Renal sulfatides: sphingoid base-dependent localization and region-specific compensation of CerS2-dysfunction.

Mammalian kidneys are rich in sulfatides. Papillary sulfatides, especially, contribute to renal adaptation to chronic metabolic acidosis. Due to differences in their ceramide (Cer) anchors, the structural diversity of renal sulfatides is large. However, the underling biological function of this complexity is not understood. As a compound’s function and its tissue location are intimately connected, we analyzed individual renal sulfatide distributions of control and Cer synthase 2 (CerS)2-deficient mice by imaging MS (IMS) and by LC-MS2 (in controls for the cortex, medulla, and papillae separately). To explain locally different structures, we compared our lipid data with regional mRNA levels of corresponding anabolic enzymes. The combination of IMS and in source decay-LC-MS2 analyses revealed exclusive expression of C20-sphingosine-containing sulfatides within the renal papillae, whereas conventional C18-sphingosine-containing compounds were predominant in the medulla, and sulfatides with a C18-phytosphingosine were restricted to special cortical structures. CerS2 deletion resulted in bulk loss of sulfatides with C23/C24-acyl chains, but did not lead to decreased urinary pH, as previously observed in sulfatide-depleted kidneys. The reasons may be the almost unchanged C22-sulfatide levels and constant total renal sulfatide levels due to compensation with C16- to C20-acyl chain-containing compounds. Intriguingly, CerS2-deficient kidneys were completely depleted of phytosphingosine-containing cortical sulfatides without any compensation.

like galactosylceramide I 3 -sulfate (SM4s) and lactosylceramide II 3 -sulfate (SM3), one sulfate ester is bound to the C3 atom of the terminal galactosyl residue. In contrast to the minor bis-sulfated form gangliotetraosylceramide-II 3 , IV 3 bis-sulfate found in mouse kidney, SM4s as well as SM3 are present in both human and mouse kidneys ( Fig. 1 ). Sulfatide SM4s is commonly known to stabilize the myelin sheaths of the central and peripheral nervous system. In glandular epithelia, sulfatide SM4s concentrations appear synchronized with Na + -K + -ATPase activities and are elevated upon NaCl loading (1)(2)(3)(4). In the kidney, SM4s and SM3 are postulated to function as essential players in the transport of sodium chloride ( 5 ), but evidence is missing. Recently, analysis of mutant mice revealed an important function of sulfatides (SM4s and SM3) for the ammonia exchange process ( 6 ). A role in the process of osmotic balance, which had been postulated previously ( 1,7 ), could be excluded in the recent study of mutant mice ( 6 ). With respect to their Cer anchor, sulfatides appear to be very diverse within the renal system .
Modifi cations by ␣ -hydroxylation of the acyl chain, as well as variations in its chain length from C16 up to C26, are commonly known. Next to the major sphingoid base C18-sphingosine (d18:1), 3 C18-phytosphingosine (t18:0) also appears in SM3. As described in our previous work ( 6 ), hydroxylated forms [Cer anchor with an alpha-hydroxy acyl chain and a sphingosine (AS)] of SM4s are mainly located to the renal medulla while nonhydroxylated forms [Cer anchor with a nonhydroxy acyl chain and a sphingosine (NS)] show the highest concentration in the renal papillae. Until now, there was no systematic description of sulfatide location in the kidney with respect to the different fatty acid acyl chain and sphingoid base lengths. On one hand, the length of the acyl moiety depends on acyl-CoA availability as substrates for sphingolipid (SL) synthesis and, on the other hand, on the substrate specifi city of the individual Cer synthases (CerSs). Seven members of mammalian elongases [elongation of very long chain fatty acids (Elovl)1-7] regulate differentially, in a cell typespecifi c manner, the elongation degree for saturated and polyunsaturated fatty acids, and by that determine the acyl-CoA pool available for SL synthesis ( 8 ). In mammals, six CerSs (CerS1-6) condense these acyl-CoAs with sphingoid bases to form (dihydro)ceramides. Whereas human and mouse CerS1, -5, and -6 seem to be restricted to a certain set of long chain acyl-CoAs (C14-C20), CerS2 and -4 accept very long acyl-CoAs ( у C22) up to 26 C-atoms in length (9)(10)(11)(12)(13), and CerS3, being rather promiscuous, further on can introduce ultra-long acyl-CoAs with up to 38 C-atoms in length ( 12 ). The length of the sphingoid base is guided by the serine palmitoyl-CoA transferase (SPT) complex (14)(15)(16) and impacts the fi nal Cer anchor patterns observed. Hence, the acyl chain-and sphingoid base length-dependent localization of SLs to different parts of an organ should correlate with the regional specifi c expressions of elongases, CerS, and the SPT complex composition. Indeed, CerS2, for example, is expressed in myelin-forming cells where sulfatides with very long Cer anchors are found, but not in neurons which synthesize gangliosides with long chain Cer anchors ( 17 ). On the other hand, Cers3 is expressed in differentiating epidermis and spermatocytes, where unique ultra-long chain acyl moieties are found in respective SLs ( 10,12,(18)(19)(20). However, the majority of ultra-long chain acyl moieties found in the epidermis are of saturated and monounsaturated nature ( 21 ), whereas in spermatocytes these ultra-long acyl chains are polyunsaturated with up to six double bonds ( 22,23 ). One reason for this is differential expressions of Elovl enzymes, Elovl1 and -4 in the epidermis and Elovl5 and -2 in spermatocytes ( 20,21,(24)(25)(26)(27). Recently, it was demonstrated that the small subunits of the SPT guide the selectivity of the SPT complex for the acyl-CoA. In this context, a single amino acid residue, Met25 in ssSPTa and Val25 in ssSPTb, was identifi ed, which confers specifi city for palmitoyl-or stearoyl-CoA ( 15 ). Here we demonstrate the abundant and selective expression of C20-sphingosinecontaining sulfatides/SLs, which correlates with mRNA expression of ssSPTb in the papillae.

