Fig 1.
HMGA2 protects against HU-induced DSBs.
(A) A Western blot shows corresponding HMGA2 expression levels in HeLa cells (parental and three recombinant human HMGA2-expressing cell lines) (top left panel). Representative PFGE analysis of DSB formation in HeLa cells in response to 24 h incubation with HU (left panel). Quantification of HU-induced DNA fragments (>1 Mb and 30–100 kb fractions) (right panel) was done by ImageJ software with each fragment fraction normalized to total DNA loaded (n = 4 independent experiments). Error bars show SD. Unpaired two-tailed t-tests. ns not significant, * p < 0.05, ** p < 0.01, *** p < 0.001. (B) A Western blot shows corresponding HMGA2 expression levels in A549 cells (parental and three recombinant human HMGA2-expressing cell lines) (top left panel). Representative PFGE analysis of DSB formation in A549 cells in response to 24 h incubation with HU (left panel). Quantification of HU-induced DNA fragments (>1 Mb and 30–100 kb fractions) (right panel) was done by ImageJ software with each fragment fraction normalized to total DNA loaded (n = 3 independent experiments). Error bars show SD. Unpaired two-tailed t-tests. * p < 0.05, ** p < 0.01. (C) A Western blot shows corresponding HMGA2 expression levels in H1299 cells (parental and HMGA2 KO) (top left panel). Representative PFGE analysis of DSB formation in H1299 cells in response to 24 h incubation with HU (left panel). Quantification of HU-induced DNA fragments (>1 Mb and 30–100 kb fractions) (right panel) was done by ImageJ software with each fragment fraction normalized to total DNA loaded (n = 3 independent experiments). Error bars show SD. Unpaired two-tailed t-tests. *** p < 0.001.
Fig 2.
HMGA2 expression controls sensitivity to TOP1 poison SN38.
(A) PFGE analysis of DSB formation in H1299 cells (parental and HMGA2 KO) in response to 48 h incubation with SN38 (left panel). Quantification of SN38-induced DNA fragments (>1 Mb and 30–100 kb fractions) was done by ImageJ software (right panel) with each fragment fraction normalized to total DNA loaded (n = 3 independent experiments). Error bars show SD. Unpaired two-tailed t-tests. ** p < 0.01. (B) Representative PFGE analysis of DSB formation in A549 cells (parental and three recombinant human HMGA2-expressing cell lines) in response to 48 h incubation with SN38 (left panel). Quantification of SN38-induced DNA fragments (>1 Mb and 30–100 kb fractions) (right panel) was done by ImageJ software with each fragment fraction normalized to total DNA loaded (n = 3 independent experiments). Error bars show SD. Unpaired two-tailed t-tests. * p < 0.05, ** p < 0.01. (C) Representative PFGE analysis of DSB formation in HeLa cells (parental and three recombinant human HMGA2-expressing cell lines) in response to 48 h incubation with SN38 (left panel). Quantification of SN38-induced DNA fragments (>1 Mb and 30–100 kb fractions) (right panel) was done by ImageJ software with each fragment fraction normalized to total DNA loaded (n = 4 independent experiments). Error bars show SD. Unpaired two-tailed t-tests. * p < 0.05, ** p < 0.01. (D) Representative PFGE analysis of DSB formation in HT1080 C2 cell line in response to 48 h incubation with SN38 (left panel). HMGA2 expression was down-regulated by doxycycline (Dox)-induced shRNA for 96 h in conjunction with SN38 treatment for the last 48 h. Quantification of SN38-induced DNA fragments (>1 Mb and 30–100 kb fractions) was done by ImageJ software (right panel) with each fragment fraction normalized to total DNA loaded (n = 3 independent experiments). Error bars show SD. Unpaired two-tailed t-tests. * p < 0.05.
Fig 3.
Caspase 3/7 activities correspond with the generation of 30–100 kb fragments in response to SN38 treatment.
