Global site-specific N-glycosylation analysis of HIV envelope glycoprotein

Abstract

HIV-1 envelope glycoprotein (Env) is the sole target for broadly neutralizing antibodies (bnAbs) and the focus for design of an antibody-based HIV vaccine. The Env trimer is covered by ∼90N-linked glycans, which shield the underlying protein from immune surveillance. bNAbs to HIV develop during infection, with many showing dependence on glycans for binding to Env. The ability to routinely assess the glycan type at each glycosylation site may facilitate design of improved vaccine candidates. Here we present a general mass spectrometry-based proteomics strategy that uses specific endoglycosidases to introduce mass signatures that distinguish peptide glycosites that are unoccupied or occupied by high-mannose/hybrid or complex-type glycans. The method yields >95% sequence coverage for Env, provides semi-quantitative analysis of the glycosylation status at each glycosite. We find that most glycosites in recombinant Env trimers are fully occupied by glycans, varying in the proportion of high-mannose/hybrid and complex-type glycans.

ARTICLE

Received 12 Oct 2016 | Accepted 15 Feb 2017 | Published 28 Mar 2017

Liwei Cao1, Jolene K. Diedrich2, Daniel W. Kulp3,4, Matthias Pauthner3,4, Lin He2, Sung-Kyu Robin Park2, Devin Sok3,4, Ching Yao Su3, Claire M. Delahunty2, Sergey Menis4, Raiees Andrabi3, Javier Guenaga3,Erik Georgeson3, Michael Kubitz3, Yumiko Adachi4, Dennis R. Burton3,4, William R. Schief3,4, John R. Yates III2 & James C. Paulson1,2,3

HIV-1 envelope glycoprotein (Env) is the sole target for broadly neutralizing antibodies (bnAbs) and the focus for design of an antibody-based HIV vaccine. The Env trimer is covered by B90 N-linked glycans, which shield the underlying protein from immune surveillance.

bNAbs to HIV develop during infection, with many showing dependence on glycans for binding to Env. The ability to routinely assess the glycan type at each glycosylation site may facilitate design of improved vaccine candidates. Here we present a general mass spectrometry-based proteomics strategy that uses specic endoglycosidases to introduce mass signatures that distinguish peptide glycosites that are unoccupied or occupied by high-mannose/hybrid or complex-type glycans. The method yields 495% sequence coverage for Env, provides semi-quantitative analysis of the glycosylation status at each glycosite. We nd that most glycosites in recombinant Env trimers are fully occupied by glycans, varying in the proportion of high-mannose/hybrid and complex-type glycans.

DOI: 10.1038/ncomms14954 OPEN

Global site-specic N-glycosylation analysis of HIV envelope glycoprotein

1 Department of Cell and Molecular Biology, The Scripps Research Institute, La Jolla, California 92037, USA. 2 Department of Chemical Physiology, The Scripps Research Institute, La Jolla, California 92037, USA. 3 Department of Immunology and Microbial Science, The Scripps Research Institute, La Jolla, California 92037, USA. 4 Department of the IAVI Neutralizing Antibody Center, The Scripps Research Institute, La Jolla, California 92037, USA. Correspondence and requests for materials should be addressed to J.C.P. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 8:14954 | DOI: 10.1038/ncomms14954 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14954

Despite intensive efforts, no effective human immunodeciency virus type 1 (HIV-1) vaccine has yet been developed. However, there is promise for the future since

broadly neutralizing antibodies (bNAbs) have been shown to protect against virus challenge in animal models13. Because HIV-1 envelope glycoprotein (Env) is the target for bNAbs, it is an attractive immunogen for eliciting a bNAb response1,46. Each protomer of the HIV-1 Env trimer contains 2630 N-glycans that shield most of the polypeptide surface from immune surveillance710. More importantly, many bNAbs show interactions with specic N-linked glycans on Env that are required for binding and neutralization potency1118. For example, structures of glycan-dependent bNAbs of the PGT121 family in complex with Env have revealed a supersite comprised of a dense array of glycans that are predominately high-mannose-type11,1517. In contrast, bNAbs PG9/PG16 recognize an epitope on the V1V2 loop involving a Man5GlcNAc2 glycan at N160 and a terminal sialylated complex-type glycan at N156 or N173 (ref. 18). Neutralization studies using glycan deletion mutants of HIV-1 pseudoviruses demonstrated the importance of glycans at specic glycosylation sites for the binding and neutralization of other bNAbs12,19. Adding to the complexity, the glycan structures on Env have been shown to be inuenced by various factors, such as the type of cell lines used to generate Env2022, purication methods23, structural constraints23,24, and whether Env is puried from virus, pseudovirus, or expressed recombinantly25,26. For all these reasons a robust and routine method for analysis of glycosylation status at each glycosite of HIV-1 Env to support rational design and development of vaccine immunogens is needed.

Analysis of the site-specic glycosylation of HIV Env is challenging due to the fact that the Env gp160 monomer contains 2630 N-linked glycosylation sites with multiple glycan structures at each site24,2731 and millions of different Env sequences in the viruses that are circulating worldwide17. Over the past decade mass spectrometry-based methods have yielded signicant insights24,25,2733. HIV Env contains a mixture of both high-mannose-type glycans (Man5-9GlcNAc2-Asn), and glycans that are further processed to the complex-type with a Man3GlcNAc2-Asn core, 24 branches and optionally terminated with sialic acid. Notably, some glycosylation sites are exclusively high-mannose while others are almost exclusively complex-type, demonstrating that the host cell processing machinery differentially processes the glycans at each site24,2731. Comparison of recombinant membrane gp160 Env trimer with the cleaved Env (gp120/gp41) analogous to the form in infectious virus revealed that gp160 glycans undergo a much higher degree of processing to complex-type glycans, which corresponded to more open congurations of the gp160 Envs that could provide greater access for the glycan processing machinery23. Immunization studies further showed that cleaved Env, but not uncleaved Env, immunogens induced neutralizing antibodies (NAbs) against the autologous tier 2 viruses5.

In the most thorough study of Env glycosylation to date, Crispin et al.22 employed matrix-assisted laser desorption/ ionizationtime of ight mass spectrometry (MS) and liquid chromatography-tandem mass spectrometry (LC-MS/MS) to analyse site-specic glycosylation of BG505 SOSIP.664 trimer, resulting in the identication of 2026 out of 28 glycosites24,34. The method involved separately elucidating the many glycans present in the mixture cleaved from the intact Env trimer. Glycosylated peptides were puried using hydrophilic interaction chromatography-based enrichment and analysed by LC-MS/MS searching for molecular weight species corresponding to molecular weigths of glycopeptides with the library of glycans, providing a detailed analysis of glycoforms for the sites detected.

Because non-glycosylated peptides were discarded, information about the site occupancy was absent in this study. While the method requires time consuming manual interpretation of a large volume of data collected during LC-MS/MS, it provides high quality detailed information about glycans at each of the glycosites detected24.

Here we present a method that enables us to assess the site occupancy and proportion of high-mannose and complex-type glycans at all glycosites of HIV-1 Env, and is in principle applicable to any glycoprotein. The method employs digestion with multiple proteases to improve sequence coverage and generate multiple peptides for each glycosylation site. Peptide mixtures are treated sequentially by two endoglycosidases, Endo H followed by PNGase F, resulting in the introduction of novel masses for peptides that contain high-mannose glycans, complex-type glycans or no glycan. Of note, if hybrid type glycans are present in target glycoproteins, they are included in the category of high-mannose glycans as Endo H is able to release both of high-mannose and complex-type glycans. Data is analysed by powerful proteomics software that has been optimized to identify and calculate the abundance of each peptide using ion intensity peak area of MS signals. This reduces the manual analysis of data and provides a semi-quantitative assessment of the three states of glycosylation for each glycosite. Here we evaluate the impact of various factors on the glycosylation of HIV-1 Env, including purication methods, quaternary structure, glycosite mutations, and alternative patterns of glycosites generated by different virus strains. The results show that this method serves as a robust tool for characterization of site-specic glycosylation of HIV-1 Env to facilitate the rational design and development of vaccine immunogens.

ResultsStrategy. We devised a proteomics-based strategy to assess the degree of glycan processing and the degree of site occupancy of each glycosite, dened as the N-glycosylation sequon N-X-S/T (Fig. 1a). To this end, we used sequential treatment with two endoglycosidases common to glycoproteomics workows to introduce novel mass signatures for peptides that contain the two types of glycans29,35,36. First, Env peptides are digested with Endo H to cleave high-mannose (and hybrid) glycans between the innermost GlcNAc residues, leaving a GlcNAc attached to the Asn (N 203). Subsequent PNGase F treatment removes any

remaining complex-type glycans, and in the process converts Asn to Asp resulting in a 3 Da mass shift when the reaction is

conducted in O18-water (N 3)3741. For peptides with

unoccupied glycosites these treatments produce no mass shift (N 0). With this strategy, the analytical problem is focused on

generating suitable peptides and assessing the relative abundance of the three states of glycosylation.

To maximize sequence coverage, we employed digestion with multiple proteases. Each sample was subjected to chymotrypsin, trypsin and chymotrypsin, or triple digestion developed by Gatlin et al.42 involving digestion with trypsin and the nonspecic proteases, elastase and subtilisin (Fig. 1b). Each of the three proteolytic digestions was conducted in triplicate. Proteases were denatured by heating to prevent incorporation O18 at the C-terminus in subsequent endoglycosidase treatments37. Peptides were digested sequentially with Endo H and PNGase F in O18-H2O to generate novel masses for the three states of glycosylation as described above. Conditions for each enzyme digestion were optimized to ensure complete digestion.

