Skip to main content

Drug Assays

The Antifungal Literature Has Problems

Antifungals are one of the areas I’ve never actually done any drug research in. But I’ve always heard that it’s a hard row to hoe, to use an expression from my upbringing on the Mississippi Delta. Fungi have some unusual biochemical pathways, and thus some good potential drug targets, but getting those to work in the real world is another thing entirely (much like the situation with gram-negative bacteria). Most antifungal drugs work through just a short list of mechanisms, which has (naturally) encouraged the development of resistance. And if you are unlucky enough to be dealing with a resistant fungal infection, your options are limited indeed. (There’s another big side to antifungal discovery, in agriculture, that shares the same concerns).

You’d think that discovering antifungals (or antibacterials) would be easier. After all, as opposed to a lot of disease areas, you can set up an extremely relevant assay without too much trouble. If you’re trying to see if anything on your list of compounds kills fungi or bacteria or what have you, an assay at that level will indeed tell you if they do. So in that sense, a fungicidal compound in such an assay is not a false positive; it really does kill fungi. But it almost certainly does not kill fungi in a useful manner. It may well also kill other forms of life (bleach comes to mind) or it will turn out to work through one of the existing mechanisms, with all the baggage that comes along with that. These are why the most direct approach to trying to find new compounds against bacteria or fungi (screen ’em and see) isn’t done very much any more, because you would end up spending the overwhelming amount of your time wading through compounds that apparently do just what you want, but are in fact useless for your purpose of finding new drugs. You have to set things up so that you’re not re-inventing a zillion wheels.

Target-based drug screening against individual proteins has its problems, too. As with antibacterial drug discovery, the sorts of compounds that hit your targets of interest can be rather odd looking, so it’s worth your time to cast the net broadly. (From one standpoint, that’s good news, but when you consider that decades of this have only led to the situation we have now, perhaps it isn’t). And whenever you run a big list of varied compounds against an unusual or difficult target, you can expect that far too many of the hits from your screen will turn out to be false positives, and this time they’re the “really false” ones: compounds that make your assay readout do what you’re looking for, but are not acting in the way you want at all. (Some of these do overlap into cell-based assays, killing cells for reasons that you’re not going to be able to make anything out of, either).

The authors of this new paper in J. Med. Chem. have a problem with the way antifungal research has been going recently, for just these reasons. They’re from Syngenta, a company with a long history (under various banners!) in ag-chem R&D, and from their standpoint, things are not looking too good:

Since the turn of the century there has been a significant increase in the number of publications with the word or concept “antifungal”. . .The number of these papers has increased by more than 2-fold in less than 14 years. It could easily be perceived that this is a reflection of a similar increase in the number of hits or leads and subsequently numbers of antifungal projects in development in the pharmaceutical or agrochemical industry. Unfortunately, this has not been the case. The reality is that very few compounds are reaching development and this raises the question: Why?

Many factors may come into play. To debate of all of them here would be more of a philosophical debate than a scientific endeavor. However, we believe that one key reason is the meager quality of some of these new inhibitors reported in the antifungal literature; many of which contain undesirable features.

They’re estimating that up to 80% of the antifungal compounds reported in the literature over the last five years are actually false positives. And it doesn’t have to be this way – attention to detail and a willingness to challenge your own results would deal with most of these:

This review sets out to highlight a small sample of undesirable structural features that are repeatedly published in the literature. While the science described is more often than not excellent, key questions are often ignored: Why are the compounds active? The desired mode of action might have been confirmed in vitro, but is it truly responsible for the observed antifungal activity? Do some undesirable chemical features confer fungicidal activity owing to nonspecific binding? If the latter is true, what is the likelihood that such an unselective compound would ever reach the market when the huge hurdle of registration is taken into account?

Those are exactly the sorts of uncomfortable questions that need to be asked, for sure. And as this paper goes on to show, all too many of the recent compounds reported turn out to be a parade of rhodanines, Michael acceptors (and other thiol-reactive electrophilic structures), quinones and other redox-cyclers, and on and on. I have not been following the antifungal literature closely, but if this paper has drawn from a representative selection, it’s not in very good shape.

You won’t find the likes of Syngenta working on these compounds for long, for the very clear reasons that compounds like these tend to fail the moment you push them a bit further into actual development. (Indeed, the authors mention that “None of the above-mentioned compounds have shown useful activity once promoted to our in-house higher tier tests“).

Here’s the wrap-up:

We realize that the task of discovering an antifungal is challenging and the funding of research limited, and this is why we believe that all researchers should be aware of certain pitfalls before embarking in extensive (and expensive) research programs. While we are happy to see the number of papers dealing with antifungal research increasing over the years, we would prefer even more to see an increasing number of successful cures being discovered. Ultimately, this should be our only goal.

