Role of Heme Oxygenase-1 in Human Endothelial Cells

Heme oxygenase-1 (HO-1) is an enzyme degrading heme to iron ions, carbon monoxide (CO), and biliverdin; the latter is subsequently converted to bilirubin by biliverdin reductase. Products of HO-1 activity perform important physiological functions in the vascular system, which, together with the removal of toxic heme, are ultimately linked to the protection of endothelium.¹

Accordingly, endothelial cells isolated from HO-1 knockout mice are more sensitive to oxidized lipid–induced injury and more susceptible to H₂O₂-induced cell death than those isolated from wild type individuals.² Also, in the sole known human case of HO-1 deficiency, endothelial cell damages were prominent because of the heme-mediated oxidation of low-density lipoprotein and lack of adaptive responses.³ It seems, however, that a much more important and common phenomenon in the human population is variability in HO-1 expression levels, resulting from allelic variants of the HO-1 promoter.

The 5′-flanking region of the human HO-1 gene contains a fragment of (GT)n microsatellite DNA. A number of dinucleotide repeats, ranging from 11 to 42,^4,5 can modulate the level of transcription.⁶ For example, reporter assays demonstrated that the HO-1 promoter constructs harboring (GT)₂₉ or (GT)₃₈ sequences were less active than those with (GT)₁₁, (GT)₁₆, or (GT)₂₀.^5–7 However, such experiments were performed only with fragments of the HO-1 promoter, whereas some regulatory sequences are located far upstream of the transcription initiation site, as well as in the introns of HO-1 gene.⁸ Until now, only 2 reports have been published in which the lymphoblastoid cells or mononuclear blood cells possessing short alleles displayed a higher level of HO-1 than cells carrying long alleles.^9,10

Importantly, though, although large scale analysis did not confirm a meaningful effect of HO-1 promoter polymorphism on coronary artery disease or myocardial infarction,¹¹ there are many clinical data indicating its influence on cardiovascular complications, at least in some groups of patients. Thus, the presence of longer, less active alleles was associated with an increased risk of arteriovenous fistula failure in people subjected to hemodialysis,¹² higher incidence of coronary artery disease in type 2 diabetic or hemodialyzed patients,^7,10,12,13 elevated rate of restenosis after balloon angioplasty,¹⁴ or more frequent aortic aneurysms¹⁵ and cerebrovascular events.¹⁶ Moreover, among patients with peripheral artery disease, those carrying longer HO-1 alleles had higher rates of myocardial infarction, percutaneous coronary interventions, and coronary bypass operations.¹⁷

Some reports have suggested that polymorphism of the HO-1 promoter may also influence the outcomes of stenting, as the presence of more active alleles correlates with reduced adverse cardiac events.^18,19 These observations have not been confirmed by larger-scale studies.^4,20 Nevertheless, patients carrying the short alleles had less pronounced serum lipid peroxidation^7,4 and a milder postintervention inflammatory response.⁴

Thus, the number of GT repeats seems to influence the progression of some cardiovascular diseases, especially those associated with oxidative stress, inflammation, and endothelial dysfunction. Our aim was to elucidate whether (GT)n promoter allelic variants can actually modulate the expression and biological effects of HO-1 in primary endothelial cells.

Materials and Methods

For a more detailed description of methods, please see the supplemental text, available online at http://atvb.ahajournals.org.

Cell Culture

Experiments were carried out using human umbilical vein endothelial cells (HUVEC) freshly isolated from healthy, anonymous newborns. Collecting the umbilical cords was conducted according to the guidelines of the ethical commissions on human research of the authors’ hospitals in Vienna and Lodz.

Genotyping of HO-1 Promoter

Determination of the number of GT repeats in the HO-1 promoter was performed as described earlier.¹⁸

Analysis of mRNA Expression

Analysis of mRNA was performed using quantitative RT-PCR.

Measurement of the HO-1 Protein

Concentration of the HO-1 protein in cell lysates was determined using enzyme-linked immunosorbent assay (ELISA) according to the vendor’s protocol.

Measurement of Inflammatory Mediators

Concentrations of interleukin (IL)-1β, IL-6, IL-8, tumor necrosis factor-α, and soluble E-selectin were determined using ELISA.

Measurement of Glutathione

Each HUVEC batch was cultured in 2 25-cm² flasks. Cells in one flask remained intact, but cells in the other flask were exposed to H₂O₂ (100 μmol/L) for 16 hours. Reverse-phase high-performance liquid chromatography was used to quantify glutathione (GSH) in the cell lysates according to the procedure described elsewhere,²¹ with a slight modification.

