Lumina Probiotic still exists! When we noticed that the probiotic bacteria wasn’t doing a good enough job at surviving on the teeth, we paused sales. Now, we’re back to the lab, re-engineering the strain to make it better withstand the brutal, Darwinian environment that is the human oral cavity. The reengineering phase will likely take six months, after which point we’ll be seeking volunteers to test colonization. Nonzero chance AI kills us before we reach the end of this probably-years-long-process, but we plant trees in whose shade we may never sit.
In keeping with our spirit of crowdsourced review, we are once again offering a bug bounty if folks can spot any novel issues with this new plan. (Though have your assessment be something more substantive than a Claude paste, please.)
The Non-Technical Explanation for Non-Biologists
The bacterium S. mutans lives on your teeth, eats sugar, and craps lactic acid. This is bad for the teeth. This proposed modified strain uses a different metabolism to produce less acid.
While the original strain produced ethanol as its waste metabolite, the modified strain produces 2,3-butanediol. Additionally, it also borrows a gene system from Streptococcus salivarius — a friendly species naturally associated with cavity-free individuals. This breaks down urea into a base, to actively raise local pH via neutralization.
Further, the bacteria is being engineered to be a better neighbor to other beneficial mouth bacteria. It produces a small protein which, through recruiting salivary amylase, breaks down starches into maltodextrins for its neighbors.
Additionally, it has a few other tweaks, to make it more resistant to the hydrogen peroxide secreted by its neighbors, to make it repel unmodified S. mutans, and of course the same type of “don’t change your genome any more” mechanism present in the original strain.
The Longer Technical Explanation
The current strain is based on Streptococcus mutans variant UA159. UA159 is a standard reference strain in the field of microbiology and one of the most well-characterized genomes in the field. Unmodified UA159 is a prolific acid and biofilm producer. It contains Mutacin IV (effective against the mitis group and induced by the quorum-sensing response), Mutacin V (effective against non-streptococcal targets), and Mutacin VI. Importantly, UA159 has been shown to lack the cnaB-cbpA-cmn collagen-binding locus—the sort of thing which causes pericarditis—and is characterized as a relatively low-invasive strain.
Phase one of the strain design is to produce a strain with a more robust metabolism, which doesn’t result in acidification to the point that it dissolves tooth enamel. This will be accomplished via a two-part strategy, where a novel operon is introduced to offload pyruvate metabolism into a pathway where the final end product is 2,3-butanediol. 2,3-butanediol is innocuous to the human oral environment and is pH neutral.
In oxygenic environments, NoxE is active and performs this reaction:
2 NADH + O2+ 2 H+ 2 H2O + 2 NAD+
NoxE is derived from Lactococcus lactis and is a close homolog of Sm Nox-2, however it has been established that NoxE is a robust enzyme with effective pH optimal range spanning from pH 6.0-9.0.
Under anaerobic conditions we must contend with the massive reduction capacity of the naturally high-expressing LDH pool within Sm. Furthermore, the LDH variant from Sm is allosterically activated by high levels of fructose 1,6-bisphosphate. Under conditions of naturally high levels of sugar exposure, Sm is able to quickly convert those sugars to lactate, acidifying the local environment to kill off competing commensal strains to establish its niche. The lesson here is that LDH activity is not bad in and of itself (many strains produce LDH and acid byproducts) - it’s that unopposed LDH activity leads to excessive acidification, healthy microbiome disruption, and enamel erosion.
The goal of the metabolic redesign of the Strain is not to eliminate LDH activity altogether, but to attenuate the activity below a level that produces harmful acid production, and supplements alternative metabolic pathways of carbon reduction (ex. the 2,3-butanediol operon). This will be accomplished by trialing several different ribosome-binding sequences to tune down translation of the LDH protein.
