Divalent metal transporter-1 (DMT1) mediates dietary iron uptake across the intestinal mucosa and facilitates peripheral delivery of iron released by transferrin in the endosome. Here, we report that classical cannabinoids (Δ9-tetrahydrocannabinol, Δ9-THC), nonclassical cannabinoids (CP 55,940), aminoalkylindoles (WIN 55,212-2) and endocannabinoids (anandamide) reduce 55Fe and 54Mn uptake by HEK293T(DMT1) cells stably expressing the transporter. siRNA knockdown of cannabinoid receptor type 2 (CB2) abrogated inhibition. CB2 is a G-protein (GTP-binding protein)-coupled receptor that negatively regulates signal transduction cascades involving serine/threonine kinases. Immunoprecipitation experiments showed that DMT1 is serine-phosphorylated under basal conditions, but that treatment with Δ9-THC reduced phosphorylation. Site-directed mutation of predicted DMT1 phosphosites further showed that substitution of serine with alanine at N-terminal position 43 (S43A) abolished basal phosphorylation. Concordantly, both the rate and extent of 55Fe uptake in cells expressing DMT1(S43A) was reduced compared with those expressing wild-type DMT1. Among kinase inhibitors that affected DMT1-mediated iron uptake, staurosporine also reduced DMT1 phosphorylation confirming a role for serine phosphorylation in iron transport regulation. These combined data indicate that phosphorylation at serine 43 of DMT1 promotes transport activity, whereas dephosphorylation is associated with loss of iron uptake. Since anti-inflammatory actions mediated through CB2 would be associated with reduced DMT1 phosphorylation, we postulate that this pathway provides a means to reduce oxidative stress by limiting iron uptake.
Divalent Metal Transporter-1 (DMT1; Slc11A2) mediates ferrous (Fe2+) iron uptake across the small intestine  and the delivery of iron to peripheral tissues by transferrin . Electrophysiological studies have shown that DMT1 not only transports Fe2+, but that it also interacts with other divalent metals . DMT1 activity is voltage and pH-dependent . The SLC11A2 gene encoding DMT1 produces multiple isoforms through alternative promoters (1A or 1B) and/or alternative splicing to produce transcripts with or without an iron regulatory element (+IRE or −IRE) . While 1A isoforms are predominantly expressed in duodenum and kidney, 1B isoforms are ubiquitously expressed . The 1A isoform is regulated by hypoxia-inducible factor 2α response elements that increase intestinal DMT1 expression in response to hypoxia [6,7]. Protein levels of the +IRE isoforms increase during times of deficiency, and iron-dependent post-transcriptional control of IRE-containing DMT1 transcripts by iron regulatory proteins has been demonstrated [8,9]. The membrane proteins generated by various DMT1 transcripts have different subcellular targeting. All DMT1 isoforms display similar function and equal efficiency, but the 1A/+IRE isoform is predominantly detected at the plasma membrane, whereas the 1A/−IRE, 1B/+IRE and 1B/−IRE isoforms are more abundant in intracellular compartments [10,11]. The tissue-specific expression and subcellular localization of different DMT1 isoforms may regulate iron transport by distinct membranes and organelles in different cell types. In addition, DMT1 is regulated post-translationally by ubiquitination mediated through Ndfips (NEDD4 family-interacting protein 2) and WWP2 (WW domain-containing E3 ubiquitin protein ligase 2), which control its degradation . In summary, DMT1 expression and activity are highly regulated at the transcriptional, post-transcriptional and post-translational levels. Since excess iron is associated with oxidative stress and cytotoxicity, these various mechanisms controlling iron assimilation by DMT1 are critical to maintaining cellular homeostasis.
Small molecule inhibitors can help to further identify factors involved in the regulation of iron transport activity. Our previous screen of bioactive molecules revealed that the cannabinoid Δ9-tetrahydrocannabinol (Δ9-THC) inhibits DMT1 function . Δ9-THC effects are mediated through cannabinoid receptors (cannabinoid receptor type 1, CB1 and cannabinoid receptor type 2, CB2), members of the G-protein (GTP-binding protein)-coupled receptor class . The cannabinoid system has been implicated in many physiological functions, both in the central and peripheral nervous system and in peripheral organs, providing a therapeutic target for many pathological conditions . In particular, the cannabinoid system has been shown to negatively couple to N- and P/Q-type voltage-operated Ca2+ channels [16–18] and positively couple to A-type and inwardly rectifying K+ channels [19,20]. Cannabinoids have also been shown to inhibit acid-sensing channels  and Na+-dependent amino acid transporters . We undertook this study to determine the mechanism through which Δ9-THC inhibits DMT1-mediated iron transport, since this pathway would also provide therapeutic opportunities to limit assimilation to prevent iron overload diseases.
