Genetic polymorphisms in arginase I and II and childhood asthma and atopy

Abstract

Background

A recent microarray study implicated arginase I (ARG1) and arginase II (ARG2) in mouse allergic asthma models and human asthma.

Objectives

To examine the association between genetic variation in ARG1 and ARG2 and childhood asthma and atopy risk.

Methods

We enrolled 433 case-parent triads, consisting of asthmatics 4 to 17 years and their biologic parents, from the allergy clinic of a public hospital in Mexico City between 1998 and 2003. Atopy to 24 aeroallergens was determined by skin prick tests. We genotyped 4 single nucleotide polymorphisms (SNPs) of ARG1 and 4 SNPs of ARG2 with minor allele frequencies over 10% using the TaqMan assay.

Results

ARG1 SNPs and haplotypes were not associated with asthma but all four ARG1 SNPs were associated with the number of positive skin tests (P = 0.007 to 0.018). Carrying two copies of minor alleles for either of two highly associated ARG2 SNPs was associated with a statistically significant increased relative risk (RR) of asthma [1.5, 95% confidence interval (CI) 1.1–2.1 for arg2s1; RR = 1.6, 95% CI = 1.1–2.3 for arg2s2]. The association was slightly stronger among children with a smoking parent (arg2s1 RR = 2.1, 95% CI = 1.2 – 3.9 with a smoking parent; RR =1.2, 95% CI = 0.8–1.9 without, interaction P = 0.025). Haplotype analyses reduced the sample size but supported the single SNP results. One ARG2 SNP was related to the number of positive skin tests (P = 0.027).

Conclusions

Variation in arginase genes may contribute to asthma and atopy in children.

INTRODUCTION

Candidate genes for asthma have arisen from linkage studies as well as knowledge of mechanisms involved in the pathogenesis of asthma. Recently, microarray analyses have identified novel asthma candidate genes¹. In two murine asthma models, Zimmermann et al. identified increased expression of genes coding for two arginase isoforms, arginase I (ARG1), and arginase II (ARG2) in lung tissue¹. Further, ARG1 expression was increased in lung macrophages from peribronchial and perivascular inflammatory sites in affected mice and in bronchoalveolar lavage cells from asthmatic humans. ARG1 mRNA was strongly detected in human asthmatic lung biopsy tissue with predominant localization to submucusal inflammatory cells as well as epithelial cells.

Arginases may contribute to asthma pathogenesis through various mechanisms, including inhibition of nitric oxide (NO) generation, reduced arginine bioavailability, and increased ornithine production, which could lead to airway remodeling due to altered polyamine and proline synthesis^{2, 3}. Of note, ARG2 lies close to an asthma linkage region on chromosome 14q24⁴.

Based on the increasing evidence for a role of arginases in asthma pathogenesis ^{2, 5}, we investigated single nucleotide polymorphisms (SNPs) in ARG1 and ARG2 in relation to asthma risk in a case-parent triad study of childhood asthma in Mexico City.

METHODS

Study Design and Subject Enrollment

Using the case-parent triad design^{6, 7}, we recruited nuclear families consisting of a case and both parents. Cases were children aged 4–17 years with asthma, diagnosed by a pediatric allergist at the allergy referral clinic of a large public pediatric hospital in central Mexico City (Hospital Infantil de Mexico Federico Gomez), between June 1998 and November 2003. Children and parents provided blood samples as sources of DNA. We enrolled 433 complete case-parent triads.

A pediatric allergist based the diagnosis of asthma on clinical symptoms and response to treatment⁸. The severity of asthma among the cases was rated by a pediatric allergist according to symptoms in the Global Initiative on Asthma schema as mild (intermittent or persistent), moderate and severe⁹. Pulmonary function was measured using the EasyOne spirometer (ndd Medical Technologies, Andover, MA, USA) for 332 subjects according to ATS specifications¹⁰. The best test out of three technically acceptable tests was selected. Spirometric prediction equations from a Mexico City childhood population were used to calculate the percent predicted forced expiratory volume in one second (FEV₁)¹¹. Children were asked to hold asthma medications on the morning of the test.

