Article Text

Original article
Dietary haem stimulates epithelial cell turnover by downregulating feedback inhibitors of proliferation in murine colon
  1. Noortje IJssennagger1,2,
  2. Anneke Rijnierse1,2,
  3. Nicole de Wit1,3,
  4. Denise Jonker-Termont1,4,
  5. Jan Dekker1,
  6. Michael Müller1,2,3,
  7. Roelof van der Meer1,2
  1. 1Top Institute Food and Nutrition, Wageningen, The Netherlands
  2. 2Nutrition, Metabolism and Genomics Group, Division of Human Nutrition, Wageningen University, Wageningen, The Netherlands
  3. 3Netherlands Nutrigenomics Centre, Wageningen, The Netherlands
  4. 4NIZO Food Research, Health Department, Ede, The Netherlands
  1. Correspondence to Noortje IJssennagger, Nutrition, Metabolism and Genomics Group, Division of Human Nutrition, Wageningen University, Bomenweg 2, NL-6703 HD Wageningen, The Netherlands; noortje.ijssennagger{at}


Objective Colon cancer is a leading cause of cancer deaths in Western countries and is associated with diets high in red meat. Haem, the iron-porphyrin pigment of red meat, induces cytotoxicity of gut contents and damages the colon surface epithelium. Compensatory hyperproliferation leads to epithelial hyperplasia which increases the risk of colon cancer. The aim of this study was to identify molecules signalling from the surface epithelium to the crypt to initiate hyperproliferation upon stress induced by haem.

Methods C57Bl6/J mice (n=9/group) received a ‘westernised’ control diet (40 en% fat) with or without 0.5 μmol/g haem for 14 days. Colon mucosa was used to quantify cell proliferation and for microarray transcriptome analysis. Gene expression profiles of surface and crypt cells were compared using laser capture microdissection. Protein levels of potential signalling molecules were quantified.

Results Haem-fed mice showed epithelial hyperproliferation and decreased apoptosis, resulting in hyperplasia. Microarray analysis of the colon mucosa showed 3710 differentially expressed genes (false discovery rate (q) <0.01), with many involved in the cell cycle. Expression levels of haem- and stress-related genes showed that haem affected surface cells but did not directly affect crypt cells. Injured surface cells should therefore signal to crypt cells to induce compensatory hyperproliferation. Haem downregulated the inhibitors of proliferation, Wnt inhibitory factor 1, Indian Hedgehog and bone morphogenetic protein 2. Interleukin-15 was also downregulated. Haem upregulated amphiregulin, epiregulin and cyclo-oxygenase-2 mRNA in surface cells. Their protein/metabolite levels were, however, not increased as haem induced surface-specific inhibition of translation by increasing 4E-BP1.

Conclusions Haem induces colonic hyperproliferation and hyperplasia by inhibiting the surface to crypt signalling of feedback inhibitors of proliferation.

  • Colon cancer
  • cytotoxic stress
  • translation inhibition
  • Wnt inhibitory factor 1
  • dietary—colon cancer
  • cell signalling
  • nutrition
  • epithelial proliferation
  • intestinal stem cell
View Full Text

Statistics from

Significance of this study

What is already known about this subject?

  • The risk of colon cancer is associated with diets high in red meat but not with diets containing white meat.

  • The porphyrin pigment of red meat, haem, increases cytotoxicity of colonic contents and induces compensatory hyperproliferation in rodents. However, the signalling pathways inducing this compensatory hyperproliferation are unknown.

What are the new findings?

  • Using laser capture microdissection, differential gene expression profiles were obtained for epithelial surface and crypt cells in murine colon.

  • Signalling from the injured surface epithelium to the proliferative crypt occurs via downregulation of feedback inhibitors of proliferation (eg, Wif1, Ihh, Bmp2 and IL-15).

  • Upregulated mRNA levels of growth factors in the haem-stressed surface epithelium were not translated to protein, as protein translation was inhibited by increased 4E-BP1 levels.

How might it impact on clinical practice in the foreseeable future?

  • This study provides further insight into the molecular mechanism of the relationship between red meat consumption and the risk of colon cancer.

  • If validated in humans, Wif1, Ihh, Bmp2 and IL-15 may be used as early biomarkers of diet-modulated colon cancer risk.


