Anti-tumor efficacy of naked siRNAs for ERBB3 or AKT2 against lung adenocarcinoma cell xenografts


  • This article was published online March 8, 2011. An error was subsequently identified. This notice is included in the online and print versions to indicate that both have been corrected November 22, 2011.


The use of siRNAs against specific molecular targets has potential for cancer therapy but has been thought to be limited by the need for formulation to improve cellular uptake. Lung adenocarcinoma cells are markedly suppressed in culture by siRNAs to the receptor ERBB3 or its downstream signaling partner AKT2. We now demonstrate that naked, unformulated siRNAs to ERBB3 or AKT2, administered i.v. as saline solutions, 2 μg/g five times per week for 3 weeks (total dose 30 μg/g), were effective suppressors of growth of A549 human lung adenocarcinoma cell xenografts in athymic mice, 12 mice per group, in four different experiments. ERBB3 and AKT2 siRNAs each inhibited growth by 70–90% on average, compared to saline-treated or untreated controls; a nonsilencing siRNA was without significant effect. Lesser but significant effects were noted with a total dose of 12 μg/g. With the higher dose, effects persisted for several weeks after the end of treatment. Expected reductions of ERBB3 and AKT2 mRNAs and proteins occurred and correlated with decrease in tumor volume. There were no significant changes in serum cytokines. These results show that naked siRNAs to ERBB3 or AKT2 may have potential for lung cancer therapy.

Lung cancer is the leading cause of cancer-related deaths in the United States, and its incidence is increasing rapidly in developing countries. Non-small cell lung carcinoma accounts for 80% of all lung cancers. Most patients have advanced stage at the time of diagnosis, and the survival rates have not changed significantly over the years.1 Intrinsic and acquired resistance of lung cancer to chemotherapy remains a challenge for lung cancer treatment. Currently, there is emphasis on improvement in lung cancer treatment by therapy based on specific molecular targets.2

ERBB3 is a member of the epidermal growth factor receptor family and is unique among this family in that it has impaired kinase activity. Its kinase domain functions as a specialized allosteric activator of other ERBB family members.3, 4 ERBB3 is activated by ligand heregulin and by heterodimerization with other ERBB family members and is thought to be active only as a heterodimer. It is centrally involved in resistance of some cancers to therapies directed at EGFR or ERBB2.5, 6

Downstream, tyrosine-phosphorylated ERBB3 has a uniquely high capacity to activate phosphatidylinositol 3-kinase (PI3K) directly, via six direct binding motifs for the regulatory unit of PI3K,7 which in turn activates AKT signaling. The ERBB3 receptor has a central role in maintenance and malignancy of cancers of the lung and several other tissues, often involving signaling through the PI3K/AKT pathway.8 Transgenic ErbB3 mice developed a high incidence of lung tumors compared to nontransgenic mice.9 High ERBB3 expression was linked to poor prognosis in lung cancer patients.10 Moreover, ERBB3 is directly involved in connecting EGFR to the PI3K/AKT pathway in a subset of EGFR-driven lung tumors.11 EGFR tyrosine kinase inhibitors blocked ERBB3 activation in these lung tumor cells. Reactivation of ERBB3 through the oncogene MET has been implicated in the resistance of EGFR-targeted therapy.12

Inactivation of ERBB3 could therefore be therapeutically useful. However, targeting of ERBB3 therapeutically with the traditional approach, use of kinase domain inhibitors, has been regarded as not practicable, because of impairment of its kinase domain. An alternative approach is to suppress ERBB3 protein synthesis specifically, using siRNA. Our earlier studies implicated ERBB3 and AKT in growth of malignant lung cells and demonstrated that small inhibitory RNAs (siRNAs) to ERBB3 and to the AKT isoforms were active in cultured lung adenocarcinoma cells, effectively suppressing tumor cell growth, survival and invasiveness.13–15 The next important question was whether these siRNAs would be effective against tumors in vivo. It has been thought that naked siRNAs are of limited usefulness in vivo, because of the lack of stability and poor entry into cells.16, 17 However, recently, unformulated doubled-stranded siRNA administered i.p. was detected and quantified in mouse tissues, at levels sufficient to bring about gene silencing.18 Such siRNAs, given i.p. or i.v., have shown effectiveness against a variety of xenografted tumors,18 including non-small cell lung cancer.19 In the study reported here, we demonstrate that i.v. administration of unmodified siRNA to ERBB3 or AKT2 resulted in the expected specific effects on their respective molecular targets, and significantly reduced the growth of A549 human lung adenocarcinoma cell xenografts.

