News

2026

New preprint: mycobacteria catabolise lanosterol, not only cholesterol

Our preprint “CYP123, a sterol C4-methyl oxidase, acts with CYP51 to initiate mycobacterial lanosterol catabolism” is now on bioRxiv.

Mycobacteria are known to rely on host cholesterol during infection and persistence. We show that this is not the whole story. They also catabolise lanosterol, the first sterol intermediate of cholesterol biosynthesis, through a route regulated independently of the cholesterol one. Two cytochrome P450s encoded side by side carry out the first steps: CYP51 removes the C14-methyl group, and CYP123 then oxidises the C4-methyl group to a carboxylic acid.

The locus is conserved in more than 90% of Mycobacteriaceae, and several species grow on lanosterol as their only carbon source. One that barely does is Mycobacterium bovis var. BCG, the vaccine strain, which carries an ancestral frameshift in cyp123.

Two things follow. The first is a question. Cholesterol catabolites are not only food: ring-degradation products such as androstenedione are proposed to suppress inflammatory responses in the host. Are lanosterol catabolites immunometabolites too? We cannot say yet, but the locus switches on in conditions where lanosterol offers the cell no growth advantage, which is hard to explain by carbon alone.

The second is for drug discovery. If mycobacteria reach host sterols by two independent routes, blocking cholesterol catabolism alone will not deprive them of host sterols, and current efforts target only that route. Lanosterol utilisation deserves a look. Both enzymes bind azoles, so there is approved chemistry to start from.

Thank you to Dongxu Yuan, Lauren Hatfield, Hannah Ryan, Harry Williams, Mochen Dong, Josie Parker and Yi Jin for the work behind this.

Read the preprint: 10.64898/2026.10.31.760369

EFI web tools 2.0

The Enzyme Function Initiative (EFI) genomic enzymology web tools have been relaunched at efi.igb.illinois.edu. With the 2.0 release, the software behind the tools has been rewritten and modernised.

The tools still support the same genomic enzymology approach, which combines sequence similarity with genome context to explore protein families. They allow users to easily create and exploit sequence similarity networks (SSNs), genome neighbourhood networks (GNNs) and genome neighbourhood diagrams (GNDs). Researchers can use them to propose new biological functions for uncharacterised proteins, and to draw on what is known about an enzyme family for biotechnological applications.

If you used the tools before, you will need to create a new account on the new website. Please do not hesitate to contact the EFI team if you have any questions. Work is led by Nils Oberg with Russell Davidson and Hunter DeMeyer. John Gerlt and I provided scientific guidance. The work was funded by the National Institutes of Health/National Institute of General Medical Sciences (grant R24GM141196 to John A. Gerlt).

CD-HIT in Google Colab: an install-free alternative to the CD-HIT web server

The CD-HIT web server has been unavailable since 2022. As an alternative, I have released CD-HIT in Google Colab, a notebook that builds CD-HIT from source and clusters your sequences in the browser. You don’t need to install anything.

The parameters are set through form fields with explanatory notes, and every run records its own provenance (CD-HIT version, command line, input checksum and timestamp), so results are easy to reproduce and report.

If you use it, please cite CD-HIT and the notebook: 10.5281/zenodo.23503573.

EFI workshop at the launch of the Singapore National Centre for Engineering Biology

In October I travelled to the National University of Singapore for the opening of the Singapore National Centre for Engineering Biology (NCEB) on 5 October, followed by the Singapore Enzyme Biology Week. I am grateful to Wen Shan Yew for inviting me. Wen Shan, John Gerlt (University of Illinois Urbana-Champaign) and I work together on the ongoing development of the Enzyme Function Initiative (EFI) web tools.

On 8 October I ran a hands-on EFI workshop for graduate students who were new to the tools. We built sequence similarity networks (SSNs) to map protein families, then used genome neighbourhood networks (GNNs) and diagrams (GNDs) to look at the genomic context of their proteins of interest, with the aim of applying these approaches to their own projects.

It was a privilege to take part in the launch of the centre, and I look forward to seeing what these students discover.

New paper in Microbial Genomics: bacteriophages against citrus canker

Our paper “Characterization of putatively lytic bacteriophages able to infect Xanthomonas citri subsp. citri and identification of novel putative exopolysaccharide depolymerases” is out in Microbial Genomics.

Together with colleagues in Brazil, the US and the UK, we isolated and sequenced 72 bacteriophages that infect Xanthomonas citri, the bacterium that causes Asiatic citrus canker. Almost all of them are closely related, like the bacterial population they infect. Using structure-based annotation (HHsearch and AlphaFold), we found conserved tail-fibre proteins that may be depolymerases, able to break down the bacterium’s xanthan gum capsule. In greenhouse trials, one phage cut canker lesions by 60%. This makes phages a promising, sustainable alternative to copper sprays.

Read the paper: 10.1099/mgen.0.001821

New paper in PNAS: a cloaked glutamate decarboxylase in Mycobacterium tuberculosis

Our paper “A cloaked glutamate decarboxylase sustains the GABA shunt in Mycobacterium tuberculosis” is out in Proceedings of the National Academy of Sciences.

Mycobacterium tuberculosis Rv2531c had long been annotated as a lysine/ornithine/arginine decarboxylase, a superfamily of more than 26,000 sequences. It isn’t one. Combining bioinformatics, microbiology, metabolomics and enzymology, we demonstrate that it is instead an L-glutamate decarboxylase that sustains carbon flux through the γ-aminobutyric acid (GABA) shunt in M. tuberculosis.

I contributed the creation and analysis of the sequence similarity network that placed Rv2531c in the context of its enzyme family, showing that the superfamily is composed of distinct forms with different domain organisations, and that Rv2531c belongs to a group in which no enzyme had been experimentally characterised.

A good reminder of how easily annotation bias propagates across homologous sequences, and of what network-based approaches can reveal when we look at a family as a whole rather than one sequence at a time.

Work led by Hai Minh Thai and Luiz Pedro S. de Carvalho, with Debbie M. Hunt, Yugen Miyahara, Manisha Priya and Htin L. Aung.

Read the paper: 10.1073/pnas.2619778123

2024

Three new papers published in 2024