Atopy was determined using skin prick test. The following battery of 24 aeroallergens common in Mexico City was used: Aspergillus fumigatus, Alternaria, mucor, Blattella germanica, Periplaneta americana, Penicillium, cat, dog, horse, Dermatophagoides (pteronyssinus and farina), Ambrosia, Artemisia ludoviciana, Cynodon dactylon, Chenopodium album, Quercus, Fraxinus, Helianthus, Ligustrum vulgare, Lolium perenne, Plantago lanceolata, Rumex crispus, Schinus molle, Salsola, and Phleum pratense. Histamine was used as positive control and glycerin as negative control. Children were considered atopic if the diameter of the skin reaction to at least one allergen exceeded 4 mm. The test was considered valid if the reaction to histamine was >= 6 mm according to the grading of skin prick test recommended by Aas and Berlin¹². Skin test data were available on 420 (of 433) triads with complete genotyping data.

The Institutional Review Boards of the Mexican National Institute of Public Health, the Hospital Infantil de Mexico Federico Gomez and the U.S. National Institute of Environmental Health Sciences approved the protocol. Parents provided consent for the child’s participation.

SNP Selection

To select SNPs in ARG1 and ARG2, we searched the dbSNP Variation Database from NCBI LocusLink (http://www.ncbi.nlm.nih.gov/SNP) and Applied Biosystem SNP genotyping products (http://www.appliedbiosystems.com/catalog). Because there is no functional data for arginase SNPs and because complete resequencing has not been done in our Mexican population, nor others, we selected SNPs based on adequate minor allele frequency (at least 10%) and spacing throughout the gene. We selected 4 SNPs spaced at 1–3kb, including 3kb upstream genomic region of gene ARG1 of which the genomic size is about 11kb. For ARG2, we selected 4 SNPs with 10–25kb interval including 10kb upstream and 5kb downstream of genomic region of the gene. The genomic size of ARG2 is about 32kb.

Genotyping of ARG1 and ARG2 SNPs

Peripheral blood lymphocytes were isolated, and DNA was extracted using Gentra Puregene kits (Gentra Systems, Minneapolis, MN). Primers and probes were purchased from Assay-by-Design and Assay-on-Demand (Applied Biosystems, Foster City, CA). All PCR amplifications were performed using the 5′-nuclease assay on GeneAmp PCR Systems 9700. Details of the genotyping methods are included in the Journal’s Online Repository.

We ascertained nonparentage with a set of short-tandem repeats (AmpFLSTR Profiler Plus; Applied Biosystems, Foster City, CA) analyzed using Pedcheck software¹³. The number of complete triads with data on all 4 SNPs was 412 for ARG1 and 405 for ARG2.

Statistical Analysis

Linkage disequilibrium (LD) between pairs of SNPs was measured by computing squared correlation coefficient (r²) and Lewontin’s standardized disequilibrium coefficient (D′)^{14, 15}.

To analyze associations with individual SNPs, we used a log-linear-based method implemented in SAS version 9.1 for Windows (SAS Institute, Cary, NC) and STATA version 8.0 (Stata Press, College Station, TX) to test for asymmetric distribution of a particular variant allele among affected offspring compared to Mendelian expectation based on their biologic parents⁶. The method uses genotypes of cases and their parents (triads) stratified into the 15 possible types of triads. Genetic effects produce frequency distortions within parental mating types that permit estimation of genotype relative risks (RR) and 95% confidence intervals (CI) through the log-linear model. A null hypothesis that a SNP is not associated or linked with the outcome is tested by means of the likelihood ratio test. We calculated relative risks for individual SNPs without restricting to a specific genetic model and also under the dominant and recessive genetic models. We also calculated likelihood ratio tests and relative risks for the joint effects of genotype and parent smoking, using the method of Umbach and Weinberg¹⁶.

We analyzed associations with haplotypes, as opposed to single SNPs, using HAPLIN Version 1.0 (http://www.uib.no/smis/gjessing/genetics/software/haplin) running under S-PLUS Version 6.2 for Windows (Seattle, Washington). HAPLIN estimates single- and double-dose effects of haplotypes using maximum likelihood¹⁷. HAPLIN is a direct extension of the log-linear models for a single diallelic locus⁶ although it currently requires complete triad data. The haplotypes of individuals with unknown phase are reconstructed from the triad family information whenever possible, and the remaining are estimated using the expectation-maximization algorithm. The increase in standard error due to the unknown phase is accounted for using a jackknifing procedure. For effects of one specific haplotype the reference group consists of individuals with all remaining haplotypes. The single-dose relative risk then measures the change in risk for an individual carrying exactly one copy of the specific haplotype, whereas the double-dose estimate is the risk for homozygotic individuals relative to the reference group. The rare haplotypes (below 5%) had to be excluded from the analyses because it is difficult to obtain valid relative risk estimates for these haplotypes. Thus, triads including individuals with one or more of these rare haplotypes were excluded, leading to smaller numbers of triads available for the full haplotype analyses than for the single SNP analyses. The HAPLIN analysis includes 360 complete triads for ARG1 and 310 complete triads for ARG2.