Colon cancer is a leading cause of cancer deaths in Western countries.1 The risk of colon cancer is strongly associated with nutrition, especially with diets high in red meat.2 3 Consumption of white meat (poultry and fish), however, is not associated with an increased colon cancer risk.4 5 Kinzler and Vogelstein6 argued that dietary factors increasing the risk of colon cancer are probably not mutagens but rather luminal irritants that damage colonic epithelial cells. Red meat contains the iron-porphyrin pigment haem. Several epidemiological studies show an association between haem intake and the risk of colon cancer.7 8 Intact haem—but not its constituents porphyrin and iron—induces the cytotoxicity of gut contents, indicating an increased exposure of the colonic mucosa to luminal irritants and inducing compensatory epithelial hyperproliferation.9 Hyperproliferation increases the risk of endogenous mutations in tumour suppressor genes and oncogenes and thus the cancer risk.6

Intestinal epithelial cell turnover is determined by the balance between cell proliferation and cell death. Colonic cell proliferation occurs from stem cells near the crypt bottom and is controlled by the Wnt signalling pathway.10 The newly formed cells migrate up to the surface epithelium while they undergo mitosis and differentiation. Cells reach the surface epithelium and disappear after about 4 days by means of exfoliation and/or cell death (necrosis or apoptosis). It was recently shown that dietary haem increases cell death by damaging the surface cells, resulting in luminal necrosis and inhibition of active exfoliation.11 This increased cell death at the surface was compensated by inhibition of apoptosis and by hyperproliferation of crypt cells causing hyperplasia. How these compensatory mechanisms are initiated in the crypt upon surface damage is not known. Most logically, signals from surface to crypt should increase proliferation. The aim of this study was therefore to identify haem-modulated molecules that signal from the surface epithelium to the crypt to increase cell proliferation and/or inhibit apoptosis.

Materials and methods

Animals and diets

Eight-week-old male C57BL6/J mice (Harlan, Horst, The Netherlands) were housed individually in a room with controlled temperature (20–24°C), relative humidity (55±15%) and a 12 h light-dark cycle. Mice were fed diets and demineralised water ad libitum. To study the effects of haem on colonic epithelium, mice (n=9/group) received either a ‘westernised’ control diet (40 en% fat (mainly palm oil), low calcium (30 μmol/g)) or this diet supplemented with 0.5 μmol/g haem (Sigma-Aldrich, St Louis, Missouri, USA) for 14 days, as previously described.12 Faeces were quantitatively collected during days 11–14, frozen at −20°C and subsequently freeze-dried. After 14 days of intervention the colon was excised, mesenteric fat was removed and the colon was opened longitudinally, washed in phosphate buffered saline and cut into three parts. The middle 1.5 cm colon tissue was snap-frozen in liquid nitrogen for cryostat sections for laser capture microdissection (LCM) or formalin-fixed and paraffin embedded for histology. The remaining proximal and distal parts were scraped. Scrapings were pooled per mouse, snap-frozen in liquid nitrogen and stored at −80°C until further analysis.

Cytotoxicity measurement

Faecal water was prepared by reconstituting freeze-dried faeces with double-distilled water to obtain a physiological osmolarity of 300 mOsm/l, as described previously.9 The cytotoxicity of faecal water was quantified by potassium release from human erythrocytes after incubation with faecal water, as previously described,9 and validated with human colon carcinoma-derived Caco-2 cells.13 Cytotoxicity was expressed as percentage of maximal lysis.


Immunohistochemical and immunofluorescent stainings were performed on paraffin embedded colon sections. Details of antibodies and the procedure are shown in the online supplement. To quantify Ki67-positive colonocytes, 15 well-oriented crypts (longitudinal section, displaying the total crypt) per animal (n=9/group) were counted. These crypts were equally distributed over the middle 1.5 cm of the stained colon. Cells were scored Ki67-positive when their nucleus was distinctly brown. The number of Ki67-positive cells per crypt, the total number of cells per crypt and the labelling index (percentage of Ki67-positive cells per crypt) were determined.

RNA isolation and microarrays

RNA was isolated from colon scrapings and individually hybridised (n=7 control; n=9 haem) to Affymetrix Mouse genome 430 2.0 arrays (Affymetrix, Santa Clara, California, USA) (see online supplement for details). Genes that satisfied the criterion of false discovery rate <1% (q<0.01) were considered significantly differentially expressed between control and haem diets. Genes with signal intensities <20 in both treatments were considered absent and were excluded from further analysis. Array data were submitted to the Gene Expression Omnibus, accession number GSE27849.

