Department of Medicine, Albert Einstein College of Medicine, Bronx, NY
Department of Pathology, Marion Bessin Liver Research Center, Diabetes Center, Cancer Center, Ruth L. and David S. Gottesman Institute for Stem Cell and Regenerative Medicine Research, and Institute for Clinical and Translational Research, Albert Einstein College of Medicine, Bronx, NY
Albert Einstein College of Medicine, 1300 Morris Park Ave., Ullmann Bldg., Rm. 625, Bronx, NY 10461
Potential conflict of interest: Nothing to report.
Supported in part by NIH grants R01 DK071111, R01 DK088561, and P30-DK41296. R.B. received a postdoctoral fellowship from German Research Foundation (DFG).
To optimize strategies for liver-directed cell therapy, prevention of initial transplanted cell losses is particularly important for subsequent liver repopulation. After cell transplantation in hepatic sinusoids, perturbations in hepatic microcirculation along with changes in various liver cell types are among the earliest changes. Therefore, for advancing further concepts in cell engraftment we studied vascular and related events in the liver after transplanting syngeneic hepatocytes into dipeptidyl peptidase IV-deficient rats. We treated rats with vascular drugs to define whether deleterious cell transplantation-induced events could be controlled followed by improvements in transplanted cell engraftment and proliferation. We found cell transplantation altered liver gene expression related to vessel tone, inflammation, cell adhesion, thrombosis, or tissue damage/remodeling. This was due to hepatic ischemia, endothelial injury, and activation of neutrophils, Kupffer cells, and hepatic stellate cells. Treatment of rats before cell transplantation with the angiotensin converting enzyme blocker, lisinopril, or angiotensin II receptor blocker, losartan, did not improve cell engraftment. By contrast, direct-acting nitroglycerine or prostacyclin improved cell engraftment and also kinetics of liver repopulation. These drugs lowered hepatic ischemia and inflammation, whereas pretreatment of rats with the dual endothelin-1 receptor blocker, bosentan, improved cell engraftment independently of hepatic ischemia or inflammation, without improving liver repopulation. However, incubation of hepatocytes with bosentan protected cells from cytokine toxicity in vitro and produced superior cell engraftment and proliferation in vivo. Conclusion: Cell transplantation-induced changes in hepatic microcirculation contributed to transplanted cell clearances from liver. Vascular drugs, such as nitroglycerine, prostacyclin, and bosentan, offer opportunities for improving cell therapy results through superior cell engraftment and liver repopulation. Ongoing clinical use of these drugs will permit rapid translation of the findings in people. (HEPATOLOGY 2013)
Transplanting cells into liver sinusoids is the best way to initiate liver repopulation for cell therapy.1, 2 However, 80%-90% of transplanted cells are cleared within 1 or 2 days.2 Transplanted cells serve as emboli in sinusoids with hepatic ischemia, injury, and inflammation.3-6 The role of vascular regulators in these processes has not been defined. This should be significant for interventions to prevent initial loss of transplanted cells.
Homeostatic mechanisms regulating hepatic microcirculation are complex,7 including vasoconstrictors, e.g., angiotensin (AGT), endothelin (EDN), norepinephrine, etc., and vasodilators, e.g., nitric oxide (NO), carbon monoxide, prostacyclin (PGI2), etc. Hepatic sinusoidal vasodilatation by nitroglycerine (NTG), an NO donor, or phentolamine, an α-adrenergic blocker, improved cell engraftment,8 suggesting the possibility of pharmacological manipulations for cell therapy. Further benefits could result from the simultaneous decrease by vascular drugs in release of inflammatory cytokines/chemokines or increase in release of beneficial substances. The latter will be similar to the role of the cyclooxygenase-blocker, naproxen,9 which improved cell engraftment via vascular endothelial growth factor (VEGF) release from hepatic stellate cells (HSCs). Longer-acting vascular drugs are of particular interest because short-acting drugs, such as NTG, did not prevent rebound ischemia and delayed transplanted cell clearance.8
Here we characterized vascular gene expression and associated changes in liver cell types, followed by studies with drugs directed at vessel tone modulators, i.e., AGT, EDN1, NO, and PGI2, which affect liver sinusoidal endothelial cells (LSECs), HSCs, and other cells.10-16 This allowed analysis of the role of vascular mechanisms in cell engraftment. The studies were facilitated by dipeptidyl peptidase IV-deficient (DPPIV−) F344 rats, because these provide convenient methods for identifying DPPIV+ transplanted cells. Also, liver repopulation is readily studied in DPPIV− rats preconditioned with the DNA-damaging alkaloid, retrorsine, plus partial hepatectomy (PH).1-5 The findings provided new insights into the potential of vascular drugs for cell transplantation.
