The labelling of Fig. 2A (and, as a consequence, of Fig. 1C and of Fig. 7, top frames) is misleading because it seems to imply that all chromatin loops correspond to GC-poor isochores and all chromatin boundaries correspond to GC-rich isochores. This is not so for the following reasons (see Table S1):
1. ~35% of the genome is made of GC-poor, gene-poor sequences that correspond to LADs (lamina associating domains; see ref. 66) and comprise all the L1 isochores (19% of the genome) plus L2 isochores corresponding to ~ 16% of the genome;
2. ~65% of the genome is made of GC-rich, gene-rich sequences from L2 (corresponding to ~20% of the genome), H1, H2 and H3 isochores that correspond to TADs (topologically associating domains);
3. According to a new estimate of gene density (Jabbari K, Nürnberg P . 2015. A genomic view on epilepsy and autism candidate genes. Genomics 108:31-36), H1 isochores belong to the gene-rich genome core and not to the gene-poor genome desert as estimated in the basis of the gene density results previously available (Consortium IHGS. Initial sequencing and analysis of the human genome, Nature. 2001; 409: 860-921; see Bernardi G. Chromosome Architecture and Genome Organization, PLoS One. 2015 Nov 30;10:e0143739.).

The same corrections apply to Fig. 2 of Bernardi G. (2015) Genome Organization and Chromosome Architecture. Cold Spring Harb Quant Biology 80:83-91.