Laser capture microdissection

Colon surface and crypt cells of four control mice and three haem-fed mice were separately isolated using LCM. RNA was isolated and analysed with microarray to obtain surface and crypt-specific gene expression profiles (see online supplement).

Protein/metabolite levels

Protein levels of amphiregulin (Areg), epiregulin (Ereg) (both n=8/group) and interleukin-15 (IL-15) (n=9/group) were determined by ELISA in colon homogenates. 4E-BP1 levels of homogenate (pools of n=8/group) were determined by western blot analysis. Prostaglandin (PG) levels were measured by enzyme immunoassay in tissue culture supernatants (n=8/group) (see online supplement for details of method).

Statistical analysis

Data are presented as mean±SEM. Differences between mean values were tested for statistical significance by a two-tailed Student t test. p Values <0.05 were considered significant. Statistical considerations for microarray analysis are described above and in the online supplement.


Physiological changes induced by haem

After 2 weeks of diet intervention, haem-fed mice had a lower body weight than controls (25.4±0.2 g vs 29.1±0.8 g, p<0.05). Faecal water of haem-fed mice was significantly more cytotoxic than control faecal water (table 1). Histological examination of colon tissue showed that the surface epithelium of control mice was intact. However, haem-fed animals had a ruffled surface epithelium indicating disruption of the surface epithelium architecture (figure 1). There were no signs of inflammation as infiltration of leucocytes was absent. This is in accordance with gene expression data, as Ingenuity pathway analysis showed no changes in inflammation-related pathways (see figure 2 in online supplement).

Table 1

Effect of dietary haem on faecal parameters and mucosal proliferation

Figure 1

Stainings of mouse colonic mucosa after 14 days of control versus haem diet. Following histological H&E staining, immunohistochemistry was performed using Ki67, haem oxygenase 1 (Hmox1) and cleaved caspase-3 (Casp3)-specific antibodies. Arrows in the latter staining indicate Casp3-positive cells.

To measure colonocyte proliferation, colon tissue was stained with an antibody against Ki67. Stainings revealed a twofold increase in the number of Ki67-positive cells by haem and a 1.3-fold increase in the total number of cells per crypt (table 1). Moreover, the labelling index increased significantly by 40%, indicating haem-induced increased cell proliferation. This increased cell proliferation expanded the proliferative compartment of the crypts and increased crypt depth (figure 1). Besides increased proliferation, decreased apoptosis also contributed to increased crypt depth and increased cell numbers. Immunostaining for active caspase-3 (figure 1) showed that apoptotic cells were present in control mice, although the numbers were low as apoptosis is a rapid process. However, there were no caspase-3-positive cells in the surface epithelium of haem-fed mice. This is in line with our earlier study showing that haem abrogates apoptosis in the colon mucosa.11 All these physiological and histological changes are in line with the effects observed in haem-fed rats11 and indicate that, also in mice, dietary haem injures the surface epithelium which is compensated by crypt cell hyperproliferation and by inhibition of apoptosis, resulting in hyperplasia.

Haem modulated transcriptomics

Transcriptome analysis was performed on colonic mucosal scrapings of individual mice to determine haem-induced differentially expressed genes. Applying the following selection criteria (false discovery rate (q) <0.01 and signal intensity >20 in at least one of the arrays), we found 3710 differentially expressed genes. The 50 most upregulated and downregulated genes are shown in tables 2 and 3, respectively, in the online supplement. Categorising Gene Ontology Biological Process annotations of the differentially expressed genes showed that predominantly processes related to cell proliferation and differentiation were changed (see figure 2A in online supplement). Furthermore, stress- and immune response-related processes were influenced. Metabolism-related genes were mainly involved in lipid metabolism. The results were confirmed by Ingenuity canonical pathway analysis (supplementary figure 2B), showing that pathways related to the cell cycle (control of chromosomal replication and G2/M DNA damage checkpoint regulation) and stress (NRF2-mediated oxidative stress response) were especially affected in the colonic mucosa. Figure 2C in the online supplement shows that genes involved in the progression of the cell cycle were almost entirely upregulated by haem (eg, cyclins, cyclin-dependent kinases, as well as ORC- and MCM-related genes involved in DNA replication), whereas genes controlling cell cycle arrest (Smad3/4) and DNA damage checkpoint (Prkdc and Atm) were downregulated. Together, the microarray data clearly indicate that proliferation was stimulated in the colonic mucosa by haem, which is in accordance with the physiological findings described above.