Material and Methods

Establishment of xenograft tumors and siRNA injection

Xenograft procedures and analyses followed the advice of Hollingshead.20 Human lung adenocarcinoma cell line A549 was obtained from the American Type Culture Collection and cultured in RPMI 1640 medium with 10% fetal bovine serum, according to the cells' supplier's protocol, for a maximum of four passages before use. The A549 cells were harvested at 70–80% confluence, washed with phosphate-buffered saline, suspended in phosphate-buffered saline and implanted subcutaneously at 5 × 106 cells/0.2 ml into Swiss athymic nude mice, obtained from the Animal Production Area at NCI-Frederick. All animals used in this research project were cared for and used humanely according to the following policies: The U.S. Public Health Service Policy on Humane Care and Use of Animals (1996), the Guide for the Care and Use of Laboratory Animals (1996) and the U.S. Government Principles for Utilization and Care of Vertebrate Animals Used in Testing, Research, and Training (1985). All NCI-Frederick animal facilities and the animal program are accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International.

When the tumors reached ∼3 × 3 mm, the mice were distributed randomly into groups of 10–12 for treatment. HPLC-purified ERBB3, AKT2 and nonsilencing siRNAs were obtained from Qiagen. The sequences used were as follows: ERBB3 siRNA : sense: 5′ r(CCAAUACCAGACACUGUAC)d(TT) 3′, antisense: 5′ (GUACAGUGUCUGGUAUUGG)d(TT) 3′; AKT2 siRNA : sense : 5′ r(GAGCGACGGCUCCUUCAUU)d(TT) 3′, antisense: 5′ r(AAUGAAGGAGCCGUCGCUC)d(TT) 3′; Nonsilencing siRNA: sense: 5′ r(UUCUCCGAACGUGUCACGU)d(TT)3′, antisense: 5′ r(ACGUGACACGUUCGGAGAA)d(TT)3′ as reported previously (15). The siRNAs were freshly resuspended in 0.15 N NaCl, mixed by vortexing for 10 sec, incubated at room temperature for 10 min and finally vortexed vigorously for 5–10 sec. They were then injected into the warmed tail veins of the mice, 2 μg/g body weight in 0.2 ml, five times per week for 3 weeks. Other groups were treated with saline or were untreated. This experiment was repeated four times. Mice were weighed, and tumors were measured three times per week. Tumor volumes in mm3 were estimated by the formula ½(length × width2). In three experiments, mice were euthanized 2 hr after the last treatment. Tumors were divided and portions either fixed in formalin for histology or frozen at −80°C for protein and RNA analysis. In the fourth experiment, the mice were maintained after the last treatment until moribund. In a fifth experiment, a reduced total dose was tried, utilizing the same daily dose but only twice per week. In some of the experiments, a few early deaths occurred that were not related to specific treatment; no mice surviving to 18 days were excluded from the dataset.

QuantiGene assay to determine mRNA levels

QuantiGene 1.0 reagent system kit (Panomics/Affymetrix) was used to quantify the mRNA levels of ERBB3 and AKT2 relative to a housekeeping gene, human peptidylprolyl isomerase B (PPIB), in the siRNA treated or control tumor homogenates. Tumor homogenates were prepared according to the procedure described in the QuantiGene kit. Briefly, 400 μl of homogenizing solution with 4 μl of proteinase K (50 μg/μl) was added to 20-mg tumor sample and homogenized using a mini glass grinder (Radnoti grinders). The tissue homogenates were incubated at 65°C for 30 min, and the samples were vortexed every 5 min for 15 sec at maximal speed. The homogenate was centrifuged at 16,000 g for 5 min to remove debris, and the homogenate was frozen at −80°C and stored until further use.