For each of the nine samples, LC-MS/MS data were acquired. Peptides were identied using ProLuCID4346. The abundance of each peptide, as determined by the sum of the peak areas from all

2 NATURE COMMUNICATIONS | 8:14954 | DOI: 10.1038/ncomms14954 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14954 ARTICLE

a

b

Treatment MS Protease treatment:

triple digestion, chymotrypsin,a combination of trypsin and chymotrypsin

Endoglycosidase treatment: Endo H, PNGase F

LC-MS/MS analysis: three technical replicates for all samples (n=9)

Analyse data: Identify all peptides, and assess N+0, N+3 and N+203

Convert to peak area:sum peak area for each peptide, use peptides with peak area >5E8

Glycan shield

Endo H PNGaseF

(O18H2O)

Predicted mass

No glycan

Proteases (peptides)

N

N N N

N N D

N N +0

+203

+3

High mannose (hybrid)

Complex type

c

1.0

0.8

0.6

0.4

0.2

0.0

Proportion of total

Viral membrane

1 88 133 137 156 160 185e

185h

197

234

262

276

295

301

332

339

355

363

386

392

398

406

411

448

462

611

618

625

637

d

0 88 133 137 156 160 185e

185h

197

234

262

276

295

301

332

339

355

363

386

392

398

406

411

448

462

611

618

625

637

Glycosite

Keys: High mannose Complex No glycan

Figure 1 | Schematic representation of a method and to determine site-specic glycosylation of BG505 SOSIP.664 trimer. (a) Introduction of novel masses for peptides with glycosites that contain high-mannose (hybrid) glycans, complex-type glycans, or are not glycosylated, by using Endo H treatment followed by PNGase F deglycosylation. (b) The workow of the method present in this study. (c) Scatter plot of the site-specic glycosylation of BG505 SOSIP.664 trimer puried with Ni2 /SEC. A set of peptides with N 0 (grey dots), N 3 (purple dots), and N 203 (green dots) modications that were

identied when at least one of the three had a peak area of at least45E8 was displayed. (d) Colour-coded bar graph of the site-specic glycosylation of BG505 SOSIP.664 trimer. Means.e.m. were plotted.

identied charge states24, was calculated using Census47. Ion injection time was used to further normalize the resulting peak area. In order to improve the accuracy of the method, a set of peptides with N 0, N 3, and N 203 modications were

considered only when at least one of the three had a peak area of at least 5E8. This value was determined empirically as optimal to distinguish information from spectral noise as discussed further in Supplementary Fig. 5. For each glycosite site, the proportion of high-mannose, complex-type and unoccupied was determined for each peptide containing that glycosite detected in each MS run. Then the results were averaged over all peptides in all MS runs of that sample, with the mean and standard error used as the nal result for proportion of the three glycosylation states at each site. In the event that no peptide for a given glycosite met the threshold of 5E8, spectral hits were manually assessed as a basis for quantitation for that glycosite only. Otherwise, no manual interpretation of the data was needed, allowing for high-throughput analysis of the glycosylation of HIV-1 Env.

To test the general utility of the method, we analysed three model glycoproteins, bovine fetuin, invertase produced by the yeast Saccharomyces cerevisiae, and a recombinant inuenza virus

haemagglutinin (HA) from H3N2 strain A/Victoria/361/2011 produced in HEK 293 F cells48. As expected, the three N-linked glycosites (Asn-X-Thr/Ser) of fetuin were entirely complex-type glycosylation, and site occupancy was 499% for two sites (N99, N156) and 89% for the third site (N176; Supplementary Fig. 1), consistent with the previous report on the glycosylation of bovine fetuin49. In constrast, the 14 N-glycosites of invertase produced by the yeast S. cerevisiae were completely high-mannose type glycosylation, and site occupancy was only 42 and 77% for the sites N275 and N64, respectively, while the remaining 12 were with 490% site occupancy (Supplementary Fig. 2). Of the 12 glycosites in the inuenza haemagglutinin of A/Victoria/361/ 2011, one was not occupied (N122), another was only 32% occupied (N144) and the remaining ten were fully occupied, in line with glycosylation site occupancy of H3N2 strain observed in previous studies50,51. Of those that were glycosylated, four were completely complex-type (N22, N38, N63 and N483), two were 495% high-mannose (N45 and N285) and the rest were a combination of high-mannose and complex-type (N126, N133, N165 and N246) (Supplementary Fig. 3). For glycosites N165 and N246 that are associated with interactions

NATURE COMMUNICATIONS | 8:14954 | DOI: 10.1038/ncomms14954 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14954

between HA and SP-D lung collectin, we found that both of them were mainly occupied by high-mannose glycans, in agreement with the observations in previous studies50,51. Although peak areas are primarily used for quantitation, there were 85699 spectral hits detected for each glycosylation site as summarized in Supplementary Tables 1 and 2.

We next applied this method to assess the site-specic glycosylation of the soluble recombinant HIV-1 Env trimer BG505 SOSIP.664 (ref. 52), the structure of which has been determined by crystallography and cryo-EM11,5355. To avoid biasing the specic glycan structures during purication, we started with C-terminal His-tagged trimer puried by Ni2 -NTA afnity chromatography followed by size exclusive chromatography (Ni2/SEC). Each of the 28 glycosites on the

Env monomer was identied with multiple MS/MS spectra and spectral hits ranging from 90 to more than 2,000 per site (Supplementary Table 3). Because both specic and nonspecic proteases were used, multiple peptides were identied for each glycosite. Each peptide generates a set of up to three endoglycosidase-processed peptides representing the three states of glycosylation: unoccupied (N 0), high-mannose (N 203)

and complex-type (N 3). Previous analysis of N-glycans of

recombinant BG505 SOSIP.664 trimer revealed small amounts of hybrid glycans (320%) at several glycosites24, and in our analysis these would be included in the high-mannose category (N 203)

due to the specicity of Endo H. The proportion of the three states of glycosylation for peptide sets that reached the threshold peak area of 5E8 are shown in a scatter plot in Fig. 1c. The results suggest that the sum of the information for all peptides at a single glycosite will provide a much more unbiased assessment of the glycosylation status than any single peptide. For example, with sites that have a mixture of both high-mannose and complex-type glycans (for example, 137, 355), most peptide sets show a mixture of the two types, but a few show completely high-mannose or complex-type. Even more extreme, while most of the peptide sets at glycosites N234, N276 and N339 detect predominantly high-mannose glycans ( 203), one peptide set for each site

detected only complex-type glycans ( 3). These observed biases

did not appear to be due the specic proteases used (Supplementary Fig. 4). Such analysis underscores the importance of analysing multiple sets of peptides for each glycosite to avoid erroneous conclusions. Accordingly, as an unbiased measure of all the data, we have used an average value calculated from all peptide sets for each glycosite. The results are displayed in a colour-coded bar graph that allows visual comparisons between glycosites (Fig. 1d).

The results provide a robust semi-quantitative analysis of site-specic glycosylation of HIV-1 Env at each glycosylation site. It is immediately evident that the majority of the BG505 SOSIP glycosites are fully glycosylated. All but four out of the 28 sites are greater than 90% occupied and none are 450% unoccupied.

Previous studies on site-specic analysis have not included an analysis of site occupancy24, or have conducted qualitative analysis that imply site occupancy to be much less28. Although most sites have a mixture of high-mannose and complex-type glycans, 14 sites have 475% high-mannose, six sites have 475% complex-type, and the remaining eight sites have a mixture of both types. In summary, it is clear that processing of glycans from high-mannose type to complex-type by the host cell glycosylation machinery is highly site-specic.

The results for all 28 glycosites are summarized using a model of the BG505 SOSIP trimer with glycans colour-coded using the system proposed by Behrens et al.24 to distinguish glycosites with predominately high-mannose-type, complex-type, or a mixture of the two (Fig. 2). For the glycosites detected and analysed by Behrens et al.24,34 we nd excellent qualitative agreement in the

proportion of high-mannose and complex glycans found in the BG505 SOSIP.664. The high-mannose patch comprised of glycosites N295, N332, N339, N386 and N392 and the adjacent glycans at N301 and 411 are almost exclusively high-mannose structures17,56. Previous structural studies have suggested that a sialylated glycan present at site N156 plays a key role for PG9/ PG16 binding13,57, so it was surprising that this site was mainly occupied by high-mannose structures. One potential explanation is that glycans on V1/V2 scaffolds crystallized with PG9 and PG16 may be processed differently than those on native Env trimer23,24. On the basis of PGT121 recognition of complex-type glycans in the glycan array, the crystal structures of PGT121 bound to complex glycans, and the proximity of PGT121 to glycosite N137, it was predicted that N137 would be complex-type11,15,58. Here we showed that this site is indeed mainly occupied by complex-type glycans (B60%). Glycosite N276 on the edge of the CD4-binding site24,56, and the highly conserved glycan N262 needed for folding of gp120 (ref. 59), were both found to be exclusively high-mannose glycans. The epitope of bNAb PGT151 is comprised of the gp120-gp41 interface of one protomer and glycans N611 and N637 of the adjacent protomer that were predicted to be complex-type glycans12,60, consistent with results here. The other glycosites located in the gp41 region, N618 and N625, were found to be partially non-glycosylated, in keeping with earlier biochemical studies suggesting incomplete of occupancy glycosites in gp41 (refs 61,62).