To examine whether arginase polymorphisms influenced the degree of atopy, as assessed by the number of positive skin tests out of 24 done, we tested for linkage and association with each of the four SNPs in ARG1 and ARG2 using the polytomous logistic method of Kistner and Weinberg¹⁸. P values were calculated from likelihood ratio tests. For readers more familiar with transmission disequilibrium test (TDT) based methods, we also calculated P values for the individual SNP analyses using FBAT¹⁹. As expected, P values were very close by the two methods.

RESULTS

Cases had mean age of 9.1 years (standard deviation 2.5 years) and 58% were male. Most had mild (69%) as opposed to moderate or severe asthma (31%). Nearly all cases (98.4%) had used medication for asthma in the past 12 months. Wheezing in the past 12 months was reported by 93% and chronic dry cough was reported by 62.5%. For 71.8% of cases, asthma symptoms had interfered with daily activities or school attendance in the past 12 months. Exacerbation of asthma symptoms by exercise was reported by 72.1% of cases. Among the 331 cases with spirometry data available, FEV₁ was > 80% predicted for 23.5%, 60–80% predicted for 72.3% and < 60% predicted for 4.2%. Atopy to aeroallergens, defined as at least one positive skin test, was present in 93% of cases and 55% of cases had 5 or more positive skin tests. The highest rates of skin test positivity were seen for cockroach (42.6%) and dust mite (71.9%). A smoking parent was present for 52.1% of cases; only 6% of mothers reported smoking during pregnancy.

For the 4 SNPs genotyped for each of ARG1 and ARG2, NCBI SNP ID, the SNP location in the gene, and minor allele frequency are presented in TABLE E1 in the Online Repository. Measures of linkage disequilibrium (D′ and r²) are shown in TABLE E2 in the Online Repository for ARG1 and TABLE E3 in the Online Repository for ARG2. For ARG1, arg1s1 and arg1s2 were highly associated (r² = 0.974). For ARG2, the arg2s1 and arg2s2 SNPs were highly associated (r² = 0.937).

The frequency distributions of the 15 possible mating types of triads for the polymorphisms of ARG1 and ARG2 are shown in TABLE I. The log-linear analysis of individual SNPs in ARG1 showed no association with asthma (TABLE II). For ARG2, carrying 2 copies of minor alleles, for either arg2s1 or the highly associated SNP arg2s2 was associated with a statistically significantly increased risk of asthma (RR = 1.5, 95% CI = 1.1–2.1 for arg2s1, P = 0.018; RR = 1.6, 95% CI = 1.1–2.3 for arg2s2, P = 0.007) (TABLE II). The other two ARG2 SNPs were not appreciably associated with asthma risk. The corresponding P values from TDT-based analysis were 0.020 for arg2s1 and 0.008 for arg2s2.

TABLE I.

Distribution of case-parent triad genotypes for ARG1 4 SNPs and ARG2 4 SNPs

	Triad Counts (ARG1)				Triad Counts (ARG2)
Mother-Father-Child*	s1	s2	s3	s4	s1	s2	s3	s4
000	27	28	30	70	65	71	79	206
010	29	28	28	45	34	39	39	35
011	21	22	29	43	28	31	41	36
021	21	20	31	25	19	15	20	12
100	32	33	38	29	37	41	34	42
101	29	29	29	36	36	36	33	34
110	24	24	30	11	23	22	21	5
111	49	48	50	47	42	40	52	15
112	32	34	27	21	27	28	18	3
121	26	26	19	15	12	10	8	2
122	29	28	22	9	18	17	16	0
201	30	30	20	29	27	23	21	13
211	22	23	20	15	7	6	9	1
212	20	20	23	8	20	18	7	1
222	21	19	16	9	10	8	7	0
Total	412	412	412	412	405	405	405	405

^*Number of copies of minor allele in mother-father-child

TABLE II.

Childhood asthma risk in relation to ARG1 and ARG2 SNPs.