Human-specific ERBB3 and AKT2 probes were custom synthesized by the company. The levels of mRNA were quantified according to manufacturer's recommended protocol. The luminescence signal was measured in a microplate luminometer (MicroLumet Plus LB, Berthold Technologies). The no-template background values were subtracted from each probe set signal. Linearity levels were established for each probe.

Western immunoblotting

Tumor tissues were homogenized in CelLytic tissue lysis buffer (Sigma) with freshly added phenylmethylsulfonyl fluoride (1 mmol/L), Na3VO4 (1 mmol/L) and a cocktail of protease inhibitors (Sigma). Tumor lysates (50-μg protein for ERBB3 blots or 20 μg for AKT2 blots) were separated by gel electrophoresis (4–12% NuPAGE gel for ERBB3 and 12% Tris-glycine gel for AKT2), transferred to PVDF membranes and probed with anti-ERBB3 antibody (Santa Cruz, SC-285, 1:1000) or anti-AKT2 antibody (Cell Signaling #2692, 1:1000). The lower half of ERBB3 blots was cut off and probed with anti-β-actin (Abcam, 1:10,000). AKT2 blots were stripped and reprobed for β-actin. Bands were scanned with a Visioneer one-touch 9,320 scanner (Pleasanton, CA) and quantified with UnScanIt software (Silk Scientific, Orem, UT).


Immunohistochemical staining of 5-μm sections of the formalin-fixed paraffin-embedded tumors was carried out using antibodies and protocols from Cell Signaling Technology, anti-phospho-ERBB3 (pERBB3; # 4791, 1:250) and anti-AKT2 (# 4059, 1:150), with rabbit IgG monoclonal antibody (# 3900, 1:400) as the secondary probe. For anti-total ERBB3, Santa Cruz SC-285, 1:200 and rabbit IgG polyclonal were used, with Zymed # 02-6102, 1:250 as the secondary antibody. Nonimmune controls with the monoclonal rabbit IgG antibody at 1:150 or 1:400 were used to confirm specificity of the staining with anti-ERBB3 and anti-AKT2 antibodies.

Cytokine analysis

Sera were collected 2 hr after the last treatment from mice from two different experiments in which tumor-inhibitory doses of ERBB3 and AKT2 siRNAs were utilized. All of the cytokines except for interferon-α were analyzed using the multiplex kit from Meso Scale Discovery, following the manufacturer's protocol. Interferon-α was quantified with an ELISA kit (product #42100-1) from Endogen (Piscataway, NJ).

Statistical analysis

Median tumor volumes were compared using Kruskal–Wallis nonparametric analysis of variance and Dunn's multiple comparison test (GraphPad Instat version 3.0, GraphPad Software, San Diego, CA). Confidence intervals for medians were obtained from published tables.21 Correlations between endpoints were assessed by the Pearson linear correlation test or the Spearman nonparametric rank sum test, as appropriate. Tumor growth slopes for each mouse were calculated by linear regression, using tumor volumes measured twice per week. Analysis of parametric data, which included the average growth slopes and the average cytokine levels, utilized the analysis of variance t-test.


ERBB3 or AKT2 siRNA inhibited A549 xenograft tumor growth

Previously, we demonstrated that inactivation of ERBB3 or AKT2 by siRNA promoted apoptosis and attenuated growth and invasiveness of A549 human lung adenocarcinoma cells in vitro.15 Now, we tested whether naked siRNAs to ERBB3 or to AKT2 would suppress the growth of A549 cells in vivo as subcutaneous xenografts in athymic nude mice. Effects of the ERBB3 or AKT2 siRNAs were compared to nonsilencing siRNA, saline or no treatment. Data from four different experiments were analyzed collectively (total N per group = 44–48) and are presented together in Figure 1 and Tables 1 and 2. Two measures of effect were used, tumor growth rate (Table 1) and median tumor size at each time point (quantitative data for the final time point in Table 2). Naked ERBB3 siRNA or naked AKT2 siRNA treatments for 3 weeks at 2 μg/g/5 days per week resulted in highly significant, reproducible inhibition in tumor growth compared to that of nonsilencing siRNA- or saline-treated mice or mice with no treatment. During treatment, tumor growth rates were suppressed by ERBB3 or AKT2 siRNA by about 90% compared to untreated mice and by about 84% compared to nonsilencing siRNA or saline treatment (Table 1). By the last treatment on day 18, ERBB3 siRNA or AKT2 siRNA reduced tumor volume by 73–88% compared to the control groups (Table 2). The volumes of tumors from mice treated with control nonsilencing siRNA did not differ significantly from those mice receiving saline alone. The nonsilencing siRNA and saline only controls had reduced average tumor growth and final tumor size compared to the no-treatment controls, but these differences were not significant.