Localisation of haem modulated cell turnover genes

To determine the localisation of haem-modulated genes in the colonic mucosa, LCM was applied to separately isolate surface cells and transit-amplifying (TA) cells of the lower part of the crypts (see figure 1 in online supplement). Gene expression of these two compartments was analysed by whole genome microarrays to obtain surface and crypt-specific expression profiles. Using LCM, almost no lamina propria cells were isolated. This is in contrast to total scrapings where the complete epithelial lining is present as well as lamina propria cells.

As microarray analysis of mucosal scrapings showed that the cell cycle is changed by haem, we first focused on the localisation of cell turnover and apoptosis-related genes. In both total scrapings and LCM samples, gene expression levels of Ki67 were analysed to validate haem-induced hyperproliferation. Ki67 gene expression was 1.7-fold significantly upregulated in scrapings by haem (table 2). Using LCM, Ki67 was shown to be predominantly expressed in crypt cells where proliferation is expected (table 2). Haem doubled Ki67 expression only in the crypt compartment. This twofold upregulation of Ki67 mRNA is in line with its protein levels (figure 1 and table 1).

Table 2

Identification and localisation of haem-modulated cell turnover genes

The progression of cells through the cell cycle is controlled by cyclins. In scrapings, cyclins E1, A2 and B2 (reflecting the G1/S, S and M phases of the cell cycle) were all increased approximately twofold by haem. LCM showed that this is a crypt-specific upregulation (table 2). Cyclins can be activated by growth stimulating factors such as Wnt. However, no differences were observed in the gene expression level of Wnts and their receptors, except for Frizzled 5 (table 2). This Wnt receptor is mainly expressed and downregulated in surface cells and therefore cannot mediate the hyperproliferative effect on crypt cells.

Array data showed that haem also modulated apoptosis-related genes. The apoptosis inhibitor survivin (Birc5) was upregulated 2.6-fold in scrapings. This increase was crypt-specific as survivin was upregulated 3.1-fold at the crypt level only (table 2). Moreover, apoptosis inhibitors Xiap, Bcl2l1 and Ier3 were significantly upregulated by haem, but their upregulation was surface-specific. Thus, inhibition of apoptosis in both surface cells and crypt cells would contribute to the increased number of cells per crypt and therefore to the observed hyperplasia.

Haem modulated stress response genes in surface epithelium

It is not known whether haem can reach the bottom of the crypt and directly modulate TA cells. To determine which cells were exposed to haem we studied the expression of genes involved in haem metabolism and their localisation. The enzyme haem oxygenase 1 (Hmox1), known to be induced by its substrate haem and by various non-haem stressors,14 was upregulated 35-fold by haem in scrapings (figure 2A). Hmox1 expression was specifically induced in surface epithelium (figure 2A), which is in line with its protein expression (figure 1). Apical haem uptake may occur through the putative haem transporter Slc46a1.15 This transporter was downregulated 10-fold by haem, probably to protect colonocytes from entry of large amounts of haem. This downregulation occurred only at the surface (figure 2A). Abcg2 is proposed to transport the excess of haem and haem breakdown products out of enterocytes into the lumen.16 In mucosal scrapings of haem-fed mice there was a 1.5-fold upregulation of Abcg2, and this upregulation occurred most profoundly at the surface (figure 2A).

Figure 2

Differential expression of (A) haem metabolism-related and (B) stress-related genes. Microarray analysis was performed on total scrapings of individual mice (n=7 controls, n=9 haem-fed mice). Relative gene expression changes (fold changes) on haem were calculated by setting RNA levels of control mice to 1 (mean±SEM, *q<0.01). Surface- and crypt-specific microarray gene expression profiles are from pooled tissue samples obtained by laser capture microdissection (n=4 controls, n=3 haem-fed mice). Relative gene expression changes were calculated by setting RNA levels in the surface of controls to 1.

In addition to genes of the haem pathway being specifically changed in the surface epithelium, stress-related genes such as the NF-κB essential modulator Nemo (Ikbkg) also showed a significant upregulation almost exclusively at the surface (figure 2B). Upregulation of immediate early response genes c-Fos and Tis7 and the stress activated transcription factor Creb3l317 was also surface-specific (figure 2B). Although Nrf2 was not itself changed by a haem diet, Nrf2-interacting proteins c-Fos, Creb3l3 and Atf4 (see below) were significantly increased in surface cells, indicating Nrf2-dependent transcription of stress-related genes such as Hmox1. The changes in both haem- and stress-related genes show that dietary haem exerts its primary effect on the surface epithelium and that it has little or no direct effect on crypt cells. This raises the question of how haem-stressed surface cells signal to crypt TA cells to induce compensatory hyperproliferation.