ARG1				ARG2
SNPs	Model	Genotype	RR (95% CI)	SNPs	Model	Genotype	RR (95% CI)
s1				s1
	Unrestricted	AA	1.0		Unrestricted	GG	1.0
		AG	0.9 (0.7–1.2)			GC	0.9 (0.7–1.2)
		GG	1.1 (0.7–1.6)			CC	1.4 (0.9–2.0)
	Dominant	AG + GG	0.9 (0.7–1.2)		Dominant	GC + CC	0.9 (0.7–1.2)
	Recessive	AA + AG	1.0		Recessive	GG + GC	1.0
		GG	1.2 (0.9–1.5)			CC	1.5 (1.1–2.1) *
s2				s2
	Unrestricted	TT	1.0		Unrestricted	TT	1.0
		TA	0.9 (0.7–1.2)			TC	0.8 (0.6–1.1)
		AA	1.1 (0.8–1.6)			CC	1.4 (0.9–2.1)
	Dominant	TA + AA	1.0 (0.7–1.3)		Dominant	TC + CC	0.9 (0.7–1.2)
	Recessive	TT + TA	1.0		Recessive	TT + TC	1.0
		AA	1.2 (0.9–1.6)			CC	1.6 (1.1–2.3) †
s3				s3
	Unrestricted	CC	1.0		Unrestricted	GG	1.0
		CT	0.9 (0.7–1.1)			GA	1.1 (0.8–1.4)
		TT	1.0 (0.7–1.4)			AA	1.0 (0.6–1.5)
	Dominant	CT + TT	0.9 (0.7–1.1)		Dominant	GA + AA	1.0 (0.8–1.4)
	Recessive	CC + CT	1.0		Recessive	GG + GA	1.0
		TT	1.1 (0.8–1.5)			AA	0.9 (0.6–1.4)
s4				s4
	Unrestricted	TT	1.0		Unrestricted	TT	1.0
		TC	1.3 (1.0–1.7)			TC	1.0 (0.7–1.3)
		CC	1.0 (0.6–1.6)			CC	0.5 (0.2–1.4)
	Dominant	TC + CC	1.2 (0.9–1.6)		Dominant	TC + CC	0.9 (0.7–1.3)
	Recessive	TT + TC	1.0		Recessive	TT + TC	1.0
		CC	0.8 (0.6–1.2)			CC	0.5 (0.2–1.4)

CI, confidence interval

RR, relative risk

^*P = 0.018

^†P = 0.007

Sample sizes are reduced for analyses of haplotypes. We did not observe any significant associations for ARG1 (TABLE III). For ARG2, the haplotype results reflected the single SNP finding in magnitude and direction, although given the reduced sample size the results were not statistically significant. The relative risk was 1.40 (95% CI = 0.84–2.30) for individuals carrying two copies of the haplotype containing the minor alleles for the arg2s1 and arg2s2 with the major alleles at the other two SNPs.

TABLE III.

Haplotype relative risks for childhood asthma in relation to ARG1 and ARG2.

		Relative Risk (95% CI)
Haplotype*	Frequency (95% CI)	Single Copy	Double Copy
ARG1
ATTT	0.46 (0.43–0.50)	0.87 (0.63–1.20)	0.85 (0.54–1.30)
GACC	0.29 (0.26–0.33)	1.00 (0.73–1.40)	1.00 (0.61–1.80)
GACT	0.20 (0.17–0.23)	1.10 (0.82–1.50)	0.60 (0.27–1.30)
ATCC	0.05 (0.04–0.07)	1.00 (0.61–1.60)	1.90 (0.41–8.80)
ARG2
GTAT	0.37 (0.34–0.41)	1.10 (0.81–1.60)	1.10 (0.65–1.80)
CCGT	0.31 (0.28–0.35)	1.00 (0.76–1.50)	1.40 (0.84–2.30)
GTGT	0.17 (0.15–0.21)	0.82 (0.58–1.20)	1.00 (0.48–2.20)
GTGC	0.14 (0.12–0.17)	1.00 (0.72–1.50)	0.34 (0.08–1.50)

CI, confidence interval

^*Haplotypes formed by arg1s1, arg1s2, arg1s3, arg1s4 in order for ARG1; by arg2s1, arg2s2, arg2s3, arg2s4 in order for ARG2.