Absolutely right. Unfortunately, incentives are not always aligned towards this.Looking through the references in this paper, the examples are overwhelmingly from academic labs or startups that appear to be recently emerged from them. If your goal is instead to maintain your funding, get tenure, or persuade some investors that your small startup company is on to something, then compounds of the sorts described might work out just fine.

12 comments on “The Antifungal Literature Has Problems”

  1. me says:

    Doesn’t sound dissimilar to the antibac paper published by my former dept a few years back – really not an easy area to work in.
    I remember colleagues at a later job saying that antibac needed to ‘move with the times’ and ‘use modern methods’. My response was always,
    ‘yeah, good luck with that.’

  2. Lynn says:

    And, as you say, it is much the same for antibacterials. People have to know [or at least learn] what it is that they don’t know. It’s easy to kill fungi and bad bugs – that’s not the problem. In my experience [with antibacterials, anyway], almost all in vitro hits [enzyme inhibitors] from chemical libraries that can kill bacteria do so for reasons other than inhibiting the presumed target.

  3. Not to fear! The Burke Synthesis Machine is on it!

  4. Peter Kenny says:

    The difficulty in dealing with dealing with ‘uglies’ from screening is in distinguishing ugliness that has actually been observed from an opinion that something is ugly. The thiol activity is asserted for compound 1 (an alkylidene-TZD) but were any of the observations of thiol activity for the chemotype made for alkylidene-TZDs lacking a substituent on nitrogen. These are likely to be anionic under some assay conditions and anionic species might be expected to be less reactive as Michael acceptors.

    Discussing the original PAINS article, the authors notes, “…report on pan assays interference compounds (PAINS), with a list of structural features of frequent hitters from six different and independent assays”. I couldn’t help wondering what they meant by “independent” since if the assays are different, shouldn’t they also be independent? I’m guessing that the authors are reading the comments and it would be useful to get some clarification.

  5. HFM says:

    @Peter Kenny: I’m not the authors, but I read that as “different assay methods used on different targets” – ruling out artifacts specific to a certain assay or target class.

    Also, as an ignorant biologist, I’m curious as to how much consensus on “ugliness” there really is. It seems to me that a large percentage of screening hits get an immediate (and fairly unanimous) veto once shown to actual chemists. Which makes me scratch my head as to why the darn things are in the library at all. My best guess is that the compound libraries are legacy and it’s not worth the effort to streamline them, but I’m not sure of that.

  6. Peter Kenny says:

    @HFM The data for the original PAINS article came from 6 AlphaScreen assays (i.e. same detection technology) for inhibitors of protein-protein interactions and only three of the targets were disclosed. Compounds capable of quenching/scavenging singlet oxygen sufficiently rapidly would show activity in an AlphaScreen assay and one cannot in general equate taking out singlet oxygen with reacting with proteins. I’ve linked the first of a 3 part series of blog posts on PAINS as the URL for this comment and there are links at the top of each of these to help you find the others.

    Ugliness in compounds tends to be a relative term and people are happier to use the captive bolt on a compound if they believe that they have better. My response to one project manager in my Pharma days who asked if I thought we had a lead was, “depends on how desperate you are”. One big challenge in dealing with screening hits is separating fact from myth and I will certainly admit to making decisions based on percived ugliness of compounds. Establishing rules for ugliness has the benefit of making things more reproducible but it is a mistake to assume that the existence of rules implies that they are necessarily objective and relevant. I should also say that in three of the campaigns that I worked on, we decided that none of the screening output represented a viable way forward.

  7. Thomas says:

    See my website for the obvious XKCD: When you see a claim that a common drug or vitamin “kills cancer cells in a petri dish”, keep in mind: [cartoon] so does a handgun.

  8. Nile says:

    It’s going to be a long haul, then: basic research, probably awaiting the development of tools for studying activity in soil – which is *the* place to look for antibacterial activity – and possibly in plants.

    Essentially, go looking for places where fungi live, and study organisms which would do better if fungi didn’t.

    From what I know of soil biology, it’s incredibly difficult to study, and the lead time from observed antibacterial activity to clinic is 15-20 years.

    It is not, in any way shape or form, a commercially attractive prospect.

    , and other o

  9. Jonathan says:

    @Nile – Michael Fischback gave a really nice talk at AGBT this year on using genomics to find small molecules and synthesis pathways from the microbiome.

  10. Vader says:


    I’m pretty sure a handgun kills cancer cells in vivo as well.

    It’s the side effects that are the problem.

Comments are closed.