Haem downregulated epithelial expression of feedback inhibitors of proliferation

To identify signalling molecules triggering the proliferative capacity in the crypt, differentially expressed genes were subjected to secretome analysis. Selection criteria and the secretome table are shown in table 4 in the online supplement. We hypothesised that loss of feedback signals—and thus downregulation of gene expression—due to haem-induced stress at the surface is a likely way to induce hyperproliferation of crypt cells. Indeed, we found in our secretome analysis more downregulated than upregulated signalling molecules (62 vs 25). We further selected downregulated genes coding for feedback inhibitors of proliferation or cytokines. In this way, we identified Wnt inhibitory factor 1 (Wif1), Indian Hedgehog (Ihh), bone morphogenetic protein 2 (Bmp2) and IL-15 as the most likely candidates to signal from surface to crypt to increase proliferation (figure 3A).

Figure 3

Gene and protein expression of downregulated signalling molecules Wnt inhibitory factor 1 (Wif1), Indian Hedgehog (Ihh), bone morphogenetic protein 2 (Bmp2) and interleukin 15 (IL-15). (A) Gene expression levels in total scrapings (mean±SEM, *q<0.01) and in surface and crypt compartments. (B) Immunohistochemical staining for Wif1 and chromogranin A (ChgA) in mouse colon. (C) Immunofluorescent staining of mouse colon for Wif1 (Cy3.5-conjugated antibody: red) and ChgA (ALEXA488-conjugated antibodies: green). Stainings were merged, including nuclear DAPI staining (blue). Co-localisation was indicated by yellow staining. (D) Number (mean±SEM) of ChgA- and Wif1-positive cells per crypt counted (n=6 mice per group); *p<0.05. (E) Immunohistochemical staining for β-catenin in mouse colon. (F) IL-15 protein levels measured by ELISA in control and haem colon scraping homogenates (n=9 for both groups) represented as mean±SEM, *p<0.001. LCM, laser capture microdissection.

Wif1 antagonises Wnt signalling by binding Wnts and thereby blocking binding of Wnts to their receptor. On the haem diet, Wif1 was downregulated fivefold. LCM analysis showed that the large overall downregulation of Wif1 could not be explained by the slight decrease in its expression in surface and TA cells (figure 3A), indicating that Wif1 expression is concentrated in between these locations. Immunohistochemical staining for Wif1 confirmed LCM data by showing that Wif1 was located in the upper half of the crypt (figure 3B), the part not included in LCM analysis (see figure 1 in online supplement). The localisation and morphology of Wif1-positive cells led us to hypothesise that Wif1 originates from enteroendocrine cells. Staining for chromogranin A (ChgA; enteroendocrine-specific) showed the same staining pattern as Wif1 (figure 3B). Moreover, immunofluorescent double-staining for Wif1 and ChgA visualised co-localisation (figure 3C) and demonstrated that all Wif1-positive cells were of enteroendocrine origin (figure 3C upper panels), but not all enteroendocrine cells express Wif1 (figure 3C lower panels). Quantification of Wif1-positive cells showed that Wif1-positive cells were less abundant in haem-fed mice (figure 3C), which is in line with the haem-induced downregulation of Wif1 gene expression in scrapings. Quantification of ChgA-positive cells showed that there were fewer enteroendocrine cells present per crypt in haem-fed mice (figure 3D). To investigate whether there was a downstream effect on Wnt signaling, β-catenin staining was performed (figure 3E). Staining showed that, in controls, β-catenin was mainly located on the cell membrane while, in haem-fed mice, it was concentrated more in the cytosol and nucleus, indicating activation of Wnt signalling.

Another downregulated secreted inhibitor of proliferation is Ihh. Ihh was downregulated 1.5-fold in total scrapings, and this was a surface-specific effect (figure 3A). Ihh antagonises proliferation by stimulating the secretion of Bmps by lamina propria cells.18 These Bmps are Wnt antagonists. In line with this, haem also decreased the expression of Bmp2. This downregulation was not reflected in surface or crypt cells, illustrating that LCM samples hardly contain lamina propria cells. Unfortunately, decreased Ihh protein expression could not be verified due to lack of an appropriate antibody. It should be noted that Bmp8a and b were upregulated rather than downregulated by haem (see table 4 in online supplement). The function of these Bmps in colon is unknown, and there are no indications that they play a role in epithelial cell proliferation.