Animals
All animal procedures were performed in accordance with the guidelines for the care and use of laboratory animals and were approved by Department 35 of the Regierungspräsidium Karlsruhe. Female wild-type (wt) mice (C57BL/6N) were obtained from the German Cancer Research Center. Cers2 Ϫ / Ϫ mice have been described before ( 29 ), and were backcrossed prior to this study to a C57BL/6N background of 99.25%.

Preparation of lipid extracts from kidneys of wt and mutant mice
Lipids were extracted according to a previously reported procedure ( 30 ), with slight modifi cations. Briefl y, one frozen kidney of either 11-or 6-week-old (fe)male Black-6 (C57BL/6N) or Cers2 Ϫ / Ϫ mice were homogenized for 2 min on ice in 2 ml methanol and 3 The ceramide anchor of SLs is often detailed in parentheses following the abbreviation for ceramide (Cer), e.g., Cer(d18:1;16:0) or (t18:0;h16:0). In parentheses, the description of the sphingoid base is followed by the description of the N-linked acyl chain after the semicolon. The sphingoid base is described fi rst with a letter referring to the amount of hydroxy groups within the base (m, mono; d, di; t, tri; and te, tetra), then follows the number of C-atoms separated by a colon to the number of double bonds within the sphingoid base. After the semicolon, the fatty acid is described. Fig. 1. Anabolic pathway of renal sulfatides. SL synthesis starts out at the endoplasmic reticulum with the condensation of serine and a long chain acyl-CoA, in most cases palmitoyl-CoA. This is also the rate-limiting step, regulated by ORMDL proteins (ORM1-like proteins). After reduction of 3-keto-sphinganine to sphinganine, condensation with a variety of acyl-CoAs yields dihydroceramides, which are modifi ed 300 l water with an Ultra Turrax T25 basic (IKA Labortechnik, Staufen, Germany) at 24,000 rpm in a 15 ml polypropylene vial. For further extraction, each homogenate was transferred into a glass vial and the polypropylene vial was rinsed two times with 500 l methanol. Afterwards, 3 ml of chloroform were added to obtain a solvent mixture of chloroform:methanol:water (10:10:1, v:v:v). The extract was centrifuged for 10 min at 3,000 rpm, and the supernatant was collected in a separate glass vial. Extraction was completed by repetition of this procedure twice, once with 3 ml chloroform:methanol:water (10:10:1, v:v:v) and once with a 30:60:8 (v:v:v) mixture. Every extraction step included 2 min of sonication. The pooled extracts were dried under nitrogen stream (37°C). Finally, the extract was dissolved in 100 l chloroform:methanol:water (10:10:1, v:v:v) per 100 mg kidney wet weight and stored at Ϫ 20°C. For lipid extraction from separated renal regions (papillae, medulla, cortex), fresh mouse kidneys (C57BL/6N) were rapidly prepared after euthanization . Between 7.4 and 18.4 mg of each region were carefully separated under a light microscope. Lipids were extracted from homogenate as described above. The pooled extracts were dried under nitrogen stream (37°C). Finally, the extracts were dissolved in 100 l chloroform:methanol:water (10:10:1, v:v:v) per 100 mg kidney wet weight and stored at Ϫ 20°C.

