Cyclic undecapeptide cyclo-[MeBmt(1)-Abu(2)-MeGly(3)-MeLeu(4)-Val(5)-MeLeu(6)-Ala(7)-D-Ala(8)-MeLeu(9)-MeLeu(10)-MeVal(11)], the immunosuppressive and antifungal antibiotic cyclosporin A (CsA), was reported to interfere with the MDR1 P-glycoprotein (Pgp), a transmembranous adenosine 5'-triphosphate binding cassette (ABC) transporter with phospholipid flippase or "hydrophobic vacuum cleaner" properties that mediate multidrug resistance (MDR) of cancer cells. By use of photoaffinity-labeled cyclosporins and membranes from Pgp-expressing cells, it was recently shown that in vitro, Pgp molecules could bind a large cyclosporin domain involving residues 4-9 as well as the side chain of residue 1. Tumor cell MDR can also be reversed by a product more distantly related to cyclosporin with the structure [Thr(2), Leu(5), D-Hiv(8), Leu(10)]-CsA (SDZ 214-103). In a standardized assay that measures Pgp function in vivo (on intact live cells) by the Pgp-mediated efflux of the calcein-AM Pgp substrate and uses human lymphoblastoid MDR-CEM (VBL(100)) cells as highly resistant Pgp-expressing cells, SDZ 214-103 was found to be one of the most active Pgp inhibitors among naturally occurring cyclosporins, with an IC(50) of 1.6 microM in an assay where CsA gives an IC(50) of 3.4 microM. Using the in vivo assay, 60, mostly natural, cyclosporin analogues were analyzed to establish structure-activity relationships (SAR). Our SAR are compatible with the in vitro-defined Pgp binding domain model and further disclose that in vivo Pgp inhibition is favored by larger hydrophobic side chains on cyclosporin residues 1, 4, 6, and 8 and a smaller one on residue 7, although with no effect on the residue 5 side chain; moreover, larger hydrophobic side chains on other residues 2, 3, 10, and 11 (outside the in vitro-defined Pgp binding domain) also favor the eventual inhibition of Pgp function. The N-desmethylation of any of the seven N-methylated amides, as naturally occurring in numerous cyclosporins, regularly leads to a decreased Pgp inhibitory activity (Pgp-InhA), up to its abrogation if it occurs at residues 4 and 9. Nevertheless, despite unfavorable use of [Thr(2)] and [Leu(10)] residues, all [D-Hiv(8)] analogues whose lead is SDZ 214-103 show a large Pgp-InhA. The SAR for Pgp inhibition by cyclosporins are thus very complex. Because CsA and SDZ 214-103 show largely different conformations when free in solution, but remarkably similar ones when bound to the cytosolic cyclophilins, SAR for Pgp inhibition must similarly include requirements for occurrence of suitable conformers for insertion in the cell membrane, sufficient conformational plasticity for gaining access to Pgp binding sites, and an adequate conformer structure there to achieve such binding with a high enough affinity and possibly escape from sequestration on cyclophilins.