3-(4,5-Dimethylthiazol-2-yl)-2,5-Diphenyl Tetrazolium Bromide Reduction Assay

Cells were cultured in 96-well plates (100 μL medium per well) and exposed to H₂O₂ (100 to 800 μmol/L) for 3 to 24 hours. Then, a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide solution (final concentration of 0.5 mg/mL) was added for 2 hours. Reaction was stopped by adding 50 μL of lysis buffer (20% SDS, 50% dimethylformamide). Absorbance was read at a wavelength of 570 nm.

Annexin-V Staining

Cells were cultured in 96-well plates and exposed to H₂O₂ (100 μmol/L) for 24 hours. Annexin-V staining was performed using a TiterTACS 96-well apoptosis detection kit, according to the manufacturer’s instructions.

Cell Migration Assay

For each HUVEC batch, the spontaneous and vascular endothelial growth factor (VEGF)-A₁₆₅–induced migration was analyzed using modified Boyden chambers.

Capillary Sprouting

The measurements of capillary sprouting were performed according to the procedure described by Korff and Augustin.²² HUVEC spheroids were embedded in collagen gels and remained intact or were stimulated with VEGF-A₁₆₅ (30 ng/mL). The sprouting of capillary-like structures was quantified 24 hours later by measuring the length of the sprouts that had grown out of each spheroid.

Cell Proliferation Assay

Cells were seeded in 96-well plates (5000 per well) in a medium devoid of endothelial cell growth supplement but supplemented with 10% FCS, and they remained intact or were stimulated with VEGF-A₁₆₅ (30 ng/mL). After a 42-hour incubation period, the 5-bromodeoxyuridine solution (1 μmol/L) was added for 6 hours, and then proliferation was measured using colorimetric ELISA, according to the vendor’s protocol.

Statistical Analysis

All data are shown as the mean±SEM. One-way ANOVA followed by a Tukey posteriori test (for comparison of multiple samples) or Student t test (for comparison of 2 samples) was used to analyze the statistical differences.

Results

Distribution of (GT)n alleles in the population studied and the response of HUVEC to the stimuli are described in the supplemental text and shown in Supplemental Figures I through III.

Effect of HO-1 Promoter Allelic Variants on HO-1 Expression

We compared the expression of HO-1 mRNA and protein in HUVEC batches classified according to the number of GT repeats in the shorter allele of the HO-1 promoter. Under control conditions, the HO-1 mRNA level was the highest in cells possessing the short (S) (22 or 23 repeats) allele (Figure 1). Basal expressions in medium (M) (24 to 28 repeats) and long (L) (29 to 37 repeats) carriers were lower by ≈40% (P<0.05) and ≈60% (P<0.001), respectively. Measurements of HO-1 protein concentrations in cell lysates generally confirmed the results of quantitative RT-PCR, although differences between groups were smaller, reaching statistical significance only when S and L alleles were compared (Figure 1). This was possibly caused by weaker sensitivity of ELISA.

• Download figure
• Download PowerPoint

Comparison of the results obtained for individual HUVEC batches showed that there was a significant positive correlation between the expression of HO-1 measured at the mRNA and protein levels (r=0.63, P=0.002).

Exposure to H₂O₂, cobalt protoporphyrin (CoPP) and 15-Deoxy-Δ^12,14-prostaglandin-J₂ (15d-PGJ₂) led to much stronger upregulation of HO-1 mRNA in carriers of S alleles than in cells possessing M or L types (Figure 1). Analysis of protein showed the same tendency, but with statistically significant differences apparent only between the S and L groups. The influence of HO-1 promoter polymorphism was also observed in cells incubated with lipopolysaccharide (LPS). Here, the meaningful differences were noted only at the mRNA level between the S and L carriers. A similar but not statistically significant trend was visible in HUVEC treated with interferon γ. In contrast, the response to hemin or hypoxia was not affected by the number of GT repeats (Figure 1). A general comparison of HO-1 inducibility of different allelic variants is shown in Supplemental Figures IVA, IVB, and V. Interestingly, differences in HO-1 expression between HUVEC carrying S, M and L alleles also influenced the upregulation of ferritin (Supplemental Figure IVC and supplemental text).