MALDI imaging MS
Mouse kidneys from female (C57BL/6N) and Cers2 Ϫ / Ϫ mice were frozen without prior perfusion and sliced into 10 m sections using a Leica CM1510S cryostat (Leica Biosystems, Nussloch, Germany) at Ϫ 20°C. To prevent interference of embedding medium with mass spectrometric analysis, the frozen kidneys were fi xed on the carrier by carefully dipping only one side of the organ in the embedding medium. By doing that, the side from which sections were obtained was free of embedding medium. Tissue sections were mounted onto indium tin oxidecoated conductive glass slides (Bruker Daltonics), dried by vacuum desiccation without prior washing steps, and used immedi ately or stored at Ϫ 80°C. The 9-AA [20 mM in acetonitrile:water (80:20, v:v)] was deposited on slides using an ImagePrep matrix sprayer (Bruker Daltonics). Mass spectrometric measurements were performed using an Autofl ex III MALDI TOF/TOF instrument (Bruker Daltonics) equipped with a Smartbeam laser (200 Hz) and controlled by fl exControl 3.4 software (Bruker Daltonics). The extraction voltage was 19 kV, and gated matrix suppression (<650 Da) was applied to prevent saturation of the detector with matrix ions. Mass spectra were obtained in negative ion-refl ector mode in the m/z range from 700 to 1,200 Da using delayed extraction. Images were acquired at a spatial resolution of 50 m with 200 laser shots per position. Spectra were saved and the images constructed using fl exImaging 3.0 software (Bruker Daltonics). Mass fi lters were chosen with a width of 0.2 Da. In all analyses blood-derived lipids were not separately taken into account. However, serum sulfatide levels make up less than 0.2% of kidney sulfatide levels and may be negligible ( 31 ).
The system was controlled by MassLynxs software v 4.1. Separation was performed on an Aquity UPLC bridged-ethylene-hybrid C18 1.7 m column (2.1 × 50 mm) at 40°C. The fl ow rate was set to 500 l/min. Elution solvent A consisted of methanol:water (95:5) and solvent B of isopropanol:water (99:1). The chromatographic conditions were as described in Table 1 .
Injection volume was 10 l. MS 2 detection was performed with argon as collision gas at a fl ow rate of 0.15 ml/min. The source temperature was set to 150°C, while the desolvation temperature was set to 350°C. The spray was started by applying 2,500 V to the fi xed capillary. Cone voltage was set to 50 V and the optimal collision energy for parent scan of 96.8 Da was 70 eV in negative ion mode. Quantifi cation was performed by single reaction monitoring (SRM) (see supplementary Table I for the transition list). For quantifi cation, nonendogenous species (d18:1,14:0; d18:1,19:0; d18:1,27:0) of SM4s and SM3 were used as internal standards ( 28 ). The response correction was performed as described in the fi rst section ( 32 ), which is based on a previous mass spectrometric study of sulfatides ( 33 ). Retention times and corresponding correlation curves are plotted in supplementary Fig. I.
For the detection of neutral glycosphingolipids and sulfatide SM4s in the positive mode, the column was equilibrated with solvent A + , which consisted of methanol:water (95:5) with 1 mM ammonia formate and 0.05% formic acid . Compounds were eluted running a gradient with increasing percentage of solvent B + [isopropanol:methanol (99:1) with 1 mM ammonia formate and 0.05% formic acid] as described in Table 2 . A volume of 10 l was injected per sample. MS 2 detections were performed with argon as collision gas at a fl ow rate of 0.15 ml/min. The source temperature was set to 90°C, while the desolvation temperature was set to 250°C. The spray was started by applying 2.5 kV to the fi xed capillary. The cone voltage was set to 50 V. The optimal collision energy for each transition is listed in Table 3 .
Quantifi cation was performed by SRM (see supplementary  Table I for the transition list). For quantifi cation, the following nonendogenous species were used as described before ( 34 )

RNA isolation and quantitative real-time PCR
Kidneys from 6-week-old wt (C57BL/6N) mice were rapidly dissected into papillae, cortex, and medulla. Total RNA was extracted using Trizol reagent (Life Technologies) followed by DNase digest with TURBO DNA-free kit (Ambion), both according to the manufacturers' instructions. Subsequently, RNA integrity was assessed using the Agilent Bioanalyzer1000 (Agilent) prior to reverse transcription with SuperscriptII (Life Technologies). The resulting cDNA was diluted 1:10 in diethylpyrocarbonatetreated water and analyzed by SYBR-green-based quantitative real-time PCR. Expression values were normalized to Gapdh and relative mRNA expression was calculated according to Livak and Schmittgen ( 35 ). Primers and annealing temperatures are specifi ed in Table 4 .
