Synthesis of “Nereid,” a new phenol‐free detergent to replace Triton X‐100 in virus inactivation

Abstract In the 1980s, virus inactivation steps were implemented into the manufacturing of biopharmaceuticals in response to earlier unforeseen virus transmissions. The most effective inactivation process for lipid‐enveloped viruses is the treatment by a combination of detergents, often including Triton X‐100 (TX‐100). Based on recent environmental concerns, the use of TX‐100 in Europe will be ultimately banned, which forces the pharmaceutical industry, among others, to switch to an environmentally friendly alternative detergent with fully equivalent virus inactivation performance such as TX‐100. In this study, a structure–activity relationship study was conducted that ultimately led to the synthesis of several new detergents. One of them, named “Nereid,” displayed inactivation activity fully equivalent to TX‐100. The synthesis of this replacement candidate has been optimized to allow for the production of several kg of detergent at lab scale, to enable the required feasibility and comparison virus inactivation studies needed to support a potential future transition. The 3‐step, chromatography‐free synthesis process described herein uses inexpensive starting materials, has a robust and simple work‐up, and allows production in a standard organic laboratory to deliver batches of several hundred grams with >99% purity.

The pharmaceutical industry as well as other large consumers of TX-100 (I) 7,8 now face the challenge to find an alternative detergent to eventually replace TX-100 (I), and indeed, some alternative detergents have recently been suggested. [9][10][11] In order to not to adversely affect pharmaceutical bioprocesses, the best solution would be to replace TX-100 (I) with a detergent of analogous activity and behavior, thus minimizing any impact on the virus inactivation capacity, physical behavior, and removability during subsequent manufacturing steps, to ultimately support a smooth transition away from the phenol-containing detergent TX-100 (I). As detergents with a close structural similarity to TX-100 (I) were not available, an attempt was made to develop non-phenolic TX-100-like detergents and to evaluate their ability to inactivate lipid-enveloped viruses.

| Process intermediates
As a model matrix for plasma-derived products, Fraction II was used, that is, the process intermediate obtained during immunoglobulin production at the stage before virus inactivation/reduction steps. 22 The starting material had a pH of 5.2. Immediately before virus inactivation runs, the material was filtered through a 0.2 µm filter, and the protein concentration was adjusted to 28.9 AU 280-320 /cm with 30 mM NaCl. As a model matrix for recombinant protein production, human albumin (25%, Baxter AG) was diluted to 0.6 mg/ml using a buffer that contained 396 mM NaCl, 20 mM MES acid monohydrate, 10 mM CaCl 2 dihydrate, and 0.099% (v/v) PS80. This matrix is representative for an intermediate of the production process of recombinant factor VIII (ADVATE; Baxter AG). The starting material had a pH of 6.4 and was filtered through a 0.2 µm filter immediately before virus inactivation runs were performed. The choice of viruses reflects requirements as defined in the relevant guidelines 23,24 : PRV was included as a model for the family of Herpesviridae, while X-MuLV served as a model for retroviruses, which are of concern as retroviral-like particles have been reported in rodent cell lines utilized for recombinant protein production. Virus stocks were produced from infected susceptible cell lines essentially as earlier described. 25