The association between carrying two copies of the minor allele for arg2s1 or arg2s2 was slightly stronger among moderate to severe asthmatics (RR = 2.1, 95% CI = 1.2 – 3.9 for arg2s1, RR = 2.0, 95% CI = 1.1 – 3.6 for arg2s2) than among mild asthmatics (RR = 1.2, 95% CI = 0.8–1.9 for arg2s1, RR = 1.4, 95% CI = 0.9 – 2.2 for arg2s2).

Because parental smoking is a consistent risk factor for childhood asthma, we examined whether the associations differed by whether the case had a smoking parent. The increased risk for carrying two copies of the minor allele for arg2s2 SNP appeared to be limited to cases with smoking parents (RR = 2.39, 95% CI = 1.48–3.86 with smoking parents, RR = 1.06, 95% CI = 0.62–1.81 with nonsmoking parents, P value for likelihood ratio test of gene by environment interaction = 0.025 under recessive model). A similar result was obtained for the very highly associated arg2s1 SNP.

We examined the relationship between individual SNPs of ARG1 and ARG2 and the degree of atopy to aeroallergens, assessed by the number of positive skin tests out of a battery of 24 tests. To select the most appropriate genetic model for analyses of these data, we examined the mean number of positive skin tests by number of copies of the variant alleles for the four ARG1 SNPs (TABLE IV). For all four ARG1 SNPs, the number of positive tests varied significantly by genotype. The change in number of positive skin tests across the ARG1 genotypes suggests similar differences in magnitude for going from 0 to 1 copy as for going from 1 to 2, consistent with an additive genetic model. Under the additive model, we observed significant evidence of linkage plus association for all 4 ARG1 SNPs (TABLE IV). Only one of the four ARG2 SNPs (arg2s4) showed evidence of linkage and association with the number of positive skin tests (P = 0.027). Due to the limited number of children with two copies of the variant allele (only four) and the sign of the estimate associated with the model, it can be concluded that the significant difference corresponds to an increase in the number of positive skin tests for children with 1 copy as compared to children with no copies of the variant. We calculated odds ratios, under the additive model, for a five unit change in the number of positive skin tests. The magnitude of the associations was similar for the five SNPs (4 in ARG1 and 1 in ARG2) that were statistically significantly associated with number of positive skin tests. P values from TDT based analyses were virtually identical (range 0.009 to 0.021 for the 4 ARG1 SNPs and 0.027 for arg2s4).

TABLE IV.

Relation between number of positive skin tests and arginase genotypes

SNP	# copies of minor allele	# of cases	# positive skin tests (mean)	OR* (95% CI)	P value for linkage and association†
ARG1
s1	0	109	4.96	1.00
	1	194	5.77	1.34(1.05–1.71)
	2	97	6.72	1.80(1.10–2.92)	0.017
s2	0	110	4.96	1.00
	1	194	5.77	1.35(1.05–1.72)
	2	96	6.75	1.81(1.11–2.95)	0.016
s3	0	121	6.47	1.00
	1	194	5.72	0.75(0.59–0.95)
	2	85	4.95	0.56(0.34–0.91)	0.018
s4	0	151	5.32	1.00
	1	203	5.91	1.42(1.09–1.85)
	2	46	6.76	2.02(1.20–3.41)	0.007
ARG2
s1	0	153	6.10	1.00
	1	166	5.42	0.89(0.69–1.14)
	2	73	5.86	0.78(0.47–1.29)	0.343
s2	0	167	6.07	1.00
	1	156	5.46	0.86(0.67–1.11)
	2	69	5.72	0.75(0.45–1.23)	0.245
s3	0	171	5.63	1.00
	1	176	6.16	0.94(0.74–1.20)
	2	45	4.73	0.88(0.54–1.43)	0.612
s4	0	277	5.51	1.00
	1	111	6.48	1.48(1.04–2.10)
	2	4	4.00	2.18(1.08–4.41)	0.027

CI, confidence interval

^*The odds ratio (OR) is interpreted as the change in odds of inheriting one or two copies of the variant allele per 5 unit change in the number of positive skin tests.

^†P values and odds ratios are based on the additive genetic model

DISCUSSION

We found genetic variation in ARG2 to be associated with childhood asthma risk in our Mexican population. Carrying two copies of the minor allele of either of two highly associated SNPs in ARG2 (arg2s1 or arg2s2) conferred a modest but statistically significantly increased risk to asthma. The modest effect size, with a 40% increase in risk for minor allele homozygotes, is of the magnitude expected for individual SNPs in genes related to asthma²⁰, a complex disease involving genetic and environmental causes and their interaction. Haplotype results agreed with the single SNP findings, although given the reduced sample size, they were less precise. We also found evidence that ARG1 SNPs were related to the degree of atopy to aeroallergens, as defined by the number of positive skin tests. All four of the ARG1 SNPs tested showed evidence of linkage plus association. In addition, one SNP in ARG2 showed a statistically significant result for atopy to aeroallergens.