Importantly, gas chromatography measurements of CO concentrations performed in media harvested from different HUVEC batches, suggest that differences in HO-1 expression are reflected in HO-1 activity (Supplemental Figure VI). Namely, untreated cells carrying the S alleles released more CO to the culture media than those with L alleles. A similar tendency was observed after the stimulation of cells with H₂O₂, whereas the most significant differences were found in cells stimulated with 15d-PGJ₂ (P=0.050, 0.067, and 0.004 for untreated, H₂O₂-stimulated, and 15-PGJ₂-stimulated cells, respectively; N=10 for the L group and N=17 for the S group).

Effect of HO-1 Promoter Allelic Variants on Inflammatory Response

Release of IL-1β, IL-6, and soluble intercellular adhesion molecule-1 was significantly lower in the control HUVEC carrying short (GT)n fragments than in their counterparts with longer HO-1 alleles (Figure 2). The same tendency was observed in cells treated with LPS, but here only differences in IL-1β production were statistically significant. On the other hand, the HO-1 promoter polymorphism did not seem to affect basal production of soluble E-selectin and tumor necrosis factor-α, although there was some tendency for a more pronounced response to LPS in cells from the M and L groups. Finally, it did not modulate expression of IL-8 (Figure 2).

• Download figure
• Download PowerPoint

Effect of HO-1 Promoter Allelic Variants on the Oxidative Status of Endothelial Cells

To assess the antioxidative efficacy of HO-1, we measured the concentration of total GSH, as well as GSH and oxidized glutathione (GSSG), in cells cultured under control conditions or exposed for 16 hours to H₂O₂ (100 μmol/L). The levels of total GSH in the control cells were similar in the S, M, and L groups (Figure 3A) and were slightly increased in the M and L carriers treated with H₂O₂ (to 2.03±0.65 and 2.48±0.39 nmol/mg, respectively). Concentrations of reduced GSH (Figure 3B) were also comparable in the HUVEC of all genotypes (P>0.5 and P>0.4 for control and H₂O₂-treated cells, respectively). In contrast, we observed the strong effect of HO-1 promoter polymorphism on the level of GSSG, which was the highest in the L group (Figure 3C). As a consequence, oxidative status measured as a GSH:GSSG ratio was much more favorable in cells with S allele than in carriers of M or L variants (12.62±3.63 versus 4.11±1.19 or 3.55±0.82 in control cells, P<0.05) (Figure 3D). Unexpectedly, we did not find significant differences between untreated and H₂O₂-treated cells. Possibly, to observe the response to H₂O₂ we should perform analyses at earlier time points.

• Download figure
• Download PowerPoint

The effect of H₂O₂ on expression of other cytoprotective genes are described in the supplemental text and shown in Supplemental Figure VII.

Effect of HO-1 Promoter Allelic Variants on Viability of Endothelial Cells

The oxidative status of cells was reflected by the survival of HUVEC exposed to H₂O₂ (100 to 800 μmol/L, 3 to 24 hours). Results of the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide reduction assay demonstrated that endothelial cells carrying the S allele are much more resistant to oxidative stress than those with the less active HO-1 promoter (Figure 4A through 4D). Staining for annexin-V performed after 24 hours of incubation with H₂O₂ (100 μmol/L) confirmed the highest rate of apoptosis in endothelial cells with L HO-1 allele and thus with the lowest activity of HO-1 promoter (Figure 4E).

• Download figure
• Download PowerPoint

Effect of HO-1 Promoter Allelic Variants on Angiogenic Potential

Finally, we assessed the response of HUVEC to stimulation with VEGF-A, that being the crucial proangiogenic factor. The 5-bromodeoxyuridine incorporation assay showed that proliferation after 48 hours of incubation with VEGF-A (30 ng/mL) was much more pronounced in cells carrying the S allele than in those with the M or L allele (Figure 5A). The fold of induction was, respectively, 3.38±0.34, 1.98±0.41, and 1.53±0.23 (P<0.001). A similar relationship was demonstrated in which cells from SS and LL genotypes were compared (Supplemental Figure VIIIA).

• Download figure
• Download PowerPoint

In contrast, the migration of cells in response to VEGF-A was not significantly influenced by HO-1 promoter polymorphism, as estimated by using Boyden chamber assay (Figure 5B). This observation was confirmed by measuring the sprouts growing out of endothelial spheroids embedded in collagen gel, a process relying primarily on cell motility (Figure 5C). Again, a similar lack of influence was demonstrated when the HUVEC of SS and LL cells were compared (Supplemental Figure VIIIB).