The hemagglutinin (HA)-tagged mouse DMT1 (non-IRE exon 1B isoform) plasmid  was generously provided by Dr Philippe Gros (McGill University). In this construct, the HA tag (YPYDVPDYA) is inserted in the predicted extracellular domain between transmembrane domains 7 and 8 . Previous work by this group has characterized the transport activity [24,25] and membrane trafficking [26,27] of this construct to demonstrate its compatibility with properties of wild-type DMT1 (non-IRE 1B). In the present study, site-directed mutations on this plasmid (S21A-HA, S23A-HA, S32A-HA, S39A-HA and S43A-HA) were constructed using the QuikChange® II XL kit (Stratagene, La Jolla, CA, USA) according to the manufacturer's protocol. The site-directed mutation, orientation and fidelity of the inserts, and incorporation of the epitope tags, were confirmed by directed sequencing (Dana Farber/Harvard Cancer Center DNA Resource Core).
Cell culture and transient transfection
HEK293T(DMT1) cells  were cultured in α-minimal essential medium supplemented with 50 U/ml penicillin, 50 μg/ml streptomycin and 10% fetal bovine serum. To transiently express HA-tagged DMT1 (DMT1-HA), parental HEK293T cells were transfected with 0.5 µg (24-well plates), 4 µg (6-well plates) or 24 µg (10 cm2 dishes) of wild-type or mutant alleles of DMT1 plasmid in serum- and antibiotic-free medium using Lipofectamine 2000 (Invitrogen) at a DNA:transfection reagent ratio of 1:2.5 according to the manufacturer's specifications. Transfection medium was replaced 4–6 h later with antibiotic-free growth medium, and cells were cultured for an additional 24 h prior to experiments.
Transport studies with radioisotopic tracer
HEK293T(DMT1) cells were grown to 60–80% confluence in 10-cm plates, and washed three times with PBS containing 1 mM MgCl2 and 0.1 mM CaCl2 (PBS++), and lifted by carefully pipetting up and down into PBS++ or uptake buffer (25 mM Tris, 25 mM MES, 140 mM NaCl, 5.4 mM KCl, 1.8 mM CaCl2, 0.8 mM MgSO4 and 5 mM glucose, pH 6.75). Cell counts were determined and samples were aliquoted for studies. Δ9-THC (Sigma), CP 55,940 (Tocris Bioscience), WIN 55,212-2 (Tocris Bioscience), anandamide (Tocris Bioscience), staurosporine (LC Laboratories), H-7, GDPβS, Sp-8-bromo-cAMP (Sigma Chemical Co.), U0126-EtOH, trametinib or PD0325902 (Selleck Chemicals) were added to cell aliquots at the indicated concentrations and compared with control cell aliquots treated with the appropriate vehicle (water or 0.5% DMSO, v/v). In the standard assay, cells were incubated at 37°C for 20 min in uptake buffer containing 1 μM 55Fe and 50 μM ascorbic acid (pH 6.75). Cells were chilled on ice, collected on nitrocellulose filters and washed three times by centrifugation or vacuum filtration with ice-cold PBS to remove any unbound 55Fe or 54Mn. Cells were lysed with solubilization buffer (0.1% Triton X-100/0.1% NaOH), and cell-associated radioactivity was determined by scintillation counting. Uptake was calculated as pmol per million cells per min and was normalized to control (vehicle alone) to determine the extent of inhibition in the presence of a drug as indicated in the legends. In experiments employing HEK293T cells transiently expressing DMT1 and various mutants, transport assays were carried out for the times indicated in the figures and legends by using individual wells that were each separately transfected. In these experiments, cells from each well were lysed and uptake was measured by scintillation counting, then adjusted to lysate cell protein for that individual well measured using the Bradford assay. Uptake is shown as pmol/mg protein.