Validation of MS detection of glycotypes. The current method employs the specicity of endoglycosidases to introduce novel masses for peptides that contain high-mannose or complex-type glycans. One major assumption is that enzyme treatments would go to completion, avoiding miss-assignment of the glycosylation status. For example, if high-mannose glycans were not completely removed by Endo H and were subsequently cleaved by PNGase F, they would be scored as complex-type glycans. Thus, we rst optimized digestion conditions to completely remove glycans on whole glycoproteins (for example, fetuin and invertase) as monitored by gel shift assays. Another major assumption is that the endoglycosidase-processed peptides with glycosites (Asn-X-Thr/Ser) representing the three glycosylation states with Asn (N 0; unoccupied), GlcNAc-Asn (N 203; high-mannose)

or Asp (N 3) would be detected equally during LC-MS/MS

analysis63. These assumptions were directly assessed with a strategy outlined in Fig. 3a using BG505 SOSIP.664 trimer expressed in the presence of kifunensine (Kif_BG505), which should be exclusively occupied by high-mannose glycans25. High-mannose glycans on Kif_BG505 were either removed by Endo H followed by PNGase F, generating homogeneous N 203, or

alternatively by PNGase F treatment only, generating N 3 tags.

With Endo-H only, glycosites comprised 495% N 203

(green bars), conrming that the sample prepared in the presence of Kif had high-mannose glycans, and that Endo H treatment was highly effective (Fig. 3b). The low percentage of N 3 (purple bars) could be either from incomplete removal of

high-mannose glycans by Endo H, incomplete inhibition of glycan processing by Kifunensine, or false positive results generated from data processing. With PNGase F treatment only, mimicking a sample with entirely complex-type glycans, 498% of the glycosites were N 3 (Fig. 3c). We presume

low percentages of N 203 appearing at several glycosites,

(for example, N398 and N137) are likely false positives generated from data processing. The percentages of false positives increased when the threshold of peak area was reduced from 5E8 to 5E7 (Supplementary Fig. 5), whereas further raising this threshold to 5E9 resulted in losing signals from some glycosites.

4 NATURE COMMUNICATIONS | 8:14954 | DOI: 10.1038/ncomms14954 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14954 ARTICLE

a

b

185e

137

160

133

398 339 406

392

156

185h

185h

133 185e

392

363

Oligomannose content:

75%100%

0%25%

25%75%

363

462

276 185h 332

133 332

386

185e

156

411

137

411 392

411

137

637

133

197 301

185e

185e

406 363 339

462

332

301

301

197 276

406

156 160

133

262 295

137

411

398

276

185h

160 160

156

355 234

448

637

398

339

392 386

363

406

462 276 301

156

133 386

234

339

185e

392 462

406

185h

355

88

625

88 625

611

398 137 411

398

618

618

Top view

Side view - interface

c

185h

185e

392

386

185h 160

156

133

185h

185e

156

386

462

392 363

339

276

137

406

462 363

301

137 411

332

301

332

398

355

295 276234

618 611

448

295

406

625 637 411

398

637

234

355

625

88

618

611

Side view - high mannose patch

Figure 2 | Mapping of site-specic glycosylation onto the structure of BG505 SOSIP.664 trimer. The full glycosylated model was constructed using PDB: 4TVP, RosettaRemodel and GlycanRelax8890. The trimer is shown as a grey surface and the glycans are shown as spheres coloured by proportion of oligomannose content at that site. (a) Top view of the model. (b) Side view of the model: the interface of two protomers. (c) Side view of the model: the high-mannose patch.

Having two samples containing Env peptides that are identical except for glycosites with N 203 (Fig. 3b) or N 3 (Fig. 3c), we

could assess MS detection efciency by mixing them at a molar ratio of 1:1 prior to LC-MS/MS analysis (Fig. 3d). The results show that both the N 203 and N 3 glycosites were detected

for each site, with a ratio of 1.0 to 1.2, respectively, suggesting slightly increased sensitivity for the N 3. The red dotted line

represents the average proportion of high-mannose and complex-type glycans obtained in this experiment. For the most part, the site-specic ratios of high-mannose and complex glycans were similar to the average for most glycosites. To determine how corrections for differences in detection would impact the results obtained for the BG505 SOSIP.664 trimer shown in Fig. 1d, we

rst applied site-specic corrections for high-mannose and complex-type glycans (Supplementary Fig. 6b). The corrected values were within the standard error of uncorrected values except for a single site, N637, and even for that site, the proportion of high-mannose and complex-type glycans were qualitatively similar before and after correction (Supplementary Fig. 6a,b). Because spectra counts are often used as a quantitative measure, we assessed spectral counts for the three forms of the peptides ( 0, 3, 203) to determine the relative abundance

of no glycan, high-mannose and complex-type at each glycosite (Supplementary Fig. 6c). In contrast to the site-specic correction, quantitative assessment using spectral counts gave signicant differences at 14 out of 28 sites. In every case, there

NATURE COMMUNICATIONS | 8:14954 | DOI: 10.1038/ncomms14954 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14954

a

High mannose

Proteases (peptides)

Treatment

Endo H PNGaseF (O18H2O)

Predicted mass

N N N

D

N

MS

+203

+3

None

b

Endo H + PNGaseF

PNGaseF only

Mix 1:1

1

0.8

0.6

0.4

0.2

Proportion of total

0

88 133 137 156 160 185e 185h 197 234 262 276 295 301 332 339 355 363 386 392 398 406 411 448 462 611 618 625 637

c

1

0.8

0.6

0.4

0.2

0

88 133 137 156 160 185e 185h 197 234 262 276 295 301 332 339 355 363 386 392 398 406 411 448 462 611 618 625 637

d

1

0.8

0.6

0.4

0.2

0 88 133 137 156 160 185e 185h 197 234 262 276 295 301 332 339 355 363 386 392 398 406 411 448 462 611 618 625 637

Glycosite

Figure 3 | Validation of endoglycosidase treatments and MS detection for peptides that contain N 3 and N 203 modications. (a) Schematic

representation of the process. (b) Validation of Endo H treatment by using Kif_BG505. (c) Validation of PNGase F treatment was conducted by using Kif_BG505 without Endo H treatment. (d) MS detection for peptides that contain homogeneous N 3 and N 203 modications at a molar ratio of

1:1. Peptides that had potential glycosites, but were not glycosylated were not included. The proportions of high-mannose and complex-type glycans at those glycosites highlighted in yellow were assigned based on the proportion of spectra hits since peak area did not reach the threshold of 5E8 Supplementary Tables 57.

was an apparent increase in the proportion of complex-type glycans particularly in glycosites that are known to be occupied predominantly by high-mannose glycans24,56, such as N156, N262, N295, N332, N339, N386 and N392. We attribute to sensitive detection, since counts for low and high abundant species would be counted as equivalent, giving a bias for the lower abundant complex-type glycans. For these reasons, we have used peak area for all results presented, and have not applied site-specic corrections since they were not found to signicantly change the result (Supplementary Fig. 6b).

The impact of purication methods on Env glycosylation. With the method established we investigated the inuence of several factors that could affect results from recombinant HIV-1 Env samples. We rst assessed the potential for batch variation in preparations of the BG505 SOSIP.664 trimer with a C-terminal hexa histidine-tag produced in 293 F HEK cells puried using Ni2 -NTA afnity chromatography followed by size exclusion chromatography. Compared with the reference batch (Fig. 1d) we saw virtually identical results with the second batch (Fig. 4a), with only a single glycosite (N137) showing signicantly, but only marginally, lower proportion of complex-type glycans.

Use of glycan-dependent bNAbs for purication of HIV-1 Env raises the prospect of selection of and enrichment of particles that carry a favourable glycan recognized by the antibody22,23. To test this possibility directly, we compared the glycosylation of the Ni2 -NTA afnity column-puried material (Fig. 1d) with the same batch of recombinant BG505 SOSIP.664 trimer that was alternatively puried using afnity columns using bNAbs 2G12 (ref. 64), PGT145 (ref. 65) or PGT151 (refs 12,60), in each case followed by size exclusion chromatography (Fig. 4b). We found that BG505 SOSIP.664 trimer puried using 2G12, PGT145 or PGT151 had an almost identical glycosylation pattern compared with the trimer puried using Ni2 -NTA afnity chromatography, indicating that different purication methods have little impact on the glycosylation pattern of BG505 SOSIP.664 trimer.

Impact of glycans on compositions at neighbouring glycosites. Mutants of HIV-1 pseudovirus, with specic glycan sites removed, have been widely used in neutralization studies to assess the degree to which specic glycans are involved in glycan-dependent bNAbs epitopes12,15,19. Furthermore, design of candidate vaccine priming immunogens to activate bNAb

6 NATURE COMMUNICATIONS | 8:14954 | DOI: 10.1038/ncomms14954 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14954 ARTICLE

a

BG505 SOSIP (second biological batch)

BG505 SOSIP (alternative purification)

10.80.60.40.2 0

10.80.60.40.2 0

10.80.60.40.2 0

10.80.60.40.2 0

10.80.60.40.2 0

10.80.60.40.2 0

10.80.60.40.2 0

10.80.60.40.2 0

8888 133

137

156

301

332

339

406

411

448

462

611

618

625

637

b

mAb 2G12

mAb PGT145

mAb PGT151

133

137

156

160

160

185e

185e

185h

185h

197

234

262

276

197

234

262

276

295

295

301

332

339

355

363

386

392

398

355

363

386

392

398

406

411

448

462

611

618

625

637

88 133

137

156

160

185e

185h

197

234

262

276

295

301

332

339

355

363

386

392

398

406

411

448

462

611

618

625

637

Proportion of total

88 133

137

156

160

185e

185h

197

234

262

276

295

301

332

339

355

363

386

392

398

406

411

448

462

611

618

625

637

c

N276D

N197D, N276D

N197D, N276D, N301A, N386D

BG505 mutants

88

133

137

156

160

185e

185h

197

234

262

276

295

301

332

339

355

363

386

392

398

406

411

448

462

611

618

625

637

88

133

137

156

160

185e

185h

197

234

262

276

295

301

332

339

355

363

386

392

398

406

411

448

462

611

618

625

637

88 133

137

156 160 185e

185h

197

234

262

276

295

301

332

339

355

363

386

392

398

406

411

448

462

611

618

625

637

d

BG505 gp120 monomer

Glycosite

88 133

137

156 160 185e

185h

197

234

262

276

295

301

332

339

355

363

386

392

398

406

411

448

462

Keys: High mannose

Significant differences from reference sample

Complex No glycan

Figure 4 | Site-specic glycosylation of BG505 Envs. (a) second biological batch of BG505 SOSIP.664 trimer that was puried with Ni2 /SEC; (b) BG505 SOSIP.664 trimers that were puried over different bNAbs columns followed by SEC; (c) the mutants of BG505 SOSIP.664 trimer;

(d) BG505 gp120 monomer. The glycosylation of BG505 SOSIP.664 trimer shown in Fig. 1d was selected as reference for comparison. The proportions of high-mannose and complex-type glycans at those glycosites highlighted in yellow were assigned based on the proportion of spectra hits since peak area did not reach the threshold of 5E8 (Supplementary Tables 823).