Recent work by Zimmermann et al. implicates arginase in the etiology of asthma in mouse models and humans¹. Expression profiling of lung tissue from mice with experimental asthma in response to two distinct allergens revealed markedly increased expression of the two arginase isoforms, ARG1 and ARG2¹. Total arginase protein activity also increased notably following the induction of asthma in these mouse models. Using in situ hybridization for ARG1 mRNA, Zimmermann et al.¹ found localization to macrophages in areas of lung inflammation from the affected mice. Arginase protein by immunohistochemistry also localized to macrophages in areas of peribronchiolar and perivascular inflammation. In contrast, mice without the allergic asthma phenotype had undetectable arginase protein expression. Extending the finding to humans, there was significantly higher arginase protein expression in bronchoalveolar lavage macrophages from asthmatic than nonasthmatic individuals. ARG1 mRNA was strongly detected in bronchial biopsies of asthmatic patients but virtually undetectable in tissue from normals. In contrast to the murine findings, ARG1 mRNA was also observed in epithelial cells in the human bronchial biopsy. Localization studies were not done for ARG2 mRNA or protein¹.

Arginase could impact asthma pathogenesis in several ways. Inhibition of NO generation, due to reduced arginine bioavailability, provides one mechanism^{2, 3}. Arginine serves as a common substrate for both nitric oxide synthase (NOS) and arginase. Arginase catalyzes the transformation of arginine to ornithine, thereby reducing arginine availability for NO formation by NOS. Because arginine is a rate-limiting factor for NOS activity in the normal airway, decreased bioavailabity of arginine to NOS can reduce NO synthesis²¹. Deficiency of NO can induce airway hyperresponsiveness, as shown in allergen challenged guinea-pigs^22–24. With low concentrations of arginine, NOS generates superoxide and NO that interact to form peroxynitrite which leads to airway cell injury²⁵.

Another potential mechanism for a role of arginase in asthma pathogenesis relates to increasing production of ornithine, a precursor of polyamines (i.e., putrescine, spermidine and spermine) and proline^{2, 3}. Polyamines can control cell proliferation and are important in the airway. The polyamine putrescine increased in asthmatic mice along with increased expression of ARG1 and ARG2 genes¹. Circulating levels of putrescine and spermidine were found to be elevated in patients during asthmatic attacks²⁶. Proline is a rate limiting substrate for synthesis of collagen, which is deposited in the subepithelium of asthmatic airways in humans²⁷. Production of collagens in asthmatic airway cells can influence proliferation and likely contribute to airway remodeling, a key process in asthma²⁸. Interestingly, a recent study revealed elevated serum arginase activity and low plasma arginine concentration in human asthma patients²⁹. Taken together, these studies support a role for arginase in asthma pathogenesis via NO-related mechanisms and increased polyamines and prolines, favoring airway remodeling.

We found ARG1 to be associated with the degree of atopy, but not asthma. ARG2 was predominantly associated with asthma and to a lesser degree, atopy. Various lines of data support a greater role for ARG1 in allergic response compared with ARG2. ARG1 is localized to macrophages whereas ARG2 is induced in lung myofibroblasts and epithelium as well³⁰. An in vitro study using murine macrophages, showed that interleukin-4 (IL-4)/interleukin-13 (IL-13) stimulates arginase activity that depletes arginine levels leaving none available for inducible NOS to generate NO³¹. In a study of immunoglobulin E (IgE) sensitized mouse mast cell lines³², NO inhibited antigen-induced pro-inflammatory cytokine production. If this NO inhibition of antigen-related cytokine release occurs in human macrophages, then decreased NO could lead to increased pro-inflammatory cytokine production following antigen exposure. Although the T helper lymphocytes type 2 (Th2) cytokines IL-4 and IL-13 induce both ARG1 and ARG2 in mouse lung¹, induction of ARG1 mRNA expression by IL-4 is signal transducer and activator of transcription 6 (STAT6) dependent, whereas ARG2 mRNA expression is STAT6 independent. Mice deficient in STAT6 do not make IgE³³, which also supports our finding that ARG1 may contribute to allergic inflammation.