Discussion

We have demonstrated that allelic variants of (GT)n repeats in human HO-1 promoter can modulate the level of gene transcription in primary endothelial cells, cultured in basal conditions or treated with H₂O₂, CoPP, 15d-PGJ₂, and LPS. In contrast, these variants do not significantly affect HO-1 expression in cells exposed to interferon γ, hemin, and hypoxia. This may suggest that activities of general transcription factors and transcription initiation complex are not affected by the length of the (GT)n repeat sequence.

One might notice that the Nrf2 transcription factor is the common mediator involved in HO-1 induction on treatment with CoPP,²³ 15d-PGJ₂,²⁴ H₂O₂,²⁵ and LPS.²³ Nrf2 was shown to recruit BRG1 (Brahma-related gene-1) protein to the proximal, GT-repeat-containing fragment of the HO-1 promoter, which thereby may interact with the distal enhancer motifs.^26,27 Our data might suggest that long microsatellite fragments reduce such transactivation. This speculation, however, needs further experimental verification.

Many articles have described the significant effects of (GT)n polymorphism in the HO-1 promoter on cerebrovascular and cardiac disorders,^{7,10,12–16,18,19} the outcomes of transplantations,²⁸ and progression of some cancers.^29,30 Strikingly, in all the earlier articles, the criteria for classification of HO-1 promoter alleles as “short” or “long” were chosen arbitrarily. Thus, the maximal, threshold number of GT repeats in alleles regarded as short varied widely: 24,^11,16–18 25,¹⁹ 26,¹³ 27,¹¹ 29,³⁰ 30,^12,14 32,⁷ or 36 repeats.¹⁵ In some analyses, the alleles were divided into 3 groups: short, medium, and long. In these cases, the cutoffs were also different, being located at 24 and 30 repeats,^6,30 25 and 31,³¹ 26 and 33^9,29 or 29 and 38.⁴

We have demonstrated for the first time the real effect of different numbers of GT repeats on basal and induced HO-1 expression (see supplemental text). The reasonable threshold values for allele classification are 23 repeats (for S, the most active alleles) and 28 repeats (for moderately active alleles). L alleles with microsatellite DNA fragments carrying 29 or more GTs are the least active. We chose a threshold of 23 for the S allele, similar to the most common cutoff (24) chosen in other studies. Because alleles with 24 GTs are relatively rare, our study generally confirms that the arbitrary classification used in most articles reflects the actual effect of (GT)n on HO-1 expression. Furthermore, we showed that the most important factor is the number of repeats in the shorter allele. The length of the (GT)n fragment in the second allele of the HO-1 promoter is negligible (Supplemental Figure IVB).

One of the most studied functions of HO-1 is cytoprotection, resulting from both the removal of prooxidant heme and generation of biologically active products. It has been postulated, however, that too high an activity of HO-1, especially when not associated with sufficient expression of ferritin, may be detrimental to the cells, probably because of an increase in the free iron pool.³²

To investigate the HO-1-mediated cytoprotection, various models were used, including cells with enforced HO-1 overexpression and HO-1 knockout or transgenic mice.¹ The question that was still unresolved, however, was whether relatively subtle changes in HO-1 expression resulting from promoter allelic variants can exert a biologically significant impact. Our data show that variation of HO-1 inducibility observed in the human population influences the sensitivity of primary endothelial cells to oxidative stress and that induction of HO-1 within the physiological range does play a protective and antiapoptotic role. Moreover, cells carrying the more active HO-1 variant display a more appropriate GSH:GSSG ratio, which may reflect the antioxidative potential of the HO-1/biliverdin reductase pathway. A similar relationship was observed in transgenic mice after cardiovascular damage induced by angiotensin-II, in which enforced cardiac overexpression of HO-1 protected cells from decrease in the GSH:GSSG ratio.³³

Interestingly, experiments performed in a pheochromocytoma cell line exposed to the CO-releasing molecule indicate that CO may increase the concentration of GSH by Nrf2-dependent upregulation of the catalytic subunit of glutamate-cystein ligase, which represents the rate-limiting enzyme in GSH synthesis.³⁴ However, it seems that this mechanism does not play a role in our experimental setting. Instead, we suppose that higher expression of HO-1 in endothelial cells carrying the S allelic variant of the HO-1 promoter was associated with prevention of GSH oxidation, leading to a decreased requirement as well as decreased synthesis of GSH. This resulted in a better GSH:GSSG ratio and lower concentration of GSH in the cells.