| Virus inactivation by three-component S/D treatment
Each combination of S/D mix, virus, and matrix was investigated in duplicate runs using 30-50 ml of filtered process material per run.
Throughout the entire experimental runs, the process material was kept at 17 ± 1°C (plasma-derived model matrix) or 1 ± 1°C (recombinant model matrix) and continuously mixed by a magnetic stirrer. Spiking with the respective virus stock solution was performed at a ratio of 1:31 (v/v).
Subsequently, two samples termed spike control (SC) and hold control (HC) were drawn. SC was titrated immediately while HC was incubated in the same cooling unit as the SD-treated process material and titrated at the end of the respective run. For the vast majority of experiments, the difference in virus titer between SC and HC pairs was ≤0.5 log 10 [TCID 50 /ml], indicating that neither matrix constituents nor the chosen physicochemical parameters led to virus inactivation. As for manufacturing, the final target concentrations of PS80, TNBP, and TX-100 are 0.3%, 0.3%, and 1% (w/w), respectively. Taking the ratio of SD components into account, different S/D mixes were prepared by combining PS80 (Merck; 817061) and TNBP (Merck; 100002) with either TX-100 (Merck; 108643) or the potential surrogate compounds. Among these substances, TX-100 reduced (Sigma Aldrich; X100RS), Ecosurf EH-9, and Tergitol TNM-100X (both Dow Chemicals) were obtained commercially, while the other detergents were synthesized in-house (see ESI).
The spiked process material was weighed to calculate the amount of S/D mix to be added to reach 5% (plasma-derived model matrix) or 10% (recombinant model matrix) of the concentration as specified for manufacturing. These considerably reduced final concentrations enable to demonstrate virus inactivation kinetics, a regulatory requirement, 27 as opposed to virtually immediate virus inactivation at manufacturing concentrations. The detergent mix was added with a Hamilton syringe, and the actual amount of detergent added was determined by back-weighing the syringe. Samples for TCID 50 virus titration were drawn at 1-2, 10 ± 1, 30 ± 1, and 59 ± 1 min after detergent addition.
For all used process materials, the cytotoxicity of the respective S/D mix, as well as any possible matrix effects on cell lines used for FARCET ET AL.
| 3881 virus detection were tested and taken into consideration for the calculation of reduction factors.

| Looking for a bioisostere
Bioisosteres are surrogate compounds designed to exhibit similar physical properties with broadly similar biological activity to another chemical compound. The purpose of exchanging one bioisostere for another can be multiple such as improved potency, enhanced selectivity, altered physical properties, reduced or redirected metabolism, acquisition of novel intellectual property or circumvention of undesired metabolites. 12 The present investigation was based on the need to eliminate or modify the phenol toxicophore present in TX-100 (I) while retaining its full inactivation activity against viruses as well as its well-established process compatibility.
Bioisosteres possess near-equal molecular shapes and volumes and have approximately the same electron distribution. 13 As a previous study revealed that none of the tested, commercially available detergents fulfilled these prerequisites, 11 a library of new detergents for virus inactivation was established, to screen for novel TX-100 (I) replacement compounds.

| Structure-activity relationship (Sar)
It is appealing to understand which structural elements of TX-100 (I) are responsible for its outstanding virus inactivation activity, 14 as this subject has not been addressed in the public domain since its discovery. 1 Ultimately, understanding the role of each functional group in the structure might be crucial to find the most suitable replacement detergent.
In addition to its polyethylene glycol chain, TX-100 (I) features structural elements such as the 1,1,3,3-tetramethylbutyl alkyl chain, the aromatic ring, and the linkage to the polyethylene glycol chain.
All those elements can be chemically modified one by one by designed chemical synthesis, to reveal their contribution to the activity in subsequent virus inactivation studies.

