Resistance to DDT and pyrethroids is widespread and has hampered malaria control efforts throughout Africa [2–9]. Artificial insecticide resistance selection on laboratory colonies is useful as it allows one to study the resistance mechanism on a population not influenced by other environmental selection pressures. Furthermore, artificial selection in the laboratory allows us to mimic the development of insecticide resistance from repeated and continuous exposure to insecticides, a situation that wild vector populations are frequently exposed to.
The two resistant An. arabiensis colonies used in this study, one from South Africa and the other from Sudan, have been under DDT selection pressure in the laboratory. Bioassay data confirmed that both SENN-DDT and MBN-DDT are highly resistant to DDT. In addition to a high level of DDT resistance, the two colonies were found to be resistant to pyrethroids (deltamethrin and permethrin). The South African population showed additional resistance to carbamates, which was not present in the Sudanese colony.
The development of multiple insecticide resistance in the above mentioned colonies is supported by subsequent studies published on the same laboratory populations. The MBN colony was colonized in 2002 without detecting pyrethroid resistance in the population. However, three years later Mouatcho et al.  reported the presence of pyrethroid resistance, which was rapidly selected for (within four generations) in the laboratory and has been shown to be P450 based. The same author also showed that carbamate tolerance could be selected for from the same colonized field population. Ranson et al.  recently published a country wide study and showed that An. arabiensis populations from Sudan are resistant to both DDT and pyrethroids, but remained fully susceptible to carbamates and the organophosphate, fenitrothion. This supports what was observed in the SENN-DDT colony.
The fact that DDT and pyrethroid resistance in An. gambiae are linked has been well-documented and has been attributed to the presence of kdr mutations [23, 25]. Specifically, kdr is strongly linked with DDT and permethrin resistance, and less so with deltamethrin resistance [27, 40]. In An. arabiensis, the relationship between the presence of kdr mutations and resistance phenotype is also complicated [2, 41]. The SENN-DDT colony is fixed for the L1014F mutation. The South African An. arabiensis population has previously been confirmed not to carry any kdr mutations [7, 8]. However, the continued selection pressure from exposure of MBN-DDT to DDT has resulted in this colony being fixed for the L1014F mutation. The L1014S mutation is absent from both laboratory colonies.
The detoxification enzyme profiles of the two laboratory selected DDT-resistant An. arabiensis strains was investigated using cross-species hybridizations of An. arabiesnsis genetic material with the An. gambiae detoxification microarray (detox chip). Of the 98% of probes that hybridized, only one gene in the SENN-DDT colony was over-transcribed. This was a cytochrome P450, CYP9L1. This was in contrast to the MBN colony where a similar success rate of probe hybridization was recorded, but 20 genes were highly transcribed in the resistant phenotype.
The use of the An. gambiae detox chip allows for the evaluation of transcription of a large number of genes simultaneously, but the criteria one uses to find significance will determine how many genes are of interest for further study. In other studies (both same- and cross-species hybridizations) the cut-off for significance in terms of fold change ranged from >1.5 to 2.0, and the p-value cut-off for significance ranged form < 0.001 to <0.05 [32, 33, 42–44]. Generally, where a higher fold-change was used as criteria to identify over-transcribed genes, a lower p-value cut-off was also used to determine significance, and vice versa. In this study, the stringency was adjusted for the wash solutions by increasing the required amount of each solution (i.e. higher than what was recommended by the supplier). The experimental conditions selected produced the best arrays, but because the experiment was based on cross-species hybridizations, we chose to use less strict criteria for identifying those genes with a significant level of differential transcription.
The action of the P450-dependent monooxygenases is one of the ways in which insects become resistant to insecticides . Only one gene, CYP9L1, showed high expression levels in the SENN resistant phenotype and is likely to play a key role in the observed resistance to deltamethrin. The CYP9 gene family is closely related to the CYP6 family (highly expressed in the MBN resistant phenotype)  and members have been linked to insecticide resistance in a number of insects [44–46]. Although not likely to be the case here, it is interesting to note that a single P450 enzyme has been implicated in resistance to DDT [47, 48].