NATURE COMMUNICATIONS | 8:14954 | DOI: 10.1038/ncomms14954 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14954

precursor B cells has generally involved removal of one or more glycans from Env or Env-derived proteins6671. A gap in current knowledge is the impact of removing glycosylation sites on neighbouring glycosites and the overall glycosylation of Env. To this end, we analysed three mutants of recombinant BG505 SOSIP.664 trimer with the deletion of 14 glycan sites near the CD4bs, including glycan sites N276, N197, N301 and N386 (refs 21,24,72; Fig. 4c). When compared with the glycosylation of BG505 SOSIP.664 trimer (Figs 1d and 4a), deletion of glycan site at N276 (N276D) produced only slight but signicant changes at N462 and N197. Further deletion of site N197 (N197D, N276D) gave a virtually identical glycosylation pattern. The mutant with four glycosites removed (N197D, N276D, N301A and N386D) showed signicant differences at ve sites (N160, N295, N339, N406 and N462), but again the differences were slight, and qualitatively the same as the fully glycosylated reference BG505 trimer (Figs 1d and 4a). Thus, perhaps surprisingly, removal of glycans close to CD4bs did not change access of the glycan processing machinery sufciently to alter the ratio of total high-mannose/hybrid and complex-type glycans at neighbouring sites, or the overall glycosylation of the BG505 SOSIP.664 as assessed by this method.

The impact of quaternary structure on Env glycosylation. Recombinant monomeric gp120 has been used as a common immunogen in various vaccine trials22,73. Structural constrains imposed by trimerization have been shown to impact the glycosylation of HIV-1 Env, resulting in higher content of high-mannose glycans in the trimer22,25. To assess the impact of trimerization on site-specic glycosylation, we assessed the glycosylation of the BG505 monomer (Fig. 4d). In contrast to minimal differences in glycosylation of the BG505 trimer glycosylation mutants, signicant changes in glycosylation were observed in 17 out of the 24 glycosites of the monomer relative to the trimer. As observed by others, the monomer had increased levels of complex-type glycans. We also observed a striking decrease in site occupancy at 10 glycosites relative to the trimer. Of note, highly conserved glycosites, for example, N156, N262, N295, N332, N386 and N392, were still predominantly high-mannose structures. Glycans with a dramatic shift to complex-type included N88, N160, N276 and N411 (Figs 1d and 4a,d). Overall, the site occupancy of BG505 monomer gp120 was 86%, similar to that reported previously74. The results show that the glycosylation pattern of monomeric gp120 is dramatically different from that of the native-like Env trimer, which can be attributed in part to greater accessibility of glycans at the subunit interface by the glycosylation machinery25,26. Such changes will impact the binding of glycan-dependent antibodies and should be considered when using gp120 proteins as immunogens.

Variation in site-specic glycosylation of Env trimers. The glycosylation of cleaved, native-like Env trimers derived from diverse genotypes, including JR-FL SOSIP.664 (subtype B)75, B41 SOSIP.664 (subtype B)65, CRF02_AG_250 SOSIP.664 (subtype AG)76, 327c SOSIP.664 (subtype C)77 and simian immunodeciency virus (SIV) trimer SIVcpzMT145 SOSIP.66478 were analysed and compared with that of BG505 SOSIP.664 trimer (subtype A) (Fig. 5). To facilitate comparison of glycosites, sequences were aligned using a reference HIV Env (Suppleme ntary Fig. 7). As a group, the Env trimers have similar glycosylation patterns. Almost all glycosites are fully occupied in the gp120 portion of the six trimers. The conserved glycosites17, N156, N262, N276, N301, N295, N332/334, N411/413 and N448, were mainly occupied by high-mannose glycans, and in contrast, complex-type

glycans were predominately found at sites N406 and N460-463 and in most of the glycosites of the gp41 subunit.

In addition to the similarities there were also signicant differences among the trimers. The proportions of high-mannose glycans ranged from a low of B50% for JR-FL SOSIP.664 and

CRF02_AG_250 SOSIP.664 to somewhat higher (6066%) for other isolate trimers. For example, glycans at N137 consisted predominantly of complex-type glycans on most of the trimers, but were largely high-mannose glycans on B41 SOSIP.664 and 327c SOSIP.664. Although complex-type glycans at the glycosite N137 have been shown to be involved in the formation of bNAb epitopes such as those for the PGT121 family11,15,58, complex-type glycans are apparently not essential at this position for neutralization of these strains. There was also variation in glycosites located in the canonical high-mannose patch, including N339, N386 and N392. For instance, BG505 SOSIP.664 trimer, B41 SOSIP.664, CRF02_AG_250 SOSIP.664 and SIVcpzMT145 SOSIP.664 each were occupied predominantly by high-mannose glycans at the site N386, while a high proportion of complex-type glycans was found at this site on two other trimers. In addition, high-mannose glycans dominated the site N356 on SIVcpzMT145 SOSIP.664, whereas a high proportion of complex-type glycans was detected at this site on ve other trimers. There were also variations in the site occupancies and types of glycans in the gp41 region among the six trimers. For instance, the glycosite N637 was mainly occupied by high-mannose glycans on BG505 SOSIP.664 trimer, JR-FL SOSIP.664, B41 SOSIP.664 and 327c SOSIP.664, while a high proportion of complex-type glycans was found at this site on CRF02_AG_250 SOSIP.664 and SIVcpzMT145 SOSIP.664. The glycosites located on gp41 subunits were almost completely occupied on SIVcpzMT145 SOSIP.664, while this region was largely unoccupied on other trimers. In summary, the results show a conservation of an overall glycosylation pattern for Envs of different HIV strains, but there are also signicant differences.

DiscussionWe have developed a mass spectrometry-based workow for global analysis of protein glycosylation that determines the degree of glycan occupancy and the proportion of high-mannose/hybrid and complex-type glycans at each glycosite. Although used here to analyse HIV Env, the method is in principle applicable to any glycoprotein. The strategy employs a combination of proteases, including triple digestion, chymotrypsin and a combination of trypsin and chymotrypsin, to generate multiple overlapping peptides, resulting in 495% sequence coverage of HIV-1 Env.

For those glycoproteins with fewer N-glycosites, such as fetuin and invertase, triple digestion alone is able to generate detectable peptides that contain all N-glycosites for MS analysis. A combination of triple digestion and chymotrypsin would be enough for detecting all N-glycosites in most glycoproteins. Sequential treatments with Endo H and PNGase F were employed to create different mass signatures at the Asn of each glycosite based on occupancy with high-mannose/hybrid glycan (N 203),

complex-type glycan (N 3) or no glycan (N 0). We could then

apply standard database searching and label-free analysis4345,47 to conduct high-throughput identication and semi-quantitative analysis of peptides representing the three glycosylation states of each glycosite. A complete analysis can be accomplished in a few days with only 30 mg of glycoprotein, representing a small fraction of the material and time needed to do a more detailed glycomics analysis to identify all glycoforms at each site.

Several groups have conducted detailed analysis of Env glycosylation by LC-MS/MS analysis of intact glycopeptides24,2731,79. In contrast to the method described here, glycoproteomics

8 NATURE COMMUNICATIONS | 8:14954 | DOI: 10.1038/ncomms14954 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14954 ARTICLE

a

b

c

d

e

f

BG505 SOSIP.664 trimer

B41 SOSIP.664

CRF02_AG_250 SOSIP.664

327c SOSIP.664

SIV MT145 SOSIP.664

Keys: High mannose

1

0.80.60.40.2 0

10.80.60.40.2 0

10.80.60.40.2 0

10.80.60.40.2 0

10.80.60.40.2 0

88

133 156 160 185e185h 197 234 262 276 295 301 332 339 355 363 386 392 398 406 411 448 462 611 618 625 637

137

JR-FL SOSIP.664

88 135 138 141 156 160 187 241 262 276 295 301 332 339 355 362 386 392 397 406 448 463 611 616 625 637

10.80.60.4

Proportion of total

0.2 0

88

137 140 143a 156 160 186e 187 197 234 241 262 276 295 301 332 339 355 362 386 392 396 413 448 463 611 616 625 637

88 135 141 156 160 197 241 262 295 301 332 339 354 386 392 397 401a 406 411 460 465 611 618 625 637

625

625

88

135

141

156

160 637

189

197

230

241

262

276

289

301

332

339

355

386

393

398

413

442

448

461

465

611

616

88 136 141 156 160 186 197 236 241 262 268 301 334 343 356 386 392 397 406 442 448 465 611 616 637