Small interfering RNA-mediated gene suppression of cannabinoid receptor
HEK293T(DMT1) cells (1 × 105 cells per well) were plated in six-well plates overnight and then transfected with 100 nM of control siRNA (Ambion), CB1 siRNA (CNR1-s3261, Ambion) or CB2 siRNA (CNR2-s3263, Ambion) using Lipofectamine 2000 (Invitrogen) as described above. After 24 h incubation, the transfection medium was replaced with serum-containing, antibiotic-free medium, and cells were cultured for another 16 h before experiments. Cells were collected for the determination of CB1 and CB2 protein abundance and 55Fe uptake.
Cells were washed twice with ice-cold PBS, and the total cell lysates were prepared in lysis buffer containing 150 mM NaCl, 1 mM EDTA, 50 mM Tris and 1% Triton containing a protease inhibitor cocktail (Complete Mini, Roche). Protein concentration was determined by the Bradford assay. Samples (25 µg) were separated by electrophoresis on 10% SDS–PAGE gels and transferred to PVDF membranes, and immunoblotted with rabbit anti-CB1 or -CB2 antibody (Cayman Chemical Company). The primary antibody was detected with IRDye 800-conjugated secondary antibodies (Rockland Immunochemicals, Inc.). After further washing, immunoreactivity was detected by an Odyssey Infrared Imaging System (Li-COR).
To determine HA-DMT1 phosphorylation, cells transfected with HA-DMT1 or empty vector were scraped into ice-cold PBS and pelleted by centrifugation, then lysed in Nonidet-P40 (NP-40) buffer (40 mM HEPES, pH 7.4, 120 mM NaCl, 5% glycerol, 1% NP-40 and 1 mM EDTA, pH 8.0) containing phosphatase inhibitors (10 mM pyrophosphate, 10 mM β-glycerophosphate, 50 mM NaF and 0.5 mM orthovanadate) for 30 min at 4°C with rotation. Samples were centrifuged for 10 min at 15 000×g, and the protein concentration of the lysates was determined by the Bradford assay. The cell lysates were preincubated with anti-HA agarose (Pierce) overnight at 4°C with rotation and washed with TBS before protein elution at 95°C for 5 min. Eluates were electrophoresed on 4–20% gradient gels, transferred to PVDF membranes and the extent of HA-DMT1 phosphorylation was assessed by immunoblotting with mouse antiphosphoserine antibody (Millipore) followed by IRDye 800-conjugated secondary antibodies (Rockland Immunochemicals). Meanwhile, total cell lysates were immunoblotted with either anti-HA antibody or by antiactin antibody followed by IRDye800-conjugated secondary antibodies. After further washing, immunoreactivity was detected by an Odyssey Infrared Imaging System (Li-COR).
Statistical comparisons were determined using unpaired Student's t-test (Prism Graph Pad, Berkeley, CA, USA). In Figure 1, dose-response curves were best fit by a Hill four-parameter sigmoidal model with analysis made using Sigmaplot (Systat software). In Figure 2B, data were analyzed by two-way ANOVA followed by Holm–Sidak's multiple comparisons test for pairwise comparisons. In Figure 4C, where comparisons were made by one-way ANOVA followed by Tukey post hoc analysis. Results shown are means ± SEM unless otherwise indicated. For all statistical analyses, results were considered significant at P < 0.05.
Dose–response studies of DMT1 inhibition by cannabinoids.
Loss of CB2 receptor abrogates Δ9-THC inhibition of iron transport.
Inhibitors and activators of DMT1-mediated iron uptake.
Serine phosphorylation of DMT1 is blocked by Δ9-THC.
Cannabinoids inhibit DMT1-mediated iron (II) uptake
Our previous study showed that Δ9-THC inhibited DMT1 activity with an IC50 of ∼0.45 μM . We further examined the effect of different types of cannabinoids on iron uptake using an 55Fe tracer assay. Inhibition of HEK293T cells stably overexpressing DMT1 by the endogenous cannabinoid anandamide, and synthetic cannabinoids structurally unrelated to the ‘classic cannabinoid Δ9-THC, including the so-called nonclassical cannabinoid CP 55,940 and the aminoalkylindole agonist WIN 55,212-2, was examined (Figure 1A). The IC50 values were all below 10 µM (see table in Figure 1). The values for y(min) determined from regression analysis suggest an index of efficacy of CP 55,940 (30.9%) > anandamide (36.1%) > WIN 55,121-2 (56.8%). These combined data indicate that both exogenous and endogenous cannabinoids inhibit DMT1-mediated iron uptake in a dose-dependent manner. Since DMT1 transports other divalent metals, we also confirmed the effects of cannabinoids on 54Mn uptake, which was reduced by anandamide and CP 55,940 (Figure 1B).
