https://www.science.org/content/blog-post/go-make-me-fifty-kilos-stuff

on the details of scale-up of a common reaction, and even if you’re not into organic synthesis it’ll show you the sorts of issues that drug manufacturing has to deal with. The authors (from Takeda in Japan) were looking at the reaction shown at right, which the medicinal chemists in the audience will assure you is a 100%-right-down-the-middle example of a typical drug synthesis step. The Suzuki-Miyaura palladium-catalyzed coupling conditions are the obvious choice for this transformation, but that phrase encompasses hundreds, thousands, who knows how many variations in catalysts, solvents, temperatures, additives and so on.

The team found that a set of very standard conditions worked well (70/30 mix of isopropanol/water, bis(triphenylphosphine) palladium dichloride as catalyst, potassium carbonate as base), and indeed that setup is almost bound to give you some product for most of these reactions. By adjusting the ratios of reagents, the concentrations, and the temperature the reaction was well-optimized for yield and for purity. The amount of boronic acid coupling partner was the biggest variable, as it turned out, and the peskiest side product was dimerization of the starting chloro compound. Another step that was optimized was treatment of the crude product with a solution of L-cysteine as a palladium scavenger (getting the Pd out of there is always a consideration after these reactions).

But then came time to do all of this on a fifty-kilo scale, and that’s when folks like me wave goodbye and wish everyone well, because I have never done a fifty-kilo reaction in my life and never plan to. (I think my experience has maxed out at between one and two kilos, and I did not enjoy those reactions very much, in retrospect). The first interesting wrinkle was that the contract manufacturing organization that was hired to do this work was located at notably higher altitude, so the boiling-solvent-mixture conditions for the original reaction took place at a lower temperature for them. The reaction just could not be driven to completion under those conditions. Just as you do when baking a cake at altitude, adjustments had to be made - in this case, it meant running things in a sealed vessel to get the temperature back up again under a bit of pressure.

Unfortunately, the first runs at 50kg scale, while going to completion, produced material that had too many impurities as compared to the smaller reference runs. Specifically, the leftover palladium levels were “alarmingly high”. And there were a whole list of things that could have changed: there’s that overpressure, to start with. There’s stirring and mixing, which is always going to be different on large scale. There’s heating, very much likewise (for example, the lab-batch reactions were immersed in a 100C oil bath, while the 50KG reactions were brought up to temperature in steam-jacketed reactors whose walls reached higher temperatures).

Further experiments showed that the only one of these that really affected the reaction seemed to be the higher-temperature heating, and that wasn’t hurting the yield, just the Pd-residue purity. Even further work showed that a big factor in those palladium levels was the presence (or absence) of air and oxygen. The lab-scale batches were exposed to ambient air at several points, while the 50kg reactors were strictly nitrogen-purged. And it turns out that you need some air in there, especially at higher temperatures. Pd species were produced at those hot spots in the walls of the reactor that were difficult to remove under the standard cysteine workup if they had not seen oxygen, but were much easier to clear out if air was introduced (probably because they were being oxidized to Pd (II). So a tube bubbling air into the reactors was introduced, although that had to be done as a separate step at the end of the reaction time (which was still effective at oxidizing Pd species). Furthermore, the nitrogen atmosphere was switched to static as opposed to nitrogen-flow, to keep from stripping residual oxygen out of the reaction.

Normally you feel safer keeping air (and especially oxygen) out of your reaction mixtures, not least because you don’t want anything igniting, but in this case it was crucial. This is a potential blind spot for scale-up, as the paper notes, particularly with the temperature changes producing new Pd species that had to be dealt with. Every little detail counts in this work!

https://www.science.org/content/blog-post/go-make-me-fifty-kilos-stuff

https://www.science.org/content/blog-post/lithium-orotate-revisited

After that big lithium-and-Alzheimer’s paper recently, I thought a look at the chemistry of the lithium orotate used therein would be worthwhile. So let’s get into ion behavior for a bit:

As the chemists in the crowd know, there are several general behaviors that you see for ionic compounds in solution. If you think of all ionic substances as fully solvent-separated solvated ions once they're in solution, just ions, all the same, the other possibilities are going to sneak up on you. And these vary according to both the anion and cation, naturally, and according to the concentration, and very much so with the nature of the solvent and whatever other species might be floating around in there (overall ionic strength is certainly a factor, for one). Let’s stick with water as the solvent for the three most distinct classifications:

1. A fully solvated ion pair. That’s what you’d see with (for example) a low concentration of sodium chloride in water. The most energetically favorable state has the sodium cation and the chloride anion each surrounded by their own “solvation shells” of water molecules; it’s like they are each in their own bubbles of slightly-more-orderly water. The ions are not really “seeing” each other at all.