It should also be remembered that treatment of HUVEC with H₂O₂ induces many other genes, some of them directly involved in H₂O₂ inactivation, such as catalase, Thrx, and ThrxR. Thus, the protective effects of HO-1 can be supported by the activity of additional antioxidative pathways, although induction of HO-1 seems to be one of the strongest responses to the oxidative stress.

Interestingly, HO-1 upregulation on treatment with 15d-PGJ₂, CoPP, or H₂O₂ is accompanied by an augmented expression of ferritin (see supplemental text). An association between HO-1 induction and synthesis of ferritin has been already described.³⁵ It has also been suggested that type 2 diabetes patients carrying short (GT)n repeats in HO-1 promoter may have a higher serum ferritin concentration.³⁶ We demonstrated that ferritin upregulation in endothelial cells correlates with HO-1 inducibility and is significantly augmented in carriers of S alleles. This may provide protection from a possible increase in the free iron pool.

HO-1 is commonly regarded as an antiinflammatory enzyme. Its importance is elegantly illustrated in HO-1 knockout mice, in which HO-1 deficiency leads to increased production of proinflammatory cytokines.³⁷ Also, in patients subjected to bypass surgery, a higher activity of HO-1 resulted in a lower concentration of IL-6.⁴ Our data fully confirm the antiinflammatory potential of HO-1 and show that even relatively small differences in HO-1 expression in cells carrying distinct promoter alleles are enough to modulate generation of IL-1β, IL-6, and soluble intercellular adhesion molecule-1. Similar trends were observed for tumor necrosis factor-α and soluble E-selectin.

The relationship between HO-1 and IL-8 raises more controversy. Some data suggest that activation of HO-1 by CoPP and CoCl₂ leads to a decrease in production of IL-8.³⁸ In contrast, we demonstrated that upregulation of IL-8 in response to 15d-PGJ₂, CoPP, and CoCl₂ was HO-1-independent.^39,40 Measurements of concentrations of IL-8 released from HUVEC of different HO-1 genotypes provided evidence that generation of IL-8 was not influenced by HO-1 expression.

Finally, we analyzed the effect of HO-1 promoter allelic variants on the angiogenic potential of HUVEC. The link of HO-1 and angiogenesis was first indicated by augmented endothelial proliferation resulting from HO-1 overexpression.⁴¹ Accordingly, inhibition of HO-1 disrupted the response of endothelial cells to growth factors.⁴² This defect was also seen in vivo, as the neovascularization of wounds was impaired in HO-1 knockout mice.^43,44 Current results demonstrate the importance of small changes in HO-1 expression for the endothelial mitogenic response to VEGF and may suggest the weaker angiogenic potential of endothelium in patients with less active variants of HO-1.

In contrast, HO-1 promoter polymorphism did not influence VEGF-induced migration of HUVEC. This was unexpected, as the tin protoporphyrin, HO-1 inhibitor, markedly decreased endothelial cell motility.⁴² Moreover, recent experiments have shown the essential role of HO-1-derived CO in stromal cell-derived factor-1-induced migration of endothelial progenitors, as in the cells isolated from HO-1^−/− mice, the migration was potently attenuated.⁴³

Therefore, we suppose that some activity of HO-1 is necessary for cell migration, but after crossing a threshold value, further upregulation of HO-1 is not so important. This suggestion is supported by our unpublished observation that migration measured by scratch assay, although strongly reduced in HO-1^−/− endothelial progenitors, is very similar in endothelial progenitor cells isolated from HO-1^+/+ and HO-1^+/− mice (data not shown). Thus, analysis of HUVEC indicates that possibly, the impairment of migration is not a significant factor inhibiting the angiogenic potential of endothelial cells in patients with a less active HO-1 promoter.

In summary, (GT)n allelic variants of the HO-1 promoter directly modulate the level of HO-1 expression in human primary endothelial cells. HUVEC batches carrying the shorter, more active HO-1 alleles survive better under oxidative stress, proliferate more effectively in response to VEGF-A, and produce lower levels of proinflammatory mediators. Thus, the lesson from HO-1 promoter polymorphism confirms the cytoprotective, promitogenic, and antiinflammatory role of HO-1 in endothelium and indicates that the efficacy of this enzyme can vary significantly in the human population.

Footnotes