(A) Caspase 3/7 activity (left panel) in H1299 cells (parental and HMGA2 KO) after SN38 treatment at indicated doses for 24 h and cell survival (CCK8) assay (right panel) with H1299 and HMGA2 KO cells after treatment with SN38 for 48 h (n = 3 independent experiments with 3 technical replicates in each). Mean of untreated triplicates used for normalization. Error bars show SD. Unpaired two-tailed t-tests. ** p < 0.01, **** p < 0.0001. (B) Caspase 3/7 activity in A549 cells (parental and three recombinant human HMGA2-expressing cell lines) after SN38 treatment at indicated doses for 24 h (n = 3 independent experiments with 3 technical replicates in each). Mean of untreated triplicates used for normalization. Error bars show SD. Unpaired two-tailed t-tests. **** p < 0.0001. (C) Caspase 3/7 activity (top panel) in HeLa cells (parental and three recombinant human HMGA2-expressing cell lines) after SN38 treatment at indicated doses for 24 h and cell survival (CCK8) assay (bottom panel) with HeLa and P2 cells (recombinant HMGA2 expressing cell line) after treatment with SN38 for 48 h (n = 3 independent experiments with 3 technical replicates in each). Mean of untreated triplicates used for normalization. Error bars show SD. Unpaired two-tailed t-tests. *** p < 0.001, **** p < 0.0001. (D) Caspase 3/7 activity in HT1080 C2 cells (Dox-/+) after SN38 treatment at indicated doses for 24 h (n = 2 independent experiments with 3 technical replicates in each). Mean of untreated triplicates used for normalization. Error bars show SD. Unpaired two-tailed t-tests. *** p < 0.001, **** p < 0.0001.
Fig 4.
HMGA2 expression levels determine differential chemosensitivity to SN38.
(A) Representative Western blot analysis comparing endogenous (H1299 and HT1080 C1/C2 cells) and His-tagged (HeLa P2 and A549 1.3 cells) exogenous HMGA2 levels across various cell lines (left panel). β-actin was used as a loading control. Quantification (right panel) of HMGA2 expression relative to H1299 (set as 1) using ImageJ software (n = 3 independent experiments). The dotted line indicates an arbitrary functional threshold level for high versus low/moderate HMGA2 expression. Error bars show SD. Unpaired two-tailed t-tests. * p < 0.05, *** p < 0.001, **** p < 0.0001. (B) Western blot of HMGA2 expression (left panel) after transfection with indicated amounts of HMGA2 expression vector in H1299 HMGA2 KO cells (mock vector served as control) using the anti-HMGA2 antibody (Ab41878). Representative PFGE analysis of DSB formation in transfected cells after SN38 treatment (2 μM) for 48 h (centre panel). Quantification of SN38-induced DNA fragments (>1 Mb and 30–100 kb fractions) (right panel) was done by ImageJ software with each fragment fraction normalized to total DNA loaded (n = 4 independent experiments). Error bars show SD. Unpaired two-tailed t-tests. * p < 0.05. (C) Cell survival (CCK8) assay in H1299 HMGA2 KO cells transfected with 0.25 μg (low) and 1.5 μg (high) amounts of HMGA2 expression vector (mock vector served as control) followed by SN38 treatment (2 μM) for 48 h. Data plotted as fold change with mock as 100% (n = 4 independent experiments with 3 technical replicates in each). Error bars show SD. Unpaired two-tailed t-tests. ** p < 0.01, *** p < 0.001. (D) Representative higher-order ternary complex formation of HMGA2 and 2,3M HMGA2 carrying mutated AT-hooks 2 and 3 with supercoiled plasmid DNA at indicated HMGA2:DNA molar ratios. HMGA2-DNA complexes were formed during 30 min incubation and subjected to electrophoresis in 0.8% agarose gels. DNA was stained with ethidium bromide (n = 2 independent experiments). Positions of sc (supercoiled) and oc (open circular/nicked) DNA are indicated. (E) Representative in vivo complex of enzyme (ICE) assay (see Methods), comparing the intracellular formation of TOP1cc between H1299 and HMGA2 KO cells after 20 μM SN38 treatment for 30 min (top panels). The amount of genomic DNA (μg) loaded is indicated and DMSO-treated cells were used as controls. Quantification of SN38 induced TOP1cc formation (bottom panel) with intensities normalized to amounts of spotted DNA that were visualized by staining with Ethidium bromide and plotted as fold change relative to parental H1299 cells (n = 3 independent experiments). Error bars show SD. Unpaired two-tailed t-tests. ** p < 0.01. (F) Representative in vivo complex of enzyme (ICE) assay, comparing the intracellular formation of TOP1cc between HeLa and HeLa P2 cells after 20 μM SN38 treatment for 30 min (top panels). DMSO treated cells were used as controls. Quantification of SN38 induced TOP1cc formation (bottom panel) with intensities normalized to the amounts of spotted DNA that were visualized by staining with Ethidium bromide and plotted as fold change relative to parental HeLa cells (n = 3 independent experiments). Error bars show SD. Unpaired two-tailed t-tests. * p < 0.05.
