Fig 2: Schematic depiction p110 CUX1 effect on EGFR–MEK-ERK signaling. p110 CUX1 synergizes with EGFR-KRASmut signaling by reinforcing MEK-ERK signaling via upregulation of the EGF ligand sheddase ADAM17 and the serine/threonine kinase MOS.

Fig 3: Generation and validation of KCCux1p110 mice. (A) Breeding scheme for the generation of KrasLSL-G12D/+; Ptf1aCre/+; Cux1LSL-p110/+ mice. KrasLSL-G12D/+; Cux1LSL-p110/LSL-p110 mice were mated with Ptf1aCre/+ mice to generate KrasLSL-G12D/+; Ptf1aCre/+ and KrasLSL-G12D/+; Ptf1aCre/+; Cux1LSL-p110/+ animals. (B) Survival of tumor-bearing (n = 4) and tumor-free (n = 5) KCCux1p110 mice. Kaplan–Meier curves showed a median survival for tumor-bearing mice of 21 days and for tumor-free mice of 83 days with significant difference by log-rank test, p = 0.0027. (C) Representative H&E stainings of 4-week-old (left) KC (i) and tumor-bearing KCCux1p110 (ii) and 3-month-old (right) KC (iii) and tumor-free mPanIN-bearing KCCux1p110 (iv) mice. Scale bar, 100µm. (D) Analysis of PanIN burden in 3-month-old KC (grey; n = 9) and KCCux1p110 (black; n = 5) mice. *** p ≤ 0.0001 by t-test. Data shown as mean ± SD. (E) Percentage of tumor-bearing KC and KCCux1p110 mice. In the KC cohort, 0 of 10 mice, and in the KCCux1p110 cohort (black bar), 4 of 9 mice showed a PDAC. * p ≤ 0.02 by Fisher’s exact test.

Fig 4: CUX1 has different isoforms which interact with SOX2. (A) Schematic representation of the CUX1 protein, N-terminally tagged with a myc peptide, and C-terminally with a HA peptide. The tagged CUX1 was co-transfected with a FLAG-tagged SOX2 construct, and total input was analyzed with antibodies against the myc-tag (α-myc), the HA-tag (α-HA), or a CUX1 raised against a peptide mapping at the C-terminus [α-CUX1 (aa521-621)] asteriks indicates CUX1 isoforms. (B) HEK cells were transfected with myc-CUX1-HA alone or together with FLAG-SOX2, and protein extracts were subsequently precipitated with one of two CUX1 antibodies (CUX1 peptide C-terminus or CUX1 aa521-621). Western blots were subsequently labeled with antibodies against the FLAG peptide to identify co-precipitating SOX2 protein, or with antibodies against the myc or HA tag to detect the CUX1 protein isoforms. Red arrows indicate the precipitated SOX2, black arrows indicate the precipitated CUX1 isoforms. (C) Protein extracts of tracheas isolated from birA mice (BirA) or bioSOX2/birA mice were incubated with streptavidin beads to isolate biotinylated SOX2 proteins. Total input (I), bound (B), and unbound (U) fractions were immunoblotted with antibodies against CUX1 (top, white asterisks indicate specific bands in the bound fraction) or against SOX2 (bottom, black asterisks). (D) Identical streptavidin precipitation as in panel (C), using brain protein extracts. (E) Bacterial produced glutathione S-transferase (GST) or GST-fused to full length SOX2 (GST-SOX2) were incubated with nuclear protein extracts of HEK cells. CUX1 specifically precipitated with the GST-SOX2 fusion protein. U = eluate, B = precipitated fraction.

Fig 5: Immunostainings of later stage ZIP13K5 iPSC derived organoids highlight the pervasive presence of cortical units in Triple-i organoids.Immunostaining for cortical markers PAX6 and FOXG1, iSVZ/oSVZ markers TBR2 and HOPX along with SOX2, oRG markers LIFR and PTPRZ1 along with SOX2, upper layer neuronal marker CUX1 along with deep layer neuronal marker TBR1, and upper layer neuronal marker SATB2 along with deep layer neuronal marker CTIP2 in day 80 Dual SMAD-i organoids (N=3) and day 80 Triple-i organoids (N=2) derived from the ZIP13K5 iPSC line. Scale bar: 200 µm. Zoom-ins of the bottom Triple-i organoid are shown in the bottom panel. Note the lack of FOXG1 expression across all three Dual SMAD-i organoids and the presence of cortical units covering both Triple-i organoids, with co-localization of SOX2 and oRG markers, as well as interspersed upper and deep layer neuronal gene expression.