Additionally, IL-15 mRNA was downregulated fourfold by haem, exclusively in the surface epithelium (figure 3A). IL-15 is a pleiotropic cytokine that inhibits proliferation of tumour cells independent of natural killer cells.19 IL-15 protein was eight times lower in haem-exposed colons than in control animals (figure 3F), so protein levels were decreased to an even greater extent than IL-15 mRNA levels.

Haem increased transcription of growth factor genes but this was not translated into mitogenic signals

Growth factors are obvious candidates to trigger hyperproliferation in the crypt. As mentioned above, Wnts were not changed but the growth factors Areg and Ereg were dramatically upregulated at the mRNA level (figure 4A). However, these increased mRNA levels were not translated into increased Areg and Ereg protein levels (figure 4A). Furthermore, mitogenic metabolites such as prostaglandins (PGs) might influence cell proliferation. Haem increased cyclo-oxygenase-2 (Cox-2) expression and the expression of the PGE2 efflux transporter Abcc4 in surface epithelium (figure 4B). Genes involved in PGE2 catabolism, such as the organic anion transporter Slco2a1 and the hydroxyprostaglandin dehydrogenase-15 (Hpgd), were significantly downregulated in total scrapings (data not shown). Since production and secretion of PGE2 was increased and catabolism decreased, this should lead to higher extracellular PGE2 concentrations. However, PGE2 levels of tissue culture supernatant were significantly decreased by haem (figure 4B). PGE2 can easily be converted to PGF, but PGF levels were also reduced by haem (figure 4B).

Figure 4

Gene expression and protein/metabolite levels of upregulated signalling molecules amphiregulin (Areg), epiregulin (Ereg) and prostaglandins (PGs) and of translation-regulating proteins. (A) Gene expression levels in total scrapings (mean±SEM, *q<0.01) and in surface and crypt compartments. Protein levels measured in homogenates (mean±SEM, n=8 mice/group). (B) Expression levels of PG-related genes and PGE2 and PGF levels in colon tissue culture supernatant measured by enzyme immunoassay (mean±SEM, n=8 mice/group,*p<0.01). (C) Expression levels of Ddit4 and 4E-BP1 (mean±SEM, *q<0.01). Western blot analysis using 4E-BP1 and β-actin specific antibodies. Immunohistochemical staining for 4E-BP1. LCM, laser capture microdissection.

Taken together, these results show that upregulation of mRNAs for mitogenic signals was not translated into effective products. Furthermore, as mentioned above, protein levels of the downregulated lL-15 were also twofold lower than expected from mRNA levels. This suggests that translation of mRNAs coding for secretory proteins is compromised by haem. In line with this, we found that the endoplasmic reticulum (ER) protein synthesis inhibitor Redd1 (Ddit4) was upregulated 2.5-fold, mainly in surface cells (figure 4C). Moreover, the cap-dependent translation inhibitor Eif4ebp1 (4E-BP1) was upregulated threefold, also mainly in the surface epithelium. Western blot analysis of colon homogenates showed that 4E-BP1 was increased at the protein level (figure 4C). Immunohistochemical staining showed that 4E-BP1 protein expression is increased by haem in the surface epithelium (figure 4C). Together these results indicate that the haem-stressed surface cells repress the ER translation of messengers for secretory proteins. We therefore conclude that haem increases proliferation by downregulating feedback inhibitors of proliferation and not by upregulating mitogenic signals.


To our knowledge, this is the first study showing that a non-absorbed nutrient, haem, induces colonic hyperproliferation and hyperplasia by repressing surface secretion of feedback inhibitors of proliferation. Microarray analysis of colonic mucosa showed that cellular stress response and cell cycle control were the processes most prominently changed by haem. LCM analysis revealed that stress responses occurred in surface cells only, whereas cell cycle control was changed near the bottom of the crypt. Furthermore, our results show that haem did not reach the TA crypt cells as gene expression levels of haem-sensing genes (such as Hmox1, Slc46a1 and Abcg2) and stress-related genes (such as Nemo, Cox-2 and Tis7) are surface-specific differentially expressed. Haem probably cannot reach TA cells because fluid efflux from the crypt counteracts the diffusion-driven influx of haem. Consequently, haem only enters the crypt lumen over a limited length. This implies that modulated signals from the stressed surface epithelium have to trigger compensatory cell proliferation in the crypt compartment.

