A nine-residue cyclic peptide called BLMP6 selectively homes to triple-negative breast cancer metastases in mice. Tag it with a PET-imaging radioisotope and it lights up the lungs where the cancer has spread. Tag the same peptide with a cytotoxic warhead and it kills those same metastatic cells in place. And along the way, the team that just published in Molecular Therapy: Oncology ↗ identified what BLMP6 is actually grabbing onto: fibulin-4, an extracellular-matrix protein elevated in invasive and metastatic human breast tumors but barely noted as a cancer marker before.

That is three findings stacked into one preclinical paper, from Mikhail Kolonin's group at UTHealth Houston's Institute of Molecular Medicine.

The peptide

BLMP6's sequence is CRAGVGRGC. Nine amino acids, with the two cysteines linked into a loop so the molecule holds a fixed shape. The Kolonin lab originally pulled it out of a combinatorial peptide library by screening for sequences that home to lung metastases in mice, and synthesized it with D-amino acids so circulating proteases cannot chew it up. In tissue sections it stains the subpopulation of cancer cells that have undergone the epithelial-to-mesenchymal transition (EMT), the program that lets a primary tumor cell detach, migrate, and seed new sites. It does not stain the lung vasculature, the normal stroma, or non-metastatic tissue.

That selectivity is the load-bearing fact. A targeted-therapy peptide is only as useful as the populations it ignores.

The imaging arm

To turn BLMP6 into an imaging probe, the team attached a DOTA chelator and radiolabeled it with gallium-68, the same short-lived PET isotope used in approved peptide diagnostics like the Ga-68-DOTATATE scan for neuroendocrine tumors. In nude mice grafted with human MDA-MB-231 cells (the standard triple-negative breast cancer line) and given lung metastases by tail-vein injection, Ga-68-BLMP6 accumulated in metastatic lungs above background and stayed at background in tumor-free animals. Bladder and kidney signal was high (the radiolabel clears through urine) and nothing else lit up, which is roughly the cleanest result a metastasis-specific PET tracer can produce. They saw similar results in an orthotopic 4T1 mouse breast cancer model with spontaneous lung metastases.

For a normie reader: PET scans need a molecule that travels to cancer cells and sits there long enough for the scanner to spot it. Most clinically used cancer PET tracers detect glucose uptake or generic protein expression, not metastatic cells specifically. BLMP6 was designed to do exactly that.

The therapy arm

For the cytotoxic version, the group built a peptide-drug conjugate (a PDC, the small-molecule-targeted cousin of antibody-drug conjugates) by coupling monomethyl auristatin E (MMAE) to BLMP6 through a PEG4 linker, a valine-citrulline cleavable bond, and a self-immolating p-aminobenzyl unit. That is the exact linker chemistry used in approved MMAE antibody-drug conjugates like Adcetris (brentuximab vedotin) and Padcev (enfortumab vedotin). The peptide does the targeting; cathepsins inside the cancer cell cleave the linker; free MMAE poisons the microtubules.

The numbers: MMAE-BLMP6 killed human MDA-MB-231 cells with a half-maximal inhibitory concentration of 120 nanomolar (very potent for a PDC), mouse 4T1 cells at 500 nanomolar, and lost most of its activity in fibulin-4-knockout cells. Scrambling the BLMP6 sequence eliminated the killing. Substituting BLMP5, a different metastasis-homing peptide from the same library, raised the IC50 to 400 nanomolar against MDA-MB-231 cells. In two mouse lung-metastasis models (B16F10 melanoma and MDA-MB-231 breast), subcutaneous MMAE-BLMP6 suppressed metastatic growth on bioluminescence imaging and significantly extended survival, p=0.003 in the B16F10 cohort and p=0.02 in an orthotopic 4T1 cohort against a scrambled-MMAE-BLMP6 control. Crucially, free MMAE at its maximum tolerable dose did not extend survival. The peptide is doing the delivery work.

The target

The other half of the paper is the molecular biology. The Kolonin group ran the reverse BLMP6 sequence through structural homology searches and AlphaFold3 modeling and found a match in residues 725 to 730 of latent transforming growth factor beta binding protein 4 (LTBP4), inside its EGF-like calcium-binding domain 6. That LTBP4 region is already known to bind the fibulin family of ECM proteins, which led them straight to fibulin-4 (also called EFEMP2) as the receptor BLMP6 is mimicking. AlphaFold3 and ClusPro docked BLMP6 and the LTBP4 fragment onto the same pocket of fibulin-4 with RMSD around 1.3 angstroms across species.

The biology checks: knocking out fibulin-4 in cancer cells abolished BLMP6 homing in mice. Fibulin-4 stains heavily in invasive and metastatic human breast cancer sections, and AZDye555-BLMP6 binding on those same sections correlated with fibulin-4 expression. So fibulin-4 is plausibly a metastasis-specific surface mark on human tumors, accessible to a small peptide ligand, and BLMP6 binds the same pocket on it that LTBP4 does.

One small bonus

In treated tumors, MMAE-BLMP6 produced a roughly five-fold increase in CD8+ T-cell infiltration relative to controls. Auristatin warheads have been reported to cause immunogenic cell death in other settings, and the effect on tumor-resident macrophages here was much smaller, which fits. A peptide-drug conjugate that also flips a cold tumor warmer is more interesting than one that just kills.

What this changes

This is mouse work, with cell-line tumors, in one lab. The paper is not a clinical readout and BLMP6 has not been in a human. The translational bridge from a 4T1 orthotopic survival curve to an approvable triple-negative breast cancer therapy is years of toxicology, IND-enabling chemistry, and dose-finding. Many "homing peptide" stories have ended at that bridge.

What is uncommon here is the package. A single short cyclic peptide that has now been characterized as a metastasis-selective ligand, validated as a PET tracer with a clinically used isotope, validated as an MMAE delivery vehicle with the same linker chemistry as approved antibody-drug conjugates, and walked back to a named molecular target with structural modeling and a genetic knockout. The peptide-as-theranostic-scaffold concept has been talked about for fifteen years; this is one of the cleaner end-to-end demonstrations of it, and it pulls fibulin-4 into the anticancer ↗ target conversation in a way the field has not really had it before.