Glycosite

Complex

No glycan Significant differences from reference sample

Figure 5 | Site-specic glycosylation of native-like Env trimers from different HIV-1/SIV isolates. (a) BG505 SOSIP.664 trimer (subtype A), which is the same data as shown in Fig. 1d; (b) JR-FL SOSIP.664, subtype B; (c) B41 SOSIP.664, subtype B; (d) CRF02_AG_250 SOSIP.664, subtype AG;(e) 327c SOSIP.664, subtype C; (f) SIV trimer SIVcpzMT145 SOSIP.664. N-linked glycosites from multiple HIV Env trimers were aligned according to their relative position to HXB2 and their predicted functional importance. The glycosylation of BG505 SOSIP.664 trimer shown in Fig. 1d was selected as reference for comparison. The proportions of high-mannose and complex-type glycans at those glycosites highlighted in yellow were assigned based on the proportion of spectra hits since peak area did not reach the threshold of 5E8 (Supplementary Tables 2433).

workows typically provide little information about site occupancy24,79, unless a separate analysis is conducted using PNGase F-treated Env peptides and screening is carried out for peptides with Asn versus Asp, as we have done here28,31. Indeed, analysis of site occupancy is a routine outcome of our method, and we have found that there is 490% site occupancy for all but 25 of 2529 sites in each of the soluble Env trimers of 6 HIV/SIV strains examined. The one or more of the glycosites in the gp41 subunits of all 6 Env trimers except SIVcpzMT145 SOSIP.664 were not fully glycosylated, perhaps as a consequence of being released from the signal anchor before the oligosaccharyltransferase had a chance to act80,81. A number of glycosites on the BG505 gp120 monomer were also only partially glycosylated, presumably due to the different kinetics of synthesis and folding compared with the SOSIP trimer81,82. Beyond site occupancy, quantication of the glycoforms at each site by glycoproteomics is also a signicant challenge due to the multiple glycoforms at each site, and the unknown efciency for detection of peptides with each glycoform due to the impact of glycans with different number of sialic acid moieties on ionization efciency during MS analysis63,83. Thus, detection of a glycoform at a given site provides evidence for the types of glycans found at that site, but without reliable means to routinely assess the relative abundance2731,79.

In a thorough glycoproteomics analysis of glycopeptide enriched fractions from a BG505 SOSIP.664 trimer, Berhens et al.24,34 detected 2026 of the 28 glycosites and summed peak areas for each high-mannose and complex-type glycoform detected as a measure of their abundance at each site. The high-mannose and complex-type glycoforms detected at each site were consistent with the ordered processing pathway that rst trims the high-mannose structures from Man9 to Man5 before adding terminal sequences that dene complex glycans. Thus, for sites with 100% high-mannose, there was mainly Man9. For sites with a mixture of high-mannose and complex-type, there were mainly mixtures of processed high-mannose structures (Man8 to Man5) and simple complex-type structures24. Thus, as would be expected, dening the ratio of high-mannose to complex-type glycans, as we have done here, provides insight into the degree of processing that has occurred at each site. In this regard, it is notable that there is quite good concordance between the ratio of high-mannose versus complex-type glycans observed for all sites in the BG505 SOSIP trimers analysed here and by Berhrens et al.24. Thus, we suggest that the global analysis of site-specic glycosylation described here provides a great deal of information about access of each site to the processing machinery.

Given the importance of glycosylation on the immunogenicity and biological activity of HIV Env20,21,84,85, we believe that this

NATURE COMMUNICATIONS | 8:14954 | DOI: 10.1038/ncomms14954 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14954

semi-quantitative proteomics-based method lls a need to routinely get detailed information about site-specic glycosylation of Env. Indeed, the glycosylation of HIV-1 Env is potentially impacted by various factors, such as purication methods23, batch-to-batch variation, cell type used for expression24,25,31, expression as soluble or membrane form25,28, structural constraints24 and strain/genotype26,27,30. Using the native-like Env soluble trimers as a model, the method revealed little difference in site-specic glycosylation in different biological batches, or in the method of purication (Fig. 4), but signicant differences in site-specic glycan processing across various strains of HIV (Fig. 5). Such information will also inform high resolution X-ray and cryo-EM structure analysis of Env where electron density for glycans can be visualized. In one recent report on the crystal structure of BG505 SOSIP.664 trimer, we note that glycans at sites N301 and N276 were assigned complex-type, while predominately high mannose type was seen here86. So clearly there will be opportunity for these methods to provide complementary information for rening Env structure analysis relevant for vaccine design. In this regard, it will be of immediate interest to further establish the degree to which glycan processing differs between soluble, membrane bound and viral forms of Env, and the degree to which glycosylation of viral Env matches that of recombinant Env trimers or Env fragments being considered as vaccine candidates.

Finally, although we have used HIV Env as a model, the method is suitable for analysis of site-specic glycosylation of any glycoprotein. It should be particularly useful for analysis of the glycosylation of cell surface glycoproteins, viral glycoproteins and soluble recombinant glycoproteins and could be an additional tool for analysis of biotherapeutics to assess in one-step site-specic occupancy and degree of glycan processing.

Methods

Expression and purication of Env trimers and gp120. SOSIP.664 trimer52 and gp120 constructs were transiently transfected using 293 Fectin (Invitrogen) into 293F cells, cultured in 293 Freestyle media (Life Technologies) and grown for 96144 h. SOSIP.664 timers were co-transfected with furin plasmids. For the BG505 SOSIP.664 trimer produced with kifunensine (BG505_Kif), 15 uM of kifunensine (Cayman Chemical) was added to 293 Freestyle media. After harvest and sterile ltration on a 0.22 mM lter (Nalgene), the supernatants were puried using the following protocols:

His-tag purication. Unless indicated otherwise, all N-terminally His-tagged BG505 SOSIP.664 and BG505_Kif constructs were puried in two steps usinga HIS-TRAP afnity column (GE Healthcare) with a linear elution gradient from 20 mM Imidazole to 500 mM Imidazole, followed by a Superdex 200 Increase SEC column (GE Healthcare) in Hanks Buffered salt solution (HBE) (10 mM HEPES 150 mM NaCl)71. The oligomeric state and purity were assessed by size exclusion chromatography-mutli-angle light scattering (Wyatt Technology DAWN HELEOS II, Wyatt Technology Optilab T-rEX).

Antibody afnitypositive selection. In this approach, N-terminally Avi-tagged B41 SOSIP.664 and 327c SOSIP.664, as well as untagged CRF02_AG_250 SOSIP.664 and SIVcpzMT145 SOSIP.664 constructs were transfected as described above. Supernatants of B41 SOSIP.664 and 327c SOSIP.664 were subsequently puried over 2G12 antibody afnity column, while CRF02_AG_250 SOSIP.664 and SIVcpzMT145 SOSIP.664 SOSIP.664 were puried by using PGT145 antibody afnity column. Supernatants were slowly run over a CNBr-activated Sepharose4 Fast Flow (GE) conjugated antibody afnity column at 4 C overnight, eluted in3 M Magnesium chloride and buffer exchanged into TBS. The eluate was then run over a Superdex 200 Increase SEC column (GE Healthcare) in TBS and trimer-size fractions were collected. The trimeric state of all proteins was conrmed by blue-native PAGE.

Antibody afnity - negative selection. In this approach, untagged JR-FL SOSIP.664 trimers were puried by negative selection. The ltered culture supernatants were run over a Galanthus nivalis lectin (GNL) afnity chromatography column overnight at 4 C to isolate the heavily glycosylated envelope glycoproteins. The eluate containing the JR-FL SOSIP protein was run over a Superdex 200 10/300 GL SEC column to separate trimer-size oligomers from aggregates and gp140 monomers. Trimer-size JR-FL SOSIP containing fractions were loaded onto a protein A agarose column previously loaded with two-fold weight excess of non-neutralizing mAb F105 over the amount of trimeric SOSIP loaded. The column was gently rocked at 4 C for 45 min, the solid phase was

allowed to settle for 5 min and the well-ordered trimeric JR-FL SOSIP protein was recovered from the ow-throw in a negative selection manner.

Combination proteolytic digestion of Envs. The proteins (B50 mg) were denatured with 8 M urea in 100 ml of 100 mM ammonium acetate (pH 6) to minimize nonenzymatic deamidation87. The denatured proteins were reduced with 10 mM of dithiothreitol (DTT) at 56 C for 1 h, followed by iodoacetamide treatment (50 mM, 45 min, in the dark) to alkylate cysteine residues. The reduced and alkylated samples were buffer exchanged using 10 kDa centrifugal lters (Millipore). The resulting proteins dissolved in 100 mM ammonium bicarbonate (pH 8) were divided into ve equal aliquots for the following proteolytic digestion.

Triple digestion. Three out of ve aliquots were digested with different proteases by using a modied triple digestion protocol, described previously42. Arg-C followed by trypsin. One fraction was rst digested with Arg-C (Promega) at an enzyme/substrate ratio of 1:20 (w/w) in 100 ml of 100 mM ammonium bicarbonate (pH 8) containing 5 mM DTT and 0.2 mM EDTA. The reaction solution was placed at 37 C for 4 h. The resulting sample was lyophilized to remove volatile salt ammonium bicarbonate, and was re-dissolved in 500 ml of 100 mM ammonium acetate (pH 6) subsequently. Sequencing grade modied trypsin (Promega) was added to the solution at an enzyme/substrate ratio of 1:10 (w/w). The protein was digested at 37 C for 16 h. After incubation, the sample was lyophilized. Elastase. A second fraction was digested with elastase (Promega) at an enzyme/substrate ratio of 1:20 (w/w) in 500 ml of 100 mM ammonium bicarbonate (pH 8). The reaction was incubated at 37 C for 16 h. After incubation, the sample was lyophilized. Subtilisin. A third fraction was digested with subtilisin (Sigma) at an enzyme/substrate ratio of 1:20 (w/w) in 500 ml of 100 mM ammonium bicarbonate (pH 8). The digestion was conducted at 37 C for 4 h. After incubation, the sample was lyophilized. Three fractions were then re-dissolved in water and combined into one microcentrifuge tube, and the protease enzymes were denatured at 100 C for 5 min.