Our data show that haem activates the oxidative stress-sensing transcription factor Nrf2 (Nfe2l2) and upregulates the expression of other stress-sensing transcription factors, as summarised in figure 5A. This is corroborated by our earlier studies showing that haem induces stress by generating reactive oxygen species (ROS) in the colonic lumen, reflected by a haem-dependent increase in faecal thiobarbituric acid reactive substances (TBARS).9 Moreover, dietary antioxidants inhibit the effects of haem on TBARS, luminal cytotoxicity and on the colonic mucosa.20 This is because antioxidants block the radical-mediated synthesis of the cytotoxic haem factor which is a covalently modified porphyrin molecule derived from haem.12 Unfortunately, the detailed structure of this haem metabolite remains uncertain as this modified porphyrin is refractory to ionisation and thus cannot be identified by mass spectrometry. We think that this cytotoxic factor, together with haem-induced ROS, triggers the activation of the stress-dependent transcription factors and their targets in surface cells (summarised in figure 5A). For instance, the significant upregulation of glutathione transferases (Gsta2, 3 and 4), synthases (Gclc and Gclm) and reductase (Gsr) indicates protection against noxious compounds that oxidise cellular SH groups. It has been shown that toxic and oxidative stress inhibits ER protein synthesis.21 In line with this, upregulated Ddit4 levels lead to inactivation of mTOR and thereby inhibit protein synthesis.22 Furthermore, increased mRNA and protein levels of 4E-BP1 can contribute to the inhibition of cap-dependent translation specifically in the surface epithelium. In our study the increase in mRNA transcripts of secreted proteins such as Areg and Ereg was not translated into protein. The increased synthesis of haem metabolism- and stress-related proteins (eg, Hmox1 and 4E-BP1, respectively) is not in conflict with this mechanism because stress response genes can be translated cap-independently by using internal ribosome entry sites.23

Figure 5

Model of proposed mechanism by which haem stresses surface cells and increases crypt cell proliferation. (A) Overview of main pathways changed by haem in surface cells. Fold changes in surface cells are shown in italics. (B) Model of haem-modulated signalling from surface to crypt cells. EEC, enteroendocrine cell; TA, transit-amplifying cell. Bold arrowheads indicate direction of haem-induced changes.

The observed upregulation of the anti-apoptotic NF-κB target genes Xiap and Bcl2l1 in surface cells can be either a response to oxidative or cytotoxic stress. It has recently been shown that upregulation of these NF-κB targets is a protective response in the surface epithelium to compensate for injured cells.24 In concert with Ier3, this may explain the inhibition of apoptosis observed in surface cells. Combined with the crypt-specific upregulation of survivin, this also explains the haem-reduced amount and activity of cleaved caspase-3 in mucosal homogenate, as reported previously.11

The mucosal signalling of surface cells might not only be compromised by inhibition of ER protein translation but also by Tis7-mediated repression of transcription which reduces histone-deacetylase activation.25 It has recently been shown that the Tis7-null mutant is not able to initiate compensatory hyperproliferation after mucosal injury, indicating that Tis7 is crucial in the signalling from injured cells to TA crypt cells.26 Our search for injury signalling molecules was focused on downregulated feedback inhibitors of proliferation. This is because homeostasis of epithelial cell turnover is theoretically only possible if differentiated surface cells feed back negatively on the TA cells. Furthermore, it is not so likely that damaged surface cells will invest much energy in producing a signalling molecule. Our secretome analysis showed that injured cells indeed regulate more signals down than up. The most relevant surface-specific downregulated candidate signalling molecules were Wif1, Ihh and IL-15. Our hypothesis about how these proteins can signal to the TA cells is shown in figure 5B.