In our data, ARG2 was associated with both clinical diagnosis of asthma and asthma severity. Both ARG1 and ARG2 are expressed in mouse lung; in human lung, expression of ARG2 mRNA may be greater than ARG1³⁴. In bleomycin-induced lung fibrosis, ARG2 is increased earlier and more persistently than ARG1³⁰. Similar to the picture in bleomycin-induced fibrosis, in the allergic mouse model, ARG2 mRNA production occurred earlier after allergen challenge than ARG1¹. ARG2 appears to foster the creation of prolines and polyamines to a greater degree than ARG1³⁵ and thus might play a greater role in airway remodeling and thus asthma and asthma severity. Of note, ARG2, but not ARG1, induction by IL-4 is STAT6 independent as are features of chronic asthma³⁶. Given the apparently greater importance of ARG2 in processes relevant to airway remodeling, and the time course of ARG2 expression in relation to antigen challenges, ARG2 may be more important in the development of chronic asthma than ARG1, which may be more involved in the acute allergic asthmatic response. Alternatively, given the lack of complete resequencing data, it may be that our coverage of the both ARG1 and ARG2 was not sufficient to rule out that both genes are consistently involved in both asthma and atopy.

Our asthma cases were diagnosed by pediatric allergists at a pediatric allergy specialty clinic of a large public hospital. A limitation of the study is that we did not have tests of bronchial hyperreactivity. However, referral to this pediatric allergy clinic is a tertiary referral and thus children had already been seen by a generalist and then a pediatrician over time for recurrent symptoms. Diagnoses were made on clinical grounds according to previous guidelines⁸. Of note, physician diagnosis of asthma has been used as the major endpoint in linkage studies of asthma that have resulted in the positional cloning of asthma genes^37–40. We had objective data on atopy; skin prick tests revealed the vast majority of these asthmatic children (93%) to be atopic to aeroallergens.

A major limitation of our study was our inability to generate true haplotypes for ARG1 and ARG2 because complete resequencing data were not available. However, we felt that it was of interest to examine arginase SNPs taking advantage of public access data with frequency information. In the absence of complete resequencing data, we selected SNPs spaced widely apart from the 5′ to the 3′ end from existing databases. We genotyped four SNPs spread across the locus regions of each gene. In our Mexican population common SNPs were not in the coding region and thus we genotyped SNPs in noncoding regions. Although intronic SNPs may affect gene expression and function, we have no functional data on these SNPs. A limitation of our study is that there is no functional data available on any SNPs in either gene. A further limitation is that some of our SNPs were highly associated in our Mexican population and thus did not provide independent information. Although arg2s1 and arg2s2 are very highly associated, arg2s1 has been classified as being in exon 1 of the predicted gene LOC441689 with a product similar to protein kinase C. It is possible that a SNP in this nearby gene is functionally important or that there is a gene-gene interaction with ARG2. It is also possible that the LD block including arg2s1 and arg2s2 along with LOC441689 and other genes, contains a SNP in a regulatory or coding region that might influence gene expression or function. Although ARG2 lies relative close (2 megabase) to an asthma linkage region on chromosome 14, the r² values between SNPs closest to the linkage peak and arg2s1 ranged from 0.05 to 0.13 making it unlikely that our findings were due to linkage equilibrium with SNPs in this region.

We found suggestive evidence of a stronger association between asthma and SNPs in ARG2 among children from homes where a parent smokes. Environmental tobacco smoke is a stimulus for airway inflammation⁴¹ that might magnify genetic effects that influence this pathway. In addition to effects on inflammatory cells, smoking can decrease endothelial nitric oxide synthase (eNOS) activity in smooth muscle⁴² which, if occurring in the airway, would result in decreased bronchodilatory NO. Because arginase activity depletes arginine, leading to decreased substrate for eNOS, there could be a synergistic decrease in NO production. There may also be synergism based on airway remodeling, which both smoking and arginase activity can promote⁴³. A limitation of the case-parent triad design in the assessment of gene-environment interaction is that the independence of exposure and genotype must be assumed. However, we have no reason to suspect that the arginase genotype would alter smoking behavior.

Our data suggest associations with between childhood asthma risk and two highly associated SNPs in ARG2 and the degree of atopy and all four SNPs in ARG1. Given the strong biologic rationale for a role of arginases in allergic asthma risk, these findings merit consideration in other populations.