| Sar#1: Removing the aromaticity
By removing the aromaticity from the structure of TX-100 (I), we obtained the structure of TX-100 reduced (II), a commercially available detergent ( Figure 1).
The compound shows excellent virus inactivation when S/D treatment was performed at standard process temperature and in a matrix representative for a plasma-derived product. As for recombinant proteins, lower temperatures are preferable during the downstream processes, and therefore, it is also important to investigate the effectivity of S/D treatment under these conditions. However, TX-100 reduced (II) failed to effectively inactivate a model retrovirus in a recombinant model matrix at such cold temperature ( Figure 2).
It is tempting to speculate that TX-100 reduced (II) has a diminished penetrating activity compared to TX-100 (I), which is both efficient at low and elevated temperature in plasmatic and recombinant matrices. A possible cause for the more moderate inactivation at low temperature might be the decrease of the virus membrane fluidity and its porosity as described in the fluid mosaic model. 15 3.4 | Sar#2: Removing the entire ring structure In a further SAR step, the whole ring was removed to yield model compound III, having an almost identically substituted alkyl chain as TX-100 (I) directly linked to the hydrophilic polyethylene glycol chain.
To access compound III, the corresponding highly substituted alcohol was mesylated followed by nucleophilic attack of a monoprotonated PEG polymer, which delivered the candidate III in a satisfactory 64% overall yield after a chromatographic purification in multigram scale (Scheme 1).
Interestingly, the inactivation properties of compound III were rather limited at warm temperature and negligible at cold conditions ( Figure 3). These results corresponded to analyses of the recently proposed TX-100 (I) surrogates Ecosurf EH-9 and Tergitol TNM-100X (both containing branched relatively short alkyl chains similar to III as lipophilic part) which equally failed to show significant inactivation, especially at a cold temperature (data not shown).
A hypothesis for the low activity of structures without the aromatic ring might be that the planar geometry of the aromatic ring enables to create a denser interaction with the virus membrane, by π-π stacking and/or hydrophobic interaction with membrane proteins, for instance, thereby favoring the formation of mixed micelles and disruption of the lipid envelope.
3.5 | Sar#3: Adding π-electrons to the system Next, an electron richer detergent IV containing two double bonds, based on the Geraniol structure as well as its fully hydrogenated F I G U R E 1 Structures of Triton X-100 (TX-100) and TX-100 reduced F I G U R E 2 Virus inactivation by S/D mixes containing TX-100 versus compound II (TX-100 reduced). A, Plasma-derived model matrix was spiked with PRV before the addition of S/D mix consisting of PS80, TNBP, and TX-100 or compound II. Duplicate runs were performed at 17°C at a final S/D mix concentration of 5% as specified for manufacturing. B, Recombinant model matrix was spiked with Xenotropic murine leukemia virus (X-MuLV) before the addition of S/D mix consisting of PS80, TNBP, and TX-100 or compound II. Duplicate runs were performed at 1°C at a final S/D mix concentration of 10% as specified for manufacturing. A, B, Samples were drawn after 1, 10, 30, and 60 min; viral loads of these samples were compared to a sample drawn before S/D mix addition to calculate the respective RF. Asterisks indicate viral inactivation below the detection limit. SD only shown if larger than the height of symbols). PRV, pseudorabies virus; RF, reduction factor; S/D, solvent/ detergent; SD, standard deviation; TNBP, tri-n-butyl-phosphate; TX-100, Triton X-100

S C H E M E 1 Synthesis route of detergent III
F I G U R E 3 Virus inactivation by S/D mixes containing novel compound III. Matrices were spiked with the respective virus before the addition of S/D mix (PS80, TNBP, compound III). Samples were drawn 1, 10, 30, and 60 min after S/D mix addition; viral loads of these samples were compared to a sample drawn before S/D mix addition to calculate the respective RF. A, Duplicate runs were performed at 17°C using a plasma-derived model matrix and PRV; the final S/D mix concentrations were 5% of manufacturing. Asterisks indicate viral inactivation below detection limit. B, Duplicate runs were performed at 1°C using a recombinant model matrix and Xenotropic murine leukemia virus (X-MuLV); the final S/D mix concentrations were 10% of manufacturing. SD only shown if larger than the height of symbols). PRV, pseudorabies virus; RF, reduction factor; S/D, solvent/detergent; SD, standard deviation; TNBP, tri-n-butyl-phosphate; TX-100, Triton X-100 FARCET ET AL.