CB2 mediates inhibition by Δ9-THC
Cannabinoids produce many biological effects primarily mediated through interactions with cannabinoid receptors CB1 and CB2 . To determine whether iron transport inhibition by Δ9-THC was due to interactions of either cannabinoid receptor, we used siRNA to knockdown CB1 and CB2 in HEK293T(DMT1) cells. Western blot analysis showed that CB1 and CB2 receptors were effectively reduced compared with control cells (Figure 2A). DMT1-mediated 55Fe uptake and its response to Δ9-THC were unaffected by CB1 knockdown (Figure 2B). Knockdown of CB2 reduced the total amount of iron taken up in the absence of Δ9-THC, at 2 μM Δ9-THC and at 10 μM Δ9-THC in CB2 siRNA-treated cells compared with the control siRNA-treated cells (Figure 2B). These results indicate that inhibition of iron transport is dependent on the expression of CB2. Under basal conditions, the presence of this receptor appears to enhance DMT1 activity, supporting the idea that cannabinoid agonists act through CB2 to suppress transporter function.
CB2 is a member of the G-protein-coupled receptor family and is known to negatively regulate adenylate cyclase through its interactions with Gi/Go α subunits. To further delineate its potential role in the regulation of DMT1-mediated iron transport, we studied activators and inhibitors of this signaling pathway (Figure 3). The treatment of HEK293T(DMT1) cells with the GDP analog GDPβS reduced iron uptake. In contrast, the Gi/o protein activator pertussis toxin enhanced transport. Likewise the cAMP agonist sp-8-Br-cAMP enhanced iron transport, consistent with the idea that negative regulation of adenylate cyclase leads to inhibition of DMT1 activity. H7, a cAMP-protein kinase (PKA) inhibitor, reduced iron transport by a small but reproducible extent. Staurosporine, a broad spectrum PKA and protein kinase C (PKC) inhibitor, had a greater dose-dependent effect (see below). Through interactions with the G-protein βγ subunit, CB2 can also couple to mitogen-activated protein kinase/extracellular signal-regulated kinases (MAPK/ERK). We, therefore, also tested transport inhibition by the kinase inhibitors U0126-EtOH (50 μM), trametinib (5 μM) or PD0325902 (50 μM), but did not observe any significant or reproducible effects (data not shown).
Serine phosphorylation of DMT1 is blocked by Δ9-THC
As discussed above, CB2 receptor signal transduction mechanisms regulate serine/threonine phosphorylation by many different kinases. To more directly examine possible phosphorylation of DMT1 and effects of Δ9-THC, HEK293T cells were transfected to transiently express HA-tagged DMT1. These cells and control cells (transfected with empty plasmid) were treated with or without Δ9-THC. Anti-HA antibodies were used to immunoprecipitate HA-DMT1 from cell lysates, which was then immunoblotted to detect phosphoserine. Antiphosphoserine antibodies were detected phosphorylation of HA-DMT1 in the absence of Δ9-THC. The extent of DMT1 serine phosphorylation was reduced in cells treated with the cannabinoid (Figure 4A). The results are consistent with the idea that under basal conditions, active DMT1 is serine-phosphorylated and in response to CB2 activation, the inactivated transporter is dephosphorylated.
Two independent computer-assisted approaches that predict phosphorylation sites were used to examine potential DMT1 target residues: PhosphoSitePlus  and NetPhos 2.0 server . Both algorithms revealed possible phosphosites in the N-terminal cytoplasmic domain (S21, S23, S32, S39 and S43) on DMT1 (Figure 4A). These residues are all conserved in four known protein isoforms of DMT1. To determine if these serine residues play a role in DMT1 phosphorylation, we replaced each serine residue individually with alanine by site-directed mutagenesis and tested HA-DMT1 phosphorylation. Our results showed that point mutation of serine to alanine at 43 (S43A) completely abolished basal phosphorylation of DMT1 (Figure 4B).