Fig 5.
Synergistic effects of SN38 and HMGA2 on supercoil relaxation by human topoisomerase I.
(A-B) In the presence of 5 nM TOP1 and 5 μM SN38, fast relaxation of (-) scDNA and slow relaxation of (+) scDNA are observed indicated by segmented extension increases linearly. The scatter plot of the raw data is fitted by piecewise linear regression (also see S4 Fig). (C-D) In the presence of 5 nM TOP1 and 500 nM HMGA2, both (-) scDNA and (+) scDNA are relaxed in a jump-pause manner for a similar duration. (E-F) In the presence of 5 nM TOP1, 5 μM SN38 and 500 nM HMGA2, many of the relaxation events are greatly impeded. The mixed features of extension increase (linear and pause-jump) can be observed. (G) Box plot of relaxation time (grey circle) summarized from (A-F). The mean is shown by the red line, grey and dark grey areas represent the distribution of SD and SEM. p-value is calculated by Wilcoxon rank-sum test. A different scale of the same plot is shown in S4 Fig. in order to highlight the different relaxation times due to the sole presence of either SN38 or HMGA2.
Fig 6.
Genome-wide mapping of the SN38-induced double strand break (DSB) fractions in H1299 and HMGA2 KO cells.
The SN38-induced 30–100 kb fragment fraction (DSB fraction) and total genomic DNA were sequenced, and the reads were aligned to the human genome as described in the Materials and Methods section. Regions showing differential enrichment or depletion (false discovery rate (FDR) <1%) in the DSB fraction relative to untreated genomic DNA are illustrated by black and red colour, respectively in plots 1 and 2 (for H1299) and plots 3 and 4 (for HMGA2 KO); while regions with differential enrichment/depletion (FDR < 1%) in parental versus HMGA2 KO cells are shown in plots 5 and 6. Subtelomeric/telomeric regions were similarly enriched in the 30–100 kb fraction of both SN38-treated cell samples. As an example, the last 200 kb of chromosome 1 is enlarged (boxed region), and the normalized coverage values for the DSB fraction (blue lines) and the untreated total genomic DNA (orange lines) are shown for both H1299 (plot 8) and HMGA2 KO (plot 9) cells. Copy number variations (CNV) in untreated HMGA2 KO cells compared to untreated parental H1299 cells are shown on plot 7, with purple indicating copy number gain and green indicating loss. Values for plots 1–7 were calculated using 20 kb bins, while plots 8–9 utilize 500 bp bins. Figure generated using Circos [62].
Fig 7.
Expression of HMGA2 affects human subtelomeric domains upon induced replication stress.