| Sar#5: Modification of the alky side chain
To determine whether the length of the lipophilic part was also key to activity, compounds XIII and XIV were synthesized. Although n-butyl benzyl alcohol could be purchased from a chemical supplier, benzyl alcohol XII, with a biphenyl structure had to be prepared by an uneventful and high yielding (94%) Suzuki coupling step. Following standard protocol, both detergents XIII and XIV were obtained in gram quantities in respectively 34% and 24% overall yield (Scheme 5).
According to measurement calculations, the two detergents Detergent VI has several advantages over the other synthesized detergents from this library, as an alternative to TX-100 (I).
Firstly, VI showed effective and reliable virus inactivation for the process matrices and viruses tested. Secondly, its very similar structure to TX-100 (I) suggests that also the chemo-physical properties of the detergent (foaming, critical micelle concentration, solubility, detectability), as well as interaction with the protein(s) of interest will be highly similar. This should ensure that changes in the affected processes, such as analytical method or removal process, will be only minimal. Finally, from an ecological point of view, after release of compound VI into the environment, the polyethylene glycol chain is broken down in a similar manner as for TX-100; however, the revealed benzyl alcohol will be easily oxidized to the corresponding benzoic acid, [17][18][19] which is expected to interact less with estrogen receptors in animal toxicology studies than the corresponding octyl-phenol, as this metabolite has a completely different polarity and geometric structure than the phenol derivatives produced during TX-100 (I) degradation. In addition, further differences between phenols et benzyl alcohols in their abilities to be metabolized either through oxidative or bacterial processes might favor benzylic compounds in regard to the impact on their biodegradations. 20,21 S C H E M E 4 Synthesis route of detergent VII FARCET ET AL.
F I G U R E 5 Virus inactivation by S/D mixes containing novel compounds VI and VII. Matrices were spiked with the respective virus before addition of S/D mix (PS80, TNBP, compound VI/compound VII). Samples were drawn 1, 10, 30, and 60 min after S/D mix addition; viral loads of these samples were compared to a sample drawn before S/D mix addition to calculate the respective RF. A, B, Duplicate runs were performed at 17°C using a plasma-derived model matrix and PRV; the final S/D mix concentrations were 5% of manufacturing. Asterisks indicate viral inactivation below the detection limit. C, D, Duplicate runs were performed at 1°C using a recombinant model matrix and Xenotropic murine leukemia virus (X-MuLV); the final S/D mix concentrations were 10% of manufacturing. SD only shown if larger than the height of symbols). PRV, pseudorabies virus; RF, reduction factor; S/D, solvent/detergent; SD, standard deviation; TNBP, tri-n-butyl-phosphate; TX-100, Triton X-100 S C H E M E 5 Synthesis route of detergents XIII and XIV

| Scalable synthesis of VI
The laboratory-scale 6-step synthesis (Scheme 3) yields hundreds of grams of detergent within a couple of months, however, this route utilizes 4 chromatography steps as well as expensive (Tf 2 O, Pd catalyst), toxic (octylphenol, Zn(CN) 2 , DMF, MsCl), and hazardous (LiAlH 4 ) chemicals. Moreover, as the demand for material grew rapidly to accommodate feasibility studies, the synthetic route was upgraded to a convenient 3-step synthesis using cheap, nontoxic and safe starting materials available in bulk (Scheme 6). The work-up was optimized to ultimately replace the chromatography steps by convenient large-scale work-up procedures which reduced the volume of required solvents.

| Step 1
Taking advantage of the inducing effect of the methyl group of toluene to achieve the para-substitution of the target molecule, an acid-catalyzed Friedel-Craft alkylation using diisobutylene was envisaged to access the hydrocarbon XV. The original sulfuric acid was replaced by a more soluble nonafluoro-1-butanesulfonic acid.
Due to the bulkiness of the electrophile generated from the diisobutylene as a mixture 2,4,4-trimethyl-1-pentene /2,4,4-trimethyl-2pentene (3:1), no ortho product was detected. The work-up included a simple quench of the acid followed by extractive aqueous washes. Other isomers originating from likely cation rearrangements of the alkyl chain were also formed and could be removed F I G U R E 6 Virus inactivation by S/D mixes containing novel compounds XIII and XIV. Duplicate runs were performed at 17°C using a plasma-derived model matrix and PRV. PS80 and TNBP were combined with (A) compound XIII or (B) compound XIV and added to the virus-spiked matrix to reach a final concentration of 5% as specified for manufacturing. Samples were drawn 1, 10, 30, and 60 min after S/D mix addition; viral loads of these samples were compared to a sample drawn before S/D mix addition to calculate the respective RF. Asterisks indicate viral inactivation below detection limit. SD only shown if larger than the height of symbols). PRV, pseudorabies virus; RF, reduction factor; S/D, solvent/detergent; SD, standard deviation; TNBP, tri-n-butyl-phosphate; TX-100, Triton X-100 | 3887 entirely during the purification of the following Step 2. Exhaustive evaporation of the solvent is needed to remove the totality of the toluene, as it might react competitively with the NBS reagent during the next step.

