Altered Complement and Tubular Mitochondrial Proteins Are Linked to Graft Loss in DSA+ Antibody-Mediated Rejection

Antibody-mediated rejection (AMR) is the main driver of late kidney allograft loss. Anti-HLA donor-specific antibodies (DSA) are strongly associated with AMR. However, 30-60% of DSA+ patients do not develop AMR, raising the possibility of local tissue adaptations to DSA.

We performed unbiased proteomic analysis on glomeruli and tubulointerstitium isolated using laser-capture microdissection from the kidney biopsies of 116 DSA+ kidney transplant recipients with Banff-classified active AMR, chronic active AMR, mixed AMR or no rejection. The 'Significant Latent factor Interaction Discovery and Exploration' (SLIDE) machine-learning algorithm was used to identify latent factors linked to graft loss amongst individuals with AMR.

We quantified 1917 glomerular and 2758 tubulointerstitial proteins by mass spectrometry, with 127 glomerular and 389 tubulointerstitial proteins significantly differentially expressed between all 4 patient groups (one-way ANOVA p<0.05, Tukey FDR<0.05). In glomeruli, chronic active AMR had the most differentially expressed proteins, consisting primarily of complement and basement membrane components. Tubulointerstitial terminal complement proteins were also significantly increased in chronic active AMR but did not correlate with C4d deposition. Mixed AMR contained the most differentially expressed proteins in the tubulointerstitium, which were linked to IFNγ-associated immune activation, stress responses and mitochondrial metabolism. DSA+ biopsies without rejection were characterized by preservation of podocyte and tubular proteins in glomeruli and tubulointerstitium, respectively, but did not show conventional markers of accommodation. SLIDE identified 3 latent factors from the glomerular proteome and 4 from the tubulointerstitial proteome that significantly distinguished AMR patients with and without graft loss, correlated significantly with allograft injury, and improved prediction of graft loss beyond standard clinical variables in a multivariable risk model. Top proteins within the glomerular SLIDE model included complement components while top proteins for the tubulointerstitial SLIDE model included tubular and mitochondrial proteins.

AMR and graft loss following kidney transplantation were defined by compartment-specific disruptions in complement and mitochondrial proteins.