Chymotrypsin. A fourth fraction was digested with chymotrypsin (Promega) at an enzyme/substrate ratio of 1:13 (w/w) in 500 ml of 100 mM ammonium bicarbonate (pH 8). The reaction was conducted in a 30 C water bath for 10 h. The reaction buffer was lyophilized, and the peptides were re-dissolved in water. The protease enzyme was denatured at 100 C for 5 min.

Trypsin and chymotrypsin. The fth was digested with the combination of trypsin and chymotrypsin at an enzyme/substrate ratios of 1:20 (w/w) and 1:13 (w/w) respectively in 500 ml of 100 mM ammonium bicarbonate (pH 8). The reaction was conducted at 37 C for 16 h. The reaction buffer was lyophilized, and the peptides were re-dissolved in water. The protease enzymes were denatured at 100 C for 5 min.

Deglycosylation. The samples generated from different proteolytic digestions were separately de-glycosylated with Endo H, followed by PNGase F treatment.

Endo H. Endo H (New England Biolabs) was added to the (glyco)peptides (250 units per 10 mg) re-dissolved in 20 ml of 100 mM ammonium acetate (pH 5.5).

The reaction was incubated at 37 C for 1 h.

PNGase F. The Endo H-treated (glyco)peptides were lyophilized and then re-dissolved in 100 mM ammonium bicarbonate (pH 8) prepared with O18-H2O (97%, Sigma). PNGase F (lyophilized, New England Biolabs) re-dissolved in O18-H2O (97%, Sigma) was added to the (glycol)peptide solution (500 units/10 mg).

The reaction was allowed to proceed at 37 C for 1 h in a sealed microcentrifuge tube. The endoglycosidase enzymes were denatured at 100 C for 5 min.

Endoglycosidase validation experiment using BG505_Kif. BG505_Kif was denatured, reduced, and alkylated as described above. The protein solution was then divided into three aliquots, which were digested with triple digestion, chymotrypsin, and the combination of trypsin and chymotrypsin, respectively. The reaction buffer was lyophilized, and the peptides were re-dissolved in water. The protease enzymes were denatured at 100 C for 5 min and lyophilized. The resulting peptides generated from each proteolytic digestion were re-dissolved in 40 ml of 100 mM ammonium acetate (pH 5.5), and then divided into two equal aliquots with equal volume. One aliquot was mock digested with 100 mM ammonium acetate (pH 5.5) buffer only (no Endo H), followed by PNGaseF treatment. Another one was de-glycosylated with sequential Endo H and PNGase F treatments. The resulting samples were either directly analysed to validate endoglycosidase treatments or equal volumes were mixed to generate a sample with N 3 and N 203 modied peptides at a molar ratio of 1:1 for characterizing

MS detection of glycotypes.

Mass spectrometric analysis. The samples from HIV-1/SIV Envs and the recombinant inuenza virus haemagglutinin were analysed on a Fusion Orbitrap tribrid mass spectrometer (Thermo Fisher Scientic). Approximately 1 ug of de-glycosylated peptides were injected directly onto a 30 cm, 75 um ID column packed with BEH 1.7 mm C18 resin (Waters). Samples were separated at a ow rate of 200 nl min 1 on a nLC 1000 (Thermo Fisher Scientic). Buffer A and B were0.1% formic acid in water and acetonitrile, respectively. A 240-minute gradient was run, consisting of the following steps: 0150 min, 522% B; 150200 min, 2232% B; 200210 min, 3290% B and hold at 90% B for a nal 30 min of run time.

10 NATURE COMMUNICATIONS | 8:14954 | DOI: 10.1038/ncomms14954 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14954 ARTICLE

Column was re-equilibrated with 20 ml of buffer A prior to the injection of sample. Peptides were eluted directly from the tip of the column and nanosprayed directly into the mass spectrometer by application of 2.5 kV voltage at the back of the column. The Orbitrap Fusion was operated in a data-dependent mode. FullMS1 scans were collected in the Orbitrap at 120 K resolution with a mass range of 3501,500 m/z and an AGC target of 4e (ref. 5). The cycle time was set to 3 s, and within this 3 s the most abundant ions per scan were selected for CID MS/MS in the orbitrap with an AGC target of 5e (ref. 4) and minimum intensity of 5,000. Maximum ll times were set to 50 and 100 ms for MS and MS/MS scans, respectively. Quadrupole isolation at 1.6 m/z was used, monoisotopic precursor selection was enabled and dynamic exclusion was used with exclusion duration of 5 s.

Bovine fetuin was separately digested with triple digestion, chymotrypsin, and the mixture of trypsin and chymotrypsin. The peptides that were generated from three digestions were combined at a molar ratio of 1:1:1 after deglycosylation. Approximately 2 ug of the peptides were pressure-loaded onto a 100 mm i.d fused silica capillary MudPIT column containing 1.5 cm of 10 mm Jupiter C18 (Phenomenex) followed by 1.5 cm of 5 mm Partisphere strong cation exchange (SCX, Whatman). The MudPIT column was connected to a 75 mm capillary separation column packed with 15 cm of 4 mm Jupiter C18 (Phenomenex) and was attached to an Agilent 1200 quaternary HPLC placed in line with the heated capillary of an LTQ Orbitrap XL Mass Spectrometer (Thermo Fisher Scientic). The buffers used were A: 5% acetonitrile/0.1% formic acid, B: 80% acetonitrile/0.1% formic acid, and C: 500 mM ammonium acetate in 5% acetonitrile/0.1% formic acid. Peptides were separated using a nine-step separation. The rst step was59 min with the following gradient: 05 min, 015% B; 535 min, 1570% B; 3540 min, 70100% B; 100% B, 4045 min; 1000% B, 4546 min; 0% B,4659 min. The next eight steps were 120 min each with the following gradient: 100% A, 04.9 min; 58 min, x% C; 100% A, 8.110 min; 10085% A, 1012 min; 8550% A, 12100 min; 500% A, 100105 min; 0% A, 105110 min; 100% A, 110.1120 min. The 3 min buffer C percentage (x) in steps 29 were as follows: 20, 30, 40, 50, 60, 70, 90 and 100. As peptides were eluted from the microcapillary column, they were ionized by the application of a distal 2.5 kV spray voltage. Ionized peptides were detected with one survey scan (3002,000 m/z) in the Orbitrap with nominal resolution of 60,000 followed by 10 data-dependent MS/MS scans at 35% normalized collision energy. Charge state rejection was set to omit singly charged ion species for MS/MS but to include ions for which a charge state could not be determined. Dynamic exclusion was enabled with a repeat count:1, duration: 30.0, list size: 500, exclusion duration 120.0, exclusion mass with high/low: 1.51/0.51 m/z, early expiration enabled.

Data processing. MS and MS/MS data were extracted from RAW les by using RawConverter46, enabling monoisotopic correction. The MS/MS spectra were searched using the ProLuCID algorithm from software package Integrated Proteomics Pipeline-IP2 (version 4)4345,47 against the European Bioinformatic Institute (IPI) Bos taurus protein database including the sequences of HIV-1/SIV Envs analysed in this study. A target/decoy database containing the reversed sequences of all the proteins appended to the target database was used so that we could estimate peptide probabilities and FDRs accurately. MS tolerances were set at 50 p.p.m. for precursor ions and 20 p.p.m. (was 600 p.p.m. for the bovine fetuin data) for fragment ions. No enzyme specicity was considered for the searching. Carboxyamidomethylation ( 57.02146 C) was considered as a xed modication.

In addition, oxidation ( 15.9994 M), deamidation ( 2.988261 N), GlcNAc

( 203.079373 N), and pyroglutamate formation from N-terminal glutamine

residue ( 17.026549 Q) were searched as variable modications. The results were

ltered by DTASelect (version 2.0). The parameters were set as: minimum number of peptide per proteinZ2, spectrum false positive rater0.05, and precursor delta mass cutoffr10 p.p.m. (was 20 p.p.m. for the bovine fetuin data). The remaining peptides were further ltered to remove those peptides with N 3 and N 203

modications that were not located at the consensus motif (N-X-S/T, X can be any amino acid residue except proline). We used Census47 to reconstruct chromatograms for identied peptides. Each peak was smoothed and t to Gaussian distribution to calculate the abundance of peptide using peak area. Peak area was further normalized by using ion injection time. We aligned spectra from different samples by using precursor mass tolerance and retention time to calculate relative ratios.

Statistical analysis. The following statistical analyses were performed with GraphPad Prism 5. The ion intensity peak area of the peptides from each raw le was generated by summing the peak area of this peptide over all identied charge states24, and that was used for statistical analysis. A set of peptides with N 0,

N 3 and N 203 modications was chosen only when at least one out of the

three was identied with peak area more than 5E8. If no peptide for a given glycosite met the threshold of 5E8, proportions of the three glycosylation states at this site were generated based on spectra hits28. BG505 SOSIP.664 trimer puried with Ni2 /SEC that was shown in Fig. 1d was selected as reference, to which proportions of the three glycosylation states at each glycosite on other Envs analysed in this study were compared by using a Mann-Whitney test. Signicant differences were concluded only when the P value was o0.05 and the difference on the proportion of no glycan, high-mannose, or complex-type glycans was 410%.

Data availability. The data that support the nding of this study can be downloaded from the MassIVE site (http://massive.ucsd.edu/ProteoSAFe/dataset.jsp?task=05eced07b0f9486fb04509a4aec1197a

Web End =http://massive.ucsd.edu/ProteoSAFe/data- http://massive.ucsd.edu/ProteoSAFe/dataset.jsp?task=05eced07b0f9486fb04509a4aec1197a

Web End =set.jsp?task=05eced07b0f9486fb04509a4aec1197a ) or available from the corresponding author upon reasonable request.