DPPIV− F344 rats, 6-8 weeks old, were from the Special Animal Core of Marion Bessin Liver Research Center. F344 rats were from the National Cancer Institute (Bethesda, MD). The Animal Care and Use Committee at Albert Einstein College of Medicine approved the protocols, according to institutional and National Institutes of Health guidelines.
Drugs and Chemicals.
We purchased lisinopril (LIS) (Sigma Chemical, St. Louis, MO), losartan (LOS) (Fluka Chemical, Ronkonkoma, NY), NTG (American Regent Laboratories, Shirley, NY), and PGI2 (Sigma). Bosentan (BOS) was from Actelion Pharmaceuticals (Allschwil, Switzerland). BOS monohydrate (free base) was administered according to the manufacturer as microsuspension in 5% gum arabicum (Fluka). LIS, LOS, NTG, and sodium BOS were dissolved in normal saline. PGI2 was dissolved in Tris-buffered saline, pH 9.0. All reagents and chemicals were from Sigma.
Rat hepatocytes were isolated by 2-step collagenase perfusion.5 One × 105 cells were cultured for 16-48 hours in 35 mm rat tail-collagen-coated dishes in Dulbecco's Modified Eagle's Medium with 10% fetal bovine serum (FBS) and antibiotics (DMEM). Culture conditions included normoxia (21% O2, 5% CO2) and hypoxia (5% O2, 5% CO2, 11% N2). CFSC-8B clone of rat HSC was cultured in DMEM for up to 3 days.17 Conditioned medium from CFSC-8B cells was obtained in serum-free DMEM after 16-18 hours. For tumor necrosis factor (TNF)-α cytotoxicity, hepatocytes were incubated for 2 hours in 200 ng/mL actinomycin D (Sigma) followed by 10 ng/mL TNF-α (Serologicals, Atlanta, GA) for 24 hours. For pro-oxidant cytotoxicity, hepatocytes were cultured for 16-18 hours and incubated in 2 mM H2O2 (Sigma) for 90 minutes. Cell viability was analyzed by thiazolyl blue (MTT) (Sigma).18
DPPIV− rats were anesthetized with ketamine/xylazine. Two × 107 F344 rat hepatocytes were injected intrasplenically. For liver repopulation, DPPIV− rats were given 30 mg/kg retrorsine at 6 and 8 weeks of age, followed 4 weeks later by two-thirds PH, and intrasplenic injection of 5 × 106 hepatocytes.
Blood Pool Analysis.
Red blood cells (RBCs) from F344 rats were labeled with 99mTechnetium (Tc) (UltraTag RBC Kit, Covidien, St. Louis, MO) as described.19 Radiolabeled RBCs were injected intrasplenically. Regions of interest were drawn for 99mTc-RBC activity before and after drugs.
Serum Hyaluronic Acid (HA).
Serum stored at −20°C was assayed by commercial kit (Corgenix, Westminster, CO).20
VEGF was measured by commercial kit (Rat VEGF Quantikine ELISA, RRV00; R&D Systems, Minneapolis, MN).9
Total liver RNAs were extracted by Trizol Reagent (Invitrogen, Carlsbad, CA). Complementary DNA (cDNA) was prepared (Omniscript RT PCR Kit; Qiagen, Valencia, CA) for reverse-transcription polymerase chain reactions (RT-PCR) (Platinum PCR Kit; Invitrogen). PCR primers were for rat AT1A receptor,21 ETA and ETB receptors,22 and β-actin.23 Products were resolved in 1% agarose.