Additionally, there is a ~24 kb integrative and conjugative foreign element called TnSmu1 that appears to function to induce growth arrest under stress. It also carries machinery for conjugative transfer. This entire locus will be deleted and replaced with the urease operon from S. salivarius. This ~ 7 kb operon consists of ureI (acid-gated urea transporter), ureABC (urease trimer), ureEFGD (accessory proteins for nickel insertion into the apoenzyme), and ureMQO (nickel-specific ABC transporter). Urea is highly abundant in saliva and we do not need to burn ATP for the reaction other than initial protein synthesis.
The reaction is as follows:
Urea ( CO(NH2)2) ) + H20 2NH3 + CO2
We predict that the native Sm CodY regulator will be active in regulating this operon in similar fashion as it is natively regulated in S. salivarius (CodY is a repressor sensing branched-chain amino acid and GTP pools). The conserved regulatory motif is “AATTTTCNGAAAATT”.
Once the metabolic background is formalized (robust growth under aerobic and anaerobic conditions, terminal pH > 5.4), we will further “commensalize” the Strain to promote the colonization of healthy microbes to the tooth surface (Phase two). This will be done by adding the secreted amylose-binding protein (abpA from S. gordonii) the genome. This tiny 20 kDa protein strongly interacts with human salivary α-amylase for the host pioneer colonizer. That enzyme functions to break down starches. Sm GtfB is known to interact with AbpA to promote GtfB enzymatic activity and biofilm scaffolding. The colocation of GtfB-AbpA with sequestered amylase would also release sugars such as maltodextrins as a food source for the local microbes to ingest (including the Strain).
The pioneer colonizer S. gordonii also has two isoforms of the SpaP adhesin called SspA and SspB. This duo can, in theory, outcompete SpaP for immobilized gp340, and introduction of N1182G/V1185P to SspB should abrogate binding to P. gingivalis. A broader substrate-binding repertoire is possible in a Sm SspA/B genetic background, in addition to co-aggregation with commensal microbes such as Actinomyces. The secreted Gtf’s play a substantial role in the organism’s attachment to surfaces, however, and SpaP is known to form biofilm matrix stabilizing functions. Thus, it is unclear at this time if a SspA/B-expressing strain has a clear fitness advantage.
The Strain, if successful, will inhabit an ecological niche with a higher number of commensal microbes and streptococci than typical for a Sm niche. This will require augmentation of the ability to withstand peroxide stress most likely (Phase three). Sm are naturally catalase-negative and do not encode heme biosynthesis pathways for typical catalase enzymes. We will express a low amount of manganese catalase originally found in Lactiplantibacillus plantarum. This 30 kDa enzyme forms a homohexamer to perform the reaction 2 H202 2 H20 + O2. Together with the native alkyl hydroperoxide reductase, thiol peroxidase, and Dpr peroxidase defense systems, this addition will bolster resistance to reactive oxygen species typically used to attack Sm.
Lastly, an aspirational addition to the Strain’s native anti-Sm arsenal would be the secretion of a Spa-targeting VHH nanobody (Phase four). A conserved epitope has been identified and in silico generative VHH models are being designed. Our Strain would express a point-mutated version of the epitope immune to VHH targeting, while native strains would be susceptible. If a high-affinity VHH (or multivalent polypeptide) were to be developed and proven effective in the low pH environment, it would be a substantial Sm-specific addition to preventing Sm colonization specifically. Such a mechanism could also be copied to inhibit other harmful microbes from colonizing, such as P. gingivalis. For example, secreted VHH could be designed to target the catalytic cores of the gingipains Kgp and RgpB.
The final genomic change to the Strain will be the deletion of comX, the master regulator of competence genes in the organism. This will render the strain virtually incapable of uptake of foreign DNA.
Timelines and Signups
This reengineering work will likely take months. Our new lab is operating out of Portland. If at some point in the future you’d like to try the new strain and help us confirm that it can survive indefinitely in the human mouth, let us know! By signing up here: Signups.
Hi folks. It’s been a while.