| Step 2
The crude oily hydrocarbon XV was then engaged at 70°C with AIBN and NBS to form benzyl bromide XVI in a radical bromination in cyclohexane, during which after radical propagation, succinimide precipitates as a white solid. After filtration and concentration, the crude product containing some starting material as well as dibrominated product XVII was precipitated in IPA to yield a product with an approximate 90% purity.
Dibrominated product XVII was synthesized separately to confirm the identity of this main side-product. Further two recrystallizations in IPA delivered the benzyl bromide intermediate XVI in moderate yield over two steps (25%-30%) but at >99% purity (200 nm). Thus, all side products from the first and second step could be removed and a crystalline colorless solid (large plates) was obtained after drying.

| Step 3
Without solvent, a 5-molar excess of PEG400 (polyethylene glycol with an average MW of 380-420 g/mol) was deprotonated at 60°C by addition of tBuOK. To the mono-deprotonated PEG400, the benzyl bromide intermediate XVI was added in one portion before the reaction mixture was cooled 30 min later by addition of ice/water followed by pH adjustment, removal of the discoloration, and addition of EtOAc. Several aqueous washes assured that the excess of PEG400 was entirely removed. After concentration of the EtOAc, the residue containing some bi-functional product XVIII was redissolved in EtOH/water and non-polar washes were performed with cyclohexane to remove a large portion of the bi-functional product XVIII. An independent synthesis of bifunctional side-product XVIII This material is currently in use for feasibility studies, engineering and confirmation runs. Further chemical development studies and improvements are in progress to continuously optimize the process and prepare for large-scale industrial production of this detergent, which was named "Nereid" after Amphitrite, the Greek sea goddess and wife of Poseidon but also the mother of Triton, from whom comes the prefix "amphi-" (ἀμφί) meaning "both kinds, on both sides" used in modern words such as amphibious, amphitheater or amphiphile (detergent).

| CONCLUSIONS
The limited number of commercially available detergents that meet the pre-defined structural requirements of similarity with TX-100 (I) dramatically narrowed the screening possibilities described in this study. In this regard, the plan arose to develop a tailor-made detergent that fulfills the prerequisites of the virus inactivation potency of TX-100 (I) without being harmful to the environment. Structurefunction analyses indicated that the alkyl chain substitution must follow the 1,1,3,3-tetramethylbutyl pattern of TX-100 (I) and must be connected to an aromatic phenyl ring to retain full virus inactivation properties. The installation of a -CH 2between the aromatic ring and the polyethylene glycol chain resulted in conservation of the inactivation potential, thus the phenol-free detergent Nereid was discovered. A convenient 3-step synthesis was developed, which allows the production of hundreds of grams of pure detergent using reliable protocols and standard equipment. Further work is in progress to adapt the synthesis route to the multi-kg, and ultimately ton scales. Investigations are underway to analyze and dismantle the exact molecular mechanism of the S/D treatment. Although protein compatibility as well as process compatibility have been verified on a considerable range of products (gene therapy product, recombinant, and plasma-based biologics), 11 extended studies such as toxicity studies have recently been initiated.

DATA AVAILABILITY STATEMENT
The data that supports the findings of this study are available in the supplementary material of this article. Thomas R. Kreil https://orcid.org/0000-0001-9970-0987