Mechanisms underpinning signals from genome-wide association studies remain poorly understood, particularly for noncoding variation and for complex diseases such as type 2 diabetes mellitus (T2D) where pathogenic mechanisms in multiple different tissues may be disease driving. One approach is to study relevant endophenotypes, a strategy we applied to the UBE2E2 locus where noncoding single nucleotide variants (SNVs) are associated with both T2D and visceral adiposity (a pathologic endophenotype). We integrated CRISPR targeting of SNV-containing regions and unbiased CRISPR interference (CRISPRi) screening to establish candidate cis-regulatory regions, complemented by genetic loss of function in murine diet-induced obesity or ex vivo adipogenesis assays. Nomination of a single causal gene was complicated, however, because targeting of multiple genes near UBE2E2 attenuated adipogenesis in vitro; CRISPR excision of SNV-containing noncoding regions and a CRISPRi regulatory screen across the locus suggested concomitant regulation of UBE2E2, the more distant UBE2E1, and other neighborhood genes; and compound heterozygous loss of function of both Ube2e2 and Ube2e1 better replicated pathological adiposity and metabolic phenotypes compared with homozygous loss of either gene in isolation. This study advances a model whereby regulatory effects of noncoding variation not only extend beyond the nearest gene but may also drive complex diseases through polygenic regulatory effects.