For quantitative RT-PCR, RNAs were reverse transcribed by RT2 PCR Array First Strand Kit (SABiosciences, Frederick, MD), followed by PCR with RT2 Real-Time SYBR Green PCR master mix (SABiosciences). Expression of 84 genes each was examined by Endothelial Cell Biology PCR Array (PARN-015; http://www.sabiosciences.com/rt_pcr_product/HTML/PARN-015A.html) or Signal Transduction Pathway Finder Array (PARN-014; http://www.sabiosciences.com/rt_pcr_product/HTML/PARN-014A.html). Each analysis used three replicate samples. The 2ˆΔΔ cycle threshold method was used to first normalize for β-actin expression, as described.6 Gene expression differences of 2-fold or more were considered significant.
Cryostat sections of 5 μm were prepared. Sections were fixed in chloroform-acetone (1:1, vol/vol) for DPPIV histochemistry. Combined staining for DPPIV and glycogen (Gly) or DPPIV and glucose-6-phosphatase (G6P) was as described.1 For myeloperoxidase (MPO), sections were fixed in 4% paraformaldehyde and 85% ethanol, and stained with peroxidase indicator reagent (Sigma).6 For γ-glutamyl transpeptidase (GGT), sections were fixed in chloroform-acetone and stained as described.3 Desmin was stained with anti-desmin IgG (DAKO Cytomation, Carpinteria, CA).24 For Kupffer cell (KC) activity, 0.1 mL carbon in India ink (Pelican number 17, Hanover, Germany), was given as a 30-minute pulse intrasplenically 6 hours after cell transplantation.5 Ki67 staining used a commercial antibody (Novocastra, Newcastle-upon-Tyne, UK).25
MPO+ neutrophils and carbon-containing KC were counted under 200× magnification. HSC were counted under 400× magnification. GGT+ areas were measured under 100× magnification. DPPIV+ cells were counted in sections under 100× magnification. At least 50-100 areas centered on portal radicles were studied per tissue.5, 6 For liver repopulation, the size of DPPIV+ transplanted cell clusters was measured in consecutive microscopic images.5 Each analysis included three to four animals.
Data are shown as mean ± standard deviation (SD). Differences were analyzed by t tests, Mann-Whitney rank sum tests, or analysis of variance (ANOVA) with Holm-Sidak pairwise comparisons. P < 0.05 was considered significant.
Cell Transplantation Caused Vascular Changes.
First, we analyzed cell transplantation-induced changes in liver (n = 30) (Fig. 1A). After 1 to 6 hours, most portal radicles contained transplanted cells, range 52% to 63%, with 7 to 8 transplanted cells adjacent to each portal radicle in liver lobules (Fig. 1B). Most of these transplanted cells were lost, because after 1, 3, and 7 days transplanted cells were in only 16 ± 11%, 3 ± 0.6% and 4 ± 2% portal radicles, P < 0.05, with their numbers declining to 1.5 ± 0.9%, 1.5 ± 0.9%, and 1 ± 0.4% adjacent to each portal area, respectively, P < 0.05. This accompanied tissue changes (discussed below). Of 84 genes in endothelial cell biology, vascular, or inflammatory pathways, 31 (37%), 9 (11%), and 3 (4%) were up-regulated 6 hours, 1 day, and 3 days after cell transplantation, respectively. Therefore, cell transplantation induced rapid but transient changes in gene expression. The 31 genes expressed differentially at the 6 hours timepoint were up-regulated by 2- to >220-fold (Fig. 1C). These included genes involved in regulation of vessel tone (Ace, Agt, Blr1, Edn1, Nos2, Ptgis), inflammation (Ccl2, Csf2, Cxcl1, -2, -4, Il1b,-6, -7, Tnf-α, Tnip2), cell adhesion (Icam1, Rhob, Sell, Selp, Vcam1), thrombosis (Plat, Plau, Serpine1, Thbd, Thbs1), and tissue injury/remodeling (Cflar, Mmp9, Pgf, Timp1). Genes in renin-angiotensin system (Ace, Agt), and direct-acting vascular regulators, i.e., NO, PGI2, and Edn1, were perturbed. Because Ace, Agt, and Edn1 can be manipulated by long-acting drugs, we analyzed receptor messenger RNA (mRNA) expression. We found greater expression of EdnRA and EdnRB receptors but not of Agt receptor (Fig. 1D). We were unable to verify endothelin receptor expression at the protein level by western or immunostaining because several commercial antibodies gave nonspecific results (not shown).
Vascular Drugs and Cell Engraftment.