Furthermore, MALDI IMS ( Fig. 2 ) and LC-MS 2 [ Fig. 5B, D , see also ( 32 )] analyses suggested the majority of sulfatides to contain very long acyl chains (C20 р acyl chain р C26) within their Cer anchor (C40-C44, and in part C38). When analyzing renal sections from CerS2-defi cient mice in parallel, we found a strong reduction of C44-sulfatide species together with C42-sulfatides ( Figs. 2, 5 ; including unsaturated species, Fig. 5 , supplementary Fig. III), but no complete depletion. Surprisingly, in striking contrast to previously published data for Cers and sphingomyelins ( 29 ), C40-sulfatide levels appeared almost unchanged and those of C38-sulfatides even seemed to increase. Similar to the previous results of increasing C34-Cer and C34-sphingomyelin levels in CerS2-defi cient kidneys, we observed a compensating increase of C34-species for NSsulfatides. However, AS-sulfatides and NP-sulfatides failed to compensate in this manner, which resulted in an almost complete depletion of cortical NP-sulfatides (SM3) ( Fig. 2 , CerS2 KO). Parallel analysis of total renal extracts from wt and CerS2-defi cient kidneys corroborated these fi ndings: only a signifi cant reduction of NS-species and a complete depletion of NP-species were observed for the minor compound SM3, whereas the overall levels of other SLs did not alter or even increased slightly ( Fig. 5 ). Intriguingly, compensation was achieved not only by increased expression of C34/C16-species, as had been initially observed for sphingomyelins and Cers, but substantially by C38-species. AS-sulfatides, especially, compensated almost exclusively with C38-species ( Fig. 5 ).
Most strikingly, sulfatides with the largest Cer anchor (C44) were highly restricted to the papillae ( Fig. 2 , NS-SM4s and NS-SM3 of wt). Here, roughly every third NS-sulfatide, whether SM4s or SM3, contained this C44-backbone (45% of NS-sulfatides and 43% of all sulfatides within the papillae, as determined by LC-MS 2 ). In total renal extracts, these species contributed only 10% to all NS-sulfatides and made up less than 2% of all renal sulfatides ( Fig. 5 ). Hence, C44-sulfatides were roughly 20-fold enriched over all sulfatides in the papillae, which refl ects the papillae making up a very small amount of the total kidney by weight. Analyzing 6-week-old females (supplementary Fig. IV) version 5.04 software. Isotope distributions were calculated with the isotope distribution calculator from the webpage of Scientifi c Instrument Service, Ringoes, New Jersey .

Imaging MS reveals CerS2-and Cer anchor-dependent specifi c expression patterns for sulfatides in the cortex, medulla, and papillae
Analysis of mouse kidneys by imaging MS (IMS) demonstrated a Cer anchor-dependent expression of sulfatides within the major regions of the kidney. Sulfatides containing nonhydroxy fatty acids and phytosphingosine in their Cer anchor [Cer anchor with a nonhydroxy acyl chain and a phytosphingosine (NP)-SM3] were almost exclusively found in special cortical regions. The medulla, however, was about 4-fold (10-to 18-fold by LC-MS 2 ) enriched in sulfatides carrying alpha-hydroxy acyl chains connected to sphingosines (AS-SM4s). These AS-SM4s appeared to emanate in stripes into the cortical regions. In correlation with the LC-MS 2 data, signals for corresponding AS-SM3 were below detection limits in normal kidneys by IMS, which is in contrast to data obtained from kidneys of the storage disease ASA Ϫ / Ϫ mouse ( 36 ). Finally, sulfatides with nonhydroxy fatty acid moieties N-linked to sphingosines (NS-SM4s and NS-SM3) were most abundant in the papillae, but almost undetectable in the cortex ( Fig. 2 ). The only major unsaturated sulfatide, NS-SM4s containing nervonic acid (C24:1), was distributed similarly to the saturated counterpart with lignoceric acid (C24:0; see supplementary Fig. III).