Wif1 is a secreted Wnt antagonist that inhibits the Wnt signalling pathway.27 Haem-induced downregulation of Wif1 gene expression would lead to a more active Wnt pathway and to increased proliferation. Immunohistochemistry demonstrated that Wif1 protein resides in the upper half of the crypt in enteroendocrine cells. This is supported by previously reported Wif1 expression by Colo320 cells, a cell line of enteroendocrine origin.27 The fact that not all enteroendocrine cells express Wif1 reflects the different types of enteroendocrine cells in the gastrointestinal tract.28 Our novel finding of basic Wif1 expression in mouse colon contrasts with an earlier study indicating Wif1 expression only in colonic and small intestinal tumour tissue but not in normal wild-type tissue.29 However, our results are supported by two separate studies showing Wif1 expression in normal human colon tissue and its downregulation in primary colon cancer.27 30 The question remains whether Wif1 is a primary signalling molecule influencing proliferation rate or whether its downregulation is a secondary effect. For Wif1 to be an initiating factor, enteroendocrine cells should be able to sense the haem-induced cytotoxicity in the lumen. Wif1 downregulation can be a secondary effect if there is a molecule signalling from the surface to cells in the upper half of the crypt (eg, Ihh, see below). Another possibility is that the downregulation of Wif1 can be attributed to cell fate decisions and/or cell maturation under the influence of dietary haem as the number of ChgA-positive enteroendocrine cells per crypt also decreases. These different—but not mutually exclusive—possibilities require further studies focusing on epithelial cells in the upper half of the crypt.

Crypt cell proliferation can also be controlled by epithelium-to-mesenchyme interactions—for instance, Ihh can influence Bmp secretion from the lamina propria cells.10 Bmps are members of the transforming growth factor β family of growth factors that act as Wnt antagonists. Bmp4 has previously been identified as an Ihh-modulated mediator of mesenchymal-epithelial interactions,31 but Bmp2 and 7 might also play a role.18 Our data indicate that haem downregulated Bmp2 mainly in lamina propria (ie, mesenchymal) cells. We propose that this lowering of Bmp2 is caused by the lowered expression and secretion of Ihh from surface cells. Ihh also plays a role in epithelial cell differentiation.31 Whether this is a direct effect of Ihh on nascent epithelial cells or occurs via mesenchymal interactions is not known. Our LCM dataset shows that the Ihh receptor Patched-1 (Ptch1) was expressed in both surface and crypt cells and downregulated by haem (data not shown). This suggests that Ihh may also directly affect epithelial cells. We therefore speculate that Ihh supports the differentiation into Wif1-positive enteroendocrine cells, which might explain their reduced cell number on the haem diet. Mucosal signalling of Ihh requires further investigation, but pertinent to our study is the consistent finding that activation of Ihh signalling inhibits proliferation18 whereas blocking of Ihh signalling induces hyperproliferation.31

Our third surface-specific downregulated signal is IL-15. IL-15 is a pleiotropic cytokine that can inhibit tumour growth in murine models of cancer by stimulating natural killer cells and CD8+ cells.32 However, IL-15 also has direct antiproliferative effects as it inhibits tumour cell proliferation in mice lacking these cells.19 IL-15 knockout mice are protected against colitis induced by dextran sulfate sodium that injures the surface epithelium.33 This suggests that IL-15 must be downregulated for proper compensatory hyperproliferation of colonic epithelium. Alternatively, IL-15 stimulates cell differentiation34 and thus may affect the number of Wif1-positive enteroendocrine cells. Note that epithelial cells can sense IL-15 as its receptor (IL-15Ra) is expressed in surface and crypt epithelium (data not shown). These proposed effects of IL-15 require further nutritional studies in IL-15 knockout mice using control and haem-supplemented diets.

In conclusion, this study shows that haem induces hyperproliferation and hyperplasia by repressing the feedback inhibition of proliferation. The haem-induced aberrant turnover of crypt cells observed in this study is identical to that in rats,11 indicating that this haem effect is species-independent. This is further supported by the finding that the protective effect of chlorophyll against haem found in rats is also observed in epidemiological studies in humans.8 These species-independent effects suggest that the mechanism proposed in this study can be extrapolated to humans. Studies on repression of mucosal signalling by haem in humans would provide further insight into the molecular mechanisms of the relationship between red meat consumption and the risk of colon cancer observed in epidemiological studies.


The authors thank Mechteld Grootte Bromhaar, Jenny Jansen, Philip de Groot and Mark Boekschoten for microarray analysis and Bert Weijers for biotechnical assistance.


View Abstract

Supplementary materials


  • Funding This study was funded by the program council of TIFN (grant A-1001). The council had no vote in study design, data collection and analysis, preparation of the manuscript or decision to publish.

  • Competing interests None.

  • Ethics approval The experiments were approved by the Ethical Committee on Animal Testing of Wageningen University and were in accordance with national law.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement Array data have been submitted to the Gene Expression Omnibus, accession number GSE27849.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • Digest
    Emad El-Omar William Grady Alexander Gerbes