We studied the benefits of drugs on cell engraftment 3 days after cell transplantation (n = 30). Suitable doses were identified by dose-ranging studies. LIS and LOS were injected intraperitoneally 2 hours before cell transplantation. Some animals (n = 6) received LIS and LOS daily for 3 days before and for 2 days after cells. NTG and PGI2 were infused into splenic pulp starting 5 minutes before and ending 10 minutes after cells. BOS was given by gavage daily for 3 days, 2 hours before and for 2 days after cells, totaling six doses, according to manufacturer (Fig. 2A). NTG served as control because it previously increased cell engraftment.8 After 0.5 μg/kg/min NTG, cell engraftment improved by 1.8 ± 0.5-fold, P < 0.05 (Fig. 2B). LIS and LOS in single 1 to 10 mg/kg doses improved cell engraftment by only 1.4-fold, P > 0.5, although portal vein radicles with transplanted cells increased >2-fold, P < 0.05. This reflected greater effects of LIS and LOS on presinusoidal vessels. Despite six doses of 10 mg/kg LIS or LOS, three doses before, and three doses after cell transplantation, cell engraftment was unchanged (not shown). By contrast, 1 μg/kg/min PGI2 increased cell engraftment by 1.9 ± 0.2-fold, P < 0.05. We found a single dose of 100 mg/kg BOS 18 hours before cell transplantation did not improve cell engraftment (not shown). However, 100 mg/kg BOS daily for 3 days before and for 3 days after cell transplantation increased cell engraftment by 2.5 ± 0.3-fold, P < 0.05. Costaining for DPPIV and Gly or G6P confirmed hepatic functions in transplanted cells.
To determine whether sinusoidal dilatation contributed in improved cell engraftment, we measured hepatic blood pools with 99mTc-labeled RBCs (n = 50). After 0.5 μg/kg/min NTG, 1 μg/kg/min PGI2, or six doses of 100 mg/kg BOS, blood pools increased 11 ± 0.4%, 8 ± 3%, and 18 ± 6% above basal values, respectively, P < 0.05, Mann-Whitney tests. However, LIS or LOS did not increase blood pools; therefore, we stopped further study of these two drugs.
Effects of Drugs on Cells.
Of 31 genes up-regulated 6 hours after cell transplantation alone, cell transplantation in rats pretreated with NTG, PGI2, or BOS showed up-regulation of fewer genes, 1, 3, and 10 genes, respectively (Fig. 3A). The remaining genes were expressed at lower levels (Fig. 3B). Although all drugs decreased changes in genes regulating vessel tone, cell damage, adhesion, or thrombosis, PGI2 and BOS did not affect Ace expression and BOS did not affect Edn1 expression. Among thrombosis genes, NTG and BOS lowered Plau and Thbd expression less, and BOS did not lower Serpine1 expression. Therefore, NTG and PGI2 were most effective in controlling gene expression related to ischemia or inflammation, whereas BOS had fewer effects on gene expression.
Changes in gene expression were mirrored in tissues, including GGT expression, which represents liver ischemia in the cell transplantation setting.3, 8 In normal rats, only bile duct cells expressed GGT, whereas 6 hours after cells, many hepatocytes expressed GGT (Fig. 4A). After NTG or PGI2 plus cells, GGT expression was 3- to 6-fold lower, P < 0.05. Remarkably, BOS had no effect on cell transplantation-induced GGT expression. Onset of endothelial injury was shown by increases in serum HA levels,20, 25 which 6 hours after cells was 66 ± 4 ng/mL versus 43 ± 4 ng/mL in normal controls, P < 0.05 (n = 3 each). After NTG plus cells, HA levels were lower, 30 ± 21 ng/mL, P < 0.05. However, in recipients of PGI2 or BOS, HA levels were above normal levels, 50 ± 0.2 ng/mL and 62 ± 13 ng/mL, respectively. This was similar to no decreases after BOS in expression of ischemia-related genes or GGT.