These results were validated by LC-MS 2 analysis of surgically separated cortical, medullar, and papillar renal tissue ( Fig. 3 ). IMS, however, is able to identify more subtle differences in distribution than LC-MS 2 . Thus, when plotting the sum of NP-sulfatide against the sum of AS-sulfatide signals within a single IMS picture, we could demonstrate that the cortical fraction of AS-sulfatides was restricted to

Cers1
Forward 5-TGA CTG GTC AGA TGC GTG A-3  60  Reverse  5-TCA GTG GCT TCT CGG CTT T-3  Cers2  Forward  5-TCA TCA TCA CTC GGC TGG T-3  56  Reverse  5-AGC CAA AGA AGG CAG GGT A-3  Cers3  Forward  5-ATC TCG AGC CCT TCT TCT CC-3  58  Reverse  5-CTG GAC GTT CTG CGT GAA T-3  Cers4  Forward  5-TGC GCA TGC TCT ACA GTT TC-3  60  Reverse  5-CTC GAG CCA TCC CAT TCT T-3  Cers5  Forward  5-TCC ATG CCA TCT GGT CCT A-3  56  Reverse  5-TGC TGC CAG AGA GGT TGT T-3  Cers6  Forward  5-GGG TTG AAC TGC TTC TGG TC-3  56  Reverse  5-TTT CTT CCC TGG AGG CTC T In papillae, Cers and GSLs, including sulfatides, are highly enriched in C20-sphingosine content Regarding the C44-Cer anchor of papillar sulfatides, we initially thought of a combination of the most prominent sphingoid base C18-sphingosine linked to a rather rare C26-acyl chain (cerotic acid). Astonishingly, other SLs, i.e., sphingomyelins and especially the sulfatide precursor Cers and HexCers [in kidneys basically galactosylceramide (GalCer)] with a C18-sphingosine did not contain significant amounts of this C44-backbone in any renal region [ Fig. 6A (Cer), Fig. 6B (HexCer), data for sphingomyelin not shown]. Therefore, we thought of an alternative structure consisting of an elongated C20-sphingosine linked to the prominent C24-fatty acid (lignoceric acid). Indeed, screening for Cers and HexCers with a C20-sphingosine moiety using the C20-sphingosine-specifi c MS 2 transition to m/z 292 revealed high concentrations of C20-sphingosine containing Cers and HexCers, especially in combination with C22-(behenic acid) and C24-acyl (lignoceric acid) moieties within the papillae ( Fig. 6A, B ). As we had quantifi ed sulfatides by LC-MS 2 in the negative mode using the MS 2 transition to m/z Ϫ 97, which refl ects the sulfate group, we initially did not obtain any information about the composition of the Cer anchor. Furthermore, only AS-sulfatides, but not NS-sulfatides, produce minor product ions in ESI-MS 2 that are indicative of the incorporated sphingoid base [( 33 ) and supplementary Fig. II in ( 32 )]. Surprisingly, every HexCer (likely Gal-Cer, which is the precursor of SM4s) showed two instead of one peak in the SRM chromatograms on the LC-MS 2 system. These transitions were obtained in positive ESI mode and the transition is indicative for the incorporated sphingoid base. For each of these transitions, the second peak with higher retention time refl ected the expectations of the standard HexCer retention times, whereas the fast eluting fi rst peak refl ected exactly the retention times of the corresponding sulfatides, when measured with the identical gradient by LC-MS 2 ( Fig. 7 ). Hence, with a certain probability, protonated sulfatides undergo in source decay (ISD) in positive ESI conditions and lose the sulfate group. Then they behave exactly like their biosynthetic precursor GalCer and are detected with the transitions specifi c for the group of GalCers (HexCers). We had already earlier observed an ISD for Cers and HexCers, which resulted especially in water loss ( Fig. 8 ). The advantage of the observed ISD for sulfatides yielding mainly HexCer in positive mode was that these transitions are specifi c for the size of the sphingoid base. Hence, positive mode detection allowed sphingoid base-dependent sulfatide quantifi cation demonstrating only marginal signals for sulfatides with a C26-acyl chain, but very prominent signals for C20-sphingosine-containing sulfatides in extracts of the renal papillae. The latter appeared especially in combination with C22-and C24-acyl chains. These profi les were very similar to those observed for Cers and HexCers ( Fig. 6C, 7 ).
With this information and using ISD, we reanalyzed the composition of Cers, HexCers, and SM4s sulfatides in total extracts of wt and CerS2-defi cient kidneys. Indeed, C20sphingosine-containing SLs were minor components of total renal extracts. These data also confi rmed a significant compensation of sulfatides with acyl chains longer than C22 by those with C20-acyl chains in CerS2-defi cient kidneys ( Fig. 9 ).

Region-specifi c mRNA expression of enzymes involved in SL biosynthesis
Local variations in the expression and activity of enzymes involved in GSL anabolism might be responsible for the observed local different sulfatide structures. Therefore, we sought to investigate regional mRNA expressions of the corresponding (iso-)enzymes.
SL biosynthesis starts with the activity of SPT, a heterodimer consisting of two of the three subunits serine palmitoyl-CoA transferase long chain (Sptlc)1, Sptlc2, and Sptlc3 ( 38 ). Sptlc3 appears to synthesize shorter C16-sphingoid bases ( 16 ), which were not present in renal sulfatides. Expression of Sptlc1 and of Sptlc2 was observed in all renal regions, whereas that of Sptlc3 was not detectable ( Fig. 10 ,  supplementary Fig. VI).