References

1. Burton, D. R. & Hangartner, L. Broadly neutralizing antibodies toHIV and their role in vaccine design. Annu. Rev. Immunol. 34, 635659

2016:

2. Hessell, A. J. et al. Broadly neutralizing monoclonal antibodies 2F5 and 4E10 directed against the human immunodeciency virus type 1 gp41 membrane-proximal external region protect against mucosal challenge by simian-human immunodeciency virus SHIVBa-L. J. Virol. 84, 13021313 (2010).

3. Moldt, B. et al. Highly potent HIV-specic antibody neutralization in vitro translates into effective protection against mucosal SHIV challenge in vivo. Proc. Natl Acad. Sci. USA 109, 1892118925 (2012).

4. Kwong, P. D. & Mascola, J. R. Human antibodies that neutralize HIV-1: identication, structures, and B cell ontogenies. Immunity 37, 412425 (2012).

5. Sanders, R. W. et al. HIV-1 neutralizing antibodies induced by native-like envelope trimers. Science 349, aac4223 (2015).

6. Jardine, J. G. et al. Priming a broadly neutralizing antibody response to HIV-1 using a germline-targeting immunogen. Science 349, 156161 (2015).

7. Reitter, J. N., Means, R. E. & Desrosiers, R. C. A role for carbohydrates in immune evasion in AIDS. Nat. Med. 4, 679684 (1998).

8. Wei, X. P. et al. Antibody neutralization and escape by HIV-1. Nature 422, 307312 (2003).

9. Chackerian, B., Rudensey, L. M. & Overbaugh, J. Specic N-linked and O-linked glycosylation modications in the envelope V1 domain of simian immunodeciency virus variants that evolve in the host alter recognition by neutralizing antibodies. J. Virol. 71, 77197727 (1997).

10. Back, N. K. T. et al. An N-glycan within the human-immunodeciency-virus type-1 Gp120 V3 loop affects virus neutralization. Virology 199, 431438 (1994).

11. Julien, J. P. et al. Crystal structure of a soluble cleaved HIV-1 envelope trimer. Science 342, 14771483 (2013).

12. Falkowska, E. et al. Broadly neutralizing HIV antibodies dene a glycan-dependent epitope on the prefusion conformation of gp41 on cleaved envelope trimers. Immunity 40, 657668 (2014).

13. Pancera, M. et al. Structural basis for diverse N-glycan recognition by HIV-1-neutralizing V1-V2-directed antibody PG16. Nat. Struct. Mol. Biol. 20, 804 (2013).

14. Alam, S. M. et al. Recognition of synthetic glycopeptides by HIV-1 broadly neutralizing antibodies and their unmutated ancestors. Proc. Natl Acad. Sci. USA 110, 1821418219 (2013).

15. Julien, J. P. et al. Broadly neutralizing antibody PGT121 allosterically modulates CD4 binding via recognition of the HIV-1 gp120 V3 base and multiple surrounding glycans. PLoS Pathog. 9, e1003342 (2013).

16. Garces, F. et al. Structural evolution of glycan recognition by a family of potent HIV antibodies. Cell 159, 6979 (2014).

17. Sok, D. et al. Promiscuous glycan site recognition by antibodies to the high-mannose patch of gp120 broadens neutralization of HIV. Sci. Transl. Med. 6, 236ra63 (2014).

18. Amin, M. N. et al. Synthetic glycopeptides reveal the glycan specicity of HIV-neutralizing antibodies. Nat. Chem. Biol. 9, 521526 (2013).

19. Walker, L. M. et al. Broad neutralization coverage of HIV by multiple highly potent antibodies. Nature 477, 466U117 (2011).

20. Raska, M. et al. Glycosylation patterns of HIV-1 gp120 depend on the type of expressing cells and affect antibody recognition. J. Biol. Chem. 285, 2086020869 (2010).

21. Kong, L. et al. Expression-system-dependent modulation of HIV-1 envelope glycoprotein antigenicity and immunogenicity. J. Mol. Biol. 403, 131147 (2010).

22. Pritchard, L. K., Harvey, D. J., Bonomelli, C., Crispin, M. & Doores, K. J. Cell- and protein-directed glycosylation of native cleaved HIV-1 envelope.J. Virol. 89, 89328944 (2015).23. Pritchard, L. K. et al. Structural constraints determine the glycosylation of HIV-1 envelope trimers. Cell Rep. 11, 16041613 (2015).

24. Behrens, A. J. et al. Composition and antigenic effects of individualglycan sites of a trimeric HIV-1 envelope glycoprotein. Cell Rep. 14, 26952706 (2016).

25. Doores, K. J. et al. Envelope glycans of immunodeciency virions are almost entirely oligomannose antigens. Proc. Natl Acad. Sci. USA 107, 1380013805 (2010).

26. Bonomelli, C. et al. The glycan shield of HIV is predominantly oligomannose independently of production system or viral clade. PLoS ONE 6, e23521 (2011).

27. Go, E. P. et al. Glycosylation site-specic analysis of clade C HIV-1 envelope proteins. J. Proteome. Res. 8, 42314242 (2009).

NATURE COMMUNICATIONS | 8:14954 | DOI: 10.1038/ncomms14954 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 11

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14954

28. Go, E. P. et al. Comparative analysis of the glycosylation proles of membrane-anchored HIV-1 envelope glycoprotein trimers and soluble gp140. J. Virol. 89, 82458257 (2015).

29. Go, E. P. et al. Characterization of glycosylation proles of HIV-1 transmitted/ founder envelopes by mass spectrometry. J. Virol. 85, 82708284 (2011).30. Go, E. P. et al. Glycosylation site-specic analysis of HIV envelope proteins (JR-FL and CON-S) reveals major differences in glycosylation site occupancy, glycoform proles, and antigenic epitopes accessibility. J. Proteome Res. 7, 16601674 (2008).

31. Go, E. P. et al. Characterization of host-cell line specic glycosylation proles of early transmitted/founder HIV-1 gp120 envelope proteins. J. Proteome Res. 12, 12231234 (2013).

32. Leonard, C. K. et al. Assignment of intrachain disulde bonds and characterization of potential glycosylation sites of the type-1 recombinant human-immunodeciency-virus envelope glycoprotein (Gp120) expressed in Chinese hamster ovary cells. J. Biol. Chem. 265, 1037310382 (1990).

33. Zhu, X. G., Borchers, C., Bienstock, R. J. & Tomer, K. B. Mass spectrometric characterization of the glycosylation pattern of HIV-gp120 expressed in CHO cells. Biochemistry 39, 1119411204 (2000).

34. Behrens, A. J. et al. Molecular architecture of the cleavage-dependent mannose patch on a soluble HIV-1 envelope glycoprotein trimer. J. Virol. 91, e0189416 (2017).

35. Go, E. P. et al. Methods development for analysis of partially deglycosylated proteins and application to an HIV envelope protein vaccine candidate. Int. J. Mass Spectrom. 305, 209216 (2011).

36. Hagglund, P. et al. An enzymatic deglycosylation scheme enabling identication of core fucosylated N-glycans and O-glycosylation site mapping of human plasma proteins. J. Proteome Res. 6, 30213031 (2007).

37. Angel, P. M., Lim, J. M., Wells, L., Bergmann, C. & Orlando, R. A potential pitfall in O-18-based N-linked glycosylation site mapping. Rapid Commun. Mass Sp 21, 674682 (2007).

38. Kaji, H. et al. Lectin afnity capture, isotope-coded tagging and mass spectrometry to identify N-linked glycoproteins. Nat. Biotechnol. 21, 667672 (2003).

39. Khemiri, A. et al. N-Glycosidase treatment with O-18 labeling and de novo sequencing argues for agellin FliC glycopolymorphism in Pseudomonas aeruginosa. Anal. Bioanal. Chem. 405, 98359842 (2013).

40. Chen, R. et al. Development of a combined chemical and enzymatic approach for the mass spectrometric identication and quantication of aberrant N-glycosylation. J. Proteomics 75, 16661674 (2012).

41. Kolarich, D., Jensen, P. H., Altmann, F. & Packer, N. H. Determination of site-specic glycan heterogeneity on glycoproteins. Nat. Protoc. 7, 12851298 (2012).

42. Gatlin, C. L., Eng, J. K., Cross, S. T., Detter, J. C. & Yates, J. R. Automated identication of amino acid sequence variations in proteins by HPLC/ microspray tandem mass spectrometry. Anal. Chem. 72, 757763 (2000).43. Eng, J. K., Mccormack, A. L. & Yates, J. R. An approach to correlate tandem mass-spectral data of peptides with amino-acid-sequences in a protein database. J. Am. Soc. Mass Spectrom. 5, 976989 (1994).

44. Xu, T. et al. ProLuCID: an improved SEQUEST-like algorithm with enhanced sensitivity and specicity. J. Proteomics 129, 1624 (2015).

45. Tabb, D. L., McDonald, W. H. & Yates, J. R. DTASelect and contrast: tools for assembling and comparing protein identications from shotgun proteomics.J. Proteome Res. 1, 2126 (2002).46. He, L., Diedrich, J., Chu, Y. Y. & Yates, J. R. Extracting accurate precursor information for tandem mass spectra by raw converter. Anal. Chem. 87, 1136111367 (2015).

47. Park, S. K., Venable, J. D., Xu, T. & Yates, J. R. A quantitative analysis software tool for mass spectrometry-based proteomics. Nat. Methods 5, 319322 (2008).

48. Lee, P. S. et al. Receptor mimicry by antibody F045-092 facilitates universal binding to the H3 subtype of inuenza virus. Nat. Commun. 5, 3614 (2014).

49. Zhu, Z. K., Go, E. P. & Desaire, H. Absolute quantitation of glycosylation site occupancy using isotopically labeled standards and LC-MS. J. Am. Soc. Mass Spectrom. 25, 10121017 (2014).