Figure 1.

Suppression of in vivo growth of A549 lung adenocarcinoma cells as xenografts in nude mice by siRNAs for ERBB3 or AKT2. Doses of the siRNAs were 2 μg/g body weight in 0.2 ml, five times per week for 3 weeks. Data shown are combined from four separate experiments, with a total of 44–48 mice/group (see Table 1). Tumor growth rates are compared quantitatively in Table 1. Median tumor volumes at the last time point (day 18) are shown in Table 2. There were significance differences (p < 0.0001) between the groups by day 2. By day 16, ERBB3 siRNA and AKT2 siRNA treated tumors were each significantly different from each of the control groups.

Table 1. Tumor growth rates and duration of effect of siRNAs to ERBB3 or AKT2
inline image
Table 2. Effects of ERBB3 and AKT3 siRNAs on median tumor volume of A549 lung adenocarcinoma xenografts after 18 days, and dose-response
inline image

The specific effects of the ERBB3 and AKT2 siRNAs were not associated with reduced body weights. Average body weights (g) ± SE on day 18 were 24.5 ± 0.2, 24.4 ± 0.3, 24.6 ± 0.3, 24.2 ± 0.3 and 25.6 ± 0.3 for the ERBB3 siRNA, AKT siRNA, nonsilencing siRNA, saline and no treatment groups, respectively.

These results suggest that naked ERBB3 or AKT2 siRNA administered by tail vein injection is capable of specific growth suppression of established lung adenocarcinoma xenograft tumors.

A lower dose of the siRNAs, 2 μg/g/2 days per week for 3 weeks, was tested in Experiment 5. Median tumor growth slopes were 1.09 (ERBB3 siRNA), 1.79 (AKT2 siRNA), 6.57 (nonsilencing siRNA), 7.10 (saline) and 4.60 (untreated; overall p = 0.020). There were also significant differences in median tumor volume on day 18 (Table 2).

To investigate persistence of effect, in Experiment 4, mice were observed after the end of treatment until termination due to tumor burden. Tumor growth rate was used as the indicator because of differences in the survival times for individual mice. Although tumor growth rate increased in all groups during the post-treatment terminal interval (Table 1), it was lower in the ERBB3 and AKT2 siRNA groups, by about 40%, compared to the controls. The difference was significant for the whole dataset and for comparison with the untreated mice. Median survival times also differed (Table 1, legend). The results indicate a long-lasting effect of the siRNAs after termination of treatment, though a larger experiment would be required to be certain that the ERBB3 and AKT2 siRNAs were having a larger persistent effect compared to the nonsilencing siRNA.

ERBB3 or AKT2 siRNAs specifically reduced their respective mRNAs and proteins in the tumors

To validate that the ERBB3 or AKT2 siRNA-mediated suppression of tumor growth was directly associated with their respective gene silencing effects, we quantified ERBB3 and AKT2 mRNA levels in lysates of xenograft tumors, collected 2 hr after the last siRNA treatment. Levels of ERBB3 and AKT2 mRNA were normalized to the housekeeping gene peptidylprolyl isomerase (Table 3). Representative Western immunoblots are shown in Figures 2a and 3a, with protein levels normalized to actin given in Table 3.

Figure 2.