The Sptlc-heterodimer is further regulated by the serine palmitoyl-CoA transferase small subunits (Sptss)a and Sptssb. Sptssa supports the production of C18-sphingoid bases and Sptssb guides the complex toward C20sphingoid base production ( 15 ). We observed Sptssa in all renal regions with highest expression in the papillae and cortex. Sptssb expression was low and could only be detected in the papillae ( Fig. 10 , supplementary Fig. VI), which is in line with the particular presence of C20-SLs in the papillae.
For the synthesis of higher SLs, activated fatty acids are required. It is currently assumed that fatty acids are ␣ -hydroxylated by fatty acid 2-hydroxylase (Fa2h) before being incorporated into Cers, resulting in renal AS-sulfatides. Fa2h expression was highest in the medulla and papillae and lowest in the cortex ( Fig. 10 , supplementary Fig. VI), correlating with the high medullary AS-sulfatide levels.
SLs contain a high degree of very long chain fatty acids which requires the action of special elongases, Elovl1-7. Except for Elovl3, all Elovl-mRNAs were detected in the three renal compartments. Elovl7 and Elovl1 can elongate saturated acyl chains from C16/18 up to C24/C28 (Elovl7 also elongates polyunsaturated fatty acids) ( 8,11,39 ), and saturated fatty acids are especially found in SLs ( 40 ). Both Elovl1 and Elovl7 were present in all regions, with Elovl1 concentrating in the cortex and Elovl7 being mostly expressed in the papillae and medulla ( Fig. 10 , supplementary Fig. VI).
The condensation of sphingoid bases with activated fatty acids is catalyzed by one of the six CerSs in mice. CerS2 (preferring C22-and C24-acyl-CoAs) was the most highly expressed, followed by CerS4 (preferring C20-acyl-CoA), CerS5, and CerS6 (both preferring C14-and C16acyl-CoAs), whereas CerS1 (preferring C18-acyl-CoA) and CerS3 (being quite promiscuous) were almost undetectable (supplementary Fig. VI). Out of the prominent CerSs, only CerS4 concentrated signifi cantly in the papillae over the other regions with the lowest levels found in the cortex.
Subsequent to the condensation of sphingoid bases with acyl chains via the CerSs, the desaturases, gene of . The highest concentrations of sulfated GSLs were recorded in the medulla followed by the papillae (76% of the medulla) and more than 10-fold less sulfatide concentrations in the cortex (8% of the medulla). In the medulla, 74% of all sulfatides are AS-sulfatides, whereas in the papillae, 91% are NS-sulfatides. Finally the concentration of NP-sulfatides is 17-fold higher in the cortex than in any other region. The loss of species with C22-C26 fatty acid chain length is compensated via increased concentration of species with C16-C20 fatty acid chain length. C: The total amount of AS-SM4s shows the same behavior (D), while for these molecules no species with palmitic acid are detected (n = 3; n.s. = not signifi cant; * P < 0.05; ** P < 0.01; *** P < 0.001 ).

Reducing acyl chain length in the Cer anchor of sulfatides by about two carbon units does not affect urinary pH
Inspired by our latest fi ndings ( 6 ) that: i ) papillar sulfatides are important for the renal ammonia exchange process, ii ) urinary pH decreased in mice with renal sulfatide defi ciency, and iii ) strong reduction of the characteristic very long chain sulfatides especially in the papillae of CerS2deficient mice; we monitored the urine pH of CerS2defi cient mice over 24 h of diet. No signifi cant differences compared with the wt were observed, however (wt, pH = 5.99 ± 0.38; CerS2-KO, pH = 6.04 ± 0,31; n = 14).