(A) Southern blot analysis of H1299 cells (parental and HMGA2 KO cell line) after PFGE separation using telomere-specific probes. SN38 treatment at three different doses as indicated was done for 48 h, with DMSO used as control. (B) Western blot showing exogenous expression of HMGA2 (top panel) after transfection with 1.5 μg HMGA2 expression vector in H1299 HMGA2 KO cells (mock vector served as control). Southern blot analysis of HMGA2 KO transfected cells after PFGE separation using telomere-specific probes (bottom panel). 2 μM SN38 treatment was done for 48 h, with DMSO used as control (n = 3 independent transfection experiments followed by SN38 treatment). (C) Representative Southern blot analysis of HeLa cells (parental and three recombinant HMGA2-expressing cell lines) after PFGE separation using telomere-specific probe. 2 μM SN38 treatment was for 48 h, with DMSO as control (n = 3 independent experiments). (D) Western blot showing HMGA2 expression for indicated cell lines with β-actin used as internal control (top panel). Southern blot analysis of human embryonic stem (hES) cells after PFGE separation using telomere-specific probes (bottom panel). 0.1 μM SN38 treatment was for 24 h, with DMSO as control (n = 3 independent experiments). (E) Representative Southern blot analysis of HeLa cells (parental and three recombinant HMGA2-expressing cell lines) after PFGE separation using telomere-specific probe. 100 mM HU treatment was done for 24 h (n = 3 independent experiments). (F) PFGE analysis of DSB formation in HeLa cells (DNA stained with EtBr) in response to 24 h incubation with indicated doses of TMPyP4 (top panel) and the corresponding Southern blot analysis using telomere-specific probes (bottom panel). 100 mM HU treatment for 24 h was used as control for comparing telomeric DNA fragments (>1 Mb and 30–100 kb fractions).
Fig 8.
Low-to-moderate HMGA2 expression increases irinotecan chemosensitivity of PDX models.
(A) Western blot using anti-HMGA2 antibody Ab41878 exhibiting high HMGA2 detection sensitivity (top panel). HMGA2 expression from representative propagated tumors (indicated by the corresponding mouse number) from each PDX model before inoculation into the treatment groups (two independently propagated samples are shown for model no 1030). We compared their HMGA2 expression levels with two cancer cell lines as standards for SN38 sensitivity: H1299 cells, which exhibit a degree of SN38 resistance due to high HMGA2 levels, and HT1080 C1 cells, which exhibit SN38 sensitivity due to low/moderate HMGA2 levels. The complete set of propagated tumors for each PDX model chosen for treatment is shown in S6 Fig and indicated as mouse no. Quantification (bottom panel) of HMGA2 expression relative to H1299 (set as 1) using ImageJ software (n = 3 independent experiments). Error bars show SD. Unpaired two-tailed t-tests. * p < 0.05, ** p < 0.01, **** p < 0.0001. (B) Tumor growth kinetics (Day 0 to Day 39) of five PDX models (n = 6 animals/model) in Irinotecan treatment group. Dotted lines indicate mean starting volume of all PDX models. Two-way ANOVA followed by Tukey’s multiple comparisons. Error bars show SEM. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. (C) Tumor growth kinetics (Day 0 to Day 39) of five PDX models (n = 6 animals/model) in vehicle group. Two-way ANOVA followed by Tukey’s multiple comparisons. Error bars show SEM. ns not significant. (D) Tumor growth inhibition (TGI) of each PDX model at day 39 (Plotted as percentage). TGI is calculated taking the physiological growth of the tumor (vehicle group) into account. Dotted lines indicate 100% TGI level. Two-way ANOVA followed by Tukey’s multiple comparisons. Error bars show SEM. * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. (E) Graphical representation of the correlation between HMGA2 expression and tumor outcomes due to irinotecan treatment in colorectal PDX models used in the study.
Fig 9.
Models for the regulation of subtelomere stability by HMGA2 as a result of replication stress induced by SN38 or HU.
Replication fork stalling at human subtelomeres as a result of TOP1 poisoning or DNA synthesis inhibition. (A) HMGA2 levels determine SN38 treatment responses, with low/moderate HMGA2 levels bound to (+) sc DNA potentiating the accumulation of TOP1-DNA cleavage complexes (TOP1cc), hence leading to more frequent replication run-off events and fork collapse that will ultimately separate subtelomeres from the rest of chromosomal DNA. (B) The accumulation of TOP1cc as a result of SN38 treatment is minimized at high HMGA2 levels due to a combination of constraining of (+) sc DNA and TOP1 exclusion from binding to its scDNA substrate, thereby preventing DSBs at the subtelomeres. (C) Such a varied response was not observed with DNA synthesis inhibition by HU since detectable HMGA2 expression always reduced fork collapse and promoted subtelomere stability and cell survival. Described in detail in the Discussion.