(a) ERBB3 protein was quantified relative to actin protein by Western immunoblot followed by densitometric scanning. Tumors were from Experiment 3 after 3 weeks of treatment. Representative blots are shown. There are evident reductions in ERBB3 protein by ERBB3 siRNA. Quantitative results are shown in Table 3. (b) Correlations between ERBB3 protein and ERBB3 mRNA (left panel) and tumor volume (right panel). Both correlations are highly significant. (c) Immunohistochemical staining for phospho ERBB3 in representative tumors (×200). Strong membrane staining in a tumor from a mouse treated with nonsilencing siRNA was greatly reduced in a tumor from a mouse treated with ERBB3 siRNA (left). Lack of signal with nonimmune serum confirmed specificity (right).

Figure 3.

(a) AKT2 protein was quantified relative to actin protein by Western immunoblot followed by densitometric scanning. Tumors were from Experiment 3 after 3 weeks of treatment. Representative blots are shown. There are evident reductions in AKT2 protein by AKT2 siRNA. Quantitative results are shown in Table 3. (b) Correlations between AKT2 protein and AKT2 mRNA (left panel) and tumor volume (right panel). Both correlations are highly significant. (c) Immunohistochemical staining for AKT2 protein in representative tumors (×200). Strong cytoplasmic staining in a tumor from a mouse treated with nonsilencing siRNA was greatly reduced in a tumor from a mouse treated with AKT2 siRNA (left). Lack of signal with nonimmune serum confirmed specificity (right).

Table 3. ERBB3 and AKT2 mRNAs and proteins in xenograft tumors from Experiment 3
inline image

Average ERBB3 mRNA levels in tumors from mice treated with the highest dose of ERBB3 siRNA were significantly downregulated compared to combined control tumors (Table 3). The nonsilencing and saline controls did not differ from the no-treatment controls. Average AKT2 mRNA levels were significantly downregulated by AKT2 siRNA treatment, compared to combined controls. These results confirm that the siRNAs were acting by their expected mechanism, i.e., enhanced degradation of their corresponding mRNAs.

At the lower but still effective dose of 2 μg/g/2 days per week (Table 2), normalized levels of ERBB3 mRNA were 0.384 ± 0.060 (N = 6), 0.735 ± 0.097 (N = 6), 0.724 ± 0.060 (N = 6) and 0.720 ± 0.105 (N = 4), for ERBB3 siRNA, nonsilencing siRNA, saline and no treatment, respectively, (p = 0.012). The average for all controls was 0.727 ± 0.047, a 53% difference, p = 0.0007 vs. ERBB3 siRNA. A dose of 1 μg/g/2 days per week did not affect tumor growth, and ERBB3 mRNA was not reduced in thirteen tumors tested (not shown). AKT2 mRNA levels for the lower doses were not determined.

Results at the protein level were confirmatory (Table 3). ERBB3 protein was reduced on average by 41% by ERBB3 siRNA compared to combined controls, a highly significant difference. AKT2 protein was reduced by an average 72% vs. combined controls, again highly significant. Furthermore, ERBB3 and AKT2 proteins correlated significantly with their respective mRNA values and with tumor volumes (Figs. 2b and 3b). ERBB3 and AKT2 proteins localized in the tumors by immunohistochemistry showed the expected localizations, cell membrane for phospho-ERBB3 and cytoplasm for AKT2 (Figs. 2c and 3c). Cellular staining for ERBB3 or AKT2 was reduced after specific siRNA treatment.

As an additional indicator of the specificity of the siRNAs, ERBB3 siRNA had no effect on the normalized levels of AKT2 protein (Table 3). In a separate experiment, utilizing vinculin for normalization, AKT2 siRNA had no effect on normalized levels of ERBB3 protein in five tumors (not shown).

Serum cytokine levels were not altered by siRNA treatments

A remaining possibility was that the specific siRNAs stimulated natural or B-cell related anti-tumor defenses.22, 23 Sera from two experiments were assayed for a panel of cytokines including IL-1β, IL-12p70, IFNγ, IL-6, KC, IL-10, TNFα and IFNα. These cytokines were detected in some or all of the mice (Supporting Information Table 1). There were no significant differences among the groups with regard to percent positive mice or average cytokine levels.