DISCUSSION
Using MALDI IMS for the histological analysis of renal sulfatides, we discovered special sulfatides with a C44-Cer anchor as highly concentrated markers of the papillary region. As the papillae contribute by not more than 5% to the total kidney wet weight, these sulfatides make up less than 2% of all sulfatides within total kidney extracts. To reveal the structure of the C44-sulfatides, negative ion mode MALDI-MS 2 , either directly from tissue sections or from corresponding extracts, was too insensitive to collect relevant specifi c fragments, which are among the very low abundance product ion fraction recorded by collisioninduced dissociation (CID)-MS 2 spectra from sulfatides ( 44 ). To determine the type of sphingoid base and acyl Fig. 6. Cer-anchor compositions differentiating C18-and C20sphingosine residues of renal Cers (A), HexCers (B) (mainly Gal-Cer), and SM4s sulfatides (C) with a NS-backbone from extracts of cortex (blue), medulla (red), and papillae (green) as determined by LC-(ESI)MS 2 (positive ESI). C18-sphingosine (d18:1) and C20sphingosine (d20:1) containing SLs were detected by their transition to m/z 264 and to m/z 292, respectively. For saturated N-linked acyl chains, only the number of carbon atoms, i.e., 16,18,20,22,23,24, and 26, is annotated, whereas for nervonic acid, the number of double bonds is indicated behind the colon (i.e., 24:1) in addition. Sulfatides were detected in positive ESI with their ISD to Hex-Cer and subsequent transition to either m/z 264 or to m/z 292, respectively. chain of the Cer anchor of sulfatides, we established a new LC-MS 2 method in which sulfatides were recorded indirectly in positive ESI mode as their precursor GalCers. Probably due to the acceleration of ions within the step wave of the Xevo TQ-S instrument at a relatively low vacuum, molecular ions partially undergo low energy CID. This induces ISD of sulfatides to produce protonated Gal-Cers before mass analysis is performed. Hence, sulfatides mimic their precursor lipids, GalCers, which then are specifi cally detected in SRM mode by their CID transition to sphingoid base-specifi c fragments. Combining LC with the MS 2 detection, sulfatides and their precursor GalCers, which are also present in the renal samples, are easily differentiated by their chromatographic retention times (chromatographic resolution R ≈ 6). To our knowledge, this is the fi rst method to quantify unmodifi ed sulfatides while simultaneously obtaining structural information on their sphingoid base and fatty acid composition. Analysis of papillary extracts by this method revealed the papillary C44-sulfatides to contain C20-sphingosine instead of the otherwise dominant C18-sphingosine. Also sulfatides with a shorter C40-and C42-Cer anchor were substantially built on the bases of C20-sphingosine. In fact, more than 50% of the papillary sulfatides were based on C20-sphingosine, where as in the medulla, the concentration of C20-sphingosinecontaining sulfatides was fi ve times lower, and in the cortex they were undetectable. We also observed a similar gradient of C20-sphingosine-containing sulfatides in human kidneys (supplementary Fig. VII). C20-sphingosine had been reported as a major sphingoid base of bovine renal papillae ( 45 ), but the acid hydrolysis used obscured the SLs it was originating from. Interestingly, C20-sphingosine is the major sphingoid base in sulfatide-rich salt (nasal) glands from Eider duck and Herring Gull. Here the sulfatide content correlates with changes of Na + -K + -ATPase activity upon NaCl loading ( 4 ). These results were obtained after SL hydrolysis, but the sphingolipid from which the C20-sphingosine originated was not documented . Analyzing intact compounds, we observed a similar concentration gradient for C20-sphingosinecontaining Cers and HexCers (most probably GalCers). Initially, these structures were probably not recognized because: i ) total renal extracts contain less than 2% of them; ii ) common MS 2 screenings for Cers and HexCers with the fragment m/z 264 (derived from doubly dehydrated sphingosine) are C18-sphingoid base specifi c; and iii ) there is no difference in mass between sulfatides containing C20-sphingosine and sulfatides with C18-sphingosine and an additional C 2 H 4 -unit in the fatty acid. Only for C44-sulfatides, there is no overlap with a C18-sphingosinecontaining sulfatide, as cerotic acid (C26:0) is basically not present in sulfatides. Therefore, only these C44-sulfatides refl ected a papillae-specifi c signal in IMS working in MS 1 mode . This biological example of differentially located structural isomers also demonstrates that high resolution MS will not always be suffi cient for complete compound identifi cation.   