. Then we added 100 μL CTAB solution (25 g L−1 CTAB in 2 % w/v sodium hydroxide). After 5 min at room temperature, the turbidity was determined at 400 nm with the Multi‐Mode Microplate Reader SynergyTM 2 (BioTek®. Vermont. USA). The final concentration was calculated via linear regression of diluted (6.25–200 μg mL−1) HMW HA (Contipro Biotech. M w≥2 MDa).

Author Contributions

Conceptualization, J.G. and L.E.; methodology, J.G., and M.A.; investigation, J.G., and M.A.; data curation, J.G., and M.A.; writing original draft preparation, J.G. and L.E.; writing, review and editing, J.G., M.A., and L.E.; supervision, L.E.; funding acquisition, L.E., J.K

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Click here for additional data file.