Five genes were consistently saturated when both MBN- and SENN-DDT arrays were analysed. Some of these were mainly saturated in one channel, and less so in the other, which raises the possibility that a gene is over-transcribed, but this is masked by the saturation, and might therefore be overlooked. Two of these, CYP4G16 and CO1, were investigated further using qPCR and SENN-DDT genetic material. The monoxygenase, CYP4G16 was chosen because it has previously been linked to pyrethroid tolerance in An. arabiesnis. The cytochrome oxidase gene, CO1 , was selected as it was over-transcribed in a microarray study on pyrethroid resistant An. funestus. In this study, we obtained FC values of 1.8 and 1.6 for CYP4G16 and CO1 respectively after qPCR analysis. While these values are relatively low when compared with previously reported data, their involvement, if any, in resistance and the reason for saturation on the microarrays should be investigated further.
According to our criteria, 20 genes were differentially regulated in the resistant MBN colony and most of these genes belong to the monooxygenase and GST enzyme groups. In addition, most of the over-transcribed CYP genes belonged to the CYP6 family, which is frequently associated with insecticide resistance in insects. The top four genes were selected for qPCR validation. These were, in order of significance, CYP6M2, TPX4, CYP6AK1 and CYP6P3. Recently, Munhenga and Koekemoer  used qPCR to assess the transcription of a range of monooxygenase genes in a pyrethroid-selected An. arabiensis colony from the same geographical area (KZN, South Africa). They found that CYP6Z1 (FC = 4.7), CYP6Z2 (FC = 1.7) and CYP6M2 (FC = 2.2) were significantly over-transcribed. Interestingly, in our evaluation of CYP6M2, qPCR produced a FC of 2.2, the same level as that reported by Munhenga and Koekemoer , even though a different reference gene was used between the two studies.
Of the CYP genes that were over-transcribed in this study according to microarray evaluation, a number have been implicated in insecticide resistance in An. gambiae. Djouaka et al.  found that CYP6P3 and CYP6M2 were both upregulated in pyrethroid-resistant An. gambiae populations in Benin and Southern Nigeria. In permethrin-resistant An. gambiae from Ghana, CYPM2 CYP6AK1 and CYP6P3 were amongst the top 10 differentially expressed genes in resistant mosquitoes . The authors found that the outcomes of the microarray and qPCR data were similar as was confirmed in the present study.
The GSTs also featured prominently in the enzyme profile of resistant MBN colony. The epsilon class GSTs have been specifically linked to DDT resistance in An. gambiae[29, 51–54] and delta class GSTs to a lesser extent . Furthermore, GSTs have more recently been linked to pyrethroid resistance in other insects [55, 56] and so their presence in the resistance profile of MBN-DDT might be linked directly to protection against the pyrethroid, deltamethrin. Because they help to protect cells against oxidative stress, their over-expression in the MBN-DDT colony is also likely to be linked to the action of the cytochrome P450s where the GSTs are involved in secondary metabolism through the action of glutathione peroxidase .
A number of enzymes, namely the SODs, TPXs and GRXs, counteract the effects of reactive oxygen molecules, which are harmful to the host . The SODs function by converting superoxide anions to hydrogen peroxide and oxygen . In turn, the TPXs are involved in the removal of hydrogen peroxide . Based on microarray experiments, we reported high levels of TPX4 (2.3 fold) expression in the South African population of DDT selected An. arabiensis. This enzyme was over-transcribed in An. arabiensis during the spraying season of a cotton field in Cameroon , while TPX1 was over-expressed in An. gambiae, resistant to pyrethroids, from Ghana . In the MBN colony, whether the high expression of TPX4 is related directly to the activities of the P450 enzymes (to counteract metabolic byproducts), or is a function of the insecticide resistance selection process where they are on “stand-by” to provide protection against pyrethroids, is unknown.
According to Brooke and Koekemoer , and references therein, the correlation between the presence of kdr and mosquito response to insecticide is strongest in the case of DDT, less so with permethrin, and weakest with deltamethrin. The outcome of the synergist studies performed here suggests that detoxification enzymes have no impact on DDT resistance in these strains, but are very important for protection against the pyrethroid, deltamethrin. The presence of the L1014F kdr mutations is likely to assist in protection against permethrin.