50. An, Y. M., McCullers, J. A., Alymova, I., Parsons, L. M. & Cipollo, J. F. Glycosylation analysis of engineered H3N2 inuenza A virus hemagglutinins with sequentially added historically relevant glycosylation sites. J. Proteome Res. 14, 39573969 (2015).

51. Khatri, K. et al. Integrated omics and computational glycobiology reveal structural basis for inuenza A virus glycan microheterogeneity and host interactions. Mol. Cell. Proteomics 15, 18951912 (2016).

52. Sanders, R. W. et al. A next-generation cleaved, soluble HIV-1 Env trimer, BG505 SOSIP.664 gp140, expresses multiple epitopes for broadly neutralizing but not non-neutralizing antibodies. PLoS Pathog. 9, e1003618 (2013).

53. Lyumkis, D. et al. Cryo-EM structure of a fully glycosylated soluble cleaved HIV-1 envelope trimer. Science 342, 14841490 (2013).

54. Do Kwon, Y. et al. Crystal structure, conformational xation and entry-related interactions of mature ligand-free HIV-1 Env. Nat. Struct. Mol. Biol. 22, 522 (2015).

55. Garces, F. et al. Afnity maturation of a potent family of HIV antibodies is primarily focused on accommodating or avoiding glycans. Immunity 43, 10531063 (2015).

56. Guttman, M. et al. CD4-induced activation in a soluble HIV-1 Env trimer. Structure 22, 974984 (2014).

57. McLellan, J. S. et al. Structure of HIV-1 gp120 V1/V2 domain with broadly neutralizing antibody PG9. Nature 480, 336U386 (2011).

58. Mouquet, H. et al. Complex-type N-glycan recognition by potent broadly neutralizing HIV antibodies. Proc. Natl Acad. Sci. USA 109, E3268E3277 (2012).

59. Mathys, L., Francois, K. O., Quandte, M., Braakman, I. & Balzarini, J. Deletion of the highly conserved N-glycan at Asn260 of HIV-1 gp120 affects folding and lysosomal degradation of gp120, and results in loss of viral infectivity. PLoS ONE 9, e101181 (2014).

60. Blattner, C. et al. Structural delineation of a quaternary, cleavage-dependent epitope at the gp41-gp120 interface on intact HIV-1 Env trimers. Immunity 40, 669680 (2014).

61. Depetris, R. S. et al. Partial enzymatic deglycosylation preserves the structure of cleaved recombinant HIV-1 envelope glycoprotein trimers. J. Biol. Chem. 287, 2423924254 (2012).

62. Sliepen, K., van Montfort, T., Melchers, M., Isik, G. & Sanders, R. W. Immunosilencing a highly immunogenic protein trimerization domain. J. Biol. Chem. 290, 74367442 (2015).

63. Stavenhagen, K. et al. Quantitative mapping of glycoprotein micro-heterogeneity and macro-heterogeneity: an evaluation of mass spectrometry signal strengths using synthetic peptides and glycopeptides. J. Mass Spectrom. 48, 627639 (2013).

64. Sanders, R. W. et al. The mannose-dependent epitope for neutralizing antibody 2G12 on human immunodeciency virus type 1 glycoprotein gp120. J. Virol. 76, 72937305 (2002).

65. Pugach, P. et al. A native-like SOSIP.664 trimer based on an HIV-1 subtype B env Gene. J. Virol. 89, 33803395 (2015).

66. Westrop, S. J., Jackson, A., Gazzard, B. & Imami, N. HIV-1 clade A infection and viral control: an immunological perspective on a case of underquantication. Int. J. STD AIDS 22, 690692 (2011).

67. Jardine, J. et al. Rational HIV immunogen design to target specic germline B cell receptors. Science 340, 711716 (2013).

68. McGuire, A. T. et al. Engineering HIV envelope protein to activate germline B cell receptors of broadly neutralizing anti-CD4 binding site antibodies. J. Exp. Med. 210, 655663 (2013).

69. McGuire, A. T. et al. Antigen modication regulates competition of broad and narrow neutralizing HIV antibodies. Science 346, 13801383 (2014).

70. Jardine, J. G. et al. HIV-1 broadly neutralizing antibody precursor B cells revealed by germline-targeting immunogene. Science 351, 14581463 (2016).

71. Steichen, J. M. et al. HIV vaccine design to target germline precursors of glycan-dependent broadly neutralizing antibodies. Immunity 45, 483496 (2016).

72. Pritchard, L. K. et al. Glycan microheterogeneity at the PGT135 antibody recognition site on HIV-1 gp120 reveals a molecular mechanism for neutralization resistance. J. Virol. 89, 69526959 (2015).

73. Rerks-Ngarm, S. et al. Vaccination with ALVAC and AIDSVAX to prevent HIV-1 infection in Thailand. N. Engl. J. Med. 361, 22092220 (2009).

74. Struwe, W. B., Stuckmann, A., Behrens, A. J., Pagel, K. & Crispin, M. Global N-glycan site occupancy of HIV-1 gp120 by metabolic engineering and high-resolution intact mass spectrometry. ACS Chem. Biol. 12, 357361 (2016).

75. Guenaga, J. et al. Well-ordered trimeric HIV-1 subtype B and C soluble spike mimetics generated by negative selection display native-like properties. PLoS Pathog. 11, e1004570 (2015).

76. Andrabi, R. et al. Identication of common features in prototype broadly neutralizing antibodies to HIV envelope V2 apex to facilitate vaccine design. Immunity 43, 959973 (2015).

77. Chang, H. W. et al. Generation and evaluation of clade C simian-human immunodeciency virus challenge stocks. j. Virol. 89, 19651974 (2015).

78. Barbian, H. J. et al. Neutralization properties of simian immunodeciency viruses infecting chimpanzees and gorillas. Mbio 6, e0029615 (2015).

79. Panico, M. et al. Mapping the complete glycoproteome of virion-derived HIV-1 gp120 provides insights into broadly neutralizing antibody binding. Sci. Rep. 6, e0029615 (2016).

80. Livi, G. P. et al. Secretion of N-glycosylated interleukin-1-beta in Saccharomyces-cerevisiae using a leader peptide from Candida-albicans-effect of N-linked glycosylation on biological-activity. J. Biol. Chem. 266, 1534815355 (1991).

81. Elliott, S., Chang, D., Delorme, E., Eris, T. & Lorenzini, T. Structural requirements for additional N-linked carbohydrate on recombinant human erythropoietin. J. Biol. Chem. 279, 1685416862 (2004).

12 NATURE COMMUNICATIONS | 8:14954 | DOI: 10.1038/ncomms14954 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms14954 ARTICLE

82. Ujvari, A. et al. Translation rate of human tyrosinase determines its N-linked glycosylation level. J. Biol. Chem. 276, 59245931 (2001).

83. Cao, L. W. et al. N-glycosylation site analysis of proteins from Saccharomyces cerevisiae by using hydrophilic interaction liquid chromatography-based enrichment, parallel deglycosylation, and mass spectrometry. J. Proteome. Res. 13, 14851493 (2014).

84. Li, Y., Luo, L. Z., Rasool, N. & Kang, C. Y. Glycosylation is necessary for the correct folding of human immunodeciency virus-Gp120 in Cd4 binding.J. Virol. 67, 584588 (1993).85. Quinones-Kochs, M. I., Buonocore, L. & Rose, J. K. Role of N-linked glycans in a human immunodeciency virus envelope glycoprotein: effects on protein function and the neutralizing antibody response. J. Virol. 76, 41994211 (2002).

86. Gristick, H. B. et al. Natively glycosylated HIV-1 Env structure reveals new mode for antibody recognition of the CD4-binding site. Nat. Struct. Mol. Biol. 23, 906915 (2016).

87. Hao, P. L., Ren, Y., Alpert, A. J. & Sze, S. K. Detection, evaluation and minimization of nonenzymatic deamidation in proteomic sample preparation. Mol. Cell. Proteomics 10, O111.009381 (2011).

88. Pancera, M. et al. Structure and immune recognition of trimeric pre-fusion HIV-1 Env. Nature 514, 455 (2014).

89. Huang, P. S. et al. RosettaRemodel: a generalized framework for exible backbone protein design. PLoS ONE 6, e24109 (2011).

90. Pancera, M. et al. Structure of HIV-1 gp120 with gp41-interactive region reveals layered envelope architecture and basis of conformational mobility. Proc. Natl Acad. Sci. USA 107, 11661171 (2010).

Acknowledgements

We thank Dr Ian A. Wilson and Dr Peter S. Lee for providing recombinant inuenza virus haemagglutinin. This work was supported by NIH R01AI113867 (J.C.P., J.Y., W.R.S.); NIH UM1 AI100663 (D.R.B.); the International AIDS Vaccine Initiative (W.R.S., D.R.B.); and NIH P41 GM103533 (J.Y.).

Author contributions

L.C., J.R.Y. and J.C.P. designed the research. L.C. prepared samples for MS analysis. J.K.D., L.C. and C.M.D. performed the MS analysis. L.C., L.H. and S.R.P. analysed the data. D.W.K., M.M., D.S., C.Y.S., S.M., R.A., J.G., E.G., M.K. and Y.A. expressed and puried Env proteins. D.R.B., W.R.S., J.R.Y. and J.C.P. supervised the project. L.C. and J.C.P. wrote the manuscript.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing interests: The authors declare no competing nancial interests.

Reprints and permission information is available online at http://npg.nature.com/reprintsandpermissions/

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Cao, L. et al. Global site-specic N-glycosylation analysis of HIV envelope glycoprotein. Nat. Commun. 8, 14954 doi: 10.1038/ncomms14954 (2017).

Publishers note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional afliations.

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

r The Author(s) 2017

NATURE COMMUNICATIONS | 8:14954 | DOI: 10.1038/ncomms14954 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 13