2. A solvent-separated ion pair, which can also be known as an “outer-sphere complex”. In this situation the anion and cation are separated by (pretty much) a single layer of water molecules (or indeed a single water molecule itself). In this case there certainly is an electrostatic interaction between the two anions, but the lowest energetic state of the system includes a solvent molecule in there too.

3. A contact ion pair, which can also be known as an “inner-sphere complex”. Here the anion and cation are right next to each other, fully electrostatically paired. Indeed, this situation can usually be described as “partially covalent”; the interaction is that tight. It’s like the far end of the spectrum of polarized covalent bonds, like drawing a sulfoxide as an S-plus connected to an O-minus. The two ions are surrounded by a common solvation shell of water molecules; there’s nothing between them.

There are several factors that go into the thermodynamics of these states. There’s outright Coulombic attraction (positive charges and negative ones), but note that Coulomb’s Law includes a term in the denominator for the dielectric constant of the medium (so water is going to be rather different than less polar solvents and more apt to separate things). And you’ll also have to keep in mind that your ions are going to have a polarizing effect on those nearby solvent molecules, somewhat cancelling out the situation compared to “naked charges” alone. You’ve also got enthalpic contributions from all those solvation interactions with the water molecules, balanced with the entropy changes that come from making more orderly solvation shells out of those waters. And there’s the loss of entropy that comes from having ions associated with each other rather than swimming around randomly.

OK, now what do we know about lithium orotate’s behavior? I ask because many people (in the comments here and elsewhere) have had a hard time imagining that it can be all that different from any other lithium salt. With lithium chloride or lithium carbonate, you would absolutely expect the two ions to go off on their separately solvated adventures by themselves, so why shouldn’t any lithium whateverate do the same?

It is a question with a surprisingly long and controversial history, which is very well summed up here. and in even more detail in this article. In short, claims were made in 1973 that lithium orotate dosing led to higher CNS concentrations than lithium carbonate dosing. A followup study in 1976 did not confirm this, but another in 1978 apparently did see such differences (up to threefold higher concentrations with the orotate). A 1979 followup, though, suggested that this could be an artifact of impaired renal function after the orotate dosing, and that report seems to have shut down this area of inquiry for some time. More recent toxicological investigations have not seen any such effects, however. In fact, lithium carbonate seems to have more renal toxicity problems itself - it’s possible that lithium orotate is a safer compound, pharmacokinetic and efficacy claims aside.

But what about those pharmacokinetic differences? Are they real, and if so, how does this occur? Well, the PK of lithium salts in general seems to be a battleground (see section 6.1 here). Most lithium dosing in the psychiatric field is lithium carbonate, but that’s due to its easier formulation compared to lithium chloride (it’s non-hygroscopic, i.e. it does not soak up moisture from the air). Lithium chloride itself has some regulatory issues left over from its (over)use in salt substitutes in the 1930s and 40s as well. Lithium citrate is available as a substitute for people who have difficulty swallowing the lithium carbonate caplets, and there are varying reports of whether it has any PK differences compared to the carbonate. Lithium sulfate seems to have no real differences.

Orotate salts, though, may well be a different matter. It’s been observed, for example, that magnesium orotate does not have the laxative effects of common magnesium salts, which suggests that it does not ionize under physiological conditions the way that those do. The lithium/Alzheimer’s paper showed that lithium orotate solutions showed notably lower conductivity than other lithium salt solutions, and that is indeed a measure of their degree of ionization (i.e., more contact ion pairing than for the other salts). It is possible that the lithium-orotate pair is handled as a single substance. At the destination end, there is evidence that orotate is transported via a urate receptor (URAT1) which is found in both the kidney and the choroid plexus (for entry into the brain), and it may be taken up through nucleotide transporters as well. And once in the cell, orotate is already an intermediate in pyrimidine synthesis, which might be a way to finally liberate the lithium counterion.

More needs to be done to shore up all these ideas, but they are not implausible. This paper goes a way towards that, showing that lithium-driven mouse behavioral assays are significantly different with the orotate salt, and that inhibition of anion transport pathways (or of the pentose phosphate pathway for nucleotide synthesis) seem to shut off these effects. So there is reason to think that lithium orotate could indeed be different from other lithium salts, and that these differences are exploitable for its use in lithium supplementation into the CNS. That of course is a separate issue from “Is lithium deficiency the cause of Alzheimer’s” and from “Would lithium supplementation be a useful Alzheimer’s therapy”. But it would behoove us to figure this out in case the answer to either of those latter questions is “yes”.

https://www.science.org/content/blog-post/lithium-orotate-revisited