Lumina Probiotic still exists! When we noticed that the probiotic bacteria wasn’t doing a good enough job at surviving on the teeth, we paused sales. Now, we’re back to the lab, re-engineering the strain to make it better withstand the brutal, Darwinian environment that is the human oral cavity. The reengineering phase will likely take six months, after which point we’ll be seeking volunteers to test colonization. Nonzero chance AI kills us before we reach the end of this probably-years-long-process, but we plant trees in whose shade we may never sit.
Also, for the historical record, I posted a response to our criticisms from back in the day, though that’s mostly me thinking about the apocalypse and not wanting to die unshriven.
In keeping with our spirit of crowdsourced review, we are once again offering a bug bounty if folks can spot any novel issues with this new plan. (Though have your assessment be something more substantive than a Claude paste, please.)
The Non-Technical Explanation for Non-Biologists
The bacterium S. mutans lives on your teeth, eats sugar, and craps lactic acid. This is bad for the teeth. This proposed modified strain uses a different metabolism to produce less acid.
While the original strain produced ethanol as its waste metabolite, the modified strain produces 2,3-butanediol. Additionally, it also borrows a gene system from Streptococcus salivarius — a friendly species naturally associated with cavity-free individuals. This breaks down urea into a base, to actively raise local pH via neutralization.
Further, the bacteria is being engineered to be a better neighbor to other beneficial mouth bacteria. It produces a small protein which, through recruiting salivary amylase, breaks down starches into maltodextrins for its neighbors.
Additionally, it has a few other tweaks, to make it more resistant to the hydrogen peroxide secreted by its neighbors, to make it repel unmodified S. mutans, and of course the same type of “don’t change your genome any more” mechanism present in the original strain.
The Longer Technical Explanation
The current strain is based on Streptococcus mutans variant UA159. UA159 is a standard reference strain in the field of microbiology and one of the most well-characterized genomes in the field. Unmodified UA159 is a prolific acid and biofilm producer. It contains Mutacin IV (effective against the mitis group and induced by the quorum-sensing response), Mutacin V (effective against non-streptococcal targets), and Mutacin VI. Importantly, UA159 has been shown to lack the cnaB-cbpA-cmn collagen-binding locus—the sort of thing which causes pericarditis—and is characterized as a relatively low-invasive strain.
Phase one of the strain design is to produce a strain with a more robust metabolism, which doesn’t result in acidification to the point that it dissolves tooth enamel. This will be accomplished via a two-part strategy, where a novel operon is introduced to offload pyruvate metabolism into a pathway where the final end product is 2,3-butanediol. 2,3-butanediol is innocuous to the human oral environment and is pH neutral.
The layout of this novel operon is as follows:
Lactate dehydrogenase engages in this metabolic reaction cycle:
The novel operon engages in this metabolic reaction cycle:
In oxygenic environments, NoxE is active and performs this reaction:
NoxE is derived from Lactococcus lactis and is a close homolog of Sm Nox-2, however it has been established that NoxE is a robust enzyme with effective pH optimal range spanning from pH 6.0-9.0.
Under anaerobic conditions we must contend with the massive reduction capacity of the naturally high-expressing LDH pool within Sm. Furthermore, the LDH variant from Sm is allosterically activated by high levels of fructose 1,6-bisphosphate. Under conditions of naturally high levels of sugar exposure, Sm is able to quickly convert those sugars to lactate, acidifying the local environment to kill off competing commensal strains to establish its niche. The lesson here is that LDH activity is not bad in and of itself (many strains produce LDH and acid byproducts) - it’s that unopposed LDH activity leads to excessive acidification, healthy microbiome disruption, and enamel erosion.
The goal of the metabolic redesign of the Strain is not to eliminate LDH activity altogether, but to attenuate the activity below a level that produces harmful acid production, and supplements alternative metabolic pathways of carbon reduction (ex. the 2,3-butanediol operon). This will be accomplished by trialing several different ribosome-binding sequences to tune down translation of the LDH protein.