Hepatic inflammation was evident with more carbon-containing KC and MPO+ neutrophils 6 hours after cells (Fig. 4B,C). No drug decreased numbers of MPO+ neutrophils. After NTG or PGI2, KC activation decreased by 37 ± 5% or 48 ± 2% versus controls, respectively, P < 0.05. BOS did not decrease KC activation. As noted above, inflammation genes, including monocyte-related cytokines/chemokines, were expressed less after NTG or PGI2, but were mostly unchanged after BOS. Cell transplantation activated HSC with 17 ± 2 desmin+ HSC per high-powered field (HPF) 6 hours after cell transplantation (Fig. 4D), which was greater than normal rats, P < 0.05. After PGI2 plus cells, desmin+ HSC numbers were similar to after cells alone, 15 ± 4 cells per HPF, P > 0.5. However, numbers of desmin+ HSC increased after NTG or BOS, 38 ± 5 and 32 ± 0.2 cells per HPF, respectively, P < 0.05. Previously, cell transplantation increased hepatic expression of cytoprotective HGF, VEGF, etc.2, 23 Similarly, after cell transplantation desmin expression in HSC was associated with VEGF expression.9
To determine the significance of BOS-induced desmin expression in HSC, we studied cultured CFSC-8 cells. CFSC-8B cells released VEGF after 16-18 hours under hypoxia, 44 ± 10 pg/mL versus 22 ± 4 pg/mL under normoxia, P < 0.05. After culture with BOS, CFSC-8 cells released more VEGF (Fig. 5A). A high dose of BOS (1 mM) was cytotoxic to CFSC-8B cells, similar to studies with rat hepatocytes and pancreatic stellate cells.27, 28 Conditioned medium from CFSC-8 cells containing VEGF plus BOS, protected hepatocytes from TNF-α- or H2O2-induced cytotoxicity (Fig. 5B,C). Therefore, BOS-induced VEGF secretion from HSC could have improved cell engraftment.
Vascular Drugs and Liver Repopulation.
For liver repopulation, we used retrorsine/PH preconditioned rats (n = 30) in groups (n = 6-8). Untreated control rats (n = 3-6) were included (Fig. 6A). In controls, 7 ± 2% liver was repopulated over 3 weeks (Fig. 6B,C). After NTG or PGI2, liver was repopulated more, 24 ± 7% and 19 ± 3%, respectively, P < 0.05. Surprisingly, BOS did not improve liver repopulation, which was similar to control rats, 8 ± 2%. To exclude whether BOS interfered with retrorsine/PH preconditioning, e.g., by antiproliferative effects, we gave rats six daily doses of 100 mg/kg BOS with PH on d5 (Fig. 6D). We found no changes in liver morphology or apoptosis rates. Over 30% hepatocytes expressed Ki67 at peak of liver regeneration 36 hours after PH,26 in controls as well as BOS-treated rats, P > 0.05, t test. This excluded possible interference by BOS in hepatic preconditioning.
Next, we determined whether BOS altered intracellular signaling to explain lack of transplanted cell proliferation. Cultured hepatocytes express high-affinity Edn receptors.29 Gene expression analysis with Signal Transduction Pathway Finder Array in hepatocytes cultured for 16-18 hours with or without 1 μM BOS showed BOS changed expression of only 2 of 84 genes: the cell cycle inhibitor, Cdkn2b, and nuclear factor kappa B (NF-κB) inhibitor, NF-κb1a, were up-regulated by 3.8- and 2.3-fold, respectively.
In intact animals, widespread occupancy of Edn receptors by BOS could have displaced Edn1 and increased its blood levels, as established in mouse studies.30 To determine whether Edn1 harmed cells after transplantation, we incubated hepatocytes with 1, 10, and 100 μM BOS for 1 hour in vitro, followed by transplantation for cell engraftment (n = 12) and proliferation (n = 24) in rats (Fig. 7A). BOS-incubated cells engrafted better; 2-fold above controls, P < 0.05 (Fig. 7B). In retrosine/PH-conditioned rats, BOS-incubated cells repopulated 22 ± 2% of liver versus vehicle-treated control cells or 8 ± 2% liver repopulation in rats treated with BOS in vivo, P < 0.05 (Fig. 7C). Moreover, BOS-treated hepatocytes resisted TNF-α-induced toxicity (Fig. 7D).
This study showed that cell transplantation rapidly altered expression of vascular, inflammatory, and other genes along with activation of neutrophils, KC, LSEC, and HSC. In the process, many transplanted cells were cleared from the liver, including due to vascular events, because NTG, PGI2, or BOS decreased the loss of transplanted hepatocytes. However, the mechanisms by which drugs benefited cell transplantation seemed complex, especially in the case of BOS versus NTG or PGI2. The ability of BOS to improve cell engraftment by simply incubating cells in vitro will be very helpful for cell therapy. Moreover, as vascular drugs did not normalize all cell transplantation-induced deleterious events, e.g., hepatic inflammation, additional mechanisms will be available for development of combination treatments.