In summary, reduction in tumor volume correlated with specific mRNA and protein knock down, and there was no cytokine activation in the serum of these siRNA-treated xenografted mice. These results provide evidence that the ERBB3 siRNA and AKT2 siRNA antitumoral effects which we observed in this lung adenocarcinoma model are specific and are mediated through RNA interference.


In confirmation of our findings with cultured A549 lung adenocarcinoma cells,15 we have demonstrated that siRNAs to ERBB3 or its downstream signaling partner AKT2 markedly suppress the growth of A549 xenograft tumors in vivo. These siRNAs were administered in unformulated naked form by tail vein injection under normal pressure. Highly significant effects were seen in five independent experiments, did not involve weight loss or cytokine changes and were associated with specific downregulation in the mRNAs and proteins of the molecular targets. It is a firm result. These may be the first demonstrations of in vivo anti-tumor effects of ERBB3 or AKT siRNAs administered systemically.

The cellular mechanisms involved in these in vivo effects require further study. In A549 cells in culture, the suppressive effects of ERBB3 and AKT2 siRNAs involved apoptosis, necrosis and inhibition of the cell cycle.15 Both areas of necrosis and apoptotic cells were observed in the siRNA-treated tumors (data not shown). Quantitative comparisons of these parameters, between siRNA-treated and control tumors, will best be undertaken early in the tumor suppression process.

Previously, there was a belief that naked siRNA would be too unstable and too poorly taken up by cells to be effective in vivo.24 However, a recent detailed study by Morin et al.,18 involving quantitative RT-PCR detection of an siRNA administered i.p. in saline, showed that this assumption is not correct. Both guide and passenger strands were quantified. The siRNA indeed was measurable in tissue cells of five organs 20 min after treatment, and at levels sufficient to lead to target gene mRNA degradation. The mechanisms for greater uptake of siRNA into tissue cells in vivo, compared to cultured cells, are not known18 but could involve serum proteins or lipids or components of the complex extracellular matrices in tissues. For ERBB3 or AKT2 siRNAs, lengthy stability in the blood would not be required, as even brief downregulation of these anti-apoptotic proteins could set in motion pathways to cell death.18

Use of naked siRNAs avoids many issues of side-effects and toxicities associated with formulation reagents and could facilitate specific therapeutic molecular targeting, currently regarded as a highly promising approach. Indeed naked siRNA has been demonstrated to be effective in vivo in a number of contexts.18 Five types of cancer cells were suppressed in vivo by naked siRNA treatment,18 including H358 non-small cell lung cancer cells after intraperitoneal treatment with siRNA to netrin, a ligand for proapoptotic receptors.19 We have now shown that another human lung adenocarcinoma cell line, A549, and additional molecular anti-apoptotic regulators, ERBB3 and AKT2, also can be targeted in vivo by host treatment with naked siRNAs.

ERBB3 is a particularly attractive molecular target for lung cancer, because it has been among five25 or four26 genes whose expression predicts survival of patients with non-small cell lung cancer. Increased ERBB3 expression was found in lung cancer metastases to brain.27 Systematic efforts have correspondingly been underway to target ERBB3 for therapy,28, 29 leading to development of a monoclonal antibody to human ERBB3, MM121, which was effective against ligand-dependent cancer cells, including A549.28, 29 Pertuzumab, an antibody that binds to the dimerization arm of the ERBB2 receptor, was demonstrated to act through specific blockage of the heregulin-induced ERBB2/ERBB3 heterodimer.30 MM121 and Pertuzumab were not effective for ligand-independent activation of ERBB3. Use of ERBB3 siRNA could circumvent this limitation.

In addition to its role in lung cancer, ERBB3 is the key activator of the PI3K/AKT pathway in EGFR-driven pancreatic cancer,31, 32 a particularly deadly disease, as well as in ERBB2-driven cancers of a variety of types, including breast, ovary, melanoma, kidney, prostate and colon.28, 33 Thus, there could be wide applicability of therapy based on ERBB3 or AKT2 siRNAs.


The authors thank Donna Butcher for immunohistochemistry, Helen Rager for the cytokine assays, Dr. Bhal Diwan for photomicrography, and Dr. Natasha Caplen for helpful discussion and advice.