Additionally, there is a ~24 kb integrative and conjugative foreign element called TnSmu1 that appears to function to induce growth arrest under stress. It also carries machinery for conjugative transfer. This entire locus will be deleted and replaced with the urease operon from S. salivarius. This ~ 7 kb operon consists of ureI (acid-gated urea transporter), ureABC (urease trimer), ureEFGD (accessory proteins for nickel insertion into the apoenzyme), and ureMQO (nickel-specific ABC transporter). Urea is highly abundant in saliva and we do not need to burn ATP for the reaction other than initial protein synthesis.
The reaction is as follows:
We predict that the native Sm CodY regulator will be active in regulating this operon in similar fashion as it is natively regulated in S. salivarius (CodY is a repressor sensing branched-chain amino acid and GTP pools). The conserved regulatory motif is “AATTTTCNGAAAATT”.
Once the metabolic background is formalized (robust growth under aerobic and anaerobic conditions, terminal pH > 5.4), we will further “commensalize” the Strain to promote the colonization of healthy microbes to the tooth surface (Phase two). This will be done by adding the secreted amylose-binding protein (abpA from S. gordonii) the genome. This tiny 20 kDa protein strongly interacts with human salivary α-amylase for the host pioneer colonizer. That enzyme functions to break down starches. Sm GtfB is known to interact with AbpA to promote GtfB enzymatic activity and biofilm scaffolding. The colocation of GtfB-AbpA with sequestered amylase would also release sugars such as maltodextrins as a food source for the local microbes to ingest (including the Strain).
The pioneer colonizer S. gordonii also has two isoforms of the SpaP adhesin called SspA and SspB. This duo can, in theory, outcompete SpaP for immobilized gp340, and introduction of N1182G/V1185P to SspB should abrogate binding to P. gingivalis. A broader substrate-binding repertoire is possible in a Sm SspA/B genetic background, in addition to co-aggregation with commensal microbes such as Actinomyces. The secreted Gtf’s play a substantial role in the organism’s attachment to surfaces, however, and SpaP is known to form biofilm matrix stabilizing functions. Thus, it is unclear at this time if a SspA/B-expressing strain has a clear fitness advantage.
The Strain, if successful, will inhabit an ecological niche with a higher number of commensal microbes and streptococci than typical for a Sm niche. This will require augmentation of the ability to withstand peroxide stress most likely (Phase three). Sm are naturally catalase-negative and do not encode heme biosynthesis pathways for typical catalase enzymes. We will express a low amount of manganese catalase originally found in Lactiplantibacillus plantarum. This 30 kDa enzyme forms a homohexamer to perform the reaction 2 H202 2 H20 + O2. Together with the native alkyl hydroperoxide reductase, thiol peroxidase, and Dpr peroxidase defense systems, this addition will bolster resistance to reactive oxygen species typically used to attack Sm.
Lastly, an aspirational addition to the Strain’s native anti-Sm arsenal would be the secretion of a Spa-targeting VHH nanobody (Phase four). A conserved epitope has been identified and in silico generative VHH models are being designed. Our Strain would express a point-mutated version of the epitope immune to VHH targeting, while native strains would be susceptible. If a high-affinity VHH (or multivalent polypeptide) were to be developed and proven effective in the low pH environment, it would be a substantial Sm-specific addition to preventing Sm colonization specifically. Such a mechanism could also be copied to inhibit other harmful microbes from colonizing, such as P. gingivalis. For example, secreted VHH could be designed to target the catalytic cores of the gingipains Kgp and RgpB.
The final genomic change to the Strain will be the deletion of comX, the master regulator of competence genes in the organism. This will render the strain virtually incapable of uptake of foreign DNA.
Timelines and Signups
This reengineering work will likely take months. Our new lab is operating out of Portland. If at some point in the future you’d like to try the new strain and help us confirm that it can survive indefinitely in the human mouth, let us know! By signing up here: Signups.
I will, as ever, be testing it on myself first.
Thanks.
Aaron