The rapidity of transplanted cell clearance from sinusoids, release of inflammatory cytokines/chemokines, and tissue remodeling were similar to previous studies.2, 6, 9, 23 Changes after cell transplantation, such as hepatic ischemia or endothelial injury, were undesirable.2-6, 9, 19, 23, 24 As NTG and PGI2 blocked ischemia (GGT expression) and alterations in gene expression, this suggested NO and other vasoactive molecules were likely involved.12, 13 On the other hand, BOS studies established that Edn1 also played roles in clearance of transplanted cells, which was previously unknown. Hepatic sinusoidal dilatation by NTG, PGI2, or BOS, as indicated by blood pool studies, likely contributed to cell engraftment. LIS and LOS did not increase hepatic blood pools and did not increase entry of transplanted cells in hepatic sinusoids, which should exclude targeting of renin-angiotensin system for cell transplantation strategies. However, mechanisms other than sinusoidal dilatation were clearly involved in how NTG, PGI2, and BOS improved cell engraftment.
After NTG, PGI2, or BOS, cell transplantation-induced hepatic inflammation was not abrogated. Moreover, BOS did not decrease GGT expression, endothelial damage, or activation of neutrophils and KC. Therefore, it should be reasonable to conclude that improved cell engraftment after these drugs was due to mechanisms other than these. In this respect, NTG, as well as BOS, increased the numbers of desmin+ HSC over the entire duration of the studies. Previously, naproxen activated desmin expression in HSC, along with VEGF expression, leading to greater cell engraftment.9 We found that BOS too induced VEGF release in cultured rat HSC. This suggested superior cell engraftment after BOS likely included potential contributions from activated HSC.
Our studies were performed with a regimen of BOS effective in rats because it is noteworthy that excellent Edn receptor-blockade requires adjustment in BOS dosing.31 Although BOS may alter bile salt handling,32 we did not observe BOS toxicity in animals.
The extent of liver repopulation with NTG and PGI2 was in agreement with their hepatic proliferation-promoting effects.12, 33 However, we were surprised that BOS did not increase liver repopulation. The specificity of BOS for EdnRA and EdnRB receptors is well established. In intact animals, plasma Edn1 levels rose in EdnRB-deficient mice.30 Consistent with this mechanism, transplanted cells should have been exposed to greater Edn1 levels in BOS-treated rats, which may have produced unknown effects. More Edn1 might be synthesized after BOS-induced release of inducers, e.g., interleukin (IL)-6 or TNF-α.16, 34 These cytokines were expressed more in BOS-treated rats following cell transplantation. Continued administration of BOS to animals after cell transplantation might have produced other changes. Greater Cdkn2b expression in BOS-treated hepatocytes should be noteworthy because this cell cycle suppressor could have inhibited transplanted cell proliferation in BOS recipients.
Transplantation studies with BOS-incubated hepatocytes indicated that EdnRA and EdnRB receptor blockade, which would have abrogated binding of Edn1, increased the liver repopulation ability of cells. Interestingly, Edn1 may regulate cell proliferation,35 including through complex intracellular signaling mechanisms,36 but little is known about the effects of Edn1 on hepatocytes. In renal tubular cells, Edn1 induced NF-κB, which activates cell injury pathways.37 In smooth muscle cells, Edn1 induced TGF-β signaling, which inhibits hepatocyte proliferation.38 Perhaps up-regulation of NF-κB1a in BOS-treated hepatocytes in vitro suggested protective signaling to block NF-κB, as shown in assays of TNF-α-mediated cytotoxicity.
These studies offer lessons for including vascular drugs in cell therapy strategies. Even the relatively small extent of improved cell engraftment with vascular drugs will be of clinical significance, e.g., for repeated cell transplants or for liver repopulation. Vascular drugs can be combined with other drugs or strategies. NTG or PGI2 are widely used in people. BOS and similar drugs are also in clinical use. Incubating cells with drugs in vitro before transplantation will be simple and should decrease the potential for systemic drug toxicity.
Ms. Chaoying Zhang and Ms. Gertrude Ukpong provided technical assistance.