DMT1 phosphorylation at serine 43 promotes iron uptake
To confirm that S43 is a phosphorylation site important for DMT1 function, cells were transfected to express S43A-HA. Control immunoprecipitation experiments confirmed that this mutant failed to be phosphorylated when compared with serine phosphorylation of wild-type DMT1 (Figure 5). Subsequently, the transport of iron by cells expressing the serine mutant was tested using the 55Fe tracer assay. As shown in Figure 5B,C, expression of the S43A-HA mutant produced lower 55Fe uptake compared with cells expressing wild-type DMT1. The input controls (Figure 5A) confirm that the expression of wild-type and the S43A mutant in transfected HEK293T cells are similar, suggesting that the change in iron uptake reflects altered activity of the latter. Both the rate and extent of iron transport were reduced in cells expressing S43A-HA compared with wild-type DMT1-HA.
DMT1 phosphorylation at serine 43.
To test the potential role of phosphorylation in DMT1 transport function, the effects of the serine/threonine protein kinase inhibitor staurosporine on iron uptake were determined. Staurosporine inhibited DMT1-mediated 55Fe uptake in a dose-dependent manner (Figure 6A). To confirm the inhibition of protein kinase activity not only blocked Fe transport but was also associated with reduced serine phosphorylation, HEK293T cells transiently transfected to express HA-tagged DMT1 were treated with or without staurosporine, and anti-HA antibodies were used to immunoprecipitate HA-DMT1 from cell lysates to detect phosphoserine. As predicted, serine phosphorylation of HA-DMT1 was reduced in cells treated with staurosporine compared with vehicle-treated control cells (Figure 6B). Figure 6C shows densitometric data obtained from two independent experiments with a 50% decrease in HA-DMT1 phosphorylation in staurosporine-treated cells. It has been shown that staurosporine induces apoptotic cell death in many different cell lines [30–32]. However, in our experiments, we did not observe any signs of cell death over the short 20 min time course of incubation (Y.A.S., personal observation).
Staurosporine inhibits DMT1 transport and reduces serine phosphorylation.
Previous studies from our research group applied chemical genetics to advance our understanding of different pathways of iron uptake through the discovery of transport inhibitors [13,33–35]. The bioactive small molecule Δ9-THC was identified to block iron uptake by DMT1 in HEK293T(DMT1) cells . The present study shows that the CB2 receptor is responsible for mediating the effects of Δ9-THC on DMT1 activity in this cell line. In Δ9-THC-treated cells, serine phosphorylation of DMT1 was significantly reduced. These observations indicate that the transporter's function can be regulated by limiting its phosphorylation or promoting its dephosphorylation. Consistent with this model, site-directed mutagenesis studies revealed that conversion of DMT1 Ser43 to Ala results in loss of phosphorylation and iron uptake activity. All of these functional effects correlate with inhibition of uptake observed in the presence of cannabinoid agonists.
The fact that Ser43 is present in all DMT1 isoforms suggests that regulation at this site is independent of the C-terminus of the protein; i.e. occurring in both +IRE and −IRE isoforms. Previous studies reported that Nramp1 (natural resistance-associated macrophage protein) , a macrophage-specific homolog of Nramp2 or DMT1 , is also post-translationally modified by phosphorylation . Although specific phosphorylation site(s) have not been determined, the Nramp1 protein contains several possible PKC and casein kinase II (CK-2) sites [38,39]. Barton et al.  have shown that a peptide corresponding to the N-terminal domain of Nramp1 is phosphorylated and suggested that this modification reflects PKC activity. Beyond the preliminary identification of Nramp1 as a phosphoprotein, little is known about functional effects of this modification. Our observation is the first report of DMT1 (Nramp2) phosphorylation. Computer-based programs, such as KinasePhos , GPS (group-based prediction system)  and PPSP (prediction of PK-specific phosphorylation site) , all indicate that the Ser43 in DMT1 is a potential site for protein kinase A/G/C family members as well as CK-2. Further research is necessary to precisely define the kinase(s) involved in recognition of the Nramp transporter family and how this post-translational modification affects their transport mechanism. It has been suggested that PKCβ could play a role in Nramp1 regulation, based on regulation of this kinase by iron . We have found that staurosporine, a broad spectrum inhibitor of both protein kinases PKA and PKC , reduces both DMT1 transport activity and phosphorylation, but further study is necessary to identify the DMT1 serine kinase and its relationship to other Nramp family kinases.