However, the simultaneous documentation of sulfatide tissue distributions according to their membrane anchor compositions reveals new aspects in the complexity of cell type-specifi c pathways, which were not observed by previous immunocytochemical studies ( 46 ), as the anti-sulfatide antibodies recognize only the sulfatide head groups (sulfated glycan moieties). Taking quantitative aspects into account, we could document with MALDI IMS systematic distribution differences of sulfatides, which were based exactly on these differences of their Cer anchor moieties. These differences refl ect cell-specifi c enzyme expression/ activity variations for sulfatide biosynthesis for which we analyzed mRNA expression patterns. Hence, NS-sulfatides (containing nonhydroxy fatty acids and sphingosines, no matter if SM4s or SM3) with very long acyl chains were synthesized in the papillae, which was refl ected by high CerS2-mRNA, Degs1-mRNA, cerebroside galactosyltransferase (Cgt; UDP-galactosyltransferase 8A or UDPgalactose-ceramide-galactosyltransferase)-mRNA, and cerebroside sulfotransferase (Cst)-mRNA, but lower Fa2h-mRNA levels. In addition, the medulla expressed high mRNA levels of Fa2h, which introduces the ␣ -hydroxy group into fatty acids ( 47,48 ). Indeed, the bulk of AS-sulfatides concentrates in this region. In all regions, sulfatides with C18-sphingoid bases were found, which is in line with a general expression of the small a-subunit of the SPT complex (Sptssa). The longer C20-sphingosine, however, concentrated highly in papillary NS-sulfatides, which was refl ected by the presence of the Sptssb-mRNA (encoding the small b-subunit) in the papillae, but not in the medulla and cortex. When incorporated into the SPT complex, the b-subunit had been demonstrated recently to trigger the production of C20-sphingosine, whereas the a-subunit leads to the production of C18-sphingoid bases ( 15 ). Even in the papillae, Sptssb-mRNA levels were quite low as compared with those of the a-subunit. For a detailed study, protein levels of the two subunits would have to be compared. In the cortex, in line with minor Cst-mRNA levels, overall sulfatide levels were low, but NP-sulfatides expressing phytosphingosine were exclusively detected here. This was not refl ected by the relatively minor cortical mRNA levels of Degs2, which is needed for phytosphingosine synthesis ( 42 ). As only special minor subcortical regions contained these sulfatides, it may be reasoned that the corresponding cell-specifi c mRNA levels are not represented correctly by the average cortical mRNA levels. Furthermore, mRNA levels may not directly correlate with enzyme activities and always have to be discussed with caution.
Mice are often taken as model organisms for human diseases, which works only if the corresponding molecules facilitate the same functions in both organisms. One prerequisite for this is a similar local expression. Therefore, we investigated the local sulfatide anchor composition in the papillae, medulla, and cortex of two human cases. Here also, the highest expression of NS-sulfatides was found in the papillae. More interestingly, C20-sphingosinecontaining sulfatides were also restricted to the papillae in human kidney. Furthermore, they contained, basically, only very long acyl chains like NS-sulfatides of mouse papillae (supplementary Fig. VII). This similarity triggers speculations on special papillary sulfatide function in mammals that might depend on these particular Cer anchors.
While recently the involvement of renal sulfatides for papillary ammonia transport could be demonstrated in vivo ( 6 ), functional reasons for such diverse sulfatide anchor expressions within different renal cell types remain elusive. As Stettner et al. ( 6 ) discussed sulfatides of the papillae to regulate ammonia homeostasis, we hypothesized that the abundant very long acyl chains of papillary sulfatides may contribute to these sulfatide functions. Therefore, we investigated CerS2-defi cient mice that had lost most of the very long acyl chains in neutral SLs ( 29 ), but could not observe similar differences in urinary pH as had been documented in sulfatide-defi cient kidneys. Detailed analysis revealed no changes in the overall sulfatide levels, only the minor population of cortical NP-sulfatides were absent. As expected from previous investigations of Cers, sphingomyelins, and HexCers in CerS2-defi cient mice ( 29,49 ), sulfatides with C41-up to C44-acyl moieties were signifi cantly decreased, however, those with C40-acyl moieties did not change signifi cantly. In contrast to NS-ceramides and NS-sphingomyelins, sulfatides and SLs with AS-moieties compensated C41-to C44-SLs signifi cantly with C20-acyl chains. This may refl ect a compensatory role of CerS4, which preferentially incorporates C22-acyl chains, but may also use C20-CoA ( 13,50 ). Besides in the papillae, CerS4 is especially expressed in the medulla of control kidneys, where AS-sulfatides concentrate. An additional upregulation of CerS4-mRNA levels in mutant tissue cannot be excluded, and may further contribute to the rather unchanged levels of C22-sulfatides. However, CerS4 principally could also use C24-acyl-CoAs. Why this potential is not utilized in vivo remains to be investigated. A compensatory role by CerS3 in mutant mice is rather unlikely due to the very low mRNA levels detected. Furthermore, especially with CerS3, we would have expected a higher capacity to compensate for C24-than for C22-sulfatides, and especially for C20-sulfatides ( 12 ).
In summary, IMS is a unique and valuable tool to dissect local tissue lipid changes that may occur upon metabolic disease, metabolic stimulus, or in corresponding model organisms. By that, it reveals a new level of complexity of cell type-specifi c lipid expression patterns which hint at the individual fi ne tuning of cell functions. To dissect the physiological functions of lipids in defi ned cell populations, cell type-specifi c mouse models have been and will be developed. The phenotype of fi ne tuning defects, however, cannot be expected to be obvious, and revealing it will often remain diffi cult. Nevertheless, the precise evaluation of cell type-specifi c mouse models of lipid metabolism again will depend on location-dependent lipid data which can be generated by IMS, but no other MS technique.