Although the C-termini of DMT1 isoforms have been shown to direct subcellular trafficking , the possibility that phosphorylation of the N-terminal Ser43 redirects subcellular localization of DMT1 to modulate its function must be considered . It is interesting to note that the peripheral benzodiazepine receptor-associated protein 7 (PAP7) has been implicated in DMT1 trafficking . This factor is functionally linked with the G-protein-coupled benzodiazepine receptor and associates with a cAMP-dependent kinase . In a similar fashion, CB2 might regulate phosphorylation or dephosphorylation of DMT1 through a third partner akin to PAP7. We have attempted to study changes in DMT1 cellular localization upon cannabinoid treatment, as well as in response to kinase inhibitors, but failed to observe major redistribution of transporters in HEK293T(DMT1) cells or in HEK cells transiently overexpressing the HA-tagged DMT1. Both fluorescence confocal microscopy and cell surface biotinylation did detect surface transporters (data not shown) as previously demonstrated . Further work is necessary to determine how phosphorylation of DMT1's N-terminus may affect iron transport. It has been shown that the copper transporter ATP7B is basally phosphorylated and becomes hyperphosphorylated in response to increased copper levels and although the functional consequences have yet to be fully determined, this activity does not affect membrane trafficking .
To our knowledge, the demonstration that CB2 modulates iron uptake activity is also the first report of endocrine receptor regulation of DMT1 transport. A growing body of evidence has shown the expression of CB2 receptors in microglial cells, astrocytes and some neuron populations . The function of CB2 in microglia, the macrophages of the brain, is of particular interest since receptor activation decreases the in vitro production of proinflammatory molecules in rat [50,51] and human  microglial cells. CB2 activation also reduces the release of proinflammatory factors in a model of hypoxic–ischemic brain damage in newborn rats  and in Huntington's disease . Moreover, recent evidence shows that microglia express a functional endocannabinoid signaling system and that CB2 receptors control their immune-related functions . Iron is also an important regulator of immune cell function , and CB2 regulation of DMT1 iron uptake activity could potentially reflect an important physiological response associated with iron withdrawal during anti-inflammatory responses . Our work in an immortalized microglial cell line has recently demonstrated that the CB2 selective agonist JWH-015 markedly reduces inflammation promoted by the Alzheimer's disease protein amyloid-β . A decrease in iron import would be consistent with this effect. While there are many studies on iron regulation and its effects on peripheral macrophages and immune function, very little is known regarding microglial iron handling . Thus, it will be interesting to test whether microglia utilize the cannabinoid signaling system to modulate cellular iron status through the regulation of DMT1 function.
Further studies are also warranted to determine the detailed mechanism through which DMT1 phosphorylation is responsible for controlling its activity. DMT1 is regulated at transcriptional [6,60], post-transcriptional [8,9] and post-translational levels . Our study demonstrates that phosphorylation at S43 is linked to a functionally active state of the transporter's function. Understanding how DMT1 function is modified by this post-translational modification may provide insights about the membrane transport of iron. Moreover, since Ser43 phosphorylation activates DMT1 under basal conditions, dephosphorylation at this site provides an interesting potential target for pharmacological tools to treat diseases of iron overload.
cannabinoid receptor type 1
cannabinoid receptor type 2
casein kinase II
divalent metal transporter-1
iron responsive element
mitogen-activated protein kinases/extracellular signal-regulated kinases
peripheral benzodiazepine receptor-associated protein 7
protein kinase C
Y.A.S., R.K. and H.W. conducted experiments. All authors contributed to conceptual design of experiments. Y.A.S. and R.K. wrote sections of this manuscript and M.W.R. is responsible for final editing.
Research reported in this publication was supported by grants from the National Institutes of Health under the awards R01 DK064750 and R21 DA0255. Y.A.S. was supported by a grant from the National Institutes of Health under award number [K99 ES024340].
We appreciate the technical help from Jose Carlo Sosa in conducting the MAPK/ERK inhibition study, and appreciate advice from Dr Jonghan Kim, Northeastern University, on Sigmaplot analyses.
The Authors declare that there are no competing interests associated with the manuscript.
Both authors contributed equally to this work.